Neurosurgery

NEUROSURGERY TABLE OF CONTENTS BY TOPIC FEBRUARY 2007 VOLUME 60 NUMBER 2 Pages 213–416 Articles are grouped by categories, complete author listings can be found on preceding Table of Contents CELLULAR BIOLOGY 392 History of Spine Biomechanics: Part II—From 346 Multiple Differentiation Potentials of Neonatal Dura the Renaissance to the 20th Century: Sait Naderi Mater-derived Cells: Ioana A. Peptan MICROVASCULAR 353 Interaction Between Krit1 and Malcavernin: Implications 227 Aneurysms of the Distal Anterior Cerebral Artery: Results for the Pathogenesis of Cerebral Cavernous Malformations: in 59 Consecutively Managed Patients: David A. Steven Jun Zhang 235 No Long-term Excess Mortality in 280 Patients with CEREBROVASCULAR Ruptured Distal Anterior Cerebral Artery Aneurysms: 213 Genetics of Intracranial Aneurysms: Brian V. Nahed Martin Lehecka 227 Aneurysms of the Distal Anterior Cerebral Artery: Results 242 Results of Microsurgical Clipping of 50 High in 59 Consecutively Managed Patients: David A. Steven Complexity Basilar Apex Aneurysms: Ali F. Krisht 235 No Long-term Excess Mortality in 280 Patients with MOLECULAR & CELLULAR BIOLOGY Ruptured Distal Anterior Cerebral Artery Aneurysms: 213 Genetics of Intracranial Aneurysms: Brian V. Nahed Martin Lehecka 338 Expression of Hypoxia-inducing Factor-1α and Endoglin 242 Results of Microsurgical Clipping of 50 High in Intimal Hyperplasia of the Middle Cerebral Artery of Complexity Basilar Apex Aneurysms: Ali F. Krisht Patients with Moyamoya Disease: Yasushi Takagi 253 Treatment of Cavernous Sinus Dural Arteriovenous Fistulae 372 The In Vivo Antitumoral Effects of Lipopolysaccharide by External Manual Carotid Compression: Yutaka Kai against Glioblastoma Multiforme Are Mediated in Part 259 Subarachnoid Clot Volume Correlates with Age, by Toll-like Receptor 4: Michael R. Chicoine Neurological Grade, and Blood Pressure: David S. Rosen PEDIATRICS 268 Role of Radiosurgery in the Management of Cerebral 268 Role of Radiosurgery in the Management of Cerebral Arteriovenous Malformations in the Pediatric Age Group: Arteriovenous Malformations in the Pediatric Age Group: Data from a 100-patient Series: Nicolas Reyns Data from a 100-patient Series: Nicolas Reyns 338 Expression of Hypoxia-inducing Factor-1α and Endoglin RADIOSURGERY in Intimal Hyperplasia of the Middle Cerebral Artery of 268 Role of Radiosurgery in the Management of Cerebral Patients with Moyamoya Disease: Yasushi Takagi Arteriovenous Malformations in the Pediatric Age Group: 353 Interaction Between Krit1 and Malcavernin: Implications Data from a 100-patient Series: Nicolas Reyns for the Pathogenesis of Cerebral Cavernous Malformations: 277 A Pilot Study of Neurocognitive Function in Patients with Jun Zhang One to Three New Brain Metastases Initially Treated with ENDOVASCULAR Stereotactic Radiosurgery Alone: Eric L. Chang 227 Aneurysms of the Distal Anterior Cerebral Artery: Results SPINE in 59 Consecutively Managed Patients: David A. Steven 382 History of Spine Biomechanics: Part I—The Pre-Greco- 235 No Long-term Excess Mortality in 280 Patients with Roman, Greco-Roman, and Medieval Roots of Spine Ruptured Distal Anterior Cerebral Artery Aneurysms: Biomechanics: Sait Naderi Martin Lehecka 392 History of Spine Biomechanics: Part II—From EPIDEMIOLOGY the Renaissance to the 20th Century: Sait Naderi 213 Genetics of Intracranial Aneurysms: Brian V. Nahed SPORT 307 Objectifying When to Halt a Boxing Match: FUNCTIONAL NEUROSURGERY A Video Analysis of Fatalities: Vincent J. Miele 296 Expressive and Receptive Language Areas Determined by a Non-invasive Reliable Method Using Functional 405 The War of the Gods: Lissa C. Baird Magnetic Resonance Imaging and Magnetoencephalography: STEREOTAXY Kyousuke Kamada 268 Role of Radiosurgery in the Management of Cerebral Arteriovenous Malformations in the Pediatric Age Group: GENERAL NEUROSURGERY Data from a 100-patient Series: Nicolas Reyns 317 Risk Factors Associated with Postcraniotomy Meningitis: Irene S. Kourbeti 277 A Pilot Study of Neurocognitive Function in Patients with One to Three New Brain Metastases Initially Treated with HYDROCEPHALUS Stereotactic Radiosurgery Alone: Eric L. Chang 327 Long-term Outcome in Patients with Suspected Normal TRAUMA Pressure Hydrocephalus: Babar Kahlon 307 Objectifying When to Halt a Boxing Match: 333 Adjustable Shunt Valve Reprogramming at Home: A Video Analysis of Fatalities: Vincent J. Miele Safety and Feasibility: Christian W. Sikorski 405 The War of the Gods: Lissa C. Baird IMAGING TUMOR 296 Expressive and Receptive Language Areas Determined 277 A Pilot Study of Neurocognitive Function in Patients with by a Non-invasive Reliable Method Using Functional One to Three New Brain Metastases Initially Treated with Magnetic Resonance Imaging and Magnetoencephalography: Stereotactic Radiosurgery Alone: Eric L. Chang Kyousuke Kamada 285 Surgery for Temporal Mediobasal Tumors: Experience 360 In Vivo Imaging in a Murine Model of Glioblastoma: Based on a Series of 235 Patients: Johannes Schramm Sarah C. Jost 360 In Vivo Imaging in a Murine Model of Glioblastoma: INFECTION Sarah C. Jost 317 Risk Factors Associated with Postcraniotomy Meningitis: 372 The In Vivo Antitumoral Effects of Lipopolysaccharide Irene S. Kourbeti against Glioblastoma Multiforme Are Mediated in Part LEGACY by Toll-like Receptor 4: Michael R. Chicoine 382 History of Spine Biomechanics: Part I—The Pre-Greco- VASCULAR Roman, Greco-Roman, and Medieval Roots of Spine 253 Treatment of Cavernous Sinus Dural Arteriovenous Fistulae Biomechanics: Sait Naderi by External Manual Carotid Compression: Yutaka Kai

Entire journal content can be viewed online at www.neurosurgery-online.com ARTICLES IN RED indicate free access for non-subscribers. N4 CLINICAL REVIEW

GENETICS OF INTRACRANIAL ANEURYSMS

Brian V. Nahed, M.D. DESPITE ADVANCES IN the treatment of intracranial aneurysms (IA) in recent years, Department of Neurosurgery, the overall outcome of patients with aneurysmal subarachnoid hemorrhage has shown Yale University School of Medicine, only modest improvement. Given this poor prognosis, diagnosis of IA before rupture New Haven, Connecticut is of paramount importance. Currently, there are no reliable methods other than screen- Mohamad Bydon, B.A. ing imaging studies of high-risk individuals to diagnose asymptomatic patients. Multiple Department of Neurosurgery, levels of evidence suggest that environmental factors acting in concert with genetic Yale University School of Medicine, New Haven, Connecticut susceptibilities lead to the formation, growth, and rupture of aneurysms in these patients. Epidemiological studies have already identified aneurysm-specific risk factors such as Ali K. Ozturk, M.D. size and location, as well as patient-specific risk factors, such as age, sex, and pres- Department of Neurosurgery, ence of medical comorbidities, such as hypertension. In addition, exposure to certain Yale University School of Medicine, New Haven, Connecticut environmental factors such as smoking have been shown to be important in the formation of IA. Furthermore, substantial evidence proves that certain loci contribute geneti- Kaya Bilguvar, M.D. cally to IA pathogenesis. Genome-wide linkage studies using relative pairs or rare fam- Department of Neurosurgery, ilies that are affected with the Mendelian forms of IA have already shown genetic Yale University School of Medicine, New Haven, Connecticut heterogeneity of IA, suggesting that multiple genes, alone or in combination, are important in the disease pathophysiology. The linkage results, along with association studies, Fatih Bayrakli, M.D. will ultimately lead to the identification of IA susceptibility genes. Identification of the Department of Neurosurgery, Yale University School of Medicine, genes important in IA pathogenesis will not only provide novel insights into the primary New Haven, Connecticut determinants of IA, but will also result in new opportunities for early diagnosis in the preclinical setting. Ultimately, novel therapeutic strategies based on biology will be Murat Gunel, M.D. developed, which will target these newly elucidated genetic susceptibilities. Department of Neurosurgery, Yale University School of Medicine, KEY WORDS: Environmental, Genetics, Intracranial aneurysms, Pathogenesis New Haven, Connecticut Neurosurgery 60:213–226, 2007 DOI: 10.1227/01.NEU.0000249270.18698.BB www.neurosurgery-online.com Yale Brain Aneurysm and AVM Center, The Anylan Center for Human Genetics and Genomics (BVN, MB, AKO, KB, FB, MG) ubarachnoid hemorrhage (SAH) owing to According to various estimates, aneurysmal Reprint requests: the rupture of an intracranial aneurysm SAH accounts for 3 to 11% of all strokes, (48, 92), Murat Gunel, M.D., S(IA) is a serious neurosurgical emer- 5% of stroke deaths, and more than one-quarter Yale University School of Medicine, gency with poor prognosis; approximately of potential life years lost through stroke (32). Department of Neurosurgery, TMP 4, 12% of patients die before reaching medical Although the 20th century has seen great 333 Cedar Street, attention (91) and 40% die in the hospital (31, advances in the diagnosis, treatment, and pre- New Haven, CT 06510. Email: [email protected] 35, 61). Only 25% of those who live past the vention of complications of SAH, the overall out- first month recover completely, leaving the come has shown only modest improvement (30). Received, December 28, 2005. majority of survivors requiring long-term care Given the devastating sequelae of aneurys- Accepted, September 8, 2006. (27, 30, 92). The incidence of SAH is 6 in mal SAH, surgical or endovascular interven- 100,000, with approximately 28,000 ruptures tion before rupture is considered to be of para- per year in the United States. This number mount importance. Guidelines have been may be even higher in other countries, such established to assist in the decision between as Finland, where the incidence has been treatment and careful monitoring, with the reported to be as high as 33 in 100,000 among goal of prophylactically treating those aneu- Finnish men, which is likely owing to higher rysms that are likely to rupture. Attempts to rates of rupture of existing aneurysms, rather identify risk factors and the pathophysiology than a greater number of aneurysms per se leading to aneurysm formation and rupture (82, 87). In the absence of trauma, IA is the have had limited success, although there has leading cause of SAH and accounts for ap- been noteworthy progress of late, particularly proximately 85% of the cases (41). in the study of aneurysm genetics.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 213 NAHED ET AL.

Intracranial Aneurysms of patients present with SAH owing to aneurysms smaller than IA affects 5 to 10% of the general population (54) and repre- 7 mm in daily clinical practice. A recent study analyzing 280 sents a major public health problem. It is estimated that 2.3% of ruptured aneurysms showed that 74% were smaller than the population have undetected aneurysms (33), a fraction of 10 mm (mean, 7.6 mm) (68), suggesting that factors other than which will rupture and lead to devastating consequences. aneurysm size are important in determining the risk of Progress in understanding the pathogenesis of IA has been aneurysmal rupture. hampered by its multifactorial nature. For many years, the decision to operate on an unruptured aneurysm was based Other Risk Factors solely on the size of the aneurysm as determined by imaging studies. One large multicenter trial suggested that IAs larger Many studies have attempted to identify risk factors, other than 10 mm had a risk of rupture of 1% per year, with smaller than size, that predict the formation, growth, and rupture of aneurysms having a much lower risk. However, this initial data IAs (20, 39, 60, 86, 105). Factors such as sex, hypertension, ath- conflicted with clinical experience in which a significant num- erosclerosis, diabetes, and vascular anatomic differences have ber of patients present with SAH owing to aneurysms smaller been implicated in its pathogenesis (40, 69). than 10 mm in size. Furthermore, the data contradicted a pre- Sex is an important risk factor for IA and for SAH in general. viously published series in which aneurysms smaller than A Finnish study found that, among lean, hypertensive smokers, 10 mm were at risk of rupture (38). women were three times more likely to experience SAH than Other than aneurysm size, the conditions that lead to men after adjusting for age (46). A more recent study found aneurysm formation and rupture have not been fully delin- female sex to be a significant independent risk factor for de eated. Contributing to the disease are a number of well- novo aneurysm formation (adjusted odds ratio, 4.73) (35). established risk factors such as sex, hypertension, smoking, A Johns Hopkins study of 419 patients with multiple IAs found alcohol, low body mass index, and drug use (35, 36). These that female sex was significantly associated with the risk of risk factors, however, fall short of a complete explanation. developing multiple aneurysms (odds ratio, 1.9) (79). Recent studies suggest that, in addition to these, genetic factors In addition, social factors such as smoking and diet have also contribute to the pathogenesis of IA. Given the large num- also been suggested to play a role in the disease (38). A review ber of familial cases and increased incidence with other genetic by Teunissen et al. (96) revealed only hypertension, cigarette diseases such as adult polycystic kidney disease (ADPKD) (52), smoking, and alcohol consumption (greater than 150 g/wk) as a genetic contribution to aneurysm formation and rupture has significant risk factors. These results were confirmed in a long been speculated. review of five trials in North America, Canada, and Europe The degree to which each contributes to an individual’s which demonstrated cigarette smoking to be a major risk factor aneurysm is likely patient-specific. In this review, we consider (106). Although family history and the above-mentioned mod- both the genetic mechanisms and environmental factors that ifiable risk factors have been suggested to increase the risk of work independently and/or synergistically to form IA. rupture, there is insufficient data to support a definitive clinical recommendation for the surgical treatment of small Environmental Risk Factors aneurysms (2, 3, 6, 34, 37, 38, 63, 66, 68, 80, 106). In a study of 280 ruptured aneurysms, the mean size of rup- Based on a number of association studies linking environ- tured aneurysms in patients who were normotensive, had med- mental risk factors with the formation, growth, and rupture of ically controlled hypertension, and had poorly controlled aneurysms, guidelines have been established to identify which hypertension were 8.3, 7.4, and 6.5 mm, respectively (68). aneurysms should be treated and which need to be watched. Furthermore, patients with a family history of SAH or who These guidelines have been largely based on aneurysm size had poorly controlled hypertension were more likely to have and location. ruptured aneurysms smaller than 5 mm. With respect to location, ruptured aneurysms of the anterior communicating artery Aneurysm Size were smaller on average (6.6 mm) than aneurysms in other The notion that aneurysm size correlates with rupture risk is locations. These results confirmed the results of a previous as old as aneurysm surgery itself. The prospective arm of the single-center retrospective analysis that showed that 50% of International Study of Unruptured Intracranial Aneurysms ruptured aneurysms were 6 to 10 mm in size (20). suggested that treatment with either microsurgical or endovas- However, the results of these studies have not yet been incor- cular techniques should be considered with aneurysms larger porated into the treatment guidelines. Our group extended than 7 mm in size, with the greatest benefit when aneurysms these findings specifically to a subgroup of patients who harbor larger than 7 mm of the posterior communicating artery are aneurysms smaller than 7 mm and showed that hypertension, surgically treated in young patients (Ͻ50 yr) (107). Further- younger age, and posterior circulation are significant risk fac- more, it recommended that aneurysms smaller than 7 mm of tors for rupture in the subgroup of patients who have small the anterior circulation in patients with no history of SAH aneurysms (64). Interestingly, other studies, including the should be left untreated, a view that has been reinforced by prospective arm of the International Study of Unruptured others (63). Despite this recommendation, a significant number Intracranial Aneurysms study, showed that aneurysms of the

214 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com PATHOGENESIS OF INTRACRANIAL ANEURYSMS

posterior circulation were more likely to rupture at smaller patients (20%) had a first- or second-degree relative with sizes (Յ7 mm) (107). aneurysmal SAH (88). The number of observed first-degree rel- A recent study by Frosen et al. (22) advanced our under- atives with aneurysmal SAH was 11, compared with an standing of the pathogenesis of aneurysm rupture. The authors expected number of 2.66, resulting in a relative risk of 4.14. used immunohistochemical staining to examine the cellular Nakagawa et al. (65) found a significantly higher incidence of mechanisms that precede rupture of saccular cerebral artery asymptomatic cerebral aneurysms among Japanese patients aneurysms. Using surgically resected aneurysms (24 unrup- with family history of SAH within the second degree of consan- tured and 42 ruptured) of similar size, the authors were able to guinity versus healthy volunteers (13.9 versus 6%) (65). When identify types of aneurysm wall associated with rupture. They combined with other risk factors such as hypertension and confirmed that the blood vessel undergoes morphological habitual smoking, these patients were found to have the high- changes before rupture, including endothelial apoptosis, de- est incidence. endothelialization, and luminal thrombosis, as well as smooth Although there has been much debate regarding this issue, muscle cell proliferation and macrophage infiltration, both of we advocate screening in patients with two or more first-degree which are likely to be involved in the insufficient repair mech- family members with IA. In patients participating in our genet- anisms that precede rupture (22). Such molecular studies will ics studies, we found that 20% of at-risk, first-degree family ultimately prove to be useful in understanding the pathogene- members had positive radiological findings (unpublished sis of aneurysms, but they cannot address the etiological ques- data). In our experience, magnetic resonance angiography is a tions that surround the disease. useful screening tool, followed by computed tomographic Although there is evidence for environmental factors con- angiography or digital subtraction angiography to confirm or tributing to the pathogenesis of IAs, they fail to explain the exclude equivocal cases. complete picture, particularly in young adults. Thus, genetic Several studies have reported that cases of familial IA behave factors have been suggested to play a role in the pathogenesis differently than cases of sporadic IA. Familial cases were of aneurysm formation. detected at an earlier age and ruptured at a smaller size when compared with sporadic cases (8, 42, 53, 58, 81). In fact, 70% of Genetics familial IAs rupture by the age of 50 versus 43% of non-famil- It is now accepted that, in addition to environmental risk ial aneurysms (58). Additionally, a study by Leblanc et al. (53) factors, genetic risk factors contribute to the formation and pos- found higher than expected concordance of the age at rupture sible rupture of cerebral aneurysms. In this section, we high- in a prospective study of 30 individuals in 13 families with light familial intracranial aneurysms and previous linkage multiple affected individuals. In the Saguenay-Lac Saint Jean analysis studies. region of the Province of Quebec, Canada, Mathieu et al. (59) found that siblings of patients with ruptured IA had a greater Familial Aneurysms risk of ruptured IA than the general population. The notion that aneurysms cluster in families was first The genetic mode of IA transmission from generation to gen- noticed in identical twins during the 1960s (99). Ullrich (99) eration has been difficult to determine. Schievink et al.’s (89) reported four families, each of which had at least two mem- review of the literature covering 238 affected families did not bers with cerebral aneurysms (99). This was followed by sev- find any one pattern of inheritance nor a Mendelian model that eral case reports of multiple familial IAs (5, 7, 18, 25, 97, 98). applied uniformly. The most commonly affected relatives were In 1980, Fox and Ko (21) reported the largest family to date: siblings. Twenty-two percent of siblings of male probands had 13 siblings, six of whom had proven IA and five of whom had an IA compared with 9% of siblings of female probands. normal findings on cerebral angiograms; two of the siblings Interestingly, angiographic screening in 12 families detected refused angiography. Subsequently, one of the two who refused IAs in 29% of 51 asymptomatic relatives (89). These studies workup experienced a SAH and was found to have two show the complexity of intracranial aneurysm genetics and aneurysms. Interestingly, there was no disease in their parents point to locus heterogeneity. and relatives. Since then, we have ascertained and recruited this family to our study (IA 20) and have found that there are newly Candidate Genes affected members in subsequent generations (64). Although several candidate studies have implicated the role In 1993, Ronkainen et al. (80) reported a 10% incidence of of various genes in the pathogenesis of aneurysms, none have familial IAs in family members of 1130 patients with proven led to any significant findings. These genes range from those aneurysmal SAH from east Finland. Similarly, Kojima et al. associated with vascular wall formation to those that are (47) reported a 10% prevalence of IAs in families with positive mutated in connective tissue disorders. By studying known history. Two subsequent reports, a prospective and a retrospec- genetic diseases with high concordance of IA, researchers have tive study, found a three- to fivefold increase in the incidence of identified proteins associated with the genetic disease as poten- IAs in first-degree relatives in comparison with the general tially related to the pathogenesis of aneurysms. Diseases such population (89). as ADPKD (Online Mendelian Inheritance in Man database A study spanning from 1970 to 1989 evaluated the families of [OMIM] #173900) (11), Marfan syndrome (OMIM #154700) (95), patients with aneurysmal SAH, reporting that 15 out of 76 glucocorticoid remediable aldosteronism (OMIM #103900) (55),

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 215 NAHED ET AL.

SAH and SNPs in exon 22 and intron 5 of the ELN gene. In TABLE 1. Candidate genes suggested to play a role in intracranial addition, haplotypes of intron 5/exon 22 and intron 4/intron aneurysm pathogenesis 5/ exon 22 were also associated with SAH (85). Chromosomal Associated The Japanese and Dutch studies notwithstanding, the role of Candidate gene location disease elastin in IA is far from definitive. A German study separately Elastin (ELN) 7q11 Williams-Beuren examined eight SNPs within the ELN gene and found no signif- syndrome icant association between them and the incidence of IA among Elastase 2 (ELA) 19p13,3 120 affected Caucasian patients. In contrast to the results of the ␣1-antitrypsin (AAT) 14q32 Chronic obstructive groups of Onda et al. (71, 72) and Ruigrok et al. (85), this study pulmonary disease failed to identify a haplotype in the 7q11.2 locus that is associ- Collagen III (COL3A1) 2q31 Ehlers-Danlos syndrome ated with the aneurysm phenotype (49). Endothelial nitric oxide 7q35–36 Nevertheless, the Japanese and Dutch findings tentatively synthase (eNOS) support the body of literature implicating the ELN locus in Endoglin (ENG) 9q3 Hereditary hemorrhagic aneurysm genesis, more specifically the locus on chromosome telengiectasia 7q11.2. However, no frank mutation has been identified. Polycystin (PKD1) 16p13 Adult polycystic kidney disease Elastase and α1-antitrypsin Fibrillin (FBN1) 15q21 Marfan syndrome Transforming growth 9q; 3p; 1p Thoracic aortic Elastase is a proteolytic enzyme that degrades elastin, colla- factor-␤ receptors I, II, III aneurysm gen, and other proteins within the ECM (103) (62). Secreted by polymorphonuclear leukocytes, it is inactivated once it binds to α1-antitrypsin (AAT), forming a serum complex with the protease (49, 50). The balance between active and inactive elastase and Ehlers-Danlos syndrome Type IV (OMIM #130050) (16) has been implicated in aneurysmal formation. Tartara et al. (94) seem to increase the risk of IA formation. This phenomenon have suggested that AAT activity is decreased in the walls of prompted researchers to study candidate genes and gene prod- IAs (94). Accordingly, elevated levels of elastase are found at ucts responsible for the disease with the hopes of explaining the site of IAs (50). Although controversial, these complimen- the increased IA formation (Table 1). tary studies suggest that AAT levels are decreased at the site of Recent advances in genetic disorders and vascular abnor- aneurysms, leading to increases in elastase activity. malities have revealed several alterations in genes and gene Two studies contradict these findings demonstrating ele- products involved in the remodeling of the extracellular matrix vated elastase levels in healthy individuals without aneurysms (ECM). The dynamic nature of the ECM has been theorized to (4, 14). In addition, a Japanese case-control study evaluated go awry, leading to diminished support of the vasculature, ulti- 384 affected patients and 289 unaffected patients and found no mately resulting in an aneurysm. Supporting this notion, the significant difference in the AAT allelic frequencies of the two content and structure of collagen and elastin, the predominant groups (111). Debate over the importance of the ratio of AAT to elements in aneurysmal walls, is significantly altered. The elastase will require further work delineating aneurysmal ECM-related proteins discussed below have been identified in causes from those of normal physiology. genetic disorders and could be related to IA formation. Collagen (I and III) Elastin Studies have uniformly demonstrated a decrease and alter- In 2001, a genome-wide linkage study of 104 Japanese ation in the collagen within aneurysmal walls. In particular, a affected sibling pairs identified an area near the elastin gene deficiency in Type III collagen has been found within (ELN) as the best evidence of linkage (72). Linkage was found aneurysm walls (23, 24, 67, 73, 78, 100). Subsequent studies at a total of three sites: 5q22-q31 with a maximum logarithm of have proposed a specific defect in Type III collagen (COL3A1) the odds (LOD) score (MLS) of 2.24, 7q11 (MLS, 3.22), and (16, 77). Ehlers-Danlos Syndrome Type IV is a connective tis- 14q22 (MLS, 2.31). None of the fourteen single nucleotide poly- sue disorder caused by mutations in the COL3A1 gene on morphisms (SNPs) identified within ELN co-segregated with chromosome 2q31. Ehlers-Danlos Syndrome Type IV is char- the aneurysmal phenotype. However, the haplotype between acterized by vascular abnormalities consisting of increased intron-20 / intron-23 polymorphism of ELN was strongly asso- ruptures, thin transparent skin, and ligament weakness. ciated with intracranial aneurysms [P ϭ 3.81 ϫ 10(-6)]. Ostergaard and Oxlund (73) demonstrated that six of the 14 Furthermore, patients who were homozygous for the mutation patients who died from ruptured IAs had Type III collagen were at highest risk (P ϭ 0.002), with an odds ratio of 4.39. deficiency in the middle cerebral and brachial arteries post- Recently, a Dutch group seemed to confirm the linkage to the mortem (73). However, others have suggested that mutations elastin locus. The Dutch study evaluated patients with spo- in the COL3A1 gene are not a common cause of intracranial radic aneurysmal SAH and unaffected controls for polymor- aneurysms or of cervical artery dissections (51). Although col- phisms in the ELN gene. The authors checked for 18 exonic lagen constitutes the majority of the ECM, its role in aneurysm and intronic polymorphisms and found an association between formation remains to be proven.

216 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com PATHOGENESIS OF INTRACRANIAL ANEURYSMS

Endothelial Nitric Oxide Synthase mutations in both genes have been linked to intracranial aneurysms, work on PKD1 has revealed a specific mutation in Endothelial nitric oxide synthase (eNOS) has been implicated chromosome 16p13.3, exon 15 of PKD1 in two patients with as a gene of interest by a Mayo Clinic group that looked into aneurysms (104). Furthermore, the position (rather than the the presence of three eNOS mutations in 58 patients with rup- type of PKD1 mutation) influences a patient’s likelihood of tured IAs and 49 with unruptured IAs. The group found that developing an aneurysm. Rossetti et al. (84) demonstrated that varying alleles of the three eNOS polymorphisms studied were mutations in the 5؅ half of the gene were associated with poorer found two to four times more frequently among patients with prognoses (84). The prevalence of asymptomatic IAs in patients ruptured IAs than those with unruptured IAs (43). with ADPKD is five times that of the general population (10, Previously, the same group had conducted a study of eNOS 76). The average age of aneurysmal rupture is 41 years, a gene polymorphisms which found that one eNOS intron poly- decade earlier than sporadic cases (12, 75, 90). One could spec- morphism was significantly over-represented in patients with ulate that the PKD1 gene either encodes multiple proteins or ruptured aneurysms when compared with healthy volunteers. that mutations in the 5؅ end produce a dominant negative The authors also reported that among patients with aneurysms, effect. The relationship with PKD1 and IAs is rather complex those with a heterozygous eNOS T-786C SNP have larger rup- and could possibly be owing to one of the multiple splice vari- tured IAs than those with a homozygous genotype (44). ants encoded by the PKD1 gene. A subsequent larger Japanese study failed to confirm the link between eNOS mutation and the size of rupture. The study, Fibrillin which enrolled 336 Japanese and 191 Korean patients, found no Marfan syndrome, an autosomal dominant disorder of the significant differences in the size of aneurysms of heterozy- connective tissue, affects the cardiovascular, skeletal, and ocu- gous eNOS T-786C SNP patients as opposed to their homozy- lar systems. It is characterized by arachnodactyly, unusual gous counterparts. The study concludes that the eNOS T-786C height, pectus abnormalities, enlargement of the aorta, and pos- polymorphism does not influence the size of aneurysms in the sibly aortic aneurysms. The disease is caused by a mutation in particular ethnic groups that were investigated (1). the gene encoding fibrillin (FBN1) located on chromosome 15q21.1. There is debate over the association between FBN1 Endoglin mutations and intracranial aneurysms. A recent study of 25 Endoglin, a component of the transforming growth factor- autopsy cases with Marfan syndrome revealed no statistical β receptor complex is highly expressed on endothelial cell difference in the prevalence of IAs in patients with and without surfaces. One study demonstrated an association between Marfan syndrome (15). intracranial aneurysms and a six-base insertion polymor- Although there has been no association between FBN1 muta- phism in intron 7 of the endoglin gene in a Japanese populations and IA, this is not the case for thoracic aortic aneurysms. tion (93). This region codes for a component of the transform- Two studies have linked chromosomes 5q13-q14 and 11q23.2- ing growth factor-Β receptor complex and is also mutated in q24 with aortic and thoracic aneurysms, respectively (26, 102). hereditary hemorrhagic telengiectasia 1 (HHT1). However, The pathogenesis of aortic and thoracic aneurysms could prove two other studies, one on a Caucasian population (50) and to be similar to that of IAs and the identification of the genes another on a separate Japanese population (71) failed to repli- and gene products of one may divulge a wealth of knowledge cate this result. about the other. Our group and others have encountered quite The discrepancy in results from the above-mentioned studies a number of families with a clear genetic basis of IAs who are suggests that polymorphism sequences are likely influenced also plagued with abdominal aortic aneurysms (45, 64). by genetic and environmental factors which differ according to A group at Brigham and Women’s Hospital conducted the ethnicity. To prove this point, ethnicity-related differences are largest study to date of families afflicted with both cerebral reported for polymorphisms and are considered within the and aortic aneurysms (45). The study found that both cerebral norm. Whereas the studies of Krex et al. (49, 50) and Onda and aortic phenotypes can be inherited from a parent with an et al. (71, 72) refute the findings of Takenaka et al.’s (93) study, aortic aneurysm. The findings raise the possibility that a single it is possible that endoglin encodes multiple splice variants, gene defect may lead to the manifestation of either lesion. one of which could lead to aneurysm formation. The polymor- β phisms expressed in endoglin represent one of the many rea- Transforming Growth Factor- Receptors sons genetic testing is complex. Transforming growth factor (TGF)-β is a member of a group of cytokines that are collectively referred to as the TGF super- Polycystin family. This family is involved in a variety of cellular processes, ADPKD is characterized by renal cysts, renal failure and vas- including proliferation, extracellular matrix deposition, gene cular pathology. Disease results from a mutation in one of two expression, differentiation, and morphogenesis. Recently, Loeys genes, PKD1 and PKD2, which encode polycystin 1 and 2, et al. (57) described a new syndrome that includes widespread respectively. Polycystin participates in protein-protein perturbations in cardiovascular, craniofacial, neurocognitive, (multiprotein membrane-spanning complex) and protein- and skeletal development in 10 families with mutations in carbohydrate interactions in the extracellular matrix. Although genes encoding either Type I or II TGF-β receptors (TGF-β R)

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 217 NAHED ET AL.

(57). Affected members in all families had aortic root A B aneurysms and aneurysms in different locations of the arterial system as well as intracranial cerebral aneurysms. Based on this finding, we hypothesized that the TGF-β R family might also be mutated in familial forms of IA. To test this hypothesis, we screened the index cases in all of our families (n ϭ 33) with two or more members affected by IA to look for possible mutations associated with TGF-β R 1, 2, and 3. Of the 33 index cases we screened, we found several intronic and exonic polymorphisms. None of these, however, co-segregated with the disease trait. Furthermore, the same polymorphisms were observed in FIGURE 1. Representative screening control groups. Thus, we concluded that the TGF-β R family imaging tests. Magnetic resonance does not directly cause non-syndromic, familial IAs in our angiogram (A) and three-dimensional computed tomographic angiogram (B) patient database (Bademci et al., unpublished data). A sum- revealing an M2 to M3 junction aneurysm in addition to the superior cerebel- mary of candidate gene studies in familial IA can be found in lar artery aneurysm (arrows). Table 1. nation rather than an inherited variation in a single gene. Limitations of Candidate Gene Analysis However, this observation also makes it highly likely that one Discrepancies in the results of the aforementioned studies will be able to identify a sizable cohort of patients relatively reveal the limitations of candidate gene analyses. Although rapidly by ascertaining mutigenerational extended families several associations have been made between IAs and specific through affected index cases. Moreover, the availability of large polymorphisms in genes selected based on this method, none pedigrees still provides the opportunity to identify the unusual of these associations have been consistently reproducible. In a circumstance in which IAs are transmitted as a consequence of comprehensive review of this method, Hirschhorn et al. (28) a mutation in a single gene that imparts a major effect. compiled 166 positive association studies that were analyzed Consistent with these observations, since 1994, we have col- three or more times between SNPs and complex genetic traits. lected nearly 142 families, both large and small, to give us the Only six out of the 166 were robust and consistently replicated. power to positionally map the IA gene or genes. In the major- This alludes to the potential weakness in the candidate gene ity of these families, we were able to increase the number of approach, whereby the more genes that are analyzed using this affected patients by using imaging studies (Fig. 1) (Goksu et al., method, the higher the likelihood of false positive results. unpublished data). Whereas approaches driven by hypotheses regarding disease Large families with a high prevalence of IAs are likely owing mechanisms are intuitively attractive, experience has repeat- to a single gene responsible for the large effect. Therefore, edly shown that genome-wide linkage analysis is a more fruit- genome-wide linkage analysis of these families could allow ful method to isolate susceptibility genes of a complex disorder, one to find the chromosomal position of the disease gene by the pathophysiology and molecular biology of which are not using genetic markers. Confining analysis to a single or a few established. large families minimizes the chance of obscuring linkage owing to genetic heterogeneity in which the disease is caused by Genome-wide Linkage Analysis mutations in various genes in different families. This method Although there is no question that genetic factors contribute has been successful in identifying disease-causing genes in to the pathophysiology of IAs, only a few extended, multigen- complex disorders such as diabetes, obesity, and hypertension erational IA families have been reported. Other than the lethal- (13, 56, 108). Furthermore, this technique has also shed light on ity of the trait, a number of explanations may account for this the pathophysiology of cerebral cavernous malformations, observation. First, IAs may be attributable to the inheritance of another type of malformation involving the cerebral vascula- a single gene with major effect, but of incomplete penetrance, ture. Using the above mentioned techniques, three distinct in which case, only a fraction of recipients of the mutant gene genes have been cloned that, when mutated, lead to familial will develop the disease. Secondly, IAs may be caused by muta- forms of cerebral cavernous malformations. The cloning of tions in multiple genes of minor effect. Alternatively, these genes has not only resulted in rapid genetic tests that can aneurysms might form in genetically susceptible individuals as determine whether or not an individual is at risk for develop- a result of exposure to certain environmental factors, such as ing cerebral cavernous malformations but has also helped hypertension or smoking. researchers begin to understand the underlying pathology of The advent of imaging studies has revealed that IAs are this disease. much more common in the general population than had been Although genome-wide scanning followed by mutational previously recognized, with estimates of the population preva- analysis has produced noteworthy results, it remains unclear lence ranging up to 10% (54). With a high prevalence such as whether or not loci or genes identified in such families will this, it becomes increasingly likely that, as a general rule, a prove to play a role in the general population or if they will be majority of the trait will prove to be of multifactorial determi- of significance only in rare families. Once IA susceptibility

218 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com PATHOGENESIS OF INTRACRANIAL ANEURYSMS

TABLE 2. Genetic linkage studies published in the literature identifying suggestive intracranial aneurysm locia Chr band Mb Marker Max LOD P value Series (ref. no.) 1p36.1–34.3 19.6–35.1 D1S199-D1S496b 4.2 Nahed et al., 2005 (64) 1p33 46.6 D1S2797 0.037 Onda et al., 2001 (72) 1p22.1 93 D1S2868 0.022 Onda et al., 2001 (72) 2p15–14 62.3–66.6 D2S2206-D2S166 3.55 Roos et al., 2004 (83) 2q13 112.7 D2S160 0.044 Onda et al., 2001 (72) 3q29 198.5 D3S1311 0.014 Onda et al., 2001 (72) 4p16.1–15.3 pter-6.7 D4S2366—D4S403 1.27 Olson et al., 2002 (70) 4q34.1 174.8 D4S2991 0.04 Onda et al., 2001 (72) 5q14.3 85.4 D5S428 0.001 Onda et al., 2001 (72) 5q15 95.8 D5S644 0.018 Onda et al., 2001 (72) 5q22–31 118.9–140.9 D5S471-D5S2010b 2.24 Onda et al., 2001 (72) 6q14.1 77.6 D6S1031 1.20 Olson et al., 2002 (70) 7p22.2 4.2 D7S1819 1.67 Olson et al., 2002 (70) 7p14.1 38.96 D7S510 0.032 Onda et al., 2001 (72) 7q11 71.1–92.3 D7S2415-D7S657b 3.22 Onda et al., 2001 (72) 7q21.1 77.5 D7S669 0.021 Onda et al., 2001 (72) 7q21.1 88.1 D7S630 0.04 Onda et al., 2001 (72) 7q22.1 101.3 D7S515 0.014 Onda et al., 2001 (72) 7q34 139.4 D7S1824 0.90 Olson et al., 2002 (70) 9p24.2 3.9 D9S288 0.011 Onda et al., 2001 (72) 11q13.2 67.6 D11S987 0.042 Onda et al., 2001 (72) 11q25 131.2 D11S910 0.023 Onda et al., 2001 (72) 11q25 125–131 D11S4151 4.30 Ozturk et al., 2006 (74) 13q14.2 47.8 D13S153 0.034 Onda et al., 2001 (72) 14q22 69.6–77.6 D14S258-D14S74b 2.31 Onda et al., 2001 (72) 14q22 63.6–83.7 D14S74 3.00 Ozturk et al., 2006 (74) 14q24.3 77.7 D14S74 0.003 Onda et al., 2001 (72) 14q24.2 69.6 D14S258 0.034 Onda et al., 2001 (72) 14q34.2 99.7 D14S1426 1.36 Olson et al., 2002 (70) 17p12-q11.2 14.2–27.1 D17S921-D17S1800c 3.00 Yamada et al., 2004 (109) 17p11.2 24.3 D17S925 0.027 Onda et al., 2001 (72) 18p11.4 0.6 D18S59 0.049 Onda et al., 2001 (72) 19q13.1–13.3 38.7–55.6 D19S245—D19S246 2.58 Olson et al., 2002 (70) 19q13.2–13.4 39.8–57.4 D19S587-D19S601 van der Voet et al., 2004 (101) 19q13.2–13.3 46.7–54 D19S198-D19S596c 2.15 Yamada et al., 2004 (109) Xp22.2 14.4 DXS987 2.08 Olson et al., 2002 (70) Xp22.2–22.1 14.4–21.8 DXS987-DXS7593c 2.16 Yamada et al., 2004 (109)

a Chr, chromosome; Mb, megabases; Max, maximum; LOD, logarithm of the odds. b LOD-1 interval. c P Ͻ 0.05. genes are identified and cloned, it will be interesting to see if Genetic Locus for IAs: Nonparametric Linkage these genes are also mutated in sporadic patients. The exis- To date, a number of studies have used parametric and non- tence of polymorphisms in the same genes in sporadic patients, parametric linkage approaches in an effort to identify loci con- as well as familial IA, will be strong support in favor of the tributing to IA formation and possible rupture (16, 19, 51, 70, hypothesis that the gene product has an important in role in the 72, 78, 93, 112). Overall, non-parametric linkage analyses have integrity of the cerebral vasculature. Further complicating mat- identified multiple loci that may be contributing to IAs on sev- ters, the ability to collect large extended families may prove to eral chromosomes, including 5q22 to 31 (72, 112), 7q11 (19, 72, be exceptionally difficult owing to the lethality of the trait, such 112), 17cen (109), 19q (70, 101, 109), and Xp (109) (Table 2). The that samples from known affected subjects are unavailable. strongest evidence to date implicates regions on chromosomes Such analysis is confounded by the uncertain genotype of unaf- 7q and 19q, both of which have been suggested to contain an fected subjects. IA locus in independent studies.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 219 NAHED ET AL.

In 2001, a genome-wide linkage study of 104 Japanese affected sibling pairs using 404 polymorphic markers throughout the human genome identified a total of 14 loci with statistically significant linkage to the IA phenotype (72) (Table 2). Among other loci, the study reported suggestive linkage on chromosomes 5q22 to 31 (MLS 2.24), 7q11 (MLS 3.22), 11q and 14q22 (MLS 2.31). Our recent analysis of two extended IA kindreds confirmed linkage in two of these loci on 11q and 14q (74). In addition, Farnham et al. (19) confirmed a third locus on 7q11 based on a study of 13 extended pedigrees from Utah, consisting of 39 IA cases. A second group, however, failed to replicate linkage to the 7q locus (29, 110). The group analyzed a total of 14 Japanese families with 64 members for linkage to the 7q11 locus, obtaining total LOD and total non-parametric LOD scores of –8.04 and –0.643, respectively. The finding suggests that this locus, which contains the ELN gene, is not responsible for IAs in a majority of aggregated IA Japanese families. A similar study of 48 affected relative pairs in the Finnish population identified an IA susceptibility locus on chromosome 19q (70). Seven regions, on chromosomes 4, 6, 7, 14, 19 and X were identified as potential IA susceptibility loci (Table 2). A follow-up study by the same group using 222 affected relative pairs confirmed the major FIGURE 2. Family pedigrees. Unaffected and affected members are shown as white and black symbols, respectively. locus on chromosome 19q13 Subjects with unknown phenotype status are shown as gray symbols, whereas patients harboring aortic aneurysms are (101). Multipoint linkage designated with an a. Genotyping of kindred members are listed under the subjects with segments linking to disease phe- analysis suggested linkage of notype enclosed in a box. A, Kindred IA 20. Genotyping of all kindred members with seven short tandem repeat mark- IA to a 11.6 cM region on 19q ers that span 1p34 to 36. B, Kindred IA 100. All family members were genotyped with six short tandem repeat markers centered 2 cM proximal to that span 11q24 to 25. C, Kindred IA 101. Genotyping of all family members with seven short tandem repeat markers D19S246 (101). This locus was that span 14q23 to 31. Chr, chromosome.

220 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com PATHOGENESIS OF INTRACRANIAL ANEURYSMS

later confirmed in a Japanese study that used 29 families with three or more affected individuals (109). This study identified two other IA loci on chromosomes 17cen and Xp22 (109) (Table 2). The authors of this study proceeded to test candidate genes in the relevant regions but found no polymorphisms that segregate with the IA phenotype (109). Genetic Locus for Intracranial Aneurysms: Parametric Linkage Although non-parametric approaches are attractive with respect to IAs as they are robust in the face of mis-specification of inheritance and do not rely on recruitment of multigenera- tional families, an alternative strategy to gene discovery in complex genetic disorders, including IA, involves the use of traditional parametric linkage analysis in unusual families. By confining analyses to a single or a few large families that seem to demonstrate simple Mendelian inheritance, one minimizes the chance of obscuring linkage owing to genetic heterogeneity or environmental factors. Although there have been fewer studies using this approach, the preliminary results have been quite promising: A recent report by Roos et al. (83) identified an IA locus on 2p13 by studying a consanguineous Dutch family of 20 siblings (seven of whom are affected with IA) with a maximum LOD score of 3.55, linking IA to a 7-cM region containing 150 genes. Our group has used genome-wide linkage analysis on the largest IA family (IA 20) reported to date. In total, the family includes 11 affected individuals, 10 with documented IAs and one with distinctive multiple intracranial vessel occlusions and extensive collateral vessel formation of unknown etiology (Fig. 2A). Genetic analysis identified significant linkage to the locus 1p35 to 36 with a maximum LOD score of 4.2 (Fig. 3A) (64). Examination of the locus identified approximately 240 genes. Among these, a number of genes have been identified as plausible candidate genes including polycystic kidney disease like-1 gene, brain specific angiogenesis inhibitor 2, fibronectin Type III domain containing gene, and collagen Type XVI α1 gene. We sequenced the coding region of these and 12 additional genes (total, 16) in two affected individuals. Although this analysis revealed many coding region polymorphisms, none were found to be segregating on the affected chromosome. We are currently continuing sequencing of additional genes located within the interval. Our group has also recently completed analysis on two large families that yielded significant linkage to chromosomes 11 and 14 (74). The first family includes four living members affected with IA and two members affected with abdominal aortic aneurysms (Fig. 2B); this latter trait is sometimes associated FIGURE 3. Genome-wide linkage analysis of kindreds IA 20, IA 100, and IA with IA (9, 45). Linkage analysis demonstrated linkage to the 101. The y-axis corresponds to the LOD score and the x-axis shows the genetic locus on chromosome 11q24 to 25. Genotyping of all unaffected distance (in centimorgans [cM]). A, analysis of linkage with STR markers on 1p34 to 36 localizes an IA gene to a 12.5 cM region between markers D1S199 individuals revealed a maximum LOD score of 4.3 on chromo- and UT5144 with a maximum LOD score of 4.2. B, analysis of all family some 11q if individuals with abdominal aortic aneurysms are members in IA 100 reveals a maximum LOD score of 4.3 at 11q24 to 25, spec- considered to be affected (Fig. 3B). Assigning an affection sta- ifying 99% penetrance where members II-2 and III-4, known to harbor abdom- tus unknown phenotype to these two individuals resulted in an inal aortic aneurysms, are designated as affected. C, analysis of four affected LOD score of 3.6 without any effect on the chromosomal local- and seven unaffected family members in IA 101 gives a maximum LOD score ization. The LOD-1 interval for the identified locus lies between of 3.0 at 14q23 to 31. Chr, chromosome; ICA, internal carotid artery.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 221 NAHED ET AL.

125.6 and 131.4 million base pairs (mbp) on chromosome 11. REFERENCES Interestingly, this is one of the 14 regions linking to IA in a Japanese sibling pair study (72). 1. Akagawa H, Kasuya H, Onda H, Yoneyama T, Sasahara A, Kim CJ, Lee JC, Yang TK, Hori T, Inoue I: Influence of endothelial nitric oxide synthase T- The second family includes a total of nine members with 786C single nucleotide polymorphism on aneurysm size. J Neurosurg documented IAs, four of whom are deceased and one of 102:68–71, 2005. whom refused to provide a blood sample (Fig. 2C). Linkage 2. Alberts MJ, Quinones A, Graffagnino C, Friedman A, Roses AD: Risk of analysis revealed linkage to locus 14q23 to 31 with an MLS of intracranial aneurysms in families with subarachnoid hemorrhage. Can J 3.0 (Fig. 3C). The LOD-1 interval falls between 75.5 and 85.6 Neurol Sci 22:121–125, 1995. 3. Andrews RJ, Spiegel PK: Intracranial aneurysms. Age, sex, blood pres- mbp. This interval also lies within a locus found to have statis- sure, and multiplicity in an unselected series of patients. J Neurosurg tically significant linkage with the IA phenotype in the 51:27–32, 1979. Japanese study, which suggested an IA susceptibility locus 4. Baker CJ, Fiore A, Connolly ES Jr, Baker KZ, Solomon RA: Serum elastase between 63.6 to 87.6 mbp on chromosome 14q (72). The fact and alpha-1-antitrypsin levels in patients with ruptured and unruptured that both loci identified by our group on chromosomes 11q cerebral aneurysms. Neurosurgery 37:56–62, 1995. 5. Beumont PJ: The familial occurrence of berry aneurysm. J Neurol Neurosurg and 14q, were suggested to contain IA susceptibility genes in Psychiatry 31:399–402, 1968. the aforementioned study of Japanese sibling pairs strongly 6. Bonita R, Thomson S: Subarachnoid hemorrhage: epidemiology, diagnosis, suggests that IA susceptibility genes lie within these regions. management, and outcome. Stroke 16:591–594, 1985. Whereas demonstrating the genetic heterogeneity of IA, these 7. Brisman R, Abbassioun K: Familial intracranial aneurysms. J Neurosurg results are also an important step toward cloning IA genes 34:678–682, 1971. 8. Bromberg JE, Rinkel GJ, Algra A, van Duyn CM, Greebe P, Ramos LM, van and ultimately understanding its pathophysiology. The above Gijn J: Familial subarachnoid hemorrhage: distinctive features and patterns studies prove the genetic heterogeneity of IA and show that of inheritance. Ann Neurol 38:929–934, 1995. different loci located throughout the human genome con- 9. Cannon Albright LA, Camp NJ, Farnham JM, MacDonald J, Abtin K, Rowe tribute to IA pathogenesis (Table 2). KG: A genealogical assessment of heritable predisposition to aneurysms. J Neurosurg 99:637–643, 2003. 10. Chapman AB, Johnson AM, Gabow PA: Intracranial aneurysms in patients with autosomal dominant polycystic kidney disease: how to diagnose and CONCLUSION who to screen. Am J Kidney Dis 22:526–531, 1993. 11. Chapman AB, Rubinstein D, Hughes R, Stears JC, Earnest MP, Johnson AM, Although the etiology of intracranial aneurysm remains Gabow PA, Kaehny WD: Intracranial aneurysms in autosomal dominant poorly understood, there is significant evidence associating polycystic kidney disease. N Engl J Med 327:916–920, 1992. environmental and genetic factors with its pathogenesis. 12. Chauveau D, Pirson Y, Verellen-Dumoulin C, Macnicol A, Gonzalo A, Grunfeld JP: Intracranial aneurysms in autosomal dominant polycystic kid- Aneurysm size has long been considered the main risk factor ney disease. Kidney Int 45:1140–1146, 1994. for rupture. However, other factors such as sex, hypertension, 13. Concannon P, Erlich HA, Julier C, Morahan G, Nerup J, Pociot F, Todd JA, aneurysm location, and relatively young age also contribute Rich SS; Type 1 Diabetes Genetics Consortium: Type 1 diabetes: evidence for to its pathogenesis. In addition, inheritance of genetic sus- susceptibility loci from four genome-wide linkage scans in 1,435 multiplex ceptibility loci plays an important role. Based on genome- families. Diabetes 54:2995–3001, 2005. 14. Connolly ES Jr, Fiore AJ, Winfree CJ, Prestigiacoma CJ, Goldman JE, wide linkage analysis, there is an IA susceptibility gene on Solomon RA: Elastin degradation in the superficial temporal arteries of chromosomes 1p34.3 to 36.13, 2p13, 7q22.1, 11q25, 14q22, and patients with intracranial aneurysms reflects changes in plasma elastase. 19q13.3. These regions have been identified not only from rel- Neurosurgery 40:903–909, 1997. ative pair studies, but also from the study of large families 15. Conway JE, Hutchins GM, Tamargo RJ: Marfan syndrome is not associated with likely significant genetic contribution in the absence of with intracranial aneurysms. Stroke 30:1632–1636, 1999. 16. de Paepe A, van Landegem W, de Keyser F, de Reuck J: Association of mul- known environmental factors. This combination of genetic tiple intracranial aneurysms and collagen type III deficiency. Clin Neurol and environmental factors ultimately plays a role in the com- Neurosurg 90:53–56, 1988. plex pathway to IA. 17. Deleted in proof. The identification of susceptibility genes in aneurysm 18. Edelsohn L, Caplan L, Rosenbaum AE: Familial aneurysms and infundibu- Neurology development will shed light on the pathways leading to this lar widening. 22:1056–1060, 1972. 19. Farnham JM, Camp NJ, Neuhausen SL, Tsuruda J, Parker D, MacDonald J, disease. It is likely that both locus and allelic heterogeneity, as Cannon-Albright LA: Confirmation of Chromosome 7q11 Locus for well as environmental factors, play a role in IA pathogenesis, Predisposition to Intracranial Aneurysm. Hum Genet 114:250–255, 2004. complicating efforts at disease gene identification. Certainly, 20. Forget TR Jr, Benitez R, Veznedaroglu E, Sharan A, Mitchell W, Silva M, the identification of new genes important in IA pathogenesis Rosenwasser RH: A review of size and location of ruptured intracranial aneurysms. Neurosurgery 49:1322–1326, 2001. will provide insight into the primary determinants of this dis- 21. Fox JL, Ko JP: Familial intracranial aneurysms. Six cases among 13 siblings. ease and will result in new opportunities for early diagnosis J Neurosurg 52:501–503, 1980. in the preclinical setting. This will also assist in the under- 22. Frosen J, Piippo A, Paetau A, Kangasniemi M, Niemela M, Hernesniemi J, standing of disease pathogenesis while allowing clinicians Jääskeläinen J: Remodeling of saccular cerebral artery aneurysm wall is asso- the opportunity to modify treatment based on risk. ciated with rupture: Histological analysis of 24 unruptured and 42 ruptured cases. Stroke 35:2287–2293, 2004. Ultimately, it is anticipated that novel therapeutic strategies 23. Gaetani P, Tartara F, Grazioli V, Tancioni F, Infuso L, Rodriguez y Baena R: based on biology will be developed, which will target these Collagen cross-linkage, elastolytic and collagenolytic activities in cerebral newly elucidated genetic susceptibilities. aneurysms: a preliminary investigation. Life Sci 63:285–292, 1998.

222 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com PATHOGENESIS OF INTRACRANIAL ANEURYSMS

24. Gaetani P, Tartara F, Tancioni F, Rodriguez y Baena R, Casari E, Alfano M, 48. Kondo R, Matsumori Y, Kato N, Tsuchiya D, Kokubo Y, Sato S, Kurita K, Grazioli V: Deficiency of total collagen content and of deoxypyridinoline in Kato T, Kubota I, Kayama T: Trends in the incidence of stroke in Yamagata, intracranial aneurysm walls. FEBS Lett 404:303–306, 1997. Japan [in Japanese]. No To Shinkei 58:213–218, 2006. 25. Graf CJ: Familial intracranial aneurysms. J Neurosurg 25:304–308, 1966. 49. Krex D, Konig IR, Ziegler A, Schackert HK, Schackert G: Extended single 26. Guo D, Hasham S, Kuang SQ, Vaughan CJ, Boerwinkle E, Chen H, Abuelo nucleotide polymorphism and haplotype analysis of the elastin gene in D, Dietz HC, Basson CT, Shete SS, Milewicz DM: Familial thoracic aortic Caucasians with intracranial aneurysms provides evidence for racially/eth- aneurysms and dissections: genetic heterogeneity with a major locus map- nically based differences. Cerebrovasc Dis 18:104–110, 2004. ping to 5q13-14. Circulation 103:2461–2468, 2001. 50. Krex D, Ziegler A, Schackert HK, Schackert G: Lack of association between 27. Hackett ML, Anderson CS, The Australian Cooperative Research on endoglin intron 7 insertion polymorphism and intracranial aneurysms in a Subarachnoid Hemorrhage Study Group: Health outcome 1 year after sub- white population: Evidence of racial/ethnic differences. Stroke 32:2689– arachnoid haemorrhage: An international population-based study. 2694, 2001. Neurology 55:658–662, 2000. 51. Kuivaniemi H, Prockop DJ, Wu Y, Madhatheri SL, Kleinert C, Earley JJ, 28. Hirschhorn JN, Lohmueller K, Byrne E, Hirschhorn K: A comprehensive Jokinen A, Stolle C, Majamaa K, Myllyla VV, Norregård Ö, Schievink WI, review of genetic association studies. Genet Med 4:45–61, 2002. Mokri B, Fukawa O, ter Berg JWM, De Paepe A, Lozano AM, Leblanc R, 29. Hofer A, Ozkan S, Hermans M, Kubassek N, Sitzer M, Burtscher J, Knopp U, Ryynänen M, Baxter BT, Shikata H, Ferrell RE, Tromp G: Exclusion of muta- Schoch B, Wanke I, Huebner F, Raabe A, Steinmetz H, Auburger G: tions in the gene for type III collagen (COL3A1) as a common cause of Mutations in the lysyl oxidase gene not associated with intracranial intracranial aneurysms or cervical artery dissections: results from sequence aneurysm in central European families. Cerebrovasc Dis 18:189–193, 2004. analysis of the coding sequences of type III collagen from 55 unrelated 30. Hop JW, Rinkel GJ, Algra A, van Gijn J: Case-fatality rates and functional patients. Neurology 43:2652–2658, 1993. outcome after subarachnoid hemorrhage: a systematic review. Stroke 52. Lackner E, Jobges M, Schirmer F, Hummelsheim H: Intracranial aneurysms 28:660–664, 1997. and autosomal dominant polycystic kidney disease [in German]. Fortschr Neurol Psychiatr 70:438–442, 2002. 31. Huang J, van Gelder JM: The probability of sudden death from rupture of 53. Leblanc R, Melanson D, Tampieri D, Guttmann RD: Familial cerebral intracranial aneurysms: a meta-analysis. Neurosurgery 51:1101–1107, 2002. aneurysms: A study of 13 families. Neurosurgery 37:633–639, 1995. 32. Johnston SD, Robinson TJ: Subarachnoid haemorrhage: Difficulties in diag- 54. Linn FH, Rinkel GJ, Algra A, van Gijn J: Incidence of subarachnoid hemor- nosis and treatment. Postgrad Med J 74:743–744, 1998. rhage: role of region, year, and rate of computed tomography: a meta- 33. Juvela S: Risk factors for multiple intracranial aneurysms. Stroke analysis. Stroke 27:625–629, 1996. 31:392–397, 2000. 55. Litchfield WR, Anderson BF, Weiss RJ, Lifton RP, Dluhy RG: Intracranial 34. Juvela S: Recommendations for the management of patients with unrup- aneurysm and hemorrhagic stroke in glucocorticoid-remediable aldostero- tured intracranial aneurysms. Stroke 32:815–816, 2001. nism. Hypertension 31:445–450, 1998. 35. Juvela S: Natural history of unruptured intracranial aneurysms: risks for 56. Liu YJ, Xiao P, Xiong DH, Recker RR, Deng HW: Searching for obesity genes: aneurysm formation, growth, and rupture. Acta Neurochir Suppl Progress and prospects. Drugs Today (Barc) 41:345–362, 2005. 82:27–30, 2002. 57. Loeys BL, Chen J, Neptune ER, Judge DP, Podowski M, Holm T, Meyers J, 36. Juvela S: Prehemorrhage risk factors for fatal intracranial aneurysm rup- Leitch CC, Katsanis N, Sharifi N, Xu FL, Myers LA, Spevak PJ, Cameron DE, ture. Stroke 34:1852–1857, 2003. De Backer J, Hellemans J, Chen Y, Davis EC, Webb CL, Kress W, Coucke P, 37. Juvela S, Hernesniemi JA: Clinical manifestations and survival rates among Rifkin DB, De Paepe AM, Dietz HC: A syndrome of altered cardiovascular, patients with saccular intracranial aneurysms: population-based study in craniofacial, neurocognitive and skeletal development caused by mutations Olmsted County, Minnesota, 1965 to 1995. Neurosurgery 50:1167–1169, 2002. in TGFBR1 or TGFBR2. Nat Genet 37:275–281, 2005. 38. Juvela S, Porras M, Heiskanen O: Natural history of unruptured intracranial 58. Lozano AM, Leblanc R: Familial intracranial aneurysms. J Neurosurg aneurysms: a long-term follow-up study. J Neurosurg 79:174-182, 1993. 66:522–528, 1987. 39. Juvela S, Porras M, Poussa K: Natural history of unruptured intracranial 59. Mathieu J, Hebert G, Perusse L, Prevost C, Cantin L, Bouchard JM, aneurysms: probability of and risk factors for aneurysm rupture. J Neuro- DeBraekeleer M: Familial intracranial aneurysms: recurrence risk and acci- surg 93:379-387, 2000. dental aggregation study. Can J Neurol Sci 24:326–331, 1997. 40. Juvela S, Siironen J, Kuhmonen J: Hyperglycemia, excess weight, and history 60. Matsumoto H, Terada T, Tsuura M, Itakura T, Ogawa A: Basic fibroblast of hypertension as risk factors for poor outcome and cerebral infarction after growth factor released from a platinum coil with a polyvinyl alcohol core aneurysmal subarachnoid hemorrhage. J Neurosurg 102:998–1003, 2005. enhances cellular proliferation and vascular wall thickness: an in vitro and 41. Kassell NF, Torner JC, Jane JA, Haley EC Jr, Adams HP: The International in vivo study. Neurosurgery 53:402–408, 2003. Cooperative Study on the Timing of Aneurysm Surgery. Part 2: Surgical 61. Mayberg MR, Batjer HH, Dacey RG Jr, Diringer M, Haley EC, Heros RC, results. J Neurosurg 73:37–47, 1990. Sternau LL, Torner J, Adams HP Jr, Feinberg W, Thies W, Ex Officio 42. Kasuya H, Onda H, Takeshita M, Hori T, Takakura K: Clinical features of Members: Guidelines for the management of aneurysmal subarachnoid intracranial aneurysms in siblings. Neurosurgery 46:1301–1306, 2000. hemorrhage. A statement for healthcare professionals from a special writing 43. Khurana VG, Meissner I, Sohni YR, Bamlet WR, McClelland RL, group of the Stroke Council, American Heart Association. Stroke Cunningham JM, Meyer FB: The presence of tandem endothelial nitric oxide 25:2315–2328, 1994. synthase gene polymorphisms identifying brain aneurysms more prone to 62. McDonald JA, Kelley DG: Degradation of fibronectin by human leukocyte rupture. J Neurosurg 102:526–531, 2005. elastase. Release of biologically active fragments. J Biol Chem 255:8848– 44. Khurana VG, Sohni YR, Mangrum WI, McClelland RL, O’Kane DJ, Meyer 8858, 1980. FB, Meissner I: Endothelial nitric oxide synthase T-786C single nucleotide 63. Mitchell P, Gholkar A, Vindlacheruvu RR, Mendelow AD: Unruptured polymorphism: A putative genetic marker differentiating small versus large intracranial aneurysms: benign curiosity or ticking bomb? Lancet Neurol ruptured intracranial aneurysms. Stroke 34:2555–2559, 2003. 3:85–92, 2004. 45. Kim LG, Scott RA, Thompson SG, Collin J, Morris GE, Sutton GL, Wilson 64. Nahed BV, Seker A, Guclu B, Ozturk AK, Finberg K, Hawkins AA, DiLuna NM; Multicentre Aneurym Screening Study Group: Implications of screen- ML, State M, Lifton RP, Gunel M: Mapping a Mendelian form of intracranial ing for abdominal aortic aneurysms on surgical workload. Br J Surg aneurysm to 1p34.3-p36.13. Am J Hum Genet 76:172–179, 2005. 92:171–176, 2005. 65. Nakagawa T, Hashi K, Kurokawa Y, Yamamura A: Family history of sub- 46. Knekt P, Reunanen A, Aho K, Heliovaara M, Rissanen A, Aromaa A, arachnoid hemorrhage and the incidence of asymptomatic, unruptured cere- Impivaara O: Risk factors for subarachnoid hemorrhage in a longitudinal bral aneurysms. J Neurosurg 91:391–395, 1999. population study. J Clin Epidemiol 44:933–939, 1991. 66. Nanda A, Vannemreddy PS, Polin RS, Willis BK: Intracranial aneurysms 47. Kojima M, Nagasawa S, Lee YE, Takeichi Y, Tsuda E, Mabuchi N: and cocaine abuse: analysis of prognostic indicators. Neurosurgery Asymptomatic familial cerebral aneurysms. Neurosurgery 43:776–781, 1998. 46:1063–1069, 2000.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 223 NAHED ET AL.

67. Neil-Dwyer G, Bartlett JR, Nicholls AC, Narcisi P, Pope FM: Collagen defi- 90. Schievink WI, Torres VE, Piepgras DG, Wiebers DO: Saccular intracranial ciency and ruptured cerebral aneurysms. A clinical and biochemical study. aneurysms in autosomal dominant polycystic kidney disease. J Am Soc J Neurosurg 59:16–20, 1983. Nephrol 3:88–95, 1992. 68. Ohashi Y, Horikoshi T, Sugita M, Yagishita T, Nukui H: Size of cerebral 91. Schievink WI, Wijdicks EF, Parisi JE, Piepgras DG, Whisnant JP: Sudden aneurysms and related factors in patients with subarachnoid hemorrhage. death from aneurysmal subarachnoid hemorrhage. Neurology Surg Neurol 61:239–245, 2004. 45:871–874, 1995. 69. Ohkuma H, Fujita S, Suzuki S: Incidence of aneurysmal subarachnoid hem- 92. Sudlow CL, Warlow CP: Comparable studies of the incidence of stroke and orrhage in Shimokita, Japan, from 1989 to 1998. Stroke 33:195–199, 2002. its pathological types: results from an international collaboration. 70. Olson JM, Vongpunsawad S, Kuivaniemi H, Ronkainen A, Hernesniemi J, International Stroke Incidence Collaboration. Stroke 28:491–499, 1997. Ryynanen M, Kim LL, Tromp G: Search for intracranial aneurysm suscepti- 93. Takenaka K, Sakai H, Yamakawa H, Yoshimura S, Kumagai M, Nakashima bility gene(s) using Finnish families. BMC Med Genet 3:7, 2002. S, Nozawa Y, Sakai N: Polymorphism of the endoglin gene in patients with 71. Onda H, Kasuya H, Yoneyama T, Hori T, Nakajima T, Inoue I: Endoglin is intracranial saccular aneurysms. J Neurosurg 90:935–938, 1999. not a major susceptibility gene for intracranial aneurysm among Japanese. 94. Tartara F, Gaetani P, Tancioni F, Guagliano A, Klersy C, Forlino A, Marzatico Stroke 34:1640–1644, 2003. F, Rodriguez y Baena R: Alpha 1-antitrypsin activity in subarachnoid hem- 72. Onda H, Kasuya H, Yoneyama T, Takakura K, Hori T, Takeda J, Nakajima T, orrhage. Life Sci 59:15–20, 1996. Inoue I: Genomewide-linkage and haplotype-association studies map intracra- 95. ter Berg HW, Bijlsma JB, Veiga Pires JA, Ludwig JW, van der Heiden C, nial aneurysm to chromosome 7q11. Am J Hum Genet 69:804–819, 2001. Tulleken CA, Willemse J: Familial association of intracranial aneurysms and 73. Ostergaard JR, Oxlund H: Collagen type III deficiency in patients with rup- multiple congenital anomalies. Arch Neurol 43:30–33, 1986. ture of intracranial saccular aneurysms. J Neurosurg 67:690-696, 1987. 96. Teunissen LL, Rinkel GJ, Algra A, van Gijn J: Risk factors for subarachnoid 74. Ozturk AK, Nahed BV, Bydon M, Bilguvar K, Goksu E, Bademci G, Guclu B, hemorrhage: a systematic review. Stroke 27:544–549, 1996. Johnson MH, Amar A, Lifton RP, Gunel M: Molecular genetic analysis of two 97. Thierry A, Ballivet J, Dumas R, Binnert D, Nivelon A, Foissac JC, Faure R: large kindreds with intracranial aneurysms demonstrates linkage to 11q24- Familial cases of intracranial aneurysms. Review of the literature and case 25 and 14q23-31. Stroke 37:1021–1027, 2006. report [in French]. Neurochirurgie 18:267–276, 1972. 75. Pirson Y, Chauveau D: ADPKD-associated intracranial aneurysm: new 98. Toglia JU, Samii AR: Familial intracranial aneurysms. Dis Nerv Syst insights and unanswered questions. Contrib Nephrol 115:53–58, 1995. 33:611–613, 1972. 76. Pirson Y, Chauveau D, Torres V: Management of cerebral aneurysms in 99. Ullrich DP: Familial cerebral aneurysms including one extracranial internal autosomal dominant polycystic kidney disease. J Am Soc Nephrol carotid aneurysm. Neurology 10:288–294, 1960. 13:269–276, 2002. 100. van den Berg JS, Limburg M, Pals G, Arwert F, Westerveld A, Hennekam 77. Pope FM, Kendall BE, Slapak GI, Kapoor R, McDonald WI, Compston DA, RC, Albrecht KW: Some patients with intracranial aneurysms have a Mitchell R, Hope DT, Millar-Craig MW, Dean JC: Type III collagen mutations reduced type III/type I collagen ratio. A case-control study. Neurology cause fragile cerebral arteries. Br J Neurosurg 5:551–574, 1991. 49:1546–1551, 1997. 78. Pope FM, Limburg M, Schievink WI: Familial cerebral aneurysms and type 101. van der Voet M, Olson JM, Kuivaniemi H, Dudek DM, Skunca M, Ronkainen III collagen deficiency. J Neurosurg 72:156–158, 1990. A, Niemela M, Jääskeläinen J, Hernesniemi JA, Helin K, Leinonen E, Biswas 79. Qureshi AI, Suarez JI, Parekh PD, Sung G, Geocadin R, Bhardwaj A, M, Tromp G: Intracranial Aneurysms in Finnish Families: Confirmation of Tamargo RJ, Ulatowski JA: Risk factors for multiple intracranial aneurysms. Neurosurgery 43:22–27, 1998. Linkage and Refinement of the Interval to Chromosome 19q13.3. Am J Hum 80. Ronkainen A, Hernesniemi JA, Ryynanen M: Familial subarachnoid hemor- Genet 74:564–571, 2004. rhage in east Finland, 1977-1990. Neurosurgery 33:787–797, 1993. 102. Vaughan CJ, Casey M, He J, Veugelers M, Henderson K, Guo D, Campagna 81. Ronkainen A, Hernesniemi JA, Tromp G: Special features of familial R, Roman MJ, Milewicz DM, Devereux RB, Basson CT: Identification of a intracranial aneurysms: report of 215 familial aneurysms. Neurosurgery chromosome 11q23.2-q24 locus for familial aortic aneurysm disease, a genet- 37:43–47, 1995. ically heterogeneous disorder. Circulation 103:2469–2475, 2001. 82. Ronkainen A, Miettinen H, Karkola K, Papinaho S, Vanninen R, Puranen M, 103. Watanabe H, Hattori S, Katsuda S, Nakanishi I, Nagai Y: Human neutrophil Hernesniemi JA: Risk of harboring an unruptured intracranial aneurysm. elastase: degradation of basement membrane components and immunolocal- Stroke 29:359–362, 1998. ization in the tissue. J Biochem (Tokyo) 108:753–759, 1990. 83. Roos YB, Pals G, Struycken PM, Rinkel GJ, Limburg M, Pronk JC, van den 104. Watnick T, Phakdeekitcharoen B, Johnson A, Gandolph M, Wang M, Briefel Berg JS, Luijten JA, Pearson PL, Vermeulen M, Westerveld A: Genome-wide G, Klinger KW, Kimberling W, Gabow P, Germino GG: Mutation detection linkage in a large Dutch consanguineous family maps a locus for intracranial of PKD1 identifies a novel mutation common to three families with aneurysms to chromosome 2p13. Stroke 35:2276–2281, 2004. aneurysms and/or very-early-onset disease. Am J Hum Genet 65:1561–1571, 84. Rossetti S, Chauveau D, Kubly V, Slezak JM, Saggar-Malik AK, Pei Y, Ong 1999. AC, Stewart F, Watson ML, Bergstralh EJ, Winearls CG, Torres VE, Harris 105. Weir B, Disney L, Karrison T: Sizes of ruptured and unruptured aneurysms PC: Association of mutation position in polycystic kidney disease 1 in relation to their sites and the ages of patients. J Neurosurg 96:64–70, 2002. (PKD1) gene and development of a vascular phenotype. Lancet 361:2196– 106. Weir BK, Kongable GL, Kassell NF, Schultz JR, Truskowski LL, Sigrest A: 2201, 2003. Cigarette smoking as a cause of aneurysmal subarachnoid hemorrhage and 85. Ruigrok YM, Rinkel GJE, Algra A, Raaymakers TW, van Gijn J: Characteris- risk for vasospasm: A report of the Cooperative Aneurysm Study. J tics of intracranial aneurysms in patients with familial subarachnoid hemor- Neurosurg 89:405–411, 1998. rhage. Neurology 62:891–894, 2004. 107. Wiebers DO, Whisnant JP, Huston J 3rd, Meissner I, Brown RD Jr, Piepgras 86. Russell SM, Lin K, Hahn SA, Jafar JJ: Smaller cerebral aneurysms producing DG, Forbes GS, Thielen K, Nichols D, O’Fallon WM, Peacock J, Jaeger L, more extensive subarachnoid hemorrhage following rupture: A radiological Kassell NF, Kongable-Beckman GL, Torner JC; International Study of investigation and discussion of theoretical determinants. J Neurosurg Unruptured Intracranial Aneurysms Investigators: Unruptured intracranial 99:248–253, 2003. aneurysms: natural history, clinical outcome, and risks of surgical and 87. Sarti C, Tuomilehto J, Salomaa V, Sivenius J, Kaarsalo E, Narva EV, Salmi K, endovascular treatment. Lancet 362:103–110, 2003. Torppa J: Epidemiology of subarachnoid hemorrhage in Finland from 1983 108. Wilson FH, Disse-Nicodeme S, Choate KA, Ishikawa K, Nelson-Williams C, to 1985. Stroke 22:848–853, 1991. Desitter I, Gunel M, Milford DV, Lipkin GW, Achard JM, Feely MP, Dussol 88. Schievink WI, Schaid DJ, Michels VV, Piepgras DG: Familial aneurysmal B, Berland Y, Unwin RJ, Mayan H, Simon DB, Farfel Z, Jeunemaitre X, Lifton subarachnoid hemorrhage: a community-based study. J Neurosurg 83:426– RP: Human hypertension caused by mutations in WNK kinases. Science 429, 1995. 293:1107–1112, 2001. 89. Schievink WI, Schaid DJ, Rogers HM, Piepgras DG, Michels VV: On the 109. Yamada S, Utsunomiya M, Inoue K, Nozaki K, Inoue S, Takenaka K, inheritance of intracranial aneurysms. Stroke 25:2028–2037, 1994. Hashimoto N, Koizumi A: Genome-wide scan for Japanese familial intracra-

224 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com PATHOGENESIS OF INTRACRANIAL ANEURYSMS

nial aneurysms: Linkage to several chromosomal regions. Circulation The authors do not comment much on the risk of hemorrhage ver- 110:3727–3733, 2004. sus the risk of aneurysm formation or on potential genes that modulate 110. Yamada S, Utsunomiya M, Inoue K, Nozaki K, Miyamoto S, Hashimoto N, risks of bleeding from aneurysms of similar size and location. In unrup- Takenaka K, Yoshinaga T, Koizumi A: Absence of linkage of familial intracra- tured aneurysm natural history studies, there is definitely an increased nial aneurysms to 7q11 in highly aggregated Japanese families. Stroke risk of bleeding in cohorts who have already bled from another 34:892–900, 2003. aneurysm. Even in cases with well defined genetic risk such as autoso- 111. Yoneyama T, Kasuya H, Onda H, Akagawa H, Hashiguchi K, Nakajima T, Hori T, Inoue I: Collagen type I alpha2 (COL1A2) is the susceptible gene for mal dominant polycystic kidney disease, there is familial clustering of intracranial aneurysms. Stroke 35:443–448, 2004. hemorrhage in some kindreds but not others. Hence, a biological pre- 112. Yoneyama T, Kasuya H, Onda H, Akagawa H, Jinnai N, Nakajima T, Hori T, disposition to hemorrhage is inherent to some, but not all, patients Inoue I: Association of positional and functional candidate genes FGF1, with IAs, including familial cases. FBN2, and LOX on 5q31 with intracranial aneurysm. J Hum Genet Not all biological predispositions to aneurysm formation and hem- 48:309–314, 2003. orrhage are necessarily genetic. Such a complex disease may also be affected by hormonal or inflammatory mechanisms, converging and Acknowledgement contributing to the final phenotype of vascular wall failure. Beyond the Brian Nahed, M.D., and Mohamad Bydon, B.A., contributed equally to the cre- genetic risks of lesion genesis, molecular studies of the aneurysm wall ation of this manuscript. Brian Nahed, M.D., is currently in the Department of and potential molecular imaging are likely to identify mechanistic sig- Neurosurgery at Massachusetts General Hospital, Boston, MA. natures differentiating aneurysms destined to grow and rupture from more static lesions. Genetics of aneurysms will not solve all mysteries of this disease or cure aneurysms directly, but it is surely helping us COMMENTS improve diagnosis and management. his is an excellent and timely review by a group that has led mod- Issam A. Awad Tern studies on the genetics of neurovascular disease. Numerous Evanston, Illinois studies have appeared in the recent literature but had not been summarized in a coherent fashion for the neurosurgical readership. The his article reviewing the genetics of IAs demonstrates the complex- authors are congratulated on doing a wonderful job synthesizing a Tity of the topic. The list of candidate genes is long, the methodology very complex and often confusing topic. yields results that are often conflicting, and the work required to under- The frustration in studying the genetics of intracranial aneurysms stand aneurysm pathogenesis is daunting. The authors deserve recog- (IAs) relates in part to their wide genetic heterogeneity, whereby more nition for their many contributions in this area. It is heartening to see than one gene can result in aneurysmal disease. In fact, this extends factors other than aneurysm size getting their due consideration in the into true disease heterogeneity, in which more than one “disease” can evaluation of rupture risk. Size is clearly important, but so are genetics, exhibit the aneurysm phenotype (polycystic kidney disease, collagen environmental factors, hemodynamic forces, and arterial wall biology. vascular diseases, and syndromes including familial chronic obstruc- The interaction of these many factors ultimately leads to aneurysm for- tive pulmonary disease and abdominal aortic aneurysms). Beyond mation and rupture, and the current fixation on size alone is overly genotypic nonspecificity, there is also phenotypic nonspecificity, in simplistic and probably inappropriate. I look forward to future work which aneurysms develop in the setting of various genetic and nonin- from this group, which will likely lead to genetic screening tests, earlier herited diseases, including atherosclerosis, dissections, smoking, hemo- aneurysm diagnosis, and more timely therapeutic interventions. dynamic stress, and hypertension. The interaction between these factors and their relevant genetic susceptibilities and environmental Michael T. Lawton interactions over time add further complexity to this problem. For San Francisco, California example, advancing age as a factor contributing to aneurysm formation is likely related to genetic susceptibilities, environmental factors, mod- ubarachnoid hemorrhage owing to the rupture of IAs is one of the ifiable risks such as smoking and hypertension control, hormonal fac- Smost catastrophic forms of stroke. Despite its importance, the mecha- tors, or mere vascular wear and tear and senescence. nisms of initiation, progression, and rupture of IAs are not yet fully The authors review modern genetic studies of candidate genes, understood. It is evident that both environmental and genetic factors sibling pairs, and parametric and nonparametric linkage with a wide contribute to the pathogenesis of IAs. In this review article, Nahed et al. range of findings and limitations. They note that extended kindreds discuss the data from previous reports related to environmental and with familial aneurysm are uncommon and do not seem to share genetic risk factors of IAs. They focused especially on the results of common linkage. There are likely different genes in different genome-wide linkage analysis. Genetics is one of the promising aneurysm families with much variable penetrance. Yet systematic approaches that elucidate the mechanism of initiation and progression of information is rapidly emerging and groups of genes may emerge as IAs. Genome-wide linkage analyses of kindreds with high prevalence of genuine susceptibility genes. Clinicians remain impressed when con- IAs have identified some loci highly associated with the IA phenotype. fronting a family in which more than one relative died from an Some plausible IA-susceptible genes are included in these loci. aneurysm at young age or in which twins are affected by nearly iden- On the other hand, it is still unknown how functional alteration in IA- tical aneurysms detected at the same age. We remain limited in our susceptible genes has an influence on IA pathogenesis. There are some ability to screen relatives at risk and assess genetic susceptibility. discrepancies in the results of genome-wide linkage analyses. Some loci Instead, we are forced to screen directly for the final aneurysm phe- that have a strong linkage to the IA phenotype in one study were not notype, despite limitations of poor penetrance and age-related sus- replicated in another study. These shortcomings of the genetic approach ceptibility. Hopefully, this will change with genome-wide screening of should be covered by molecular biological methods. Analysis of gene hundreds of potential susceptibility genes and with pattern recogni- expression in human operative aneurysmal samples provides us with tion statistics identifying particularly deadly combinations of genetic important information on the pathogenesis of IAs. By using an experi- and environmental risks. mentally induced IA model in mice, we can examine the effect of a cer-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 225 NAHED ET AL.

tain gene on initiation or progression of IAs. The combination of these 3. Frösen J, Piippo A, Paetau A, Kangasniemi M, Niemelä M, Hernesniemi J, two different approaches is indispensable in the clarification of the com- Jääskeläinen J: Growth factor receptor expression and remodeling of saccular plicated molecular mechanisms of IA progression and to the develop- cerebral artery aneurysm walls: Implications for biological therapy preventing ment of a novel treatment modality for prevention of aneurysm rupture. rupture. Neurosurgery 58:534–541, 2006. 4. Ronkainen A, Hernesniemi J, Puranen M, Niemitukia L, Vanninen R, Ryynänen Hiroharu Kataoka M, Kuivaniemi H, Tromp G: Familial intracranial aneurysms. Lancet Nobuo Hashimoto 349:380–384, 1997. Kyoto, Japan 5. Tulamo R, Frösen J, Junnikkala S, Paetau A, Pitkaniemi J, Kangasniemi M, Niemelä M, Jääskeläinen J, Jokitalo E, Karatas A, Hernesniemi J, Meri S: Complement activation associates with saccular aneurysm wall degeneration n the aneurysm database created in the late 1980s by Dr. Hernesniemi and rupture. Neurosurgery 59:1069–1077, 2006. in Kuopio, Finland, one variable (family history) led to the first study, I 6. van der Voet M, Olson JM, Kuivaniemi H, Dudek DM, Skunca M, Ronkainen published in The Lancet (4), attempting to identify the so-called A, Niemelä M, Jääskeläinen J, Hernesniemi J, Helin K, Leinonen E, Biswas M, aneurysm gene. This first study led to several others (6, 7) in a search Tromp G: Intracranial aneurysms in Finnish families: Confirmation of linkage that is still in progress. and refinement of the interval to chromosome 19q13.3. Am J Hum Gen In Finland, the incidence of cerebral aneurysms, the most common of 74:564–571, 2004. which are middle cerebral artery aneurysms, is the same as elsewhere. 7. Wills S, Ronkainen A, van der Voet M, Kuivaniemi H, Helin K, Leinonen E, However, they tend to rupture more easily and often in a very small Frösen J, Niemelä M, Jaaskelainen J, Hernesniemi J, Tromp G: Familial size. It seems that risk factors also affect different populations in differ- intracranial aneurysms: An analysis of 346 multiplex Finnish families. Stroke 34:1370–1374, 2003. ent ways. Not all aneurysms rupture, so how could we identify those that are more prone to or those that may never need treatment (2)? his review from an outstanding neurovascular genetics group is a This excellent report by Nahed et al. summarizes the current valuable contribution to our literature surveying the considerably thoughts in dealing with these problems. By better understanding the T complex landscape of intracranial aneurysm genetics in great detail. genetics of cerebral aneurysms, we may elucidate the mechanisms Among a number of interesting findings reviewed is the authors’ iden- behind aneurysm growth and rupture. However, whether or not we tification of the 1p35–36 locus discovered in a large intracranial will ever find one single causative mutation is another story. Studies on aneurysm family. Although the 240 genes that reside on this locus are the role of inflammation (5), growth factors (3), and proteomics of the too cumbersome to investigate, the authors have triangulated the evi- aneurysm wall, not to mention thrombus (1), together with markers dence using other genomic interrogative methods to pare this list down that are thus far unknown, are still needed to better understand the to a manageable number of genes to query. Genome-wide sibling pair very dynamic process occurring in the wall. and kindred studies have also incriminated 2p13, 7q22.1, 11q25, 14q22, Mika Niemelä and 19q13.3 as harboring susceptibility genes. The authors review addi- Juhana Frösen tional recent findings from a number of groups in this field in which a Aki Laakso handful of loci have emerged as realistic candidate regions. Reza Dashti As in any genome-wide survey, which may include admixture map- Juha Hernesniemi ping or interfacing with the International HapMap Project in the future, Helsinki, Finland the delineation of lists of candidate genes is just one part of the process to identify disease genes. We look forward to the continued refinement of these lists by this group and others and to the subsequent biological 1. Frösen J, Marjamaa J, Myllärniemi M, Abo-Ramadan U, Tulamo R, Niemelä M, validation required to hone in on a handful of bonafide disease-caus- Hernesniemi J, Jääskeläinen J: Contribution of mural and bone marrow- ing genes that may be used either as biomarkers or as targets for ther- derived neointimal cells to thrombus organization and wall remodeling in a apy. The authors should be congratulated for an expansive review of a microsurgical saccular aneurysm model. Neurosurgery 58:936–944, 2006. highly relevant topic. 2. Frösen J, Piippo A, Paetau A, Kangasniemi M, Niemelä M, Hernesniemi J, Jääskeläinen J: Remodeling of the saccular cerebral aneurysm wall is associ- Ian F. Dunn ated with rupture. Histological analysis of 24 unruptured and 42 ruptured Robert M. Friedlander cases. Stroke 35:2287–2293, 2004. Boston, Massachusetts

226 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CLINICAL REPORT

ANEURYSMS OF THE DISTAL ANTERIOR CEREBRAL ARTERY: RESULTS IN 59 CONSECUTIVELY MANAGED PATIENTS

David A. Steven, M.D., M.P.H. OBJECTIVE: The aim of this study was to present the clinical and radiological charac- Division of Neurosurgery, teristics, surgical management, and outcome in a large series of patients with aneurysms Department of Clinical of the distal anterior cerebral artery (DACA) managed in the microsurgical era. Neurological Sciences, University of Western Ontario, METHODS: The records of 1109 patients with anterior circulation aneurysms managed University Hospital, at the authors’ institution between 1970 and 1998 were reviewed. London Health Sciences Centre, London, Canada RESULTS: Fifty-nine patients (5.3%) were identified with 67 DACA aneurysms. Seventy- three percent of the patients were women. The mean age of all patients was 47 years. Stephen P. Lownie, M.D. Multiple aneurysms were identified in 51% of all patients, most commonly on the mid- Division of Neurosurgery, dle cerebral artery. Thirty-six patients had ruptured DACA aneurysms and 23 had unrup- Department of Clinical tured aneurysms. In those with ruptured aneurysms, the admission grade was Grade I Neurological Sciences, University of Western Ontario, in 10 patients (27.8%), Grade II in three patients (8.3%), Grade III in 10 patients (27.8%), University Hospital, Grade IV in seven patients (19.4%), and Grade V in six patients (16.7%). Frontal lobe London Health Sciences Centre, hematomas occurred in 28% of the patients with ruptured aneurysms and carried a London, Canada poor prognosis. In those with unruptured aneurysms, 11 were incidental and 12 were Gary G. Ferguson, M.D., Ph.D. identified after a subarachnoid hemorrhage from another aneurysm. The mean diame- Division of Neurosurgery, ter was 10 mm in ruptured aneurysms and 5.8 mm in unruptured aneurysms. Fifty-eight Department of Clinical patients underwent surgery and one patient was treated with endovascular coiling. Six Neurological Sciences, patients, all with ruptured aneurysms, died. Seventy percent of survivors with ruptured University of Western Ontario, University Hospital, aneurysms had a favorable outcome. London Health Sciences Centre, CONCLUSION: DACA aneurysms possess a number of characteristics that distinguish London, Canada them from the more common intracranial aneurysms. With modern neurosurgical and Reprint requests: endovascular techniques, an acceptable operative morbidity and mortality can be David A. Steven, M.D., M.P.H., achieved. Division of Neurosurgery, Department of Clinical KEY WORDS: Distal anterior cerebral artery, Intracranial aneurysm, Multiple aneurysms, Pericallosal artery, Neurological Sciences, Subarachnoid hemorrhage University Hospital, London Health Sciences Centre, Neurosurgery 60:227–234, 2007 DOI: 10.1227/01.NEU.0000249267.33945.E7 www.neurosurgery-online.com 339 Windermere Road, Room A10-323, London, Canada N6A 5A5. Email: [email protected] requently referred to as pericallosal vascular anomalies present the neurosurgeon Received, August 1, 2006. aneurysms, distal anterior cerebral artery with a considerable technical challenge. We Accepted, October 4, 2006. F(DACA) aneurysms arise on the anterior present 59 patients with 67 surgically managed cerebral artery distal to the anterior communi- DACA aneurysms. The clinical features and cating artery (AComA). Aneurysms in this surgical results of these patients are reviewed. region are relatively uncommon, accounting for approximately 5% of all intracranial Anatomy and Nomenclature aneurysms (7, 10, 12, 16, 24, 28). They occur at The AComA is normally divided into five major branching points of the distal AComA, segments, A1 through A5 (Fig. 1) (21). Seg- most commonly at the origin of the calloso- ments A2 through A5 form the DACA. The marginal artery. Their location distal to the cir- A2 segment extends from the AComA to the cle of Willis in the interhemispheric fissure and junction between the rostrum and genu of the their frequent association with intraparenchy- corpus callosum. The A3 segment courses mal hemorrhage, multiple aneurysms, and around the genu of the corpus callosum and

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 227 STEVEN ET AL.

TABLE 1. Glasgow Outcome Scale score Score Description 1 Death 2 Persistent vegetative state 3 Severe disability (unable to live independently) 4 Moderate disability (able to live independently) 5 Good recovery

tion was obtained by reviewing the hospital record or, in the case of radiological data, the original films. Outcome, measured by the Glasgow Outcome Scale (GOS) score (13), was determined retrospectively based on the description in the summary sheets or hospital record (Table 1). Outcome was FIGURE 1. Schematic representation of the DACA and possible sites for classified further as either favorable (GOS score Ն 4) or unfa- aneurysms: 1, orbitofrontal artery; 2, frontopolar artery; 3, anterior internal vorable (GOS score Ͻ 4). frontal artery; 4, middle internal frontal artery; 5, posterior internal frontal artery; 6, paracentral artery; 7, superior parietal artery; 8, inferior parietal Statistical Methods artery. The most common site for DACA aneurysms is at the origin of the cal- The size of ruptured and unruptured DACA aneurysms was losomarginal artery (CMA). The asterisk indicates a typical site for a traumatic DACA aneurysm (from, Steven DA, Ferguson GG: Distal anterior compared using the unpaired Student’s t test. A univariate cerebral artery aneurysms, in Winn HR (ed): Youmans Neurological Surgery. analysis was performed to identify risk factors for an unfavor- Philadelphia, Saunders, 2004, vol 2, pp 1945–1957 [27]). able outcome. The unpaired Student’s t test was used for the continuous variables and a Pearson χ2 statistic was used for the categorical variables. For the categorical variables, if at least ends at the point where the vessel takes a sharp turn posteri- one cell on a 2 ϫ 2 table had an expected value of less than 5, orly. The A4 segment runs superior to the corpus callosum, Fisher’s exact test was used to obtain P values. The unadjusted extending to a point just behind the coronal suture. The A5 odds ratio and 95% confidence intervals were obtained for each segment is the portion of the AComA that lies posterior to the of the categorical variables. A significance level of α ϭ 0.05 was coronal suture. chosen for all statistical tests, and all P values in this report are As they do elsewhere in the intracranial circulation, DACA two tailed. A multivariate analysis using a backward elimina- aneurysms occur at the main branching points of the parent tion logistic regression then was used to obtain odds ratios for vessel (6). The most consistent cortical branches of the DACA factors that predicted an unfavorable outcome. For the multi- are the orbitofrontal, frontopolar, anterior internal frontal, mid- variate analysis, standard demographic data (age and sex), as dle internal frontal, posterior internal frontal, paracentral, and well as any risk factor identified as significant during the uni- the superior and inferior parietal arteries (21). The largest variate analysis, was included in the regression model. branch, the callosomarginal artery (CMA), usually arises from Adjusted odds ratios with 95% confidence intervals were used the A3 segment and gives rise to a variable number of the eight to summarize the results. All statistical analyses were com- main cortical branches, the three internal frontal branches being pleted with SAS software version 8.02 for Windows (SAS the most common (21). It is at the origin of this vessel that most Institute, Cary, NC). DACA aneurysms arise (12, 17, 20, 24, 26, 30). RESULTS MATERIALS AND METHODS Of 1109 patients with anterior circulation aneurysms man- All data were collected from patients managed at Victoria aged at the University of Western Ontario between 1970 and and University Hospitals in London, Canada between 1970 1998, 59 patients (5.3%) with 67 DACA aneurysms were iden- and 1998. During this period, summary sheets were com- tified. Forty-three (73%) of the patients in the series were pleted prospectively on most patients with surgically man- women, and the mean age of all the patients was 46.8 years. aged aneurysms. These sheets were completed by the princi- Thirty-six patients had ruptured aneurysms and 23 had unrup- pal surgeon and maintained in the archives of Dr. Charles G. tured aneurysms. Twelve of the patients with unruptured Drake. The summary sheets of all patients with anterior cir- aneurysms experienced a subarachnoid hemorrhage (SAH) culation artery aneurysms were searched and patients with from another intracranial aneurysm and the remaining 11 had DACA aneurysms were identified. For the most part, the rel- incidental aneurysms. evant clinical and radiological characteristics of each patient Among the 36 patients with ruptured aneurysms, the Hunt were extracted from the summary sheets. Missing informa- and Hess (11) clinical grade on admission was Grade I in 10

228 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com ANEURYSMS OF THE DISTAL ANTERIOR CEREBRAL ARTERY

patients (27.8%), Grade II in was not associated with a branching point, but arose at the genu three patients (8.3%), Grade of the CMA. Two of the three traumatic aneurysms were located III in 10 patients (27.8%), in this area. In two patients, the exact branching point could not Grade IV in seven patients be identified. The mean aneurysm diameter was 7.9 mm for the (19.4%), and Grade V in six 55 aneurysms in which this information was available. Twenty- patients (16.7%). All patients four of the aneurysms were less than 6 mm in diameter, 21 meas- with ruptured aneurysms ured 6 to 10 mm in diameter, and eight measured 10 to 25 mm presented with signs and in diameter. There were two giant aneurysms (Ͼ25 mm). The symptoms consistent with mean diameter of ruptured aneurysms (10 mm) was signifi- SAH. Hemiparesis or hemi- cantly larger than that of unruptured aneurysms (5.8 mm; plegia was a presenting fea- P Ͻ 0.014). ture in 13 patients (36%). In addition to recording the origin and size of the distal Nine of these 13 patients AComA aneurysms, the presence of additional aneurysms, vas- were found to have signifi- FIGURE 2. Axial CT scan from a 56- cular anomalies, or AVMs was also determined by angiogra- cant lateralized intrapar- year-old woman with a ruptured phy. Thirty patients (51%) had more than one aneurysm, and a enchymal hematomas (IPH) DACA aneurysm. total of 138 aneurysms were identified in the 59 patients in this on a computed tomographic series. Twenty-nine patients had a single aneurysm, 10 patients (CT) scan or at surgery; one had an acute subdural hematoma; had two aneurysms, seven patients had three aneurysms, six one had focal motor seizures; and in two patients, the cause of patients had four aneurysms, four patients had five aneurysms, the weakness was unknown. Three patients, all children, expe- one patient had seven aneurysms, one patient had eight rienced an SAH resulting from posttraumatic DACA aneurysms, and one patient had nine aneurysms. The distribu- aneurysms. All three patients sustained a severe head injury in tion of additional aneurysms in this and other published series the month before the SAH. The mean interval from head injury is shown in Table 2. Variations of the DACA were noted in nine to SAH was 2 weeks, ranging from 1 day to 1 month. patients (15%; Fig. 4). Six patients had azygous anomalies, all with DACA aneurysms arising at the distal end of the unpaired Radiographic Features The diagnosis of SAH was determined by CT scan or lumbar puncture. CT scanning was performed in 23 out of the A B 36 patients with ruptured aneurysms (Fig. 2). In the remaining 13 patients who were treated before the universal availability of CT scanning, the diagnosis of SAH was established by lumbar puncture or at surgery. The CT imaging features were similar to that seen for ruptured AComA aneurysms with basal and anterior interhemispheric SAH; however, the interhemispheric hemorrhage generally was much thicker and often was associated with hemorrhage within the pericallosal cistern and the cistern of the lamina terminalis. Hemorrhage into the corpus callosum was present in three patients and intraventricular hemorrhage occurred in four patients. Frontal lobe IPHs were seen in 10 patients. The IPH was contralateral to the side of the aneurysm in six patients, ipsilateral in three patients, and bilateral in one patient. All patients underwent four-vessel cerebral angiography C (Fig. 3). Sixty-seven DACA aneurysms were identified. Fifty-two patients had a single DACA aneurysm, six had two DACA aneurysms, and one patient had three DACA aneurysms. The FIGURE 3. The same patient as distribution of aneurysms along the DACA is shown in Figure 1. Figure 2. Anteroposterior (A) and Thirty-eight aneurysms (57%) arose from the origin of the CMA; lateral (B) cerebral angiograms eight aneurysms (12%) arose from the frontopolar branch; six demonstrating a DACA aneurysm aneurysms (9%) arose at the distal end of an azygous A2 seg- arising from the right callosomarginal artery. C, the aneurysm was ment; four aneurysms (6%) arose from the orbitofrontal artery or obliterated successfully with other vessels proximal to the frontopolar branch; two aneurysms Guglielmi detachable coils. (3.0%) arose from the anterior internal frontal artery; two aneurysms (3.0%) arose from the paracentral artery; and one aneurysm (1.5%) arose from the feeding vessel of an arteriovenous malformation (AVM). In four patients (6.0%), the aneurysm

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 229 STEVEN ET AL.

A2 vessel. A triplicate DACA was seen in three patients. was performed in 56 patients. Two patients underwent surgical Angiography revealed AVMs in three patients. In one patient, exploration without clipping. In one patient, a tiny unruptured the aneurysm was on an abnormal feeding vessel, and in the DACA aneurysm was explored, but could not be clipped other two patients, it was located at a normal DACA branching because of its size. In the second, a child in a moribund state, point proximal to the AVM. who eventually died, underwent a posterior fossa exploration to investigate the cause of intraventricular hemorrhage. Surgical Management and Findings Although a preoperative angiogram revealed only hydro- The details of the surgical technique used at our institution cephalus, a large, thrombosed DACA aneurysm was identified have been described in detail elsewhere (27). The approach at autopsy. One patient had her aneurysm treated by endovas- taken depended on the aneurysm’s relationship to the corpus cular coil embolization. callosum. For aneurysms located above the corpus callosum, a The presence of adhesions and blood in the interhemi- right parasagittal craniotomy centered on the coronal suture spheric fissure was universal and often made the dissection was used. For aneurysms arising below the corpus callosum, a difficult, even after draining parenchymal hematomas. Even in bicoronal scalp flap and low frontal parasagittal craniotomy patients with unruptured aneurysms, adhesions were a nui- was used. All 59 patients were managed either microsurgically sance. In a few cases, the anatomy of the A2 and A3 segments or with endovascular methods. Direct microsurgical clipping was difficult to determine because the vessels crossed over

TABLE 2. Incidence of multiple aneurysms in patients with distal anterior cerebral artery aneurysms in 10 reportsa

No. of patients Distal Other No. of Middle Anterior Series No. of with anterior internal Posterior additional cerebral communicating Other (ref. no.) patients multiple cerebral carotid circulation aneurysms artery artery aneurysms artery artery (%)

Current series 59 30 (51) 79 32 8 22 10 6 1

de Sousa et al., 72 32 (44) 50 25 2 15 2 6 — 1999 (4) Inci et al., 14 5 (36) 7 3 1 — 2 1 — 1998 (12) Proust et al., 43 12 (28) 12 3 7 2 — — — 1997 (23) Hernesniemi 84 39 (46) 78 37 8 20 9 4 — et al., 1992 (10) Ohno et al., 42 18 (43) 29 12 7 4 — — 6 1990 (20) Sindou et al., 19 7 (37) 8 4 2 2 — — — 1988 (25) Wisoff and 20 11 (55) 17 8 2 5 1 1 — Flamm, 1987 (28)

Yas¸argil, 1984 (29) 23 12 (52) 13 1 10 — 1 1 —

Snyckers and 24 6 (25) 9 3 1 2 3 — — Drake, 1973 (26) Total 400 172 (43) 302 128 (42%) 48 (16%) 72 (24%) 28 (9.3%) 19 (6.3%) 7 (2.3%)

aAdapted from, Steven DA, Ferguson GG: Distal anterior cerebral artery aneurysms, in Winn HR (ed): Youmans Neurological Surgery, Philadelphia, Saunders, 2004, vol 2, pp 1945–1957 (27).

230 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com ANEURYSMS OF THE DISTAL ANTERIOR CEREBRAL ARTERY

with IPH, only two patients were able to live independently at follow-up (GOS score, Ն4). Of the remaining eight patients, three died and five were severely disabled or vegetative. In a univariate analysis, younger age, higher clinical grade, and the presence of an intracerebral hematoma all were predictive of a poor outcome (GOS score, Ͻ4). When the three pediatric patients with traumatic aneurysms were excluded, the association between poorer outcome and younger age persisted. In a multivariate analysis, only the admission clinical grade was predictive of outcome (P Ͻ 0.0001). In the 12 patients with unruptured DACA aneurysms who experienced an SAH from another intracranial aneurysm, the GOS score was 2 in one patient, 3 in one patient, 4 in four patients, and 5 in six patients. All of the morbidity in this group of patients was associated with the non-DACA aneurysm. All 11 patients with incidental aneurysms had a favorable outcome. Ten patients had no deficits after surgery. One patient with a DACA aneurysm on the feeding vessel of a large frontal AVM who underwent simultaneous clipping of the aneurysm and removal of the AVM had mild postoperative deficits that did not impair his independence (GOS score, 4).

DISCUSSION FIGURE 4. Diagrams illustrating the usual pattern and common variations of the anatomy of the distal anterior cerebral artery based on Baptista’s stud- Aneurysms of the DACA are relatively uncommon. In 11 ies (2) (from, Steven DA, Ferguson GG: Distal anterior cerebral artery series reported between 1960 and 1999, 399 DACA aneurysms aneurysms, in Winn HR (ed): Youmans Neurological Surgery. Philadelphia, were identified among 8266 intracranial aneurysms, for a com- Saunders, 2004, vol 2, pp 1945–1957 [27]). bined incidence of 4.8% (4, 10, 12, 15, 17, 20, 23, 26, 28–31). In the current series, DACA aneurysms formed 5.3% of anterior circulation aneurysms. The number of DACA aneurysms as a each other repeatedly. In two patients, proximal occlusion was percentage of all aneurysms (anterior and posterior circulation) performed because of uncontrollable bleeding at surgery. In was not calculated because of the high number of patients with another patient, the aneurysm could not be clipped without posterior circulation aneurysms treated at our institution (5). occluding the ipsilateral CMA. In patients with bilateral The smaller size of DACA aneurysms relative to more com- DACA aneurysms, it was occasionally impossible to deter- mon aneurysms has been reported (4, 10, 12, 20). With the mine which aneurysm had ruptured. In one patient who exception of two giant aneurysms, de Sousa et al. (4) reported exemplified the confusion that can occur with bilateral that all of the aneurysms in their series measured 10 mm or aneurysms, a preoperative angiogram demonstrated an appar- less. Similarly, Ohno et al. (20) found that 73% of their ent single left DACA aneurysm that was clipped successfully. aneurysms were less than 5 mm in diameter and 94% were less A postoperative angiogram, however, showed a filling than 10 mm in diameter. In the present series, although 44% of aneurysm. At reoperation, the clipped left-sided aneurysm was found to be completely collapsed and a second aneurysm that was not seen at the initial surgery was identified on the right DACA and also clipped. TABLE 3. Outcome in 36 patients with ruptured distal anterior cerebral artery aneurysms Outcome Admission No. of Glasgow Outcome Scale score Outcome was assessed by determining GOS scores for all a grade patients 12 3 45 patients at the time of discharge or at the last available follow- up, whichever was later. The mean follow-up was 13 months, Grade I 10 — — — 1 9 with a range of 3 days to 14 years. Six patients, all with rup- Grade II 3 — — 1 — 2 tured DACA aneurysms, died. In the surviving patients with Grade III 10 1 1 3 2 3 ruptured aneurysms, the GOS score was 2 in one patient, 3 in Grade IV 7 1 — 3 3 — eight patients, 4 in seven patients, and 5 in 14 patients (Table 3). Grade V 6 4 — 1 1 — Including deaths, the mean GOS score in patients with rup- Total 36 6 1 8 7 14 tured aneurysms was 3.6. The presence of an IPH was associ- a Hunt and Hess grade. ated with a significantly poorer outcome. In the 10 patients

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 231 STEVEN ET AL.

TABLE 4. Admission grade and surgical outcome in patients with ruptured distal anterior cerebral artery aneurysms in eight reports Outcome Preoperative Grade I to IIIa Preoperative Grade IV to Va Series (ref. no.) No.b Favorable (%)c Unfavorable (%)d No.e Favorable (%)c Unfavorable (%)d Current series, 2006 23 17 (74) 6 (26) 13 4 (31) 9 (69) de Sousa et al., 1999 (4) 64 53 (83) 11 (17) 1 1 (100) — Inci et al., 1998 (12) 14 13 (93) 1 (7) — — — Proust et al., 1997 (23) 26 20 (77) 6 (23) 9 3 (33) 6 (67) Hernesniemi et al., 1992 (10) 55 38 (69) 17 (31) 10 — 10 (100) Sindou et al., 1988 (25) 18 12 (67) 6 (33) — — — Wisoff and Flamm, 1987 (28) 9 8 (89) 1 (11) 5 — 5 (100) Yas¸argil, 1984 (29) 20 18 (90) 2 (10) 1 — 1 (100) Total 229 179 (78) 50 (22) 39 8 (21) 31 (79)

a Hunt and Hess Grade (11). b Number of patients with ruptured distal anterior cerebral artery aneurysms with preoperative Grades I to III. c Favorable outcome, Glasgow Outcome Scale score greater than or equal to 4. d Unfavorable outcome, Glasgow Outcome Scale score less than 4. e Number of patients with ruptured distal anterior cerebral artery aneurysms with preoperative Grades IV–V. the aneurysms measured less than 6 mm and 82% measured aneurysm. In another report, Hamilton and Falconer (9) noted less than 10 mm, the mean diameter of ruptured DACA that the IPH usually was contralateral to the aneurysm. The aneurysms was 10 mm. This is comparable with ruptured reason for these disparate observations may be because it can aneurysms at other intracranial sites (7) and suggests that there sometimes be extremely difficult to determine from which may be no basis for the general belief that these aneurysms DACA an aneurysm arises (29). rupture at a smaller size than other aneurysms. Many authors have reported an increased incidence of mul- The tendency of ruptured DACA aneurysms to cause IPH tiple aneurysms in patients with DACA aneurysms (4, 10, 12, has been noted in previous reports (9, 17, 25, 26, 28, 29). Dandy 20, 23, 25, 26, 28, 29). In 400 patients described in 10 major (3) theorized that the small pericallosal cistern and the inti- reports, including the present series, 43% had one or more mate relationship of these aneurysms to the adjacent cerebral additional intracranial aneurysms (Table 2). The most frequent hemispheres were responsible for this tendency. In the current site was the middle cerebral artery (42%), followed by the inter- series, IPH was common and carried a decidedly poor progno- nal carotid artery (24%). Additional DACA aneurysms were sis. Ten out of the 36 patients (28%) with ruptured aneurysms identiﬁed in 16% of patients. Interestingly, AComA aneurysms had signiﬁcant frontal hematomas. In most of these patients, were uncommon, accounting for only 9.3% of additional evacuation of the hematoma and clipping of the aneurysm was aneurysms. performed on an emergent basis with minimal or, in one The DACA is a common site for traumatic intracranial patient, no preoperative angiography. Despite rapid surgical aneurysms (1, 8). Asari et al. (1) reviewed 60 reported cases of intervention, the outcome in this subset of patients was dismal. traumatic, peripheral cerebral artery aneurysms, 30% of which Three patients died and, of the seven who survived, only two occurred on the distal AComA. In Fleischer et al.’s (8) review of were able to live independently. We found that the side of the 41 cases of traumatic intracranial aneurysms, 25% were dis- IPH did not depend on the side of the aneurysm. Yas¸argil (29), tributed on the branches of the distal anterior cerebral artery. in a review of his personal experience, made the same observa- These aneurysms seem to be most common in the pediatric tion. Conversely, Sindou et al. (25) found that all of the age group and, in their series, four out of the nine pediatric hematomas in their series occurred ipsilateral to the side of the patients had traumatic aneurysms. Similarly, all three traumatic

232 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com ANEURYSMS OF THE DISTAL ANTERIOR CEREBRAL ARTERY

DACA aneurysms identified by Nakstad et al. (19) occurred in 4. de Sousa AA, Dantas FL, de Cardoso GT, Costa BS: Distal anterior cerebral children. All three traumatic aneurysms in our series occurred artery aneurysms. Surg Neurol 52:128–136, 1999. in children. Two of these children died and the third was 5. Drake CG, Peerless SJ, Hernesniemi JA: Surgery of Vertebrobasilar Aneurysms. London, Ontario Experience on 1767 Patients. Vienna, Springer-Verlag, 1996, severely disabled. pp 7–16. Yas¸argil (29) summarized the surgical difficulties specific to 6. Ferguson GG: Physical factors in the initiation, growth, and rupture of human aneurysms in this location. The interhemispheric fissure and the intracranial saccular aneurysms. J Neurosurg 37:666–677, 1972. callosal cistern are narrow and exposure can be difficult without 7. Ferguson GG: Intracranial arterial aneurysms-a surgical perspective, in Toole JF (ed): Handbook of Clinical Neurology. New York, Elsevier Science, 1989, adequate brain relaxation. The cingulate gyri may be adherent, pp 41–87. making exposure of the corpus callosum and distal AComA ves- 8. Fleischer AS, Patton JM, Tindall GT: Cerebral aneurysms of traumatic origin. sels difficult. This is compounded by the presence of thick, Surg Neurol 4:233–239, 1975. densely adherent clot in the pericallosal cistern, which frequently 9. Hamilton JG, Falconer MA: Immediate and late results of surgery in cases of obscures the normal vascular anatomy and may disorient the saccular intracranial aneurysms. J Neurosurg 16:514–541, 1959. 10. Hernesniemi JA, Tapaninaho A, Vapalahti MP, Niskanen M, Kari A, surgeon and make the dissection slow and tedious. At times, it Luukkonen M: Saccular aneurysms of the distal anterior cerebral artery and may be impossible to tell from which DACA the aneurysm its branches. Neurosurgery 31:994–999, 1992. arises. Even bilateral DACA aneurysms may not be appreciated 11. Hunt WE, Hess RM: Surgical risk as related to time of intervention in the on angiography. Indeed, in one patient in our series, an addi- repair of intracranial aneurysms. J Neurosurg 28:14–20, 1968. 12. Inci S, Erbengi A, Ozgen T: Aneurysms of the distal anterior cerebral artery: tional operation was required to clip a second DACA aneurysm Report of 14 cases and a review of the literature. Surg Neurol 50:130–140, 1998. that was not appreciated either on the initial angiogram or dur- 13. Jennett B, Bond M: Assessment of outcome after severe brain damage. Lancet ing the first procedure. The dome of the aneurysm may be 1:480–484, 1975. densely adherent to the pia of the cingulate gyrus or even the 14. Keston P, White PM, Horribine L, Sellar R: The endovascular management of J Neuroradiol opposite DACA, risking premature rupture during retraction. pericallosal artery aneurysms. 31:384–390, 2004. 15. Laitinen L, Snellman A: Aneurysms of the pericallosal artery: A study of 14 Most commonly, these aneurysms are approached distally and cases verified angiographically and treated mainly by direct surgical attack. one comes upon the fundus of the aneurysm before dissecting J Neurosurg 17:447–458, 1960. the proximal DACA, which also increases the risk of premature 16. Locksley HB: Natural history of subarachnoid hemorrhage, intracranial rupture. With these surgical difficulties in mind, the endovascu- aneurysms and arteriovenous malformations. Parts I and II, in Sahs A, Perret G, Locksley HB, Nishioka H (eds): Intracranial Aneurysms and Subarachnoid lar management of these aneurysms is appealing. However, ini- Hemorrhage: A Cooperative Study. Philadelphia, Lippincott Williams and tial attempts at coiling these aneurysms frequently were unsuc- Wilkins, 1969, pp 37–108. cessful (22). In an early series, Pierot et al. (22) reported that only 17. Mann KS, Yue CP, Wong G: Aneurysms of the pericallosal-callosomarginal two out of eight DACA aneurysms could be coiled successfully. junction. Surg Neurol 21:261–266, 1984. 18. Menovsky T, van Rooij WJ, Sluzewski M, Wijnalda D: Coiling of ruptured Possibly owing to an improvement in endovascular technology, pericallosal artery aneurysms. Neurosurgery 50:11–15, 2002. more recent investigators have reported much higher rates of 19. Nakstad P, Nornes H, Hauge HN: Traumatic aneurysms of the pericallosal success (14, 18). arteries. Neuroradiology 28:335–338, 1986. The outcomes obtained in this series are comparable to those 20. Ohno K, Monma S, Suzuki R, Masaoka H, Matsushima Y, Hirakawa K: of previous reports (Table 4). Despite the difficulties associated Saccular aneurysms of the distal anterior cerebral artery. Neurosurgery 27:907–913, 1990. with aneurysms in this location, good results can be achieved 21. Perlmutter D, Rhoton AL Jr: Microsurgical anatomy of the distal anterior in most patients who have a good clinical grade. Poor grade cerebral artery. J Neurosurg 49:204–228, 1978. patients remain a significant challenge, particularly those 22. Pierot L, Boulin A, Castaings L, Rey A, Moret J: Endovascular treatment of patients with large parenchymal hemorrhages. Although it is pericallosal artery aneurysms. Neurol Res 18:49–53, 1996. 23. Proust F, Toussaint P, Hannequin D, Rabenenoina C, Le Gars D, Freger P: tempting to recommend surgery to evacuate the hematoma, in Outcome in 43 patients with distal anterior cerebral artery aneurysms. Stroke our experience, such patients usually have a dismal outcome 28:2405–2409, 1997. despite rapid surgery. As is the case for other intracranial 24. Royand F, Carter L, Guthkelch A: Distal anterior cerebral artery aneurysms, aneurysms, we anticipate that most of these aneurysms will in Carter L, Spetzler RF, Hamilton M (eds): Neurovascular Surgery. New York, eventually be treated by endovascular means. Nonetheless, McGraw-Hill, 1995, pp 717–728. 25. Sindou M, Pelissou-Guyotat I, Mertens P, Keravel Y, Athayde AA: Pericallosal knowledge of the microsurgical treatment of these aneurysms aneurysms. Surg Neurol 30:434–440, 1988. and the expected outcomes remains important for surgeons 26. Snyckers FD, Drake CG: Aneurysms of the distal anterior cerebral artery. A managing intracranial aneurysms. report on 24 verified cases. S Afr Med J 47:1787–1791, 1973. 27. Steven DA, Ferguson GG: Distal anterior cerebral artery aneurysms, in Winn HR (ed): Youmans Neurological Surgery. Philadelphia, Saunders, 2004, vol 2, pp REFERENCES 1945–1957. 28. Wisoff JH, Flamm ES: Aneurysms of the distal anterior cerebral artery and 1. Asari S, Nakamura S, Yamada O, Beck H, Sugatani H: Traumatic aneurysm associated vascular anomalies. Neurosurgery 20:735–741, 1987. of peripheral cerebral arteries. Report of two cases. J Neurosurg 46:795–803, 29. Yas¸argil MG: Distal anterior cerebral artery aneurysms, in Yasargil MG (ed): 1977. Microneurosurgery. Stuttgart, Georg Thieme Verlag, 1984, vol 2, pp 224–231. 2. Baptista A: Studies on the arteries of the brain. II. The anterior cerebral artery; 30. Yas¸argil MG, Carter LP: Saccular aneurysms of the distal anterior cerebral Some anatomic features and their clinical implications. Neurology 13: artery. J Neurosurg 40:218–223, 1974. 825–835, 1963. 31. Yoshimoto T, Uchida K, Suzuki J: Surgical treatment of distal anterior cerebral 3. Dandy WE: Intracranial Arterial Aneurysms. Ithaca, Comstock, 1944. artery aneurysms. J Neurosurg 50:40–44, 1979.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 233 STEVEN ET AL.

COMMENTS review of more than 1000 aneurysm patients treated at one institution during a 28-year period. The authors highlight the fact that, although n this report, Steven et al. reviewed their experience with 59 consec- DACA aneurysms are rare, certain clinically relevant characteristics Iutively managed patients with distal anterior cerebral artery (DACA) unique to these lesions exist. Outcome analysis emphasizes the poor aneurysms. Based on their abundant experience with the treatment of prognosis associated with intraparenchymal hemorrhage. The obser- cerebral aneurysms, the authors presented clinical characteristics of vation that DACA aneurysms tend to be associated with multiple DACA aneurysms regarding multiplicity, sex, size, location, and surgi- aneurysms is confirmed by this series. However, it refutes the general cal outcome, comparing them with previous reports. Poor surgical out- belief that these aneurysms tend to be of a smaller diameter when comes were shown to be correlated with preoperative clinical grade they rupture. and large parenchymal hemorrhages. Because more than half of the The discussion of surgical considerations may be the most important aneurysms are located at the anterior part of the anterior cerebral arter- part of this work because these aneurysms require a different surgical ies, brain injury at the frontal lobe may cause higher cognitive dysfunc- strategy than those of the circle of Willis. In our experience, careful tion. Future studies with long-term evaluation of cognitive function are scrutiny of preoperative imaging is critical to determine whether or not necessary to discuss the surgical variations between direct surgery and the anterior cerebral artery is azygous and to identify the number of effer- endovascular means. ent vessels leaving the area. It is also critical to clarify preoperatively whether the aneurysm originates from the right, left, or azygous vessel. Nobuhiro Mikuni We disagree with the authors that these lesions are likely to be pri- Nobuo Hashimoto marily treated endovascularly in the foreseeable future. The technical Kyoto, Japan challenges from the standpoint of catheter maneuverability are formidable, whereas surgical access is relatively straightforward and well he authors have reported the results of a retrospective review of 67 tolerated with the caveats mentioned in this work. TDACA aneurysms in 59 patients treated at a single institution during a 28-year period. In reviewing their series, the authors have iden- Edward Duckworth tified a number of unique characteristics that distinguish DACA H. Hunt Batjer aneurysms from aneurysms in other typical locations. In reviewing Chicago, Illinois other series of DACA aneurysms, it is evident that most of the characteristics outlined in this review are similar to characteristics identified ACA aneurysms are a subset of aneurysms with a number of in previous publications. characteristics that represent a special challenge for their micro- Distinguishing features of DACA aneurysms are the high incidence D surgical management. Their profound location in the colossal cistern of presentation with intracerebral hemorrhage, more frequent multi- and the need to dissect both medial portions of the frontal lobes in a plicity of aneurysms, and trauma as an etiology. Previous reports have narrow interhemispheric fissure can cause bilateral frontal lobe lesions. suggested that DACA aneurysms are more likely to rupture at a Their usual small size and broad neck could make clipping the neck smaller size than aneurysms in other locations. In this series, the mean difficult. The high incidence of multiple aneurysms associated with diameter of ruptured aneurysms was 10 millimeters and the mean DACA aneurysms is another source of a possibly poor outcome in diameter for unruptured aneurysms was 5.8 millimeters, which does management. not confirm this perception. The authors report a retrospective study of a series of 59 consecu- In the current series, only one aneurysm was treated with endovascu- tively treated patients who presented with DACA aneurysms. Only lar therapy. This may be owing to the fact that endovascular therapy was one patient in this series was treated with endovascular coiling. Their not widely used during a significant portion of the retrospective review. results are very good compared with those previously reported in the It has been our own experience that a significant majority of DACA literature. DACA aneurysms were initially very difficult to treat using aneurysms have a configuration that does not make them readily the endovascular approach. With developments in technology, how- amenable to endovascular therapy and we continue to treat the majority ever, most aneurysms at this location can now be appropriately of these lesions by surgical clip ligation. The high incidence of associated approached using endovascular coiling, leaving microsurgery for very intracerebral hemorrhage also makes surgery the treatment of choice if small or giant aneurysms. evacuation of the hematoma is required for therapeutic purposes. The authors address the issue of difficulty of exposure of these Atos Alves de Sousa aneurysms because of the presence of adhesions and blood in the inter- Belo Horizonte, Brazil hemispheric fissure. Although the authors did not describe their surgical approach and positioning in detail, most textbooks and previous he authors present a series of 59 patients with DACA aneurysms articles describe the interhemispheric approach with the patient in the Twho were treated at two Canadian institutions over a 30-year period supine position and the head slightly flexed. We prefer to approach and likely include a large number of patients operated on by Dr. Charles these aneurysms with the head in the lateral position and the operative Drake, a pioneering aneurysm surgeon. As such, the series has signifi- side parallel to the floor with the vertex slightly elevated. This position cant historical interest. Since the start of this series, the neurosurgical has the advantage of using gravity to allow the ipsilateral frontal lobe world has changed dramatically. We have seen the introduction of early to fall away from the falx and improves the field of view by orienting aneurysm surgery, triple-H therapy for vasospasm, a whole new world the interhemispheric fissure with the plane of the surgeon’s eyes. of computerized imaging and stereotactic guidance, balloon angioplasty for vasospasm, endovascular coiling and stenting, and improved micro- Daniel L. Barrow surgical techniques. It will be interesting to learn if these modern Atlanta, Georgia approaches and new technology can improve these results. teven et al. present a detailed retrospective analysis of 59 patients Robert A. Solomon Swith 67 DACA aneurysms. These patients were culled from a New York, New York

234 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CLINICAL STUDIES

Martin Lehecka, M.D. Department of Neurosurgery, Helsinki University Central Hospital, Helsinki, Finland NO LONG-TERM EXCESS MORTALITY IN

Mika Niemelä, M.D., Ph.D. 280 PATIENTS WITH RUPTURED DISTAL Department of Neurosurgery, NTERIOR EREBRAL RTERY NEURYSMS Helsinki University Central Hospital, A C A A Helsinki, Finland OBJECTIVE: The aim of this study was to assess the long-term excess mortality after the Johanna Seppänen, M.Sc. rupture of distal anterior cerebral artery (DACA) aneurysms compared with that of a Finnish Cancer Registry, Helsinki, Finland matched general Finnish population in an unselected, population-based series. METHODS: We identiﬁed 280 consecutive patients (119 men, 161 women) treated for Hanna Lehto, M.D. ruptured DACA aneurysms (clipped, 262; coiled, 10; no intervention, 8) at two neuro- Department of Neurosurgery, surgical centers serving solely the southern and eastern parts of Finland from 1976 to Helsinki University Central Hospital, 2003. All patients were followed from subarachnoid hemorrhage until death or the end Helsinki, Finland of 2004. No patients were lost to follow-up. Long-term excess mortality was estimated Timo Koivisto, M.D., Ph.D. using the annual relative survival ratio compared with the general Finnish population Department of Neurosurgery, matched by age, sex, and calendar time. Kuopio University Hospital, RESULTS: The median follow-up period was 9.6 years (range, 0.1–29 yr). The 3-year Kuopio, Finland cumulative relative survival ratio was 0.84 (95% confidence interval, 0.78–0.88), implying 16% excess mortality in the patient group during the first 3 years after subarach- Antti Ronkainen, M.D., Ph.D. noid hemorrhage. The annual relative survival ratio attained 1.0 at the fourth year Department of Neurosurgery, Kuopio University Hospital, of follow-up, indicating no excess mortality thereafter. There were four episodes of Kuopio, Finland recurrent subarachnoid hemorrhage and only one from a treated DACA aneurysm, with a 10-year cumulative risk of 1.4% (95% confidence interval, 0.0–3.0). Jaakko Rinne, M.D., Ph.D. Cardiovascular disease and cancer were the leading causes of death after 10 years Department of Neurosurgery, of follow-up. Kuopio University Hospital, Kuopio, Finland CONCLUSION: After surviving 3 years after the rupture of a DACA aneurysm, the patients’ long-term survival became similar to that of the matched general population. Risto Sankila, M.D., Ph.D. Rebleeding of treated DACA aneurysm was rare. Finnish Cancer Registry, KEY WORDS: Cerebral aneurysm, Mortality, Outcome, Subarachnoid hemorrhage, Survival Helsinki, Finland

Neurosurgery 60:235–241, 2007 DOI: 10.1227/01.NEU.0000249261.95826.8F www.neurosurgery-online.com Juha Jääskeläinen, M.D., Ph.D. Department of Neurosurgery, Kuopio University Hospital, Kuopio, Finland opulation-based long-term outcome ing as homogenous a study population as pos- studies of aneurysmal subarachnoid sible. DACA aneurysms were chosen because Juha Hernesniemi, M.D., Ph.D. Phemorrhage (SAH) are scarce (17, 23). Do the series published so far are relatively small, Department of Neurosurgery, the aneurysms occluded by microsurgical clip- and long-term follow-up data is lacking. Also, Helsinki University Central Hospital, ping or, recently, by endovascular coiling stay clipping has been the “gold standard” of treat- Helsinki, Finland closed for years and decades? How do the ment in DACA aneurysms, making it less brain and the cardiovascular system endure biased regarding the effect of treatment Reprint requests: the impact caused by SAH in the long run? To method on long-term survival (1, 5, 13, 14, 16, Martin Lehecka, M.D., Department of Neurosurgery, address these two questions, we studied the 20, 34). They are located in the interhemi- Helsinki University Central Hospital, long-term outcome after SAH in a specific spheric ﬁssure, between the frontal lobes, and Topeliuksenkatu 5, aneurysm group, namely distal anterior cere- they comprise only 3 to 7% of all intracranial P.O. Box 266, bral artery (DACA) aneurysms, which are also aneurysms (5, 20). They bleed into the adja- FIN-00029-HUS, Finland. E-mail: martin.lehecka@hus.ﬁ called pericallosal aneurysms. We decided to cent brain tissue in almost half of the cases (5), concentrate on aneurysms at only one more often than aneurysms located elsewhere Received, May 16, 2006. anatomic location to minimize the confound- (20). Mortality is not considered to be higher, Accepted, October 5, 2006. ing factors related to location, thereby obtain- but cognitive problems owing to frontal lobe

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 235 LEHECKA ET AL.

injury are more likely (5, 27). DACA aneurysms are usually RESULTS more suitable for clipping than coiling because of their distal Patients and Aneurysms location and small size. In Finland, a high-quality public health care system, compre- In the 280 patients with ruptured DACA aneurysms, the hensive vital statistics, a relatively stable population, and a mean age at rupture was 50 years (range, 14–80 yr), 47 years high incidence of SAH for adults support long-term outcome (range, 14–75 yr) for men and 52 years (range, 27–80 yr) for studies of SAH patients (9, 23, 25). In this population-based women. The total follow-up period was 2894 person-years with study of 280 patients treated for ruptured DACA aneurysms, a median follow-up period of 9.6 years (range, 1 mo–29 yr). No causes of death and excess mortality compared with a matched patients were lost to follow-up. Of the 280 patients, 87 (31%) had general Finnish population were of special interest. multiple aneurysms. However, only 121 patients underwent four-vessel angiography, 40 (33%) of whom had multiple aneurysms. Eighteen patients (6%) had multiple DACA PATIENTS AND METHODS aneurysms; 262 patients (94%) were treated with microsurgical clipping, 10 (4%) with endovascular coiling, and eight (3%) were Patients and Aneurysms treated conservatively because of poor grade on admission. Of these eight patients, seven died during the first day and one We identified 280 consecutive patients who were treated for died 5 months after the DACA aneurysm rupture. During the ruptured DACA aneurysms from 1976 to 2003 in the two neu- acute phase, only the ruptured aneurysm was treated (inter- rosurgical referral centers for the southern (Helsinki Univer- hemispheric approach). However, in 14 of the 18 patients with sity Central Hospital) and eastern (Kuopio University Hos- multiple DACA aneurysms, the unruptured aneurysm was pital) parts of Finland, with a total catchment population of clipped in the same session as well. Patients who recovered approximately 3 million. These 280 patients comprised 7% of well and had multiple aneurysms that required treatment with all admitted SAH patients during the same time period. The a different approach were treated in separate sessions 3 to 12 patients’ clinical and radiological data were reviewed. months later with no observed mortality. Characteristics of the Diagnosis of SAH was based on lumbar puncture before 1980 study population are summarized in Table 1. and on computed tomographic (CT) scans thereafter. Ruptured DACA aneurysms were identified either by conventional Recurrent Subarachnoid Hemorrhage angiography or, more recently, by CT angiography (11). Clinical status on admission was expressed using the Hunt and Hess New episodes of SAH occurred in four (95% CI, 1.09–10.2) of scale (7). the 280 patients during the follow-up period. These included one patient with re-bleeding from a subtotally clipped DACA Follow-up Data aneurysm at 17 years, one patient with rupture of a de novo middle cerebral artery aneurysm at 2 years, one patient with Available follow-up data was collected starting from the day rupture of an untreated middle cerebral artery aneurysm at of diagnosis of SAH until death or December 31, 2004. Vital sta- 3 years, and one patient with rupture of a previously undiag- tus at the end of year 2004 was ascertained from the Population nosed basilar bifurcation aneurysm at 5 years. The estimated Register Centre, which contains information on all people cumulative risk for a new episode of SAH at 10 years was 0.014 residing in Finland. The dates and causes of death were (95% CI, 0.000–0.030) and at 20 years was 0.030 (95% CI, obtained from Statistics Finland. From the same register, we 0.000–0.065). The incidence for recurrent SAH was 138 per also obtained causes of death for the entire population of 100,000 follow-up years (95% CI, 38–354), which indicates a Finland during the study period. risk ratio of 3.9 (95% CI, 1.5–7.3) compared with the general adult Finnish population with an annual incidence of SAH of Statistical Methods 35 per 100,000 (25). The relative survival ratio (RSR) provided a measure of the excess mortality for patients diagnosed with ruptured DACA Early Mortality aneurysms, irrespective of whether the excess mortality was Table 2 shows the causes of death stratified by survival time. directly or indirectly attributable to the illness (2). The RSR By the end of the follow-up period, 92 (33%) of the 280 patients was calculated by dividing the observed survival by the had died, 36 within 12 months, implying a 13% management expected survival. The expected survival was derived, using mortality at 1 year. The causes of the 36 deaths during the first the Ederer II method, from that of the comparable general pop- 12 months were mostly owing to the primary SAH in 33 (92%) ulation of Finland matched with respect to age, sex, and calen- patients. A large number of the patients (17 out of 36 [47%]) dar time based on data from Statistics Finland (2). The 95% who died during the first 12 moths were initially in Hunt & confidence intervals (CI) for annual RSR and cumulative rela- Hess Grades 4 and 5. tive survival estimates were calculated by assuming normal distribution. Statistical analysis was carried out using SAS soft- Long-term Mortality ware (SAS Institute, Cary, NC). This study was approved by No patients died of a recurrent SAH from a treated DACA the ethical committees of both university hospitals. aneurysm. Instead, one patient died owing to SAH from a pre-

236 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com NO LONG-TERM EXCESS MORTALITY IN RUPTURED DISTAL ANTERIOR CEREBRAL ANEURYSMS

TABLE 1. Characteristics of the 280 patients admitted between 1976 and 2003 owing to subarachnoid hemorrhage caused by distal anterior cerebral artery aneurysma (280 ؍ Total (n (161 ؍ Men (n = 119) Women (n Age, mean Ϯ SD (yr) 47.2 Ϯ 11.5 52.1 Ϯ 11.5 50.1Ϯ11.7 Patients with multiple aneurysms, no. (%) 28 (24%) 59 (37%) 87 (31%) Ruptured aneurysm sizes, no. (%) Ͻ7mm 55 (46%) 91 (57%) 146 (52%) 7–14 mm 52 (44%) 62 (39%) 114 (41%) 15–24 mm 11 (9%) 6 (4%) 17 (6%) Ͼ24 mm 1 (1%) 2 (1%) 3 (1%) Hunt and Hess classiﬁcation at diagnosis, no. (%) Grade 1 23 (19.3%) 25 (15.5%) 48 (17.1%) Grade 2 44 (37.0%) 48 (29.8%) 92 (32.9%) Grade 3 35 (29.4%) 53 (32.9%) 88 (31.4%) Grade 4 9 (7.6%) 21 (13.0%) 30 (10.7%) Grade 5 8 (6.7%) 14 (8.7%) 22 (7.9%) Treatment, no. (%) Microsurgery 112 (94.1%) 150 (93.2%) 262 (93.6%) Endovascular 3 (2.5%) 7 (4.3%) 10 (3.6%) Died before treatment 4 (3.4%) 4 (2.5%) 8 (2.8%) Follow-up Total person-years 1375 1519 2894 Median, yr (range) 11.5 (0.2–27.0) 7.4 (0.1–29.0) 9.6 (0.1–29.0)

a SD, standard deviation.

TABLE 2. Causes of death of the 92 patients who died during follow-up in the series of 280 consecutive patients with subarachnoid hemorrhage from distal anterior cerebral artery aneurysma Time from diagnosis to death Age at SAH diagnosis (mean ؎ SD) Total <12 mo 1–3 yr 4–10 yr >10 yr Related to primary SAH 54 Ϯ 14 33 3 5 2 43 Rebleeding from treated aneurysm 0 0 0 0 0 0 SAH from another aneurysm 46 0 0 1 0 1 Cerebrovascular 58 Ϯ 12 0 2 1 0 3 Cardiovascular 52 Ϯ 11 1 5 3 7 16 Cancer 55 Ϯ 700268 Infection or pneumonia 58 Ϯ 11 0 0 2 4 6 Trauma 49 Ϯ 12 1 1 1 1 4 Suicide 48 Ϯ 18 0 1 1 0 2 Other 52 Ϯ 15 1 2 2 3 8 Not known 72 0 0 0 1 1 All deaths 54 Ϯ 13 36 14 18 24 92

a SAH, subarachnoid hemorrhage; SD, standard deviation.

viously undiagnosed basilar bifurcation aneurysm. Of the 208 Excess Mortality patients who had survived more than 4 years, 42 died. The Table 3 shows the annual and cumulative RSRs for the entire cause of death was cardiovascular disease in 24% and cancer in study group. The 3-year cumulative RSR was 0.84 (95% CI, 19%. Of the 137 patients who survived for more than 10 years, 0.78–0.88), implying 16% excess mortality for patients with rup- 24 died. The cause of death was cardiovascular disease in 29% tured DACA aneurysm during the ﬁrst 3 years after diagnosis and cancer in 25%. when compared with the general population (Fig. 1). At the

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 237 LEHECKA ET AL.

TABLE 3. Annual relative survival ratio, cumulative relative survival ratio, and corresponding 95% confidence intervals for patients with ruptured distal anterior cerebral artery aneurysmsa No. of Interval RSR CRSR patients (yr) (95% CI) (95% CI) at risk 0–1 280 0.87 (0.83–0.91) 0.87 (0.83–0.91) 1–2 243 0.96 (0.93–0.98) 0.84 (0.79–0.88) 2–3 229 0.99 (0.97–1.00) 0.84 (0.78–0.88) FIGURE 1. Cumulative RSR of patients with ruptured DACA aneurysms 3–4 221 1.00 (0.98–1.01) 0.84 (0.79–0.88) (n ϭ 280) as a function of follow-up period in years. Error bars indicate 95% CI. 4–5 208 0.98 (0.95–1.00) 0.83 (0.77–0.87) 5–6 192 0.98 (0.95–1.00) 0.81 (0.75–0.86) 6–7 172 1.01 0.82 (0.76–0.87) 7–8 163 0.99 (0.95–1.00) 0.81 (0.75–0.86) 8–9 151 1.00 (0.96–1.01) 0.81 (0.75–0.86) 9–10 144 1.00 (0.96–1.01) 0.81 (0.75–0.87) 10–11 137 1.01 (0.96–1.01) 0.82 (0.75–0.87) 11–12 124 0.99 (0.94–1.01) 0.81 (0.74–0.87) 12–13 109 0.99 (0.94–1.01) 0.80 (0.73–0.87) 13–14 98 0.96 (0.89–0.99) 0.77 (0.69–0.84) 14–15 87 0.99 (0.92–1.01) 0.76 (0.68–0.84) 15–16 75 0.99 (0.91–1.01) 0.76 (0.67–0.83) 16–17 71 1.02 0.77 (0.68–0.85) FIGURE 2. Annual RSR of patients with ruptured DACA aneurysms (n ϭ 17–18 62 0.99 (0.87–1.01) 0.76 (0.66–0.84) 280) as a function of follow-up period in years. Error bars indicate 95% CI. 18–19 52 1.02 0.78 (0.68–0.86) 19–20 44 0.93 (0.78–0.99) 0.72 (0.60–0.82) 20–21 36 0.96 (0.80–1.01) 0.69 (0.57–0.81)

a RSR, relative survival ratio; CI, confidence intervals; CRSR, cumulative relative survival ratio. 95% conﬁdence intervals are not expressed for time intervals with no deaths fourth year of follow-up, the annual RSR attained 1.0, indicating no excess mortality thereafter (Fig. 2). Separate analysis of the patient group with good recovery at 1 year showed no excess mortality during the entire follow-up period. The 3-year cumulative RSR was 0.82 (95% CI, 0.74–0.87) for women and 0.86 (95% CI, 0.78–0.92) for men, indicating 18 and 14%, excess mortality during the ﬁrst 3 years, respectively. Annual RSRs for both sexes are showed in Figure 3. Analysis by age at diagno- ϭ ϭ sis (Ͻ45 yr, 45–60 years, Ͼ60 yr) showed that the youngest FIGURE 3. Annual RSR for men (n 119) and women (n 161) with rup- group reached an annual RSR of 1.0 1 to 2 years earlier with tured DACA aneurysms as a function of follow-up period in years. Error bars less excess mortality. indicate 95% CI. DISCUSSION In the present series of 280 ruptured DACA aneurysms with Relatively little is known about the long-term survival rates a median follow-up period of nearly 10 years, there were no after aneurysmal SAH, as only two population-based studies fatal rebleedings from the 262 clipped DACA aneurysms and have been published (17, 23). Major studies, such as the no excess mortality after the first 3 years of follow-up. DACA International Subarachnoid Aneurysm Trial (15) and the aneurysms were chosen for this study to minimize the effect International Study of Unruptured Intracranial Aneurysms of location on mortality and morbidity (21) because microsur- (33), have reported outcomes with a median follow-up period gical clipping has been the main treatment modality until of 4 years. Other studies have reported outcomes even earlier now (12) and because there is little data published on the after the initial rupture (4, 6). The two population-based stud- course of SAH in patients with ruptured DACA aneurysms (1, ies on long-term survival of SAH patients used the standard- 5, 13, 14, 16, 20, 34). ized mortality ratio (SMR) as a measure of excess mortality.

238 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com NO LONG-TERM EXCESS MORTALITY IN RUPTURED DISTAL ANTERIOR CEREBRAL ANEURYSMS

Neither used annual RSR (2). In Iceland, a study of 44 SAH undiagnosed aneurysm, in a median follow-up period of almost patients with a median follow-up period of 10 years suggested 10 years. In a Finnish study by Juvela et al. (10), de novo that the patients with good recovery at 6 months subsequently aneurysms developed with an annual rate of 0.84% during a had the same mortality as the general population (17). In median follow-up period of 19 years. The International Study of Finland, a study of 1537 patients by Ronkainen et al. (23) with Unruptured Intracranial Aneurysms suggested a 5-year cumu- a median follow-up period of 7.5 years showed that the lative rupture rate of 1.5% for unruptured anterior circulation patients with good recovery at 1 year still had increased long- aneurysms in patients with previous SAH (33). In a Japanese term mortality compared with the general population (SMR, study, the risk for SAH from de novo aneurysm was 1.4% at a 2.02). Ronkainen et al.’s series consisted of ruptured aneurysms 10-year follow-up period (28). However, a recent study from of all loci, with a worse prognosis in posterior circulation The Netherlands states the incidence of recurrent SAH after aneurysms (18). In the present Finnish series of DACA clipping of ruptured aneurysms in a 10-year follow-up as 3.2% aneurysms, RSR did not suggest excess mortality for the 172 with 77% caused by de novo aneurysms (31). Our results for (61%) patients with good recovery at 1 year, whereas SMR recurrent SAH with a 1.4% rate of cumulative incidence at 10- (1.53) did. Unlike SMR, which is a quotient of the observed years were closer to the Japanese study, although within the over the expected number of deaths, RSR takes into account the confidence intervals of the Dutch study as well. Altogether, temporal variation of excess mortality (2). This makes it a more recurrent SAH seems to be an infrequent cause of death for sophisticated method for evaluation of excess mortality. patients with long-term follow-up after surgically treated rup- The management mortality at 12 months was 13%, similar to tured DACA aneurysms, even if the cumulative rupture rate previous studies (23, 24), and most of these early deaths (92%) increases with longer follow-up periods and the relative risk were directly owing to the primary SAH and its sequelae. In the compared with the general population is higher (9, 30). present and previous studies, increased age at the time of SAH Based on our study, routine angiographic control studies and poor preoperative clinical condition increased early comor- during long-term follow-up are unlikely to be effective in pre- bidity and mortality in SAH patients (18, 19). Later in the follow- venting new episodes of SAH in most patients. Young patients up period, the distribution of death causes started to resemble with multiple aneurysms or family history may behave differ- that of the general Finnish population, with the two major causes ently. Our data supports the view that, after a recovery period, of death being cardiovascular disease and cancer. Several studies ruptured and treated DACA aneurysms should not cause any have addressed the acute extracranial complications after SAH, long-term effects on life expectancy. Aneurysms at other loca- including cardiac arrhythmias, dysfunction and myocardial tions might behave differently. Unfortunately, at the moment, injury, pulmonary edema and acute lung injury, and renal and we have no such data available allowing us to perform similar hepatic dysfunction (26, 29). There are no studies on the long- survival analysis for other locations. However, we are plan- term effect of SAH on the cardiorespiratory system. Hypertension ning to do so in the future. and smoking are important risk factors for SAH, but they also predispose to cardiovascular disease and cancer (3, 8). CONCLUSION The risk of rebleeding from a treated aneurysm is of major concern for a patient after microsurgical or endovascular occlu- After surviving 3 years from diagnosis, patients in this series sion. In our series, there was only one case of rebleeding and no with ruptured DACA aneurysms had survival rates similar to deaths from the 262 clipped aneurysms. DACA aneurysms are those of a matched general population. During the ﬁrst 3 years, located relatively distally in the cerebrovascular tree with nar- the deaths were mainly owing to the primary SAH or its seque- row parent arteries, which could indicate a smaller shear stress lae. After 3 years, the leading causes of death were cardiovas- onto the vessel wall, making DACA aneurysms less prone to cular disease and cancer. This study indicates that clipping is a rebleeding than aneurysms at other anatomic locations (31). long-lasting method of treatment for DACA aneurysms. Based on our results, microsurgical clipping is a long-lasting Ruptures of de novo or other aneurysms in the long-term treatment for DACA aneurysms, as also suggested by the follow-up period were infrequent. Recurrent episodes of SAH International Subarachnoid Aneurysm Trial with a shorter leading to death were exceptional. follow-up period (15). Multiple aneurysms, which are usually already present at REFERENCES the ﬁrst SAH or developing later (de novo), are detected in approximately one-third of the SAH patients (22). They are 1. de Sousa AA, Dantas FL, de Cardoso GT, Costa BS: Distal anterior cerebral considered a predisposing factor for recurrent SAH, along with artery aneurysms. Surg Neurol 52:128–135, 1999. 2. Ederer F, Axtell LM, Cutler SJ: The relative survival rate: A statistical method- smoking and hypertension (32). In the present series, 33% of ology. Natl Cancer Inst Monogr 6:101–121, 1961. the 121 patients who underwent four-vessel angiography had 3. Feigin VL, Rinkel GJ, Lawes CM, Algra A, Bennett DA, van Gijn J, Anderson multiple aneurysms as opposed to the 31% of the patients in CS: Risk factors for subarachnoid hemorrhage: An updated systematic review the whole series. Obviously, a small number of posterior circu- of epidemiological studies. Stroke 36:2773–2780, 2005. 4. Hackett ML, Anderson CS: Health outcomes 1 year after subarachnoid hem- lation aneurysms were missed in the remaining 159 patients (5). orrhage: An international population-based study. The Australian In our series, only one of the 92 deaths (1%) was owing to Cooperative Research on Subarachnoid Hemorrhage Study Group. Neur- recurrent SAH, which resulted from a de novo or a previously ology 55:658–662, 2000.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 239 LEHECKA ET AL.

5. Hernesniemi JA, Tapaninaho A, Vapalahti MP, Niskanen M, Kari A, Luuk- rhage: A report of the multicenter, cooperative aneurysm study. Participants konen M: Saccular aneurysms of the distal anterior cerebral artery and its of the Multicenter Cooperative Aneurysm Study. Crit Care Med 23:1007– branches. Neurosurgery 31:994–998, 1992. 1017, 1995. 6. Hop JW, Rinkel GJ, Algra A, van Gijn J: Case-fatality rates and functional out- 27. Stenhouse LM, Knight RG, Longmore BE, Bishara SN: Long-term cognitive come after subarachnoid hemorrhage: a systematic review. Stroke 28:660–664, deficits in patients after surgery on aneurysms of the anterior communicating 1997. artery. J Neurol Neurosurg Psychiatry 54:909–914, 1991. 7. Hunt WE, Hess RM: Surgical risk as related to time of intervention in the 28. Tsutsumi K, Ueki K, Usui M, Kwak S, Kirino T: Risk of subarachnoid hemor- repair of intracranial aneurysms. J Neurosurg 28:14–20, 1968. rhage after surgical treatment of unruptured cerebral aneurysms. Stroke 8. Juvela S: Risk factors for aneurysmal subarachnoid hemorrhage. Stroke 30:1181–1184, 1999. 33:2152–2153, 2002. 29. Tung P, Kopelnik A, Banki N, Ong K, Ko N, Lawton MT, Gress D, Drew B, 9. Juvela S, Porras M, Poussa K: Natural history of unruptured intracranial Foster E, Parmley W, Zaroff J: Predictors of neurocardiogenic injury after aneurysms: probability of and risk factors for aneurysm rupture. J Neurosurg subarachnoid hemorrhage. Stroke 35:548–551, 2004. 93:379–387, 2000. 30. van der Schaaf IC, Velthuis BK, Wermer MJ, Majoie C, Witkamp T, de Kort G, 10. Juvela S, Poussa K, Porras M: Factors affecting formation and growth of Freling NJ, Rinkel GJ, ASTRA Study Group: New detected aneurysms on intracranial aneurysms: a long-term follow-up study. Stroke 32:485–491, 2001. follow-up screening in patients with previously clipped intracranial 11. Kangasniemi M, Makela T, Koskinen S, Porras M, Poussa K, Hernesniemi JA: aneurysms: Comparison with DSA or CTA at the time of SAH. Stroke Detection of intracranial aneurysms with two-dimensional and three- 36:1753–1758, 2005. dimensional multislice helical computed tomographic angiography. 31. Wermer MJ, Greebe P, Algra A, Rinkel GJ: Incidence of recurrent subarach- Neurosurgery 54:336–340, 2004. noid hemorrhage after clipping for ruptured intracranial aneurysms. Stroke 12. Keston P, White PM, Horribine L, Sellar R: The endovascular management of 36:2394–2399, 2005. pericallosal artery aneurysms. J Neuroradiol 31:384–390, 2004. 32. Wermer MJ, van der Schaaf IC, Velthuis BK, Algra A, Buskens E, Rinkel GJ, 13. Laitinen L, Snellman A: Aneurysms of the pericallosal artery: A study of 14 ASTRA Study Group: Follow-up screening after subarachnoid haemorrhage: cases verified angiographically and treated mainly by direct surgical attack. Frequency and determinants of new aneurysms and enlargement of existing J Neurosurg 17:447–458, 1960. aneurysms. Brain 128:2421–2429, 2005. 14. Mann KS, Yue CP, Wong G: Aneurysms of the pericallosal-callosomarginal 33. Wiebers DO, Whisnant JP, Huston J 3rd, Meissner I, Brown RD Jr, Piepgras junction. Surg Neurol 21:261–266, 1984. DG, Forbes GS, Thielen K, Nichols D, O’Fallon WM, Peacock J, Jaeger L, 15. Molyneux AJ, Kerr RS, Yu LM, Clarke M, Sneade M, Yarnold JA, Sandercock Kassell NF, Kongable-Beckman GL, Torner JC, International Study of P, International Subarachnoid Aneurysm Trial (ISAT) Collaborative Group: Unruptured Intracranial Aneurysms Investigators: Unruptured intracranial International subarachnoid aneurysm trial (ISAT) of neurosurgical clipping aneurysms: Natural history, clinical outcome, and risks of surgical and versus endovascular coiling in 2143 patients with ruptured intracranial endovascular treatment. Lancet 362:103–110, 2003. aneurysms: A randomised comparison of effects on survival, dependency, 34. Yas¸argil MG, Carter LP: Saccular aneurysms of the distal anterior cerebral seizures, rebleeding, subgroups, and aneurysm occlusion. Lancet artery. J Neurosurg 40:218–223, 1974. 366:809–817, 2005. 16. Ohno K, Monma S, Suzuki R, Masaoka H, Matsushima Y, Hirakawa K: Saccular aneurysms of the distal anterior cerebral artery. Neurosurgery Acknowledgments 27:907–912, 1990. We thank Ms. Maarit Alalahti and Ms. Katariina Helin for their excellent tech- 17. Olafsson E, Hauser WA, Gudmundsson G: A population-based study of prog- nical assistance. This study was funded by Helsinki University Central Hospital nosis of ruptured cerebral aneurysm: Mortality and recurrence of subarach- research funds (EVO grant TYH5204). We declare that there were no conflicts of noid hemorrhage. Neurology 48:1191–1195, 1997. interest. We have nothing to disclose. 18. Osawa M, Hongo K, Tanaka Y, Nakamura Y, Kitazawa K, Kobayashi S: Results of direct surgery for aneurysmal subarachnoid haemorrhage: Outcome of 2055 patients who underwent direct aneurysm surgery and pro- COMMENTS file of ruptured intracranial aneurysms. Acta Neurochir (Wien) 143:655–663, 2001. ehecka et al. have reviewed a series of patients with ruptured dis- 19. O’Sullivan MG, Dorward N, Whittle IR, Steers AJ, Miller JD: Management Ltal anterior cerebral artery (DACA) aneurysms. With 100% follow- and long-term outcome following subarachnoid haemorrhage and intracra- up in 280 consecutive patients, the study demonstrates that patients nial aneurysm surgery in elderly patients: an audit of 199 consecutive cases. surviving at least 3 years after treatment of a ruptured pericallosal Br J Neurosurg 8:23–30, 1994. aneurysm had long-term survival similar to that of the general popu- 20. Proust F, Toussaint P, Hannequin D, Rabenenoina C, Le Gars D, Freger P: lation. There were only a few rare deaths from causes other than the Outcome in 43 patients with distal anterior cerebral artery aneurysms. Stroke original subarachnoid hemorrhage (SAH) during the first 3 years. My 28:2405–2409, 1997. 21. Raaymakers TW, Rinkel GJ, Limburg M, Algra A: Mortality and morbidity of conclusion from these data is that there is no special mortality risk for surgery for unruptured intracranial aneurysms: A meta-analysis. Stroke patients who are successfully treated and recover from an aneurysmal 29:1531–1538, 1998. SAH. Because essentially all treated patients in this series were surgi- 22. Rinne J, Hernesniemi JA, Puranen M, Saari T: Multiple intracranial cally clipped, the data further reinforce the durability of clipping as a aneurysms in a defined population: Prospective angiographic and clinical cure for ruptured aneurysms. study. Neurosurgery 35:803–808, 1994. 23. Ronkainen A, Niskanen M, Rinne J, Koivisto T, Hernesniemi JA, Vapalahti Robert A. Solomon MP: Evidence for excess long-term mortality after treated subarachnoid hem- New York, New York orrhage. Stroke 32:2850–2853, 2001. 24. Ropper AH, Zervas NT: Outcome 1 year after SAH from cerebral aneurysm. n this interesting study, the authors identify 280 consecutive patients Management morbidity, mortality, and functional status in 112 consecutive treated for ruptured DACA aneurysms. The authors explain why good-risk patients. J Neurosurg 60:909–915, 1984. I they chose this population of patients, most of whom are still treated 25. Sarti C, Tuomilehto J, Salomaa V, Sivenius J, Kaarsalo E, Narva EV, Salmi K, Torppa J: Epidemiology of subarachnoid hemorrhage in Finland from 1983 to with open surgery with minimal surgical manipulation and morbidity 1985. Stroke 22:848–853, 1991. from the actual aneurysm location. The major finding is that there was 26. Solenski NJ, Haley EC Jr, Kassell NF, Kongable G, Germanson T, Truskowski no excess mortality related to the intracranial aneurysm after 3 years. L, Torner JC: Medical complications of aneurysmal subarachnoid hemor- Up until 3 years, there was an increase in expected mortality. These

240 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com NO LONG-TERM EXCESS MORTALITY IN RUPTURED DISTAL ANTERIOR CEREBRAL ANEURYSMS

findings are in keeping with what is observed in long-term follow-up lobe parenchymal hemorrhage with the attendant potential behavioral of aneurysm patients. The authors have a well-controlled population and neuropsychological consequences. Therefore, although these data and are in a unique situation to perform such studies. They are to be are highly relevant to the target aneurysm, they should not be gener- commended for their long-term follow-up of this population of alized to predict long-term outcomes from the more common conduct- patients. It would be interesting to see if similar results could be ing system aneurysms. obtained from aneurysms at other locations thought to have a higher H. Hunt Batjer incidence of recurrence such as basilar tip or paraclinoid aneurysms. Chicago, Illinois Christopher S. Ogilvy Boston, Massachusetts his study examines the long-term follow-up of patients treated for DACA aneurysms after presenting with SAH. It is very interesting he authors report an extraordinary series of 280 patients treated T and important for several reasons. It is a population-based analysis, after SAH from DACA aneurysms from 1976 to 2003. The reasons T which is very uncommon among studies dealing with cerebral for this analysis were to determine the durability of clipping or coiling aneurysms that generally report on surgical or neuroradiological series and to detect the long-term implications of SAH on the brain and car- (hospital populations). Moreover, the authors use sophisticated statis- diovascular system. Several important points can be gleaned from their tical analysis and relative survival ratio to evaluate the patients’ sur- analysis using relative survival ratio and a few questions emerge. As vival. This method takes more factors affecting excess mortality of expected, 50% of the early deaths were related to poor clinical grade. patients into account compared with the more common standardized This factor together with treatment morbidity resulted in 16% excess mortality rate (ratio between observed and expected cases) and, thus, mortality in 3 years. This excess mortality was not detectable after the provides more reliable results. fourth year. The durability of treatment was exceptional, with only one This study is clear and easy to read and understand. With 18% of the treated aneurysm rebleeding after a mean follow-up period of 9.6 patients being classified as Grades IV or V at the time of admission and years. This rebleed was from a known subtotally clipped aneurysm. a mortality rate of 13% at 1 year, the results of this study are excellent. Overall, the risk of recurrent SAH was 1.4% at 10 years. All other recur- The vast majority of patients (all but 10) underwent operation, suggest- rent bleeding was from either previously undetected aneurysms or the ing that surgery is superior to the endovascular approach for DACA de novo development of new lesions. When one examines the deaths aneurysms. occurring between 1 and 3 years, there was a somewhat high risk of I would have liked to see more data or clarification concerning the death from cardiovascular disease and one suicide. This implies that following. How many aneurysms did the authors observe in the same there may be some intermediate-term jeopardy of the cardiovascular period? Among the 280 DACA aneurysms, what was the spectrum of system and possibly neuropsychological sequelae. dimensions? Did the authors treat multiple aneurysms? If so, in what It is important to note that this specific subgroup would typically way? Have any nonbleeding multiple aneurysms ever been operated? have lower risks of diffuse basal SAH, diffuse vasospasm, and retraction injury in important regions such as the anterior communicating Alessandro Ducati artery. Aneurysms in this region would have a higher risk of frontal Turin, Italy

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 241 CLINICAL STUDIES

RESULTS OF MICROSURGICAL CLIPPING OF 50 HIGH COMPLEXITY BASILAR APEX ANEURYSMS

Ali F. Krisht, M.D. OBJECTIVE: Complex basilar aneurysms (large size, wide base, low bifurcation, and Department of Neurosurgery, dysmorphic posteriorly projecting domes) frequently fail endovascular treatment. We University of Arkansas for Medical Sciences, report our experience using the pretemporal transzygomatic transcavernous approach Little Rock, Arkansas with 50 complex basilar aneurysms. METHODS: Using the pretemporal transcavernous route, opening the occulomotor Niklaus Krayenbühl, M.D. trigone, and removing the anterior clinoid and the posterior clinoid when necessary, a Department of Neurosurgery, wide exposure of the interpeduncular fossa is achieved. Temporary clips are applied University of Arkansas for Medical Sciences, to a perforator-free zone of the basilar trunk, proximal to the superior cerebellar artery. Little Rock, Arkansas Complexity criteria in the 50 aneurysms included large or giant size in 27 patients, wide dysmorphic base in 18 patients, low bifurcation in 21 patients, posteriorly project- David Sercl, M.D. ing dome in 11 patients, and dolichoectasia of the apex in three patients. Department of Neurosurgery, RESULTS: Twenty-ﬁve patients presented with subarachnoid hemorrhage. There were 14 University of Arkansas for Medical Sciences, men and 36 women between the ages of 32 and 76 years (mean, 52.2 yr). Forty-nine Little Rock, Arkansas aneurysms (98%) were successfully clipped. There was no procedure-related mortality. Two patients died (one from delayed bowel ischemia and one from a vasospasm-related Kerem Bikmaz, M.D. complication). There were three ischemia-related events, two of which were procedure- Department of Neurosurgery, related (medial thalamic lacunar infarct, superior cerebellar distribution ischemia) and University of Arkansas for Medical Sciences, one which was a third distal middle cerebral cardiac embolus after stopping Coumadin Little Rock, Arkansas (DuPont Pharmaceuticals, Wilmington, DE) for atrial fibrillation. Transient partial or complete occulomotor palsies occurred in all patients with full recovery as the rule, Paulo A.S. Kadri, M.D. except in one patient. At discharge, Glascow Outcome Scale scores were 4 or 5 in 88% Department of Neurosurgery, of the patients. At the 6-month follow-up examination, Rankin Outcome Scale scores University of Arkansas for Medical Sciences, were 0 to 2 in 92% of the patients. Little Rock, Arkansas CONCLUSION: Our experience reintroduces microsurgery as a safe and more durable treatment option for the management of complex basilar apex aneurysms that tend to Reprint requests: Ali F. Krisht, M.D., have a higher rate of failure with endovascular therapy. Department of Neurosurgery, KEY WORDS: Aneurysm, Basilar, Outcome, Transcavernous University of Arkansas for Medical Sciences, Neurosurgery 60:242–252, 2007 DOI: 10.1227/01.NEU.0000249265.88203.DF www.neurosurgery-online.com 4301 W. Markham, #507, Little Rock, AR 72205. Email: [email protected]

Received, July 10, 2006. asilar apex aneurysms continue to re- who consider microsurgical clipping of basilar Accepted, November 3, 2006. main a treatment challenge. The recent apex aneurysms as a treatment option. Btrend has been to refer most, if not all, Endovascular therapy of basilar apex region basilar apex region aneurysms for endovascu- aneurysms has proven its safety (3, 16, 18, 37, lar treatment (3, 16–18, 37, 38, 41, 44, 48, 49, 54, 41, 44, 48–50, 62, 64). However, its efficacy in 62, 64). This is driven by several factors, the terms of its durability remains in question. This most important of which is the higher reported is best reflected in the high rate of recanaliza- morbidity of microsurgical clipping of poste- tion and regrowth of coiled aneurysms (20). rior circulation aneurysms when compared Basilar apex aneurysms are more prone to with anterior circulation aneurysms (2, 8, recanalization, regrowth, and coil compaction 10–15, 21–24, 31–34, 39, 45, 47, 51, 52, 56–60, 65, owing to a hemodynamic disadvantage created 67–70). This trend is further strengthened by by their anatomic location in relation to the the diminishing number of neurosurgeons direction of blood flow. In addition, coiled basi-

242 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MICROSURGICAL CLIPPING OF 50 HIGH COMPLEXITY BASILAR APEX ANEURYSMS

lar aneurysms have a persistent risk of future bleeding, which was reported at an annual rate of as high as 1.3% per year, with TABLE 2. High complexity criteria the annual risk of bleeding in partially coiled aneurysms reach- Criteria No. of aneurysms Percentage ing as high as 2.1% (20). This relatively high annual risk of bleeding is similar to, and possibly higher than, the reported Size annual risk of bleeding of unruptured aneurysms (27, 28, 66). Large 16 33% This raises the concern of whether or not coiling an unruptured Giant 11 22% aneurysm provides any protection. This issue becomes more of Posterior projecting dome 11 22% a concern when dealing with the more complex, large-sized Low bifurcation 21 41% aneurysms that are more likely to be partially coiled and more Wide dysmorphic base 18 35% prone to both recanalization and rebleeding (20). Dolichoectasia 3 6% In this study, we present our experience with the microsurgical treatment of high complexity basilar apex aneurysms. These are aneurysms that are more prone to fail endovascular (small size, small neck, and/ therapy. They also have features that make them difﬁcult to or anteriorly projecting with treat with the classic subtemporal and pterional approaches. the neck above the level of The microsurgical details of this approach were previously the posterior clinoid process) reported as they relate to the ﬁrst 21 patients treated; these were not included in the patients are also included in this series (30). In this study, we study (Fig. 1). These patients report and analyze the outcome of microsurgical clipping of 50 were treated with the classic high complexity basilar apex aneurysms treated with the pterional transsylvian ap- pretemporal transcavernous approach. proach. One patient with a growing, previously coiled CLINICAL PATIENTS AND METHODS aneurysm who was explored FIGURE 1. Angiogram showing a and deemed too risky to clip, small basilar tip aneurysm with a small Between July 1998 and June 2006, 90 patients with verte- was excluded from the analy- neck, which was treated with the pteri- brobasilar aneurysms (a total of 139 aneurysms) were treated at sis. The data was prospec- onal transsylvian approach without the need for a transcavernous approach. the Cerebrovascular Clinic of the University of Arkansas for tively collected and retrospec- Medical Sciences. Fifty-one patients with a total of 82 aneurysms Such aneurysms were not included in tively analyzed. this study. (Table 1) met the criteria of high-complexity basilar apex region The criteria of high com- aneurysms (Table 2), in which the pretemporal transcavernous plexity used included: 1) large (1.5–2.4 cm) or giant (Ͼ2.5 cm) approach was used. Patients with non-complex aneurysms size (Fig.2); 2) posterior projecting dome of the aneurysm (Fig. 3); 3) low bifurcation, in which the basilar bifurcation is

TABLE 1. Distribution of basilar tip aneurysms and location of additional aneurysmsa (%) 51 ؍ Location of basilar tip aneurysms n Basilar tip 42 (82) Basilar tip/SCA 7 (14) Basilar tip/P1/P2 2 (4) (%) 31 ؍ Location of other aneurysms n ICA cavernous sinus 2 (6.5) ICA paraclinoid 2 (6.5) ICA PComA 6 (19.4) ICA ant. chor. 1 (3.2) ICA bifurcation 2 (6.5) AComA 6 (19.4) MCA bifurcation 10 (32.3) MCA distal 1 (3.2) PICA 1 (3.2)

a SCA, subclavian artery; ICA, internal carotid artery; PComA, posterior communicating artery; ant. chor., anterior choroidal artery; AComA, anterior communicating artery; MCA, middle cerebral artery; PICA, posterior-inferior FIGURE 2. Angiograms showing a giant-sized basilar tip aneurysm. communicating artery. Preoperative anteroposterior (A) and lateral (B) views, as well as postoperative anteroposterior (C) and lateral (D) views obtained 6 months postoperatively.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 243 KRISHT ET AL.

of the occulomotor nerve along its course from the brainstem to the superior orbital fissure. After opening the sylvian fissure, the Liliequist membrane and the arachnoid of the interpeduncular fossa are dissected and an assessment of the position of the basilar bifurcation in relationship to the posterior clinoid process is made. When necessary, the posterior clinoid process is removed. This provides the space to apply a temporary clip to the basilar trunk region. The clip is applied proximal to the take-off of the superior cerebellar artery and at a perforator-free zone. If need FIGURE 3. A–C, computed tomographic angiograms with three-dimensional reconstruction showing a giant-sized be, the posterior communicat- basilar tip aneurysm that projected posteriorly. These angiograms also show an associated bilobed paraclinoid aneurysm. ing artery is divided when it is D and E, postclipping intraoperative angiograms of the basilar tip aneurysm. F, anteroposterior angiogram of the inter- not of the fetal type. It is usu- nal carotid artery obtained after clipping the paraclinoid aneurysm. ally divided at a perforator- free zone, which is commonly located behind the posterior clinoid process to where proximal located at the junction of the posterior communicating artery control is not easy to achieve without drilling (Fig. 4); 4) wide with the posterior cerebral artery. By the time these steps are dysmorphic base, in which the aneurysm (neck) base is wide performed, a wide exposure of the interpeduncular fossa is and not suitable for endovascular coiling with the origin of one achieved at the depth of the surgical field. Removal of the or both posterior cerebral arteries incorporated in the wide base medial aspect of the posterior clinoid may sometimes be complex (Fig. 4); and 5) dolichoectatic changes in the basilar needed to better visualize the contralateral P1 segment, espe- apex region, in which the whole basilar apex complex is dilated cially in a low lying basilar bifurcation. The approach and its (angiographic dolichoectatic) with an aneurysmal dilatation details were published in a previous study (30). The advan- (Fig. 5). At least one or more of the complexity criteria occurred tages gained in relation to the clip application process are fur- in each of the aneurysms. Also, at least one or more of the com- ther discussed in the discussion section. Figure 6 provides an plexity criteria that render an aneurysm more prone to regrow illustration and an accompanying intraoperative photograph or compact with endovascular therapy, such as large or giant showing the wide view of the interpeduncular fossa achieved size and/or wide neck, were present in 48 aneurysms. using the pretemporal transcavernous approach. Figure 7 is a sequence of photographs demonstrating the clip application Surgical Approach process for the aneurysm shown in Figure 4. It reflects the ade- The surgical approach used is the transzygomatic pretempo- quate visualization of a basilar trunk for applying a proximal ral transcavernous approach, which is performed by drilling temporary clip before attacking the aneurysm. The clips are and widening the zygomatic notch to allow more inferior applied proximal to the take-off of the superior cerebellar reflection of the temporalis muscle. This is followed by a pteri- artery at a perforator-free zone. This location of the clip will onal craniotomy with a temporal extension. These steps allow continue to allow collateral blood flow through the superior a more pretemporal approach to the interpeduncular fossa. The cerebellar artery system to continue perfusing the perforator lateral wall of the cavernous sinus is exposed by separating while significantly decreasing the flow into the aneurysm. The the dura propria of the temporal lobe from the lateral wall of location of the temporary clip is away from the basilar bifurca- the cavernous sinus. Hemostasis within the cavernous sinus is tion region and provides ample space for efficient maneuver- achieved by injecting fibrin glue (Tisseal; Baxter Healthcare ability, resulting in better final clip application. After the final Corp., Deerfield, IL) between the V1 and V2 branches of the clip is applied, the microDopplers are used to confirm ade- trigeminal nerve. The sphenoid wing and the anterior clinoid quate flow in the posterior cerebral branches. If there is any process are then removed. The dura is then opened in a question in this regard, intraoperative angiography is per- T-shaped fashion with a vertical arm of the T extending along formed. For a detailed discussion of the approach and its the indentation of the sphenoid wing, all the way to the level of advantages, we refer readers to the previously published arti- the occulomotor trigone. This opening allows full mobilization cle on the technical details of the approach (30).

244 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MICROSURGICAL CLIPPING OF 50 HIGH COMPLEXITY BASILAR APEX ANEURYSMS

FIGURE 4. A and B, preoperative computed tomographic scans of a patient with a large ruptured basilar tip aneurysm. C, anteroposterior view showing the dysmorphic wide FIGURE 5. A, anteroposterior angiogram showing a dolicoectatic dilation of base of the aneurysm. D, view show- the basilar tip region with an associated aneurysm that presented with a his- ing the low location of the neck of tory of multiple ruptures. B, three-dimensional computed tomographic the aneurysm hiding behind the pos- angiogram of the same aneurysm. terior clinoid process, which was col- ored with green. E, postoperative angiogram with one clip used after resecting and reconstructing the aneurysm neck with bipolar coagulation, as surviving patients had a Rankin Outcome Scale score of 0 to shown in Figure 7. 2 (Fig. 8, Table 4). Two (4%) of the surviving patients had a Rankin Outcome Scale score of 3. Follow-up studies using RESULTS angiography (40 patients [80%]) and CT angiography (five patients [10%]) showed no evidence of any residual or recur- The patients’ ages ranged between 32 and 76 years (mean, rent aneurysm in those patients. The average initial radiolog- 52.2 yr). There were 36 women and 14 men. Twenty-ﬁve of the ical follow-up time is 1 year. Five patients had no long-term patients (50%) presented with subarachnoid hemorrhage radiological follow-up because of death in two, the aneurysm (Table 3) and 25 (50%) presented with other symptoms, includ- not being clipped in one, and a Rankin score greater than 3 in ing headaches, transient ischemic attacks, and symptoms two. The longest follow-up period is 9 years. The median related to mass effect. follow-up period is 33 months. None of the patients devel- Complete clipping with no evidence of residual aneurysm oped a rebleed during a total follow-up period of 137.9 was achieved in 49 aneurysms (98%). One unruptured basilar patient-years. aneurysm was not clipped, owing to perforators attached to the neck region in a patient who was operated on for a ruptured Complications posterior communicating artery. Although there was no procedure-related mortality, there Forty-nine out of the 50 surviving patients were discharged was one case of postsurgical-related mortality. This occurred from the hospital. At discharge, 44 (88%) of the patients had in a patient with a giant dolichoectatic basilar apex aneurysm a Glascow Outcome Scale score of 4 to 5 and five patients and a history of multiple subarachnoid hemorrhages who (10%) had a Glascow Outcome Scale score of 3 (Table 4) (26). died during her hospitalization 10 days after her initial hem- At the 6-month follow-up examination, 46 (92%) out of the 48 orrhage owing to vasospasm-related cerebral ischemia com-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 245 KRISHT ET AL.

FIGURE 6. A, illustration showing a wide exposure of the interpeduncular fossa region achieved using the transcavernous approach. B, intraoperative photograph of the transcavernous approach showing a wide exposure in the region of the neck of a very large wide based aneurysm. PCOM, posterior communicating artery; ICA, internal carotid artery; III, oculomotor nerve; P1, P1 segment of the posterior cerebral artery; SCA, superior cerebral artery; BA, basilar artery; IV, trochlear nerve; VI, first branch of the trigeminal nerve; ON, optic nerve. plicated with increased intracranial pressure in spite of triple- FIGURE 7. Intraoperative photographs demonstrating the clipping process of the aneurysm shown in Figure 4. A, the wide view achieved with the transcav- H therapy (Fig. 2). Another patient died 2 months after dis- ernous approach after removal of the posterior clinoid process. It also shows the charge owing to unrelated complications of bowel ischemia first step of applying the temporary clips to the basilar trunk. B, the temporary (Table 5). clip applied to the basilar trunk and both P1 segments (asterisks). The Three of the surviving patients experienced ischemia- aneurysm dome has been resected away from the coagulated neck region. related complications. One patient had a small mesial thala- Notice the full visualization of the region of the basilar apex. C, the final clip mic infarct with full recovery and employment by 6 months, in place after achieving full visualization of the surrounding anatomy as another had ischemia in the superior cerebellar artery distri- shown in B. III, oculomotor nerve; An, aneurysm; BA, basilar trunk bution resulting in postoperative ataxia, and a third experi- aneurysm; P1, posterior cerebral arteries. enced right middle cerebral artery distribution embolic stroke related to atrial fibrillation after stopping his anticoagulation therapy for surgery. One patient developed postoper- Oculomotor Palsy ative cisternal and subdural hematoma needing evacuation Partial (n ϭ 25) and complete (n ϭ 25) transient oculomotor due to leakage from the stump of the cut posterior communi- palsies occurred in all patients. In two patients with giant cating artery. Two patients developed a postoperative cere- aneurysms, transient right complete and left partial oculomo- brospinal fluid leak related to the bony sinuses and pneuma- tor palsies occurred in the immediate postoperative period. On tized posterior clinoid process. One patient had a minor long-term follow-up evaluation (up to 1 yr postoperatively), wound revision. recovery of the oculomotor palsy was the rule in all patients,

246 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MICROSURGICAL CLIPPING OF 50 HIGH COMPLEXITY BASILAR APEX ANEURYSMS

TABLE 3. Hunt and Hess scores for patients with subarachnoid (25 ؍ hemorrhage (n Hunt and Hess Grade No. of patients 14 214 35 42 50

FIGURE 8. Bar graph showing the Rankin Outcome Scale scores at discharge TABLE 4. Outcome at discharge and at 6 monthsa and at 6 months. Outcome at discharge GOS No. of patients, n ϭ 50 (%) TABLE 5. Complicationsa 4–5 44 (88) 3 5 (10) Procedure-related transient complications No. of patients (%) 1 1 (2)b Ischemia (thalmus) 1 (2) Outcome at 6 months Ischemia (cerebellum SCA) 1 (2) Rankin Score No. of patients, n ϭ 50 (%) Perioperative complications 0–2 46 (92) Subdural hematoma (with decreased 1 (2) 3 2 (4) consciousness) 6 2 (4)c CSF leak 2 (4) a GOS, Glasgow Outcome Scale. Wound healing problem 1 (2) b One patient died during hospitalization because of therapy refractory Ischemia, embolic (MCA territory) 1 (2) vasospasm. owing to atrial fibrillation c A second patient died 2 months after hospitalization owing to mesenteric Severe vasospasm leading to mortality 1 (2) infarction and bowel ischemia. a SCA, subclavian artery; CSF, cerebrospinal fluid; MCA, middle cerebral artery. Refer to text for third nerve palsy. except one who had a mild partial residual palsy, which was easily corrected with prism glasses. Seven out of the 48 surviving patients, five of whom had partial oculomotor palsy and advanced cranial base techniques; and 6) the perception that, if two of whom had complete oculomotor palsy in the immediate a procedure is less involved such as endovascular therapy, it postoperative period, have not been followed long enough to means that it is less risky. document their full recovery. At the time of the study, five of It is becoming clear that ideal aneurysms for coiling are those seven patients have only intermittent diplopia small aneurysms with small necks and an ideal dome-to-neck ratio. Large and giant aneurysms and/or wide-neck DISCUSSION aneurysms have a high rate of recanalization and growth (3, 16, 18, 20, 37, 38, 41, 44, 48–50, 62, 64). The location of the Nowadays, the treatment of basilar apex aneurysms is dom- aneurysm is another important factor in terms of recanaliza- inated by endovascular coiling. In a majority of centers, these tion and growth as it relates to the direction of blood flow. A patients are routinely referred for endovascular coiling (3, 16, large number of basilar apex aneurysms are not ideal 18, 37, 41, 44, 48–50, 62, 64). This trend is owing to several fac- aneurysms for endovascular therapy because of their location, tors, including: 1) the relatively higher complication rate with shape, and size. This is why they have a significantly high rate microsurgical clipping of basilar aneurysms compared with of recanalization. This comes mostly in the form of com- anterior circulation aneurysms; 2) the heightened severity of paction. Henkes et al. (20) reported on their experience with the complications because of the involvement of perforators to 316 basilar bifurcation aneurysms treated with coiling. Coil the brainstem region; 3) the limitations of the classic pterional compaction occurred in 24% of patients on follow-up angiog- and subtemporal approaches in achieving proper visualization raphy. This necessitated a second treatment in 15% of the of the neurovascular anatomy in this deep region and their patients during a mean follow-up period of 27 months. They ability to provide adequate space for proximal control; 4) also reported a cumulative annual bleeding rate of 1.3%, which improved safety of endovascular procedures; 5) the lack of increases to 2.1% in partially coiled aneurysms (20). knowledge by many about the safety of and what can be Drake’s unmatched experience and results in treating basilar achieved with microsurgery of basilar aneurysms using artery aneurysms were achieved using the subtemporal

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 247 KRISHT ET AL.

approach in the overall majority of aneurysms. The main argu- the clip location allows collateral flow from the superior cere- ment against the subtemporal approach is centered around the bellar artery, which is usually enough to continue perfusing excessive retraction of the temporal lobe, which is especially the perforators, but, at the same time, it softens the aneurysm needed when operating in the acute phase after subarachnoid and makes it easy to clip; and 3) should a rupture occur, the hemorrhage (10–15, 47). Drake’s argument in support of the amount of bleeding is very manageable and, often, the use of subtemporal approach over the pterional approach is based on bipolar coagulation can seal the bleeding site and convert the his experience with the pterional approach being narrow and case to an unruptured-like situation. confining, as well as providing less visualization of the This strategy has helped us achieve a major shift in the clip- aneurysm, especially its posterior aspect in the interpeduncu- ping application process. Yas¸argil (69, 70) has always advocated lar fossa “where most of the trouble lurks with the perforating that aneurysm clipping is not a single step, but rather a process vessels” (9, 14, p 29). He also felt that the pterional approach that involves multiple steps with the aim of complete oblitera- presents a difficulty in applying the clip along the parent bifur- tion and then final perfect clipping of the aneurysm. The steps cation and carries the risk of kinking the P1 segment (9, 14). in this process involve intermittent application of temporary Drake recognized the potential benefit of an approach combin- clips and gradual shrinkage of the aneurysm when possible, ing both approaches. He especially recognized its potential in followed by the final clipping step after full visualization of the that “the clip may be applied from in front of, over or beside circumferential anatomy. The same concept was described by the dorsum, with all vessels in view” (9, 14, p 29). Drake et al. (10) and was reflected in their good results. The pterional approach for the treatment of basilar apex The benefits of the advances in cranial base microsurgery for aneurysms was popularized by Yas¸argil (68, 69, 70). His reasons the treatment of complex vascular lesions should continue to in support of this approach over the subtemporal approach are be recognized and used. The role of endovascular therapy in as follows: 1) to avoid excessive retraction of the temporal lobe; the treatment of cerebral aneurysms, especially small aneur- 2) to provide a shortcut direct approach along the sylvian fis- ysms with small necks, is well established. Advances are con- sure; 3) to provide the ability to better visualize the contralat- tinually being made to improve on the treatment of large and eral P1 segment of the posterior cerebral artery; 4) to provide more complex aneurysms. However, the trend of relying solely better visualization of the perforators, especially those arising on endovascular therapy may lead to the dominance and from the contralateral P1 segment; and 5) to clip more availability of only one treatment modality to the detriment of aneurysms in patients with multiple aneurysms. the other. This trend results in a diminishing number of neu- Using the classic pterional and subtemporal approaches, rosurgeons who are able to safely clip basilar apex region there was significant limitation in both the space availability aneurysms. This fact and the disappearing role of micro- and the ability to achieve adequate proximal control without surgery as an option in treating aneurysms is, in general, mak- compromising the view to the neck of the aneurysm and its ing us accept the shortcomings of endovascular therapy with surroundings (2, 10–15, 21–23, 61, 69–70). This becomes more their risks and disadvantages. This also led to the introduction compounded when the aneurysm is large or giant in size and of more complicated endovascular procedures, such as the filling the interpeduncular fossa region, thus leading to limited insertion of multiple stents and the need for anticoagulation, in maneuverability during the clip application process as well as addition to coiling, resulting in more involved and complex limited visualization of the anatomy after the clips were steps with the added morbidity and without proven record of applied. This uncertainty was a main source for the higher their durability (4, 5, 63). morbidity reported when clipping basilar apex aneurysms. The This situation reminds us of when endovascular therapy wide exposure achieved with the pretemporal transcavernous started being used mainly to help overcome the limitations of approach is able to overcome these limitations. The approach microsurgery. We consider using the recent advances in micro- provides ample space to maneuver the clips to the extent that, surgery as a way to overcome the limitations of endovascular in the overall majority of patients, we were able to reconstruct therapy (1, 6, 7, 9, 19, 21, 22, 25, 29, 30, 35, 36, 39, 40, 42, 43, 46, the neck region and prepare it for final clipping with full 360- 53, 55, 56, 68–70). The better approach is to maintain a healthy degree visualization of the anatomy of the neck region. competition with complimentary roles of both treatment In a previous publication, we discussed these advantages at options. This approach will continue to raise the bar of our length (30). In brief, the most important aspects of this expectations and can only benefit our patients. approach are increasing the space and widening the surgical In the long term, the decision to coil an aneurysm that is view at its depth. This provides us with the security of gaining known to have a high rate of recanalization and failure with proximal control before ever getting close to the aneurysm. It endovascular therapy should not be based on the lack of avail- also provides the space to visualize both P1 segments of the ability of the microsurgical option, which is able to provide a posterior cerebral arteries. It helps in visualizing the perfora- superior result. The trend should change; if there is a way to tors. The site of the temporary clip on the basilar trunk is usu- achieve a curable and more durable outcome, then we should ally at a perforator-free zone below the superior cerebellar focus on the training and education of more young and inter- artery. The application of the temporary clip at this location has ested neurosurgeons who are sure to achieve a more superior many advantages including: 1) the clip not obstructing the result than what we can achieve. It is true that such training is view and not being in the way of the neck of the aneurysm; 2) demanding and requires visiting the laboratory frequently and

248 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MICROSURGICAL CLIPPING OF 50 HIGH COMPLEXITY BASILAR APEX ANEURYSMS

visiting other neurosurgeons who are familiar with these 23. Heros RC, Lee SH: The combined pterional/anterior temporal approach for approaches, but this will ensure that our patients will receive aneurysms of the upper basilar complex: Technical report. Neurosurgery the best treatment option available. 33:244–251, 1993. 24. Hillman J, Säveland H, Jakobsson KE, Edner G, Zygmunt S, Fridriksson S, Brandt L: Overall management outcome of ruptured posterior fossa REFERENCES aneurysms. J Neurosurg 85:33–38, 1996. 25. Inao S, Kuchiwaki H, Hirai N, Gonda T, Furuse M: Posterior communicating artery cutting during surgery for basilar tip aneurysm. Acta Neurochir 1. Aziz KM, van Loveren HR, Tew JM Jr, Chicoine MR: The kawase approach to (Wien) 138:853–861, 1996. retrosellar and upper clival basilar aneurysm. Neurosurgery 44:1225–1236, 26. Jennett B, Snoek J, Bond MR, Brooks N: Disability after severe head injury: 1999. Observations on the use of the Glasgow Outcome Scale. J Neurol Neurosurg 2. Batjer HH, Samson DS: Causes of morbidity and mortality from surgery of Psychiatry 44:285–293, 1981. aneurysms of the distal basilar artery. Neurosurgery 25:904–915, 1989. 27. Juvela S, Porras M, Heiskanen O: Natural history of unruptured intracranial 3. Bavinzski G, Killer M, Gruber A, Reinprecht A, Gross CE, Richling B: aneurysms: A long-term follow-up study. J Neurosurg 79:174–182, 1993. Treatment of basilar artery bifurcation aneurysms by using Guglielmi detach- 28. Juvela S, Poras M, Poussa K: Natural history of unruptured intracranial able coils: A 6-year experience. J Neurosurg 90:843–852, 1999. aneurysms: probability of and risk factors for aneurysm rupture. J Neurosurg 4. Chow MM, Woo HH, Masaryk TJ, Rasmussen PA: A novel endovascular treat- 93:379–387, 2000. ment of a wide-necked basilar apex aneurysm by using a Y-configuration, 29. Kawase T, Toya S, Shiobara R, Mine T: Transpetrosal approach for aneurysms double-stent technique. AJNR Am J Neuroradiol 25:509–512, 2004. of the lower basilar artery. J Neurosurg 63:857–861, 1985. 5. Cross DT 3rd, Moran CJ, Derdeyn CP, Mazumdar A, Rivet D, Chicoine MM: 30. Krisht AF, Kadri PA: Surgical clipping of complex basilar apex aneurysms: A Neuroform stent deployment for treatment of a basilar tip aneurysm via a posterior strategy for successful outcome using the pretemporal transzygomatic tran- communicating artery route. AJNR Am J Neuroradiol 26:2578–2581, 2005. scavernous approach. Neurosurgery 56:261–273, 2005. 6. Day JD, Fukushima T, Giannotta S: Cranial base approaches to posterior cir- 31. Lawton MT: Basilar apex aneurysms: Surgical results and perspectives from culation aneurysm J Neurosurg 87:544–554, 1997. an initial experience. Neurosurgery 50:1–8, 2002. 7. Day DJ, Giannotta SL, Fukushima T: Extradural temporopolar approach to 32. Lubicz B, Leclerc X, Gauvrit JY, Lejuene JP, Pruvo JP: Giant vertebrobasilar lesions of the upper basilar artery and infrachiasmatic area. J Neurosurg aneurysms: endovascular treatment and long term follow-up. Neurosurgery 81:230–235, 1994. 8. de Sousa AA: Basilar bifurcation aneurysms. Neurol Med Chir (Tokyo) 55:316–323, 2004. 38:70–73, 1998. 33. Lupret V, Vidovic D, Negovetic L, Novak M: Surgical approach to a large 9. Dolenc VV, Skrap M, Sustersic J, Skrbec M, Morina A: A transcavernous- basilar artery bifurcation and upper basilar trunk aneurysm: Case report. transesellar approach to the basilar tip aneurysms. Br J Neurosurg 1:251–259, Surg Neurol 33:404–406, 1990. 1987. 34. Lusseveld E, Brilstra EH, Nijssen PC, van Rooij WJ, Sluzewski M, Tulleken 10. Drake CG, Peerless SJ, Hernesniemi JA: Surgery of Vertebrobasilar Aneurysms. CA, Wijnalda D, Schellens RL, van der Graaf Y, Rinkel GJ: Endovascular coil- New York, Springer Verlag, 1996. ing versus neurosurgical clipping in patients with a ruptured basilar tip 11. Drake CG: Bleeding aneurysms of the basilar artery: Direct surgical manage- aneurysm. J Neurol Neurosurg Psychiatry 73:591–593, 2002. ment in four cases. J Neurosurg 18:230–238, 1961. 35. MacDonald JD, Antonelli P, Day AL: The anterior subtemporal, medial 12. Drake CG: Surgical treatment of ruptured aneurysms on the basilar artery: transpetrosal approach to the upper basilar artery and ponto-mesencephalic Experience with fourteen cases. J Neurosurg 23:457–473, 1965. junction Neurosurgery 43:84–89, 1998. 13. Drake CG: The surgical treatment of aneurysms of the basilar artery. J 36. Matsuyama T, Shimomura T, Okumura Y, Sakaki T: Mobilization of the inter- Neurosurg 29:436–446, 1968. nal carotid artery for basilar artery aneurysm surgery. Technical note. 14. Drake CG: The surgical treatment of the vertebral-basilar aneurysms. Clin J Neurosurg 86:294–196, 1997. Neurosurg 16:114–169, 1969. 37. McDougall CG, Halbach VV, Dowd CF, Higashida T, Larsen DW, Hieshima 15. Drake CG: The treatment of aneurysms of the posterior circulation. Clin GB: Endovascular treatment of basilar tip aneurysms using electrolytically Neurosurg 26:96–144, 1979. detachable coils. J Neurosurg 84:393–399, 1996. 16. Eskridge JM, Song JK: Endovascular embolization of 150 basilar tip 38. Moller V, Axmann C, Reith W: Clinical course of a partially thrombosed, aneurysms with Guglielmi detachable coils: Results of the Food and Drug symptomatic aneurysm of the basilar artery tip with partial recanalization Administration multicenter trial. J Neurosurg 89:81–86, 1998. subsequent to coiling [in German]. Radiologe 46:417–420, 2006. 17. Friedman JA, Nichols DA, Meyer FB, Pichelman MA, Melver JL, Toussaint 39. Morcos JJ, Heros RC: Distal basilar artery aneurysms: Surgical techniques, in LG 3rd, Axley PL, Brown RD Jr: Guglielmi detachable coil treatment fro rup- Batjer HH (ed): Cerebrovascular Disorders. Philadelphia, Lippincott-Raven, tured saccular cerebral aneurysms: Retrospective review of a 10-year single 1997, pp 1055–1078. center experience. AJNR Am J Neuroradiol 24:526–533, 2003. 40. Neil-Dwyer G, Lang DA, Evans BT: The effects of orbitozygomatic access for 18. Guglielmi G, Viñuela F, Duckweiler G, Dion J, Lylyk P, Berenstein A, Strother ruptured basilar and related aneurysms on management outcome. Surg C, Graves V, Halbach V, Nichols D, Hopkins N, Ferguson R, Sepetka I: Neurol 47:354–359, 1997. Endovascular treatment of posterior circulation aneurysms by electrothrom- 41. Nichols DA, Brown RD Jr, Theilen KR, Meyer FB, Atkinson JL, Piepgras DG: bosis using electrically detachable coils. J Neurosurg 77:515–524, 1992. Endovascular treatment of ruptured posterior circulation aneurysms using 19. Halbach VV, Higashida RT, Dowd CF, Barnwell SL, Fraser KW, Smith TP, electrolytically detachable coils. J Neurosurg 87:374–380, 1997. Teitelbaum GP, Hieshima GB: The efficacy of endosaccular aneurysm occlu- 42. Nukui H, Mitsuka S, Hosaka T, Kakizawa T, Horikoshi T, Miyazawa N, sion in alleviating neurological deficits produced by mass effect. J Neurosurg Yagishita T, Fukamachi A, Shimizu T: Technical points to improve surgical 80:659–666, 1994. results in cases with basilar tip aneurysms. Neurol Med Chir (Tokyo) 20. Henkes H, Fischer S, Mariushi W, Weber W, Liebig T, Miloslavski E, Brew S, [Suppl 38]:74–78, 1998. Kuhne D: Angiographic and clinical results in 316 coil-treated basilar artery 43. Nutik SL: Pterional craniotomy via a transcavernous approach for the treat- bifurcation aneurysms. J Neurosurg 103:990–999, 2005. ment of low-lying distal basilar artery aneurysms. J Neurosurg 89:921–926, 21. Hernesniemi JA, Ishil K, Niemela M, Kivipelto L, Fujiki M, Shen H: 1998. Subtemporal approach to basilar bifurcation aneurysms: Advances technique 44. Ogilvy CS, Hoh BL, Singer RJ, Putnam CM: Clinical and radiographic out- and clinical experience. Acta Neurochir Suppl 94:31–38, 2005. come in the management of posterior circulation aneurysms by use of direct 22. Hernesniemi JA, Vapalahti MP, Niskanen M, Kari A: Management outcome surgical or endovascular techniques. Neurosurgery 51:14–22, 2002. for vertebrobasilar artery aneurysms by early surgery. Neurosurgery 45. Ojemann RG, Crowell RM: Surgical Management of Cerebrovascular Disease. 31:857–862, 1992. Baltimore, Williams & Wilkins, 1988.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 249 KRISHT ET AL.

46. Origitano TC, Anderson DE, Tarassoli Y, Reichman OH, Al-Mefty O: Skull 69. Yas¸argil MG: Microneurosurgery. New York, Thieme-Stratton, 1984, vol. 1. base approaches for complex cerebral aneurysms. Surg Neurol 40:339–346, 70. Yas¸argil MG: Microneurosurgery. New York, Thieme-Stratton, 1984, vol. 2, pp 1993. 232–295. 47. Peerless SJ, Hernesniemi JA, Gutman FB, Drake CG: Early surgery for ruptured posterior circulation aneurysms. J Neurosurg 80:643–649, 1994. 48. Pierot L, Boulin A, Castaings L, Rey A, Moret J: Selective occlusion of basilar COMMENTS artery aneurysms using controlled detachable coils: Report of 35 cases. he results of this series are comparable to the “gold standards” of Neurosurgery 38:948–953, 1996. 49. Raymond J, Roy D, Bowjanowski M, Moumdjian R, L’Esperance G: TYas¸argil (2), the founder of microsurgery, and of Drake et al. (1) in Endovascular treatment of acutely ruptured and unruptured aneurysms of their never-to-be-repeated experience. When dealing with these chal- the basilar bifurcation. J Neurosurg 86:211–219, 1997. lenging lesions, it is wise to refer the patients to high-volume centers 50. Raymond J, Roy D: Safety and efficacy of endovascular treatment of acutely with dedicated neurovascular surgeons who are also knowledgeable rupture aneurysms. Neurosurgery 41:1235–1245, 1997. about cranial base approaches when necessary. 51. Redekop GJ, Duity FA, Woodhurst WB: Management-related morbidity in unselected aneurysm of the upper basilar artery. J Neurosurg 87:836–840, Mika Niemelä 1997. Reza Dashti 52. Rice BJ, Peerless SJ, Drake CG: Surgical treatment of unruptured aneurysms Juha Hernesniemi of the posterior circulation. J Neurosurg 73:165–173, 1990. Helsinki, Finland 53. Sano K: Tempora-polar approach to aneurysms of the basilar artery at and around the distal bifurcation: Technical note. Neurol Res 2:361–367, 1980. 54. Sauvageau E, Raymond J, Roy D, Juravsky L, Guilbert F, Weill A: Postcoiling aneurysm tilting: A disturbing finding. AJNR Am J Neuroradiol 25:601–603, 1. Drake CG, Peerless SJ, Hernesniemi JA: Surgery of Vertebrobasilar Aneurysms: 2004. London, Ontario Experience on 1767 Patients. Vienna, Springer Verlag, 1997. 55. Seoane E, Tedeschi H, de Oliveira EP, Wen HT, Rhoton AL Jr: The pretempo- 2. Yas¸argil MG: Reflections on the thesis “Prospective outcome study of aneurys- ral transcavernous approach to the interpeduncular and prepontine cisterns: mal subarachnoid hemorrhage” of Dr. Timo Koivisto. http://www.uku.fi/ Microsurgical anatomy and technique application. Neurosurgery 46:891–897, tutkimus/vaitokset/2002/isbn951–780–338–9.pdf. Accessed December 2006. 2000. 56. Shiokawa Y, Saito I, Aoki N, Mizutani H: Zygomatic temporopolar approach e read with interest Krisht et al.’s study describing their experi- for basilar artery aneurysms. Neurosurgery 25:793–796, 1989. Wence treating 50 high complexity basilar apex aneurysms via a 57. Solomon RA, Stein BM: Surgical approaches to aneurysms of the vertebral pretemporal transcavernous approach. Although they note the short- and basilar arteries. Neurosurgery 23:203–209, 1988. comings of the pterional approach in this region, we favor the use of 58. Spetzler RF, Hadley MN, Rigamonti D, Carter LP, Raudezens PA, Shedd SA, the orbitozygomatic craniotomy to approach basilar apex aneurysms at Wilkinson E: Aneurysms of the basilar artery treated with circulatory arrest, our institution. In our experience, the additional bone removal of this hypothermia, and barbiturate cerebral protection. J Neurosurg 68:868–879, approach decreases the amount of brain retraction needed and widens 1988. 59. Sugita K, Kobayashi S, Shintani A, Mutsuga N: Microneurosurgery for the corridor to the surgical target (1). Nonetheless, the excellent results aneurysms of the basilar artery. J Neurosurg 51:615–620, 1979. reported here attest to the safety and efficacy of their approach. 60. Tanaka Y, Kobayashi S, Kyoshima K, Gibo H: Factors influencing surgical out- Both microsurgery and endovascular surgery have strengths and lim- come of the basilar bifurcation aneurysms. Neurol Med Chir (Tokyo): itations. The potential exists for these two modalities to complement 38:79–82, 1998. each other, resulting in superior care for patients compared with options 61. Tanaka Y, Kobayashi S, Sugita K Gibo H, Kyoshima K, Nagasaki T: Character- available in the past. The advantages offered by microsurgery in treating istics of pterional routes to basilar bifurcation aneurysm. Neurosurgery 36: difficult lesions such as complex basilar apex aneurysms presupposes the 533–538, 1995. presence of surgeons capable, in both skill and experience, of harnessing 62. Tateshima S, Murayama Y, Gobin YP, Duckwiler GR, Guglielmi G, Viñuela F: those advantages. Thus, the authors highlight the continued importance Endovascular treatment of basilar tip aneurysms using Guglielmi detachable coils: Anatomic and clinical outcomes in 73 patients from a single insti- of training young neurosurgeons in cerebrovascular surgery. tution. Neurosurgery 47:1332–1342, 2000. Francisco Ponce 63. Thorell WE, Chow MM, Woo HH, Masaryk TJ, Rasmussen PA: Y-configured Robert F. Spetzler dual intracranial stent-assisted coil embolization for the treatment of wide- Phoenix, Arizona neck basilar tip aneurysms. Neurosurgery 56:1035–1040, 2005. 64. Vallee JN, Aymard A, Vicaut E, Reis M, Merland JJ: Endovascular treatment of basilar tip aneurysms with Guglielmi detachable coils predictors of immediate and long-term results with multivariable analysis 6-year experience. 1. Gonzalez LF, Crawford NR, Horgan MA, Deshmukh P, Zabramski JM, Spetzler Radiology 226:867–879, 2003. RF: Working area and angle of attack in three cranial base approaches: 65. Wascher TM, Spetzler RF: Saccular aneurysms of the basilar bifurcation, in Pterional, orbitozygomatic, and maxillary extension of the orbitozygomatic Carter LP, Spetzler RF, Hamilton MG, (eds): Neurovascular Surgery. New York, approach. Neurosurgery 50:550–557, 2002. McGraw-Hill, 1995, pp 729–752. 66. Wiebers DO, Whisnant JP, Huston J 3rd, Meissner I, Brown RD Jr, Piepgras DG, Forbes GS, Thielen K, Nichols D, O’Fallon WM, Peacock J, Jaeger L, risht et al. report their surgical experience using the pretemporal Kassell NF, Kongable-Beckman GL, Torner JC; International Study of Ktranszygomatic transcavernous (PTT) approach for 50 complex Unruptured Intracranial Aneurysms Investigators. Unruptured intracranial basilar aneurysms. They have already published several studies on the aneurysms: Natural history, clinical outcome, and risks of surgical and usefulness of this approach in basilar apex aneurysms; in this study, endovascular treatment. Lancet 362:103–110, 2003. 67. Wright CD, Wilson CB: Surgical treatment of basilar aneurysm. Neurosurgery they add many cases, making their series the largest on this issue. This 5:325–333, 1979. is based on their conviction and continuous confirmation that com- 68. Yas¸argil MG, Antic J, Laciga R, Jain KK, Hodosh RM, Smith RD: Micro- plex basilar aneurysms can be adequately treated by clipping using this surgical pterional approach to aneurysm of the basilar bifurcartion. Surg approach if the surgery is performed by experienced surgeons. Their Neurol 6:83–91, 1976. message is that using recent advances in microsurgery is a way to over-

250 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MICROSURGICAL CLIPPING OF 50 HIGH COMPLEXITY BASILAR APEX ANEURYSMS

come the limitations of endovascular therapy. This does not mean that intimate relationship of the aneurysm and vital neurovascular struc- clipping is superior to endovascular treatment in general, but rather tures. Another problem is their rarity; neurosurgeons who are not ded- reconfirms that both modalities are complementary even in treating icated to vascular procedures are exposed to very few cases of posterior very complex vascular lesions. Their message is in sharp contrast to the circulation aneurysms in their careers. Because of these factors, surgi- recent tendency of referring patients with aneurysms that are difficult cal results for aneurysms in this location are usually worse than those to clip for coiling without consideration to avoid risk. This results in for aneurysms in any other location. Complex basilar apex aneurysms surgeon’s losing their confidence and skill in the treatment of such are even more difficult to treat surgically if the complexity is related to complex lesions. We congratulate Krisht et al. for their efforts and suc- a giant and/or partially thrombosed lesion. cessful results of their surgical treatment of such difficult cases. In this study, Krisht et al. report outstanding results in a large series of 50 complex basilar apex aneurysms treated microsurgically using a Nobuo Hashimoto pretemporal trancavernous approach. With this technique, 49 Yasushi Takagi aneurysms (98%) were successfully clipped. The two deaths reported in Kyoto, Japan this series were not related to the procedure. Perforator injuries occurred in only two patients; another patient experienced superior risht et al. present an interesting update of their previous report cerebelar artery ischemia. At the 6-month follow-up examination, the published in April 2005 (2). In this study, they analyzed a series of K Rankin Outcome Scale scores were 0 to 2 in 92% of the patients. 50 patients with high complexity basilar apex aneurysms. As endovascular therapy is increasingly used in the treatment of The term “pretemporal approach” was introduced by de Oliveira most posterior circulation aneurysms, neurosurgeons are treating only et al. (1) in 1995 and its combination with the transcavernous route was the most complex cases. With this increase in the complexity of surgi- published by our group in 2000 (3). The modification published by the cal cases, vascular neurosurgeons must develop excellence in microsur- authors in 2005 and the update in this new report show several impor- gical techniques and meticulous knowledge of microsurgical anatomy tant concepts in modern neurovascular microneurosurgery. and cranial bases approaches. The application of the cranial base approach for vascular lesions is The authors’ results are probably the best found in the literature for the tailoring of the approach to each patient rather than tailoring the complex basilar apex aneurysms and are better than the results in my patient to a particular approach. Modern cranial base techniques are experience in which the morbidity and mortality rate is 20%. These minimally invasive because the amount of bone removal is joined with results are much better than endovascular therapy for basilar apex a minimal brain retraction, and their secondary damage is achieved aneurysms, suggesting that microsurgical clipping performed by high- with a wide exposure of a particular lesion, a basilar apex aneurysm in this case. The technical difficulties of a particular treatment are not an level vascular neurosurgeons would be the first treatment option for argument for choosing an “easier” method of treatment, such as the cases of complex basilar aneurysms. endovascular route in this case. Atos Alves de Sousa As is pointed out by the authors, the training of young neurosur- Belo Horizonte, Brazil geons in these techniques is very demanding and sometimes frustrating. Nevertheless, this report compels us to permanently invest in the risht et al. present a continuation of their work on the microsurgi- education and training of new generations of modern microsurgeons Kcal treatment of complex basilar aneurysms using the PTT who will treat more complex and challenging cases. approach. The authors reported their specific procedure and analyzed Microsurgery and cranial base techniques are evolving every day. the results of 21 patients in 2005 (3). In this study, they present impres- This report is a very good example that superior results in the manage- sive clinical findings and raise important questions regarding treatment of very complex neurovascular pathologies are possible. Our ment patterns for such aneurysms. We think there are several points in patients deserve the effort of vascular neurosurgeons to evolve, this article deserving of commentary. develop new techniques, and improve on the classic lessons learned First, the authors somewhat overstate the novelty of the PTT from the masters in neurovascular surgery. approach. They imply that the use of cranial base techniques to access Jorge M. Mura complex basilar aneurysms is a new concept. In fact, the neurosurgical Evandro de Oliveira literature over the past two decades is full of varied and creative mod- São Paulo, Brazil ifications to the original basilar apex approaches popularized by Drs. Drake and Yas¸argil. The authors’ approach is a modification of Dolenc’s original description of the transcavernous approach (1). This approach 1. de Oliveira E, Tedeschi H, Siqueira MG, Peace DA: The pretemporal approach and its modifications have been, and continue to be, analyzed and to the interpeduncular and petroclival regions. Acta Neurochir (Wien) debated (2, 4, 5). Moreover, the PTT approach is a rather conservative 136:204–211, 1995. cranial base approach when compared with numerous more “radical” 2. Krisht AF, Kadri PA: Surgical clipping of complex basilar apex aneurysms: A descriptions in the literature. The PTT approach involves the drilling of strategy for successful outcome using the pretemporal transzygomatic tran- the upper portion of the zygomatic arch rather than removing it, leaves scavernous approach. Neurosurgery 56 [Suppl 2]:261–273, 2005. the orbital rim in place, and does not necessarily involve posterior cli- 3. Seoane E, Tedeschi H, de Oliveira E, Wen HT, Rhoton AL Jr: The pretemporal transcavernous approach to the interpeduncular and prepontine cisterns: noidectomy. Microsurgical anatomy and technique application. Neurosurgery 46:891–899, The real value of this work is found in its remarkable clinical results. 2000. The aneurysms treated were complex (defined by the authors as being large in size, lying low, having a wide base, or having a dysmorphic asilar apex aneurysms are considered the most difficult aneurysms posterior projecting dome) and the outcomes were very good. There Bto treat surgically. The difficulties approaching these aneurysms were two procedure-related ischemic events, as well as one vasospasm include the deep location in the center of the cranial base, the very event and a cardioembolic event, all of which were clinically mild. tight space in the interpeduncular and prepontine cisterns, and the Functional outcomes, as defined by Rankin Outcome Scale scores, were

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 251 KRISHT ET AL.

good in 92% of the patients. Angiographic outcomes were also excellent aneurysms using the PTT approach. Patient outcomes continue to be and there were no residual or recurrent aneurysms seen in any of the remarkable, leading the authors to advocate this approach. I agree with patients who underwent clipping. There was oculomotor palsy in all the authors that we need basilar aneurysm surgeons who maintain patients, but this resolved in most cases. There were a handful of other proficiency with complex basilar apex aneurysms in the face of increas- complications, including cerebrospinal fluid leak in two patients and a ing endovascular case volumes, diminishing microsurgical case vol- subdural in another. umes, and increasing technical complexity. We seem to be headed to a In their discussion, the authors emphasized the value of microsurgery future of a limited number of centers with sufficient clinical volume to versus endovascular management of these lesions. The decision to use support a small group of basilar aneurysm surgeons who can manage a specific endovascular or microsurgical technique in the management these complex cases. If we fail to preserve the microsurgical option, of high complexity basilar apex aneurysms is an intricate one based on endovascular therapies will become more complex, with increased numerous important factors, including patient condition and age, exact reliance on adjuncts (e.g., stents, biologically modified coils, and novel aneurysm morphology, and the elevation of the basilar apex within the embolic agents) and increased risk. interpeduncular cistern. The recommendation for treatment modality This collection of cases represents the future of basilar aneurysm should not be predicated based on an individual practitioner’s comfort surgery. Our surgical cases will be those that are endovascularly unde- level with one particular therapy, but rather should occur in a cere- sirable, i.e., those that are large or giant in size, posteriorly projecting, brovascular center with high-level expertise in all therapies. The central- low lying relative to the posterior clinoid processes, wide based, and ization of high-end neurovascular services has already occurred. dolichoectatic. Aneurysms with small, narrow necks are increasingly Although most neurosurgical graduates are comfortable treating ante- selected for coiling. However, it is important to remember that patients rior circulation aneurysms, basilar aneurysms do not fall into this cate- with these aneurysms also do exceptionally well with microsurgical gory and should, in our view, be treated at recognized centers. We clipping. Therefore, selecting therapy in young patients with good should also keep in mind that more and more frequently, complex Hunt and Hess grades or unruptured aneurysms should consider the lesions such as the ones described in this study are found to benefit from advantages of microsurgical clipping, including durability. The rush multimodality (endovascular and microvascular) strategies. toward endovascular therapy must be tempered with a careful consideration of recurrence after coiling and the possibility of retreatment, Edward Duckworth which frequently occur at the basilar apex. H. Hunt Batjer As positive as the results in this report are, it is important to remem- Chicago, Illinois ber that cranial base techniques alone do not address the difficulties associated with perforators at the basilar apex. Widened exposures are essential, but, in the end, the management of these perforators deter- 1. Dolenc VV, Skrap M, Sustersic J, Skrbec M, Morina A: A transcavernous-trans- sellar approach to the basilar tip aneurysms. Br J Neurosurg 1:251–259, 1987. mines outcomes. There are countless ways that a perforator can be 2. Figueiredo, EG, Zabramski JM, Deshmukh P, Crawford NR, Pruel MC, Spetzler compromised during one of these procedures. It may adhere to the RF: Anatomical and quantitative description of the transcavernous approach aneurysm base in such a way that it cannot be freed; it may become to interpeduncular and prepontine cisterns. Technical Note. J Neurosurg spastic with the slightest manipulation; it may kink after the clip is 104:957–964, 2006. applied; it may be compressed by the clip shank as it lies next to the 3. Krisht AF, Kadri PA: Surgical clipping of complex basilar apex aneurysms: A perforator; or it may escape visualization on the back side of the strategy for successful outcome using the pretemporal transzygomatic tran- aneurysm. These perforator problems are what limit the microsurgical scavernous approach. Neurosurgery 56 [Suppl 2]:261–273, 2005. management of basilar apex aneurysms because it only takes one per- 4. Seoane E, Tedeschi H, de Oliveira E, Wen HT, Rhoton AL Jr: The preteompo- forator to debilitate a patient. ral transcavernous approach to the interpeducular and prepontine cisterns: Microsurgical anatomy and technique application. Neurosurgery 46:891–899, Michael T. Lawton 2000. San Francisco, California 5. Youssef AS, Abdel Aziz KM, Kim EY, Keller JT, Zuccarello M, van Loveren HR: The carotid-oculomotor window in exposure of upper basilar artery aneurysms: A cadaveric morphometric study. Neurosurgery 54:1181–1189, 2004. 1. Krisht AF, Kadri PA: Surgical clipping of complex basilar apex aneurysms: A his report provides an update on the first half of this clinical series strategy for successful outcome using the pretemporal transzygomatic tran- T(1) and presents excellent results with 50 complex basilar tip scavernous approach. Neurosurgery 56 [Suppl 2]:261–273, 2005.

252 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CLINICAL STUDIES

TREATMENT OF CAVERNOUS SINUS DURAL ARTERIOVENOUS FISTULAE BY EXTERNAL MANUAL CAROTID COMPRESSION

Yutaka Kai, M.D. OBJECTIVE: External manual carotid compression is a noninvasive method to treat cav- Department of Neurosurgery, ernous sinus (CS) dural arteriovenous fistulae (DAVF). We studied a group of patients with Graduate School of Medical Sciences, CS-DAVF to identify what factors made complete resolution of their clinical symptoms Kumamoto University, and closure of the DAVF on magnetic resonance angiography (MRA) by compression Kumamoto, Japan therapy possible. Jun-ichiro Hamada, M.D., Ph.D. METHODS: We treated 23 patients with CS-DAVF without cortical venous drainage or Department of Neurosurgery, a recent decline in visual acuity by compression therapy. All were followed up by mag- Graduate School of Medical Sciences, netic resonance angiography at 1, 3, 6, and 12 months after treatment and the charac- Kanazawa University, Kanazawa, Japan teristics of the imaging findings, their neurological symptoms, and the patterns of symptom improvement were examined. Motohiro Morioka, M.D. RESULTS: In Group A (n ϭ 8), complete resolution was achieved by manual carotid Department of Neurosurgery, compression. In the other 15 patients (Group B), complete resolution was not obtained. Graduate School of Medical Sciences, Group B manifested significantly higher ocular pressure and a significantly longer inter- Kumamoto University, Kumamoto, Japan val between symptom onset and compression treatment. In Group A, venous drainage was via the superior orbital vein with or without involvement of the inferior petrosal Shigetoshi Yano, M.D. sinus. Closure of the CS-DAVF occurred within 4.1 months after the start of treatment. Department of Neurosurgery, In three patients, symptom improvement progressed steadily and gradually. The other Graduate School of Medical Sciences, five patients with complete resolution experienced transient worsening of their symp- Kumamoto University, Kumamoto, Japan toms at 2 to 4 months after the start of treatment and symptom resolution occurred within 4 to 7 months. Jun-ichi Kuratsu, M.D., Ph.D. CONCLUSION: We identified lower ocular pressure, a shorter interval between symp- Department of Neurosurgery, tom onset and compression treatment, and venous drainage solely via the superior orbital Graduate School of Medical Sciences, vein without involvement of the inferior petrosal sinus as the factors in our CS-DAVF Kumamoto University, Kumamoto, Japan patients that made complete resolution by manual carotid compression possible. KEY WORDS: Carotid artery, Cavernous sinus, Dural arteriovenous fistulae, External compression Reprint requests: Yutaka Kai, M.D., Neurosurgery 60:253–258, 2007 DOI: 10.1227/01.NEU.0000249274.49192.3B www.neurosurgery-online.com Department of Neurosurgery, Graduate School of Medical Science, Kumamoto University, Honjo 1-1-1, n cavernous sinus (CS) dural arteriovenous and the closure and healing of these lesions Kumamoto 860-8556, Japan. fistulae (DAVF), dural branches of the inter- remain unclear. Therefore, we studied the Email: [email protected]. kumamoto-u.ac.jp Inal and/or external carotid arteries commu- imaging findings, neurological symptoms, and nicate with the cavernous sinus. Clinical symp- whether or not the patient experienced com- Received, May 2, 2006. toms include conjunctival chemosis, proptosis, plete resolution of their clinical symptoms and Accepted, October 6, 2006. extraocular nerve palsy, retro-orbital pain, and closure of the DAVF on magnetic resonance subjective/objective bruit (7, 11). Although the angiography (MRA) by manual carotid com- prognosis of CS-DAVF is generally favorable pression. (5), some patients with retrograde draining of the CS-DAVF presented with brain swelling or MATERIALS AND METHODS subcortical hematoma (2, 4, 9). Although external manual carotid compres- Between 1996 and 2004, we treated 48 sion is often used to treat patients with CS- patients (35 women and 13 men; age, 48–76 yr) DAVF (6), factors that contribute to its success with CS-DAVF at our department. Baseline

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 253 KAI ET AL.

angiograms were obtained for all 48 patients to document the showed remarkable improvement (Group A). In the latter size and location of the lesions and to document the type of patient, the persistent symptom was slight diplopia. The other venous drainage of the fistulae. All patients also underwent 15 patients did not obtain resolution (Group B). Their symp- ophthalmological examination, including visual acuity, visual toms progressed or failed to improve in the course of 6 months. field, and ocular pressure testing by a neuro-ophthalmologist. Ten of these patients underwent transvenous embolization, two Of the 48 patients, 23 highly compliant and motivated patients underwent transarterial embolization, and three patients without cortical venous drainage or a recent decline in patients underwent combined transarterial-transvenous visual acuity were treated with manual carotid compression embolization. Transvenous and transarterial embolization was therapy. The patients included in this study provided informed achieved with Guglielmi detachable coils (Target Therapeutics, consent and the investigation was approved by the Ethics San Jose, CA), with particulate materials, or liquid coils (Target Committees of Kumamoto University. Therapeutics), respectively. These interventions produced complete resolution in 13 patients; the other two patients showed Manual Carotid Compression improvement. None of the patients subjected to interventional The patients were instructed to compress the carotid artery surgery experienced complications. and jugular vein with the contralateral hand for 10 seconds All eight patients in Group A had chemosis; six manifested several times each hour. Then, using gradually increasing pres- diplopia from ophthalmoplegia, and one had proptosis and sure, they compressed the carotid artery and jugular vein until diminished visual acuity. Their ocular pressure ranged from 16 pulsation was no longer palpable. We required use of the hand to 28 mmHg (mean, 22.5 mmHg). The interval between symp- contralateral to the lesion site so that, if weakness developed tom onset and the inception of manual carotid compression unbeknownst to the patient, the compressing hand would fall ranged from 1.5 to 5.5 months (mean, 3.5 mo). Of the and treatment would automatically stop. The patients per- 15 patients in Group B, 13 (86.7%) had chemosis, 11 (73.3%) formed the external self-treatment while sitting or lying down. had diplopia from ophthalmoplegia, five (33.3%) had propto- The initial sessions were monitored by a physician who sis, and four (26.7%) manifested diminished visual acuity. watched for signs of neurological symptoms and bradycardia. Their ocular pressure ranged from 18 to 32 mmHg (mean, 26.7 If the initial 10-second compression treatments were tolerated mmHg); the interval between symptom onset and the start of without eliciting symptoms, their duration was gradually manual carotid compression ranged from 3.5 to 7.5 months increased to a maximum of several minutes per compression. (mean, 5.8 mo). The two groups did not differ with respect to Patients who reported audible orbital bruit were instructed to their clinical symptoms; however, in Group B, the mean ocu- continue the compression until the bruit ceased. lar pressure was significantly higher (P Ͻ 0.02) and the inter- All 23 patients were followed up after treatment with MRA val between onset and treatment was longer (P Ͻ 0.03) than in studies at 1, 3, 6, and 12 months. Clinical complete resolution Group A (Table 1). was defined as complete resolution of clinical symptoms and Angiographically, the CS-DAVF in Group A was on the right closure of the DAVF on MRA. Patients experienced a complete side in two patients, was on the left side in three patients, and resolution if their symptoms disappeared, no new symptoms was bilateral in three patients. In Group B, four lesions were on developed, and follow-up imaging showed disappearance of the right side, five were on the left side, and six were bilateral. the CS-DAVF. They were considered improved if their symp- We categorized the feeding arteries according to the classification toms improved significantly and MRA showed a decrease in of Barrow et al. (3). Type B arteries were fed only by the external the size of the CS-DAVF. Patients were judged to have incom- carotid artery, Type C arteries were fed only by the dural plete resolution if their symptoms failed to disappear or pro- branches of the internal carotid artery, and Type D arteries were gressed in the course of the 6-month manual carotid compres- fed by both the external and internal carotid arteries. As shown sion therapy. Patients whose ocular pressure increased beyond in Table 1, there was no difference between the two groups with 25 mmHg or those who developed intolerable orbital pain respect to the laterality of the lesions or their feeders. within 6 months after the start of compression therapy under- In our study population, drainage was into the superior went interventional surgery before the end of the treatment orbital vein (SOV), the superior petrosal sinus, the inferior pet- period. In the absence of ocular pressure elevation beyond rosal sinus (IPS), and the contralateral cavernous sinus 25 mmHg and intolerable orbital pain, manual carotid com- through the intercavernous sinus. None of the patients mani- pression therapy was continued for the entire 6-month treat- fested retrograde venous drainage. In Group A, drainage ment period, even in patients whose symptoms did not disap- involved only the SOV (n ϭ 6), the SOV plus the IPS (n ϭ 2), pear or progressed. We used the Mann-Whitney U test to or the SOV plus the contralateral CS (n ϭ 1). In Group B, 12 of determine the statistical significance of differences. Statistical the 15 fistulae drained into the SOV and in six cases, the IPS significance was defined as a P value of less than 0.05. was involved (Table 1). Although there was no significant difference in the drainage route between the two groups, the SOV RESULTS was the sole drainer of significantly more Group A than Group B fistulae (P Ͻ 0.02). Of 23 patients treated by external manual carotid compres- In patients with complete resolution (Group A), closure of sion, seven achieved complete clinical resolution and one the fistula occurred at 2.5 to 7 months (mean, 4.1 mo) after the

254 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MANUAL CAROTID COMPRESSION FOR CAVERNOUS SINUS DURAL ARTERIOVENOUS

TABLE 1. Summary of 23 patients with cavernous sinus dural arteriovenous fistulae who underwent external manual carotid compression therapya Group A Group B Significance No. of patients 815 Symptoms Chemosis 8 (100%) 13 (86.7%) N.S. Ciplopia 6 (75%) 11 (73.3%) N.S. Proptosis 1 (12.5%) 5 (33.3%) N.S. Cisual disturbance 1 (12.5%) 4 (26.7%) N.S. Ocular pressure (mmHg) 16–28 18–32 Mean ocular pressure 22.5 26.7 0.02 (mmHg) Duration between onset 1.5–5.5 3.5–7.5 and treatment (mo) Mean (m) 3.5 5.8 0.03 FIGURE 1. Clinical course of symptom abatement in Group A patients. Angiographical findings Patients 4–8 experienced transient symptom exacerbation before achieving complete resolution by external manual carotid compression. Site Right 2 (25.0%) 4 (26.7%) N.S. Left 3 (37.5%) 5 (33.3%) N.S. Bilateral 3 (37.5%) 6 (40.0%) N.S. AB Artery Barrow Type B 2 (25.0%) 3 (20.0%) N.S. Barrow Type C 1 (12.5%) 2 (13.3%) N.S. Barrow Type D 5 (62.5%) 10 (66.7%) N.S. Drainer SOV 8 (100%) 12 (80.0%) N.S. SPS 0 2 (13.3%) N.S. IPS 2 (25.05) 6 (40.0%) N.S. Con-CS 1 (12.5%) 3 (20.0%) N.S. Multiple draining 3 (37.5%) 8 (53.3%) N.S. system FIGURE 2. A, right external carotid angiogram, anterior-posterior view, SOV only 6 (75%) 4 (26.7%) 0.02 showing shunt flow into the right cavernous sinus. B, right internal carotid angiogram, lateral view, showing a dural arteriovenous fistula at the posterior a N.S., not significant; SOV, superior orbital vein; SPS, superior petrosal cavernous sinus. Drainage is toward the right inferior petrosal sinus. sinus; IPS, inferior petrosal sinus; Con, contralateral; CS, cavernous sinus.

carotid arteries; drainage was via the right IPS (Fig. 2). She was instructed to perform intermittent manual compression of the right start of manual carotid compression therapy. In three patients, cervical carotid artery. MRA performed 3 months after the start of ther- improvement was steady and gradual and required compres- apy demonstrated shrinkage of the CS-DAVF; her double vision was sion therapy ranging from 0.5 to 2.5 months (mean, 1.7 mo). also improved. She continued compression therapy and MRA per- The other ﬁve patients experienced transient symptom exacer- formed 6 months later demonstrated closure of the right CS-DAVF bation before achieving complete resolution and therapy was (Fig. 3). She demonstrated opaciﬁcation of the right CS-DAVF and her required for a mean of 5.6 months (range, 4–7 mo). The tran- double vision disappeared. siently worsening symptoms, consisting of visual disturbance exacerbation, manifested at 2 to 4 months (mean, 3.2 mo) after Patient 6 the start of therapy. Continued compression therapy for 2 to 3 On admission, a 62-year-old woman experienced left orbital pain months (mean, 2.5 mo) led to their resolution (Fig. 1). that had begun suddenly 3 months earlier. Chemosis was absent and she had no visual disturbance. A cerebral angiogram demonstrated a left CS-DAVF fed mainly by multiple small meningeal branches of the bilat- ILLUSTRATIVE CASES eral external and internal carotid arteries. Drainage was via the bilateral IPS and left SOV (Fig. 4). She was instructed to perform intermittent Patient 3 manual compression of the left cervical carotid artery. At 3 months after A 63-year-old woman had a 2-month history of double vision. the start of therapy, she reported gradual improvement of her orbital Cerebral angiography demonstrated a right CS-DAVF fed mainly by pain. However, MRA failed to show disappearance of the lesion (Fig. 5) multiple small meningeal branches of the right external and internal and she developed proptosis and orbital congestion. Angiography

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 255 KAI ET AL.

AB AB

FIGURE 5. A, pretreatment MRA C demonstrating hyperintensity in the left cavernous and inter-cavernous FIGURE 3. A, pretreatment MRA demonstrating hyperintensity in the right sinus suspicious of abnormal shunt CS. We thus suspected abnormal shunt flow via a dural arteriovenous fistula. flow of a dural arteriovenous fistula. B, MRA obtained 6 months after the start of manual carotid compression B, MRA obtained 3 months after the treatment showing disappearance of the shunt flow. start of compression treatment shows resolution of the shunt flow. C, MRA obtained 6 months after the start of A B treatment shows disappearance of the shunt flow.

vations, we recommend external manual carotid compression in CS-DAVF patients without rapidly progressive visual deterioration, neurological symptoms, or other complicating factors. The FIGURE 4. A, left external carotid angiogram, anterior-posterior view, showing shunt flow into the left cavernous sinus. Drainage is toward the (IPS) and results of careful serial the contralateral IPS through the intercavernous sinus. B, left internal carotid follow-up and angiographic angiogram, lateral view, showing a dural arteriovenous fistula at the posterior studies help to determine cavernous sinus. Drainage is toward the left IPS and left (SOV). whether other, more defini- FIGURE 6. Late-phase left internal tive, therapy is necessary. carotid angiogram, lateral view, show- This maneuver is not with- ing no flow in the left IPS. Note resid- demonstrated obstruction of the left IPS and stasis of the contrast out risk, especially in elderly medium in the left SOV. These findings indicated the development of ual medium in the left SOV. The flow patients. Stimulation of the thrombosis in the left CS (Fig. 6). She was instructed to continue man- is slow. ual carotid compression. Three months later, she manifested striking vagus nerve, located in the improvement of her orbital findings. MRA performed 6 months later carotid sheath, may elicit vasovagal reactions, resulting in car- showed closure of her left CS-DAVF. She demonstrated opacification of diac arrhythmias including sinus bradycardia, atrioventricular the left CS-DAVF, and her double vision disappeared. block, systole, and syncope. Elderly patients with significant atherosclerotic disease are also at risk for neurological deficit DISCUSSION and stroke from compromised carotid blood flow and dislodged emboli. Therefore, contraindications for manual carotid Manual external carotid compression, a maneuver that facil- compression are hypertensive carotid sinus syndrome, athero- itates thrombus formation in the CS, is an accepted treatment sclerotic stenosis or ulceration of the carotid artery, and low- for CS-DAVF, except in cases with cortical venous drainage flow or borderline cerebral ischemia, as patients with these and progressive visual decline. Halbach et al. (5) reported that anomalies cannot tolerate the transient occlusion of the ipsilat- among 30 of their patients treated by carotid jugular compres- eral internal carotid artery. sion, seven patients achieved complete resolution and Miki et al. (10) encountered three patients with CS-DAVF 23 patients did not. Higashida et al. (6) found that 15 out of 71 who experienced retinal hemorrhage, probably because of cen- patients subjected to this treatment manifested complete fistula tral retinal vein occlusion, during manual carotid compression closure. None of these patients exhibited clinical or angio- treatment. Komiyama et al. (8) reported transient monocular graphic signs of recurrence 1 year after treatment. In their blindness during manual carotid compression in two patients series, closure occurred from several days to 6 months after with DAVF. All of these patients experienced darkness in the the start of compression therapy. Based on their and our obser- visual field immediately after manual carotid compression.

256 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MANUAL CAROTID COMPRESSION FOR CAVERNOUS SINUS DURAL ARTERIOVENOUS

According to Seeger et al. (11), external compression of the cer- CONCLUSION vical carotid artery and the jugular vein acts synergistically to produce thrombosis of the fistula. Compression of the cervical Some patients with CS-DAVF can achieve complete resolu- carotid artery induces transient collapse of the CS-DAVF, tion by external manual carotid compression therapy. Factors thereby altering the hemodynamic flow to both the CS tributar- suggestive of possible complete resolution are lower ocular ies and the CS itself. Simultaneously, compression of the jugu- pressure, a shorter interval between symptom onset and com- lar vein decreases venous outflow, raising the pressure within pression treatment, and venous drainage solely via the SOV the CS. This may temporarily equalize the pressure between without involvement of the IPS. During the process of the carotid artery inflow to the fistula and the CS outflow and compression-induced closure of the CS-DAVF, some patients result in stasis and thrombosis within the CS or its smaller trib- may experience a transient decline in visual acuity. utaries. Using MRA studies, we were able to follow the progressive development of thrombosis elicited by repeated com- REFERENCES pression of the carotid artery and jugular vein in several of our 1. ApSimon HT, Ives FJ, Khangure MS: Cranial dural arteriovenous malforma- patients. As manual carotid compression can induce thrombus tion and fistula. Radiological diagnosis and management. Review of thirty formation in the CS and occlusion in the central retinal vein four patients. Australas Radiol 37:2–25, 1993. because of intentionally reduced retinal perfusion pressure, it 2. Awad IA, Little JR, Akarawi WP, Ahl J: Intracranial dural arteriovenous mal- also reduces the perfusion pressure in the retina and anterior formations: Factors predisposing to an aggressive neurological course. optic nerve, resulting in a transient visual decline (8). J Neurosurg 72:839–850, 1990. 3. Barrow DL, Spector RH, Braun IF, Landman JA, Tindall SC, Tindall GT: Our search of the literature found few reports describing the Classification and treatment of spontaneous carotid-cavernous sinus fistulas. characteristics of CS-DAVF curable by manual carotid compres- J Neurosurg 62:248–256, 1985. sion. Based on their clinical features and imaging characteristics, 4. Brown RD Jr, Wiebers DO, Nichols DA: Intracranial dural arteriovenous fis- ApSimon et al. (1) divided 34 patients with CS-DAVF into four tulae: Angiographic predictors of intracranial hemorrhage and clinical outcome in nonsurgical patients. J Neurosurg 81:531–538, 1994. groups. The seven patients with anterior cavernous fistulae 5. Halbach VV, Higashida RT, Hieshima GB, Reicher M, Norman D, Newton draining primarily into the SOV were the most amenable to TH: Dural fistulas involving the cavernous sinus: Results of treatment in 30 manual carotid compression as the sole treatment. Complete patients. Radiology 163:437–442, 1987. resolution was achieved in four of these patients. In our group 6. Higashida RT, Hieshima GB, Halbach VV, Bentson JR, Goto K: Closure of of successfully treated patients with CS-DAVF (Group A), the carotid cavernous sinus fistulae by external compression of the carotid artery and jugular vein. Acta Radiol Suppl 369:580–583, 1986. SOV was recognized angiographically as the only venous 7. Houser OW, Baker HL Jr, Rhoton AL Jr, Okazaki H: Intracranial dural arterio- drainage route. While six out of eight (75%) patients in Group A venous malformations. Radiology 105:55–64, 1972. manifested only SOV drainage, four out of 15 (26.7%) patients 8. Komiyama M, Nishikawa M, Yasui T: Transient monocular blindness during in Group B exhibited this drainage pattern. We posit that manual carotid compression for carotid-cavernous sinus fistulas-two case reports. Neurol Med Chir (Tokyo) 36:805–807, 1996. because of the existence of multiple venous drainage systems in 9. Lasjaunias P, Chiu M, terBrugge K, Tolia A, Hurth M, Bernstein M: Group B, manual compression failed to elicit adequate throm- Neurological manifestations of intracranial dural arteriovenous malforma- botic change in the CS, rendering complete resolution by only tions. J Neurosurg 64:724–730, 1986. manual carotid compression impossible. 10. Miki T, Nagai K, Saitoh Y, Onodera Y, Ohta H, Ikoma H: Matas procedure in In Group A, the mean ocular pressure was significantly the treatment of spontaneous carotid cavernous sinus fistula: A complication of retinal hemorrhage [in Japanese]. No Shinkei Geka 16:971–976, 1988. lower and the interval between symptom onset and the start of 11. Seeger JF, Gabrielsen TO, Giannotta SL, Lotz PR: Carotid-cavernous sinus fis- manual compression treatment was significantly shorter than tulas and venous thrombosis. AJNR Am J Neuroradiol 1:141–148, 1980. in Group B. Moreover, among our eight patients who achieved complete resolution by manual compression, five experienced a transient decline in visual acuity. In four of these patients, COMMENTS venous drainage involved the SOV and IPS. We posit that in lthough the literature on therapeutic alternatives for treating these patients, the IPS was thrombosed before occlusion of the Acarotid cavernous fistulae routinely describes manual carotid com- SOV so that this vessel was temporarily the only venous pression as one of the conservative therapeutic options, the literature is drainage route before complete thrombosis of the shunting sparse with regard to the efficacy and potential complications of this point of the CS. Under these conditions, orbital pressure was option. Kai et al. have reported the results of manual carotid compres- elevated but declined gradually after complete thrombosis of sion for cavernous sinus dural arteriovenous fistulae (CS-DAVF) in 23 the CS. Patients who experience exacerbation of visual symp- patients over an 8-year period. During this time, the authors treated 48 toms after the start of manual compression therapy must be patients for dural arteriovenous fistulae of the cavernous sinus. Those monitored closely; this therapy should only be continued in the with cortical venous drainage or recent decline in visual acuity were excluded because of the risks of the natural history of the disease. absence of prolonged progressive worsening. Patients who Twenty-three patients were provided careful instruction on the appro- develop ocular pressure exceeding 25 mmHg or who experi- priate technique for manual carotid and jugular vein compression. The ence unbearable orbital pain require additional treatment, such authors identified eight patients (Group A) whose fistulae resolved as interventional surgery. Our patients with transient symp- within 4.1 months after initiating manual carotid compression. The tom exacerbation during compression therapy achieved com- other 15 patients (Group B) did not respond adequately to treatment plete resolution during the subsequent 2 to 3 month period. and subsequently underwent transvenous and/or transarterial

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 257 KAI ET AL.

embolization. Thirteen out of these 15 were cured using these endovas- already available commercially; although these pulse sequences may cular techniques. still compromise the size of the image area and the temporal resolution, This article illustrates some important points that add to our existing they might provide sufficient enough data to make reliable distinc- knowledge. The authors’ results clearly indicate that a significant per- tions between venous drainage via the SOV and the IPS. Alternatively, centage of patients will respond favorably to external manual carotid parameters for computed tomographic angiography using novel 64- compression with resolution of their fistula by this noninvasive tech- slice computed tomographic scanners could be optimized to accurately nique. The authors discuss the importance of careful instruction to the show the venous drainage. This would mean optimal timing of image patient on the appropriate method for manual carotid compression. acquisition in relation to the contrast media injection. Eventually, the Initial supervision by the treating physician not only assures that the primary diagnostics could be obtained using noninvasive imaging compression is delivered appropriately to both the carotid artery and methods. Based on those findings, only certain patients would undergo jugular vein, but also emphasizes the importance of using the contralat- digital subtraction angiography and endovascular treatment. eral hand for the compression in the event that the patient develops any Marko Kangasniemi unrecognized ischemia, which would affect the compressing hand. Mika Niemelä Despite its noninvasive nature, manual carotid compression is not Juha Hernesniemi without some risks. In addition to the potential for diminishing cere- Helsinki, Finland bral blood flow, carotid compression can result in bradycardia and arrhythmias and retinal hemorrhage. anual carotid artery compression is a simple method to treat Although the series is small, the authors discovered that patients in some CS-DAVFs, but the results associated with this method are Group A were more likely to have venous drainage via the superior M not well described in the literature. Kai et al. present their experience opthalmic vein (SOV) with or without involvement of the inferior petrosal using this treatment in 23 patients. Carotid compression therapy was sinus (IPS) and that patients in Group B had higher ocular pressures and associated with lower ocular pressures, a shorter interval between a longer interval from the onset of symptoms to the initiation of treatment. symptom onset and compression treatment, and venous drainage via These factors may assist clinicians in determining which group of patients the SOV without involvement of the IPS. The authors conclude that is most likely to respond to manual carotid compression. these factors might help select patients for this conservative approach. I would add one note of caution. The authors used the patients’ However, although differences were detected in ocular pressures and symptomatology to determine the clinical success and confirmed that time intervals to treatment in cured patients, these data do not estab- clinical success with magnetic resonance angiography (MRA). An lish criteria for selecting one therapy over another. Therefore, treat- improvement in clinical symptoms associated with dural arteriove- ment decisions remain somewhat subjective. It is important to empha- nous fistulae does not always translate into an improved condition for size that patients with high-risk features such as retrograde cortical the patient. The anatomy of the venous drainage of dural arteriovenous venous drainage, recent decline in visual acuity, or other neurological fistulae dictates the clinical symptoms. An improvement in clinical symptoms were excluded from this study and managed more aggres- symptoms may occur with spontaneous thrombosis of the fistula. On sively with an endovascular approach. In addition, ocular pressures of the other hand, I have personally witnessed patients whose improve- greater than 25 mmHg, unbearable orbital pain, or other neurological ment in clinical symptoms occurred with occlusion of the venous decline during carotid compression therapy may signal a need to drainage into the affected dural sinus with subsequent diversion of switch to a more aggressive intervention. arterialized venous drainage into leptomeningeal veins. Although the symptoms in these patients may resolve completely, the lesion now is Michael T. Lawton associated with a much more aggressive natural history. It is unclear San Francisco, California whether or not magnetic resonance imaging alone is adequate to document obliteration of a dural arteriovenous fistula. We have relied on he authors treated 23 patients with CS-DAVF who did not have cor- post-treatment angiography to document obliteration of the fistula and Ttical venous drainage or a recent decline in visual acuity using determine with certainty that venous drainage has not been converted manual carotid compression therapy. All patients were followed with to more ominous leptomeningeal venous drainage. clinical examinations and MRA for 12 months. Eight patients achieved complete resolution, defined as complete resolution of their clinical Daniel L. Barrow symptoms and MRA evidence of CS-DAVF closure, whereas 15 patients Atlanta, Georgia did not. Complete resolution was statistically more likely in patients who presented with lower ocular pressures, a shorter interval between ai et al. report some suggestive criteria to identify those CS-DAVF symptom onset and compression treatment, and venous drainage patients who might be cured by manual compression without inva- K solely via the SOV without involvement of the IPS. This information sive iatrogenic procedures. One criterion was found to be the venous suggests angiographic and clinical factors predicting the success of drainage solely via the SOV without involvement of the IPS. This cri- manual carotid compression for treating CS-DAVFs and is potentially terion makes new challenges for neurovascular imaging. Although quite useful clinically. However, it is still unclear whether or not this MRA with the conventional time-of-flight pulse sequence often shows study helps in the selection of CS-DAVF patients who will be cured by the shunt and draining vein, it is not a time series and may not always manual compression. Furthermore, MRA evidence of CS-DAVF clo- provide accurate enough data to allow specification of the type of sure does not guarantee an angiographic cure. It will be interesting and venous drainage. Although MRA was used for follow-up in this study, important to obtain longer-term follow-up data on these patients to the primary diagnostics were obtained using digital subtraction ensure that their DAVFs and clinical symptoms do not recur. angiography. However, there are novel imaging methods that could possibly be optimized to noninvasively show the type of venous Gary K. Steinberg drainage in the CS-DAVF. Dynamic MRA using contrast agents is Stanford, California

258 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CLINICAL STUDIES

SUBARACHNOID CLOT VOLUME CORRELATES WITH AGE, NEUROLOGICAL GRADE, AND BLOOD PRESSURE

David S. Rosen, M.D. OBJECTIVE: Cerebral vasospasm after aneurysmal subarachnoid hemorrhage (SAH) is Section of Neurosurgery, associated with the volume and location of subarachnoid blood clots. Factors that inﬂu- Department of Surgery, ence the volume of SAH have seldom been studied. University of Chicago Medical Center and Pritzker School of Medicine, METHODS: Two independent sets of data were analyzed. Data from 3028 patients with Chicago, Illinois SAH enrolled in four clinical trials of the drug tirilazad were analyzed in addition to data from 74 patients with SAH who underwent digital volumetric analysis of admis- Chris Amidei, R.N. sion computed tomographic scans to determine the subarachnoid clot volume. In the Section of Neurosurgery, Department of Surgery, smaller sample of 74 patients, aneurysm width, length, neck size, aspect ratio, and vol- University of Chicago Medical Center ume were measured on diagnostic cerebral angiograms. Statistical inference bearing and Pritzker School of Medicine, on the question of what factors are associated with clot volume was derived by uni- Chicago, Illinois variate methods, including analysis of variance, 2 and t tests, and polytomous logistic regression. Jocelyn Tolentino, B.A. Department of Surgery RESULTS: Of 22 clinical parameters examined by univariate analysis of the tirilazad and Section of Infectious Diseases, dataset, age, World Federation of Neurological Surgeons (WFNS) clinical grade, time Department of Medicine, from SAH to admission, history of hypertension or diabetes mellitus, aneurysm loca- University of Chicago Medical Center and Pritzker School of Medicine, tion, and admission diastolic and systolic blood pressure were correlated with the sub- Chicago, Illinois arachnoid clot volume (P 0.05). Polytomous logistic regression found that only age, WFNS grade, time to admission, admission systolic blood pressure, and history of hyper- Christopher Reilly, B.A. tension were higher in patients with larger subarachnoid clots (P 0.05). Analysis of Section of Neurosurgery, 74 patients with quantitative subarachnoid clot volumes also found that age and WFNS Department of Surgery, University of Chicago Medical Center grade were higher in patients with larger subarachnoid clots (P 0.05). No aneurysm and Pritzker School of Medicine, location or measurement of aneurysm size showed a statistically signiﬁcant relationship Chicago, Illinois to clot volume in either dataset. CONCLUSION: SAH volume is correlated with clinical characteristics, including age, R. Loch Macdonald, M.D., Ph.D. history of hypertension, admission systolic blood pressure, and WFNS grade. Anatomic Section of Neurosurgery, Department of Surgery, aneurysm characteristics such as size and location do not reliably predict clot volume. University of Chicago Medical Center KEY WORDS: Cerebral aneurysm, Hypertension, Subarachnoid hemorrhage, Vasospasm, Volumetric analysis and Pritzker School of Medicine, Chicago, Illinois Neurosurgery 60:259–267, 2007 DOI: 10.1227/01.NEU.0000249271.56816.03 www.neurosurgery-online.com

Reprint requests: R. Loch Macdonald, M.D., Ph.D., Section of Neurosurgery, elayed neurological deterioration from mortality rate among 80 patients with aneurys- Department of Surgery, cerebral arterial vasospasm is a major mal SAH. University of Chicago Medical Center, cause of morbidity and mortality after Despite the clinical significance of blood 5841 South Maryland Avenue, MC3026, D Chicago, IL 60637. aneurysmal subarachnoid hemorrhage (SAH) clot volume, little is known about the clinical Email: [email protected] (3, 4, 7). The volume, location, density, and rate and radiological characteristics that are asso- of clearance of subarachnoid clot correlate with ciated with or that predict the volume of SAH Received, July 11, 2006. the likelihood of developing cerebral vaso- a patient will have. Russell et al. (22) reported Accepted, September 18, 2006. spasm (3, 4, 6, 7, 10, 11, 19, 25). In addition, that smaller aneurysms produce more exten- several studies have demonstrated a link sive subarachnoid clots, although an editorial between the volume of subarachnoid clot and comment on the article noted that the meas- outcome (1, 2, 21). Broderick et al. (1) reported urement of SAH volume was qualitative and that the volume of subarachnoid blood meas- assessment of aneurysm size was not well de- ured on digitized computed tomographic (CT) scribed (17). Thus, the purpose of this study scans was a powerful predictor of a 30-day was to investigate the relationship between

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 259 ROSEN ET AL.

clinical and radiological patient characteristics as well as the Quantitative SAH Volume Dataset volume of the subarachnoid clot. We analyzed two distinct A prospectively collected database of patients with SAH datasets and conducted an exploratory analysis on a large admitted to the University of Chicago Hospitals between database of patients enrolled in four clinical trials of the drug January 2002 and May 2003 was analyzed. Patients were tirilazad (9, 14–16). The results of this analysis were tested on included in this study if they had a CT scan of the brain within a consecutive series of 74 patients with SAH who were treated 24 hours of SAH and a subsequent catheter angiogram show- at our institution and who had digital volumetric analysis of ing a saccular aneurysm. Exclusion criteria included a poor subarachnoid clot volumes. We determined whether or not quality admission CT scan, a CT scan without evidence of similar associations between clinical variables existed in the SAH, an unretrievable CT scan or angiogram, and SAH caused two patient groups. by an arteriovenous malformation, head injury, or another identifiable nonaneurysmal etiology. Details of data collected, MATERIALS AND METHODS patient management, collection of data, and measurement of radiological variables have been published (19). We analyzed digital images of admission CT scans from images that had a Tirilazad Dataset printed distance standard included in them. The slices through the basal cisterns were analyzed to determine the volume of Four prospective, randomized, double-blind, placebo- blood present. Subarachnoid blood clots over the convexities, controlled trials were conducted between 1991 and 1997 in neu- which tended to be absent or minimal in this patient series, rosurgical centers throughout the world to evaluate the effects were not analyzed. SAH was quantified using Adobe of tirilazad on cerebral vasospasm and outcome after aneurys- Photoshop (Version 7.0; Adobe Systems, Inc., San Jose, CA). A mal SAH (9, 14–16). Patients in these studies had SAH from an single examiner (CR) blinded to the clinical data manually angiographically documented saccular aneurysm. They were selected the entire region of interest and subtracted non-blood older than 18 years of age and began treatment within 48 hours hyperdense areas, such as bone. Density was calibrated and of SAH. Exclusion criteria included: 1) non-saccular aneurysms, pixel number and density were measured using an Image 2) severe medical, neurological, or psychiatric illness, 3) serious Processing Tool Kit (AQZ, Version 4.0; Reindeer Graphics, cardiovascular complications, 4) pregnancy or lactation, 5) use Asheville, NC). Microsoft Excel (Version XP; Microsoft, of steroids or calcium channel antagonists other than nimodip- Redmond, WA) was used to make calculations of volume using ine before randomization, and 6) endovascular treatment of the total number of selected pixels multiplied by the millime- the ruptured aneurysm. At the time of enrollment, the patients’ ter per pixel squared and the calculated slice thickness to yield medical history, clinical characteristics, and radiographical a volume. The volumes of all slices were summed. findings were recorded. A single reviewer (CA) who was blinded to the results of the For this analysis, we were interested in characteristics CT analysis measured the size of aneurysms on angiograms present on admission that were related to clot volume. All using an optical micrometer with gradations to 0.1 mm. The patients with complete data (n 3028) were analyzed aneurysm was identified on anteroposterior and lateral images because drug or placebo treatment began after admission of the appropriate angiographic arterial injection and measure- and would not affect clot volume. Variables included age, ments were obtained of the maximum length and width in sex, race, modified Glasgow Coma Scale score (26), time to two orthogonal planes. For width measurements, every admission, and the World Federation of Neurological attempt was made to obtain the width measurements at the Surgeons (WFNS) clinical grade on admission (5). Race was same point along the length of the aneurysm to maximize the recorded as Caucasian, African-American, or other. History accuracy of the volume calculation. The width of the neck was of hypertension, angina or myocardial infarction, diabetes, measured on the angiographic view that best demonstrated it, hepatic disease, thyroid disease, migraine headaches, and which was not necessarily one of the views above. If two previous SAH were recorded. Blood pressure and body tem- views showed widely disparate neck sizes, the mean was perature were documented on admission. Radiographic data taken. All measurements were corrected for variations in included the amount of subarachnoid blood on the initial CT angiographic magnification by using the cavernous internal scan, graded as none, diffuse or localized thin, or diffuse or carotid artery, which was assumed to be 5.9 mm in diameter, localized thick. No central radiological review was reported or the basilar artery, which was assumed to be 4.1 mm in and no quantitative definitions of clot volumes were agreed diameter, as internal standards (13, 23). The aneurysm aspect upon. Intracerebral and intraventricular hemorrhages, as ratio (length:neck) and volume (4 Β length width on well as hydrocephalus, were coded as present or absent. anteroposterior view width on lateral view/24) were calcu- Location of the ruptured aneurysm was determined and the lated using the corrected measurements (27). size of the aneurysm was categorized by the investigators as 12 mm or smaller, 13 to 24 mm, or 25 mm or larger in its Statistical Analysis largest dimension. The Glasgow Outcome Score was determined 3 months after SAH (12). No other data is available On the tirilazad dataset, variables were first checked for nor- from these studies. mality and homoscedasticity using normal probability plots

260 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SUBARACHNOID CLOT VOLUME CORRELATIONS

and jackknife residual plots, respectively. Homoscedasticity is between variables were assessed qualitatively. The best regres- the assumption that the variability in scores for one variable is sion using these criteria was then chosen as the predictive roughly the same as all values of the other variable. Categorical model for initial subarachnoid clot volume. In addition, poly- variables were transformed into dummy or ordinal variables. tomous logistic regression was used to assess the relation The dependent variable was SAH clot volume, which was between SAH volume measured by the Fisher scale and the assumed to increase in the following order: none, localized selected predictor variables because dependent variables were thin, localized thick, diffuse thin, and diffuse thick (the greatest categorical. For univariate models, the regression coefficient, volume). The 74 patients in the tirilazad studies with no clot simple Pearson correlation, and R2 and F test P values are seen on CT scans were used as the reference group in prelimi- reported. For multivariable models, the regression coefficient, nary analyses or were excluded; the 330 patients with local, partial correlation, and partial F test P values were obtained. thin clot were used in that case. The results presented here are Tests were considered significant when the P value was less for the latter analysis, which was not quantitatively different, than 0.05. All statistical analyses were performed using SAS and are more robust because of the larger number of patients Version 8.2 (SAS Institute, Cary, NC) or STATA Version 9 (Stata with local, thin clot. Univariate logistic regression was used to Corp., College Station, TX). calculate correlations between initial SAH clot volume and predictor variables that included: age, WFNS grade, Glasgow Coma Score, preexisting hypertension, diabetes mellitus, num- RESULTS ber of preexisting medical conditions, presence of hyperten- Tirilazad Dataset sion on admission, admission systolic, diastolic and mean blood pressure, ruptured aneurysm size and location, angina, Clinical characteristics of patients included in the explor- myocardial infarction, sex, hepatic disease, thyroid disease, atory analysis of the tirilazad database show that the mean time from SAH to admission, and body weight. Continuous patient age was 52 years and that 83% were women (two of the variables (age, time to admission, and blood pressure) were tirilazad trials enrolled only women, Table 1). A history of analyzed as continuous or ordinal variables by decade, less hypertension was present in 33% of the patients. Approxi- than or equal to 24, or more than 24 hours for admission time. mately 65% of the patients were WFNS Grades 1 or 2 and 70% Intraventricular or intracerebral hemorrhage on the admission had good recovery or moderate disability on the Glasgow CT scan could represent an increase in the volume of hemor- Outcome Scale at 3 months after SAH. The amount of sub- rhage from the aneurysm, but these variables were coded only arachnoid clot was localized thin in 330 (11%) patients, diffuse as present or absent and adjustment was made for the presence thin in 669 (22%) patients, localized thick in 584 (19%) patients, or absence of these variables. Odds ratios and 95% confidence and diffuse thick in 1445 (48%) patients. Intraventricular hem- intervals were calculated for each dependent variable and clot orrhage was present in 48% and intracerebral hematoma pres- volume. These were then examined to determine if they tended ent in 22% of patients. to increase when more clot was present. Univariate analysis was conducted on the 22 dependent vari- For the quantitative dataset, clot volume was a continuous ables listed above. Age, WFNS grade, Glasgow Coma Score, independent variable; dependent variables were age, admis- history of diabetes mellitus, time to admission, admission sys- sion WFNS grade, sex, history of hypertension, Fisher grade tolic and diastolic blood pressure, aneurysm location, and (6), intraventricular hemorrhage, aneurysm location, and history of hypertension correlated with larger clot volumes aneurysm size measurements (length, width in two planes, (P 0.005) (Fig. 1). Aneurysm size had no consistent or signif- mean neck diameter, volume, and aspect ratio). A logarithmic icant association with clot volume (Table 1). For aneurysm loca- transformation was applied when the variable was not nor- tion, internal carotid artery aneurysm location tended to have mally distributed if the transformation showed a considerable less incidence of SAH compared with anterior cerebral artery change to normal distribution. This was only required for aneurysms. The multivariable models constructed from the tir- aneurysm volume. Univariate linear regression was used to ilazad database found that age had a significant effect on clot calculate correlations between initial SAH clot volume and con- volume, although the impact was small (odds ratio 1.02, P tinuous predictor variables. For categorical and ordinal vari- 0.0005) (Table 2). WFNS grade on admission remained signifi- ables, clot volumes were compared using Student’s t tests. cantly associated with larger clot volumes (P 0.0005). Systolic Ordered logistic regression was performed on the tirilazad blood pressure on admission and history of hypertension were dataset using factors significant in the univariate analysis. significant, but the individual effect of admission systolic blood Logistic regression was performed on the quantitative analysis pressure on the model was small (P 0.006). Time to admis- dataset. The full models from each were checked for multi- sion was also significant (P 0.0005). Presence of intraventric- collinearity using the variance inflation factor. In choosing a ular hemorrhage had a significant impact on the multivariable reduced model, simultaneous influences of variables were models and was adjusted for in the analysis (P 0.0005). assessed using a stepwise, backward elimination and forward selection procedure (significance level for variable entry/stay Quantitative Database was set at P 0.10). The models were then evaluated using Clinical characteristics, location of ruptured aneurysms, and Mallow’s Cp statistic to prevent overfitting. Interactions radiological features of this dataset were consistent with the

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 261 ROSEN ET AL.

TABLE 1. Characteristics of 3028 patients with subarachnoid hemorrhage in the tirilazad databasea Characteristic Subarachnoid Clot Size Local Diffuse Local Diffuse thin thin thick thick Total (3028 ؍ n) (1445 ؍ n) (584 ؍ n) (669 ؍ n) (330 ؍ n) (11%) (22%) (19%) (48%) Clinical characteristics Age (yr) 47 13 49 13 52 13 54 13 52 13 Female 274 (83) 553 (83) 494 (85) 1186 (82) 2507 (83) Admission WFNS grade 1 1190 (58) 302 (45) 186 (32) 421 (29) 1099 (36) 2 76 (23) 210 (31) 175 (30) 455 (31) 916 (30) 3 22 (7) 62 (9) 69 (12) 204 (14) 357 (12) 4 21 (6) 40 (6) 57 (10) 175 (12) 293 (10) 5 21 (6) 55 (8) 97 (17) 190 (13) 363 (12) Hypertension on admission 83 (25) 192 (29) 184 (32) 580 (40) 1039 (34) Hypertension 79 (24) 193 (29) 179 (31) 549 (38) 1000 (33) Diabetes mellitus 9 (3) 17 (3) 22 (4) 66 (5) 114 (4) Myocardial infarction 3 (1) 7 (1) 11 (2) 37 (3) 58 (2) Angina 10 (3) 15 (2) 23 (4) 41 (3) 89 (3) Hepatic disease 7 (2) 22 (3) 13 (2) 49 (3) 91 (3) Migraine headaches 95 (29) 169 (25) 133 (23) 343 (24) 740 (24) Thyroid disease 17 (5) 51 (8) 34 (6) 115 (8) 217 (7) Weight (kg) 68 14 69 15 68 15 70 15 70 15 Admission systolic blood pressure 137 21 139 24 141 24 144 26 141 25 Admission diastolic blood pressure 76 14 76 15 76 15 77 16 77 15 Glasgow Outcome Score at 3 mo Good recovery 240 (73) 451 (67) 317 (54) 734 (51) 1742 (58) Moderate disability 38 (12) 85 (13) 91 (16) 191 (13) 405 (13) Severe disability 20 (6) 59 (9) 71 (12) 198 (14) 348 (11) Vegetative state 4 (1) 7 (1) 7 (1) 30 (2) 48 (2) Death 28 (8) 67 (10) 98 (17) 292 (20) 485 (16) Computed tomography ﬁndings Intraventricular hemorrhage 76 (23) 264 (39) 231 (40) 767 (53) 1338 (44) Intracerebral hematoma 66 (20) 94 (14) 201 (34) 315 (22) 676 (22) Aneurysm characteristics Aneurysm size 12 mm 248 (75) 517 (77) 389 (67) 1074 (74) 2228 (74) 13–24 mm 66 (20) 136 (20) 168 (29) 324 (22) 694 (23) 25 mm 16 (5) 16 (2) 27 (5) 47 (3) 106 (4) Aneurysm location Anterior cerebral 92 (28) 271 (41) 173 (30) 580 (40) 1116 (37) Internal carotid 114 (35) 212 (32) 152 (26) 422 (29) 900 (30) Middle cerebral 80 (24) 112 (17) 180 (31) 242 (17) 614 (20) Posterior circulation 44 (13) 74 (11) 79 (14) 201 (14) 398 (13) a WFNS, World Federation of Neurological Surgeons. Values are means standard deviations with percents in parentheses where appropriate. From, Drake CG, Hunt WE, Sano K, Kassell N, Teasdale G, Pertuiset B, DeVilliers JC: Report of World Federation of Neurological Surgeons committee on a universal subarachnoid hemorrhage grading scale. J Neurosurg 68:985–986, 1988 (5). spectrum seen in relatively unselected patients with aneurys- 3 hemorrhage (6). The mean volume of subarachnoid hemor- mal SAH (Table 3). The mean age was 51 years and 56 (76%) rhage clot was 10.5 8.7 ml. patients were women. Thirty-eight (51%) patients had a history Univariate analysis found a correlation between clot vol- of hypertension. Approximately half (53%) of the patients had ume and WFNS grade (P 0.05) (Fig. 2, Table 4). Patients WFNS grades of 1 or 2; the other half were distributed among with higher WFNS grades were more likely to have larger grades 3 to 5. The mean aneurysm length was 1.53 0.78 cm, clots. In addition, a correlation between clot volume and age the mean aspect ratio was 2.15 1.26, and the mean aneurysm (P 0.05) was observed (Fig. 3, Table 4). Older patients were volume was 1.76 3.19 ml. Fifteen (20%) patients had a Fisher more likely to have larger clot volumes. Univariate analysis Grade 2 hemorrhage and 59 (80%) patients had a Fisher Grade did not reveal a correlation between clot volume and

262 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SUBARACHNOID CLOT VOLUME CORRELATIONS

aneurysm location or any measurement of aneurysm size. A history of hypertension was not correlated with clot volume. Multivariate models were constructed next (Table 5). First, a full model including all variables was examined; however, this model did not significantly predict clot volume (P 0.11). A stepwise, backward elimination and forward selection procedure was used to select the most stable and parsimonious model, which included age, WFNS grade, and internal carotid artery aneurysm location (overall reduced model, P 0.01). However, the significance of internal carotid artery aneurysm location was marginal (P 0.06). It was estimated that the sample size of 74 achieved 85% power to detect R2 0.15 attributed to three independent variables using an F test with signifi- FIGURE 1. Adjusted odds ratios for univariate linear models from the exploratory analysis of 3028 patients with SAH. cance of using P 0.05. Odds ratios are adjusted for intracerebral hemorrhage and intraventricular hemorrhage. Values are odds ratios and 95% We next examined whether confidence intervals. or not classification of patients using the Fisher scale predicted clot volume. There were no patients with substantial intracerebral hemorrhage. However, DISCUSSION some patients had intraventricular hemorrhage, which was coded as present or absent. There were no Fisher Grade 1 (no The likelihood of development of cerebral vasospasm after clot seen on CT scan) patients. Because Fisher Grade 4 was aneurysmal SAH is best predicted by the location and volume defined as intraventricular or intracerebral hemorrhage with no of subarachnoid blood clot visualized on CT scans taken or minimal SAH, different methods had to be devised for clas- within a short time after the hemorrhage (3, 4, 6, 7, 10, 11, 19, sifying patients with substantial SAH and intraventricular 25). The density and duration of clot presence are also impor- hemorrhage. Two methods were used: 1) to increase the Fisher tant in predicting vasospasm (19). In addition, the volume of Grade by one category if there was intraventricular hemor- SAH is a determinant of overall outcome after aneurysm rup- rhage, or 2) to classify all patients with intraventricular hemor- ture (1, 2, 21). Given the importance of subarachnoid clot vol- rhage as Grade 4. Stepwise, backward elimination selection ume, it is surprising how little is known about what factors, if analysis showed that, for both methods, WFNS was the only any, determine how much clot is present in the subarachnoid variable that was a significant predictor of SAH volume classi- space after the hemorrhage. The new finding reported here is fied by the Fisher scale. This validates the results by showing that the volume of SAH is associated with clinical characteris- that the same variable (WFNS grade) predicts clot volume. tics, including patient age, neurological grade, and a history of Finally, interactions between variables were examined. hypertension, but not with some characteristics of the rup- Interaction between locations (as dummy variables) did not tured aneurysm, such as location and several measures of size yield valid analyses owing to the limited sample size. Interac- and shape. tions between other variables (aneurysm length, anteroposte- Russell et al. (22) reported that smaller cerebral aneurysms rior or lateral width, aspect ratio, volume, WFNS grade, hyper- produce more SAH. They measured the maximum aneurysm tension, and age) showed that only the interaction between diameter on two-dimensional cerebral angiograms and used a length and the logarithm of volume was significant (P 0.047). qualitative system (10) to grade the amount of subarachnoid The interaction, however, could not be included in the reduced blood on admission CT scans. Meyer (17) commented on the model because aneurysm length was not in the model. methods of this study, noting that the authors used only max-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 263 ROSEN ET AL.

TABLE 2. Multivariable logistic regression model for prediction of TABLE 4. Univariate linear models for predictors of subarachnoid subarachnoid clot volume in the 3028 patients in the tirilazad data- clot volume from the quantitative database of 74 patients with sub- basea arachnoid hemorrhagea 95% P Odds Standard Variables Coefﬁcient R2 rb Variable z P>|z| conﬁdence value ratio error interval Aneurysm location — 0.059 — 0.376 Age 1.02 0.0026 7.90 0.000 1.02–1.03 Anterior cerebral (reference) — — — — WFNS grade 1.24 0.030 8.84 0.000 1.18–1.30 Internal carotid 1.43 0.004 0.063 0.603 History of 1.26 0.084 3.51 0.000 1.11–1.44 Middle cerebral –2.46 0.010 –0.10 0.407 Posterior circulation –4.75 0.031 –0.18 0.142 hypertension Other aneurysm location –2.26 0.006 –0.08 0.530 Time from SAH 0.98 0.004 –3.51 0.000 0.98–0.99 Mean neck diameter –0.28 0.0001 –0.01 0.937 to admission Length –0.77 0.005 –0.07 0.562 Systolic blood 1.004 0.001 2.73 0.006 1.00–1.01 Anteroposterior width –1.98 0.025 –0.16 0.181 pressure Lateral width –1.53 0.011 –0.11 0.374 a WFNS, World Federation of Neurological Surgeons; SAH, subarachnoid Aspect ratio –0.81 0.014 –0.12 0.321 hemorrhage. From, Drake CG, Hunt WE, Sano K, Kassell N, Teasdale G, Pertuiset Log volume –0.51 0.007 –0.08 0.483 B, DeVilliers JC: Report of World Federation of Neurological Surgeons committee WFNS grade 1.34 0.052 0.23 0.050 on a universal subarachnoid hemorrhage grading scale. J Neurosurg 68:985–986, Preexisting hypertension 1.93 0.012 0.11 0.345 1988 (5). Age (yr) 0.14 0.055 0.23 0.044

a WFNS, World Federation of Neurological Surgeons. b Simple Pearson correlation. From, Drake CG, Hunt WE, Sano K, Kassell N, Teasdale G, Pertuiset B, DeVilliers JC: Report of World Federation of Neurological TABLE 3. Characteristics of 74 patients with SAH patients in the Surgeons committee on a universal subarachnoid hemorrhage grading scale. quantitative analysisa J Neurosurg 68:985–986, 1988 (5). Characteristic Value Clinical characteristics Age (yr) 51 15 Female (%) 56 (76) History of hypertension 38 (51) Admission WFNS grade 1 23 (31) 2 16 (22) 3 5 (7) 4 20 (27) 5 10 (14) Computed tomography ﬁndings Fisher Grade 2 15 (20) Fisher Grade 3 59 (80) Intraventricular hemorrhage 21 (28) Aneurysm characteristics FIGURE 2. Positive linear correlation between subarachnoid clot volume and Anterior cerebral artery aneurysm 23 (31) WFNS grade in the quantitative analysis of 74 patients with SAH. Values are means standard errors. Internal carotid artery aneurysm 18 (24) Middle cerebral artery aneurysm 14 (19) Posterior circulation aneurysm 11 (15) Other aneurysm location 8 (11) Neck size (cm) 0.58 0.30 Length (cm) 1.53 0.78 Anteroposterior width (cm) 1.13 0.69 Lateral width (cm) 1.07 0.60 Volume (ml) 1.76 3.19 Aspect ratio 2.15 1.26

a WFNS, World Federation of Neurological Surgeons. Values are means standard deviations with percents in parentheses where appropriate. From, Drake CG, Hunt WE, Sano K, Kassell N, Teasdale G, Pertuiset B, DeVilliers JC: Report of World Federation of Neurological Surgeons committee on a universal subarachnoid hemorrhage grading scale. J Neurosurg 68:985–986, 1988 (5). FIGURE 3. Positive linear correlation between SAH clot volume and age in the conﬁrmatory analysis of 74 patients with SAH.

264 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SUBARACHNOID CLOT VOLUME CORRELATIONS

that they are lower in the elderly, leading to more subarach- TABLE 5. Multivariable linear regression models for prediction of noid bleeding. subarachnoid clot volume from the quantitative database of 74 The second factor associated with SAH volume in both data- patients with subarachnoid hemorrhagea bases was WFNS grade at the time of presentation. This find- Models Coefficient rb P value ing seems logical because more bleeding is likely to cause Full model 0.114 greater initial brain injury. Initial neurological grade is a pow- Aneurysm location 0.263 erful predictor of outcome after SAH (21) and our findings Anterior cerebral (reference) ——— indicate that larger SAH volumes may be one of the primary Internal carotid 2.39 0.10 0.416 causes of early brain injury and poor neurological grade. Middle cerebral –2.52 –0.11 0.393 The difference between blood and intracranial pressure at Posterior circulation –4.75 –0.19 0.129 the time of the hemorrhage may also influence SAH volume. Other aneurysm location –1.22 –0.04 0.743 These values are rarely known at the time of aneurysm rupture, Aspect ratio –0.50 –0.07 0.571 but we did find a significant relation between a history of Log corrected volume –0.17 –0.03 0.842 hypertension and clot volume in the tirilazad database. In our WFNS grade 1.47 0.26 0.035 smaller, quantitative database, this relation was not significant. Preexisting hypertension 0.65 0.04 0.758 Age 0.13 0.23 0.068 In view of a previous study that did not find an association Reduced model 0.010 between preexisting hypertension and clot volume, we hypoth- Internal carotid aneurysm 4.40 0.23 0.058 esize that the higher the cerebral perfusion pressure at WFNS grade 1.55 0.27 0.023 aneurysm rupture, the larger the volume of bleeding will be. A Age 0.13 0.23 0.049 history of hypertension would be weakly associated with this a WFNS, World Federation of Neurological Surgeons. because these patients tend to have higher blood pressures. b Partial correlation. From, Drake CG, Hunt WE, Sano K, Kassell N, Teasdale G, Other factors, such as whether or not the hypertension is Pertuiset B, DeVilliers JC: Report of World Federation of Neurological Surgeons treated, however, probably weaken this association. Finally, committee on a universal subarachnoid hemorrhage grading scale. J Neurosurg 68:985–986, 1988 (5). clot volumes decreased with increasing time after SAH. Including this variable simply documents that the data are valid, as it is obvious that clot clears with time. There are several limitations to the present analyses. First, imal aneurysm diameter as a measure of aneurysm size. relatively crude measures of aneurysm size and morphology Secondly, they measured SAH volume qualitatively by grading were used. Inconsistencies in the magnification of angiographic the amount of blood in each of 10 subarachnoid cisterns from images were corrected by using average values for the diame- 0 to 3 and summing the values (10). The limitations of this ters of the internal carotid and basilar arteries. The mean and grading system were reported to be inflation of the values for standard deviation of the vessel diameters had been defined, hemorrhages that are more widely distributed in the basal cis- but this method of standardization almost certainly introduces terns (22). In addition, this system has high interobserver vari- error into the measurements (8, 13, 23). Aneurysm volume was ability (24, 28). The present analysis avoids these problems by calculated based on two-dimensional angiographic measure- using a quantitative method to calculate clot volumes and by ments and was not directly derived, which would be more using several measurements of aneurysm size, including accurate (18). These measurement factors are likely to lead to width, length, neck size, aspect ratio, and volume. Never- random error and, therefore, are unlikely to change the overall theless, in the large tirilazad and quantitative databases, we results. The more important issue is probably whether or not could not find a relation between aneurysm size and clot vol- finer details of aneurysm morphology influence SAH volume. ume, although the clot volumes in the tirilazad database are A simple measure is the aspect ratio (27). Observational stud- qualitative and aneurysm sizes are graded in three categories. ies suggested that the aspect ratio is higher in ruptured The reason we could not confirm the findings of Russell et al. aneurysms compared with nonruptured aneurysms, although (22) may be because of differences in the methods used. the effect on SAH volume is unknown (29). More complex SAH clot volumes were larger in older patients in both measurements of aneurysm morphology have been performed datasets. Older patients are more likely to have brain atrophy (18) and may affect SAH volume because factors estimated by and larger cerebrospinal fluid cisterns (20). This could lead to these models, such as blood flow, wall tension, and intra- more SAH by at least two mechanisms. First, blood flowing aneurysmal pressure, could be important determinants of the out of the aneurysm will spread to fill the adjacent subarach- SAH volume. Secondly, information on factors that could alter noid space, which may lead to larger volumes. Measurements blood coagulation and, thus, SAH volume were not known in of clot density might allow one to determine if there is more either database, including antiplatelet drug and alcohol use. blood present in older patients or if the same volume of blood Finally, the quantitative analysis included only the volume of is present, but is diluted into a preexisting larger subarachnoid SAH in the basal cisterns and did not measure clot over the space. Secondly, pressure or tension from arachnoid and other convexities or in the ventricles and posterior fossa. In a previ- tissues adjacent to the aneurysm are thought to be important ous quantitative volumetric analysis of SAH volumes, the total in stopping the aneurysm hemorrhage. One could hypothesize volume of cisternal hemorrhage was strongly correlated with

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 265 ROSEN ET AL.

the total volume of hemorrhage (7). Therefore, the total cister- aneurysmal subarachnoid hemorrhage: A cooperative study in Europe, nal hemorrhage volume is probably an adequate proxy for the Australia, and New Zealand. J Neurosurg 84:221–228, 1996. total hemorrhage volume. 15. Lanzino G, Kassell NF: Double-blind, randomized, vehicle-controlled study of high-dose tirilazad mesylate in women with aneurysmal subarachnoid In summary, SAH volume was higher in older patients and hemorrhage. Part II. A cooperative study in North America. J Neurosurg in patients presenting with worse neurological grade. There 90:1018–1024, 1999. were trends for larger SAH to occur in patients with hyperten- 16. Lanzino G, Kassell NF, Dorsch NW, Pasqualin A, Brandt L, Schmiedek P, sion. Further studies are needed to confirm or refute these find- Truskowski LL, Alves WM: Double-blind, randomized, vehicle-controlled study of high-dose tirilazad mesylate in women with aneurysmal subarach- ings and to determine if other clinical characteristics not noid hemorrhage. Part I. A cooperative study in Europe, Australia, New included in our analysis, such as coagulopathy, smoking, or Zealand, and South Africa. J Neurosurg 90:1011–1017, 1999. alcohol use, as well as more detailed assessment of aneurysm 17. Meyer FB: Small aneurysms. J Neurosurg 99:220–221, 2003. morphology, influence SAH volume. 18. Raghavan ML, Ma B, Harbaugh RE: Quantified aneurysm shape and rupture risk. J Neurosurg 102:355–362, 2005. Disclosure 19. Reilly C, Amidei C, Tolentino J, Jahromi BS, Macdonald RL: Clot volume and clearance rate as independent predictors of vasospasm after aneurysmal sub- This study was supported by grants from the National Institutes of Health arachnoid hemorrhage. J Neurosurg 101:255–261, 2004. (NS25946) and the American Heart Association (RLM). R. Loch Macdonald, 20. Resnick SM, Pham DL, Kraut MA, Zonderman AB, Davatzikos C: M.D., Ph.D., received grant support from Boston Scientific and consultant fees Longitudinal magnetic resonance imaging studies of older adults: A shrink- from Actelion Pharmaceuticals. ing brain. J Neurosci 23:3295–3301, 2003. 21. Rosen DS, Macdonald RL: Grading of subarachnoid hemorrhage: REFERENCES Modification of the World Federation of Neurosurgical Societies scale on the basis of data for a large series of patients. Neurosurgery 54:566–575, 2004. 1. Broderick JP, Brott TG, Duldner JE, Tomsick T, Leach A: Initial and recurrent 22. Russell SM, Lin K, Hahn SA, Jafar JJ: Smaller cerebral aneurysms producing bleeding are the major causes of death following subarachnoid hemorrhage. more extensive subarachnoid hemorrhage following rupture: A radiological Stroke 25:1342–1347, 1994. investigation and discussion of theoretical determinants. J Neurosurg 2. Brouwers PJ, Dippel DW, Vermeulen M, Lindsay KW, Hasan D, van Gijn J: 99:248–253, 2003. Amount of blood on computed tomography as an independent predictor 23. Smoker WR, Price MJ, Keyes WD, Corbett JJ, Gentry LR: High-resolution after aneurysm rupture. Stroke 24:809–814, 1993. computed tomography of the basilar artery: 1. Normal size and position. 3. Charpentier C, Audibert G, Guillemin F, Civit T, Ducrocq X, Bracard S, AJNR Am J Neuroradiol 7:55–60, 1986. Hepner H, Picard L, Laxenaire MC: Multivariate analysis of predictors of 24. Svensson E, Starmark JE, Ekholm S, von Essen C, Johansson A: Analysis of cerebral vasospasm occurrence after aneurysmal subarachnoid hemorrhage. interobserver disagreement in the assessment of subarachnoid blood and Stroke 30:1402–1408, 1999. acute hydrocephalus on CT scans. Neurol Res 18:487–494, 1996. 4. Claassen J, Bernardini GL, Kreiter K, Bates J, Du YE, Copeland D, Connolly 25. Takemae T, Mizukami M, Kin H, Kawase T, Araki G: Computed tomography ES, Mayer SA: Effect of cisternal and ventricular blood on risk of delayed of ruptured intracranial aneurysms in acute stage-relationship between cerebral ischemia after subarachnoid hemorrhage: The Fisher scale revisited. vasospasm and high density on CT scan [in Japanese]. No To Shinkei Stroke 32:2012–2020, 2001. 30:861–866, 1978. 5. Drake CG, Hunt WE, Sano K, Kassell N, Teasdale G, Pertuiset B, DeVilliers 26. Teasdale G, Jennett B: Assessment of impaired consciousness and coma: A JC: Report of World Federation of Neurological Surgeons Committee on a practical scale. Lancet 2:81–84, 1974. Universal Subarachnoid Hemorrhage Grading Scale. J Neurosurg 68:985– 27. Ujiie H, Tachibana H, Hiramatsu O, Hazel AL, Matsumoto T, Ogasawara Y, 986, 1988. Nakajima H, Hori T, Takakura K, Kajiya F: Effects of size and shape (aspect 6. Fisher CM, Kistler JP, Davis JM: Relation of cerebral vasospasm to subarach- ratio) on the hemodynamics of saccular aneurysms: A possible index for sur- noid hemorrhage visualized by computerized tomographic scanning. gical treatment of intracranial aneurysms. Neurosurgery 45:119–129, 1999. Neurosurgery 6:1–9, 1980. 28. van der Jagt JM, Hasan D, Bijvoet HW, Pieterman H, Koudstaal PJ, Avezaat 7. Friedman JA, Goerss SJ, Meyer FB, Piepgras DG, Pichelmann MA, McIver JI, CJ: Interobserver variability of cisternal blood on CT after aneurysmal sub- Toussaint LG 3rd, McClelland RL, Nichols DA, Atkinson JL, Wijdicks EF: arachnoid hemorrhage. Neurology 54:2156–2158, 2000. Volumetric quantification of Fisher Grade 3 aneurysmal subarachnoid hem- 29. Weir B, Amidei C, Kongable G, Findlay JM, Kassell NF, Kelly J, Dai L, orrhage: A novel method to predict symptomatic vasospasm on admission Karrison TG: The aspect ratio (dome/neck) of ruptured and unruptured computerized tomography scans. J Neurosurg 97:401–407, 2002. aneurysms. J Neurosurg 99:447–451, 2003. 8. Gabrielsen TO, Greitz T: Normal size of the internal carotid, middle cerebral and anterior cerebral arteries. Acta Radiol Diagn (Stockh) 10:1–10, 1970. 9. Haley EC Jr, Kassell NF, Apperson-Hansen C, Maile MH, Alves WM: A ran- COMMENTS domized, double-blind, vehicle-controlled trial of tirilazad mesylate in n this study, Rosen et al. review four separate clinical trials of tiri- patients with aneurysmal subarachnoid hemorrhage: A cooperative study in North America. J Neurosurg 86:467–474, 1997. Ilazad conducted in the 1990s and involving 3028 subarachnoid hem- 10. Hijdra A, Brouwers PJ, Vermeulen M, van Gijn J: Grading the amount of orrhage (SAH) patients. They examine a smaller sample (74 patients) at blood on computed tomograms after subarachnoid hemorrhage. Stroke their own institution to determine an association between clinical char- 21:1156–1161, 1990. acteristics and subarachnoid clot volume at presentation. The authors 11. Hirashima Y, Kurimoto M, Takaba M, Endo S, Takaku A: The use of com- go through a complex statistical analysis of more than 24 parameters to puted tomography in the prediction of delayed cerebral infarction following arrive at the conclusion that increased age, worsening neurological acute aneurysm surgery for subarachnoid haemorrhage. Acta Neurochir grade, and increased preexisting blood pressure correlate with a larger (Wien) 132:9–13, 1995. clot volume at the time of presentation. The genesis of this study is the 12. Jennett B, Bond M: Assessment of outcome after severe brain damage. A already confirmed observation that larger subarachnoid clot and vol- practical scale. Lancet 1:480–484, 1975. 13. Kamath S: Observations on the length and diameter of vessels forming the ume correlates with poor outcome as well as more delayed vasospasm. circle of Willis. J Anat 133:419–423, 1981. The authors used their statistical analysis to make an argument sug- 14. Kassell NF, Haley EC Jr, Apperson-Hansen C, Alves WM: Randomized, gesting that clinical characteristics define clot size and that clot size double-blind, vehicle-controlled trial of tirilazad mesylate in patients with defines outcomes. Logically, this relationship may be parallel and clin-

266 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SUBARACHNOID CLOT VOLUME CORRELATIONS

ical characteristics that define outcome may do so unrelated to sub- admission blood pressure, World Federation of Neurological Societies arachnoid clot size. Indeed, outcome may also be determined by other grade, and time to admission as significant. Age and admission World independent parameters. Federation of Neurological Societies grade also seemed significant in The analysis in this study is careful, the data set is quite large, and the smaller series of 74 patients. Although it seems that clots would be the conclusions are quite reasonable. The authors speculate that with larger in brains with larger subarachnoid spaces because of age-related increased age, atrophy may allow clots to expand to a greater extent. volume loss and that more intense bleeding (i.e., higher grade) would An increase in neurological grade has already been established as also result in a greater volume of blood, these associations have hereto- determining outcome, suggesting that grade and clot size are somehow fore not been so nicely described. related in this process. Blood pressure is a weaker correlation, but E. Sander Connolly, Jr. allows us to speculate regarding perfusion pressure and transmural New York, New York pressure in the processes that cause the initial leak and later seal it, allowing the patient to survive to be analyzed. he authors have evaluated the factors that affect the volume of The issues and limitations to this study are the relatively subjective SAH after an aneurysmal bleed. As the authors mention, this is an measurement of clot volume in the 3000 patients in the tirilazad studies T important topic because clot volume has clearly been related to patient as opposed to more objective computed tomographic measurements. outcome. First, patients with a larger SAH volume have a worse out- The limitations of the computed tomographic measurements include no come as a result of being a worse grade. Secondly, patients with a analysis of convexity blood or intraventricular or intraparenchymal larger SAH volume are at increased risk for vasospasm which predis- blood. The same clot could be classified as larger in one manner of poses the patients to strokes. Using two databases, the authors’ results thinking and lesser in another because blood was diverted to another have demonstrated that the patient’s age, neurological grade, and intracranial structure. Generally, these patients would be expected to do blood pressure are associated with SAH volume. In regards to hyper- worse clinically. In general, one is always concerned with the fidelity of tension and age, the authors’ results are somewhat expected because the initial data measurement, even in large studies. The resulting conclu- hypertension, smoking, and alcohol use are known risk factors for SAH sions may be less clear. Indeed, this is why the authors follow their (1, 2), and age is probably related to hypertension, which is more com- conclusions in a smaller dataset that is more rigorously measured. An mon in the elderly population. Interestingly, although hypertension additional concern with these studies is the nonrepresentative nature of has been better controlled recently, no decrease in the incidence of SAH sex because two of the tirilazad studies include only women. These has been found in the population. This does raise questions in this results may have a sex relationship that cannot be determined in this study and others regarding findings related to hypertension (3). study. The authors’ conclusions are sensible and, although they do not Despite this, this study has provided further insight into the causes of point to new treatment paradigms other than the avoidance of preexist- the poor outcome associated with aneurysmal SAH. ing hypertension in the general population, they do provide some insight into the pathophysiologies that affect poor outcome. The authors Gavin W. Britz would be wise to more carefully explain the limitations of their dataset Seattle, Washington and the implications of these results. Robert J. Dempsey Madison, Wisconsin 1. Taylor CL, Yuan Z, Selman WR, Ratcheson RA, Rimm AA: Cerebral arterial aneurysm formation and rupture in 20,767 elderly patients: Hypertension and other risk factors. J Neurosurg 83:812–819, 1995. osen et al. analyzed 3028 patients enrolled in four clinical trials of 2. Teunissen LL, Rinkel GJ, Algra A, van Gijn J: Risk factors for subarachnoid tirilizad and 74 patients undergoing volumetric computed tomo- R hemorrhage: A systematic review. Stroke 27:544–549, 1996. graphic scans of subarachnoid clot volume to determine which clinical 3. Broderick JP, Phillips SJ, Whisnant JP, Whisnant JP, O’Fallon WM, Bergstralh variable predicted clot volume. Of the 22 variables studied in the 3028 EJ: Incidence rates of stroke in the eighties: The end of the decline in stroke? patients, multivariate analysis yielded increased age, hypertension, Stroke 20:577–582, 1989.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 267 CLINICAL STUDIES

ROLE OF RADIOSURGERY IN THE MANAGEMENT OF CEREBRAL ARTERIOVENOUS MALFORMATIONS IN THE PEDIATRIC AGE GROUP: DATA FROM A 100-PATIENT SERIES

Nicolas Reyns, M.D. OBJECTIVE: To assess the safety and efficacy of radiosurgery for the management of Department of Neurosurgery, arteriovenous malformations (AVMs) in the pediatric age group. University Hospital, Lille, France METHODS: We reviewed data from 100 children (44 girls and 56 boys) presenting a total of 103 AVMs treated by linear accelerator radiosurgery between December 1988 and Serge Blond, M.D. May 2002. The median patient age was 12 years (range, 2–16 yr). Sixty-seven AVMs Department of Neurosurgery, (65%) were in functional locations and 30% were inoperable. The mean AVM volume University Hospital, was 2.8 cm3 (range, 0.9–21.3 cm3). The mean marginal dose was 23 Gy (range, 15–25 Lille, France Gy) and required between one and four isocenters. Fifty patients received multimodal treatments with embolization and/or surgery before and/or after radiosurgery. Given Jean-Yves Gauvrit, M.D. that 16 patients underwent two sessions of radiosurgery and one patient received three Department of Neuroradiology, University Hospital, sessions, a total of 119 radiosurgical treatments were delivered. We maintained our Lille, France clinical and angiographic follow-up for at least 36 months after irradiation or until the complete obliteration of the AVM was conﬁrmed by angiography (our sole end point for Gustavo Touzet, M.D. judging clinical efﬁcacy). Univariate and multivariate analysis were performed to deter- Department of Neurosurgery, mine predictive factors for obliteration. University Hospital, Lille, France RESULTS: Complete obliteration was achieved for 72 AVMs (70%). The permanent neurological deficit rate was 5%. One patient died because of rebleeding. None of our Bernard Coche, M.D. patients presented bleeding after an angiographically verified AVM obliteration. The Department of Radiation Oncology, main predictive factors for obliteration were low AVM volume and no previous Centre Oscar Lambret, embolization. Moreover, the younger the patient, the more effective the radiosurgery Lille, France seemed to be. Jean-Pierre Pruvo, M.D. CONCLUSION: Radiosurgery is a safe and effective treatment for AVMs in the pedi- Department of Neuroradiology, atric age group. One criterion for success was the use of a prescription dose similar to University Hospital, that used with adult populations. Lille, France KEY WORDS: Arteriovenous malformations, Children, Linear accelerator radiosurgery, Pediatric, Patrick Dhellemmes, M.D. Radiosurgery Department of Neurosurgery, Neurosurgery 60:268–276, 2007 DOI: 10.1227/01.NEU.0000249277.72063.BD www.neurosurgery-online.com University Hospital, Lille, France

Reprint requests: Nicolas Reyns, M.D., n view of the high hemorrhage rate associ- ding to a grim prognosis (15). Hence, a 13-year- Department of Neurosurgery, ated with their natural history, pediatric old child with a newly detected AVM has a 33% University Hospital, cerebral arteriovenous malformations cumulative risk of hemorrhage in the next University of Lille, I (AVMs) require curative treatment (18, 35). 20 years and a 55% risk in 40 years (24). Of Avenue Emile Laine, Indeed, the risk of hemorrhage for untreated course, the younger the patient, the more sub- 59037 Lille Cedex, France. E-mail: [email protected] cerebral AVMs has been estimated at 2 to 4% stantial the cumulative risk of hemorrhage, per year, with an attendant mortality rate of 5 thus, increasing the potential beneﬁts of early Received, August 16, 2005. to 10% and a 50% risk of serious neurological intervention in children (43). At present, micro- Accepted, September 18, 2006. morbidity associated with each hemorrhage surgery is still the “gold standard” treatment (15, 48). Moreover, it seems that rebleeding may for accessible, low-grade AVMs, providing an occur in almost 30% of these cases, correspon- immediate cure with generally accepted risks

268 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com RADIOSURGERY FOR PEDIATRIC ARTERIOVENOUS MALFORMATIONS

(21, 23, 43). Moreover, immediate surgical management is indicated for patients who present with intracerebral hemorrhage TABLE 1. Additional treatment modalities in 50 patientsa and associated progressive neurological deficits. It is only in Embolization Embolization the past decade that the role of radiosurgery for treating AVMs before before and No in children has been more precisely defined; previously, there RS after RS embolization was little information available regarding the use of this tech- No surgery 38 2 nique in a pediatric population (21). However, encouraging Surgery before RS 2 6 results in carefully selected patients have been reported in the Surgery before and 1 literature during the past few years (1, 27, 29, 32, 33, 43, 45, 46). after RS Nowadays, there is no real reason to avoid radiosurgery as an Surgery after RS 1 initial treatment for an inoperable nidus or after partial embolization or partial surgical resection. Indeed, if the treat- a RS, radiosurgery. ment goal is to preserve life and neurological function while curing the AVM and preserving normal cerebral circulation (21), the potential use of radiosurgery should be considered for clin- TABLE 2. Clinical summary of 100 children with arteriovenous a(103 ؍ ically stable patients who do not require emergency treatment. malformations (n Initial symptom No. of AVMs PATIENTS AND METHODS Hemorrhage 69 (67%) Seizures 10 (9.7%) Patient Selection Headaches 11 (10.7%) Between 1988 and 2002, 1050 patients presenting with cere- Asymptomatic 8 (7.7%) bral AVMs were treated by linear accelerator radiosurgery in Others 5 (4.9%) our institution. Of these, 100 individuals (9.5%) were younger a AVMs, arteriovenous malformations. than 16 years at the time of initial treatment (44 girls and 56 boys). All of these patients were included in the presently reported series. At least 36 months of clinical and imaging location and angioarchitecture, afferent vessels and draining follow-up data were available for this population, except when veins, and for grading all AVMs according to the Spetzler-Martin earlier obliteration was observed. Twenty-six patients received system. AVM locations are presented in Table 3. AVMs were clas- first-line care in our institution, with a multidisciplinary team sified as Spetzler-Martin Grade I (9.7%), Grade II (34%), Grade III of neurosurgeons and interventional radiologists deciding (22, 3%), Grade IV (4, 9%), Grade V (0%), or Grade VI (29.1% whether or not to apply previous embolization, surgery, inoperable, located in the brainstem and basal ganglia) (Table 4). and/or radiosurgery as the therapeutic approach. Seventy-four In 67 patients, the AVM was in a functional location. patients were referred to us for radiosurgical treatment by Radiosurgical Technique physicians in other institutions (either as the sole treatment or after partial surgical resection or previous embolization). In Radiosurgical Device fact, 50 patients were treated with multimodal strategies All radiosurgical procedures in this study were performed (Table 1). However, all results in this series relate only to radio- according to Betti’s technique (4), with a Saturne 18-MeV linear surgery performed on 103 residual nidi because the four accelerator (CGR, Paris, France) fitted with a stereotactic add- patients who were treated by embolization and or surgery after on device for radiosurgery and a choice of additional collima- radiosurgery were considered treatment failure vis-à-vis radio- tors (8, 16, 20 mm) for delivering x-ray minibeams. The move- surgery. The median age at the time of treatment was 12 years ments of the stereotactic frame relative to those of the (range, 2–16 yrs). The initial symptom was intracranial hemor- accelerator create a spherical sector, which represents the sur- rhage in 69 out of 100 patients—a far higher proportion of face of possible beam entry portals, extending 160 degrees in spontaneous hemorrhagic modes than in otherwise comparable the coronal plane and 120 degrees in the sagittal plane. Patients adult populations. Ten children presented with seizures, 11 pre- were seated in a Betti armchair, with the head maintained in a sented with headache, five presented with various other symp- Talairach frame (4). toms (cardiac failure in two patients due to thalamic AVMs with large draining vein of Galen, Rendu-Osler disease in two Treatment Procedure patients, and macrocrania in one patient), and five were asymp- Attachment of the Talairach stereotactic frame and stereotactic tomatic (two of the latter presented several AVMs in cases of teleangiography were performed with the patient under general Rendu-Osler disease) (Table 2). anesthesia. Four transcranial, smooth-tipped screws were used to attach the frame. Once the frame had been correctly positioned Characteristics of the AVMs and adjusted, it was removed and the patient was discharged For the purpose of this series, all pretreatment angiograms until the day of the radiosurgical operation. On readmission, the were reviewed by a radiologist (JYG) for determining nidus frame was repositioned (with the same stereotactic coordinates)

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 269 REYNS ET AL.

was recorded) via clinical evaluation, MRI scans, and annual TABLE 3. Anatomic location of arteriovenous malformations angiography. MRI scanning was performed at 6 months after a irradiation and annually thereafter. Referring physicians were(103 ؍ n) Location No. of AVMs also contacted by mail to optimize patient follow-up. The sole efficacy end point for complete obliteration was the Frontal lobe 16 (15.5%) angiographic result according to Linquist and Steiner’s criteria Parietal lobe 12 (11.65%) (normal circulation, no nidus, and no draining vein) (44). Occipital lobe 8 (7.8%) Temporal lobe 11 (10.7%) Statistical Analysis Rolandic 15 (14.6%) Univariate analysis of the factors affecting nidus obliteration Thalamus/basal ganglia 18/7 (17.5/7%) was performed. For categorical variables, a χ2 test or a Fischer Intraventricular 1 (0.9%) exact test was used, depending on the number of variables per Corpus callosum 2 (1.9%) class. For quantitative variables, a nonparametric Mann-Whitney Peduncular and pons 5 (4.9%) test or a t test was performed, again depending on the number Cerebellum 8 (7.8%) of variables per class. We studied the following parameters: age, Functional location 67 (65%) sex, nidus location and depth, nidus characteristics (compact, a AVMs, arteriovenous malformations. plexiform, and presence or absence of fistulae), previous embolization, afferent arterial vessels, type of venous drainage, size and volume of the marginal isodose, and dose prescribed to marginal isodose. A similar statistical methodology was applied TABLE 4. Correlation between Spetzler-Martin Grade and to factors that could have been correlated with radiosurgery side obliteration ratea effects. Multivariate analysis was performed using logistic Grade No. of AVMs Obliteration rate regression. The statistical significance threshold was set at 0.05. I 10 (9.7%) 7 (70%) II 35 (33.9%) 33 (94%) RESULTS III 23 (22.4%) 15 (65%) IV 5 (4.9%) 2 (40%) Neuroimaging Follow-up and Obliteration Rate V0 VI 30 (29.1%) 14 (46.5%) The median time from first-stage radiosurgery to the last imaging follow-up examination was 26 months (range, 11–126 mo). a AVMs, arteriovenous malformations. Sixty-four patients with 67 AVMs (out of 100 patients with 103 AVMs) were cured (i.e., 65%), with a mean time to obliteration of with the patient under general anesthesia, and the patient was 25.5 months. Sixteen patients underwent second-stage radio- then transferred to the irradiation site. The stereotactic frame was surgery, with five (31%) of these exhibiting complete obliteration removed immediately after the radiosurgical procedure and the (mean time delay, 18.4 mo after the second intervention) at the patient was usually discharged on the following day. end of a median follow-up period of 25 months (range, 12–51 mo). Two children underwent a third session of radiosurgical Target Delineation and Treatment Planning treatment; one patient was lost to follow-up and the other was All targets were delineated by reference to biplanar anteropos- recently treated in our institution using gamma knife radio- terior and lateral teleangiographic data. The goal of the targeting surgery. Obviously, this procedure was not included in the pres- was to match the marginal isodose (70%) to the nidus (while ent study. sparing afferent vessels and draining veins), thus obtaining con- At the time of the last follow-up examination, 69 patients formal coverage of the entire AVM nidus and the proximal sec- with 72 AVMs exhibited angiographically confirmed, complete tions of the draining veins. Dynamic angiographic imaging pro- AVM obliteration, corresponding to an overall obliteration rate vided us with identification on the nidus borders. The exact of 70% (Table 5). location of the AVM and its relationships with functional struc- Obliteration rates as a function of target volume are pre- tures were documented using magnetic resonance imaging sented in Table 6. In Group 1 (AVM volume Յ1 cm3), 30 (85.7%) (MRI) scanning. Most targets required a single isocenter treat- out of the 35 AVMs were obliterated. In Group 2 (AVM volume ment procedure. The dose delivered to the marginal isodose was Ͼ1 to Յ3 cm3), 25 (69.5%) out of the 36 AVMs were obliterated. adjusted according to the volume enclosed by the prescription In Group 3 (AVM volume Ͼ3 to Յ10 cm3) 17 (61%) out of the isodose and the distance to any nearby functional structures. 28 AVMs were obliterated. Lastly, in Group 4 (AVM volume The mean marginal dose was 23 Gy (range, 15–25 Gy). Ͼ10 cm3), none of the four AVMs was cured (0%). Obliteration rates as a function of Spetzler-Martin grade are Follow-up Period presented in Table 4. Angiographically documented oblitera- All patients underwent at least 36 months of follow-up after tion was observed for seven (70%) out of the 10 Grade I AVMs, radiosurgery (unless earlier, definitive evidence of obliteration 33 (94%) out of the 35 Grade II AVMs, 15 (65%) out of the 23

270 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com RADIOSURGERY FOR PEDIATRIC ARTERIOVENOUS MALFORMATIONS

grade (P ϭ 0.001), on the other. Obliteration was not correlated TABLE 5. Overall obliteration rate after ﬁrst and second stage of with the number of isocenters (P ϭ 0.3), sex (P ϭ 0.19), type of radiosurgerya venous drainage (P ϭ 0.34), nidus characteristics (P ϭ 0.23), ϭ No. of AVMs Obliteration rate nidus deepness (P 0.748), or dose to marginal isodose (P ϭ 0.77). Postradiosurgery morbidity was correlated with the First stage 103 67 (65%) male sex (P ϭ 0.009), left hemisphere location (P ϭ 0.04), and Second stage 16 5 (31%) nidus volume (P ϭ 0.014). Morbidity was not correlated with Total 103 72 (70%) either AVM location or previous radiosurgery. a AVMs, arteriovenous malformations. A multivariate analysis did not reveal any signiﬁcant associations between parameter combinations and the obliteration rate.

TABLE 6. Correlation between size and obliteration ratea DISCUSSION Volume (cm3) No. of AVMs Obliteration rate Multimodal Strategies Ͻ1 35 30 (85.7%) The choice between microsurgery, radiosurgery, or emboliza- 1–3 36 25 (69.5%) tion as the first-line treatment for each child treated initially in 3–10 28 17 (61%) our institution was based on a multidisciplinary approach. In Ͼ10 4 0 view of our experience and that of other groups (20), it seems a AVMs, arteriovenous malformations. that pediatric AVM management strategies often require a combination of several techniques, as has been observed in adults (9, 10). Hence, 50 patients in our series were treated using a combi- Grade III AVMs, two (40%) out of the four Grade IV AVMs, and nation of surgery and embolization before and/or after radio- 14 (46.5%) out of the 30 Grade VI AVMs. Of the 67 AVMs in surgery (Table 1). Surgery was attempted as the definitive treat- functional locations, 45 were obliterated (67%). ment whenever the size and location of the AVM enabled surgical resection under safe conditions. Radiosurgery was Functional Outcomes and Complications selected as the first-line treatment whenever the nidus was Ninety-five patients returned to their previous activity level deeply located or close to eloquent areas or in the event of a after radiosurgery. There were no treatment-related deaths. residual nidus after primary surgical resection and/or emboliza- Seven patients experienced complications with a time lag after tion. A cure is seldom obtained with embolization alone (14, 25, surgery (Table 7). Six of these exhibited a permanent neurolog- 32, 49), even though some authors consider that the hemody- ical deficit at the end of the follow-up period: two cases of namic modifications after embolization and the obliteration of slight motor palsy (a Grade II rolandic AVM and a Grade VI dangerous portions of the AVM are sufficient to protect patients thalamic AVM), one case of monocular blindness (a frontal against bleeding (25). In our population, embolization was reg- Grade III AVM close to the optic nerve), one case of hemianop- ularly performed 1) to permanently decrease the AVM volume sia (a Grade III temporomesial AVM), and two cases of diplopia and flow and 2) to allow more effective radiosurgery (17, 31, 34, (a Grade VI thalamic AVM and a Grade VI tectal plate AVM). 47); with smaller AVMs, a higher and more effective radiation Two patients experienced radiation-induced epilepsy, one of dose can be administered with a lower risk of side-effects. whom (a patient with a Grade III parietal lobe AVM) resisted However, in most instances, embolization achieves a reduction complete medical control. Hence, in view of the total number of in flow but not size. Nevertheless, we think flow reduction is radiosurgical procedures (n=119), the overall morbidity rate useful for boosting the efficiency of radiosurgery (5) even when was 6.7% (eight out of 119) and the permanent neurological the radiosurgical target volume remains unchanged, but we do deficit rate was 5% (six out of 119). Two patients (both with not have sufficient data to prove this assertion. basal ganglia AVMs) experienced delayed intracranial hemorrhaging. One died as a result (16 mo after radiosurgery) and the Follow-up and Obliteration Criteria other experienced repeated intracranial hemorrhaging in the Because of its delayed efficiency, radiosurgery requires as absence of neurological deficits during the 30-month postrad- rigorous a follow-up as possible. Moreover, in view of a child’s iosurgery latency interval. None of our patients with angio- long life expectancy, this follow-up should continue for as long graphically defined AVM obliteration experienced any bleeding as possible because of potential AVM regrowth, even in cases of whatsoever. angiographically documented obliteration (28, 42). Indeed, several authors involved in pediatric surgery have described Statistical Analysis recurrent AVM rebleeding after negative postoperative arteri- Univariate analysis demonstrated a correlation between the ography (3, 21, 22, 51). It is postulated that AVM regrowth may obliteration rate on one hand, and nidus volume (P ϭ 0.001) occur in children and could be a consequence of their relatively and size (P ϭ 0.001), no previous embolization (P ϭ0.001), age immature cerebral vasculature (21, 41). This potential for younger than 10 years (P ϭ 0.05), and low Spetzler-Martin regrowth seems to persist even after angiographically proven

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 271 REYNS ET AL.

TABLE 7. Review of literature on six major and recent pediatric series regarding radiosurgery for treatment of arteriovenous malformationsa OR, % (no.) No. of Median/ Median/ Overall Median/ Second Bleeding Permanent Series patients mean mean Ͻ1 cm3 mean stage after RS, sequelae, (ref. no.) with volume marginal 1–3 cm3 age radiosurgery no. (%) no. (%) follow-up (cm3) dose (Gy) 3–10 cm3 Ͼ10 cm3

Tanaka et 11.5c al., 1996 23 4.8c 20.5c 0 87 (20/23) 0 0 (2–15) (46) LGK

Shin et 15b al., 2002 82 1.8b 20a 0 86.6 (71/82) 4 (4) 2 (2) (4–19) (42) LGK

73.5 (39/53) Levy et al., na 12b 10 2000 (27) 53 1.7b 20b 80 4 (7.5) 1 (2) (2–17) (18.9%) LGK 64.7 0

35 (11/31) Smyth et na 11.2b 1.6b/ 12 al., 2002 31 16.7c 53 8 (25) 2 (6) (3.4–17.5) 5.4c (38.7%) (43) LGK 14 0

61.2 (30/49) Nataf et al., 100 12b/11c 3.5b/ 2003 (32) 49 28.3c 0 73 (1–4 cm3) 4 (8.2) 0 (7–15) 3c LINAC 40 (4–10 cm3) 100% (1 pt)

Nicolato et 11.7c al., 2005 47 3.8c 21.6c 0 80 (31/39) 0 (0) 1 (2) (5–16) (33) LGK

70 (72/103) 85.7 This report 12b 1.7b/ 16 100 23.3c 69.5 2 (1.7) 6 (5) LINAC (2–16) 2.8c (16%) 61 0

a OR, obliteration rate; RS, radiosurgery; LGK, Leksell gamma knife; na, not available; LINAC, linear accelerator-based radiosurgery b Median. c Mean. resection or obliteration (3, 21, 22, 28). Hence, a number of nidus. Moreover, one of the three patients presenting with these authors recommend performing an angiogram long after Rendu-Osler disease declared a new AVM location after radio- a child’s AVM has been resected (3, 22). In our opinion, this rec- surgical obliteration of three AVMs. These observations illus- ommendation is equally applicable to the follow-up of AVMs trate the possible angiogenesis phenomena evoked in cases of cured by radiosurgery, although we did not notice the phe- AVM recurrence in pediatric populations. Regarding the nomenon of nidus regrowth after obliteration in the present follow-up, we (similar to others [13, 32]) are convinced of the series. However, in one case, we did notice a new nidus com- absolute necessity for definitive, angiographic validation of ponent appearing outside the marginal isodose of a radiosurgi- AVM obliteration. At present, MRI scanning is not sensitive cal procedure that had successfully obliterated the original enough (even with new techniques, such as magnetic reso-

272 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com RADIOSURGERY FOR PEDIATRIC ARTERIOVENOUS MALFORMATIONS

A BC volume is a fully independent, predictive factor for obliteration. In relation to this, Levy et al. (27) found that volume was significantly correlated with obliteration rate only when multivariate logistic regression was applied. These authors reported a much higher obliteration rate for an AVM series with a small E median volume of 1.7 cm3. Hence, differences in oblitera- F tion rates reported for literature series do not seem to be correlated with the mean or D median AVM volume (Table 7). However, the doses applied are clearly dissimilar. In an attempt to minimize potential radiation toxicity in their patients’ developing brains, FIGURE 1. A, preradiosurgery, T2- Smyth et al. (43) adopted a weighted, cerebral MRI scan for a conservative dose prescription 14-year-old girl, showing a Rolandic policy and reduced the periph- AVM that had manifested itself eral dose, with a relatively low through seizures. B and C, lateral obliteration rate of 35%. These and anterior angiograms documenting the angioarchitecture of this AVM. D, authors reported a significant 13 months after radiosurgery, the AVM seemed to have been cured: this T2- correlation between dose pre- weighted cerebral MRI scan shows the absence of a draining vein and no resid- scription and obliteration rate ual nidus. E, lateral angiogram performed 13 months after radiosurgery show- ϭ ing an early-stage Rolandic draining vein. F, anterior angiogram showing a (P 0.025) but only observed tiny nidus at the beginning of the Rolandic draining vein. This residual AVM a (nonsignificant) trend toward was treated by a second radiosurgery procedure. a correlation between AVM size and obliteration rate. nance angiography and digital subtraction angiography) to However, some (larger) lesions received adequate prescription detect small nidi or residual microshunts with bleeding poten- doses, whereas others did not, possibly explaining why a sta- tial (Fig. 1). Unfortunately, the overall obliteration rates tistically significant correlation between response and AVM reported for several literature series include patients with size was not observed. Smyth et al. (43) also noted that when a AVMs considered to be obliterated on the basis of MRI scan- marginal dose of at least 18 Gy was delivered to the target vol- ning results (27, 38, 43). We think such a procedure could over- ume, obliteration rates were nearly 10-fold higher than those estimate the true obliteration rate. At present, we use magnetic for patients receiving lower doses, suggesting that this dose is resonance angiography digital subtraction angiography as a at or near the threshold for AVM involution, regardless of AVM routine examination during intermediate follow-up until an volume. These authors suggested that the poor obliteration rate AVM seems to have been obliterated (26). The apparent oblit- in larger AVMs may not be related so much to nidus size but eration is always confirmed by angiography. Moreover, follow- rather to the limitation in dosing a larger nidus volume within up should also assess the neuropsychological behavior of chil- a larger volume of surrounding brain tissue while also striving dren who received ionizing radiation. To date, there are few to reduce the risk of radiation-induced central nervous system data available on the long-term neuropsychological outcome toxicity. They (and Yamamoto et al. [50]) argued that the vessel- for pediatric radiosurgery patients (40). level radiobiological response to treatment is likely to be the same in large and small AVMs and that a favorable response to Predictive Factors for Obliteration radiosurgery should be expected as long as a sufficient dose is The large number of patients in the present series offers the reached, regardless of overall AVM size. Unfortunately, our advantage of statistical significance. One of the major predic- methodology did not permit us to determine potentially signif- tive factors for obliteration rate was low AVM volume. icant correlations between dose prescription and obliteration Obviously, this is not a new finding (6, 8, 12, 13, 27, 32, 33, 37, rate because we treated almost all patients identically. Indeed, 42), although our statistical analysis does suggest that AVM we adopted the same prescription dose policy in children as in

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 273 REYNS ET AL.

adults (as do other researchers [2, 27, 32, 46]) by applying a Furthermore, it is important to compare our obliteration and marginal dose of 21 to 25 Gy for 95% of patients (on the 70% morbidity data with those from other pediatric AVM radio- isodose in our study, according to the linear accelerator’s dose surgery series (Table 7) rather than surgical series. Indeed, pop- distribution). Only a few patients were treated with lower ulations having undergone either radiosurgery or conventional doses, based on their AVM volume and/or location. With the surgery should not be compared because radiosurgery is exception of Smyth et al. (43), the aforementioned authors mainly suitable for inoperable or deeply located AVMs or those report obliteration rates ranging from 61 to 95% (Table 7). In our located in functional areas. We wish to emphasize that approx- opinion, these data constitute a strong argument for a correla- imately 70% of patients in this series presented an AVM in a tion between dose prescription and obliteration rate. Hence, functional location and that 30% of the population had a Smyth et al. (43) have begun to use adult dose prescriptions in Spetzler-Martin Grade VI AVM. Only 5% of the patients pre- treating similar AVMs in children to increase the chances of sented with a permanent neurological deficit. Hence, the obliteration. Spetzler-Martin grading system clearly seems to be an unsuit- Another interesting predictive factor of obliteration seems able prognostic grading system for radiosurgery (36). We are to be young age. In our univariate analysis, an age of 10 years convinced that a child with an AVM should be treated by sur- or younger was significantly associated with a better oblitera- gery whenever the nidus can be resected under safe condition rate (P ϭ 0.05). Indeed, the significance value was even tions, in view of the immediate protection against bleeding that lower for an age of 8 years or younger (P ϭ 0.03). These find- surgery offers (5, 11, 19, 21). On one hand, the postradiosurgery ings confirm previous studies indicating young age as a predic- risk of hemorrhage during the latency interval persists until tive factor for obliteration (37, 42). obliteration is achieved. On the other hand, the less-invasive Our study highlights embolization as a negative predictive nature of the radiosurgical approach (combined with the factor for obliteration. This had already been reported in an brevity or absence of hospitalization) contributes to the adult population (37). Obviously, staged embolizations have a patient’s successful recovery from physical, mental, and emo- potentially valuable preradiosurgery role in managing large tional standpoints (16). lesions (17, 25, 31, 34, 47, 49) as long as complete elimination of a peripheral portion of the nidus is achieved (17). However, in CONCLUSION our experience, fragmenting of a residual nidus by embolization could make a potential radiosurgical target in a heteroge- The present study confirms that radiosurgery is a safe and neous environment less favorable for radiosurgery. This could effective treatment for AVMs in the pediatric age group. One be one explanation for the fact that preradiosurgery emboliza- criterion for success was the use of a prescription dose similar tion was reported in the present study as a negative predictive to that used with adult populations. Nidus volume seems to be factor for radiosurgical AVM obliteration. Another explanation the other major variable affecting successful obliteration. might be partial, postradiosurgery revascularization of the pre- Moreover, the younger the patient, the more effective the radio- embolized portion (17, 39). This would mean that radiosurgery surgery seemed to be. There are now good evidence-based rea- should be performed alone (i.e., in the absence of embolization) sons for considering this technique as an alternative to con- whenever the nidus volume permits. ventional surgery in cases of deeply located AVMs or those Our statistical analysis did not confirm single draining veins close to functional areas. Radiosurgery can be effective as a (6, 37), patient sex (32), or nidus location as predictive factors sole treatment or as a part of a multimodal strategy when com- for obliteration (30, 37). bined with other techniques. Hence, as in adult populations, the key point in managing pediatric AVMs seems to be the use Outcome and Morbidity of a multidisciplinary approach for deciding on the correct technique for the patient. We are now extending our experience With an overall obliteration rate of approximately 70% and a with gamma knife radiosurgery. permanent neurological deficit rate of 5%, the present series confirms the safety and efficacy of radiosurgery for the man- REFERENCES agement of AVMs in the pediatric age group. It could be argued that concern for potential late consequences of high doses of 1. Altschuler EM, Lunsford LD, Coffey RJ, Bissonette DJ, Flickinger JC: Gamma ionizing radiation can explain the general reluctance to use this knife radiosurgery for intracranial arteriovenous malformations in childhood technique on young patients. Nevertheless, there are increasing and adolescence. Pediatr Neurosci 15:53–61, 1989. numbers of well-documented pediatric series (2, 7, 27, 32, 42, 2. Amendola BE, Wolf A, Coy SR: Radiosurgery for intracranial arteriovenous malformations in children. J Radiosurg 2:159–164, 2000. 45, 46). To date, there is no real argument against using radio- 3. Andaluz N, Myseros JS, Sathi S, Crone KR, Tew JM Jr: Recurrence of cerebral surgery as a first-line treatment for inoperable nidi or as a com- arteriovenous malformations in children: Report of two cases and review of plementary treatment for a residual nidus after embolization or the literature. Surg Neurol 62:324–331, 2004. surgery in a pediatric population. In particular, we are not 4. Betti OO, Munari C, Rosler R: Stereotactic radiosurgery with the linear accelerator: Treatment of arteriovenous malformations. Neurosurgery 24:311–321, aware of any reported cases in the literature demonstrating 1989. that stereotactic radiosurgery for the treatment of pediatric 5. Blond S, Coche-Dequeant B, Castelain B: Stereotactically guided radiosurgery AVMs is a causative agent for new tumor formation. using the linear accelerator. Acta Neurochir (Wien) 124:40–43, 1993.

274 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com RADIOSURGERY FOR PEDIATRIC ARTERIOVENOUS MALFORMATIONS

6. Chang JH, Chang JW, Park YG, Chung SS: Factors related to complete occlu- previously proven to be obliterated after gamma knife radiosurgery. sion of arteriovenous malformations after gamma knife radiosurgery. J Neurosurgery 46:803–810, 2000. Neurosurg 93 [Suppl 3]:96–101, 2000. 29. Maity A, Shu HK, Tan JE, Ruffer J, Sutton LN, Tochner Z, Lustig R: Treatment 7. Colombo F: Linear accelerator radiosurgery of cerebral arteriovenous malfor- of pediatric intracranial arteriovenous malformations with linear-accelerator- mations in the pediatric age group, in Raimondi AJ, Choux M, Di Rocco C based stereotactic radiosurgery: The University of Pennsylvania experience. (eds): Cerebrovascular Diseases in Children. New York, Springer, 1991, pp Pediatr Neurosurg 40:207–214, 2004. 159–167. 30. Meder JF, Oppenheim C, Blustajn J, Nataf F, Merienne L, Lefkoupolos D, 8. Colombo F, Benedetti A, Pozza F, Marchetti C, Chierego G: Linear accelera- Laurent A, Merland JJ, Schlienger M, Fredy D: Cerebral arteriovenous malfor- tor radiosurgery of cerebral arteriovenous malformations. Neurosurgery mations: The value of radiologic parameters in predicting response to radio- 24:833–840, 1989. surgery. AJNR Am J Neuroradiol 18:1473–1483, 1997. 9. Deruty R, Pelissou-Guyotat I, Morel C, Bascoulergue Y, Turjman F: Reﬂections 31. Millar C, Bissonnette B, Humphreys RP: Cerebral arteriovenous malforma- on the management of cerebral arteriovenous malformations. Surg Neurol tions in children. Can J Anaesth 41:321–331, 1994. 50:245–256, 1998. 32. Nataf F, Schlienger M, Lefkopoulos D, Merienne L, Ghossoub M, Foulquier 10. Deruty R, Pelissou-Guyotat I, Mottolese C, Bascoulergue Y, Amat D: The JN, Deniaud-Alexandre E, Mammar H, Meder JF, Turak B, Huart J, Touboul combined management of cerebral arteriovenous malformations: Experience E, Roux FX: Radiosurgery of cerebral arteriovenous malformations in chil- with 100 cases and review of the literature. Acta Neurochir (Wien) dren: A series of 57 cases. Int J Radiat Oncol Biol Phys 57:184–195, 2003. 123:101–112, 1993. 33. Nicolato A, Foroni R, Seghedoni A, Martines V, Lupidi F, Zampieri P, Sandri 11. Di Rocco C, Tamburrini G, Rollo M: Cerebral arteriovenous malformations in MF, Ricci U, Mazza C, Beltramello A, Gerosa M, Bricolo A: Leksell gamma children. Acta Neurochir (Wien) 142:145–158, 2000. knife radiosurgery for cerebral arteriovenous malformations in pediatric 12. Ellis TL, Friedman WA, Bova FJ, Kubilis PS, Buatti JM: Analysis of treatment patients. Childs Nerv Syst 21:301–308, 2005. failure after radiosurgery for arteriovenous malformations. J Neurosurg 34. Perini S, Zampieri P, Rosta L, Piovan E, Micheli ED, Nicolato A, Mazza C, Benati 89:104–110, 1998. A, Ruatti P: Endovascular treatment of pial AVMs: Technical options, indications 13. Friedman WA, Bova FJ: Linear accelerator radiosurgery for arteriovenous and limits in pediatric age patients. J Neurosurg Sci 41:325–330, 1997. malformations. J Neurosurg 77:832–841, 1992. 35. Perret G, Nishioka H: Report on the cooperative study of intracranial 14. Frizzel RT, Fisher WS 3rd: Cure, morbidity, and mortality associated with aneurysms and subarachnoid hemorrhage. Section VI. Arteriovenous malfor- embolization of brain artetriovenous malformations: A review of 1246 mations. An analysis of 545 cases of craniocerebral arteriovenous malforma- patients in 32 series over a 35-year period. Neurosurgery 37:1031–1040, 1995. tions and ﬁstulae reported to the cooperative study. J Neurosurg 25:467–490, 15. Gerosa M, Cappellotto P, Licata C, Iraci G, Pardatscher K, Fiore DL: Cerebral 1966. arteriovenous malformations in children (56 cases). Childs Brain 8:356–371, 36. Pollock BE, Flickinger JC: A proposed radiosurgery-based grading system for 1981. arteriovenous malformations. J Neurosurg 96:79–85, 2002. 16. Gerszten PC, Adelson PD, Kondziolka D, Flickinger JC, Lunsford LD: Seizure 37. Pollock BE, Gorman DA, Coffey RJ: Patient outcomes after arteriovenous outcome in children treated for arteriovenous malformations using gamma malformation radiosurgical management: Results based on a 5- to 14-year knife radiosurgery. Pediatr Neurosurg 24:139–144, 1996. follow-up study. Neurosurgery 52:1291–1297, 2003. 17. Gobin YP, Laurent A, Merienne L, Schlienger M, Aymard A, Houdart E, 38. Pollock BE, Kondziolka D, Flickinger JC, Patel AK, Bissonette DJ, Lunsford LD: Casasco A, Lefkopoulos D, George B, Merland JJ: Treatment of brain arteri- Magnetic resonance imaging: An accurate method to evaluate arteriovenous ovenous malformations by embolization and radiosurgery. J Neurosurg malformations after stereotactic radiosurgery. J Neurosurg 85:1044–1049, 1996. 85:19–28, 1996. 39. Pollock BE, Kondziolka D, Lunsford LD, Bissonette D, Flickinger JC: Repeat 18. Guidetti B, Delitala A: Intracranial arteriovenous malformations. Conserva- stereotactic radiosurgery of arteriovenous malformations: Factors associated tive and surgical treatment. J Neurosurg 53:149–152, 1980. with incomplete obliteration. Neurosurgery 38:318–324, 1996. 19. Hladky JP, Lejeune JP, Blond S, Pruvo JP, Dhellemmes P: Cerebral arteriove- 40. Riva D, Pantaleoni C, Devoti M, Lindquist C, Steiner L, Giorgi C: Radio- nous malformations in children: Report on 62 cases. Childs Nerv Syst surgery for cerebral AVMs in children and adolescents: The neurobehavioral 10:328–333, 1994. 20. Hoh BL, Ogilvy CS, Butler WE, Loeffler JS, Putman CM, Chapman PH: outcome. J Neurosurg 86:207–210, 1997. Multimodality treatment of nongalenic arteriovenous malformations in pedi- 41. Schmit BP, Burrows PE, Kuban K, Goumnerova L, Scott RM: Acquired cere- atric patients. Neurosurgery 47:346–358, 2000. bral arteriovenous malformation in a child with moyamoya disease. Case 21. Humphreys RP, Hoffman HJ, Drake JM, Rutka JT: Choices in the 1990s for the report. J Neurosurg 84:677–680, 1996. management of pediatric cerebral arteriovenous malformations. Pediatr 42. Shin M, Kawamoto S, Kurita H, Tago M, Sasaki T, Morita A, Ueki K, Kirino Neurosurg 25:277–285, 1996. T: Retrospective analysis of a 10-year experience of stereotactic radio surgery 22. Kader A, Goodrich JT, Sonstein WJ, Stein BM, Carmel PW, Michelsen WJ: for arteriovenous malformations in children and adolescents. J Neurosurg Recurrent cerebral arteriovenous malformations after negative postoperative 97:779–784, 2002. angiograms. J Neurosurg 85:14–18, 1996. 43. Smyth MD, Sneed PK, Ciricillo SF, Edwards MS, Wara WM, Larson DA, 23. Kiris T, Sencer A, Sahinbas M, Sencer S, Imer M, Izgi N: Surgical results in Lawton MT, Gutin PH, McDermott MW: Stereotactic radiosurgery for pedi- pediatric Spetzler-Martin grades I-III intracranial arteriovenous malforma- atric intracranial arteriovenous malformations: The University of California tions. Childs Nerv Syst 21:69–76, 2005. at San Francisco experience. J Neurosurg 97:48–55, 2002. 24. Kondziolka D, Lunsford LD: Re: Current status of radiosurgery for arteriove- 44. Steiner L, Lindquist C, Adler JR, Torner JC, Alves W, Steiner M: Clinical out- nous malformations. Can J Neurol Sci 19:514–515, 1992. come of radiosurgery for cerebral arteriovenous malformations. J Neurosurg 25. Lasjaunias P, Hui F, Zerah M, Garcia-Monaco R, Malherbe V, Rodesch G, 77:1–8, 1992. Tanaka A, Alvarez H: Cerebral arteriovenous malformations in children. 45. Steiner L, Lindquist C, Steiner M: Stereotactic radiosurgery: Part 1—Radio- Management of 179 consecutive cases and review of the literature. Childs surgery with focused gamma-beam irradiation in children, in Edwards MS, Nerv Syst 11:66–79, 1995. Hoffman HJ (eds): Cerebral Vascular Disease in Children and Adolescents. 26. Leclerc X, Gauvrit JY, Trystram D, Reyns N, Pruvo JP, Meder JF: Cerebral arte- Baltimore, Williams & Wilkins, 1989, pp 367–388. riovenous malformations: Value of the non invasive vascular imaging tech- 46. Tanaka T, Kobayashi T, Kida Y, Oyama H, Niwa M: Comparison between niques [in French]. J Neuroradiol 31:349–358, 2004. adult and pediatric arteriovenous malformations treated by Gamma Knife 27. Levy EI, Niranjan A, Thompson TP, Scarrow AM, Kondziolka D, Flickinger radiosurgery. Stereotact Funct Neurosurg 66 [Suppl 1]:288–295, 1996. JC, Lunsford LD: Radiosurgery for childhood intracranial arteriovenous mal- 47. terBrugge KG: Neurointerventional procedures in the pediatric age group. formations. Neurosurgery 47:834–842, 2000. Childs Nerv Syst 15:751–754, 1999. 28. Lindqvist M, Karlsson B, Guo WY, Kihlstrom L, Lippitz B, Yamamoto M: 48. Wilkins RH: Natural history of intracranial vascular malformations: A review. Angiographic long-term follow-up data for arteriovenous malformations Neurosurgery 16:421–430, 1985.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 275 REYNS ET AL.

49. Wisoff JH, Berenstein A: Interventional neuroradiology, in Edwards MS, rhage or other symptoms, as well as the disability AVMs can create, we Hoffman JH (eds): Cerebral Vascular Disease in Children and Adolescents. continue to advocate an aggressive stance in their management. Baltimore, Williams and Wilkins, 1989, pp 139–157. 50. Yamamoto Y, Coffey RJ, Nichols DA, Shaw EG: Interim report on the radio- Douglas Kondziolka surgical treatment of cerebral arteriovenous malformations: The influence of Pittsburgh, Pennsylvania size, dose, time, and technical factors on obliteration rate. J Neurosurg 83:832–837, 1995. he authors present their considerable experience (100 patients) with 51. Yamamoto M, Jimbo M, Ide M, Lindquist C, Steiner L: Gamma knife radio- Tthe radiosurgical treatment (dedicated linear accelerator) of pedi- surgery in cerebral arteriovenous malformations: Postobliteration nidus atric intracranial AVMs. The indications are clearly defined and include changes observed on neurodiagnostic imaging. Stereotact Funct Neurosurg a clinically stable course and no impending intracranial hypertension. 64 [Suppl 1]:126–133, 1995. Their results are encouraging with 70% of the patients having angiographically recorded complete obliteration and no substantial difference between the pediatric and adult populations in terms of “thera- COMMENTS peutic timing.” As expected, the two major predictive factors are represented by a limited target volume and no previous endovascular he authors have compiled their extensive experience treating pedi- procedures. The latter seems to further stress the unsolved question of Tatric arteriovenous malformations (AVMs) using linear accelerator appropriate therapeutic algorithms in AVMs. After documented oblit- radiosurgery between 1988 and 2002, a period during which substantial eration, there was only 2% posttreatment bleeding and no hemorrhage. technological changes were occurring in radiosurgical practice. Among Finally, the morbidity rate was low (5%) and directly related to the these developments were improved imaging and volumetric treatment nidus volume, whereas none of these patients ever complained of car- planning and the development of radiation delivery systems capable of cinogenic sequelae. highly conformal treatment. Nonetheless, their results are very good This interesting report deserves several specific comments. Age with obliteration rates and treatment-related morbidity rates consistent definitively seems to be a predictive factor for AVM obliteration (3, 4); with contemporary expectations. This is the largest pediatric experi- this is likely owing to either the peculiar radiosensitivity of immature ence published thus far. The authors were able to obtain follow-up data vessels in the pediatric population or to focally (nidus wall, endothelial for at least 3 years for every patient treated and have confirmed AVM cells) higher levels of specific growth factors (e.g., epidermal growth obliteration angiographically in each instance. These aspects of their factor and fibroblast growth factor) in infants and children (1, 2). report eliminate potential biases that appear in some other studies and Despite the reported rewarding results, the adopted “Talairach-based” make it of considerable value in assessing the role of radiosurgery in procedure seems cumbersome, particularly when compared with cur- treating pediatric AVMs. With regard to interval follow-up of treated rent radiosurgical techniques such as double staged localization under patients, we reserve angiography for those patients in whom there is general anesthesia, single shot treatment, annual angiographic control apparent obliteration on magnetic resonance imaging scans. This is of rather than magnetic resonance imaging plus magnetic resonance particular relevance for children who require general anesthesia for angiography for at least 24 to 36 months. However, considering the angiography. We have also found computed tomographic angiography substantial case material, the appropriate clinical and radiological to be superior to magnetic resonance imaging and magnetic resonance remarks, and the long-term follow-up period, this retrospective study angiography not only in assessing the residual nidus, but also in delin- represents a relevant contribution to the literature. eating the AVM for treatment planning purposes. Massimo Gerosa Paul H. Chapman Verona, Italy Boston, Massachusetts 1. Nicolato A, Lupidi F, Sandri MF, Foroni R, Zampieri P, Mazza C, Maluta S, VMs are the most common cause of pediatric stroke. As shown in Beltramello A, Gerosa M: Gamma knife radiosurgery for cerebral arteriovenous malformations in children/adolescents and adults. Part I: Differences in this series, a wide variety of clinical problems can be associated A epidemiologic, morphologic, and clinical characteristics, permanent complica- with a brain AVM and multimodality management is often necessary. tions, and bleeding in the latency period. Int J Radiat Oncol Biol Phys This report makes a number of observations. First, the 70% overall AVM 64:904–913, 2006. 3 obliteration rate was reasonable. However, no AVM larger than 10 ml in 2. Nicolato A, Lupidi F, Sandri MF, Foroni R, Zampieri P, Mazza C, Pasqualin A, volume was obliterated (n = 4). Thus, the care of patients with larger Beltramello A, Gerosa M: Gamma Knife radiosurgery for cerebral arteriove- AVMs must be questioned. Perhaps embolization, resection, or staged nous malformations in children/adolescents and adults. Part II: Differences in radiosurgery may need to be considered for these larger lesions. obliteration rates, treatment-obliteration intervals, and prognostic factors. Int Secondly, the authors used a fairly uniform dose selection method, which J Radiat Oncol Biol Phys 64:914–921, 2006. could perhaps have been tailored individually to specific AVMs. One 3. Pollack BE, Gorman DA, Coffey RJ: Patient outcomes after arteriovenous malformation radiosurgical management: Results based on a 5- to 14-year fol- patient developed monocular blindness after radiosurgery for an AVM low-up study. Neurosurgery 52:1291–1297, 2003. near the optic nerve. Was this related to the dose received or to dose plan 4. Shin M, Kawamoto S, Kurita H, Tago M, Sasaki T, Morita A, Ueki K, Kirino T: conformality? Other centers have promoted the concept that children Retrospective analysis of a 10-year experience of stereotactic radio surgery may obliterate their AVMs faster than adults with similar lesions. This for arteriovenous malformations in children and adolescents. J Neurosurg remains a hypothetical concept. Because of the lifelong risk of hemor- 97:779–784, 2002.

276 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CLINICAL STUDIES

Eric L. Chang, M.D. Department of Radiation Oncology, The University of Texas, Houston, Texas A PILOT STUDY OF NEUROCOGNITIVE FUNCTION Jeffrey S. Wefel, Ph.D. IN PATIENTS WITH ONE TO THREE NEW BRAIN Department of Neuro-Oncology, The University of Texas, Houston, Texas METASTASES INITIALLY TREATED WITH STEREOTACTIC Moshe H. Maor, M.D. RADIOSURGERY ALONE Department of Radiation Oncology, The University of Texas, Houston, Texas OBJECTIVE: Whether to administer or omit adjuvant whole-brain radiation therapy in Samuel J. Hassenbusch III, conjunction with stereotactic radiosurgery (SRS) in the initial management of patients M.D., Ph.D. with one to three newly diagnosed brain metastases is the subject of debate. This report Department of Neurosurgery, The University of Texas, provides data from a pilot study in which neurocognitive function (NCF) was prospec- Houston, Texas tively measured for patients with one to three newly diagnosed brain metastases treated Anita Mahajan, M.D. with initial SRS alone. Department of Radiation Oncology, The University of Texas, METHODS: Fifteen patients were prospectively treated with initial SRS alone. Assessment Houston, Texas of NCF and magnetic resonance imaging scans were performed. Frederick F. Lang, M.D. RESULTS: At baseline, 67% of the patients had impairment on one or more tests of NCF. Department of Neurosurgery, The domains most frequently impaired at baseline were executive function, motor dex- The University of Texas, Houston, Texas terity, and learning/memory with an incidence of 50, 40, and 27% respectively. Brain metastasis volume (Ͼ3 cm3) measured at the time of initial SRS treatment was associ- Shiao Y. Woo, M.D. Ͻ Department of Radiation Oncology, ated with worse performance on a measure of attention (P 0.05). At 1 month, declines The University of Texas, in the learning/memory and motor dexterity domains were most common. In a sub- Houston, Texas group of ﬁve patients still alive 200 days after enrollment, four patients (80%) demon- Leni A. Mathews, R.N. strated stable or improved learning/memory, three (60%) demonstrated stable or improved Department of Radiation Oncology, The University of Texas, executive function, and three (60%) demonstrated stable or improved motor dexterity Houston, Texas relative to their baseline evaluation. Pamela K. Allen, Ph.D. CONCLUSION: Although two-thirds of the brain metastasis patients had impaired NCF Department of Radiation Oncology, at baseline, the majority of ﬁve long-term survivors had stable or improved NCF perform- The University of Texas, Houston, Texas ance across executive function, learning/memory, and motor dexterity. Almon S. Shiu, Ph.D. KEY WORDS: Brain metastasis, Neurocognitive function, Radiosurgery Department of Radiation Physics, The University of Texas, Neurosurgery 60:285–292, 2007 DOI: 10.1227/01.NEU.0000249272.64439.B1 www.neurosurgery-online.com Houston, Texas Christina A. Meyers, Ph.D. Department of Neuro-Oncology, The University of Texas, hole-brain radiation therapy (WBRT) nosed brain metastases is now the subject of Houston, Texas has historically been considered the debate. mainstay of treatment for metastatic Those who advocate adjuvant WBRT em- M.D. Anderson Cancer Center, W The University of Texas brain disease. During the past 15 years, signif- phasize its merit in reducing the risk for subse- (ELC, JSW, MHM, SJH, AM, FFL, SYW, icant advances have been made through ran- quent brain metastases. Furthermore, it has LM, PA, ASS, CM) domized clinical trials showing that focal treat- been argued that subsequent brain metastases Reprint requests: ments with either surgery or stereotactic are best prevented because they are more Eric L. Chang, M.D., radiosurgery (SRS) result in improved survival likely to cause neurological sequelae than The University of Texas, in patients with a single brain metastasis (1, WBRT (8, 11, 12). Detractors argue that WBRT M.D. Anderson Cancer Center, Division of Radiation Oncology, 10). The role of WBRT has recently been called should be avoided and that, instead, a strat- 1515 Holcombe Boulevard Unit 97, into question because of its perceived potential egy of close follow-up and neuroimaging to Houston, TX 77030. to cause neurocognitive dysfunction (14). detect recurrences treatable by focal means Email: [email protected] Whether to administer or omit adjuvant WBRT should be used. It is postulated that withhold- Received, March 3, 2006. in conjunction with SRS in the initial manage- ing WBRT from the initial management of Accepted, October 13, 2006. ment of patients with one to three newly diag- brain metastases with SRS might delay the

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 277 CHANG ET AL.

onset of neurocognitive dysfunction (5, 8). The neurocognitive Radiation Dose outcome of patients with one to three new brain metastases The dose was prescribed in a similar fashion to Radiation undergoing this strategy of SRS without initial WBRT is the Therapy Oncology Group 90-05 guidelines (13). In summary, subject of this pilot study because of the paucity of such data lesions treated with a cone diameter of less than 2 cm were using this approach. specified to receive 20 to 24 Gy, lesions treated with a cone diameter between 2 and 3 cm received 18 Gy, and lesions treated PATIENTS AND METHODS with a cone diameter between 3 and 4 cm received 15 Gy. The dose was prescribed to the 80% isodose line or higher. Lesions Patient Eligibility were contoured on 1.5-mm-thick, contrast-enhanced planning To be eligible, patients were required to be at least 18 years CT scans obtained for SRS treatment planning. For tiny or ill- of age and to have recursive partitioning analysis Class I or II defined metastases, a volumetric T1-weighted, contrast- disease (Karnofsky Performance Scale score Ͼ 70) (3). It was enhanced MRI scan was obtained before the SRS procedure for required that patients have one to three newly diagnosed fusion with the planning CT scan. In such cases, both CT and brain metastases eligible for treatment with SRS. A brain mag- MRI scans were used for target delineation. netic resonance imaging (MRI) scan was required within 1 month of registration. All patients provided signed Neurocognitive Evaluation informed consent to participate in this study. Patients were Faculty and trained staff from the Section of Neuropsy- excluded from participation if they had previously received chology performed all neuropsychological assessments. The WBRT, SRS, or surgery to the brain; had more than three brain neurocognitive domains assessed along with the correspon- metastases; had a “radiosensitive” histology, such as lym- ding specific test are listed in Table 1. The tests were 1) Digit phoma, germ-cell tumor, leukemia, or small cell lung cancer; Span to measure attention span by having the patient repeat had evidence of leptomeningeal disease; or had brain metas- numbers forward and in reverse; 2) Digit Symbol to measure tasis from an unknown primary tumor. Women of childbear- graphomotor speed by having patients code symbols under ing age were required to have a negative pregnancy test. The numbers as fast as possible; 3) Hopkins Verbal Learning Test Institutional Review Board of The University of Texas M.D. for memory, including Total Recall (number of the 12 list words Anderson Cancer Center, Houston, Texas, provided approval recalled during three trials), Recognition (number of list words for this study. recognized from distractors), and Delayed Recall (number of list words recalled after a 20-min delay); 4) Controlled Oral Treatment Plan Word Association for verbal fluency (number of words beginning with a specific letter administered during three 1-min tri- All patients enrolled in the study were required to undergo als); 5) Trail-Making Test Part A for visual-motor scanning SRS to discernible brain metastases within 1 week of registra- speed (connecting dots in order from one to 25 as fast as possi- tion. The type and duration of systemic therapy received after ble; 6) Trail-Making Test Part B for executive function (flexibly registration on the trial was left to the discretion of the patient’s shifting mental set by connecting dots alternating between medical oncologist. number and letter sequences); and 7) Grooved Pegboard for fine motor dexterity (placing 25 slotted pegs into holes as fast Radiosurgical Procedure as possible. The Functional Assessment of Cancer Therapy- A PRIMART linear accelerator (Siemens Oncology Care Brain (FACT-BR) test was administered to assess quality of life Systems, Concord, CA), installed and commissioned as a dedi- (18). Patients underwent a baseline evaluation that was gener- cated SRS unit, was used for all intracranial SRS treatments. Immobilization and intracranial target localization were a achieved using a Brown-Robert-Wells stereotactic head frame. TABLE 1. Neurocognitive domain and corresponding test Image acquisition was performed on the morning of the proce- Domain Test dure with a computed tomographic (CT) scanner (General Attention WAIS-III digit span Electric Company, Waukesha, WI or Phillips Medical Systems, Information processing WAIS-III digit symbol, Andover, MA) to obtain contrast-enhanced, 1.5-mm-thick images speed TMT Part A of the entire brain. The scans were transferred to a workstation Learning/memory HVLT, revised by a Digital Imaging and Communication in Medicine link for Verbal fluency COWA treatment planning using X-knife software (Versions 3 and 4; Executive function TMT Part B Integra Radionics, Burlington, MA). The contrast-enhanced Upper extremity fine motor Lafayette grooved metastases intended for SRS treatment were contoured without dexterity pegboard (18) margin. After treatment planning and quality assurance proce- a WAIS, Wechsler Adult Intelligence Scale (19); TMT, Trail-Making Test (6); dures were completed, the SRS procedure was performed on HVLT, Hopkins Verbal Learning Test (2); COWA, Controlled Oral Word the same day. The volume of the contoured lesion was measured Association from the Multilingual Aphasia Examination (3). in real-time using the treatment planning computer.

278 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com NEUROCOGNITIVE FUNCTION IN BRAIN METASTASIS PATIENTS

ally performed before SRS. If scheduling prevented an evalua- tests and uses the baseline level of performance of a particular tion before SRS, evaluation was performed within 1 week of individual because these values may vary from patient to SRS. Patient evaluations were performed according to the patient. For each subject, the difference between the pretreat- following schedule: 1, baseline; 2, 20 to 40 days after SRS; 3, 50 ment baseline and each follow-up assessment was coded to to 70 days; 4, 80 to 100 days; 5, 110 to 130 days; 6, 140 to 160 examine the percentages of patients who show meaningful days; 7, 230 to 280 days; and 8, 300 to 370 days. losses or gains in the various test domains during the course of The tests were selected because they are widely used, stan- the study. In terms of motor dexterity, patients were catego- dardized psychometric instruments that have demonstrated rized as impaired or changed (using the reliable change index) sensitivity to the neurotoxic effects of cancer treatment in based on any unimanual motor finding. The Stata 9 statistical other clinical trials (6, 7). The tests have published normative software package (Statcorp, College Station, TX) and the data that take age into account, as well as education, sex, and Kaplan-Meier method were used to calculate cancer-control handedness, where appropriate. The tests were also selected end points of survival, local freedom from disease progression, to minimize the effects of repeated administration. The mem- and distant freedom from disease progression. ory test has six alternate forms. The other tests measure motor and information processing speed and are relatively resistant RESULTS to the effects of practice. Intracranial Tumor Progression Patient and Treatment Characteristics Local tumor progression was scored in cases in which a The patient and tumor characteristics are listed in Tables 2 treated lesion increased in size more than 25% of its initial and 3, respectively. The various systemic agents administered diameter or was associated with worsening neurological condi- to patients after study enrollment are listed in Table 4. The tion related to the treated metastasis that was refractory to median duration of follow-up of patients enrolled in the study steroids or required surgery. Distant recurrence was defined as was 6.5 months (range, 2.5–15 mo). The median SRS cone diam- the interval appearance of any new contrast-enhanced lesion eter was 2 cm (range, 1–4 cm). The median tumor volume by 3 3 within the brain. lesion was 1.185 cm (range, 0.0014–17.3 cm ); the median tumor volume by patient was 1.76 cm3 (range, 0.161–19.98 cm3). Follow-up Schedule The median prescribed SRS dose was 20 Gy (range, 14–21 Gy). All patients were scheduled to undergo follow-up evalua- The median duration of systemic therapy administered after tion, which included a brain MRI scan, and neuropsychological study enrollment was 3 months (range, 1–12 mo). evaluation. Follow-up examinations were scheduled at 1, 2, 4, 6, 9, 12, 15, and 18 months and every 6 months thereafter until no longer feasible. TABLE 2. Patient characteristics Statistical Analysis Age Median, yr 64.9 Published normative data that adjust for age, education, and Range, yr 31.5–77 sex, where appropriate, were used to convert patients’ raw cog- Sex nitive test score performances to standardized scores (z-scores; Male, no. (%) 5 (33.3) mean, 0; standard deviation [SD], 1) to facilitate comparisons Female, no. (%) 10 (66.7) among measures. At baseline, any test performance that was Histology Ϫ 1.5 SDs or more below the mean of the normative population Lung, no. (%) 8 (53.3) was considered impaired. This corresponds to performances Renal cell, no. (%) 3 (20) at the 6.5th percentile. A composite motor dexterity index was Melanoma, no. (%) 4 (26.7) calculated by averaging the adjusted scores for the dominant Karnofsky Performance Status and nondominant hand. 100, no. (%) 5 (26.7) The reliable change index (4) was used to determine both a 90, no. (%) 8 (53.3) statistically and clinically meaningful change in function. This 80, no. (%) 2 (13.3) index is derived from the standard error of measurement 70, no. (%) 1 (6.7) (SEM) for each test in the battery and represents the 90% con- Recursive partitioning class fidence interval for the difference in scores from baseline to 1, no. (%) 0 (0) follow-up that is expected if no real change has occurred. A reli- 2, no. (%) 15 (100) able change in test scores from baseline to follow-up is consid- Brain metastasis ered significant if it falls outside of the 90% confidence interval. One, no. (%) 11 (73.4) The SEM is calculated from the test-retest reliability (r) and the Two, no. (%) 2 (13.3) ϭ ϫ Ϫ 1/2 SD of test scores: SEM SD (1 r) . The standard error of Three, no. (%) 2 (13.3) difference is then calculated as [2(SEM2)]1/2. This method accounts for both normal variation in performance on these

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 279 CHANG ET AL.

TABLE 3. Tumor characteristicsa Patient Primary Number Location Location Location Local Distant no. site of lesions lesion 1 lesion 2 lesion 3 tumor status brain status 1 Lung 1 Lt frontal C C 2 Lung 1 Rt parietal C F 3 Melanoma 3 Rt parietal Rt temporal Lt temporal F F 4 Lung 1 Lt ventricle C F 5 Renal 2 Lt parietal Lt frontal C F 6 Lung 3 Rt occipital Lt occipital Rt cerebellum C F 7 Melanoma 1 Rt parietal C F 8 Lung 1 Rt frontal C F 9 Renal 1 Lt temporal F C 10 Melanoma 1 Lt parietal C F 11b Renal 1 Lt temporal C C 12b Lung 1 Rt parietal F F 13b Lung 2 Rt frontal Rt frontal C F 14b Melanoma 1 Rt occipital C C 15b Lung 1 Rt cerebellum C F

a Lt, left; C, controlled; Rt, right; F, failure. b Long-term survivor.

Neurocognitive Evaluation The average patient had completed 13.9 years (SD, 3.4; range, TABLE 4. Systemic agents received by patients while on study 9–20 yr) of formal education. At baseline, neurocognitive func- Drug No. of patients tion (NCF) testing revealed that 67% of the study population Biochemotherapy 1 was impaired on one or more NCF tests and that 50% had Carboplatin 3 impairment on two or more NCF tests. Specifically, patients Cisplatin 1 were most frequently impaired in the domains of executive Gefitnib 2 function (47%), motor dexterity (40%), and learning/memory Gemcitabine 1 (27%). Brain metastasis volume was defined as the sum of the Paclitaxel 6 individual target volumes of all the brain metastases in a Temozolamide 3 3 patient and was categorically defined as less than 3 cm or at Thalidomide 3 3 3 least 3 cm . Patients with greater than 3 cm metastatic tumor Capcetabine 1 volume demonstrated worse performance on a measure of attention span. line; 1 month after SRS, there were equal percentages (15%) of Acute Neurocognitive Changes improvement and declines. Visual motor scanning speed was Thirteen patients underwent a follow-up NCF evaluation measured with Trail-Making Test A and showed that 13% were within 20 to 40 days after SRS (Table 5). All patients demon- impaired at baseline. At 1 month, a greater proportion declined strated a decline on one or more tests, and approximately 54% (23%) than improved (15%). The FACT-BR administered at declined on two or more tests after 1 month. Declines on meas- baseline showed a mean score of 148.6, with 15% of patients ures of learning and memory (54%) and motor dexterity (46%) declining at 1 month (Table 6). on the Grooved Pegboard were most common. Executive function as measured using the Trail-Making Test B was impaired Late Neurocognitive Outcomes in 47% at baseline. Improvement was noted in 38%, whereas Five patients were evaluable during a period of 200 days 15% experienced a decline. There was no baseline impairment after their initial baseline evaluation. Tests that demonstrated for attention as measured by the Digit Span test. At 1 month the greatest abnormality at baseline and the greatest effects after SRS, attention declines (8%) were balanced by improve- from SRS at the acute postoperative evaluation were plotted, ments (8%). Assessment of processing speed using Digit including learning/memory (HVLT-R Trial 1-3) (Fig. 1), execu- Symbol showed 7% impairment at baseline with equal percent- tive function (Trail-Making Test Part B) (Fig. 2), and motor dex- ages (8%) of improvement and declines 1 month after SRS. terity (Grooved Pegboard average index) (Fig. 3). Relative to Measurement of verbal fluency using the controlled word asso- their baseline evaluations, learning/memory was stable or ciation test showed that 7% of patients were impaired at base- improved in four patients (80%), executive function was stable

280 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com NEUROCOGNITIVE FUNCTION IN BRAIN METASTASIS PATIENTS

TABLE 5. Neurocognitive performance at baseline and acutely after stereotactic radiosurgerya Mean Impairedb Declinedc Improvedc Test standardized at baseline at 1 mo at 1 mo (13 ؍ n) (13 ؍ n) (15 ؍ score (SD) (n WAIS-III digit span 11.0 (3.5)d 0% 8% 8% WAIS-III digit 10.9 (3.3)d 7% 8% 8% symbol COWA 49.9 (29.8)e 7% 15% 15% TMT Part A –1.02 (2.11)f 13% 23% 15% TMT Part B –1.35 (2.06)f 47% 15% 38% HVLT-R trials 1–3 –0.52 (1.34)f 27% 54% 8% HVLT-R delayed –0.27 (0.93)f 13% 54% 0% recall HVLT-R recognition –0.06 (1.21)f 13% 23% 8% Grooved pegboard –0.91 (1.57)f 40% 46% 15%

a SD, standard deviation; WAIS-III, Wechsler Adult Intelligence Scale—third edition; COWA, Controlled Oral Word Fluency from the Multilingual Aphasia Examination; TMT, Trail-Making Test; HVLT-R, Hopkins Verbal Learning Test—revised. b Performance 1.5 standard deviations below the normative mean. c On the basis of the reliable change index. d Age-corrected scaled score. FIGURE 1. Scatterplot of learning/memory in long-term survivors who e Age-and education-corrected percentile score. received SRS only. Standard deviation is from the normative mean. f Age-corrected z-score. Evaluation: 1, baseline; 2, 20–40 days after SRS; 3, 50–70 days; 4, 80–100 days; 5, 110–130 days; 6, 140–160 days; 7, 230–280 days; 8, 300 to 370 days. .Patient 11; ᭛, Patient 12; ᭺, Patient 13; ᭝, Patient 14; X, Patient 15 ,ٗ

TABLE 6. Quality of life at baseline and acutely after radiosurgerya Mean (SD) Declinedb Improvedb Measure rating at base- at 1 mo at 1 mo (13 ؍ n) (13 ؍ n) (15 ؍ line (n FACT-Brain 148.6 (18.9) 15% 0%

a SD, standard deviation; FACT-Brain, Functional Assessment of Cancer Therapy—Brain module. b On the basis of the reliable change index. or improved in three patients (60%), and motor dexterity was stable or improved in three patients (60%). The FACT-BR is shown for long-term survivors (Patients 11–15) in Figure 4. Four patients self-reported relatively stable scores; Patient 11 reported a decline in the FACT-BR, consistent with hospitalization for brain edema, and bilateral leg edema after SRS. A comparison of cognitive baseline between 10 patients who did not complete long-term follow-up and ﬁve patients with long-term follow-up showed no signiﬁcant differences at baseline on any FIGURE 2. Scatterplot of executive functioning in long-term survivors who measure of neurocognitive function or quality of life. One received SRS only. Standard deviation is from the normative mean. patient (Patient 12) whose SRS-treated metastasis in the right Evaluation: 1, baseline; 2, 20–40 days after SRS; 3, 50–70 days; 4, 80–100 parietal lobe progressed from 2.5 to 2.9 cm with slightly more days; 5, 110–130 days; 6, 140–160 days; 7, 230–280 days; 8, 300 to 370 days. .Patient 11; ᭛, Patient 12; ᭺, Patient 13; ᭝, Patient 14; X, Patient 15 ,ٗ edema underwent salvage surgery to the same lesion 5 months after initial SRS. A 2-SD improvement in the learning/memory domain occurred after initial SRS, then dropped 2 SDs at the vival time between patients with one lesion versus those with time of progression, and then improved 2 SDs again after sal- more than one lesion (5.1 versus 5.9 mo; P ϭ 0.5963). The 1-year vage surgery (Fig. 1). rate of actuarial freedom from local progression of the brain metastases treated with SRS was 70%. Of the 10 patients who Cancer Control Outcomes did not complete long-term follow-up, eight patients devel- The 1-year and median survival rates for patients were 26% oped distant brain recurrences. For all patients, the 1-year rate and 7.2 months, respectively. There was no difference in sur- of actuarial freedom from distant brain metastasis and the

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 281 CHANG ET AL.

SRS treatment, declines were observed in learning/memory and motor dexterity. A more variable picture emerged with respect to executive function. Declines may have resulted from acute toxicity in the form of increased peritumoral edema. Despite baseline NCF abnormalities and acute declines in NCF, a limited number of long-term survivors showed surprising resilience with stable or even improved NCF across multiple domains of learning and memory (80%), motor dexterity (60%), and executive function (60%). The major limitation in drawing any conclusion from this observation is the small number of long-term survivors (n ϭ 5); therefore, drawing conclusions from this subset of patients must be approached with caution. However, only long-term survivors are relevant in the determination of the optimal timing of WBRT with respect to NCF. In patients with rapidly progressive brain metastases, the issue of whether or not to administer WBRT may be irrelevant because this group of patients will rapidly die before potential long-term WBRT-associated toxicity can become apparent. FIGURE 3. Scatterplot of motor dexterity in long-term survivors who received WBRT after SRS or at the Time of Recurrence? SRS only. Standard deviation is from the normative mean. Evaluation: 1, baseline; 2, 20–40 days after SRS; 3, 50–70 days; 4, 80–100 days; 5, 110–130 Regine et al. (11) reported on the risk of symptomatic brain days; 6, 140–160 days; 7, 230–280 days; 8, 300 to 370 days. ٗ, Patient 11; ᭛, tumor recurrence and neurological deﬁcit associated with using Patient 12; ᭺, Patient 13; ᭝, Patient 14; X, Patient 15. SRS alone in the treatment of newly diagnosed brain metastases in 36 patients. The median survival duration and 1-year survival rate for the entire group were 9 months and 36%, respectively. Brain tumor recurrence anywhere in the brain occurred in 17 (47%) out of 36 patients, whereas 12 (71%) out of 17 patients were symptomatic at the time of recurrence and 10 (59%) out of 17 had an associated neurological deﬁcit. Although not necessarily cost effective, the practice of more frequent MRI scan surveillance of brain metastasis patients may make it less likely that new distant brain metastases will be symptomatic. It is probably true that most practitioners regularly and frequently image the brain after initial radiosurgery making it possible to make an early diagnosis of recurrent metastases before they become symptomatic. Nevertheless, the neurocognitive outcome in Radiation Therapy Oncology Group 91-04 was measured using the Mini Mental Status Examination and showed that uncontrolled brain metastases may cause mental impairment (12). In 3 months, the average change in Mini Mental Status Examination score was a drop of 0.5 for those whose brain metastases were radiologically controlled compared with a drop of 6.3 for those with uncontrolled brain FIGURE 4. ٗ ᭛ ᭺ Scatterplot of FACT-BR. , Patient 11; , Patient 12; , Patient metastases (P ϭ 0.02). 13; ᭝, Patient 14; X, Patient 15. Chemotherapy and Extracranial Cancer median time to its development were 22% and 4.6 months, as Confounding Factors respectively. No Grade 3 or 4 neurological toxicities were Studies of NCF in the brain metastasis population are com- observed. plex to analyze because there are factors other than brain metastases and their treatment that can confound the results. In DISCUSSION a study by Wefel et al. (16), of 84 women with nonmetastatic primary breast carcinoma, comprehensive neuropsychological The majority of patients with one to three newly diagnosed evaluation before receiving adjuvant therapy showed that 35% brain metastases in our study had some degree of neurocog- of the cohort had cognitive impairment, particularly in the nitive dysfunction in at least one domain. One month after areas of verbal learning and memory function.

282 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com NEUROCOGNITIVE FUNCTION IN BRAIN METASTASIS PATIENTS

The first longitudinal trial to report evidence of an associa- 3. Gaspar L, Scott C, Rotman M, Asbell S, Phillips T, Wasserman T, McKenna tion between cognitive dysfunction and chemotherapy was WG, Byhardt R: Recursive partitioning analysis (RPA) of prognostic factors in conducted by Wefel et al. (17). In 18 patients with breast carci- three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 37:745–751, 1997. noma, neuropsychological evaluation before treatment and 4. Jacobson NS, Truax P: Clinical significance: A statistical approach to defining after chemotherapy showed that 61% of the cohort showed a meaningful change in psychotherapy research. J Consult Clin Psychol decline in cognitive function relative to baseline, most com- 59:12–19, 1991. monly in attention, learning, and processing speed. 5. Lo SS, Chang EL, Suh JH: Stereotactic radiosurgery with and without whole- brain radiotherapy for newly diagnosed brain metastases. Expert Rev On the other hand, the relationship between volume of brain Neurother 5:487–495, 2005. metastases and cognitive function has also been reported and 6. Meyers CA, Smith JA, Bezjak A, Mehta MP, Liebmann J, Illidge T, Kunkler I, demonstrated the potential impact of CNS disease on cognitive Caudrelier JM, Eisenberg PD, Meerwaldt J, Siemers R, Carrie C, Gaspar LE, function (6). In our pilot study, patients with greater than 3 cm3 Curran W, Phan SC, Miller RA, Renschler MF: Neurocognitive function and metastatic tumor volume demonstrated worse performance on progression in patients with brain metastases treated with whole-brain radiation and motexafin gadolinium: Results of a randomized phase III trial. J a measure of attention span. Clin Oncol 22:157–165, 2004. Results from two separate randomized trials showed a sur- 7. Meyers CA, Hess KR: Multifaceted end points in brain tumor clinical trials: vival advantage of administering surgery or SRS respectively in Cognitive deterioration precedes MRI progression. Neuro-oncol 5:89–95, addition to WBRT in the management of single brain meta- 2003. 8. Patchell RA, Regine WF: The rationale for adjuvant whole brain radiation stases, clearly establishing the respective roles of surgery and therapy with radiosurgery in the treatment of single brain metastases. SRS (1, 9). On the other hand, the role of WBRT has now become Technol Cancer Res Treat 2:111–115, 2003. less clear. In fact, the lack of a survival benefit for adding WBRT 9. Patchell RA, Tibbs PA, Walsh JW, Dempsey RJ, Maruyama Y, Kryscio RJ, to SRS alone has been demonstrated in a randomized trial by Markesbery WR, Macdonald JS, Young B: A randomized trial of surgery in Aoyama et al. (2). The potential risk of WBRT-related dementia the treatment of single metastases to the brain. N Engl J Med 322:494–500, 1990. has been given as a reason for omitting WBRT (14). 10. Patchell RA, Cirrincione C, Thaler HT, Galicich JH, Kim JH, Posner JB: Single In a series of newly diagnosed brain metastasis patients who brain metastases: Surgery plus radiation or radiation alone. Neurology received SRS alone or SRS plus WBRT, no difference in median 36:447–453, 1986. survival or 1-year local freedom from progression was noted 11. Regine WF, Huhn JL, Patchell RA, St Clair WH, Strottmann J, Meigooni A, Sanders M, Young AB: Risk of symptomatic brain tumor recurrence and neu- (14). A retrospective multicenter review showed that omission rologic deficit after radiosurgery alone in patients with newly diagnosed of upfront WBRT did not seem to compromise the length of brain metastases: Results and implications. Int J Radiat Oncol Biol Phys survival in patients treated with SRS for newly diagnosed brain 52:333–338, 2002. metastases (15). Although many studies address the role of 12. Regine WF, Scott C, Murray K, Curran W: Neurocognitive outcome in brain WBRT after SRS using survival and tumor control end points, metastases patients treated with accelerated fractionation vs. accelerated- hyperfractionated radiotherapy: An analysis from Radiation Therapy they are limited by the absence of neurocognitive analysis, Oncology Group Study 91–104. Int J Radiat Oncol Biol Phys 51:711–717, 2001. which is likely to be the most important factor in answering 13. Shaw E, Scott C, Souhami L, Dinapoli R, Kline R, Loeffler J, Farnan N: Single this question. dose radiosurgical treatment of recurrent previously irradiated primary brain The question of whether WBRT should be administered elec- tumors and brain metastases: Final report of RTOG protocol 90–05. Int J Radiat Oncol Biol Phys 47:291–298, 2000. tively can only be definitively answered in a randomized trial. 14. Sneed PK, Lamborn KR, Forstner JM, McDermott MW, Chang S, Park E, Our study is relevant to patients and practitioners interested in Gutin PH, Phillips TL, Wara WM, Larson DA: Radiosurgery for brain metas- choosing SRS alone and seeking additional information despite tases: Is whole brain radiotherapy necessary? Int J Radiat Oncol Biol Phys the scarcity of data on NCF outcome associated with this choice. 43:549–558, 1999. 15. Sneed PK, Suh JH, Goetsch SJ, Sanghavi SN, Chappell R, Buatti JM, Regine WF, Weltman E, King VJ, Breneman JC, Sperduto PW, Mehta MP: A multi- CONCLUSION institutional review of radiosurgery alone vs. radiosurgery with whole brain radiotherapy as the initial management of brain metastases. Int J Radiat Although two-thirds of brain metastasis patients had Oncol Biol Phys 53:519–526, 2002. impaired NCF at baseline, the majority of five long-term sur- 16. Wefel JS, Lenzi R, Theriault R, Buzdar AU, Cruickshank S, Meyers CA: vivors had stable or improved NCF performance across execu- ‘Chemobrain’ in breast carcinoma?: A prologue. Cancer 101:466–475, 2004. 17. Wefel JS, Lenzi R, Theriault RL, Davis RN, Meyers CA: The cognitive seque- tive function, learning/memory, and motor dexterity. lae of standard-dose adjuvant chemotherapy in women with breast carcinoma: Results of a prospective, randomized, longitudinal trial. Cancer REFERENCES 100:2292–2299, 2004. 18. Weitzner MA, Meyers CA, Gelke CK, Byrne KS, Cella DF, Levin VA: The 1. Andrews DW, Scott CB, Sperduto PW, Flanders AE, Gaspar LE, Schell MC, Functional Assessment of Cancer Therapy (FACT) scale: Development of a Werner-Wasik M, Demas W, Ryu J, Bahary JP, Souhami L, Rotman M, brain subscale and revalidation of the general version (FACT-G) in patients Mehta MP, Curran WJ Jr: Whole brain radiation therapy with or without stereo- with primary brain tumors. Cancer 75:1151–1161, 1995. tactic radiosurgery boost for patients with one to three brain metastases: Phase III results of the RTOG 9508 randomised trial. Lancet 363:1665–1672, 2004. 2. Aoyama H, Shirato H, Tago M, Nakagawa K, Toyoda T, Hatano K, Kenjyo M, COMMENTS Oya N, Hirota S, Shioura H, Kunieda E, Inomata T, Hayakawa K, Katoh N, Kobashi G: Stereotactic radiosurgery plus whole-brain radiation therapy vs his study evaluates cognitive functions in patients treated with stereotactic radiosurgery alone for treatment of brain metastases: A random- Tstereotactic neurosurgery (SRS) for brain metastases. The authors ized controlled trial. JAMA 295:2483–2491, 2006. emphasize the importance of investigating treatment strategies that

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 283 CHANG ET AL.

could minimize cognitive dysfunction, given the known neurotoxicity onsideration of the neurocognitive courses of SRS versus WBRT associated with whole-brain radiotherapy. The authors used a prospec- Cprovides important information concerning treatment choices for tive design including pre- and postintervention cognitive evaluations, metastases. The authors have provided pilot data suggesting that which are important to determine the specific contributions of disease patients who underwent SRS had either stable or improved neurocog- and therapy. Difficulties in executive functions, motor dexterity, and nitive functioning in the executive, learning/memory, and motor dexter- memory were detected at baseline and before treatment in 67% of ity domains. Therefore, this study provides useful data. It is noted that patients with one to three metastases. Although five patients had sta- this study used a relatively small sample with a wide range of age and ble or improved cognitive functions after therapy, the small number of educational backgrounds. Further studies using WBRT control groups patients available for long-term follow-up examination precluded more and larger cell sizes will be needed to make more definitive conclusions. definitive conclusions regarding cognitive outcome after SRS. Kenneth C. Kutner Denise Correa Neuropsychologist Neuropsychologist Hackensack, New Jersey Philip H. Gutin New York, New York his study by Chang et al. prospectively evaluates the acute neu- Trocognitive morbidity that follows radiosurgical treatment of brain his pilot study addresses one of the most important issues in neuro- metastases in a small cohort of patients. Although they have thought- Toncology, namely what are the cognitive effects after brain metas- fully chosen a battery of psychometric tests and a neurologically-based tasis management? Should whole brain radiotherapy (WBRT) be quality of life measure, the impact of this study is limited by its small avoided for this reason? There is a large literature on the cognitive size and short follow-up period. The most critical conclusions pertain- effects of WBRT in children and adults, after prophylactic cranial radio- ing to cognitive decline are based on a total of five patients followed for therapy in small-cell lung cancer, and in leukemia patients. The find- only 1 year. Because there is a latency of several years before the onset ings of these studies can be conflicting. Some show mild to minimal of noticeable neurocognitive decline with other forms of brain irradia- deficits; others indicate more pronounced disability. Avoiding WBRT in tion (WBRT), it is quite possible that meaningful cognitive deterioration the brain metastasis population and using radiosurgery alone is a con- was missed in the present investigation. Also, patients would ideally be cept with growing popularity. WBRT may not be necessary for tumor assessed by a complete battery of neurocognitive tests at multiple time control and it can cause cognitive problems in some patients who have points over an extended period of time rather than just by those tests extended survival. It also causes hair loss and can result in fatigue or that were worse in the acute phase. other symptoms, all of which our patients tell us about (1). Some other factors purported to affect neurocognitive outcomes are Although this study calls attention to this important question, it tumor progression, chemotherapy, surgery, and antiepileptic drugs. A does not provide an answer. I hope these investigators will continue more thorough study of how such factors affected neurocognition their research, particularly on those patients who survive longer than would have been interesting. 1 year. With tests for cognition, executive functioning, and quality of Despite these flaws, however, this is an important pilot study. life, their methodology is sound. I think we need to refine these tests Because of the significant challenges intrinsic to evaluating patients and make them more streamlined for this population, which would with metastatic cancer, there are few prospective studies that attempt encourage others to begin research in this area. to answer the primary question posed in this study. Moreover, the authors have arguably been able to demonstrate stable to improved Douglas Kondziolka neurocognitive outcomes in patients with brain metastases 1 year after Pittsburgh, Pennsylvania treatment with radiosurgery alone. Given the importance of the answer, confirmatory studies with larger patient populations and longer follow-up periods are warranted. 1. Kondziolka D, Niranjan A, Flickinger JC, Lunsford LD: Radiosurgery with or without whole-brain radiotherapy for brain metastases: The patients’ perspec- John R. Adler, Jr. tive regarding complications. Am J Clin Oncol 28:173–179, 2005. Stanford, California

284 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CLINICAL STUDIES

SURGERY FOR TEMPORAL MEDIOBASAL TUMORS: EXPERIENCE BASED ON A SERIES OF 235 PATIENTS

Johannes Schramm, M.D. OBJECTIVE: To describe the clinical characteristics, diagnosis, various approaches, and Department of Neurosurgery, outcomes in a retrospective review of a large series of temporomediobasal (TMB) tumors. University of Bonn Medical Center, Bonn, Germany METHODS: Charts from 235 patients with TMB tumors were identiﬁed from the glioma and epilepsy surgery database and from the electronic operations log. Preoperative Ales F. Aliashkevich, M.D. magnetic resonance imaging scans were available for all patients and postoperative Department of Neurosurgery, follow-up was available for 155 of these patients (mean follow-up period, 59 mo; range, University of Bonn Medical Center, 2–172 mo). Preoperative symptoms, approaches, technical problems, and surgical com- Bonn, Germany plications are described.

Reprint requests: RESULTS: Two hundred and thirty-five patients with intra-axial TMB tumors (mean age, Johannes Schramm, M.D., 35 yr) were collected during an 11-year period. The largest tumor groups were astrocy- Department of Neurosurgery, tomas (38.0%), gangliogliomas (29.8%), dysembryoplastic neuroepithelial tumor (11.1%), University of Bonn Medical Center, and glioblastomas (11.1%). The most frequent tumor location was the mesial Type A Sigmund-Freud-Straße 25, 53105 Bonn, Germany. tumor (45.1%), with this type also showing the highest proportion of benign (World Health Email: Johannes.Schramm@ Organization Grades I and II) histological features (91.3%). Of all tumors, 76.2% were benign. ukb.uni-bonn.de Larger tumor size was associated with higher frequency of malignant histopathological findings. The leading symptom was epilepsy in 91% of patients, followed by drug-resistant Received, December 27, 2005. Accepted, October 13, 2006. epilepsy in 71.5%. Significant preoperative neurological deficits, such as hemiparesis or aphasia, were seen in 3.8% of the patients; another 12% had visual field deficits. Thirty- eight patients with low-grade tumors had undergone surgery previously. Several surgical approaches were chosen: transsylvian in 28%, anterior two-thirds temporal lobe resection in 23%, temporal pole resection in 15.3%, subtemporal in 19%, and transcortical in 6%. The most frequent neurological complications were transient: dysphasia (4.2%), hemiparesis (5%), and oculomotor disturbance (2.5%). Permanent nonvisual neurological complications occurred in fewer than 2% of the patients and significant new hemi- anopic defects were found in another 5.4% of the patients. The most severe complication was one intraoperative internal carotid artery lesion. One patient died. CONCLUSION: Small tumor size, magnetic resonance imaging, and microsurgery have made resection of mostly benign TMB tumors possible in a large number of patients. This series supports the conclusion that these tumors can be operated on with a relative degree of safety for the patient, provided that the anatomy of the mesial temporal lobe and the variety of approaches are well known to the surgeon. However, because of the complex anatomic structures in the vicinity, transient neurological deterioration is not infrequent and certain neurological disturbances (e.g., quadrantanopia) even seem to be unavoidable, whereas permanent significant deficits are rare. KEY WORDS: Glioma, Hippocampus, Limbic glioma, Temporal lobe tumor, Temporomesiobasal lobe

Neurosurgery 60:285–295, 2007 DOI: 10.1227/01.NEU.0000249281.69384.D7 www.neurosurgery-online.com

ew series of temporomesiobasal (TMB) tumors have been approach, which was followed in 2002 by a series with 40 published. After the ﬁrst article was published by Yas¸argil patients by Russell and Kelly (26). In 1994, Fried et al. (11) pub- Fand Reeves (40), the series was enlarged in Yas¸argil’s (39) lished a series of long-term epilepsy-associated tumors, includ- chapter on limbic and paralimbic tumors, in which 132 lesions ing 22 TMB tumors. Most patients in the series had neocortical comprising 120 tumors were included. Weiner and Kelly (37) tumors and the focus was on surgery to resolve chronic ﬁrst published results from seven patients using a stereotactic epilepsy, similar to a series from the Montreal group (10) that

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 285 SCHRAMM AND ALIASHKEVICH

included 17 patients with posteriorly located TMB tumors but T2-weighted sequences. Most of the axial sections were angu- limited the resection to anterior cortical areas and the hippocam- lated either along the length axis of the hippocampus or the pus because “these lesions . . . may be relatively inaccessible anterior commissure-posterior commissure plane (35). because of their location” (10, p 1781). Clusmann et al. (4) pub- Patients with drug-resistant epilepsy in whom an additional lished a series of 78 patients with TMB lesions, of which 58 resection for epileptological reasons was a possibility (e.g., were tumors, focusing more on the epilepsy surgery aspect and hippocampectomy) were studied additionally in accordance less on surgical technique. Only the tumor series by Yas¸argil (39) with our recently published protocol, including axial and and Russell and Kelly (26) included surgical aspects, such as the coronal fluid-attenuated inversion recovery sequence (35). approach. Ture and Pamir (32) reviewed the technical challenges This group also underwent a presurgical evaluation as of removing a temporomesial lesion. In all published series, described previously (16, 20). epilepsy was a prominent sign of the underlying lesion. Although two articles from our group were devoted to the spe- Tumor Classification cific group of long-term epilepsy-associated tumors (17, 18) and We classified these patients according to a proposed classifi- to the epileptological considerations of mesiobasal temporal cation system for TMB tumors (J Schramm, A Aliashkevich, lobe lesional epilepsy (4), only the latter of the two publications unpublished data). This tumor classification distinguishes four briefly related to epilepsy surgery technique. Aspects of loca- tumor types: Type A tumors occur in the most mesiobasal tion, sizes, surgical complications, and how to choose an structures (uncus, amygdala, hippocampus, parahippocampus, approach were not addressed in our previous publications. lingual gyrus); Type B tumors are immediately lateral to these In the conclusions of his chapter on limbic and paralimbic structures but exclude the inferior and middle temporal gyrus; tumors, Yas¸argil stated that “many understandable reasons that Type C tumors occupy the regions of Type A and B tumors; mitigate against surgery in limbic and paralimbic regions are Type D tumors originate in the areas of Types A, B, or C and considered by neurosurgeons” (39, p 276). Although this pub- extend through the temporal stem into the lateral basal ganglia lication is only 10 years old, he concluded at the time that area or the subcortical insular area (Figs. 1–5). Tumor Types A “strategies for biopsy followed by radiotherapy are generally through D were further subdivided into groups with more advocated.” However as the published series demonstrate, the anteriorly or more posteriorly placed tumors. TMB area obviously has changed from an area at which many surgeons did not try complete tumor removal to an area from RESULTS which many lesions are now removed, and many of those completely. Frequently, these operations were triggered by the Patients and Symptoms necessity to treat drug-resistant epilepsy (11), as in the series presented herein in which drug-resistant epilepsy was the most The mean patient age was 35 years, the male-to-female ratio frequent reason to operate (168 out of 235 patients, 71.5%). was 131:104, and the left-to-right localization ratio was 124:112. Although this article includes a large proportion of tumors The main signs or symptoms leading to diagnosis were associated with long-term epilepsy, it is not focused on aspects seizures in 214 patients. In 168 of these, epilepsy was drug- of epilepsy surgery such as presurgical evaluation (16, 20, 35), resistant and lasted for longer than 2 years. Another 46 patients but instead presents a large surgical series of TMB tumors had occasional and treatable seizures. Preoperative focal describing the spectrum of symptoms, histological results, sur- deficits, including hemiparesis (n ϭ 3), dysphasia (n ϭ 2), gical approaches, and complications. hemianopia (n ϭ 9), and paresthesia (n ϭ 1), were rare. Prominent cognitive impairment was seen in two patients, and another two patients reported impaired memory. Nonspecific PATIENTS AND METHODS symptoms, such as headache and vertigo, led to discovery of the tumor in three patients; in another three patients, the tumor Patients was an incidental finding. Symptoms were mild in most From January 1994 through December 2003, a series of 238 patients. patients harboring tumors in the TMB area were seen. Their Thirty-eight patients had undergone surgery previously; in data were retrieved from the operative database, the glioma 37 patients, the procedure was performed outside of our database, and the epilepsy surgery database kept in the depart- department. In nine of these patients, this was a previous ment. Patient and outpatient charts were reviewed. Three biopsy, and two of these patients previously underwent radia- tumors were of extraaxial origin, thus a total of 235 intraaxial tion treatment. Of the 29 patients who underwent previous TMB tumors are included in the evaluation. resections, two more patients also had undergone radiation.

Radiological Evaluation Tumor Type and Histological Results Preoperative magnetic resonance imaging (MRI) scans were The distribution of histopathological diagnoses is shown in available for all patients and postoperative MRI scans were Table 1. There were 46.8% Grade I tumors, 29.4% Grade II tumors, available for 203 patients. Preoperative MRI scans were 10.2% Grade III tumors, and 11.1% Grade IV tumors, as well as obtained using a 1.5-T system with axial and coronal T1- and 0.8% metastases. Thus, malignant tumors made up 22.1% of the

286 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SURGERY FOR TEMPORAL MEDIOBASAL TUMORS

AB AB

C D C D

FIGURE 1. MRI scans obtained from a 34-year-old woman with a pilocytic FIGURE 2. MRI scans of a ganglioglioma (World Health Organization Grade astrocytoma before surgery (A and B) and 11 years after surgery (C and D). I) Type Bp tumor obtained from a 45-year-old woman. A and B, preoperative This was a Type Da tumor with a pronounced posterior mainly intraventric- scans demonstrating that the parahippocampal gyrus was spared as well as ular extension. Removal was carried out via an anterior two-thirds lobec- the inferior temporal gyrus. C and D, postoperative scans demonstrating the tomy, including more dorsal mesial lobe parts. preserved structures after a subtemporal approach. total and low-grade tumors comprised 77.8%. The most frequent were larger than 6 cm. The highest proportion of small tumors histological type was astrocytoma (89 patients; 38%) followed by was seen in Type A tumors. ganglioglioma (70 patients; 29.8%), DNT (26 patients; 11.1%), The distribution of histological grades versus tumor types is glioblastoma (26 patients; 11.1%), and oligodendroglioma shown in Table 2. It is obvious that the larger Type D tumors are (20 patients; 8.6%), with other types being much rarer. malignant more often (42.9% Grades III and IV) than the smaller The distribution of tumor types as described in the TMB tumors restricted to Type A or Type B. Of Type A tumors, only tumor classification scheme (J Schramm, A Aliashkevich, 8.7% were malignant; of Type B tumors, 28.2% were malignant. unpublished data) was as follows. Tumors located most If we look at the frequency of malignant versus nonmalignant mesially in the temporal lobe (Type A) were most frequent tumors for the different tumor types, allocortical Type A tumors (106 out of 235 patients, 45.1%). There were 41 Type B tumors in 64.4% of patients were Grade I tumors and another 27% were immediately lateral to the Type A area (17.4%). The larger Type Grade II tumors, i.e., they were benign in 91.3%. The neocorti- C tumors, covering the area of both Type A and B tumors, cal tumors of Types B and C were benign in 71.8% of the were found in 52 patients (22.1%). Type D tumors, with exten- patients and only 57% of grade D tumors were benign. Thus, it sion into the temporal stem and lateral insular area, were seems that most mesially located tumors (i.e., Type A) of allocor- found in only 15% of the patients (36/235). Only 26% of the tical origin are much more frequently benign. patients had tumors that (n ϭ 62) were located posteriorly (Types A, B, C, D, with sufﬁx “p”), and 74% of patients had Surgical Aspects tumors that (n ϭ 173) were located anteriorly (Types A, B, C, In all Type A, B, C, and D tumors, radical microsurgical D, with sufﬁx “a”). resection was intended. In Type D tumors, incomplete radical- Of the 235 tumors, 57 (24.3%) were 2 cm or smaller, 96 ity in the area of the temporal stem or the lateral basal ganglia (40.9%) were between 2.1 and 3.5 cm, 53 (22.6%) were between was accepted if, during surgery, parts of the tumor appeared 3.6 and 5 cm, 29 (12.3%) were larger than 5 cm, and only six inaccessible or important vascular structures were encountered.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 287 SCHRAMM AND ALIASHKEVICH

AB AB

C D C D

FIGURE 3. MRI scans demonstrating an anaplastic astrocytoma (World Health Organization Grade III), tumor Type Cp, with an anterior extension FIGURE 4. MRI scans of a ganglioglioma (World Health Organization Grade reaching the uncus. A and B, preoperative scans. C and D, postoperative I), tumor Type Ap, obtained from a 36-year-old man. A and B, preoperative scans obtained after a subtemporal approach demonstrating the extent of resec- scans demonstrating the location of the tumor just behind the anteroposterior tion with preservation of most of the neocortical part. D, coronal cut demon- partition confined to the parahippocampal gyrus and the hippocampus. C and strating the small and incomplete basal resection of the inferior temporal D, postoperative scans demonstrating the extent of resection after a pure sub- gyrus. temporal approach preserving the lateral neocortical inferior temporal gyrus and the fusiform gyrus. There were no seizures and no residual tumors at 5, 7, and 8 years of follow-up. The spectrum of approaches was wide and well differentiated (Table 3). Temporal lobe resection was either an anterior two- thirds resection (23%) or a pole resection (Ͻ3.5 cm; 15.3%). Thus, the most frequently used approaches were two variants of ante- the tumor had grown into the perimesencephalic cistern, pro- rior temporal lobectomy if counted together (90 patients; 38.3%), truding between blood vessels and cranial nerves found there. ϭ followed by the transsylvian approach (66 patients; 28%). A total More than half of these patients (n 17) had Type C tumors. of 66% of the approaches, therefore, belong to classical resection In 10 patients, problems were observed above the level at types (Fig. 1). which they can be handled easily. In one patient, rupture of the Subtemporal approaches either with or without narrow and internal carotid artery was encountered; in another, the brain- partial gyral resection were used in 45 patients (19%); in three stem surface was infiltrated. The vascular injury resulted in a patients these were in combination with an additional sylvian transient ischemic deficit that resolved completely despite a small approach. Transcortical approaches, partly with resection of infarct. The patient was seizure free and without recurrence at the smaller parts of these gyri, were used in only 6% of the patients 6-year follow-up evaluation. In another patient, major bleeding (n ϭ 14), although there were 35 Type D and 41 Type C tumors. from the main stem of the basal vein of Rosenthal occurred; in In 19 patients (8%), an approach through the old resection cav- five patients, some bleeding from deep perforators complicated ity was used and in one patient a dorsal interhemispheric surgery. In two patients, numerous and large branches of the approach to a lesion in the lingual gyrus was used. vein of Labbé made the extrapial subtemporal approach impossible. Special Surgical Problems In 150 patients, presurgical evaluation for drug-resistant In 29 patients, the growing tumor had destroyed the mesial epilepsy (16, 20) led to the recommendation of additional par- arachnoid cover of the parahippocampal gyrus or uncus and tial or classic hippocampal resection (similar to selective amyg-

288 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SURGERY FOR TEMPORAL MEDIOBASAL TUMORS

AB TABLE 1. Distribution of histopathological diagnoses for 235 patients Pathological results No. % No. % Grade I Ganglioglioma 66 28.1 Dysembryoplastic neuroepithelial 26 11.1 tumor Pilocytic astrocytoma 15 6.4 Astrocytoma 3 1.3 Total 110 46.8 Grade II Pleomorph xanthoastrocytoma 4 1.7 C D Oligoastrocytoma 14 6.0 Oligodendroglioma 18 7.7 Ganglioglioma 3 1.3 Astrocytoma 30 12.8 Total 69 29.4 Total benign tumors (Grades I and II) 179 76.2 Grade III Pleomorph xanthoastrocytoma 1 0.4 Anaplastic oligoastrocytoma 5 2.1 Anaplastic oligodendroglioma 2 0.9 Anaplastic astrocytoma 15 6.4 Anaplastic ganglioglioma 1 0.4 Total 24 10.2 Grade IV FIGURE 5. A and B, preoperative MRI scans demonstrating an oligoden- Glioblastoma 26 11.1 droglioma (World Health Organization Grade II), Type Da, compressing the Total 26 11.1 brainstem and transgressing into the basal ganglia, a good case for a transsyl- Total malignant gliomas (Grades 50 21.3 vian approach. C and D, MRI scans demonstrating a dysembryoplastic neu- III and IV) roepithelial tumor (World Health Organization Grade I) of the hippocampus, Metastasis 2 0.8 2 0.8 sparing most of the parahippocampal gyrus. This patient had previous pole Not retrievable 4 1.7 4 1.7 resection in another hospital and was not seizure free. Total 235 dalohippocampectomy), usually including the major part of the amygdaloid body, and only occasionally leaving the amyg- and was based on ictal video-electroencephalography ﬁndings, daloid body behind. Aspects of epilepsy surgery for tumors MRI findings, and semiological results. In chronic epilepsy are not the main focus of this article, but are described and patients, the hippocampus was resected with the next half cen- discussed in several publications (4, 5, 27, 42). In drug-resistant timeter if the tumor was located in or invading that structure. epilepsy patients, the extent of resection, i.e., inclusion of the If the tumor was located nearby, frequently an invasively hippocampus, was decided by the presurgical evaluation (16) recorded ictal electroencephalography was used for this deci-

a(229 ؍ TABLE 2. Tumor types versus tumor grades (n Total Grade I Grade II Grade III Grade IV Grade I ؉ II Grade III ؉ IV Tumor type No. % No. % No. % No. % No. % No. % No. % A 104 45.4 67 64.4 28 26.9 5 4.8 4 3.8 95 91.3 9 8.7 B 39 17.0 17 43.6 11 28.2 5 12.8 6 15.4 28 71.8 11 28.2 C 51 22.3 14 27.5 22 43.1 8 15.7 7 13.7 36 70.6 15 29.4 D 35 15.3 12 34.3 8 22.9 6 17.1 9 25.7 20 57.1 15 42.9 Total 229 110 48.0 69 30.1 24 10.5 26 11.4 179 78.2 50 21.8

a Two metastases and four tumors of nonretrievable histological diagnoses excluded.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 289 SCHRAMM AND ALIASHKEVICH

TABLE 3. Distribution of approaches for 235 patientsa Approach No. % No. Total (%) Transsylvian 66 28.1 Partial anterior resection 90 38.3 Temporal pole resection (<3.5 cm) 36 15.3 Classic anterior two-thirds temporal lobe resection 54 23.0 All subtemporal 45 19.1 Pure subtemporal 42 17.9 Without resection 12 5.1 With small gyral resection 30 12.8 Combined transsylvian and subtemporal 3 1.3 Without resection 2 0.9 With small gyral resection 1 0.4 Cortical 14 6.0 STG 5 2.1 MTG 6 2.6 STG + MTG 1 0.4 MTG + ITG 2 0.9 Interhemispheric (to dorsal parahippocampal and lingual gyri) 1 0.4 Via old resection 19 8.1 Only through cavity 5 2.1 With additional two-thirds temporal lobe resection 13 5.5 With transsylvian approach 1 0.4 Total 235

a STG, superior temporal gyrus; MTG, middle temporal gyrus; ITG, inferior temporal gyrus. sion. In normal oncological patients with occasional or no Visual Fields seizures, the hippocampus or amygdala was resected only if Pre- and postoperative visual field findings were available the tumor was located within the structure or invaded into this for 168 patients. In 19 (12%) out of 168 patients, field cuts were structure. Sometimes presurgical evaluation by the epileptolo- present before surgery. Visual fields remained unchanged in 98 gist led to the recommendation to preserve the hippocampus (58.3%) out of 168 patients and deteriorated in 70 (41.7%) out of because of well-preserved verbal memory—also a surgical 168 patients. A significant deterioration from normal fields to challenge (Fig. 2). Twenty-nine patients had undergone previ- hemianopia occurred in seven (4.2%) out of 168 patients. Two ϭ ous open resections (28 in outside hospitals); in most (n 19), more patients deteriorated from quadrantanopia to hemi- the approach to the second surgery was performed via the old anopia, resulting in a total of nine significant new visual field resection cavity. Of 10 patients who underwent a previous cuts that were noted by the patients themselves (5.4%). In 57 resection, in whom the reoperation was not performed through patients, a new superior quadrantanopia could be diagnosed the old defect, nine had been performed via the transsylvian only by a computer-assisted perimetry testing, representing an approach and one through a subtemporal approach. unwanted nondebilitating side effect rather than a significant complication. Surgical complications were observed in 30 Complications patients, usually without permanent sequelae except for a few A total of 55 patients were affected by one of the complica- patients, which led to and were counted among neurological tions, corresponding to 23% of the total group (Table 4). permanent complications. Complications were divided into surgical, neurological with Permanent severe sequelae were observed in only four hemianopias, and medical. Medical complications were always patients (1.7%). In 5.4% of patients, significant new field cuts transient and were observed in six patients (2.6%). Neuro- were observed. One death occurred after an unclear coma logical complications (excluding hemianopias) were seen in (0.4%) in a 63-year-old patient with a large glioblastoma. There 17.4% of patients, affecting 35 patients, with more than one was no intraoperative problem and after an initially uneventful side effect possible in the same patient. The subtle neuropsy- recovery, a progressive unexplainable loss of consciousness chological deteriorations detectable only after extensive testing developed after 9 days. Having ruled out infection, infarction, commonly used in epilepsy surgery patients were not evalu- and deep invasion, repeated imaging and laboratory studies ated because these tests were not performed in classic oncology never found a comprehensible explanation for this develop- patients. ment. Death was the result of pneumonia.

290 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SURGERY FOR TEMPORAL MEDIOBASAL TUMORS

gical complications did occur, most often related to vascular -a injury. Although five hematomas were observed, no perma(55 ؍ TABLE 4. Side effects and complications (n Temporary Permanent nent damage resulted from them. The analysis of complica- Complications tions allows the conclusion that difficult TMB tumors carry a No. % No. % certain low risk for neurological impairment, rarely including Surgical complications in 30 significant neuropsychological problems. patients (13%) Intraoperative ICA ruptureb 1 0.4 Small intracerebral hemorrhage 1 0.4 DISCUSSION (basal ganglia) Cerebellar subarachnoid 1 0.4 A limitation in the analysis of this tumor series is the fact that hemorrhage these surgeries were not planned with the retrospectively Perforator infarctionb 3 1.3 applied tumor grading system in mind. Instead, this is a retro- Isolated temporal horn, 1 0.4 Hydrocephalus, shunt spective analysis in which tumor outcomes were analyzed in a Meningitis 12 5.1 post hoc fashion. Bone flap infection 1 0.4 It is evident that subtotal temporal lobe resections in this Subdural hematoma 1 0.4 and other series still play an important role in surgery for TMB Epidural hematoma 2 0.9 tumors, frequently because of epileptological considerations Brachial plexus lesion 1 0.4 (11). However, one-fifth of these tumors have been approached Frontal branch of facial nerve 3 1.3 using subtemporal approaches (26) and 28% have been damaged approached only via a transsylvian approach. The low fre- Middle ear effusion 2 0.9 Subcutaneous CSF effusion 3 1.3 quency of transcortical approaches reflects the difficult CSF leak 1 0.4 mesiobasal location of most of these tumors and the intention Total 32 13.6 1 0.4 to preserve the optic pathway. Neurological complications in 35 It should be remembered that there is a certain degree of patients (16%)c variability with important anatomic structures that frequently Hemipareses 12 5.1 2 0.9 serve as landmarks. Ono et al. (22) and Novak et al. (21) noted Memory disturbances 4 1.7 that not all sulci at the base of the temporal lobe are consistent, Speech disturbances 14 6.0 1 0.4 with the collateral sulcus being the only one present in its typ- Ocular movement disturbances 5 2.1 1 0.4 Ataxia 2 0.9 ical position in 100% of patients. The thickness and texture of Behavioral disturbances 4 1.7 the hippocampus can vary considerably, depending on degree Total 41 17.4 4 1.7 of tumor infiltration or duration of seizure disorder. In chronic Significant visual field defectsc temporal lobe epilepsies, corresponding gyri are frequently a Permanent hemianopia 9 5.4 bit smaller on the affected side (21). From the literature, it Medical complications in six seems that the anterior extent of Meyer’s loop is somewhat patients (2.6%) variable. We, therefore, should not take simple numbers for Bronchitis 1 0.4 distances as reliable truth when planning an approach. The Pneumonia 1 0.4 Diabetes insipidus 1 0.4 integrity of the arachnoid layer on the mesiobasal surface of the Hyponatremia 1 0.4 temporal lobe cannot be assumed to be perfect because tumors Gastrointestinal infection 1 0.4 may perforate the arachnoid layer and grow in a cauliflower- Pulmonary embolism 1 0.4 like fashion, protruding between perimesencephalic structures, Total 6 2.6 such as major vessels and cranial nerves, and even caudally Unclear coma and death 1 0.4 pointing well beyond the rim of the tentorial notch. In small a ICA, internal carotid artery; CSF, cerebrospinal fluid. More than one side children, the mesial arachnoid cover is particularly vulnerable effect possible in the same patient. and clean subpial dissection is often not possible. b Partly associated with transient or permanent neurodeficit indicated below. c Quadrantanopias considered separately. Approaches As discussed previously (J Schramm, A Aliashkevich, unpub- Thus, in this truly difficult tumor group, temporary neuro- lished data) and demonstrated in Table 3, a variety of logical impairment is relatively frequent, occurring in 15.7% of approaches has been used in this series that are quite different our patients in whom temporary hemiparesis and speech dis- from that used in Yas¸argil ’s paralimbic tumor series. A consid- turbances were observed in 5 and 6%. Severe medical compli- erable part of the difference is caused by the necessity of addi- cations were infrequent and could all be handled without per- tional resections for epileptological reasons in our series. In his manent sequelae. Surgical complications were observed in 13% chapter on limbic and paralimbic tumors, which included 120 of the patients, partly overlapping with the patients experienc- true TMB tumors, Yas¸argil et al. (41) proposed that “these ing neurological complications. Permanent sequelae from sur- tumors can be approached and extirpated via the transsylvian

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 291 SCHRAMM AND ALIASHKEVICH

approach without recourse to cortical incision in superficial a significant difference in postoperative neurological status layers or temporal lobectomy” (41, p 42) Yas¸argil pointed out between three tumor sizes. that one dorsally located lesion in the lingual gyrus and the Other series of TMB tumors (6, 11, 26, 37, 39) allow a more posterior part of the parahippocampal gyrus was removed precise discussion of complications. There were no deaths in through a posterior interhemispherical approach. TMB tumors the series of Fried et al. (11) (n ϭ 22), Weiner and Kelly (37) may be approached in a variety of ways, both classical ways (n ϭ 7), Clusmann et al. (6) (n ϭ 58 tumors), and Yas¸argil (39). and recently described ways. In the past 10 years, several pub- One death occurred among the 40 patients of Russell and Kelly lications have been devoted to the description of various forms (26), compared with one death in this series. Hemipareses were of subtemporal approaches (24, 25), posterolateral approaches, seen in these TMB tumor series: 2.7% (6), 2.5% (26), and 5% (39), including those with resections of the petrous bone (14, 26, 28, which are rates comparable to this series. 31, 32), and even posterior midline approaches (23, 29, 33, 36, Because temporal lobe lesions carry a particular risk for the 41). Many authors described these approach variants reaching visual pathways, it may be acceptable to consider those sepa- for aneurysms on the large vessels circumventing the brainstem rate from other neurological complications. Considering the (12, 15, 19, 31, 34) or for arteriovenous malformations (7, 8, 13, risk for a new quadrantanopia, although this is considered by 30), but they are just as useful for small tumors. It remains a many not to be a significant complication, but rather a nondis- fact that the different tumor types defined by the new classifi- turbing acceptable side effect, it is remarkable that 28% of cation (J Schramm, A Aliashkevich, unpublished data) are asso- patients with a subtemporal approach had a new quadran- ciated with the preferential use of certain approaches. This tanopia (Fig. 5). Only two authors of previous TMB tumor seems to underline the usefulness of the proposed classifica- series mentioned preoperative visual field defects: Russell and tion. The most frequent approach in Type A tumors was trans- Kelly (26) in seven out of 40 patients (17.5%) and Yas¸argil (40) sylvian; in Type B tumors, the subtemporal was most frequent in 19 out of 97 patients (19.5%), as compared with 12% in this (Fig. 5); in Type C tumors, anterior two-thirds resection was series. Postoperative increase in field defects was specified by most frequent; and in Type D tumors, anterior lobe resection several authors: one temporary field decrease in seven patients was most frequent. (37), two hemianopias in 78 lesions (6), up to 7.5% quadrantanopia (26), and 21 quadrantanopias (16%) and 13 hemi- Complications anopias (10%) in Yasargil’s later series of 139 patients (39). In If one compares the incidence of complications in this tumor the series presented here, there were more quadrantanopias series, one has to take into account that the series may differ (36.3%) but fewer hemianopias (5.4%). from published glioma surgery series not confined to the TMB In Table 5, the surgical approaches are correlated with the side area (1–3). Two recent publications (1, 3) include tables with effects, including transient disturbances, for a total of 212 overviews of complication rates in glioma surgery. Brell et al. patients. Looking at the relative distribution of complications (1) lists five studies published between 1994 and 1999 reporting across the spectrum of resection types, it immediately becomes mortality rates between 1.7 and 5% in series of more than 200 clear that temporal pole resections and subtemporal approaches patients. The morbidity rates in those and other series varied carry a lower risk for a hemiparesis than both the transsylvian between 8 and 31.9%, usually being higher than 20% (range, and the classic anterior two-thirds lobe resection, which carry a 1–3, 9). In the Glioma Outcome Project, Chang et al. (3) similar risk of hemiparesis. This, however, does not imply that described a 24% rate of perioperative complications in a group the approach causes these unwanted effects. Concerning speech of first craniotomies and a 1.5% mortality rate. The mortality disturbances, transsylvian and two-thirds anterior lobe resection rate of 0.4% in this series compares favorably with that of again have a similar risk; however, the subtemporal approach Chang et al., but several favorable factors are relevant: the per- also carries a similar risk for speech disturbance. centage of malignant tumors and of large tumors was low; It is also quite clear that the temporal pole resection carries the infratentorial patients were not included; and the mean age in lowest chance for any of the four listed side effects. Most signif- our patient group indicates that most patients were younger icantly, perhaps, is the close similarity in the frequency of com- than 60 years. The overall complication rate of 23% in this plications when the transsylvian approach is compared with the series sounds high initially, but is comparable to that of other classic anterior two-thirds lobe resection. It can be concluded for published series (27.5% in Brell et al. and 24% in Chang et al.). this series that, after a TMB tumor was resected via the transsyl- Permanent new neurological deficits, apart from 5.4% signif- vian approach, which is usually considered to be more subtle icant new field defects, were found in our series in 1.7% of the and more minimalistic approach, the patient had similar risk for patients, compared with 8% in the Glioma Outcome Project unwanted side effects as if the patient had undergone the classic series. In that series, postoperative neurological worsening was anterior two-thirds temporal lobe resection. observed in 8 to 20% of patients, but these series partly The detailed knowledge of microsurgical anatomy, as out- included metastases and a higher number of Grade IV gliomas. lined by Wen et al. (38), is important. The surgeon must be The significance of the higher frequency of smaller tumors in familiar with specific tasks for TMB tumor surgery: to empty this series is difficult to assess. In the Glioma Outcome Study the uncus leaving the arachnoid intact, thus sparing the vessels series, univariate analysis of tumor size and multivariate analy- and IIIrd cranial nerve. The surgeon should be able to divide sis of preoperative factors, including tumor size, did not reveal the vessels entering the hippocampal sulcus and to leave the

292 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SURGERY FOR TEMPORAL MEDIOBASAL TUMORS

(212 ؍ TABLE 5. Surgical approaches versus side effects (n Side effects and neurological complicationsb Operations Speech Ocular movement Hemianopia Quadrantanopia Hemiparesis disturbances disturbances Approacha Column Column Row Column Row Column Row Column Row Column Row No. % No. % % No. % % No. % % No. % % No. % % Transsylvian 66 31.1 4 66.6 6.1 17 31.5 25.8 5 41.7 7.6 4 36.3 6.1 3 60.0 4.5 Classic anterior two-thirds 54 25.5 1 16.7 1.9 16 29.6 29.6 5 41.7 9.3 3 27.3 5.6 2 40.0 3.7 temporal lobe resection Temporal pole resection 36 17.0 5 9.3 13.9 1 8.3 2.8 1 9.1 2.8 Subtemporal 42 19.8 1 16.7 2.4 12 22.2 28.6 1 8.3 2.4 3 27.3 7.1 Transcortical 14 6.6 4 7.4 28.6 Total 212 6 2.8 54 25.5 12 5.7 11 5.2 5 2.4

a Twenty-three approaches via old resection (n ϭ 19), combined (n ϭ 3), and interhemispheric approaches (n ϭ 1) excluded. b Multiple events in one patient possible, including transient complications.

vein of Rosenthal and arteries intact. He should find the tem- approach via the transsylvian route. More mesially located Type poral horn from the sylvian fissure by entering through the A tumors are more frequently benign (World Health temporal stem at the inferior circular sulcus. For posterior sub- Organization Grades I and II), as are smaller tumors. A range of temporal approaches, the vein of Labbé must be preserved. If different approaches was used. Most of these tumors can be the temporal horn is to be entered transcortically from the lat- operated on with a relative degree of safety for the patient, but eral aspect of the temporal lobe, the ventricle should be entered the creation of new visual field defects is common. Significant through its floor if possible (Figs. 3 and 4). Anterior one-third or permanent neurological deficits were rare and the mortality was two-third lobectomy should be performed with the arachnoid low. A detailed knowledge and good orientation in this specific layer over the intact sylvian vessels. microsurgical anatomy of the mesial temporal lobe is necessary. Concerning the orientation when entering the ventricle from the inferior circular sulcus of the sylvian cistern, a too-mesial REFERENCES entry into the temporal stem not passing through the temporal horn may injure the optic radiation or perforating vessels. For 1. Brell M, Ibanez J, Caral L, Ferrer E: Factors influencing surgical complications defining resection borders (J Schramm, A Aliashkevich, unpub- of intra-axial brain tumours. Acta Neurochir (Wien) 142:739–750, 2000. lished data), the limen insulae-choroidal point line, which sep- 2. Cabantog AM, Bernstein M: Complications of first craniotomy for intra-axial brain tumour. Can J Neurol Sci 21:213–218, 1994. arates the anterolateral part of the amygdaloid body from the 3. Chang SM, Parney IF, McDermott M, Barker FG 2nd, Schmidt MH, Huang W, mesiodorsal part of the amygdaloid body in the temporal stem, Laws ER Jr, Lillehei KO, Bernstein M, Brem H, Sloan AE, Berger M; Glioma is useful. It can be easily visualized when the surgeon looks Outcomes Investigators: Perioperative complications and neurological out- through the opened sylvian fissure into the opened temporal comes of first and second craniotomies among patients enrolled in the Glioma Outcome Project. J Neurosurg 98:1175–1181, 2003. horn. Simultaneously visualizing the plexal point at the most 4. Clusmann H, Kral T, Fackeldey E, Blumcke I, Helmstaedter C, von Oertzen anterior limit of the choroidal sulcus in the temporal horn and J, Urbach H, Schramm J: Lesional mesial temporal lobe epilepsy and limited the limen insulae running parallel to the M1 segment from the resections: Prognostic factors and outcome. J Neurol Neurosurg Psychiatry superiorly located plateau of the insular cortex down to the 75:1589–1596, 2004. inferomesially located uncal surface is of considerable help. 5. Clusmann H, Kral T, Gleissner U, Sassen R, Urbach H, Blumcke I, Bogucki J, Schramm J: Analysis of different types of resection for pediatric patients with Wen et al. (38) described a related line (carotid-choroidal line), temporal lobe epilepsy. Neurosurgery 54:847–860, 2004. which is traceable from the bifurcation of the internal carotid 6. Clusmann H, Kral T, Marin G, Van Roost D, Swamy K, Schramm J: artery or the proximal segment of the M1 to the inferior Characterization of hemorrhagic complications after surgery for temporal choroidal point. lobe epilepsy. Zentralbl Neurochir 65:128–134, 2004. 7. de Oliveira E, Tedeschi H, Siqueira MG, Ono M, Rhoton AL Jr, Peace D: Anatomic principles of cerebro-vascular surgery for arteriovenous malforma- CONCLUSION tions. Clin Neurosurg 41:364–380, 1994. 8. Du R, Young WL, Lawton MT: “Tangential” resection of medial temporal This large series of TMB tumors demonstrates that small lobe arteriovenous malformations with the orbitozygomatic approach. tumor sizes are frequent and that tumors may be isolated in the Neurosurgery 54:645–652, 2004. 9. Deleted in proof. uncus, the amygdala, or the hippocampus without affecting 10. Fish D, Andermann F, Olivier A: Complex partial seizures and small poste- adjacent structures. The series demonstrates that these tumors rior temporal or extratemporal structural lesions: Surgical management. can be attacked even if located paramesially and thus difficult to Neurology 41:1781–1784, 1991.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 293 SCHRAMM AND ALIASHKEVICH

11. Fried I, Kim JH, Spencer DD: Limbic and neocortical gliomas associated with 34. Ulm AJ, Tanriover N, Kawashima M, Campero A, Bova FJ, Rhoton A Jr: intractable seizures: A distinct clinicopathological group. Neurosurgery Microsurgical approaches to the perimesencephalic cisterns and related seg- 34:815–823, 1994. ments of the posterior cerebral artery: Comparison using a novel application 12. Gerber CJ, Neil-Dwyer G, Evans BT: An alternative surgical approach to of image guidance. Neurosurgery 54:1313–1328, 2004. aneurysms of the posterior cerebral artery. Neurosurgery 32:928–931, 1993. 35. Urbach H, Hattingen J, von Oertzen J, Luyken C, Clusmann H, Kral T, 13. Heros RC: Arteriovenous malformations of the medial temporal lobe. Surgical Kurthen M, Schramm J, Blumcke I, Schild HH: MR imaging in the presurgi- approach and neuroradiological characterization. J Neurosurg 56:44–52, 1982. cal workup of patients with drug-resistant epilepsy. AJNR Am J Neuroradiol 14. Hori T, Tabuchi S, Kurosaki M, Kondo S, Takenobu A, Watanabe T: 25:919–926, 2004. Subtemporal amygdalohippocampectomy for treating medically intractable 36. Vishteh AG, David CA, Marciano FF, Coscarella E, Spetzler RF: Extreme lat- temporal lobe epilepsy. Neurosurgery 33:50–57, 1993. eral supracerebellar infratentorial approach to the posterolateral mesen- 15. Kikuchi H, Miyamoto S, Nagata I, Taki W, Nakahara I: Surgical extirpation of cephalon: Technique and clinical experience. Neurosurgery 46:384–389, 2000. the posterior hippocampal arteriovenous malformation. Surg Neurol 37. Weiner HL, Kelly PJ: A novel computer-assisted volumetric stereotactic 47:251–257, 1997. approach for resecting tumors of the posterior parahippocampal gyrus. J 16. Kral T, Clusmann H, Urbach J, Schramm J, Elger CE, Kurthen M, Grunwald Neurosurg 85:272–277, 1996. T: Preoperative evaluation for epilepsy surgery (Bonn Algorithm). Zentralbl 38. Wen HT, Rhoton AL Jr, de Oliveira EP, Cardoso AC, Tedeschi H, Baccanelli M, Neurochir 63:106–110, 2002. Marino R Jr: Microsurgical anatomy of the temporal lobe: Part 1—Mesial 17. Luyken C, Blumcke I, Fimmers R, Urbach H, Elger CE, Wiestler OD, temporal lobe anatomy and its vascular relationships as applied to amygdalo- Schramm J: The spectrum of long-term epilepsy-associated tumors: Long- hippocampectomy. Neurosurgery 45:549–592, 1999. term seizure and tumor outcome and neurosurgical aspects. Epilepsia 39. Yas¸argil MG: Limbic and paralimbic tumors, in Yas¸argil MG (ed): Micro- 44:822–830, 2003. neurosurgery IV B. New York, Thieme Medical Publishers, 1996, pp 252–290. 18. Luyken C, Blumcke I, Fimmers R, Urbach H, Wiestler OD, Schramm J: 40. Yas¸argil MG, Reeves JD: Tumours of the limbic and paralimbic system. Acta Neurochir (Wien) Supratentorial gangliogliomas: Histopathologic grading and tumor recurrence 116:147–149, 1992. 41. Yas¸argil MG, von Ammon K, Cavazos E, Doczi T, Reeves JD, Roth P: Tumours in 184 patients with a median follow-up of 8 years. Cancer 101:146–155, 2004. of the limbic and paralimbic systems. Acta Neurochir (Wien) 118:40–52, 1992. 19. Miyamoto S, Kataoka H, Ikeda A, Takahashi J, Usui K, Takayama M, Satow 42. Yonekawa Y, Imhof HG, Taub E, Curcic M, Kaku Y, Roth P, Wieser HG, T, Hashimoto N: A combined subtemporal and transventricular/tran- Groscurth P: Supracerebellar transtentorial approach to posterior temporome- schoroidal fissure approach to medial temporal lesions. Neurosurgery dial structures. J Neurosurg 94:339–345, 2001. 54:1162–1169, 2004. 43. Zentner J, Hufnagel A, Wolf HK, Ostertun B, Behrens E, Campos MG, Elger 20. Nelles M, Koenig R, Kandyba J, Schaller C, Urbach H: Fusion of MRI and CT CE, Wiestler OD, Schramm J: Surgical treatment of neoplasms associated with subdural grid electrodes. Zentralbl Neurochir 65:174–179, 2004. with medically intractable epilepsy. Neurosurgery 41:378–387, 1997. 21. Novak K, Czech T, Prayer D, Dietrich W, Serles W, Lehr S, Baumgartner C: Individual variations in the sulcal anatomy of the basal temporal lobe and its relevance for epilepsy surgery: An anatomical study performed using mag- Acknowledgments netic resonance imaging. J Neurosurg 96:464–473, 2002. We thank the surgical colleagues who have been involved in the operations 22. Ono M, Kubik S, Abernathey CD: Atlas of the Cerebral Sulci. New York, Thieme described in this series over the years: Josef Zentner, M.D., Dirk van Roost, Medical Publishers, 1990, pp 14–16. M.D., Bernhard Meyer, M.D., Elga Behrens, M.D., and Carlo Schaller, M.D., 23. Panigrahi M: Supracerebellar transtentorial approach. J Neurosurg Priv.-Doz., and Dorothee Haun, Ph.D., for assistance with the figures and tables; 95:916–917, 2001. Petra Süßmann for help with follow-up; and Erika Heunemann for help with 24. Park TS, Bourgeois BF, Silbergeld DL, Dodson WE: Subtemporal manuscript preparation. transparahippocampal amygdalohippocampectomy for surgical treatment of mesial temporal lobe epilepsy. Technical note. J Neurosurg 85:1172–1176, 1996. COMMENTS 25. Robinson S, Park TS, Blackburn LB, Bourgeois BF, Arnold ST, Dodson WE: Transparahippocampal selective amygdalohippocampectomy in children and n this well-written article, the authors present an impressive series of adolescents: Efficacy of the procedure and cognitive morbidity in patients. J Ipatients (n = 235) with temporal lobe and insular tumors. Pure tem- Neurosurg 93:402–409, 2000. poral mediobasal tumors were found in 106 patients (45.1%). The lesion 26. Russell SM, Kelly PJ: Volumetric stereotaxy and the supratentorial occipito- was located within the fusiform gyrus in 41 patients (17.4%) and within subtemporal approach in the resection of posterior hippocampus and the fusiform and mediobasal temporal areas in 52 patients (22.1%), Neurosurgery parahippocampal gyrus lesions. 50:978–988, 2002. and within the fusiform and mediobasal temporal areas; in the remain- 27. Schramm J, Kral T, Grunwald T, Blumcke I: Surgical treatment for neocorti- ing 36 patients (15%), the tumors were multicompartmental and cal temporal lobe epilepsy: Clinical and surgical aspects and seizure outcome. J Neurosurg 94:33–42, 2001. involved the insula and caudal fronto-orbital region. 28. Sindou MP, Fobe JL: Removal of the roof of the external auditory meatus in This clear, precise, and detailed study confirms that microsurgical approaching the tentorial notch through a low temporal craniotomy. removal of gliomas in the previously mentioned areas is, in fact, the Technical note. J Neurosurg 74:520–522, 1991. most important variable to achieve a reasonably effective treatment of 29. Smith KA, Spetzler RF: Supratentorial-infraoccipital approach for posterome- these lesions. Surgery can be conducted with relative safety, and dial temporal lobe lesions. J Neurosurg 82:940–944, 1995. removal of the gliomas significantly reduces the intractable seizures 30. Stein BM: Arteriovenous malformations of the medial cerebral hemisphere experienced by these patients. We commend Schramm and Aliash- and the limbic system. J Neurosurg 60:23–31, 1984. kevich on their successful surgical results. 31. Terasaka S, Sawamura Y, Kamiyama H, Fukushima T: Surgical approaches for Between 1968 and 2006, the senior commentator (MGY) had the the treatment of aneurysms on the P2 segment of the posterior cerebral artery. opportunity to operate on 161 patients with temporal mediobasal Neurosurgery 47:359–366, 2000. gliomas involving the medial sector of the temporal pole, amygdala, 32. Ture U, Pamir MN: Small petrosal approach to the middle portion of the mediobasal temporal region: Technical case report. Surg Neurol 61:60–67, hippocampus, and parahippocampus (pure limbic-paralimbic area 2004. tumors). In addition, multicompartmental gliomas (n = 176) of the lim- 33. Uchiyama N, Hasegawa M, Kita D, Yamashita J: Paramedian supracerebellar bic-paralimbic system, located mainly in the anterior mediobasal tem- transtentorial approach for a medial tentorial meningioma with supratento- poral, insular, caudal, fronto-orbital, septal and subcallosal areas, were rial extension: Technical case report. Neurosurgery 49:1470–1474, 2001. operated on. In 4.4% of these patients, the tumors were located in the

294 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SURGERY FOR TEMPORAL MEDIOBASAL TUMORS

posterior areas of hippocampus-parahippocampus with involvement of sional discussions in the future. Yas¸argil’s series with 487 operated the posterior cingulate gyrus and inferior part of the precuneus. In 12 limbic-paralimbic gliomas will be published in the near future. patients, the gliomas occurred solely within the lateral temporo-occip- Saleem I. Abdulrauf ital (fusiform) gyrus, which belongs to the isocortical neopallial system St. Louis, Missouri of the temporal lobe. Therefore, these cases have been excluded from M. Gazi Yas¸argil our limbic-paralimbic glioma series. Little Rock, Arkansas Comparison of this series by Schramm and Aliashkevich with that of the senior commentator (Yas¸argil et al., unpublished data) reveals a striking difference in the histopathological aspects of gliomas. In he authors review a large series of patients (n = 235) with temporal Schramm et al.’s series, 76.2% of the gliomas were benign, 23.8% were Tmediobasal tumors and make recommendations with respect to malignant, and larger tumor size was associated with higher frequency surgical approach and chances of postoperative neurological deficit. of malignant histopathology. In Yas¸argil’s series, 48% were benign and These recommendations are based on their previously described classi- 52% were malignant. Schramm et al. designated the size of the tumors fication system, which arose from analysis of the same series of patients. as small or moderate in 65.2% of the cases, whereas small and moder- When tumors of different histologies in a similar location are com- ate size tumors appeared in only 34 patients (21.1%) in Yas¸argil’s series. pared, the goals of surgery will differ, as will outcomes and acceptable Modern neuroimaging facilities were at the disposal of both cen- rates of morbidity. A slightly higher rate of postoperative quadran- ters. Apparently, and fortunately, the patients were admitted to surgery tanopia may be acceptable in exchange for complete resection of high- in the neurosurgery department in Bonn at an earlier phase of their grade tumors or cases of refractory epilepsy. However, care should be disease. A significant difference in surgical concept and surgical taken to avoid this complication in patients with low-grade tumors, approaches is also noted. In Bonn, the removal of tumors was achieved which can be followed for growth or recurrence. via a transsylvian approach in 28% of the patients, by an anterior two- thirds temporal lobectomy in 23.0%, by temporal pole resection in Ruth E. Bristol 15.0%, via a transcortical approach in 6.9%. Robert F. Spetzler In Yas¸argil’s series, all gliomas (n = 306) located in the temporal Phoenix, Arizona mediobasal area or multicompartmental areas were removed via a pterional anterior transsylvian approach (95.6%). In 14 patients (4.4%), chramm and Aliashkevich provide a retrospective summary of more the temporal mediobasal glioma extended much further posteriorly, Sthan 200 patients with mesiobasal temporal tumors treated surgi- herniating into the dorsal mesencephalon or involving the posterior cally. The temporal lobe anatomy and surgical approach is well described cingulate gyrus and inferior part of the precuneus. In these instances, based on an anatomical classification scheme around the collateral sul- a second operation was performed and the residual parts of the glioma cus. Complications from surgery are also nicely presented and discussed, were removed through a parieto-occipital craniotomy and posterior including detailed visual field analysis. The take-home message is that interhemispheric approach with the patient in a seated position. Using these tumors are often difficult to access and are centered around critical the anterior transsylvian approach, which was combined with a poste- brain anatomy. The surgeon must be extremely familiar with the mesial rior interhemispheric approach in some patients, a pure lesionectomy temporal lobe anatomy to safely and adequately treat these patients. of a glioma in the limbic-paralimbic compartments could be accom- Although epilepsy was not the main focus of this study, it would be plished without causing damage to healthy neopallial isocortical areas interesting to know the decision making process concerning whether or and their function. The anterior two-thirds temporal lobectomy, not to resect the mesial structures for Type B tumors. This remains a dif- transcortical approach, and subtemporal approach have never been ficult question that can only be answered with a large series. practiced on patients with temporal mesiobasal lesions. We commend the authors on this somewhat controversial but stim- William E. Bingaman ulating study. This solid data will prove a valuable source for profes- Cleveland, Ohio

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 295 CLINICAL STUDIES

EXPRESSIVE AND RECEPTIVE LANGUAGE AREAS DETERMINED BY A NON-INVASIVE RELIABLE METHOD USING FUNCTIONAL MAGNETIC RESONANCE IMAGING AND MAGNETOENCEPHALOGRAPHY

Kyousuke Kamada, M.D., Ph.D. OBJECTIVE: It is known that functional magnetic resonance imaging (fMRI) and mag- Department of Neurosurgery, netoencephalography (MEG) are sensitive to the frontal and temporal language function, The University of Tokyo, respectively. Therefore, we established combined use of fMRI and MEG to make reliable Tokyo, Japan identiﬁcation of the global language dominance in pathological brain conditions. Yutaka Sawamura, M.D., Ph.D. METHODS: We investigated 117 patients with brain lesions whose language domi- Department of Neurosurgery, nance was successfully conﬁrmed by the Wada test. All patients were asked to gener- Hokkaido University, ate verbs related to acoustically presented nouns (verb generation) for fMRI and to read Sapporo, Japan three-letter words for fMRI and MEG. Fumiya Takeuchi, Ph.D. RESULTS: fMRI typically showed prominent activations in the inferior and middle frontal Research Institute for Electric Science, gyri, whereas calculated dipoles on MEG typically clustered in the superior temporal Hokkaido University, region and the fusiform gyrus of the dominant hemisphere. A total of 87 patients were Sapporo, Japan further analyzed using useful data from both the combined method and the Wada test. Remarkably, we observed a 100% match of the combined method results with the Shinya Kuriki, Ph.D. results of the Wada test, including two patients who showed expressive and receptive Research Institute for Electronic Science, Hokkaido University, language areas dissociated into bilateral hemispheres. Sapporo, Japan CONCLUSION: The results demonstrate that this non-invasive and repeatable method is not only highly reliable in determining language dominance, but can also locate the Kensuke Kawai, M.D., Ph.D. expressive and receptive language areas separately. The method may be a potent alter- Department of Neurosurgery, native to invasive procedures of the Wada test and useful in treating patients with brain The University of Tokyo, Tokyo, Japan lesions. KEY WORDS: Expressive language function, Functional magnetic resonance imaging, Language dominance, Akio Morita, M.D., Ph.D. Magnetoencephalography, Receptive language function Department of Neurosurgery, The University of Tokyo, Neurosurgery 60:296–306, 2007 DOI: 10.1227/01.NEU.0000249262.03451.0E www.neurosurgery-online.com Tokyo, Japan

Tomoki Todo, M.D., Ph.D. rain asymmetries have been of consider- subjects with chronic epilepsy have speech Department of Neurosurgery, able interest in neurology for more than a dominance in the right and left hemispheres, The University of Tokyo, century. Based on clinicopathological respectively (3). Furthermore, several studies Tokyo, Japan B studies, the “classical mode” of language organ- suggested the possibility of atypical language ization consists of a frontal “expressive” area representation in patients with chronic epilep- Reprint requests: for planning and executing speech and writing, sy (20–30%) (9, 28). However, the procedure of Kyousuke Kamada, M.D., Ph.D., and Tomoki Todo, M.D., Ph.D., and a temporal “receptive” area for analysis successive anesthetization of each hemisphere Department of Neurosurgery, and identiﬁcation of linguistic sensory stimuli. by intracarotid injections of sodium amobar- The University of Tokyo, This basic scheme of language functions has bital requires catheterization and irradiation. 7-3-1 Hongo, Bunkyo-ku, generally been accepted, with the assumption Furthermore, the Wada test results can only Tokyo 113-8655 Japan. Email: [email protected] that both expressive and receptive functions demonstrate a relative distribution of language dominantly exist in the same hemispheric side. functions across the two hemispheres. More Received, March 23, 2006. The Wada test has been considered the most detailed information on localization of speci- Accepted, October 13, 2006. reliable method to determine language domi- ﬁed language functions within a hemisphere is nance. According to one of the largest studies important for understanding the language net- performed to date, 4 and 96% of right-handed works, as well as the treatment of brain lesions.

296 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com FUNCTIONAL MAGNETIC RESONANCE IMAGING AND MAGNETOENCEPHALOGRAPHY FOR LANGUAGE DOMINANCE

The use of functional magnetic resonance imaging (fMRI) Handedness Inventory was used to estimate the patients’ hand- has recently been developed to identify the hemisphere with edness (18). A written informed consent was obtained from the language dominance. Most language fMRI studies have patient or his/her family before participation in the study. observed activations in the inferior frontal gyrus (IFG) and middle frontal gyrus (MFG) using tasks such as word genera- Magnetic Resonance Protocols tion and categorization (16, 24, 29). Detection of the receptive Anatomic magnetic resonance imaging (MRI) and fMRI were language area by fMRI has been reported to be more difficult performed during the same session with a 1.5-T whole-body than that of the expressive language function, and the use of lis- magnetic resonance scanner with echo-planar capabilities and tening or sentence comprehension tasks has resulted in visual- a standard whole-head transmit-receiver coil (Siemens Vision, ization of only a few pixels in the temporoparietal region (8, 16, Erlangen, Germany). During the procedures, foam cushions 25, 26). In addition, a fundamental limitation of an fMRI-based were used to immobilize the head. brain mapping is the varying degrees of regional hemodynamic responses under pathological brain conditions (7, 10, 15). Therefore, a clinical interpretation of localized activations on Language fMRI fMRI remains complicated and controversial. The patients were instructed to respond to all language tasks Magnetoencephalography (MEG) reflects intracellular elec- silently. fMRI data was acquired with a T2-weighted echo- tric current flow in the brain and allows accurate localization planar imaging sequence (echo time, 62 ms; repetition time, of the current dipole sources. Dipoles of MEG deflections that 114 ms; flip angle, 90 degrees; slice thickness, 4 mm; slice gap, peaked at approximately 400 milliseconds after word presen- 2 mm; field of view, 260 mm; matrix, 64 ϫ 128; 14 slices). Each tation (late responses) have been observed to localize in the fMRI session consisted of three dummy scan volumes fol- temporoparietal regions. These late responses have been con- lowed by three activation and four baseline (rest) periods. sidered to be related to the receptive language function (19, During each period, five echo-planar imaging volumes were 20). We have also observed dense dipole clusters of the seman- collected, yielding a total of 38 imaging volumes and 2 min- tic late responses in the superior temporal gyrus (STG), supra- utes 32 seconds in measurement time for each session. fMRI marginal gyrus (SmG), and fusiform gyrus (FuG) of the sus- data of language-related semantic responses were acquired as pected dominant hemisphere (11, 12). Therefore, we sought to follows. All subjects were examined with two different lexical use MEG not only as an additional diagnostic tool for identi- semantic language paradigms; verb generation by listening to fying the language dominance, but also to localize the recep- nouns and A/C categorization by reading words. All words for tive language center. semantic tasks were selected from common Japanese words In the present study, we describe a non-invasive method to listed in the electronic dictionary of the National Institute for locate the expressive and receptive language areas by co- Japanese Language. utilizing fMRI and MEG. The language dominance determined by our method matched the results from the Wada test with Verb Generation Task 100% accuracy. The usefulness of the method was well demon- For the auditory stimuli (duration ranges were between 400 strated, especially in those patients who showed dissociated and 600 ms), common concrete nouns spoken by a native expressive and receptive language functions. The data show Japanese speaker with a flat intonation were recorded and that this method is highly reliable and may be useful in the digitized with a sampling rate of 44,000 Hz. A backward play- management of patients with brain lesions as well as in study- back of the sound files (reference sounds) was used to elimi- ing normal brain functions. nate the primary auditory activation during the rest periods with the same inter-stimuli intervals (1600–2400 ms) as the METHODS active periods. The auditory stimuli were delivered binau- rally via two 5-m-long plastic tubes terminating at a head- Patients phone. The sound intensity was approximately 95 dB sound The functional brain mapping using fMRI (with the verb gen- pressure level at the subject’s ear. Subjects were instructed to eration task) and MEG was performed in 117 patients with brain silently generate a verb related to each presented noun during lesions since August 1999 (Ͼ7 yr) after this project was approved the active periods and passively listen to the reference sounds by the Institutional Committee for Ethics (Table 1). fMRI studies during the rest periods. with the abstract/concrete (A/C) categorization task were also performed in 106 patients. Ninety-seven patients also under- A/C Categorization Task went the Wada test to confirm the dominant cerebral hemisphere Visual stimuli were presented on a liquid crystal display mon- for language functions. Six patients showed negative Wada test itor with a mirror above the head coil allowing the patients to see results owing to the steal effect of a large arteriovenous malfor- the stimuli. Words consisting of three Kana letters (Japanese mation (AVM) or an overdose. The final analyses were per- phonetic symbols) were presented in a 300-millisecond expo- formed in 87 patients (48 men, 39 women), who underwent sure time with interstimuli intervals ranging from 2800 to 3200 Wada test, fMRI, and MEG investigations. The mean age (Ϯ milliseconds. Patients were instructed to categorize the pre- standard deviation) was 43.6 Ϯ 14.1 years. The Edinburgh sented word silently into “abstract” or “concrete” based on the

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 297 KAMADA ET AL.

TABLE 1. Summary of patients’ brain lesions typesa Chronic Cavernous Cerebral Glioma AVM Meningioma Total Epilepsy malformation ischemia fMRI with VG ϩ MEG 44 39 18 6 4 6 117 fMRI with A/C 41 34 15 6 4 6 106 Amytal test 42 29 16 6 4 0 97 Final analyses 39 26 12 6 4 0 87

a AVM, arteriovenous malformation; fMRI, functional magnetic resonance imaging; VG, verb generation task; MEG, magnetoencephalography; A/C, abstract/concrete categorization task. nature of the word. During interval periods, patients passively One hundred fifty epochs of the magnetic signals were aver- viewed random dots of destructured Kana letters that were aged and digitally filtered between 0.1 to 30 Hz. Significant MEG controlled to have the same luminance as the stimuli to elimi- deflections were visually identified based on the square root nate primary visual responses. mean fields of more than 10 sensors in the frontotemporal (FT) Before scanning, all patients had a brief practice time, and the or temporo-occipital (TO) regions. Locations and dipole fMRI examinations were repeated for each task to confirm the moments of equivalent current dipoles were calculated every 2 reproducibility. After data acquisition, a motion detection pro- milliseconds from 250 to 600 milliseconds after the stimulus gram (MEDx; Medical Numerics, Sterling, VA) discarded fMRI onsets using the single equivalent dipole and sphere head mod- sessions containing motion artifacts exceeding 25% of the pixel els. Only those dipoles of which the measured and the calculated size. A Gaussian spatial filter (6 mm in half width) was applied, field distributions showed a correlation value of more than 0.85 and functional activation maps were calculated by estimating and confidence volumes less than 1000 mm3 were used. To con- the Z-scores between the rest and activation periods using Dr. firm the calculated results, the same MEG time sections were View (Asahi Kasei, Tokyo, Japan). Pixels with Z-scores higher analyzed using a current density map (low-resolution tomogra- than 2.2 (P Ͻ 0.05) were considered to indicate real activation phy; LORETA, Curry, Neuroscan, and Compumedics USA, El and were used for mapping. Image distortion of fMRI was cor- Paso, TX). The coordinates of the MEG system were transformed rected by maximizing the mutual information of the fMRI data into anatomic 3D-MRI scans by identifying external anatomic sets and three-dimensional T1-weighted MRI (3D-MRI) scans fiduciary markers (nasion, left/right preauricular points), and of the patient’s brain (morphing compensation). The result estimated dipoles were superimposed onto the 3D-MRI scans. from each fMRI session was co-registered with the 3D-MRI by Dipoles located in the temporal region, including the STG, the Affine transformation (5). After total number of the acti- MTG, SmG, and FuG, were manually counted. A patient was vated pixels in the IFG and MEG were automatically counted, considered to have unilateral language dominance when hemi- a patient was considered to have unilateral language domi- spheric dipoles of one hemisphere counted less than 70% of the nance when hemispheric pixels of one hemisphere counted less other hemisphere. Otherwise, the language dominance was than 70% of the other hemisphere. Otherwise, the language considered bilateral. dominance was considered bilateral. Determination of Language Dominance Language MEG using fMRI and MEG The MEG signals were recorded with a 204-channel biomag- On the basis of the results of language fMRI and MEG, we netometer (VectorView; Neuromag, Helsinki, Finland) in a determined language dominance for each patient. When the magnetically shielded room. To confirm the reproducibility, we semantic activation in one side of the IFG and MFG was wider acquired two data sets for each task by repeating the MEG than that of the other side during the language fMRI tasks, a recording on two different days. One hundred fifty nouns con- patient was considered to have unilateral dominance for the sisting of three Kana letters were visually presented with a 300- expressive language function. When one side of the temporal millisecond exposure time with interstimuli intervals ranging region included more MEG dipoles than the other during the from 2800 to 3200 milliseconds. Patients were instructed to language MEG task, we determined that a patient had lateral- judge whether or not the presented word was “abstract” or ity of the receptive language function. “concrete” based on the nature of the word and to push a button with the index or middle finger (Kana reading task). Each The Wada Test epoch consisted of a 500-millisecond prestimulus baseline and All patients received injections of amobarbital (100 mg in a a stimulus followed by a 1500-millisecond analysis period. 10% solution, Amytal; Eli Lilly and Co., Indianapolis, IN) Epochs with a reaction time exceeding 1200 milliseconds and through a catheter placed in the internal carotid artery. MEG examinations with a successful task performance less Language testing was performed during the observation period than 70% were discarded. of maximal amobarbital action as indicated by contralateral

298 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com FUNCTIONAL MAGNETIC RESONANCE IMAGING AND MAGNETOENCEPHALOGRAPHY FOR LANGUAGE DOMINANCE

brachial plegia. Patients were given the following tasks in the went the verb generation task, 100 patients (84.6%) completed following order and up to four points were given, depending the task and provided useful fMRI data. The results showed on the severity of the language disturbance: 0, no response; 1, that the dominant hemisphere for the expressive language meaningless utterance; 2, incorrect repetition or paraphasia; 3, function was left, right, and bilateral in 90, eight, and two self-correction; and 4, unimpaired. patients, respectively. In the epilepsy group (n ϭ 34), left, right, The tasks were as follows: and bilateral dominance was 29 (85.2%), three (8.8%), and two 1) Spontaneous counting. Patients were instructed to count, (8.5%), respectively. In the non-epilepsy group (n ϭ 66), left, starting immediately before the amobarbital administration right, and bilateral dominance was 61 (92.4%), five (7.6%), and and continuously until the next task was given. If the patient zero (0%), respectively. The activated regions on fMRI mainly could continue to count even after brachial plegia appeared, involved the IFG and MFG, the lateral precG, AG, and the sup- obvious speech arrest and no impairment indicate 0 and 4 plementary motor area (SMA) (Figs. 1 and 2). points, respectively. In some patients, activations were observed in bilateral hemi- 2) Letter reading. Patients were instructed to read aloud seven spheres. Except for two patients who showed bilateral domi- words consisting of three or four Kana letters. The maxi- nance, the activations in the non-dominant hemisphere were mum score was 28 points (seven items ϫ four points). restricted to MFG and precG and smaller in size, so the pixels 3) Naming. Patients were asked to name aloud the five objects did not reach a cluster significance (maximum values of Z- presented pictorially. The maximum score was 20 points score, Ͻ2.2 or Ͻ10 pixels). (five items ϫ four points). Compared with successful results of the Wada test, the success- 4) Auditory comprehension. Patients were asked to carry out ful rate of fMRI with the verb generation task was 90.1%. Seven three simple tasks such as blinking eyes, opening the mouth, patients with aphasia or dementia failed to complete the task. and raising the unparalyzed arm. The maximum score was Three glioma patients with marked surrounding, four patients 12 points (three items ϫ four points). with brain ischemia and three patients with large arteriovenous 5) Pointing objects. Patients were shown a picture with a set of malformations failed to exhibit significant activations in the four objects and were instructed to point to one chosen by frontal lobe (Fig. 3). These incomplete results are accounted for by the investigator (e.g., “Point to the cat.”). The maximum the reported disadvantage of fMRI that data may be affected by score was 16 points (four items ϫ four points). the pathological changes of cerebral circulation (7, 10, 15). Performance in Tasks 1 and 3 were considered to reflect the expressive language capabilities (maximum score, 24 points); fMRI with the A/C Categorization Task performance in Tasks 2, 4, and 5 reflected receptive language The A/C categorization task was designed to locate the recep- functions (maximum score, 56 points). tive language area by fMRI. Among 106 patients who performed the A/C categorization task, 71 (67.0%) completed the task and RESULTS provided useful fMRI data. Compared with the verb generation task, the A/C categorization task more often activated wider areas Handedness and the Wada Test in bilateral hemispheres (Fig. 2). Activations generally involved Ninety-one patients (80 right-, eight left-, and three bilateral- the bilateral frontal lobes, including the IFG, MFG, and precG, handers) successfully underwent the Wada test. Language with laterality. The superior temporal regions, such as the STG dominance was left, right, and bilateral hemispheres in 81, six, and SmG, demonstrated activation spots in only 45% (n ϭ 32) of and four patients, respectively. The language dominance of the the investigated patients, and the side predominance was not right-handed patients was left in 75 patients (93.8%), right in apparent in most cases. The fMRI data of the A/C categorization two patients (2.5%), and bilateral in three patients (including task were considered unsuitable to determine the receptive lan- one patient with dissociated expression and receptive func- guage areas and were not used for the final analyses. tions [3.8%]), respectively. For left-handed patients, four patients showed left and four showed right dominance. For Language MEG Profiles and Dipole Locations both-handed patients, two showed left dominance and one The Kana reading task was designed to locate the receptive bilateral (dissociated). These results were similar to those of language area by MEG. The language MEG was performed in previous reports on language dominance (3, 4). 117 patients, of whom 99 (85.4%) completed the task and pro- For further analysis, we subdivided the subjects into groups vided useful data (Figs. 1 and 2). Results showed that the dom- with chronic epilepsy and with non-epilepsy. In the epilepsy inant hemisphere for the receptive language function was left, group (n ϭ 29), left, right, and bilateral dominance was 24 right, and bilateral in 85, 11, and three patients, respectively. In (82.8%), three (10.3%), and two (6.9%), respectively. In the non- the epilepsy group (n ϭ 31), left, right, and bilateral dominance epilepsy group (n ϭ 62), left, right, and bilateral dominance was 26 (83.9%), three (9.7%), and two (6.5%), respectively. In the was 57 (91.6%), four (6.4%), and one (1.6%), respectively. non-epilepsy group (n ϭ 68), left, right, and bilateral dominance was 59 (86.8%), eight (11.8%), and one (1.5%), respectively. fMRI with the Verb Generation Task Dipole clusters of late deflections localized mainly in the The verb generation task was designed to locate the expres- superior temporal region (STG, MTG, and SmG), and 60% of sive language area by fMRI. Among 117 patients who under- investigated patients also showed dipoles in the inferior tempo-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 299 KAMADA ET AL.

A deflections at approximately 200 milliseconds with short dura- tions and little laterality. Estimated dipoles of the FT regions were densely accumulated in the left STG, MTG, and SmG (102 dipoles), whereas the right hemisphere showed fewer dipoles (54 dipoles) in the superior temporal region. This patient was thus determined to have receptive language dominance in the left temporal lobe. The successful rate of language-MEG was 82.4%. Nine out of 39 epilepsy patients (23.1%) could not provide useful MEG data owing to artifacts from constant eye movements; the Kana- B reading task was more difficult to complete than the verb generation task for patients with mental dysfunction. On the other hand, only one out of 18 AVM patients, owing to severe dyslexia, failed to provide useful MEG data, indicating that, in contrast to fMRI, MEG was not frequently affected by cerebral blood flow abnormalities (Fig. 3).

Combination of fMRI and MEG with Wada Test Verification The verb generation task fMRI data depict expressive language areas well, but may be affected by cerebral blood flow abnormalities. The MEG results indicate receptive language areas well, but the task is rather complicated and may not be suited for patients with mental disorders. We sought to establish a non-invasive and reliable method to determine the laterality of language dominance by combining the advantages of these approaches. Furthermore, in terms of language func- C tions, the results from fMRI and MEG can be integrated to locate expressive and receptive language areas and to provide reliable evidence whether or not there is dissociation. To verify the reliability of our method, 97 patients also underwent the Wada test. Useful data from the method co-utilizing fMRI and MEG could be obtained from 87 out of 91 patients (95.6%). Remarkably, regarding language dominance, the results from the combination method matched the results of the Wada test in all 87 patients. Worth noting is that two patients (one with left tempo- FIGURE 1. A 24-year-old, right-handed man with epilepsy. A, fMRI with the verb generation task showing activations predominantly in the left IFG, MFG, ral lobe epilepsy and the other with right insular astrocytoma) PrecG, and parieto-occipital regions. B, square root mean field profiles of lan- showed dissociated language areas using the combined method. guage MEG responses in the bilateral FT and TO regions. The left FT The expressive language area was depicted in the left frontal lobe responses, peaking at 450 milliseconds, were markedly greater in amplitude by fMRI, but the receptive language area was demonstrated in than the right FT. C, source localization of the late deflections showing predom- the right temporal lobe by MEG (Fig. 4). The Wada test results inant dipole clusters (arrowheads) in the left superior temporal region. The left confirmed that both patients have language functions dissoci- and right hemispheres contained 97 and 37 dipoles, respectively. ated in the bilateral hemispheres. Among the 91 patients who underwent the Wada test, these were the only two patients in ral region (FuG and inferior temporal gyrus). In 96 patients whom the Wada test detected dissociation of language functions. who showed unilateral language dominance, the total number In 12 epilepsy patients, the expressive and/or receptive lan- of dipoles in the dominant versus non-dominant hemispheres guage areas were electrophysiologically investigated via a sub- was 124.1 Ϯ 62.1 and 58 Ϯ 30.9 (mean Ϯ standard deviation), dural electrode implantation and the results were compared respectively. The ratio of the dipole number in the dominant with those determined via the combined fMRI plus MEG hemisphere to the non-dominant hemisphere in each individ- method (Fig. 5). Out of eight patients who underwent cortical ual was 2.4 Ϯ 1.7 (range, 1.43–14.4). mapping for the expressive language area, all showed a speech A typical result with all channels of MEG with the Kana- arrest by electrical stimulation to the IFG and four to the MFG. reading task is illustrated in Figure 1. Later deflections peaking All of the physiologically determined locations were confined at approximately 400 milliseconds were predominantly within the areas depicted by the combined method. Out of six observed in the left FT. Bilateral TO regions demonstrated early patients who received electrical stimuli to the temporal lobe,

300 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com FUNCTIONAL MAGNETIC RESONANCE IMAGING AND MAGNETOENCEPHALOGRAPHY FOR LANGUAGE DOMINANCE

four showed responses interpretable as impaired speech com- determined by the combined method were always broader, but prehension. In all such cases, the electrophysiologically deter- had the border within the adjacent gyri of those determined by mined location matched the area depicted by the combined electrophysiological mapping. method, although MEG-depicted receptive language areas covered relatively broad areas of the temporal lobe. The regions ILLUSTRATIVE CASES

A Patient 1 A 16-year-old, right-handed female patient had experienced transient numbness in her left upper extremity with a 2-month history. T1- weighted MRI scans demonstrated an extra-axial cystic lesion in the left frontal region. Although the lesion markedly compressed the frontal lobe, she had no impairment of language and motor functions. fMRI with the verb generation task demonstrated obvious activation in the left IFG and MFG shifted inferiorly by the lesion (Fig. 2A). The A/C categorization task activated a small area of the left IFG, but mainly the bilateral occipital lobes. Concerning MEG with the Kana reading task, RMS of the left FT was much higher than that of the right, and numbers of semantic dipoles were 117 and 30 in left and right hemispheres, respectively. The main dipole clusters were located in the left IFG and STG. The tumor was totally removed and histopathological diagnosis B was meningioma. Patient 2 A 24-year-old, right-handed male patient had a large AVM in the left frontal lobe. fMRI detected little activation in the IFG or MFG, although a part of the left angular gyrus was activated by the verb generation task (Fig. 3A). MEG, however, disclosed numerous dipole accumulations in the left superior temporal region. In the MEG examination, the left and right hemispheres contained 130 and 45 dipoles, respectively, suggesting left language dominance (Fig. 3B). Auditory comprehension and letter-reading were suppressed by administration of amobarbital into the left carotid artery, although motor language function was preserved. These ﬁndings suggested that the steal effect caused by the AVM partly interfered with functional brain mapping of fMRI and the C Wada test. In this case, MEG was helpful to decide language dominance (Fig. 3).

Patient 3 A 32-year-old, right-handed man experienced amnesia for several minutes. T1-weighted MRI scans and brain computed tomographic scans disclosed a hypointense and hypodense mass in the right insular cortex involving the surrounding white matter. Computed tomographic scans performed 6 years earlier, however, revealed no abnormality. These ﬁndings suggested that a low-grade astrocytoma might

FIGURE 2. A 16-year-old, right-handed female patient with a large meningioma in the left frontal region. The patient had no impairment of language or motor functions. A, fMRI with the verb generation task showed activations D mainly in the left IFG and MFG that shifted inferiorly by the tumor. B, fMRI with the abstract/concrete categorization task demonstrated activations in the bilateral occipital regions in addition to small active spots in the left IFG. C, square root mean field profiles of language-MEG responses demonstrated that the left FT responses, peaking at 400 milliseconds, were markedly larger in amplitude than the right FT. D, source localization of the late deflections showed predominant dipole clusters in the left posterior temporal region. The left and right hemispheres contained 117 and 30 dipoles, respectively. The combined fMRI plus MEG method indicated left language dominance, which was confirmed by Wada test.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 301 KAMADA ET AL.

A DISCUSSION

We demonstrated that our method using both fMRI with the verb generation task and MEG with the Kana reading task is highly reliable in determining the language dominance in patients with brain lesions. The accuracy of the dominance laterality was confirmed by a 100% match with the results from the Wada test. fMRI and MEG compensated B each other’s disadvantages. The tasks of fMRI were rather simple and could be accomplished even by patients with mental dysfunctions, whereas MEG results were seldom affected by cerebral blood flow abnormalities. Reliable data on language functions were also obtained by combining the advantageous features of fMRI and MEG. fMRI with the verb generation task well depicted FIGURE 3. A 24-year-old, right-handed man with a large AVM in the left frontal lobe. A, fMRI with the verb genera- the expressive language area as tion task showed little activation in the left frontal lobe where the AVM was located. B, source localization of the late FT activations in the frontal lobe, and TO deflections on MEG showed predominant dipole clusters in the left posterior STG. The left and right hemispheres most commonly in the IFG. contained 123 and 51 dipoles, respectively. MEG, on the other hand, showed dipole clusters pre- have slowly developed during the past 6 years. In the results of the dominantly in the superior temporal regions representing the verb generation task, the left hemisphere had obvious activations in the receptive language area. In the epilepsy group, left and bilateral IFG, MFG, precG, and the angular gyrus, indicating that this patient dominance were approximately 85% and more than 6%, respec- had left dominance of motor-language functions (Fig. 4A). In contrast, tively, whereas, in the non-epilepsy group, left and bilateral estimated dipoles of the FT responses were concentrated in the poste- dominance were more than 90% and less than 2%, respectively. rior part of the right STG and MTG (138 dipoles) and another dipole The combined method, including the Wada test, fMRI, and cluster (64 dipoles) of the TO region was localized in the right FuG. The MEG, clearly demonstrated bilateral dominance is more often total dipole number of the left hemisphere (48 dipoles) did not reach even a quarter of that of the right hemisphere, suggesting right-sided observed in the epilepsy group than in the non-epilepsy group. dominance of temporal language functions (Fig. 4). In our study, two out of 87 patients analyzed (2.3%) were During the Wada test, he stopped counting (0 out of 4 points; 0%) found to have dissociation of the expressive and receptive lan- and failed to name objects (6 out of 20 points; 30%) after left intrac- guage functions by co-utilization of fMRI and MEG, verified by arotid injection, whereas letter-reading (21 out of 28 points; 75%), audi- the Wada test, which best described the usefulness of our tory comprehension (12 out of 12 points, 100%), and pointing objects method in identifying the areas of the two language functions tasks (16 out of 16 points; 100%) were well preserved. In contrast, after separately. In both cases, neither modality alone demonstrated right intracarotid injection, letter reading (13 out of 28 points; 45%), the dissociation. Although several cases have been reported auditory comprehension (3 out of 12 point; 25%), and pointing objects that dissociated language functions were found by fMRI, none (4 out of 16 points; 25%) tasks were markedly suppressed, although he of those was proven by the Wada test (2, 8, 21, 23). Our results continued to count correctly without speech blockade (4 out of 4 points; show that neither fMRI nor MEG alone is sufficient to accu- 100%) and could perform naming (17 out of 20 points; 85%). These findings suggested that language functions were distributed separately rately locate the expressive and receptive language areas, and over the bilateral hemispheres, and the expressive and receptive lan- the combined use is the key to obtaining high reliability. guage functions were dissociated in the left frontal and right temporal The results from electrophysiological investigation via a sub- lobes, respectively. A striking fact was that the combination of fMRI dural electrode implantation in 12 patients further confirmed and MEG predicted the special profiles of language functions non- the accuracy of the present method. Pouratian et al. (22) invasively. reported that the sensitivity and specificity of language-fMRI

302 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com FUNCTIONAL MAGNETIC RESONANCE IMAGING AND MAGNETOENCEPHALOGRAPHY FOR LANGUAGE DOMINANCE

A cance of activations depicted on fMRI is still under debate. Language-fMRI activations may be related to various semantic components of the task, including the will to retrieve verbal materials and the memory related to articulations. Despite that the A/C categorization task was designed to detect the receptive language area, activations in the temporoparietal region was B less frequently observed than in the frontal region. Neural activities in the temporoparietal area are considered relatively scarce (25), and the dis- crepant activities of the frontal and temporoparietal regions may be owing to physiological variations of brain regions. Alternatively, the frontal and temporal lobes may have different oscillations (brain FIGURE 4. A 32-year-old, right-handed man with astrocytoma in the right insular cortex and the surrounding white matter. A, fMRI with the verb generation task showed main activations in the IFG, MFG, precG, and AG, indicating rhythms) of brain activity in left dominance of the expressive language function. B, in contrast to the fMRI results, source localization of the late FT response to verbal tasks, deflections on MEG showed predominant dipole clusters in the right temporal lobe. The left and right hemispheres con- which are reflected in changes tained 48 and 202 dipoles, respectively. The combined fMRI plus MEG method thus indicated dissociated frontal motor in neuronal currents and cere- and temporal receptive language functions. This result was confirmed by the Wada test. The patient showed impaired bral blood flow. counting and object naming after amobarbitol injection into the left carotid artery. In contrast, letter reading, auditory Our study demonstrated comprehension and object pointing tasks were markedly suppressed, without counting impairment and speech blockade, that dominance of the recep- after amobarbitol injection into the right carotid artery. tive language function could be accurately determined by were dependent on the task, lobe, and matching criterion. The MEG. For that purpose, we originally designed the task of sensitivity and specificity of fMRI activations during expressive three-letter word reading and silent categorization and used linguistic tasks in the frontal lobe were found to be up to 100 the dipoles calculated from late deflections to process the MEG and 66.7%, respectively, in the frontal lobe. FitzGerald et al. (6) results. It has been reported that cortical evoked potentials reported that sensitivity and specificity for all multiple lan- recorded by subdural electrodes showed responses at approx- guage tasks ranged from 81 to 53% (6). On the other hand, sev- imately 200 (early) and 400 (late) milliseconds in the left tempo- eral groups have reported that the language map obtained from ral lobe cortex after letter presentation (1, 17). The late poten- fMRI poorly matched the intraoperative electrical stimulation tials have been noted especially in tasks involving decisions mapping (6, 25). In our study of language-fMRI, every electri- based on visually presented words (13, 14). In this study, the cal stimuli to the IFG, where the fMRI-activation was observed, sources of late responses (250–600 ms) were located mostly in caused speech arrest. However, the stimulation to MFG caused the posterior temporal region, and the laterality of dipole clus- language-related symptoms in only half of patients. Although ters accurately reflected the receptive language dominance. It the sensitivity of fMRI might be high, there are still several issues has been reported that dipoles in the superior temporal region of individual variability of fMRI activation and semantic tasks. showed an excellent agreement with an intraoperative electri- The discrepancy can be partly accounted for by the fundamen- cal mapping (27). We also included dipoles in the FuG for lan- tal differences in methodology such that the electrical stimula- guage dominance determination based on our experience with tion directly blocks the specific language functions, whereas a case in which an injury of FuG resulted in pure dyslexia (12). fMRI picks up all activated areas involved in the language These contrivances in our method may have led to improve- tasks. Therefore, fMRI-based mapping largely depends on the ment in accuracy on language dominance determination over design of the performing task. We tested two different tasks for previous reports (20). Basic technical issues of the MEG inves- fMRI and found the verb generation task better suited for lan- tigation still remain. Eye movement artifacts were strong guage mapping than the A/C categorization task. The signifi- enough to distort the baseline of the MEG data. In our study,

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 303 KAMADA ET AL.

A we asked patients to keep gazing at the center of the screen during the semantic decision without blinking. As a result, artifacts were observed at later than 600 milliseconds after letter presentation and usually did not affect the early and late semantic responses. It is, however, important to prevent artifacts by monitoring eye movements and using rejection thresh- olds. In conclusion, by co-utilizing fMRI and MEG, we established a method to determine language dominance with a high reliability. The fMRI activations with the verb generation task iden- tiﬁed the expressive language area, whereas the language MEG dipoles located the receptive language areas. Our institution is now routinely using the combined technique to identify the language dominance. If it does not produce data on cerebral B dominance, we additionally perform the Wada test before surgery. This non-invasive and repeatable method may be an effective alternative to the Wada test and may be useful in the management of patients with brain lesions.

Disclosure This work was supported in part by the Japan Epilepsy Research Foundation, Takeda Promotion of Science Foundation, a grant-in-aid No.17591502 for scien- tiﬁc research from MEXT, a Research Grant of the Princess Takamatsu Cancer Research Fund, Terumo Promotion of Science Foundation, Brain Science foundation, and Grant-in-Aid No. 18020010 for Scientiﬁc Research on Priority Areas Integrative Brain Research from MEXT.

C REFERENCES 1. Allison T, McCarthy G, Nobre A, Puce A, Belger A: Human extrastriate visual cortex and the perception of faces, words, numbers, and colors. Cereb Cortex 4:544–554, 1994. 2. Baciu MV, Watson JM, McDermott KB, Wetzel RD, Attarian H, Moran CJ, Ojemann JG: Functional MRI reveals an interhemispheric dissociation of frontal and temporal language regions in a patient with focal epilepsy. Epilepsy Behav 4:776–780, 2003. 3. Branch C, Milner B, Rasmussen T: Intracarotid sodium amytal for the lateralization of cerebral speech dominance: Observations in 123 patients. J Neurosurg 21:399–405, 1964. 4. Brazdil M, Zakopcan J, Kuba R, Fanfrdlova Z, Rektor I: Atypical hemispheric language dominance in left temporal lobe epilepsy as a result of the reorgan- ization of language functions. Epilepsy Behav 4:414–419, 2003. 5. Chen HM, Varshney PK: Mutual information-based CT-MR brain image registration using generalized partial volume joint histogram estimation. IEEE D Trans Med Imaging 22:1111–1119, 2003.

FIGURE 5. A 40-year-old, left-handed woman with epilepsy. A, fMRI with the verb generation task showed activations predominantly in the right IFG and MFG. B, fMRI with the A/C categorization task demonstrated activations in the right MFG and the posterior STG. C, source localization of the late deﬂections on MEG showed predominant dipole clusters (white squares) in the right posterior temporal region. The left and right hemispheres showed 44 and 144 dipoles, respectively. D, three-dimensionally reconstructed MRI scans fused with activation of the verb generation-fMRI (orange) and dipoles of language- MEG (blue). After implantation of subdural electrodes (gold), cortical mapping was performed with 50Hz bipolar electrical stimulation. Stimulation with intensity of 7mA to the right IFG caused speech arrest (white circles), whereas stimulation to the posterior STG caused impairment of auditory comprehension and reading capability (black circles).

304 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com FUNCTIONAL MAGNETIC RESONANCE IMAGING AND MAGNETOENCEPHALOGRAPHY FOR LANGUAGE DOMINANCE

6. FitzGerald DB, Cosgrove GR, Ronner S, Jiang H, Buchbinder BR, Belliveau 27. Simos PG, Papanicolaou AC, Breier JI, Wheless JW, Constantinou JE, Gormley JW, Rosen BR, Benson RR: Location of language in the cortex: A comparison WB, Maggio WW: Localization of language-specific cortex by using mag- between functional MR imaging and electrocortical stimulation. AJNR Am J netic source imaging and electrical stimulation mapping. J Neurosurg Neuroradiol 18:1529–1539, 1997. 91:787–796, 1999. 7. Holodny AI, Schulder M, Liu WC, Maldjian JA, Kalnin AJ: Decreased BOLD 28. Woermann FG, Jokeit H, Luerding R, Freitag H, Schulz R, Guertler S, Okujava functional MR activation of the motor and sensory cortices adjacent to a M, Wolf P, Tuxhorn I, Ebner A: Language lateralization by Wada test and glioblastoma multiforme: Implications for image-guided neurosurgery. AJNR fMRI in 100 patients with epilepsy. Neurology 61:699–701, 2003. Am J Neuroradiol 20:609–612, 1999. 29. Yetkin FZ, Mueller WM, Morris GL, McAuliffe TL, Ulmer JL, Cox RW, 8. Holodny AI, Schulder M, Ybasco A, Liu WC: Translocation of Broca’s area to Daniels DL, Haughton VM: Functional MR activation correlated with intra- the contralateral hemisphere as the result of the growth of a left inferior operative cortical mapping. AJNR Am J Neuroradiol 18:1311–1315, 1997. frontal glioma. J Comput Assist Tomogr 26:941–943, 2002. 9. Janszky J, Ollech I, Jokeit H, Kontopoulou K, Mertens M, Pohlmann-Eden B, Ebner A, Woermann FG: Epileptic activity influences the lateralization of COMMENTS mesiotemporal fMRI activity. Neurology 63:1813–1817, 2004. his is an interesting article evaluating the complementary features 10. Kamada K, Houkin K, Iwasaki Y, Takeuchi F, Kuriki S, Mitsumori K, of functional magnetic resonance imaging (fMRI) and magnetoen- Sawamura Y: Rapid identification of the primary motor area by using mag- T netic resonance axonography. J Neurosurg 97:558–567, 2002. cephalography (MEG) to assess language lateralization in 87 patients. 11. Kamada K, Kober H, Saguer M, Moller M, Kaltenhauser M, Vieth J: Responses Whereas any test of language lateralization is suspect if 100% correla- to silent Kanji reading of the native Japanese and German in task subtraction tion is found, the authors have carefully described their techniques magnetoencephalography. Brain Res Cogn Brain Res 7:89–98, 1998. and the analysis of results. It is quite apparent that fMRI with verb gen- 12. Kamada K, Sawamura Y, Takeuchi F, Houkin K, Kawaguchi H, Iwasaki Y, eration tasks is best at activating anterior language areas, whereas Kuriki S: Gradual recovery from dyslexia and related serial magnetoen- abstract versus concrete naming tasks can be less robust. This is a good cephalographic changes in the lexicosemantic centers after resection of a article and a large experience worthy of publication. mesial temporal astrocytoma. Case report. J Neurosurg 100:1101–1106, 2004. 13. Kutas M, Hillyard SA: Reading senseless sentences: Brain potentials reflect G. Rees Cosgrove semantic incongruity. Science 207:203–205, 1980. Burlington, Massachusetts 14. Kutas M, Hillyard SA: Brain potentials during reading reflect word expectancy and semantic association. Nature 307:161–163, 1984. he authors have applied fMRI and MEG techniques to localize 15. Lehericy S, Biondi A, Sourour N, Vlaicu M, du Montcel ST, Cohen L, Vivas E, Tspeech function in a large number of patients with different brain Capelle L, Faillot T, Casasco A, Le Bihan D, Marsault C: Arteriovenous brain lesions. They were able to supplement the two noninvasive tests with malformations: Is functional MR imaging reliable for studying language reor- the Wada test in 80% of the patients. They were able to obtain useful ganization in patients? Initial observations. Radiology 223:672–682, 2002. data with the co-utilization of fMRI and MEG in 95.6% of the patients 16. Lehericy S, Cohen L, Bazin B, Samson S, Giacomini E, Rougetet R, Hertz- and found a somewhat surprisingly good match with the results of the Pannier L, Le Bihan D, Marsault C, Baulac M: Functional MR evaluation of temporal and frontal language dominance compared with the Wada test. Wada test in 100% of those. In the results section, the authors discuss a Neurology 54:1625–1633, 2000. few differences to the localization of language areas by electrophysio- 17. Nobre AC, Allison T, McCarthy G: Word recognition in the human inferior logical means. They point out the fact that atypical language domi- temporal lobe. Nature 372:260–263, 1994. nance or bilateral language representation is more frequent in patients 18. Oldfield RC: The assessment and analysis of handedness: The Edinburgh with chronic epilepsy than in those without epilepsy. This is an impor- inventory. Neuropsychologia 9:97–113, 1971. tant fact not known to many neurosurgeons who are not ordinarily 19. Papanicolaou AC, Simos PG, Breier JI, Zouridakis G, Willmore LJ, Wheless involved with epilepsy cases. The results of this study make it more JW, Constantinou JE, Maggio WW, Gormley WB: Magnetoencephalographic likely that, in the future, the invasive Wada test procedure might be mapping of the language-specific cortex. J Neurosurg 90:85–93, 1999. abolished in those institutions at which MEG is available. This consti- 20. Papanicolaou AC, Simos PG, Castillo EM, Breier JI, Sarkari S, Pataraia E, tutes a notable limitation of this noninvasive technique. If fMRI is used Billingsley RL, Buchanan S, Wheless J, Maggio V, Maggio WW: alone, the success rate for obtaining useful data is 84.6% for word gen- Magnetocephalography: A noninvasive alternative to the Wada procedure. J Neurosurg 100:867–876, 2004. eration tasks and only 67% for the abstract/concrete categorization 21. Petrovich NM, Holodny AI, Brennan CW, Gutin PH: Isolated translocation of task. This is quite an interesting study and the results are very promis- Wernicke’s area to the right hemisphere in a 62-year-man with a temporo- ing; however, the limitations are not economical. A number of patients parietal glioma. AJNR Am J Neuroradiol 25:130–133, 2004. cannot complete all the tasks necessary for fMRI study, and MEG stud- 22. Pouratian N, Bookheimer SY, Rex DE, Martin NA, Toga AW: Utility of preop- ies can be disturbed by eye movement artifacts. We look forward to erative functional magnetic resonance imaging for identifying language cor- other reports confirming these promising results. tices in patients with vascular malformations. Neurosurg Focus 13:E4, 2002. 23. Ries ML, Boop FA, Griebel ML, Zou P, Phillips NS, Johnson SC, Williams JP, Johannes Schramm Helton KJ, Ogg RJ: Functional MRI and Wada determination of language lat- Bonn, Germany eralization: A case of crossed dominance. Epilepsia 45:85–89, 2004. 24. Roux FE, Boulanouar K, Lotterie JA, Mejdoubi M, LeSage JP, Berry I: he authors present some very interesting data in the realm of func- Language functional magnetic resonance imaging in preoperative assess- Ttional imaging to determine cerebral dominance for language. ment of language areas: Correlation with direct cortical stimulation. Currently, the standard modality for determining cerebral dominance Neurosurgery 52:1335-1345, 2003. is the venerable Wada test. In this study, the authors use both MEG and 25. Rutten GJ, Ramsey NF, van Rijen PC, Noordmans HJ, van Veelen CW: fMRI to determine language dominance based on activation in the infe- Development of a functional magnetic resonance imaging protocol for intra- rior frontal gyrus and middle frontal gyrus using fMRI and dipole operative localization of critical temporoparietal language areas. Ann Neurol 51:350–360, 2002. moments reflecting or indicating receptive language fields in the tem- 26. Rutten GJ, Ramsey NF, van Rijen PC, van Veelen CW: Reproducibility of poral lobe. As expected, they had some difficulty with the fMRI data fMRI-determined language lateralization in individual subjects. Brain Lang owing to the underlying deficit in the patient, which suggests that 80:421–437, 2002. fMRI is not always as good as one might expect in terms of determin-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 305 KAMADA ET AL.

ing cerebral dominance using a verb generation silent language task. going to be able to obtain both of these functional tests. Therefore, it is We know that fMRI is not a good choice for deﬁning receptive lan- unlikely that this strategy is going to replace Wada tests completely. guage ﬁelds that correspond to intraoperative stimulation mapping. Yet, this is a very important line of investigation and a novel observa- However, when fMRI was used together with MEG, the authors were tion that points out the frailties of functional imaging for cerebral dom- able to demonstrate 100% concordance with data from the Wada test. inance localization and the potential power when the different func- Thus, this is a very important study indicating that, in the near future, tional tests are combined. it may be possible to bypass the Wada test with these two powerful Mitchel S. Berger functional imaging modalities. That being said, not every institution is San Francisco, California

Portrait of James Figg (1695–1734), by William Hogarth, (1697–1764). Acknowledged in Britain as the “Father of Boxing,” Figg popularized the sport with teaching and exhibitions and, following victories over all the other British contenders, declared himself “heavyweight champion of England” in 1719.

306 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CLINICAL STUDIES

OBJECTIFYING WHEN TO HALT A BOXING MATCH: A VIDEO ANALYSIS OF FATALITIES

Vincent J. Miele, M.D. OBJECTIVE: Although numerous prestigious medical organizations have called for its Department of Neurological Surgery, abolishment, participation in the sport of boxing has reached an all-time high among West Virginia University, both men and women, and its elimination is unlikely in the near future. Physicians School of Medicine, Morgantown, West Virginia should strive to increase boxing safety by improving the rules of competition, which have evolved minimally over the past two centuries. Currently, subjective criteria are used Julian E. Bailes, M.D. to determine whether or not a contest should be halted. Developing a standardized, Department of Neurological Surgery, objective method of determining when a contest should be halted would be a signifi- West Virginia University, cant paradigm shift and could increase the safety of the sport’s participants. This study School of Medicine, Morgantown, West Virginia analyzed the number and types of punches landed in a typical professional match, in bouts considered to be competitive and in those that ended in fatalities, to determine Reprint requests: whether or not this would be a practical method of differentiating between these groups. Vincent J. Miele, M.D., Department of Neurological Surgery, METHODS: Three groups of professional boxing matches were defined at the begin- West Virginia University, ning of the study: 1) a “fatal” group, consisting of bouts that resulted in the death of a School of Medicine, participant; 2) a “classic” group that represented competitive matches; and 3) a “con- P.O. Box 9183, Morgantown, WV 26506–9183. trol” group of 4000 professional boxing matches representing the average bout. A com- Email: [email protected] puter program known as Punchstat (Compubox, Inc., Manorville, NY) was used in the objective analysis of these matches via videotape playback. Received, April 20, 2005. RESULTS: Several statistically significant differences were discovered between matches Accepted, June 19, 2006. that resulted in fatalities and the control group. These include the number of punches landed per round, the number of power punches landed per round, and the number of power punches thrown per round by losing boxers. However, when the fatal bouts were compared with the most competitive bouts, these differences were no longer evident. CONCLUSION: Based on the data analyzed between the control and fatal-bout groups, a computerized method of counting landed blows at ringside could provide sufficient data to stop matches that might result in fatalities. However, such a process would become less effective as matches become more competitive, and implementing such a change would significantly decrease the competitive nature of the sport. Therefore, other methods of quantifying acceleration–deceleration brain injuries are necessary to improve the safety of boxing. KEY WORDS: Boxing, Compubox, Fatalities, Punchstat, Video analysis

Neurosurgery 60:307–316, 2007 DOI: 10.1227/01.NEU.0000249247.48299.5B www.neurosurgery-online.com

he sport of boxing is often a subject of controversy because called for its abolishment (4–6, 8). Nonetheless, participation the primary strategy is to disable an opponent’s central has reached an all-time high among both men and women, and Tnervous system. This is usually performed by striking an the sport has ﬂourished into a billion-dollar industry (Figs. 1 adversary with a force sufﬁcient to render him or her uncon- and 2) (7). Its elimination is unlikely in the near future. scious. Boxing has been linked to acute traumatic brain injury Because physicians are called on to treat the acutely injured with the potential for pugilistica dementia, or “punch drunken- participants of this sport, they have a duty to suggest guide- ness,” from chronic participation (1–3, 7). In recent years, lines to increase safety. In addition, ringside physicians should numerous prestigious medical organizations, including the strive to increase boxing safety by improving the rules of com- American Medical Association, the American Academy of petition, which have evolved minimally during the past two Neurology, and the American Academy of Pediatrics, have centuries. Currently, many determinations made in the sport of

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 307 MIELE AND BAILES

professional boxing are quite subjective. These include scoring, effective aggression, defense, ring generalship, and the TABLE 1. Fatal and classic bouts analyzeda numerous ranking systems. Similarly, subjective criteria are Fatal Bout 1 Emile Griffith versus Benny Kid Paret—1962 used to determine whether or not a contest should be halted. A Fatal Bout 2 Sugar Ramos versus Davey Moore—1963 match is stopped if ringside officials and physicians feel that an Fatal Bout 3 Wilfred Scypion versus Willie Classen—1979 athlete can no longer defend him- or herself or mount an Fatal Bout 4 Johnny Owen versus Jose Guadalupe “Lupe” offense. Developing a standardized, objective method of deter- Pintor—1980 mining whether or not a contest should be halted would thus Fatal Bout 5 Ray Mancini versus Deuk Koo Kim—1984 be a paradigm shift that might increase the safety of the sport’s Fatal Bout 6 Gabe Ruelas versus Jimmy Garcia—1995 participants. Fatal Bout 7 Randie Carver versus Kabary Salem—1999 One way to accomplish this goal would be to develop an Fatal Bout 8 Steve Dotse versus Bobby Tomasello—2000 upper limit of punishment an athlete could receive before a Fatal Bout 9 Paul Vaden versus Stephan Johnson—2000 match would be halted. The simplest objective method of Fatal Bout 10 Khalid Jones versus Beethavean Scottland— achieving this goal would be to track the number and types of 2001 punches landed. This study analyzed and contrasted the num- Classic Bout 1 Nino Benvenuti W 15 Emile Griffith—1967 ber and types of punches landed in the typical professional IBHOF/Ring Magazine Fight of the Year, match, in bouts considered to be very competitive and in those Boxing Illustrated “Greatest Fights in Past 35 that ended in fatalities, to determine whether or not this would Years” be a practical method of differentiating between these groups. Classic Bout 2 Matthew Saad Muhammad KO 12 Marvin All available objective demographic data were also considered Johnson I—1977 in the comparison. Boxing Illustrated “Greatest Fights in Past 35 Years” METHODS Classic Bout 3 Danny Lopez KO 15 Mike Ayala—1979 IBHOF/Ring Magazine Fight of the Year, Three groups of professional boxing matches were defined at Boxing Illustrated “Greatest Fights in Past 35 the beginning of the study. The first group consisted of bouts Years” resulting in the death of a participant (Table 1). This was named the Classic Bout 4 Matthew Saad Muhammad KO 14 Yaqui “fatal” group. Matches were excluded if an organ system failure Lopez—1980 other than the brain was the primary cause of fatality (e.g., a IBHOF/Ring Magazine Fight of the Year, myocardial infarction). Currently, no central repository for these Boxing Illustrated “Greatest Fights in Past 35 data exists, so videotape recordings of these matches were Years” solicited from various organizations and individuals involved in Classic Bout 5 Sugar Ray Leonard KO 14 Thomas Hearns— the sport. The second group represented the competitive contest 1981 and was named the “classic” group. Because no objective standard IBHOF/Ring Magazine Fight of the Year, exists to define these competitive and entertaining matches, the Boxing Illustrated “Greatest Fights in Past 35 determination of what is a competitive or classic match is quite Years” subjective. Therefore, the authors of this article relied on determi- Classic Bout 6 Roberto Duran W 12 Iran Barkley—1989 nations made previously by the writers of the sport’s most com- IBHOF/Ring Magazine Fight of the Year monly read publications, Ring Magazine and Boxing Illustrated, as Classic Bout 7 Ivan Robinson W 10 Arturo Gatti—1998 well as members of the International Boxing Hall of Fame. The IBHOF/Ring Magazine Fight of the Year “classic” group consisted of videotaped professional boxing Classic Bout 8 Paulie Ayala W 12 Johnny Tapia—1999 matches that did not result in a fatality, but that had been chosen IBHOF/Ring Magazine Fight of the Year as “bouts of the year” by the aforementioned organizations. No Classic Bout 9 Erik Morales W 12 Marco Antonio Barrera— heavyweight matches were included in the classic matches 2000 because of the low number of fatalities in this weight class. The IBHOF/Ring Magazine Fight of the Year third group was populated by 4000 professional boxing matches Classic Bout 10 Micky Ward W 10 Emanuel Burton—2001 and named the “control” group. This group had been previously IBHOF/Ring Magazine Fight of the Year analyzed using a computer program (Punchstat; Compubox, Inc., a W, win; KO, knock out. The number following the designation for a win Manorville, NY) specifically designed for boxing and was used to or knock out is the number of rounds in the bout. represent the average bout. A video analysis of the fatal and classic groups was performed using the following definitions. Any punch that made punches are determined subjectively by individual perceptions contact with the opponent’s head or torso was considered to of the effect or potential effect that the punch would have on an have connected. Power punches are those that would result in opponent and by the type of punch. the transfer of significant force if connected. Because there is There are three main types of power punches: the hook, upper- currently no method of determining this force objectively, these cut, and cross. The hook is thrown in a side arc with a bent arm.

308 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com VIDEO ANALYSIS OF BOXING FATALITIES

It can be thrown with either hand but is typically a lead-hand punch. The power in a hook comes from the explosive rotation of the hips and shoulders. The classic hook is thrown in a horizontal plane, but the punch can also be thrown at a 45-degree angle (a “Mexican hook” or “shovel hook”), blending into the uppercut. The uppercut is thrown upwards with either hand. The uppercut travels vertically up the opponent’s chest under- neath the guard and makes contact with the chin. The power in the uppercut comes from the legs and hips. A cross is a straight punch with the athlete’s dominant hand. The rear hand crosses the body, the shoulders rotate toward the target, and the rear foot makes a half step forward, with the power coming from the push-off on the rear leg and the rotation of the shoulders. A jab was defined as a quick forward punch thrown with the lead hand. The power comes from a quarter rotation of the shoulders. This is usually not considered a power punch and is more often used to maintain distance from an opponent and to set up other punches. Jabs were included in the analysis because, although it is unknown whether or not connected jabs contribute to neurological sequelae, they can deliver significant force to the opponent’s head and are often the most frequently landed punch. This is readily apparent when the opponent’s head is FIGURE 1. Muhammad Ali and Joe Frazier at Madison Square Garden on repeatedly snapped back by this type of blow. Likewise, because March 8, 1971. The first contest between these athletes was simply called, “The it would not be possible to determine the forces transmitted to Fight of the Century.” It was witnessed by 20,455 spectators at the Garden. the cortex, midbrain, and brainstem from punches landed on It has been estimated that 300 million more watched it across the world on closed-circuit television. Muhammad Ali, arguably the most influential boxer parts of the body other than the head, all landed punches were in sports history, has experienced severe neurodegenerative disease. It is equiv- included in the analysis. The authors thought that these forces ocal whether or not participation in the sport contributed to his condition. should not be neglected because fatalities are considered the result of an accumulation of subconcussive forces, which could include forces transmitted to the neuroaxis from these blows. from this control population showed that the average number The three identified groups were contrasted with each other of punches that connected per round was 9.4, with a range of to determine whether or not using an objective resource to cal- 8.8 to 10.0, based on weight class. culate the number and types of punches landed would be a The average power punch output per boxer was 17.3 per practical method of ending a contest and to analyze differences round, with a range of 16 to 19.8 hits, based on weight class. between these populations. A computer program known as Information from this control population showed that the aver- Punchstat was used in the objective analysis of these matches age number of power punches that connected per round was 6.6 via videotape playback. This system was originally used to per boxer, with a range of 6.2 to 7.3 hits, based on weight class. chart tennis shots and was adapted to the sport of boxing in Classic Group 1985. Data entry was performed by two individuals (one for each athlete) who were well experienced in the sport and who The average number of punches per round landed in classic observed the match. Each would press a button on a keypad to bouts by the winner was 35.0, with a range of 18.2 to 72.6 (Table 2). record jabs thrown and connected, power punches thrown and In this group, the winning athlete landed an average of 120% the connected, and overall punches thrown and connected. This punches of the loser. This factor varies from 76 to 200%. The aver- program is commonly used during televised bouts and by box- age number of power punches landed in classic bouts by the win- ers and their trainers to analyze future opponents. Demographic ner per round was 23.8, with a range of 11.9 to 34.4 (Table 3). The information and competitive records were also analyzed in a average number of power punches per round thrown in classic search for differences between the fatally injured athletes and bouts by the loser was 42.9, with a range of 15.8 to 78.3 (Table 3). the losers in the classic-bout group. Data analysis was performed using a commercially available statistical software pack- Fatal Group age (Statview, version 5; SAS Institute, Inc., Cary, NC). The average number of punches per round landed by the survivor in bouts resulting in a fatality was 26.6, with a range of 17.2 RESULTS to 31.6 (Table 2). In this group, the surviving athlete landed an average of 139% the punches of the loser. This factor varied Control Group between 66 and 270%. The average number of power punches per The average punch output per boxer was 29 per round, with round landed by the survivor in bouts resulting in a fatality was a range of 27 to 32 hits, based on weight class. Information 19.9, with a range of 10.5 to 28.1 (Table 3). The average number of

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 309 MIELE AND BAILES

TABLE 2. Total punch comparison Average Average Average Average Connect Connect thrown thrown connected connected Fatal bouts percentage percentage per round per round per round per round for winner for loser by winner by loser for winner for loser Fatal Bout 1—Paret 61.8 66.1 26.1 23.2 42.2 35.1 Fatal Bout 2—Moore 66.1 42.0 27.8 20.9 42.1 49.8 Fatal Bout 3—Classen 42.9 32.7 23.1 13.4 53.8 41.0 Fatal Bout 4—Owen 66.8 135.5 29.4 43.7 44.0 32.2 Fatal Bout 5—Kim 76.0 54.9 31.6 25.8 41.5 46.9 Fatal Bout 6—Garcia 53.2 85.3 28.8 28.5 45.6 33.4 Fatal Bout 7—Carver 45.8 38.7 17.2 12.1 37.6 31.3 Fatal Bout 8—Tomasello 72.1 90.6 24.5 12.8 34.0 14.1 Fatal Bout 9—Johnson 54.2 83.0 30.1 38.7 55.5 46.6 Fatal Bout 10—Scottland 68.2 54.0 26.6 9.7 39.0 18.0 Average 60.7 68.3 26.6 22.9 43.5 34.8

Average Average Average Average Connect Connect thrown thrown connected connected Classic bouts percentage percentage per round per round per round per round for winner for loser by winner by loser by winner by loser Classic Bout 1 47.3 52.0 25.5 22.2 53.9 42.7 Grifﬁth versus Benvenuti Classic Bout 2 102.7 79.3 43.2 42.0 42.1 52.9 S. Muhammad versus Johnson Classic Bout 3 153.0 58.7 72.6 35.9 47.4 61.2 Lopez versus Ayala Classic Bout 4 96.9 51.9 43.6 28.7 45.0 55.3 S. Muhammad versus Lopez Classic Bout 5 35.5 44.6 18.2 20.1 51.3 45.0 Leonard versus Hearns Classic Bout 6 69.8 75.7 24.8 24.2 35.5 32.0 Duran versus Barkley Classic Bout 7 79.5 83.5 40.0 27.4 50.3 32.8 Robinson versus Gatti Classic Bout 8 63.2 72.1 21.8 21.0 34.4 29.1 Ayala versus Tapia Classic Bout 9 72.3 51.5 26.6 24.9 36.8 48.4 Morales versus Barrera Classic Bout 10 118.2 91.8 32.0 42.1 27.1 45.9 Ward versus Burton Average 83.8 66.1 34.8 28.9 42.4 44.5

power punches per round thrown by the fatally injured athlete in than 70% in the next to last round, and four actually had fatal bouts was 41.2, with a range of 14.4 to 92.7 (Table 3). increases in punch output during this round (Table 4).

Punch Output of Fatally Injured Boxer Demographics in the Round before Last The mean age of boxers who lost classic bouts was 26.4 years. Evaluation of whether or not a fatally injured boxer demon- The mean age of fatally injured athletes was 25.9 years. No sta- strated a precipitous decline in the number of punches thrown tistical difference in age was found between boxers who lost just before the final round revealed unexpected results. classic bouts versus those who were fatally injured (P ϭ 0.72). Compared with the boxer’s average number of punches thrown Likewise, no signiﬁcant differences were found between the per round, only two athletes had decreases in output by less weight classes involved, boxers’ stance, or country of origin

310 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com VIDEO ANALYSIS OF BOXING FATALITIES

TABLE 3. Power punch comparison Average Average Average Average Connect Connect thrown thrown connected connected Fatal bouts percentage percentage per round per round per round per round for winner for loser by winner by loser by winner by loser Fatal Bout 1—Paret 44.4 52.5 22.0 19.6 49.5 37.3 Fatal Bout 2—Moore 37.1 25.1 15.5 12.3 41.8 49.0 Fatal Bout 3—Classen 30.3 27.7 18.9 12.2 62.4 44.0 Fatal Bout 4—Owen 58.8 92.7 26.7 32.4 45.4 35.0 Fatal Bout 5—Kim 61.2 47.8 28.1 24.6 46.0 51.6 Fatal Bout 6—Garcia 45.7 24.8 23.4 9.6 51.1 38.8 Fatal Bout 7—Carver 32.4 22.4 14.4 8.8 44.4 39.3 Fatal Bout 8—Tomasello 30.0 49.6 10.5 10.1 35.0 20.4 Fatal Bout 9—Johnson 36.4 56.6 19.9 28.1 54.7 49.6 Fatal Bout 10—Scottland 42.6 14.4 19.3 4.9 45.3 34.0 Average 41.9 41.2 19.9 16.3 47.6 39.9

Average Average Average Average Connect Connect thrown thrown connected connected Classic bouts percentage percentage per round per round per round per round for winner for loser by winner by loser by winner by loser Classic Bout 1 25.3 24.5 14.9 12.0 58.7 49.0 Grifﬁth versus Benvenuti Classic Bout 2 76.7 55.1 34.3 32.7 44.8 59.3 S. Muhammad versus Johnson Classic Bout 3 63.0 44.6 34.4 27.4 54.6 61.5 Lopez versus Ayala Classic Bout 4 45.2 27.6 23.9 17.1 52.9 62.2 S. Muhammad versus Lopez Classic Bout 5 19.9 15.8 11.9 6.5 59.9 41.2 Leonard versus Hearns Classic Bout 6 41.2 44.6 19.2 15.3 46.7 34.4 Duran versus Barkley Classic Bout 7 46.8 48.2 24.7 22.0 52.8 45.6 Robinson versus Gatti Classic Bout 8 53.2 47.2 19.1 16.9 35.9 35.9 Ayala versus Tapia Classic Bout 9 60.2 42.6 24.2 22.7 40.1 53.2 Morales versus Barrera Classic Bout 10 110.2 78.3 31.4 38.6 28.5 49.3 Ward versus Burton Average 54.2 42.9 23.8 21.1 47.5 49.2

between losers of classic bouts versus fatally injured athletes. knockout (TKO) in their career versus six of the losers of the The mean round at which the match ended in the classic bouts classic bouts (P ϭ 0.58). Four of the athletes who were fatally was 12.6 versus 10.9 for the fatal matches. This was statistically injured had experienced three or more previous knockouts ver- signiﬁcant (P ϭ 0.03) (Tables 5 and 6). sus two of the losers of the classic matches (P ϭ 0.36). None of the losers of the classic bouts had experienced a KO or TKO in Previous Trauma History the previous 6 months. Three of the fatally injured boxers had The average period of time between the classic-match losers’ experienced a KO or TKO within 6 months of the fatal match previous bout was found to be 3.9 versus 3.4 months in the (P ϭ 0.07) (Table 7). fatally injured athletes (P ϭ 0.68). Five of the fatally injured ath- The mean number of career losses for losers of classic bouts letes had experienced a previous knockout (KO) or technical was 4.2 versus 5.4 in the fatally injured athletes (P ϭ 0.57). The

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 311 MIELE AND BAILES

TABLE 4. Punch output of fatally injured boxer round before last round Average number Percentage of punches Percentage of punches Athlete of punches thrown in next to thrown in next to last thrown per round last round of bout round versus average Fatal Bout 1 Benny Kid Paret—1962 66 73 (Round 11) 111 Fatal Bout 2 Davey Moore—1963 42 37 (Round 9) 88 Fatal Bout 3 Willie Classen—1979 33 49 (Round 9) 148 Fatal Bout 4 Johnny Owen—1980 136 112 (Round 11) 82 Fatal Bout 5 Deuk Koo Kim—1984 55 75 (Round 13) 136 Fatal Bout 6 Jimmy Garcia—1995 85 59 (Round 10) 69 Fatal Bout 7 Randie Carver—1999 39 34 (Round 9) 87 Fatal Bout 8 Bobby Tomasello—2000 91 62 (Round 9) 68 Fatal Bout 9 Stephan Johnson—2000 83 78 (Round 9) 94 Fatal Bout 10 Beethavean Scottland—2001 54 73 (Round 9) 135

TABLE 5. Demographics of classic bout losersa Round bout Age at time Athlete Division Stance Country of origin ended/result of bout Classic Bout 1 Emile Grifﬁth—1967 L15 Middleweight Orthodox 29 Virgin Islands (U.S.) Classic Bout 2 Marvin Johnson—1977 KO12 Light heavyweight Southpaw 23 United States Classic Bout 3 Mike Ayala—1979 KO15 Super bantamweight Orthodox 21 United States Classic Bout 4 Yaqui Lopez—1980 KO14 Light heavyweight Orthodox 29 Mexico Classic Bout 5 Thomas Hearns—1981 KO14 Light middleweight Orthodox 23 United States Classic Bout 6 Iran Barkley—1989 L12 Middleweight Orthodox 29 United States Classic Bout 7 Arturo Gatti—1998 L10 Light welterweight Orthodox 26 Canada Classic Bout 8 Johnny Tapia—1999 L12 Featherweight Orthodox 32 United States Classic Bout 9 Marco Antonio Barrera—2000 L12 Super featherweight Orthodox 26 Mexico Classic Bout 10 Emanuel Burton—2001 L10 Light welterweight Orthodox 26 United States

a L, loss; KO, knock out. mean number of losses in the past six bouts for the classic of 72.8% in the opponents of the fatally injured athletes match losers was 0.6 versus 1.4 for the fatally injured athletes (P ϭ 0.28) (Table 7). (P ϭ 0.22) (Table 7). Experience and Ability DISCUSSION Athletes who lost in the classic matches had a mean of 41.5 Boxing will always be a violent sport that poses risks of previous bouts in their professional career versus 35.4 in the injuries to its athletes. Although the death or permanent dis- fatally injured boxers (P ϭ 0.41). Losers of the classic matches ability of a boxer during competition is infrequent, it is too had previously lost a mean of 4.2 matches versus 5.4 losses in common an occurrence to be overlooked. We believe that ring- the fatally injured athletes (P ϭ 0.57). The previous six oppo- side physicians should be approved and used by the state com- nents of the athletes who had lost classic bouts had a mean of mission and should have a solid knowledge of the neurologi- 23.2 wins in their previous six bouts. The previous six oppo- cal examination and the particular risks to the central nervous nents of the fatally injured athletes had a mean of 19.0 wins in system associated with the sport. However, quite frequently, an their previous six bouts (P ϭ 0.23) (Table 7). athlete does not exhibit neurological signs or symptoms until after the injury has occurred. It would, therefore, be in the ath- Opponent Ability letes’ best interests to develop an objective method of halting a The opponents of the losers in the classic bouts had inflicted competition before the development of significant injury. a mean of 25.2 KOs and TKOs on previous opponents versus Currently, ringside physicians stop a bout based on an ath- 17.8 in the opponents of the fatally injured athletes (P ϭ 0.21). lete’s ability to defend him- or herself or his or her ability to The opponents of the losers in the classic bouts had a mean of mount an offense. However, the ability of an athlete who is 64.2% KO/TKO victories in previous opponents versus a mean receiving significant trauma to throw a knockout punch does

312 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com VIDEO ANALYSIS OF BOXING FATALITIES

TABLE 6. Demographics of fatally injured boxers Round bout Age of Athlete Division Stance Country of origin ended and result death Fatal Bout 1 Benny Kid Paret—1962 12/L TKO Welterweight Orthodox 25 Cuba Fatal Bout 2 Davey Moore—1963 10/L TKO Featherweight Orthodox 29 United States Fatal Bout 3 Willie Classen—1979 10/L KO Middleweight Orthodox 29 United States Fatal Bout 4 Johnny Owen—1980 12/L KO Bantamweight Orthodox 24 United Kingdom Fatal Bout 5 Deuk Koo Kim—1984 14/L KO Lightweight Southpaw 23 Korea Fatal Bout 6 Jimmy Garcia—1995 11/L TKO Super featherweight Orthodox 23 United States Fatal Bout 7 Randie Carver—1999 10/L TKO Super middleweight Southpaw 24 United States Fatal Bout 8 Bobby Tomasello—2000 10/D PTS Featherweight Orthodox 25 United States Fatal Bout 9 Stephan Johnson—2000 10/L KO Light middleweight Orthodox 31 United States Fatal Bout 10 Beethavean Scottland—2001 10/L TKO Light heavyweight Southpaw 26 United States

TABLE 7. Records of fatally injured boxers Number of Record Number Number Number of Number of knockouts/ Months of last six of knock- of technical knockouts/ knockouts/ Record of technical since Record at opponents outs knockouts technical technical Athlete previous knockouts/ previous time of bout during experienced experienced knockouts knockouts six bouts both in pre- bout previous before before 6 months previously vious wins six bouts fatal bout fatal bout prior to bout by opponent by opponent Fatal Bout 1 Benny Kid Paret—1962 3 W36(KO11)L12D3 W2L4D0 W28L5D3 3 0 2/0 4/6 14%/21%/48% Fatal Bout 2 Davey Moore—1963 1 W59(KO30)L7D1 W6L0D0 W24L10D2 0 1 0/0 25/4 66%/11%/76% Fatal Bout 3 Willie Classen—1979 1 W14(KO9)L7D2 W1L4D1 W29L7D0 3 0 2/0 12/0 100%/0/100% Fatal Bout 4 Johnny Owen—1980 3 W42L5D1 W6L0D0 W18L17D1 0 0 0/0 32/3 76%/7%/83% Fatal Bout 5 Deuk Koo Kim—1984 4 W17(KO8)L1D1 W6L0D0 W31L11D2 0 0 0/0 12/8 50%/33%/83% Fatal Bout 6 Jimmy Garcia—1995 6 W35(KO25)L5D0 W5L1D0 W9L12D2 0 0 0/0 18/4 45%/10%/55% Fatal Bout 7 Randie Carver—1999 1 W23(KO14)L1D1 W5L0D1 W13L23D0 0 0 0/0 4/3 36%/27%/64% Fatal Bout 8 Bobby Tomasello—2000 3 W14(KO6)L0D1 W6L0D0 W5L31D0 0 0 0/0 8/8 44%/44%/89% Fatal Bout 9 Stephan Johnson—2000 1 W27(KO18)L9D1 W3L3D0 W24L9D3 0 3 0/2 7/9 25%/32%/57% Fatal Bout 10 Beethavean Scottland—2001 11 W20(KO9)L7D2 W4L2D0 W9L25D2 2 0 0/0 2/9 13%/60%/73%

not decrease his or her risk of neurological sequelae. Therefore, it is the most accepted “real-time” method of objectifying if the goal is to eliminate fatalities, this should not be the fac- punches thrown and landed during a match, and it has been tor used to determine whether or not an athlete can continue shown to be practical to perform and quickly interpreted to participate. Likewise, whereas the continuous review of (results of a round are routinely displayed before or early into scorecards could decrease injuries experienced in noncompet- the next round). The system is by no means 100% accurate, itive matches, it would have little consequence on severe but precision has been sacrificed for timeliness, simplicity, injuries in highly competitive matches. Perhaps a paradigm and practicality. Thus, although more accurate methods exist shift is needed in this standard, with matches being halted to analyze a bout after the fact, Compubox/Punchstat is the based objectively on the amount of punishment participants best method we currently have for timely analysis during the have received. match that might make a difference in preventing injuries. A goal of this report was to determine whether or not the One of the major flaws with Compubox is that it recognizes use of computerized punch counting could aid in identifying only two broad categories, jabs and power punches; the bouts that have the potential to result in fatalities. One possi- strength and effect of landed blows is minimally considered. ble use of this data could be to alert the ringside physician if Analysis of the forces generated by individual blows would the athletes are entering a “danger level” concerning the num- be the next, and more important, step in developing a stan- ber of punches thrown or landed. Although Punchstat was dardized, objective method of determining whether or not a originally designed to add to the spectators’ enjoyment of a contest should be halted. This method would likely involve boxing match by introducing numeric trivia, it was chosen to the development of accelerometers to be worn in mouthpieces be used during this study for several reasons. Although a or headgear and the establishment of an upper threshold for frame-by-frame review of the system has not been performed, forces incurred before an athlete is required to stop compet-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 313 MIELE AND BAILES

These included the number of punches landed per round, the number of power punches landed per round, and the number of power punches thrown per round by losing boxers (P Ͻ 0.001 for all). Although these results were encouraging with respect to isolating the fatal-bout group, when the fatal group was compared with the classic group (which represented the more competitive matches), all of the previous differences disappeared. Therefore, the focus was placed on differentiating between these two groups. No significant difference was found in the number of punches landed per round, and no significant discrepancy in blows landed between com- petitors was found between the fatal and classic fights. Like- wise, no significant difference was found in the number of punches landed per round or in the number of power punches thrown per round by the losers. In fact, boxers in the 10 classic bouts landed an average of 10 more punches a round than those involved in the fatal matches. Whether or not fatigue is associated with fatalities was analyzed indirectly by measuring the number of punches thrown by the loser or fatally injured athlete and the length of the match. It was assumed that there would be more likelihood for fatigue and that its role would become more rele- FIGURE 2. First row, from left to right, Benny Kid Paret and Stephan Johnson. Second row, Davey Moore, Beethavean vant as the number of punches Scottland, and Jimmy Garcia. Third row, Deuk Koo Kim, Willie Classen. Bottom row, Bobby Tomasello, Randie thrown by an athlete and the Carver, and Johnny Owen. duration of the match increased. The highest average ing, irrespective of whether or not he or she is knocked out or number of punches landed per round in the fatal matches was exhibits neurological deficits. Before moving to this next step, from an athlete who died (Fatal Bout 4); and two of the athletes however, it should be established whether or not an already who had higher averages of punches landed per round in their existing method of simply counting punches could be used to matches died (Fatal Bouts 4 and 9). A statistically significant determine when a bout should be halted. difference in the length of the match was observed, with clas- Several statistically significant differences were discovered sic bouts lasting a mean of 1.7 rounds longer than bouts in the between matches that resulted in fatalities and the control group. fatal group. This could be the result of selection; only four of

314 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com VIDEO ANALYSIS OF BOXING FATALITIES

the matches in the fatal group (versus five of the classic bouts) COMMENTS occurred before rule changes in 1984, which lowered the number of rounds allowed from 15 to 12. his fascinating study analyzed fatal boxing matches in an attempt to determine whether or not warning signs were present that could Analysis of all available demographic data was performed in T be used to alert ring officials to stop a fight before a fatality occurs in an attempt to identify the factors that place athletes at greater future matches. The authors grouped boxing matches into three cate- risk of fatality in the fatal versus the classic groups. No signif- gories: 10 matches that resulted in the death of a participant, 10 “clas- icant differences were found between age, weight classes, sic” matches that were extremely competitive and closely fought, and stance, or country of origin. Previous traumatic injury to the 4000 “average” professional bouts. The authors analyzed the types and brain, especially recent insult, could increase an athlete’s risk of numbers of punches as well as demographic data. In retrospect, the serious neurological injury. However, no significant differences results were not surprising. The fatal bouts differed from the average were found between the classic and fatal groups in the average bouts in such factors as numbers of punches landed per round, but period of time between bouts, previous KO and TKO history, or these differences were not significant when the fatal matches were previous losses. compared with the most competitive matches. This is an admirable effort, but it is obvious that more work needs to The experience and ability of the athletes was also consid- be done if we are to develop some type of warning system that can be ered in determining risk of injury. No significant differences used at ringside. Perhaps focusing on the numbers and types of blows were found between the classic and fatal groups with regard to landed to the head will yield more useful data. Laboratory studies that professional boxing experience and success. The opponent’s help elucidate the pathophysiological mechanisms underlying boxing- ability must also be considered when determining the risk of a related fatalities may yield clues about clinically useful potential warn- match; the risk should theoretically go up when an opponent is ing systems. known to throw a harder punch. Once again, there was no statistical difference between the fatal and classic groups based on Alex B. Valadka the opponent’s previous record of KO and TKO victories. Houston, Texas

CONCLUSION y carefully analyzing video of boxing matches in which one of the Bboxers died and comparing that video with other competitive The data analyzed between the control group and the fatal- matches, Miele and Bailes have found that the likelihood of death is not bout group suggest that a computerized method of counting simply a matter of the number or force of punches landed. This is landed blows at ringside could provide sufficient data to stop counterintuitive and suggests that there was an anatomic, or perhaps matches that might result in fatalities. However, such a process even physiological, predisposition to severe brain injury in at least would become less effective as matches become more compet- some of the boxers who died. A more tragic consequence of professional boxing is the cognitive and behavioral damage caused by itive. This demonstrates what many experts in the sport repeated head trauma, which is not addressed in this article. Previous already know: there is little difference between classic and fatal studies suggest a clear relationship between the length of time spent matches. In fact, based on current computerized punch data, boxing and the likelihood of becoming permanently disabled as a result this method of stopping a contest would essentially eliminate of cognitive dysfunction. Because of this and the increased risk of what many consider to be the most competitive and exciting death, it is difficult for me to endorse the sport even if it is becoming matches. Likewise, a change in the rules that would signifi- increasingly popular among both men and women. cantly decrease the competitive nature of the game for an unproven increase in athlete safety is a solution that would be Donald Marion Boston, Massachusetts as controversial as the sport itself. It may depend on the science within sports medicine to determine an effective, objective method for identifying which boxers are at highest risk before he authors analyzed three groups of professional boxing matches they experience catastrophic injury. Tincluding a “fatal” group, a “classic” group that represented very competitive matches, and a “control” group of 4000 professional box- REFERENCES ing matches representing average bouts. A computer program known as Punchstat (CompuBox, Inc., Manorville, NY) was used in the objec- 1. Bailes JE, Cantu RC: Head injury in athletes. Neurosurgery 48:26–46, 2001. tive analysis of these matches. After their analysis, the authors found a 2. Jordan B: Sports injuries. Presented at Proceedings of the Mild Brain Injury in number of variables that were significantly different between matches Sports Summit, Dallas, TX, April 16–18, 1994. that resulted in fatalities and the control group. These included the 3. Jordan BD: Chronic traumatic brain injury associated with boxing. Semin number of punches landed per round, the number of power punches Neurol 20:179–185, 2000. landed per round, and the number of power punches thrown per 4. Lundberg GD: Boxing should be banned in civilized countries. JAMA round by losing boxers. When the fatal bouts were compared with the 249:250, 1983. most competitive bouts (classic group), these differences were no 5. Lundberg GD: Boxing should be banned in civilized countries—Round 2. longer significant. The authors concluded that a computerized method JAMA 251:2696–2698, 1984. of counting landed blows at ringside could provide sufficient data to 6. Lundberg GD: Medical arguments for nonparticipation in boxing, in BD J (ed): Medical Aspects of Boxing. Boca Raton, CRC Press, 1992, pp 11–15. stop matches that could result in fatalities, but that such would also 7. Miele VJ, Bailes JE, Voelker JL: Boxing and the neurosurgeon. Clin Neurosurg decrease the competitive nature of the sport. They suggest that other 49:396–406, 2002. methods of quantifying acceleration-deceleration brain injuries are nec- 8. Richards NG: Ban boxing. Neurology 34:1485–1486, 1984. essary to improve the safety of boxing.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 315 MIELE AND BAILES

We agree that a better assay is required to help identify which missions should be experts in the assessment of head injury. In pro- matches (and which participants) will be associated with mortality. fessional boxing, there have been 232 mortalities as the result of head The problem with Punchstat is that it is a tool developed to give the injury since 1950. Since 1983, there have been 53 mortalities. The year announcers another option for discussion during the proceedings of the 1983 is of significance because championship bouts were reduced fight. It is not significantly different than the scoring system used in from 15 to 12 rounds after the death of Duk Koo Kim. As a conse- amateur boxing, which has also been compromised by confusion. The quence of that fatal fight, studies were carried out which suggested consideration of the impact associated with a “power punch” or the that a boxer usually receives the most brutal hits after Round 12. As actual contact required for a “jab” to be recorded by a depressed but- a result of this study, the World Boxing Council soon shortened its ton at ring-side is not of scientific merit. Punchstat cannot accurately title bouts to 12 rounds. The World Boxing Association and World predict when a boxer who is ahead on points will be knocked out nor Boxing Organization followed in 1988 and the International Boxing is it necessarily indicative of the actual winner in a fight. Punchstat Federation followed in 1989. The necessity of this change is question- showed Oscar De La Hoya landing approximately 300 more punches able. The mean knockout round was 7 ± 3 with the majority of knock- than Félix Trinidad despite the fact Trinidad won the fight by a split outs occurring in Round 10. Since 1983, this has not significantly decision. Punchstat also showed Lennox Lewis outlanding Evander changed. In addition, mortality by decade has not changed signifi- Holyfield by a large margin during their first fight, which was ulti- cantly: 61 in the 1950s, 62 in the 1960s, 20 in the 1970s, 29 in the 1980s, mately declared a draw. Currently, Punchstat is used by HBO, ESPN, 29 in the 1990s, and 33 to date in the 2000s. One might even consider and Top Rank Boxing. We agree completely that other methods of the need to revamp the medical care provided at ringside, given the quantifying acceleration-deceleration injuries are necessary to improve propensity of mortalities to form temporal clusters in the United the safety of boxing, but do not agree that Punchstat is a good starting States; in New York, there were eight fatalities between 1960 and point given the lack of differentiation between body shots, head shots, 1962; in Massachusetts, there were five fatalities from 1954 to 1956. At occipital shots (rabbit punches), punches thrown at a distance, in the this point, it is impossible to say whether or not the proper direction mid-range or in close proximity. should be in regard to continuous review of the scorecards, the The importance of this study is that it supports the notion that cri- inequality of the match, and the potential knockout capacity of the teria should be developed to enhance the safety of professional box- losing fighter. ing. We think the inclusion of the heavyweight class would further cloud the analysis given the significance of “power punches” in that Freddie Roach class and the relative absence of fatalities. Since 1983, mortality by Boxing Trainer weight class has been greatest in fly weight (112 pounds, 22%) and Dan Goosen bantam weight (118 pounds, 22%) divisions and lowest in the heavy- Boxing Promoter weight division (1%). Los Angeles, California The primary problems could be related to medical care and evalu- Michael L. Levy ation at the time of the fight. Physicians approved by the state com- San Diego, California

John ‘Jack’ Broughton (1704–1789), third heavyweight boxing champion of England, introduced in 1743 the ﬁrst set of rules of modern boxing (Broughton’s Rules) and invented “mufﬂers,” the precursors of modern boxing gloves. Broughton (left), depicted in this mezzotint by John Young, is remembered as the “Father of the English School of Boxing.” Broughton’s Rules were used for nearly 100 years. CLINICAL STUDIES

RISK FACTORS ASSOCIATED WITH POSTCRANIOTOMY MENINGITIS

Irene S. Kourbeti, M.D. OBJECTIVE: The authors conducted a retrospective cohort study to determine the inci- Department of Internal Medicine, dence, bacteriological features, and risk factors for postcraniotomy meningitis. University Hospital of Crete, and Graduate Program Medical School, METHODS: Patients older than 18 years who underwent nonstereotactic craniotomies University of Crete, between January 1996 and March 2000 and who survived for more than 7 days were Voutes, Heraklion, Greece included. Operations for placement of burr holes and shunts were excluded. Records of the first 30 postoperative days were abstracted. Host factors, types of craniotomy, and Anke V. Jacobs, M.D. pre- and postoperative variables were evaluated as risk factors for meningitis Department of Internal Medicine, Metropolitan Hospital Center and RESULTS: Among 453 patients, there were 25 cases of meningitis. Eight out of 12 culture- New York Medical College, positive cases were the result of gram-positive cocci. Four hundred twenty (92%) patients New York, New York received antibiotic prophylaxis, most commonly a first-generation cephalosporin. In multivariate analysis, the risk of meningitis was increased by surgery that entered a Maxim Koslow, M.D. sinus (odds ratio [OR], 4.49; P ϭ 0.018), an increase in the American Society of Department of Neurosurgery, ϭ Bellevue Hospital Center and Anesthesiologists score (OR, 1.72; P 0.023), and increases in the number of days of New York University external ventricular drainage (OR, 1.21; P ϭ 0.049) and intracranial pressure monitor- School of Medicine, ing (OR, 1.24; P ϭ 0.002). New York, New York CONCLUSION: Access of upper airway bacteria to the surgical wound, host factors as Dimitris Karabetsos, M.D. expressed by the American Society of Anesthesiologists score, and duration of device- Department of Neurosurgery, related postoperative communication of the cerebrospinal fluid and the environment are University Hospital of Crete, major risk factors for postoperative meningitis after craniotomy. Voutes, Heraklion, Greece KEY WORDS: Craniotomy, Meningitis, Neurosurgical infections Robert S. Holzman, M.D. Neurosurgery 60:317–326, 2007 DOI: 10.1227/01.NEU.0000249266.26322.25 www.neurosurgery-online.com Department of Medicine, Division of Infectious Disease and Immunology, Bellevue Hospital Center and New York University osocomial meningitis is seen most com- clean-contaminated procedures, it is higher. School of Medicine, monly in neurosurgical patients (30, Empiric treatment must be established on sus- New York, New York N50). During a 27-year period, 40% of picion, thus the prevalence and the sensitivities the bacterial meningitis cases seen at the of the pathogens must be determined institu- Reprint requests: Massachusetts General Hospital were nosoco- tion by institution. Robert S. Holzman, M.D., Department of Medicine, mial, and 83% of them were seen in patients Comprehensive studies and reviews have New York University who had undergone invasive neurosurgical been published on the risk factors for neuro- School of Medicine, procedures, carried a neurosurgical device, or surgical site infections, including meningitis (3, 550 First Avenue, were admitted with a head injury with or 11, 21, 23, 31 , 33, 51, 54). However, few of these New York, NY 10016. without a cerebrospinal fluid (CSF) leak (10). articles focus specifically on the risk factors Email: [email protected] Meningitis may result in severe morbidity with that predispose patients to meningitis after Received, May 2, 2005. a prolonged length of stay, multiple surgeries, major craniotomy. In 2002, Reichert et al. (43) Accepted, October 3, 2006. and higher hospital costs (23, 43). In a retro- conducted a case control study at a hospital in spective review of the infections in a neurosur- Brazil that identified external ventricular gical intensive care unit, patients undergoing shunting, remote site infection, and repeat cranial surgery were far more likely to have operation as risk factors for meningitis, with secondary infection associated with the proce- only repeat operation remaining significant dure than those who underwent spinal or after the multivariate analysis. We conducted a transsphenoidal surgery (21). In clean neuro- retrospective cohort study to further define the surgery, the rate of development of postopera- risk factors, manifestations, and microbiology tive bacterial meningitis is low (1–2%); in of postcraniotomy meningitis.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 317 KOURBETI ET AL.

PATIENTS AND METHODS tions outside the nervous system, including superficial wound and bone flap infections. Identification of Patients Computer-based medical record indexes were reviewed at Statistical Analysis Bellevue and Tisch Hospitals at New York University Medical Data were transferred from the collection forms to a database Center. Patients were eligible for the study if they were at least (Epi Info, Version 6.04; Centers for Disease Control, Atlanta, 18 years of age, underwent elective or emergency craniotomy GA) and were analyzed using either Epi Info or SPSS software between January 1996 and March 2000, and survived at least (Version 11; SPSS Inc., Chicago, IL). Continuous variables were 7 days after surgery. compared using Student’s t test or the Mann-Whitney U test, Only major craniotomies were included in this review. whereas categorical variables were compared using Fisher’s Patients having only cerebrospinal shunt or external ventricu- exact test. lar device implantations, burr hole trepanation, or stereotactic Rate ratios and their 95% confidence intervals (CIs) were surgery were excluded. The medical records of all eligible computed using the StatCalc module of Epi Info version 6.04. patients were surveyed retrospectively for the presence of Odds ratios and their CIs were computed by logistic regression meningitis during the first 30 postoperative days and during in SPSS. Stepwise logistic regression was used to model the the first postoperative year if a foreign body was implanted. It interactions of those variables significantly associated with was assumed that the occurrence of meningitis would result in postoperative meningitis in univariate analyses. A backward ϭ ϭ readmission; however, it is possible that admission to another elimination model was used with Penter 0.20 and Pleave 0.05. institution may have resulted in some missed cases. SPSS provides three test methods for variable selection, all of which yielded the same model. When a continuous variable, Data Abstraction such as the number of days of ventricular drainage, was shown Data were abstracted from the medical chart to a standard to be a statistically significant risk factor, the relationship was form. We recorded the presence of diabetes, cancer, atheroscle- modeled further using suitably coded dummy variables. Power rotic vascular disease, chronic renal failure, chronic obstruc- calculations were performed using PEPI for Windows version tive pulmonary disease, preoperative American Society of 1.31 (www.brixtonhealth.com). Anesthesiologists (ASA) score, indication for craniotomy (tumor, vascular, trauma, brain abscess, or other), previous presence and type of any foreign device, site of surgery (supra- RESULTS tentorial or infratentorial), procedure urgency (elective or emergent), length of surgery, concomitant procedures (simultaneous Study Population orthopedic or abdominal surgery, facial reconstruction), The study included all 453 patients who met the inclusion implantation of a foreign body (vascular clips, stimulating elec- criteria and underwent a craniotomy between January 1, 1996, trodes, acrylic implants, titanium plates, dural homografts), and March 31, 2000. Twenty-five (5.5%) patients experienced and the presence of any postoperative CSF drainage. Details of meningitis after craniotomy. Fourteen (5.1%) out of 275 men hair removal were recorded, including the method (clipping, and 11 (6.2%) out of 178 women experienced meningitis. The shaving, or no preparation) and time before incision (Յ2 h average age (Ϯ standard deviation) of meningitis patients was before or Ͼ2 h after the incision). Prophylactic antibiotic regi- 49.3 Ϯ 17.1 years; the average age for those without meningi- mens and their relation to incision time were also recorded. tis was 51.8 Ϯ 13.0 years. The median ages of those with and Reoperations and reports of CSF leak were noted. An operation without meningitis were 54 years (quartiles 43, 60) and 48 was considered a reexploration whenever it was performed years (quartiles 32, 62), respectively. Three hundred seventy- through the same incision as a previous operation, regardless of seven patients underwent a supratentorial craniotomy and the interval between the two procedures. Postoperative CSF 67 patients underwent an infratentorial craniotomy; in nine leak was recorded when such drainage was diagnosed after patients, the procedure involved both areas. Two patients surgery on the basis of otorrhea, rhinorrhea, or leakage from (8.0%) in whom meningitis developed died before discharge, the surgical wound. as did 31 (7.3%) of those without meningitis. Neither of the Meningitis was diagnosed according to the definitions of the two patients with meningitis died as a result of meningitis. National Nosocomial Surveillance System as follows: either 1) Meningitis occurred in 14 (5.4%) out of 260 patients whose a gram stain or CSF culture, or both, demonstrating a microor- hair was removed by shaving in the operating room immedi- ganism or 2) CSF leukocytosis with increased protein concen- ately before surgery and in five (5.7%) out of 87 patients who tration, decreased glucose concentration, or both associated had no hair removal. A depilatory was used in only one with fever and nuchal rigidity, with 3) antibiotic treatment pre- patient. In 105 patients, no mention of hair removal was made. scribed by the attending physician (19). We recorded the post- Four hundred twenty (92.7%) out of the 453 patients received operative day on which the diagnostic lumbar puncture was perioperative antibiotic prophylaxis. Three hundred twenty- performed; the antibiotic or steroid regimen, or both, initiated seven (78%) patients received a first-generation cephalosporin, before and at the time of the lumbar puncture; all organisms 11 (2.6%) patients received another cephalosporin, 62 (14.7%) identified by CSF culture; and any recorded concurrent infec- patients received vancomycin, two (0.4%) patients received a

318 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com POSTCRANIOTOMY MENINGITIS AQ:1

penicillinase-resistant penicillin, and one (0.2%) patient CSF leak, and placement of the intracranial pressure drain in received penicillin G. Seventeen (4%) patients received an the intensive care unit rather than in the operating room. antibiotic combination. Statistically significant variables in the univariate analyses For patients with and without meningitis, the distribution of were included in a multivariate logistic regression. Table 2 com- ASA scores and indications for surgery, duration of surgery, pares the results of univariate analysis for each variable with a and use of drains and the percentage of patients having an P value of 0.20 or less and the results of the multivariate mod- emergent procedure, a procedure through a sinus, concomi- eling. The multivariate odds ratio is shown for variables tant infections, or multiple procedures during the craniotomy remaining in the model. For the variables removed from the can be derived from Table 1. The median duration of surgery model, the table lists the P value for removal of the variable was 4.0 hours (quartiles 3.0, 6.0) for those in whom meningitis from the final model. did not develop and 6.0 hours (quartiles 3.75, 7.25) for those in Surgery through a sinus (OR, 4.49), increasing ASA score whom it did. Sixty-one percent of the patients had some kind (OR, 1.73), duration of intracranial pressure monitoring (OR, of drain inserted after surgery. 1.24), and duration of ventricular drainage (OR, 1.21) remained in the model as independent predictors with P values of less Meningitis Risk than 0.05. Use of vascular clips, duration of surgery, duration of Table 1 lists all variables that were examined for their contri- galeal drainage, and presence of a concomitant or previous bution to the risk of meningitis. Individual variables that were infection were variables that were removed from the model statistically significant at a P value of less than 0.05 included but had P values between 0.05 and 0.10. By comparing various the presence of infection at another site before diagnosis of logistic models, it was determined that the association with meningitis, performance of multiple procedures at the time of use of vascular clips was actually because patients having oper- craniotomy, emergency scheduling, increasing ASA score, ations with vascular clipping also had a significantly prolonged increasing duration of surgery, surgery for a vascular indica- course of drainage and intracranial monitoring (means, 2.38 tion, use of vascular clips, use and duration of use of an versus 0.55 d for intracranial pressure monitoring; 1.61 versus intracranial pressure monitor, use and duration of use of a 0.28 d for ventricular drainage; and 2.10 versus 0.88 d for sub- ventricular drain, and use and duration of use of a galeal galeal drainage; each difference is statistically significant with drain. Although the presence of a concomitant infection at P Ͻ 0.001, Mann-Whitney U test). another site was associated with the development of meningi- We also investigated the relation between the use of galeal tis, this concomitant infection was not a surgical site infection drains and a traumatic indication for craniotomy. In our series, in any patient. galeal drains were used in 95 (81.9%) out of 116 patients in Among 178 patients with no postoperative CSF drainage, whom trauma was an indication and in 92 (45.1%) out of 204 four patients (2.3%) experienced meningitis. The insertion of oncological patients. We examined the interaction of indica- any type of drain was associated with a 3.3-fold increase in the tion and galeal drainage in this subset of 320 patients. rate of meningitis (95% CI, 1.2–9.5). For ventricular drainage, Craniotomies for trauma had a 1.39-fold (nonsignificant) the increase was 9.2-fold (95% CI, 2.8–29.6); for intracranial increase in risk compared with those for tumor. Adjustment pressure monitoring, the increase was 5.6 (95% CI, 1.8, 17.4); for for the use of galeal drains reduced the risk associated with a subgaleal drainage, the increase was 3.1 (95% CI, 1.2–8.0); and trauma indication to 0.977, which is close to one. In contrast, for lumbar drainage, the increase was 2.1 (95% CI, 0.7–6.5). craniotomies in which galeal drains were used had had a 3.24- Risk of meningitis increased with increasing duration of fold increase in risk (P ϭ 0.07); adjusting the risk by indication drainage. The relation of the rate ratio for meningitis with dura- had essentially no effect (adjusted OR, 3.26; P ϭ 0.08). Thus, our tion of device use is depicted in Figure 1 for ventricular data suggest a role for galeal drainage in explaining any drainage, Figure 2 for subgaleal drainage, and Figure 3 for increased risk in operations for trauma, but no role for trauma intracranial pressure monitoring. in explaining the association with galeal drainage. Thirty-three patients received no prophylactic antibiotics, four (10.5%) of whom experienced meningitis, in contrast to the Clinical Manifestations and CSF Features 21 cases of meningitis seen in 415 (5.1%) patients who did Fever was the main presenting sign in 96% of the meningitis receive prophylactic antibiotics. The observed doubling of risk patients; nuchal rigidity and headache were the main present- was not statistically significant (P ϭ 0.148, Fisher’s exact test). ing signs in 16%; altered consciousness was the main present- None of the following were associated with the developing sign in 8%; and nausea and vomiting were the main pre- ment of meningitis: age, sex, duration of the hospitalization senting signs in 4%. The mean interval between the date of before craniotomy, timing of prophylactic antibiotic adminis- surgery and the development of symptoms was 6.7 days, with tration, type of the antibiotic administered, presence of cranial a minimum of 1 day and a maximum of 65 days. Although the fracture in patients with trauma, use of steroids before sur- mean interval (Ϯ standard deviation) was 4.3 Ϯ 2.6 days for gery, type of depilation, approach for the craniotomy (supra- culture-positive patients and 9.2 Ϯ 19.9 days for culture- tentorial or infratentorial), past medical history of the patient, negative patients, the difference was not statistically signifi- previous craniotomy during the same admission, placement cant. Fourteen of the infected patients received postoperative of a foreign body other than a vascular clip, postoperative antibiotics for reasons other than prophylaxis before meningi-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 319 KOURBETI ET AL.

TABLE 1. Univariate factors analyzeda No. Rate in at- Odds ratio (95% Factor P value at risk risk group (%) conﬁdence interval) Meningitis in total study population 453 5.5 Age (no. of yr) 453 5.5 1.00 (0.98–1.03) 0.463 Age greater than median age (48 yr) 209 6.3 1.33 (0.59–3.00) 0.487 Male sex 275 5.1 0.81 (0.36–1.83) 0.62 Presence of another infection before meningitis 77 16.9 6.16 (2.69–14.11) Ͻ0.001 Surgery through a sinus 29 13.8 3.06 (0.98–9.60) 0.055 Multiple procedures at time of craniotomy 40 15.0 3.85 (1.43–10.35) 0.007 Emergency procedure 155 9.7 2.48 (0.61–3.80) Ͻ0.001 ASA score (numeric) 2.02 (1.32–3.09) 0.001 1 26 0 1 2 206 2.4 0.33 (0.14–0.78) 0.012 3 144 8.3 0.90 (0.38–2.14) 0.819 4 57 12.3 5.42 (1.99–14.77) 0.001 5 11 9.1 16.00 (177–144.72) 0.014 Surgery duration (h) Total 447 5.3 1.12 (1.01–1.24) 0.031 Ͻ2 49 2.0 1b Ͼ2 and ≤4 177 4.0 1.97 (0.237–16.46) 0.529 Ͼ4 and ≤6 125 5.6 2.85 (0.341–23.77) 0.334 Ͼ6 and ≤8 61 8.2 4.29 (0.150–42.48) 0.191 Ͼ8 and ≤10 20 5.0 2.53 (0.15–42.48) 0.520 Ͼ10 15 20.0 12.00 (1.15–125.82) 0.038 Reason for surgery Oncological 204 4.4 1b Trauma 116 6.0 1.39 (0.50–3.84) 0.524 Vascular 71 11.3 2.75 (1.02–7.43) 0.046 Brain abscess 12 0 0 Other 50 2.0 0.44 (0.55–3.57) 0.444 Chronic obstructive lung disease 8 12.5 2.45 (0.29–20.75) 0.410 Diabetes 38 10.5 2.16 (0.70–6.66) 0.179 Atherosclerotic cardiovascular disease 27 3.7 0.63 (0.08–4.84) 0.630 Cancer 50 2.0 0.32 (0.04–2.38) 0.262 Intracranial pressure monitor 79 12.7 3.47 (1.50–8.04) 0.004 Duration of pressure monitoring (d) 1.29 (1.50–1.56) Ͻ0.001 No drain 178 2.3 Any drain (versus no drain) 278 7.6 3.49 (1.18–10.35) 0.024 Use of ventricular drain 34 20.6 5.78 (2.22–15.03) Ͻ0.001 Duration ventricular drainage (d) 1.30 (1.10–1.53) 0.002 Use of lumbar drain 28 10.7 2.20 (0.616–7.84) 0.225 Duration of lumbar drainage (d) 1.17 (0.882–1.562) 0.272 Use of Galeal drain 257 7.8 3.22 (1.19–8.75) 0.022 Duration of Galeal drainage (d) 1.37 (1.14–1.64) 0.001 Foreign body placement during surgery None 189 5.6 Clips 22 22.7 5.99 (2.00–17.87) 0.001 Cranium 50 6.0 1.11 (0.32–3.83) 0.875 Dural 51 3.9 0.66 (0.15–2.90) 0.586 Craniotomy type Infratentorial 67 6.0 1b Supratentorial 377 5.3 0.88 (0.29–2.67) 0.825 Both 9 11.1 1.97 (0.19–19.86) 0.566 Other (Leiberger or Synthes) 169 5.3 0.94 (0.41–2.18) 0.889 Duration from admission to surgery (d) 1.02 (0.96–1.07) 0.581 Cranial fracture among patients with trauma 45/116 4.4 0.61 (0.11–3.31) 0.570 No use of prophylactic antibiotics 20 10.0 1.98 (0.433–9.06) 0.378 CSF leak 11 9.1 1.70 (0.21–13.87) 0.618 Previous craniotomy during the same admission 47 4.3 0.74 (0.17–3.23) 0.685 Use of steroids 133 6.0 1.19 (0.492–2.88) 0.699 Placement of ICP drain in ICU rather than operating room 18/122c 11.1 1.32 (0.26–6.68) 0.738 a ASA, American Society of Anesthesiologists; CSF, cerebrospinal ﬂuid; ICP, intracranial pressure; ICU, intensive care unit. b For grouped variables, the odds ratios are in comparison with the reference group. For continuous variables, the odds ratios are the n-fold change in the risk for each unit change from the mean. c Data on site of insertion was only available for 122 patients.

320 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com POSTCRANIOTOMY MENINGITIS

isms predominated. Coagulase-negative staphylococci were isolated from five patients, Staphylococcus aureus from two patients and Enterococcus species from one patient. Other iso- lates included Bacillus species, Acinetobacter calcoaceticus, Pseudomonas aeruginosa, Serratia marcescens, Serratia species, Haemophilus inﬂuenzae, and Candida species. Three patients had two different species in their CSF. Concomitant infections were documented in 17 (48%) patients, most commonly nosocomial pneumonia (n ϭ 8; Table 4).

DISCUSSION FIGURE 1. Graph demonstrating the relation between the number of days of ventricular drainage and postoperative meningitis. CI, conﬁdence interval. Incidence and Microbiology The incidence of meningitis in this study was 5.5%, higher than those noted in some (4, 11, 22, 23) series. Reichert et al. (43) reported a rate of 8.9%. The studies in which lower rates were encountered may not be directly comparable because they did not explicitly exclude minor procedures. Our overall fatality rate of 8% (two patients) is comparable with that reported by others (11), but neither death in our cohort could be attributed to the meningitis. Gram-positive organisms predominated as meningitis pathogens in our cohort. In some studies, especially among those published after 1993, Enterobacteriaceae and other gram- negative rods have played a greater role (5, 6, 15, 17, 20, 27, 43, FIGURE 2. Graph demonstrating the relation between the number of days of 55), representing 70 to 80% of the pathogens in some series subgaleal drainage and postoperative meningitis. CI, conﬁdence interval. (48). Our results seem to reflect the general reemergence of gram-positive organisms as nosocomial pathogens and are in agreement with both older and some of the more recent series (3, 22, 23, 33, 51, 54).

Clinical Manifestations Fever was the main presenting manifestation of postcraniotomy meningitis in our cohort (96% of the patients). The mean duration from operation to the onset of symptoms was 6.7 days, an interval similar to that seen in other reports (43). Differentiation between bacterial and aseptic meningitis is still a problem after craniotomy, especially at the early stages when treatment for bacterial meningitis should be started (3, FIGURE 3. Graph demonstrating the relation between the number of days of 12, 25, 45). The culture-negative patients in our cohort were intracranial pressure monitoring and postoperative meningitis. CI, confidence treated with empiric antibiotics because of a high clinical sus- interval; ICP, intracranial pressure. picion for infection. Although aseptic meningitis has been reported to complicate neurosurgery (3, 7), modern methods, some of which use polymerase chain reaction techniques, tis developed. The mean duration for such antibiotic adminis- suggest that at least some of these apparently aseptic cases tration was 2.6 days. The characteristics of the CSF in patients may be bacterial (9, 46). The use of the BacT/Alert system with postoperative meningitis are shown in Table 3. There were (bioMérieux, Inc., Durham, NC) has also been reported as a no important or significant differences between culture-positive more sensitive culture method for the detection of micro- and culture-negative patients. organisms in patients with postcraniotomy meningitis than the conventional method (14). Microbiology Meningitis was documented by positive CSF culture in 12 Risk Factors (50%) out of 24 patients whose fluid was cultured. Five out of The independent risk factors identified by our multivariate the 12 had positive gram stains. No patients with a negative analysis were entry into the paranasal sinuses, increasing ASA culture had a positive gram stain. Overall, gram-positive organ- score, and the prolonged use of intracranial pressure monitor-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 321 KOURBETI ET AL.

TABLE 2. Odds ratios for the variables studied by multivariate logistic regressiona Univariate analysis Multivariate analysis Factor Odds ratiob P value Odds ratio (95% conﬁdence interval) P valuec Surgery through sinus 3.06 0.055 4.49 (1.30–15.58) 0.018 ASA score 2.02 0.001 1.73 (1.08–2.765) 0.023 Days of ICP monitoring 1.27 Ͻ0.001 1.24 (1.08–1.42) 0.002 Days of ventricular drain 1.30 0.002 1.21 (1.00–1.47) 0.049 Presence of another infection 6.16 Ͻ0.001 0.083 Vascular clipping 5.99 0.001 0.077 Ventricular drain 5.78 Ͻ0.001 0.621 Multiple procedures 3.85 0.007 0.191 Intracranial pressure monitor 3.47 0.004 0.772 Galeal drain 3.22 0.022 0.253 Vascular procedure 2.75 0.048 0.539 Emergency scheduling 2.48 Ͻ0.001 0.874 Lumbar drain 2.20 0.225 0.733 Diabetes 2.16 0.179 0.804 Days of galeal drainage 1.37 0.001 0.096 Days of lumbar drain 1.17 0.225 0.944 Hours of surgery 1.12 0.031 0.059 Cancer 0.32 0.262 0.482

a ASA, American Society of Anesthesiologists; ICP, intracranial pressure. b Confidence intervals may be found in Table 1. c P values are for the odds ratios of included variables or for the removal of the variables that were not in the final model (italicized). The criterion for removal was a P value less than 0.05. ing and ventricular drainage. After the multivariate analysis, association with ventricular drain placement was a frequent surgery that included the entering of a sinus was the most sig- problem and was observed in 11% of general neurosurgical nificant risk factor for the development of meningitis patients in the series of Mayhall et al. (29). In that study, the (P Ͻ 0.018). This finding is in accordance with the experience of risk of infection was 9% at Day 5, but was 21, 37, and 42% by other studies (22, 31), but not all (37). The association we Days 8, 10, and 11, respectively. There is little agreement observed was not as significant in previous reports (22, 31), among various series regarding the relation between duration and it was independent from the presence of CSF leakage that of drainage and infection incidence; in one report, no relation could predispose to meningitis. A surgical approach through was observed (28). the paranasal sinuses qualifies for the characterization of the In our study, the duration of ventricular drainage was signif- procedure as clean-contaminated, a category that has tradition- icantly associated with the risk of meningitis. It remained an ally been associated with a higher infection rate (33). The delay independent risk factor when adjusted for the presence of other in the expansion of the brain to seal the communication into the drains, especially when drainage continued for 5 days or more anterior fossa seems to be responsible for the bacterial contam- (Fig. 1). These findings are in accordance with other recent lit- ination; therefore, immediate mucosal repair is recommended erature demonstrating a significant association between the to prevent infection (34, 35, 49). duration of use of a ventricular device and the meningitis rate An increasing ASA score was an independent risk factor for (1), although this relationship has not been observed in all stud- postcraniotomy meningitis. As can be seen in Table 1, the risk ies (39). Schade et al. (47) considered the duration of drainage was elevated with scores of 4 and 5, the two highest categories; to be the most significant risk factor after both 5 and 15 days of 68 (15%) of our patients fell into these categories. The ASA drainage. In the study by Park et al. (36), the infection rate rose score is included in the Risk Index Score for the Surgical Site daily until the fourth day after insertion and plateaued there- Infections proposed by the National Nosocomial Infections after. Lozier et al. (26) documented that controversy exists in Surveillance system (19). Although it has been associated with the literature regarding the relationship between the duration the development of infections occurring after neurosurgery in of catheterization and risk of infection, but also observed that, the past (23, 43), it was not previously reported as an inde- in their experience, risk increased during the first 10 days. They pendent risk factor. make the point that evaluating the risk of infection when Craniotomies with an external ventricular drain carry a drainage exceeds a duration of 10 days is problematic because higher risk for meningitis (24, 28, 29, 43, 54). Ventriculitis in such prolonged external ventricular drainage is rare.

322 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com POSTCRANIOTOMY MENINGITIS

TABLE 3. Distribution of cerebrospinal ﬂuid values in cases of postcraniotomy meningitisa Percentile

Laboratory values Patients Mean Minimum 25 50 75 Maximum All 235 364 1190 WBC (/ml) Culture ϩ 2,129 0 53 340 2400 19,250 Culture– 270 388 910 All 31 58 80 Glucose (mg/dl) Culture ϩ 57 9 32 60 82 158 Culture– 26 55 70 All 67 167.5 211 Protein (mg/dl) Culture ϩ 268 15 46 99 213 1304 Culture– 93 182 228 All 0.21 0.36 0.55 CSF/blood Culture ϩ 0.41 0.07 0.21 0.41 0.57 1.21 glucose ratio Culture– 0.24 0.35 0.53 a WBC, whole blood count; ϩ, positive; –, negative; CSF, cerebrospinal ﬂuid.

Meningitis or positive CSF culture has been described as TABLE 4. Concomitant infections the most frequent complication of intracranial pressure mon- No meningitis Meningitis Site itoring (42, 52). During the period we studied, intracranial pressure monitoring devices were placed either in the (25 ؍ n) (428 ؍ n) Wound infection 5 0 parenchyma (Camino; NeuroCare, San Diego, CA) or in the Pneumonia 33 10 ventricles and were attached to an external transducer. In Urinary tract 18 3 our experience and that of others, infection is rare if the Blood stream 8 2 patient is monitored for less than 4 days but rises thereafter Other site 9 2 (Fig. 3) (42, 44). However, Winfield et al. (52) examined the daily risk of infection and concluded that routine replacement of pressure monitor devices after a fixed period is unwarranted; this has been confirmed with most recent stud- At our institutions, no standard practice was followed for ies (32). The role of prophylactic antibiotic administration routine changes of drainage tubing. Based on our data and that while pressure is monitored has not been resolved (41, 42), of Mayhall et al. (29), it is logical to recommend that a ventric- but these were not administered as a routine at our institu- ular catheter should be removed or changed after 5 days to tions during the survey period. minimize the risk of an infection. Arguing against this sugges- The presence and the duration of use of a subgaleal drain tion is the data of Holloway et al. (18), who noted that the rela- have not been associated with an increased risk for meningitis tionship of ventriculitis to monitoring duration is neither sim- in previous studies. In our study, it was found to be a signifi- ple nor linear. In their experience, risk of infection became cant risk factor in the univariate analysis; however, its statisti- unlikely after the first 10 days, and they concluded that there cal significance declined to less than 0.05 in the multivariate was no benefit from catheter exchange at the fifth day. It can be analysis (Table 2). We discussed the relation between galeal seen in Figure 1 that the risk of infection on Days 1 and 2 was drainage and a trauma indication for craniotomy above (see similar to the risk on Days 3 and 4 in our series. If the per-day Results). In trying to assess if subgaleal drainage covaried with risk of infection does not increase over time, leaving an unin- other risk factors, we determined that its association with fected drain in place may be the best policy. In one randomized meningitis was not because it was used in 81.9% of trauma controlled trial, a policy of routine change was found not to be cases, but rather 45.2% of oncological patients and 63.4% of of clinical benefit (53). Additional studies are needed to resolve vascular patients. Subgaleal drains were less likely to be used the issue. in higher risk patients in whom the ASA score was more than Whether the ventricular catheter was inserted in the neuro- 3 or in whom the duration of surgery was more than the surgery intensive care unit or in the operating room did not sig- seventy-fifth percentile for the group. The latter two risk factors nificantly alter our incidence of ventriculostomy-related infec- are components of the National Nosocomial Infection tion in accordance with previously reported experiences (29, Surveillance System risk stratification system. The presence 36). Prophylactic antibiotics in ventriculostomy have proven and the duration of a lumbar drain carried a low risk of menin- to be of no benefit (1, 40, 41). gitis, in accordance with previous reports (8).

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 323 KOURBETI ET AL.

The presence of an infection outside the operative site at the REFERENCES time of surgery has been reported to increase the risk of developing meningitis after surgery (22, 31, 38, 43). In our experi- 1. Arabi Y, Memish ZA, Balkhy HH, Francis C, Ferayan A, Al Shimemeri A, Almuneef MA: Ventriculostomy-associated infections: Incidence and risk fac- ence, it was associated with a sixfold increase in the infection tors. Am J Infect Control 33:137–143, 2005. incidence, a magnitude of association reported before (31). The 2. Bayston R, Lambert E: Duration of protective activity of cerebrospinal fluid presence of a concomitant infection did not remain a significant shunt catheters impregnated with antimicrobial agents to prevent bacterial independent risk factor in the multivariate analysis, although a catheter-related infection. J Neurosurg 87:247–251, 1997. trend toward significance was observed (P Ͻ 0.083; Tables 1 3. Blomstedt GC: Infections in neurosurgery: A retrospective study of 1143 patients and 1517 operations. Acta Neurochir (Wien) 78:81–90, 1985. and 2). On that basis, we recommend that, whenever possible, 4. Blomstedt GC: Craniotomy infections. Neurosurg Clin N Am 3:375–385, 1992. such infections be treated and controlled before surgery. 5. Briggs S, Ellis-Pegler R, Raymond N, Thomas M, Wilkinson L: Gram-negative In contrast to previous reports, duration of surgery did not bacillary meningitis after cranial surgery or trauma in adults. Scand J Infect remain an independent risk factor in the multivariate analysis Dis 36:165–173, 2004. (38, 54). Age, sex, diabetes, and corticosteroid administration 6. Brinquin L, Rousseau JM, Boulesteix G, Diraison Y, Bonsignour JP: Continuous infusion of vancomycin in post-neurosurgical staphylococcal before surgery were not significant risk factors for the develop- meningitis in adults [in French]. Presse Med 22:1815–1817, 1993. ment of meningitis, a finding consistent with previous data 7. Carmel PW, Greif LK: The aseptic meningitis syndrome: A complication of (22). Repeat surgery, which has been a major risk factor in posterior fossa surgery. Pediatr Neurosurg 19:276–280, 1993. recent (43) and older reports (38, 54), was not a risk factor in 8. Coplin WM, Avellino AM, Kim DK, Winn HR, Grady MS: Bacterial meningi- our cohort. Previous studies demonstrated postoperative CSF tis associated with lumbar drains: A retrospective cohort study. J Neurol Neurosurg Psychiatry 67:468–473, 1999. leak to be a significant risk factor for the development of 9. Druel B, Vandenesch F, Greenland T, Verneau V, Grando J, Salord F, Christen meningitis (3, 22, 23, 31, 38). In our series, the presence of CSF R, Etienne J: Aseptic meningitis after neurosurgery: A demonstration of bac- leak was associated with a 1.7-fold increase in the risk of terial involvement. Clin Microbiol Infect 1:230–234, 1996. meningitis, but this was not statistically significant. 10. Durand ML, Calderwood SB, Weber DJ, Miller SI, Southwick FS, Caviness VS The limitations of our study include the fact that it was ret- Jr, Swartz MN: Acute bacterial meningitis in adults. A review of 493 episodes. N Engl J Med 328:21–28, 1993. rospective and depends on the accuracy of the data in the clin- 11. Federico G, Tumbarello M, Spanu T, Rosell R, Iacoangeli M, Scerrati M, ical chart. In addition, the power to assess a risk factor depends Tacconelli E: Risk factors and prognostic indicators of bacterial meningitis in a on the frequency of its presence. For example, the presence or cohort of 3580 postneurosurgical patients. Scand J Infect Dis 33:533–537, 2001. absence of diabetes could be assessed for 445 persons in the 12. Forgacs P, Geyer CA, Freidberg SR: Characterization of chemical meningitis study. Thirty-eight were diabetic and four of them had menin- after neurological surgery. Clin Infect Dis 32:179–185, 2001. 13. Govender ST, Nathoo N, van Dellen JR: Evaluation of an antibiotic- gitis, a twofold increase in risk, which was not statistically sig- impregnated shunt system for the treatment of hydrocephalus. J Neurosurg nificant. Given these data, it may be determined that the study 99:831–839, 2003. has a power of 80% to detect a 3.75-fold increase in risk, but 14. Guelfand L, Procopio A, Manganello S, Defain V, Rivero D, DelCastillo M, only a 22% power to detect a twofold increase. Thus, this study Soloaga R: Use of the BacT/Alert System (BAS) for diagnosis of meningitis had adequate power to detect risk factors of moderate strength. postneurosurgery. Presented at the 39th Annual Meeting of the Infectious Disease Society of America, Boston, MA, September 26–29, 1999. Recent attempts to reduce the frequency of drainage- 15. Gulati S, Kapil A, Das B, Dwivedi SN, Mahapatra AK: Nosocomial infec- associated infection have included impregnation of ventricular tions due to Acinetobacter baumannii in a neurosurgery ICU. Neurol India shunt catheters with antimicrobial agents. Such catheters retain 49:134–137, 2001. their activity for at least 6 weeks (2, 13). A controlled trial of 16. Hamilton AJ, Orozco J, Narotam P, Bowersock T: Efficacy of vancomycin/tri- iododecyclemethyl ammonium chloride-coated ventriculostomy catheters in ventricular shunting has shown such treatment to be associated reducing infection. Neurosurgery 40:1043–1049, 1997. with a reduced frequency of infection and in animal models of 17. Harder E, Moller K, Skinhoj P: Enterobacteriaceae meningitis in adults: A closed internal drainage. The approach may prove effective but review of 20 consecutive cases 1977–97. Scand J Infect Dis 31:287–291, 1999. requires evaluation in the postoperative setting of external 18. Holloway KL, Barnes T, Choi S, Bullock R, Marshall LF, Eisenberg HM, Jane drainage in humans because it already looks promising in a JA, Ward JD, Young HF, Marmarou A: Ventriculostomy infections: The effect of monitoring duration and catheter exchange in 584 patients. J Neurosurg rabbit model (16). 85:419–424, 1996. 19. Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG: CDC definitions of CONCLUSION nosocomial surgical site infections, 1992: A modification of CDC definitions of surgical wound infections. Infect Control Hosp Epidemiol 13:606–608, 1992. In our population, the incidence of postcraniotomy meningi- 20. Hsu GJ, Young TG, Chou JW, Peng MY, Chang FY, Chou MY: Gram-negative bacillary meningitis in adults. J Formos Med Assoc 92:317–323, 1993. tis was 5.5%. Entry into the paranasal sinuses, the patient’s 21. Kim YS, Pons VG: Infections in the neurosurgical intensive care unit. ASA score, and the duration of the device-related postoperative Neurosurg Clin N Am 5:741–754, 1994. communication of the CSF and the environment were the major 22. Korinek AM: Prevention of meningitis after craniotomy in scheduled surgery risk factors for the development of the specific infection. The [in French]. Ann Fr Anesth Reanim 11:711–715, 1992. ASA score is a patient characteristic and not modifiable, but the 23. Korinek AM: Risk factors for neurosurgical site infections after craniotomy: conditions that determine the ASA score can be managed to A prospective multicenter study of 2944 patients. The French Study Group of Neurosurgical Infections, the SEHP, and the C-CLIN Paris-Nord. Service improve surgical risk. Other risks may be minimized by care- Epidemiologie Hygiene et Prevention. Neurosurgery 41:1073–1081, 1997. ful surgical technique and aseptic management of postopera- 24. Korinek AM: Infectious risk from ventriculostomy [in French]. Ann Fr tive drains. Anesth Reanim 18:554–557, 1999.

324 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com POSTCRANIOTOMY MENINGITIS

25. Leib SL, Boscacci R, Gratzl O, Zimmerli W: Predictive value of cerebrospinal 49. Stieg PE, Mulligen JB: Neurosurgical complications in craniofacial surgery. fluid (CSF) lactate level versus CSF/blood glucose ratio for the diagnosis of Neurosurg Clin N Am 2:703–708, 1991. bacterial meningitis following neurosurgery. Clin Infect Dis 29:69–74, 1999. 50. Tauber MG: To tap or not to tap? Clin Infect Dis 25:289–291, 1997. 26. Lozier AP, Sciacca RR, Romagnoli MF, Connolly ES Jr: Ventriculostomy- 51. van Ek B, Bakker FP, van Dulken H, Dijkmans BA: Infections after cran- related infections: A critical review of the literature. Neurosurgery iotomy: A retrospective study. J Infect 12:105–109, 1986. 51:170–182, 2002. 52. Winfield JA, Rosenthal P, Kanter RK, Casella G: Duration of intracranial pres- 27. Lu CH, Chang WN, Chuang YC, Chang HW: Gram-negative bacillary menin- sure monitoring does not predict daily risk of infectious complications. gitis in adult post-neurosurgical patients. Surg Neurol 52:438–444, 1999. Neurosurgery 33:424–431, 1993. 28. Mahe V, Kermarrec N, Ecoffey C: Infections related to external ventricular 53. Wong GK, Poon WS, Wai S, Yu LM, Lyon D, Lam JM: Failure of regular exter- drainage [in French]. Ann Fr Anesth Reanim 14:8–12, 1995. nal ventricular drain exchange to reduce cerebrospinal fluid infection: Result 29. Mayhall CG, Archer NH, Lamb VA, Spadora AC, Baggett JW, Ward JD, of a randomised controlled trial. J Neurol Neurosurg Psychiatry 73:759–761, Narayan RK: Ventriculostomy-related infections. A prospective epidemio- 2002. logic study. N Engl J Med 310:553–559, 1984. 54. Wright RL: A survey of possible etiologic agents in postoperative craniotomy 30. Metersky ML, Williams A, Rafanan AL: Retrospective analysis: Are fever infections. J Neurosurg 25:125–132, 1966. and altered mental status indications for lumbar puncture in a hospital- 55. Wroblewska MM, Dijkshoorn L, Marchel H, Van den Barselaar M, Swoboda- ized patient who has not undergone neurosurgery? Clin Infect Dis Kopec E, van den Broek PJ, Luczak M: Outbreak of nosocomial meningitis 25:285–288, 1997. caused by Acinetobacter baumannii in neurosurgical patients. J Hosp Infect 31. Mollman HD, Haines SJ: Risk factors for postoperative neurosurgical wound 57:300–307, 2004. infection. A case-control study. J Neurosurg 64:902–906, 1986. 32. Munch E, Weigel R, Schmiedek P, Schurer L: The Camino intracranial pres- Acknowledgments sure device in clinical practice: Reliability, handling characteristics and complications. Acta Neurochir (Wien) 140:1113–1120, 1998. We thank Mr. P. Karafios for his support. Presented in part at the 42nd Annual 33. Narotam PK, Van Dellen JR, du Trevou MD, Gouws E: Operative sepsis in Meeting of the Infectious Disease Society of America, Boston, MA, September neurosurgery: A method of classifying surgical cases. Neurosurgery 30–October 3, 2004 (abstract #184). 34:409–416, 1994. 34. Nibu K, Sasaki T, Kawahara N, Sugasawa M, Nakatsuka T, Yamada A: Complications of craniofacial surgery for tumors involving the anterior cra- COMMENTS nial base. Neurosurgery 42:455–462, 1998. 35. Nishizawa S, Yokota N, Yokoyama T, Mukodaka H, Watanabe T, Hoshino T, any differences still exist among neurosurgeons’ opinions about Ueda Y: Prevention of postoperative complications in skull base surgery for what is considered to be of fundamental importance to reduce nasal or paranasal sinus carcinoma invading the skull base. J Clin Neurosci M 8 [Suppl 1]:67–70, 2001. the risk of postoperative infections. This is mirrored by different behav- 36. Park P, Garton HJ, Kocan MJ, Thompson BG: Risk of infection with pro- iors in different operating rooms. Despite the intrinsic limits of this ret- longed ventricular catheterization. Neurosurgery 55:594–601, 2004. rospective study, both in the statistical power of conclusions and in the 37. Patel RS, Yousem DM, Maldjian JA, Zager EL: Incidence and clinical signifi- baseline results (e.g., possible underestimation of the infection rate cance of frontal sinus or orbital entry during pterional (frontotemporal) cran- owing to the assumption of readmittance of all patients with postoper- iotomy. AJNR Am J Neuroradiol 21:1327–1330, 2000. ative meningitis in the same hospital), the authors have been able to 38. Patir R, Mahapatra AK, Banerji AK: Risk factors in postoperative neurosurgi- carry out a useful discussion of what seems to be more clearly related cal infection. A prospective study. Acta Neurochir (Wien) 119:80–84, 1992. to the development of postcraniotomy meningitis. This is the main 39. Pfisterer W, Muhlbauer M, Czech T, Reinprecht A: Early diagnosis of external ventricular drainage infection: Results of a prospective study. J Neurol point of interest in this article which, despite a substantial lack of nov- Neurosurg Psychiatry 74:929–932, 2003. elty, confirms that the following are the main risk factors for the devel- 40. Poon WS, Ng S, Wai S: CSF antibiotic prophylaxis for neurosurgical patients opment of postcraniotomy meningitis: 1) concomitant infections, 2) with ventriculostomy: A randomised study. Acta Neurochir Suppl poor general conditions (American Society of Anesthesiologists score > 71:146–148, 1998. 3), 3) duration of surgery, and 4) the continuing device-related postop- 41. Prabhu VC, Kaufman HH, Voelker JL, Aronoff SC, Niewiadomska-Bugaj M, erative communication of the cerebrospinal fluid (CSF) and the envi- Mascaro S, Hobbs GR: Prophylactic antibiotics with intracranial pressure ronment. Because many studies agree on these points, the neurosurgi- monitors and external ventricular drains: A review of the evidence. Surg cal community should concentrate on these well-known risk factors to Neurol 52:226–237, 1999. 42. Rebuck JA, Murry KR, Rhoney DH, Michael DB, Coplin WM: Infection reduce the infection rate. In our department, a prospective study on related to intracranial pressure monitors in adults: Analysis of risk factors and surgical site infections after neurosurgical procedures is under develop- antibiotic prophylaxis. J Neurol Neurosurg Psychiatry 69:381–384, 2000. ment. From a previous survey of 1747 consecutive patients in 1999 and 43. Reichert MC, Medeiros EA, Ferraz FA: Hospital-acquired meningitis in 2000, the crude rate of surgical site infections observed in 726 cran- patients undergoing craniotomy: Incidence, evolution, and risk factors. Am iotomies was 1.52% (Valentini et al., unpublished data). This finding J Infect Control 30:158–164, 2002. probably depends on several factors, including, most importantly, the 44. Rosner MJ, Becker DP: ICP monitoring: Complications and associated factors. electiveness of surgery and the specific prophylaxis for every risk class. Clin Neurosurg 23:494–519, 1976. Moreover, subgaleal and CSF drainages are avoided when possible 45. Ross D, Rosegay H, Pons V: Differentiation of aseptic and bacterial meningi- and are interiorized as soon as possible. Antibiotic prophylaxis is tis in postoperative neurosurgical patients. J Neurosurg 69:669–674, 1988. always used during surgery and is maintained when the use of 46. Salord F, Druel B, Grando J, Verneau V, Perret C, Vandenesch F, Etienne J, Chacornac R: Aseptic meningitis. Demonstration of bacterial DNA in cere- drainage cannot be avoided. Studies further clarifying these issues are brospinal fluid by gene amplification [in French]. Ann Fr Anesth Reanim required to better elucidate the role of each individual risk factor and 14:320–325, 1995. further reduce the infection rate in our specialty, in which a postoper- 47. Schade RP, Schinkel J, Visser LG, Van Dijk JM, Voormolen JH, Kuijper EJ: ative infection can still represent a fatal complication. Bacterial meningitis caused by the use of ventricular or lumbar cerebrospinal fluid catheters. J Neurosurg 102:229–234, 2005. Paolo Ferroli 48. Singh RV, Yeh JS: Wound infection with meningitis caused by Salmonella Giovanni Broggi typhimurium. Br J Neurosurg 7:311–313, 1993. Milan, Italy

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 325 KOURBETI ET AL.

ourbeti et al. add another retrospective study examining the inci- vided no reduction in CSF infections compared with not changing the Kdence and risk factors associated with the development of menin- catheter unless clinically indicated (7). However, this study may not gitis in postcraniotomy patients to the literature. In their large cohort of have been appropriately powered to detect a difference. In a related 453 patients, they found a 5.5% incidence of meningitis, which com- study of central venous and pulmonary artery catheters, routine pares favorably with other reports. Multivariate statistical analysis indi- replacement of the catheters did not prevent infection (1). More cated that the risk of postcraniotomy meningitis was increased by recently, the use of antibiotic-impregnated catheters has been recom- transgression of an air sinus during surgery, a poor American Society mended based on randomized trials (8). In our practice, in which exter- of Anesthesiology score, and longer duration of external ventricular nal catheters are placed primarily for treating and managing hydro- drainage or intracranial pressure monitoring. cephalus after tumor resection, we try to make decisions on the need Although the statistical methods used by the investigators are sound for permanent shunting within 5 to 7 days of ventriculostomy place- and the reported risk factors are logical, the retrospective nature of the ment based on the concerns described by Kourbeti et al. In addition, study raises the usual concerns, including the extent to which the catheters are tunneled 5 to 7 cm from the insertion sight, manipulations results are universally applicable to patients undergoing craniotomy. of the sterile circuit are minimal, and prophylactic antibiotics are used. The authors report that their patients underwent a wide variety of sur- We do not exchange the catheter unless clinically indicated, although gical procedures, including elective craniotomies for tumor and vascu- we monitor the CSF for slight increases in white cell counts. lar problems and urgent craniotomy for trauma, suggesting that the Clearly, although restrospective studies are important, there is a results may be broadly applicable. Nevertheless, retrospective studies need in the neurosurgical community to increase its effort to formulate that seek to determine risk factors in defined cohorts of patients should and complete randomized prospective clinical trials, which remain the also ideally include a second, independent cohort to permit validation “gold standard” for resolving most clinical problems. (or not) of the identified factors. This approach has been used in other Frederick F. Lang retrospective studies of neurosurgical disease processes (4, 5). Although Houston, Texas the authors are to be congratulated for their careful analysis, validation of the results in an independent cohort of patients would greatly increase the strength of this study. Nevertheless, probably the most interesting result of the study is the 1. Cobb DK, High KP, Sawyer RG, Sable CA, Adams RB, Lindley DA, Pruett TL, increased risk of postoperative meningitis associated with prolonged Schwenzer KJ, Farr BM: A controlled trial of scheduled replacement of central venous and pulmonary-artery catheters. N Engl J Med 327:1062–1068, 1992. use of external ventricular drainage or intracranial pressure monitoring. 2. Holloway KL, Barnes T, Choi S, Bullock R, Marshall LF, Eisenberg HM, Jane Quite surprisingly, even the prolonged use of subgaleal drains was asso- JA, Ward JD, Young HF, Marmarou A: Ventriculostomy infections: The effect ciated with increased risk of meningitis, although this association was of monitoring duration and catheter exchange in 584 patients. J Neurosurg not supported in the multivariate analysis. Of course, many studies 85:419–424, 1996. focusing specifically on ventriculostomy-related infections have sug- 3. Lozier AP, Sciacca RR, Romagnoli MF, Connolly ES Jr: Ventriculostomy-related gested that the longer the catheters are in place, the greater the risk of infections: A critical review of the literature. Neurosurgery 51:170–182, 2002. meningitis (3). However, controversy remains because such an associa- 4. Phillips HS, Kharbanda S, Chen R, Forrest WF, Soriano RH, Wu TD, Misra A, tion was not found in other studies (3). It has been suggested that the Nigro JM, Colman H, Soroceanu L, Williams PM, Modrusan Z, Feuerstein risk may increase early on and then decrease or plateau later (2). In this BG, Aldape K: Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogene- new study, Kourbeti et al. add to the evidence that longer dwelling sis. Cancer Cell 9:157–173, 2006. times are associated with increased risk of infection. 5. Pignatti F, van den Bent M, Curran D, Debruyne C, Sylvester R, Therasse P, Of course, identification of a risk does not resolve how to manage or Afra D, Cornu P, Bolla M, Vecht C, Karim AB, European Organization for reduce that risk. It is dangerous to make recommendations concerning Research and Treatment of Cancer Brain Tumor Cooperative Group; European best practices based on retrospective studies. Randomized prospective Organization for Research and Treatment of Cancer Radiotherapy Cooperative trials are needed to help resolve issues related to interventions to Group: Prognostic factors for survival in adult patients with cerebral low- reduce the risk of infection during ventriculostomy drainage. Should grade glioma. J Clin Oncol 20:2076–2084, 2002. prophylactic antibiotic be used while ventriculostomy catheters are in 6. Poon WS, Ng S, Wai S: CSF antibiotic prophylaxis for neurosurgical patients with place? At least one randomized controlled trial compared perioperative ventriculostomy: A randomised study. Acta Neurochir Suppl 71:146–148, 1998. 7. Wong GK, Poon WS, Wai S, Yu LM, Lyon D, Lam JM: Failure of regular exter- antibiotics alone with prolonged antibiotics and found that prolonged nal ventricular drain exchange to reduce cerebrospinal fluid infection: Result of antibiotic prophylaxis reduced the incidence of CSF infection, although a randomised controlled trial. J Neurol Neurosurg Psychiatry 73:759–761, 2002. it also selected for resistant or opportunistic pathogens (6). Should 8. Zabramski JM, Whiting D, Darouiche RO, Horner TG, Olson J, Robertson C, catheters be changed regularly or left in place until there is a clinical Hamilton AJ: Efficacy of antimicrobial-impregnated external ventricular drain indication to change them? One prospective randomized study showed catheters: A prospective, randomized, controlled trial. J Neurosurg 98:725–730, that regular external ventricular catheter exchanges every 5 days pro- 2003.

326 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CLINICAL STUDIES

LONG-TERM OUTCOME IN PATIENTS WITH SUSPECTED NORMAL PRESSURE HYDROCEPHALUS

Babar Kahlon, M.D., Ph.D. OBJECTIVE: To evaluate the outcome of patients with suspected normal pressure hydro- Department of Neurosurgery, cephalus at 6 months and 5 years after shunt surgery. University Hospital, Ϯ Lund, Sweden METHODS: Seventy-ﬁve patients (mean age, 72.5 9 yr), with normal pressure hydrocephalus symptoms were included. Fifty-four patients with positive lumbar infusion Johan Sjunnesson, M.D. and/or cerebrospinal fluid tap tests received a cerebrospinal fluid shunt, whereas Department of Neurosurgery, 21 patients with negative test results did not undergo operation. Walk, reaction time, University Hospital, memory, and identical forms tests were used as baseline (before surgery) tests and were Lund, Sweden repeated at short- (6.1 Ϯ 4.6 mo) and long-term (5.5 Ϯ 1.4 yr) follow-up evaluations. Activities of daily life functions were assessed using the Barthel index. Stig Rehncrona, M.D., Ph.D. Department of Neurosurgery, RESULTS: At the 6-month follow-up examination, 83% of the operated patients improved University Hospital, in gait, 65% improved in reaction time, 46% improved in memory, and 31% improved Lund, Sweden in identical forms tests; 96% found themselves subjectively improved. Because of unrelated mortality (37%) and declining general health from comorbidity, only 27 patients Reprint requests: Babar Kahlon, M.D., Ph.D., were available for the 5-year follow-up evaluation. Twenty-three of these patients had Department of Neurosurgery, been treated with a shunt and had a remaining improvement in close to 40% in gait University Hospital, and reaction time, whereas fewer than 10% had an improvement in cognitive tests. S-221 85 Lund, Sweden. Email: [email protected] Fifty-six percent reported subjective improvement compared with preoperative findings. More patients (64%) improved if younger than 75 years; for patients older than Received, January 18, 2006. 75 years, only 11% of the patients improved. The Barthel index was higher (P Ͻ 0.05) Accepted, September 18, 2006. in improved patients. CONCLUSION: Patients with normal pressure hydrocephalus beneﬁt from shunt surgery for at least 5 years. High mortality rate, comorbidity, and old age hamper good long-term outcome and emphasize the importance of patient selection. KEY WORDS: Comorbidity, Idiopathic normal pressure hydrocephalus, Long-term outcome, Normal pressure hydrocephalus, Outcome, Shunt surgery

Neurosurgery 60:327–332, 2007 DOI: 10.1227/01.NEU.0000249273.41569.6E www.neurosurgery-online.com

ormal pressure hydrocephalus (NPH) few long-term results are reported (11, 13, 14). is characterized by the clinical presen- In our center, we have a long experience of Ntation of gait disturbance, memory treating patients with suspected NPH. In the deficit, and urinary incontinence, combined present prospective study, the patients with with dilated cerebral ventricles and normal suspected NPH were consecutively included, cerebrospinal fluid (CSF) pressure. The syn- carefully selected for surgery, and followed for drome was first described by Hakim and short (mean, 6 mo) and long periods (mean, Adams (4) in 1965. After four decades of inves- 5 yr) with respect to outcome. tigation, the understanding of the pathophysiological mechanism is poor, but the effectiveness of diversion of CSF flow by shunt PATIENTS AND METHODS operation in treating this syndrome is well documented. Lately, many authors have Patients reported relatively high improvement results, From 1996 to 2001, 89 patients were enrolled mostly with follow-up periods ranging from 3 consecutively in the study. Eighty-one of these months up to 2 years (1, 3, 9, 12). However, patients were previously described in our ear-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 327 KAHLON ET AL.

lier study concerning short-term follow-up (6). Fourteen a pressure recording device (DPT-6100; Smiths Medical, patients were excluded because they failed to attend follow- Kirchseeon, Germany). The infusion was switched off when a up visits (three patients), refused follow-up (three patients), steady plateau pressure level was achieved (mean infusion refused surgery (four patients), had incomplete data (two time was approximately 1 h). When the pressure returned to patients), had supervening disease (one patient), or moved to normal, 50 ml of CSF was removed through one of the needles. a distant location (one patient). Hence, 75 patients with a The patients were then allowed a resting time of approximately mean age of 72.5 Ϯ 9 years (32 men, 43 women) were eligible 3 hours. Walk, reaction, memory, and identical forms tests were for long-term follow-up evaluations; all data pertain to them. then repeated and compared with the baseline. Improvements Forty-five (60%) out of 75 patients had cardiovascular dis- of 5% (walk and reaction time) and 25% (memory and identi- ease, eight (11%) out of 75 patients had a history of cere- cal forms) compared with the baseline measurements were con- brovascular thrombosis, and eight (11%) out of 75 patients sidered significant improvements (6). Improvements in at least had a history of intracerebral hemorrhage. Forty-nine (65%) two of these four different tests were required for a positive out of 75 patients had a previous history of cardiovascular or CSF-TT (6, 16, 17). cerebral vascular disease. A plateau pressure (Ppl) of at least 22 mmHg was considered A neurologist, a specialist in internal medicine, or a special- a positive LIT (5, 7). Resistance to outflow (Rout) was also calcu- ist in psychogeriatrics primarily examined all patients. The lated, but this value did not influence the selection for surgery. patients were then referred to the neurosurgery department All patients with either positive CSF-TT or positive LIT were because of clinical symptoms arousing suspicion of NPH selected for surgery and operated with a ventriculoperitoneal combined with widening of the cerebral ventricles on com- or a ventriculoatrial adjustable shunt system (Codman-Hakim puted tomographic or magnetic resonance tomographic scans Model 82–3100; Johnson & Johnson Co., Raynham, MA). The Ͼ (Evans ratio, 0.30), as examined by a neuroradiologist. shunt opening pressure was usually set at 120 to 130 mm H2O Ninety-two percent had gait disturbance, 75% had memory during surgery and was later adjusted individually as needed. disturbance, 64% had urinary incontinence, and 47% had all The patients were followed as outpatients and shunt function three symptoms of the typical NPH triad. Eight patients had was examined by palpation; if dysfunction was suspected, a history of intracerebral hemorrhage, five had a history of computed tomographic scan, plane x-ray, and LIT were per- severe head trauma, two were operated on previously formed. because of benign central nervous system tumors, and two had a history of central nervous system infection. Hence, 58 Follow-up (77%) patients were considered to have idiopathic NPH and All patients (including those who did not undergo operation) 17 (23%) patients could be considered to have secondary were followed for short (mean follow-up time Ϯ standard devi- NPH. The assumed provoking factors occurred more than ation, 6.1 Ϯ 4.6 mo) and long terms (mean, 5.5 Ϯ 1.35 yr; range, 5 years before enrollment, and the relationships to symptom 2.5–7.6 yr). For outcome evaluation, the patients were tested development were unclear. The mean duration of symptoms with the same test battery as at baseline. Improvements of 5% was 1.7 Ϯ 1.9 years (range, 0.12–12.0 yr). All patients and (walk and reaction time) and 25% (memory and identical forms) their relatives were informed and gave their consent; the local compared with the best of the two baseline measurements was ethics committee approved the study. considered significant (6). The patients and their relatives were asked about their subjective impressions regarding overall Baseline Evaluation improvement or no improvement. At the long-term follow-up All patients were tested with walk (16), reaction time (16), examination, the Barthel index was used to assess the activities memory (2), and identical forms (15) tests. This test battery of daily life (10). was repeated on the following day at the same time of the day, and the better of the two results was used as the baseline for Statistics the CSF tap test (TT) and for the follow-up evaluations (6). Continuous variables are expressed as mean Ϯ standard deviation. For comparison between groups, the nonparametric Selection for Surgery Mann-Whitney test was used for calculating the two-tailed All patients were tested with a constant-rate lumbar infusion P value. The statistical data was calculated using GraphPad test (LIT) (5, 7) and CSF-TT (6, 16, 17). The procedure has been Instat Version 3.1 (GraphPad Software, San Diego, CA). previously described in some detail (6) and will only be summarized here. RESULTS With the patient in the lateral recumbent position, two needles (diameter, 0.9 mm) were inserted in the lower lumbar Fifty-four patients had either a positive LIT and/or a positive region (L3–L4 or L4–L5). Through one of the needles, a Ringer CSF-TT (mean Ppl, 28.6 Ϯ 7.4 mmHg; Rout, 20.7 Ϯ 8.1 solution (6.6 g/L NaCl, 0.3g/L KCl, and 0.33g/L CaCl; mmHg/ml/min); all were operated with a ventriculoperitoneal 290 mosm/kg) was infused at a constant rate (0.8 mL/min) by (53 patients) or ventriculoatrial (1 patient) CSF shunt. Forty-six an infusion pump (TOP syringe pump 5100; TOP Corp., Tokyo, (85%) out of the 54 patients had idiopathic NPH and eight Japan) and pressure was recorded through the other needle via (15%) had secondary NPH.

328 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com LONG-TERM OUTCOME IN NORMAL PRESSURE HYDROCEPHALUS

Twenty-one patients had normal LIT and negative TABLE 1. Short-term (6.1 ؎ 4.6 mo) outcome in shunt-operated and CSF-TT and were not oper- nonoperated patientsa ated. In this group, 57% Mean Not Improved Baseline were considered to have % imp improved (54 ؍ idiopathic NPH and 43% Operated (n had possible provoking fac- Ws 41 (76%) 52 Ϯ 27b 28 Ϯ 16 13 tors. Wt 45 (83%) 30 Ϯ 22c 31 Ϯ 16 9 R 35 (65%) 500 Ϯ 355d 29 Ϯ 17 19 Complications, Mortality, M 25 (46%) 3.0 Ϯ 1.6e 89 Ϯ 104 29 Ϯ f Ϯ and Dropouts Id 17 (31%) 15 9 117 179 37 Two tests 42 (78%) 12 The immediate postsurgi- Subj 52 (96%) 2 (21 ؍ cal period was complicated Not operated (n Ϯ Ϯ by subdural hematomas or Ws 2 (10%) 73 44 14 619 Wt 5 (24%) 30 Ϯ 24 11 Ϯ 716 effusions in three patients. R 10 (48%) 1482 Ϯ 2229 27 Ϯ 28 11 One required surgical evacu- M 6 (28%) 2.8 Ϯ 1.9 76 Ϯ 52 15 ation and two resolved after Id 3 (14%) 26 Ϯ 2 46 Ϯ 19 18 increasing the shunt opening Two tests 5 (24%) 16 Subj 4 (19%) 17 pressure to 200 mmHg for a limited period. The shunt a Improved, number and percentage of patients who were improved after shunt operation (Ͼ5% in Ws, Wt, and R tests and Ͼ25% in M and Id tests); baseline, was removed in one patient mean Ϯ standard deviation of preoperative baseline results in patients who because of shunt infection improved after shunt surgery; mean % imp, mean Ϯ standard deviation of and was reimplanted later percentage improvement in the improved patients compared with baseline; not improved, patients who did not improve at follow-up or were worse; Ws, number after the infection healed. of steps taken for walking a distance of 18 meters; Wt, time in seconds for walking One patient with pulmonary a distance of 18 meters; R, reaction time in milliseconds; M, number of points in FIGURE 1. Graph showing the per- the memory test; Id, number of points in the identical forms test; two tests, embolism was successfully improved in at least two different tests (Ws and Wt was considered as one test); treated with anticoagulants centage of patients who were lost to Subj, subjective impression of overall improvement or no improvement at follow- and one patient with a super- follow-up, who refused follow-up, up. who were not able to perform the b N ϭ 39; two patients were able to walk only postoperatively. ficial wound infection healed c ϭ tests, who were deceased, and who N 43; two patients were able to walk only postoperatively with antibiotics. All surgical d N ϭ 32; three patients could not perform the test. were available for follow-up at 5 years e ϭ and shunt-related complica- N 24; one patient could not perform the test. (mean, 5.5 Ϯ 1.35 yr) from a total of f N ϭ 16; one patient could not perform the test. tions resolved before the first 75 patients. follow-up evaluation. Between the 6-month and 5-year follow-up periods, 28 (37%) 54 patients reported that they subjectively improved. Mean out of the initial 75 patients died. The death rates in operated improvements after the shunt operation were 28% in the walk (37%) and not operated (38%) patients were similar. The causes step test, 31% in the walk time test, 29% in the reaction time of death according to death certificates were: cardiovascular test, 89% in the memory test, and 117% in the identical forms disease (nine patients), malignancy (seven patients), pneumo- test (Table 1). nia (three patients), septicemia (two patients), dementia (two The mean age in the operated group was 74 Ϯ 7.1 years. The patients), pancreatitis (one patient); in four patients, the cause improvement rates in the walk (steps) test in patients younger was unknown. All 75 patients were available for the first than 75 years and those older than 75 years of age were simi- follow-up evaluation at 6 months. At the long-term evaluation, lar, with 74 and 78% of the patients improved, respectively 11 patients (15%) refused clinical follow-up due to poor general (Table 2). health; six (8%) patients were not able to perform the tests In the group that did not undergo operation, two (10%) out because of poor general health caused by other supervening of 21 improved in the walk (steps) test and five (24%) out of 21 diseases or by severe dementia and three patients (4%) were improved in walk (time), 10 (48%) out of 21 improved in reac- lost to follow-up (Fig. 1). tion time, six (28%) out of 21 improved in memory, and three (14%) out of 21 improved in identical forms test; five (24%) out Six-month Follow-up of 21 improved in two tests. Four (19%) out of 21 patients At the 6-month follow-up (mean, 6.1 Ϯ 4.6 mo) examina- reported that they subjectively improved (Table 1). tion, 76% of the 54 operated patients improved in the walk (steps) test, 83% improved in the walk (time) test, 65% Five-year Follow-up improved in reaction time, 46% improved in memory, and Twenty-seven patients (36%) were available for the long- 31% improved in identical forms test (Table 1). Forty-two term follow-up tests. Twenty-three (21 idiopathic and two sec- (78%) out of the 54 patients in the operated group improved ondary NPH) of these 27 patients belonged to the operated in at least two out of the four tests. Fifty-two (96%) of the group and four (two idiopathic and two secondary) did not

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 329 KAHLON ET AL.

TABLE 2. Gait improvement after shunt surgery in suspected NPH patients in relation to agea Outcome at 6-mo follow-up Outcome at 5-yr follow-up Age (yr) No. of patients Imp Ws Mean % imp No. of patients Imp Ws Mean % imp Ͻ65 4 2 (50%) 24 Ϯ 11 3 2 (67%) 12 Ϯ 6 65–69 5 3 (60%) 31 Ϯ 14 3 3 (100%) 30 Ϯ 21 70–74 18 15 (83%) 27 Ϯ 11 8 4 (50%) 23 Ϯ 18 75–79 21 17 (81%) 23 Ϯ 17 9 1 (11%) 9 Ϯ 0 Ͼ80 6 4 (67%) 36 Ϯ 28 0 0 0

a Imp Ws, improvement in number and percentage of patients in walk (steps) test; mean % imp, mean Ϯ standard deviation of percentage improvement in the improved patients compared with baseline.

TABLE 3. Outcome in 23 patients who were available for both short- and long-term follow-upa (23 ؍ Outcome at 5 yr (n (23 ؍ Outcome at 6 mo (n Improved Baseline Mean % imp Improved Baseline Mean % imp B index Not improved B index Ws 21 (91%) 45 Ϯ 22 24 Ϯ 14 10 (43%) 42 Ϯ 8 21 Ϯ 16 94 Ϯ 6b 13 63 Ϯ 35 Wt 22 (96%) 29 Ϯ 14 29 Ϯ 14 9 (39%) 26 Ϯ 9 27 Ϯ 21 94 Ϯ 6b 14 65 Ϯ 34 R 15 (65%) 437 Ϯ 224 31 Ϯ 18 9 (39%) 466 Ϯ 229 27 Ϯ 20 91 Ϯ 6 14 69 Ϯ 35 M 12 (52%) 3,4 Ϯ 1,9 83 Ϯ 103 1 (4%) 3,5 Ϯ 0 57 Ϯ 0 95 22 76 Ϯ 30 Id 5 (22%) 14 Ϯ 11 196 Ϯ 313 2 (9%) 71,7 Ϯ 93,7 104 Ϯ 77 98 Ϯ 4 21 75 Ϯ 31 Two tests 19 (83%) 6 (26%) 95 Ϯ 3 17 70 Ϯ 33 Subj 22 (96%) 13 (56%) 94 Ϯ 6c 8 57 Ϯ 36

a Improved, number and percentage of patients who improved after shunt operation (Ͼ5% in Ws, Wt, and R tests and Ͼ25% in M and Id tests); baseline, mean Ϯ standard deviation of preoperative baseline results in patients who improved after shunt surgery; mean % imp, mean Ϯ standard deviation of percentage improvement in the improved patients as compared with baseline; B index, Barthel index; not improved, patients who did not improve at follow-up or were worse; Ws, number of steps taken for walking a distance of 18 meters; Wt, time in seconds for walking a distance of 18 meters; R, reaction time in milliseconds; M, number of points in the memory test; Id, number of points in the identical forms test; two tests, improved in at least two different tests (Ws and Wt was considered as one test); Subj, subjective impression of overall improvement or no improvement at follow-up. b P Ͻ 0.05. c P Ͻ 0.01 compared with patients who did not improve. undergo operation. The mean age (at the start of the study), There was no signiﬁcant difference (P ϭ 0.832) in symptom 71.0 Ϯ 8.5 years (range, 44–80 yr), did not differ significantly duration between those who continued to show improvement (P ϭ 0.210) from the 48 patients (mean, 73.4 Ϯ 8.8 yr) who in gait (mean, 1.6 Ϯ 1.0 yr) and those who deteriorated (mean, were not available for long-term follow-up. Nine were men 1.5 Ϯ 0.9 yr). At the start of the study, the incidence of cardio- and 18 were women. Of the 23 shunt-operated patients, 10 vascular and cerebrovascular disease in the patients who con- (43%) still improved in the walk step test, nine (39%) tinued to show improvement in gait and those who did not improved in the walk time test, and nine (39%) improved in was similar (80 and 84%, respectively). Of the 23 patients, 22 the reaction time test, whereas only two showed improve- had positive LIT and nine (41%) out of 22 improved in gait, ment in the identical forms test and one in the memory test. whereas, among the six patients who had positive TT, four Six patients (26%) improved in two or more of the four tests. patients (66%) continued to show improvement in gait. Five Thirteen (57%) reported that they still perceived improvement patients were positive in both LIT and TT, three (60%) of whom compared with their preoperative status (Table 3). There was continued to show improvement in gait. no significant difference (P ϭ 0.446) in follow-up time Five of the 13 patients who deteriorated in the walk test between the patients who remained improved in gait (walk between the two follow-up evaluations reported that a con- steps test) (mean, 5.0 Ϯ 1.2 yr) and those who did not (mean, comitant disease emerging after the 6 month follow-up exam- 4.6 Ϯ 1.3 yr), but the former were significantly (P ϭ 0.01) ination had decisively reduced their gait function: stroke (two younger (mean, 66.7 Ϯ 10.1 yr) than the latter (mean, 74.7 Ϯ patients), hip fracture (one patient), lower back pain (one 4.9 yr). In fact, nine (64%) out of 14 of the patients who were patient), and Parkinson’s disease (one patient). Two out of 10 younger than 75 years continued to show improvement in patients who remained improved at the time of the long-term the walk (steps) test, whereas only 11% of patients who were follow-up examination reported lower back pain (one with sci- older than 75 years of age continued to show improvement in atic nerve root compression), partially restricting them in their the same test (Table 2). walking function.

330 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com LONG-TERM OUTCOME IN NORMAL PRESSURE HYDROCEPHALUS

The Barthel index was calculated in 22 out of 23 patients the patients who continued to show improvement in gait at the (incomplete data in one patient), in cooperation with the rel- time of the long-term follow-up examination were independent atives. It showed that the patients who continued to show in their activities of daily life, as indicated by a high Barthel improvement in the walk (steps) and walk (time) tests and index. those who reported subjective improvement had a statisti- The high percentage of patients with subjectively perceived cally significant (P ϭ 0.032; P ϭ 0.025; and P ϭ 0.010, respec- improvements (96%) at the short-term follow-up examination tively) higher Barthel index than patients who did not declined to 57% at the long-term follow-up examination. improve (Table 3). However, the statistically significant higher Barthel index in One of the four patients remaining in the nonoperated group these patients compared with patients without subjective continued to improve in walk and reaction time tests (Barthel improvements (Table 3), indicates that their reported improve- index, 95), but three were worse compared with their baseline ment had significance for them in coping with activities of daily (Barthel index, 65 Ϯ 44). None of the four reported any subjec- living. tive improvement. If high mortality and adverse effects of interfering initial and acquired comorbidity are taken into account, the long-term DISCUSSION results of shunt surgery in NPH patients should not be considered unsatisfactory. In fact, long-term improvements in gait The results of this long-term follow-up study of patients with were witnessed in more than 60% of the patients younger than clinical symptoms of NPH may seem disappointing at first. 75 years and subjective improvements were observed in nearly Despite a rather favorable short-term outcome after shunt sur- 60% of all patients. Apart from showing a better long-term gery with improvements in 30 and 80% of the patients in mem- result of shunting at younger age, our results were unable to ory and gait functions, respectively, only 20% of the operated define other clinically predictive indices. patients demonstrated continued functional improvements after 5 years. The reasons were a high mortality rate and an CONCLUSION inability to participate in the long-term follow-up tests because of worsening of preexistent or supervening comorbidity. We We conclude that patients with clinical symptoms of NPH will first discuss the reasons for the high dropout rate. benefit from a CSF shunt procedure for at least up to 5 years. No deaths were reported before the first follow-up examina- Good long-term results are mainly hampered by comorbidity, tion. However, in the 6-month to 5-year postoperative period, especially in vascular and malignant diseases, which are 37% of the patients died. This corresponds to an annual death responsible for high death rates in this patient category. Long- rate of 7.4%, which is substantially higher than in the general term results of shunt surgery for NPH are better in younger Swedish population within the similar age range and during patients, with more than 60% of the patients continuing to the same time period (annual mortality rate 3.2%; Population show improvement in gait if they were younger than 75 years. Statistics Registry, Sweden). The death rate was similar in the The results emphasize the importance of taking complicating shunt-operated and nonoperated groups, indicating that the diseases and older age into account when selecting patients deaths were unrelated to surgery or to the shunt treatment. with NPH symptoms for shunt surgery. The high death rate is, however, in agreement with the reports by Malm et al. (11) and Raftopoulos et al. (13), who found the REFERENCES annual risks of death in NPH patients to be approximately 9 and 13%, respectively. Most of the deaths in the present study 1. Bech-Azeddine R, Gjerris F, Waldemar G, Czosnyka M, Juhler M: were related to vascular disorders; in fact, 65% of our patients Intraventricular or lumbar infusion test in adult communicating hydro- already had cardiovascular and/or cerebrovascular disease cephalus? Practical consequences and clinical outcome of shunt operation. before inclusion in this study. The second common cause of Acta Neurochir (Wien) 147:1027–1036, 2005. 2. Bingley T: Mental symptoms in temporal lobe epilepsy and temporal lobe death was malignancy, which was undiagnosed before enroll- gliomas with special reference to laterality of lesion and the relationship ment. Furthermore, 23% of our patients were unable to partic- between handedness and brainedness: A study of 90 cases of temporal lobe ipate in the long-term follow-up testing because of poor general epilepsy and 253 cases of temporal lobe glioma. Acta Psychiatr Neurol Scand health and other diseases. Taken together, our results indicate Suppl 120:1–151, 1958. that comorbidity has a decisive importance for the long-term 3. Boon AJ, Tans JT, Delwel EJ, Egeler-Peerdeman SM, Hanlo PW, Wurzer HA, Avezaat CJ, de Jong DA, Gooskens RH, Hermans J: Dutch Normal-Pressure outcome of shunt surgery in NPH patients, as reported by Hydrocephalus Study: Randomized comparison of low- and medium- other authors (8). pressure shunts. J Neurosurg 88:490–495, 1998. Of the 23 shunt-operated patients who were still alive and 4. Hakim S, Adams RD: The special clinical problem of symptomatic hydro- could perform the long-term follow-up tests, 40% retained cephalus with normal cerebrospinal fluid pressure: Observations on cere- improvement in gait functioning (and reaction time), whereas brospinal fluid hydrodynamics. J Neurol Sci 2:307–327, 1965. only 9% continued to show improvement in functions related 5. Hussey F, Schanzer B, Katzman R: A simple constant-infusion manometric test for measurement of CSF absorption: II. Clinical studies. Neurology to cognition. These results indicate a more stable effect of shunt 20:665–680, 1970. surgery on motor-related than cognitive-related functioning. 6. Kahlon B, Sundbarg G, Rehncrona S: Comparison between the lumbar infu- On the other hand, despite reduction in cognitive modalities, sion and CSF tap tests to predict outcome after shunt surgery in suspected

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 331 KAHLON ET AL.

normal pressure hydrocephalus. J Neurol Neurosurg Psychiatry 73:721– apparently unrelated to treatment causing a high attrition rate. If a 726, 2002. patient experienced an improved quality of life after the shunt operation 7. Katzman R, Hussey F: A simple constant-infusion manometric test for until the point at which they succumbed to a cardiac, cerebrovascular, measurement of CSF absorption: 1. Rationale and method. Neurology or malignancy event, this could be considered a favorable outcome. 20:534–544, 1970. The results from the subset of 23 surgically treated patients available 8. Klinge P, Marmarou A, Bergsneider M, Relkin N, Black PM: Outcome of for a 5-year follow-up examination should be interpreted with cau- shunting in idiopathic normal-pressure hydrocephalus and the value of tion. These patients did not represent the larger group of surviving outcome assessment in shunted patients. Neurosurgery 57 [Suppl 3]: S40–S52, 2005. shunted patients considering that the reasons for lack of follow-up 9. Larsson A, Wikkelso C, Bilting M, Stephensen H: Clinical parameters in 74 included poor general health and severe dementia, both of which likely consecutive patients shunt operated for normal pressure hydrocephalus. Acta biased the results. Nevertheless, it is important to note that prolonged Neurol Scand 84:475–482, 1991. improvement did occur and that, among this group, it was seen prima- 10. Mahoney FI, Barthel DW: Functional evaluation: The Barthel Index. Md State rily in patients younger than 75 years. Med J 14:61–65, 1965. For a patient with NPH, the goal of treatment is typically to improve 11. Malm J, Kristensen B, Stegmayr B, Fagerlund M, Koskinen LO: Three-year quality of life. Providing even 1 or 2 years of improved function can be survival and functional outcome of patients with idiopathic adult hydro- invaluable for individual patients and could be the difference between cephalus syndrome. Neurology 55:576–578, 2000. semi-independent living and a nursing home. The results of this study 12. McGirt MJ, Woodworth G, Coon AL, Thomas G, Williams MA, Rigamonti D: suggest that the intermediate-term benefits are real and meaningful in Diagnosis, treatment, and analysis of long-term outcomes in idiopathic a group of well-selected NPH patients. However, proper preoperative normal-pressure hydrocephalus. Neurosurgery 57:699–705, 2005. counseling should be provided with regards to reasonable expecta- 13. Raftopoulos C, Massager N, Baleriaux D, Deleval J, Clarysse S, Brotchi J: tions of prolonged improvement. Prospective analysis by computed tomography and long-term outcome of 23 adult patients with chronic idiopathic hydrocephalus. Neurosurgery It should be noted that the shunt-related complication rate was low, 38:51–59, 1996. a finding likely attributable to a judicious management scheme using 14. Savolainen S, Hurskainen H, Paljarvi L, Alafuzoff I, Vapalahti M: Five-year an adjustable valve. I think the results reported here suggest that a outcome of normal pressure hydrocephalus with or without a shunt: favorable risk-benefit ratio exists in the treatment of NPH and that Predictive value of the clinical signs, neuropsychological evaluation and infu- Vanneste et al.’s (1) assertion that the risk of shunting outweighed the sion test. Acta Neurochir (Wien) 144:515–523, 2002. benefit pertaining to now outdated treatment and management proto- 15. Thurstone LL: Primary Mental Abilities. Chicago, The University of Chicago cols is, therefore, no longer relevant. Press, 1943, pp 40–41. 16. Wikkelso C, Andersson H, Blomstrand C, Lindqvist G: The clinical effect of Marvin Bergsneider lumbar puncture in normal pressure hydrocephalus. J Neurol Neurosurg Los Angeles, California Psychiatry 45:64–69, 1982. 17. Wikkelso C, Andersson H, Blomstrand C, Lindqvist G, Svendsen P: Normal pressure hydrocephalus: Predictive value of the cerebrospinal fluid tap-test. 1. Vanneste J, Augustijn P, Diren C, Tan WF, Goedhart ZD: Shunting normal-pres- Acta Neurol Scand 73:566–573, 1986. sure hydrocephalus: Do the benefits outweigh the risks? A multicenter study and literature review. Neurology 42:54–59, 1992. Acknowledgments We acknowledge the Medical Faculty, University of Lund, Lund, Sweden, for PH is a mysterious condition in which gait disorder, urinary financial support. Birgitta Kahlmeter and Maria Nilsson (registered physiother- Nincontinence, and memory loss may be reversible if patient selec- apists) and Birgitta Ekelund (registered occupational therapist) provided excel- tion is made appropriately. Because this condition occurs in aging lent help with CSF-TT. Sven Söderström (technician) provided expert technical patients, it is difficult to know how long the improvement may last. assistance with the LIT. Gunnar Gunnarson and Angelos Tsalamanis provided excellent information technology support. In the present study, Kahlon et al. have made a useful contribution toward understanding this condition. This study has three important messages. First, it demonstrates that there is greater than 80% success COMMENTS of initial shunting for idiopathic NPH with lumbar drainage as a trial for shunt placement. This has also been our experience with 200 ahlon et al. provide an important contribution describing the long- patients treated for this condition after lumbar drainage. Secondly, it Kterm follow-up of a cohort of prospectively evaluated normal pres- shows that objective improvement is sustained in 40% of patients at 5 sure hydrocephalus (NPH) patients. Although some degree of attri- years and that 56% of the patients feel they are still improved overall tion is expected in clinical studies, the reason for the high attrition at 5 years. Given the aging brain, these are quite reasonable statistics for observed in this study was interesting in itself. A 37% rate of death- improvement. Finally, and perhaps most importantly, it underscores related attrition occurred largely owing to cardiac and/or cerebrovas- the need to better understand this condition and its relationship to cular disease, conditions that were present in 65% of the initial cohort. degenerative changes in the brain. There is now increasing interest in This suggests that the potential impact of these particular comorbidities NPH and it is likely to be one of the more important diseases we have should be discussed by neurosurgeons counseling NPH patients or to treat as adult neurosurgeons. This article provides important evi- their families with regard to the risks and benefits of shunt surgery. dence that treatment is worthwhile over a long period. The authors report that only 20% of shunted patients continued with objective benefits at the time of the 5-year follow-up examination. This Peter McL. Black number is somewhat misleading considering the impact of mortality Boston, Massachusetts

332 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CLINICAL STUDIES

ADJUSTABLE SHUNT VALVE REPROGRAMMING AT HOME: SAFETY AND FEASIBILITY

Christian W. Sikorski, M.D. OBJECTIVE: Shunt valve resistance changes using a specialized magnetic program- Section of Neurosurgery, ming device permit noninvasive changes to cerebrospinal fluid drainage. In selected The University of Chicago, cases between 2001 and 2005, patients and families used shunt valve programming Chicago, Illinois devices at home. This study examines the safety and efficacy of this practice. David S. Rosen, M.D. METHODS: We conducted a retrospective review of the medical records of patients Section of Neurosurgery, who had been given a shunt valve-programming device for home use. A survey was The University of Chicago, mailed to patients or family members requesting information regarding their experi- Chicago, Illinois ences with the shunt valve programming device. Patient and family responses were tabulated and a statistical analysis was performed. David M. Frim, M.D., Ph.D. Section of Neurosurgery, RESULTS: Twenty patients or families returned the survey. The median patient age was The University of Chicago, 19.6 years (range, 6–48 yr); 25% were male. Seventeen patients had pseudotumor cere- Chicago, Illinois bri, one had an arachnoid cyst, and two had slit ventricle syndrome. Fifteen patients had lumboperitoneal shunts, one had a ventriculoperitoneal shunt, three had cisterna Reprint requests: David M. Frim, M.D., Ph.D., magna shunts, and one had an arachnoid cyst-to-peritoneal shunt. No adverse events Section of Neurosurgery, were attributable to the use of the home shunt valve programmer. Thirty-five percent The University of Chicago Medical Center, of respondents used the programmer at least once every week, 40% used the program- 5841 South Maryland Avenue, MC 4066, mer between once a week and once a month, and 25% used the programmer less fre- Chicago, IL 60637. quently than once per month. Overall, 85% of respondents reported that they bene- Email: [email protected] fited “very much” from the use of a home shunt valve programmer and 15% of respondents Received, March 3, 2006. benefited “somewhat.” Accepted, October 13, 2006. CONCLUSION: Providing shunt valve programming devices to selected patients for home use is a safe practice associated with high patient satisfaction. However, the selection of appropriate patients, comprehensive patient education, and close patient- physician communication are crucial to the success of this practice. KEY WORDS: Home therapy, Hydrocephalus, Lumboperitoneal shunt, Programmable shunt valve, Pseudotumor cerebri, Ventriculoperitoneal shunt

Neurosurgery 60:333–337, 2007 DOI: 10.1227/01.NEU.0000249269.11074.CA www.neurosurgery-online.com

rogrammable shunt valves create additional treatment many of these patients were traveling long distances to our options for patients requiring cerebrospinal ﬂuid (CSF) medical center for valve reprogramming, we adopted a practice Pdrainage (1, 3–6, 8). Programmable shunt valves are of allowing selected patients to use shunt valve-programming equipped with a range of resistance settings that can be devices at home. The patients and families were trained in the changed noninvasively using a magnetic programming device. operation of the programmer and frequently communicated This feature permits active customization of shunt therapy to with our medical team regarding symptoms and changes in alleviate the symptoms of CSF over- or underdrainage (7). valve settings. This study examines our experience with home However, only a limited number of neurosurgical centers have shunt valve programming. specialized valve programming devices. Consequently, most patients must visit their treating medical center to have their PATIENTS AND METHODS valve setting changed. At our medical center, we observed a subgroup of patients The University of Chicago Hospital Institutional Review who frequently visited the neurosurgery clinic or emergency Board approved our research protocol. We conducted a retro- department for reprogramming of their shunt valves. Because spective review of the medical records of patients who had

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 333 SIKORSKI ET AL.

been supplied with a shunt valve programmer for home use. the frequency and outcome of medical center visits before and Prospective criteria were not developed to select patients for after they were given a home programmer. home programmer use; instead, patients were evaluated indi- The survey was mailed to all patients who had received vidually. Relative indications for home shunt programming home shunt valve programmers. Patients either returned the included a history of frequent shunt valve reprogramming, a survey by mail or at a scheduled office visit. Patients were history of adherence to treatment plans, a long distance from assured that the survey results would be anonymously evalu- home to the treatment center, and family and patient interest in ated. The results of the survey were tabulated and a statistical home shunt valve programming. Absolute contraindications analysis was performed using Microsoft Excel (Version XP; included a history or high likelihood of mental status changes Microsoft, Seattle, WA). associated with shunt malfunction, a history of poor patient follow-up, and, for children, inconsistent caregivers. If the hos- RESULTS pital incurred cost, the cost of the valve programmer was reimbursed by third-party payers at the time of shunt revision; if the All 20 patients or families who were sent surveys completed valve programmers were given to the hospital without charge, them. Table 1 includes the clinical characteristics of the survey there was no cost to the patient or a third-party payer. Patient respondents. The median age was 19.6 years (range, 6–48 yr) demographic information was collected and records were and 25% were male patients. The mean approximate distance reviewed for adverse events that could be attributed to the use between the patient’s home and treating medical center was of a home shunt valve programmer. 263.3 miles (range, 5–1125 miles). Seventeen patients had We developed a written survey instrument containing pseudotumor cerebri, one patient had a posterior fossa arach- 20 questions designed to evaluate patient and family experi- noid cyst, and two patients had slit ventricle syndrome associ- ences with a home shunt valve programming device. Five ated with a previously placed ventriculoperitoneal shunt. questions requested demographic information, two questions Fifteen patients had programmable lumboperitoneal (LP) were regarding home programmer education, six questions shunts, one patient had a programmable ventriculoperitoneal asked about programming practices, and two questions evalu- shunt, two patients were changed from programmable LP ated patient satisfaction. Five questions asked patients to recall shunts to programmable cisterna magna-to-peritoneal shunts,

TABLE 1. Characteristics of 20 patients who used a home shunt valve programmer Approximate distance Patient Age Shunt Programmable Home Diagnosis Shunt Valve from home to no. (yr)/sex (mo) shunt (mo) programmer (mo) medical center (miles) 1 9/M Pseudotumor LPS/ CMP Codman 14 14 14 45 2 14/M Pseudotumor LPS/CMP Codman 44 15 15 45 3 46/F Pseudotumor LPS Codman 29 5 5 165 4 11/M Pseudotumor LPS Codman 11 11 11 350 5 17/F Pseudotumor LPS Codman 8 8 8 305 6 27/F Pseudotumor LPS Codman 8 8 8 235 7 17/F Pseudotumor LPS Codman 3 3 3 90 8 19/F Arachnoid cyst Cyst-PS Strata 203 12 12 205 9 39/F Pseudotumor LPS Codman 44 30 18 675 10 20/F Pseudotumor LPS Codman 21 21 12 75 11 16/M Slit ventricle syndrome LPS Codman 28 28 3 400 12 6/F Pseudotumor LPS Codman 6 6 6 1125 13 15/M Pseudotumor LPS Codman 12 12 12 5 14 9/F Pseudotumor VPS Strata 52 52 52 805 15 6/F Pseudotumor LPS Codman 47 22 22 55 16 40/F Pseudotumor LPS Codman 18 18 18 270 17 7/F Pseudotumor LPS Codman 6 6 6 105 18 12/F Pseudotumor LPS Codman 4 4 4 170 19 48/F Pseudotumor CM-AS/ LPS Codman 38 38 18 40 20 13/F Slit ventricle syndrome LPS Codman 11 11 11 100 Mean 19.6 30.4 16.2 12.9 263.3

a LPS, lumboperitoneal shunt; CMP, cisterna magna-to-peritoneal shunt; VPS, ventriculoperitoneal shunt; Cyst-PS, arachnoid cyst-to-peritoneal shunt; CM-AS, cisterna magna-to-right atrium shunt.

334 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SHUNT VALVE REPROGRAMMING AT HOME

one patient had a programmable arachnoid cyst-to-peritoneal shunt, and one patient had both a programmable cisterna magna-to-atrial shunt and a programmable LP shunt. For patients with LP shunts, valves were placed superficially in the soft tissue of the back to facilitate valve reprogramming. Eighteen patients had Codman Hakim programmable valves (Codman & Shurtleff, Inc., Raynham, MA) and two patients had Strata programmable valves (Medtronic Neurosurgical, Goleta, CA). At the time the survey was completed, the mean number of months the patients had any type of shunt system (programmable or nonprogrammable valve) was 30.4 months. The mean length of time was 16.2 months with a programmable shunt valve and 12.9 months with a home shunt valve programmer. Survey respondents reported that the median amount of FIGURE 1. Graph showing the reported frequency of medical center visits time for education regarding the operation of the valve pro- before and after using a home shunt valve programmer. grammer was between 5 and 15 minutes. Ninety-five percent of respondents felt comfortable operating the programmer at three key elements: 1) rigorous patient selection, 2) thorough the time it was issued. A single person operated the program- patient and family education, and 3) active communication mer for 70% of patients and multiple operators for 30%. between the patient, the patient’s family, and the healthcare Thirty-five percent of respondents reported using the pro- team. grammer at least once every week, 40% used the programmer Several large series (1, 2, 4–9) and one randomized, con- between once a week and once a month, and 25% used it less trolled trial (3) have examined the use of programmable shunt frequently than once per month. Only 35% of patients always valves. Indications for programmable shunt valves include or almost always contacted the neurosurgery office for guid- hydrocephalus of various causes, normal pressure hydro- ance before reprogramming the valve. Thirty-three percent of cephalus, arachnoid cysts, and pseudotumor cerebri (6, 8). respondents reported that they were “always sure” that the However, not all patients with programmable valves are candi- valve was reprogrammed properly, 42% were “sometimes dates for home programming and we do not advocate the rou- unsure” regarding valve reprogramming, and 25% felt “mostly tine use of home valve programmers for the general shunt pop- unsure” regarding valve reprogramming. When unsure ulation. The present series of patients did not have strict regarding valve settings, 50% of respondents had gone to the inclusion criteria; instead, patients were evaluated on an indi- neurosurgery clinic or an emergency department at least once vidual basis. In general, patients had failed conventional shunt to check valve settings by x-ray. Our staff often reviewed these treatment and remained symptomatic, most commonly with x-rays using telemetric electronic media. headaches, from conditions such as pseudotumor cerebri, No adverse events were attributable to the use of the home arachnoid cysts, or slit ventricle syndrome. shunt valve programmer. Fifty-three percent of respondents Patient and family education regarding the shunt valve pro- “always” or “almost always” experienced improvement in grammer was performed before and after receiving the home symptoms after valve reprogramming, 42% “sometimes” expe- programmer. Most programmers were provided during an rienced improvement, and 5% “rarely” experienced improve- inpatient hospital admission. During the admission, a patient ment. Figure 1 shows that patients reported a significantly and family education session was conducted to demonstrate lower frequency of medical center visits after obtaining a home the operation of the programmer. Emphasis was placed on shunt valve programmer. Thirty-seven percent of respondents understanding the anatomy and location of the shunt, espe- never required urgent medical evaluation in the neurosurgery cially for patients with programmable LP shunts because the clinic or emergency room after receiving a home programmer. soft tissue of the back makes valve programming more chal- Overall, 85% of respondents reported that they benefited “very lenging. After discharge, an active effort was made to maintain much” from the use of a home shunt valve programmer, and communication through phone calls, email, and office visits to 15% of respondents reported that they benefited “somewhat.” answer questions regarding shunt valve programming. As patients and families developed familiarity with shunt valve DISCUSSION programming, many began to make programming decisions independent of the medical team. The results of this investigation demonstrate that selected Programmable and nonprogrammable valves seem to have patients can safely use shunt valve programmers at home and similar failure rates (3). Rare complications associated with pro- suggest that home programmers may decrease the number of grammable valves include misprogramming of valve settings, medical center visits and increase patient satisfaction. The suc- failure of the valve to reprogram, and inadvertent valve repro- cessful use of home shunt valve programmers is dependent on gramming, especially after magnetic resonance imaging scans

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 335 SIKORSKI ET AL.

or exposure to other powerful magnets. Home programming is REFERENCES unlikely to change shunt failure rates. The present series demonstrated no adverse events from home programming; 1. Catalan G, Bilbao G, Pomposo I, Aurrecoechea J, Garibi J: Our experience with the programmable Codman-Medos valve: Review of 125 shunts [in however, the primary potential complication of home program- Spanish]. Rev Neurol 31:1136–1142, 2000. ming is misprogramming of the valve settings, leading to CSF 2. Kay AD, Fisher AJ, O’Kane C, Richards HK, Pickard JD; United Kingdom and under- or overdrainage. Therefore, patients with a history of Ireland Medos Shunt Audit Group: A clinical audit of the Hakim program- mental status changes from shunt underdrainage or malfunc- mable valve in patients with complex hydrocephalus. Br J Neurosurg 14: tion are not candidates for home programming. Overdrainage 535–542, 2000. of CSF rarely leads to serious complications but can result in a 3. Pollack IF, Albright AL, Adelson PD: A randomized, controlled study of a programmable shunt valve versus a conventional valve for patients with subdural hematoma; patients should be educated regarding hydrocephalus. Hakim-Medos Investigator Group. Neurosurgery 45: the signs and symptoms of this complication. 1399–1408, 1999. Several series have reported a mean number of valve repro- 4. Ringel F, Schramm J, Meyer B: Comparison of programmable shunt valves vs grammings between one and two per patient (1, 2, 4, 5, 7, 8). standard valves for communicating hydrocephalus of adults: a retrospective However, the present series includes a large percentage of analysis of 407 patients. Surg Neurol 63:36–41, 2005. 5. Rohde V, Mayfrank L, Ramakers VT, Gilsbach JM: Four-year experience with patients with pseudotumor cerebri. Our clinical experience has the routine use of the programmable Hakim valve in the management of demonstrated that this group of patients requires more fre- children with hydrocephalus. Acta Neurochir (Wien) 140:1127–1134, 1998. quent valve adjustments. Our study also differs from previ- 6. Yamashita N, Kamiya K, Yamada K: Experience with a programmable valve ously reported series in that it includes a large percentage of shunt system. J Neurosurg 91:26–31, 1999. patients with LP shunts. 7. Zemack G, Bellner J, Siesjo P, Stromblad LG, Romner B: Clinical experience with the use of a shunt with an adjustable valve in children with hydro- The present study was not designed to include a cost analy- cephalus. J Neurosurg 98:471–476, 2003. sis, but the financial impact of this practice is important for 8. Zemack G, Romner B: Seven years of clinical experience with the programma- evaluating its feasibility. Our institution charges approximately ble Codman Hakim valve: A retrospective study of 583 patients. J Neurosurg $1400 for one of the Codman Hakim programming devices, 92:941–948, 2000. which was reimbursed by third-party payers at the time of 9. Zemack G, Romner B: Do adjustable shunt valves pressure our budget? A ret- shunt revision. However, the cost for an office visit with shunt rospective analysis of 541 implanted Codman Hakim programmable valves. Br J Neurosurg 15:221–227, 2001. reprogramming is $310; for an emergency department visit with shunt reprogramming (excluding x-rays), the cost is $475. Therefore, if this practice prevents five office visits or three COMMENTS emergency department visits for valve reprogramming, the ikorski et al. evaluate a group of patients who were allowed to cost savings would justify the cost of the home programmer. Sadjust their cerebrospinal fluid diverting shunts at home. The study One of the primary limitations of the present study is the use was generated by the responses of 20 patients or family members who of a survey to measure outcomes. Our survey relied on respon- completed and returned a survey. The authors report no adverse con- dent recall; however, our study population may not have sequences by changing the valve settings. In terms of frequency, 35% of remembered accurately and may have introduced recall bias the respondents changed the valve setting at least once per week, 40% into our analysis. In addition, the accuracy of home shunt valve changed the setting between once per week and once per month, and programming was not measured and, therefore, it is not known 25% changed the setting less frequently than once per month. The whether or not the benefits reported by our survey respondents authors note that 85% of the respondents indicated that they benefited were caused by actual changes in the valve settings or placebo very much from changing their valve setting at home and 15% reported effect from using the programmer. Finally, it is unclear what some benefit. Apparently, there were no responses that indicated no physiological mechanisms explain the apparent benefit some benefit or adverse effects from changing the valve setting. The authors conclude that home programming of shunt valves is feasible, safe, and patients reported from frequent changes in shunt valve settings. possibly effective in selected patients. Despite these unanswered questions, the results of this As the authors note in their discussion, the survey results were study indicate that home shunt valve programmer use is a dependent on the information supplied by the respondents. Thus, the feasible, safe, and, possibly effective therapy for selected number of times and the degree to which the valve settings were patients. We do not advocate the routine use of home shunt changed may or may not reflect reality. The fact that 85% of the respon- valve programmers for the general shunt population, but for dents benefited very much and 15% benefited somewhat from chang- patients who have failed conventional shunt therapy and ing the valve is in itself remarkable. I agree with the authors’ conclu- remain symptomatic from conditions such as pseudotumor sion that home adjustment of programmable shunts is safe in some cerebri, arachnoid cysts, or slit ventricle syndrome, home selected situations. Whether or not it is effective is questionable, as the shunt valve programming may be an additional therapeutic placebo effect may have been the predominant factor. In a pediatric option. Although no adverse events were attributed to home population, we have not found that programmable valves are of bene- shunt valve programming during the study period, this prac- fit over those that are not, except in very rare situations. Obviously, the tice has inherent risks that should be thoroughly explained to case series presented differs markedly from that of the usual pediatric patients. This report may serve as the basis for a prospective, shunt population. randomized trial with strictly defined inclusion criteria to bet- J. Gordon McComb ter evaluate the efficacy of this practice. Los Angeles, California

336 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SHUNT VALVE REPROGRAMMING AT HOME

eurosurgeons who perform a large number of shunts can gener- selected patients (mostly pseudotumor adults). Although not designed Nally identify a small subgroup of patients in their own practices as an efficacy study, the reader obviously wants to derive information who are similar to those patients discussed in this study. These patients about this. The fact that almost all of the patients subjectively benefited all have certain features in common. The condition for which they were is not particularly revealing because they seemed to reprogram their shunted is not associated with life-threatening sequelae if the valve shunts on a regular basis, which makes no physiological sense. These setting is wrong or if the shunt malfunctions. Nonetheless, they and patients are notoriously fragile from an emotional point of view and, as their families are plagued by recurring symptoms, particularly the authors point out in the discussion, all of the benefits may have been headache, that require some sort of shunt adjustments. If they live at a the placebo effect. One wonders if the same benefit would have been distance from the neurosurgeon, major effort is required to obtain help. found if these patients had been sent home with the programming Even patients who live near a major medical center may spend painful, device without programmable shunts or with a dummy programmer. frustrating, and expensive hours (usually at night) in the hospital emer- We also do not know if the adjustments made at home resulted in gency department with their families. This may lead to a tendency to any change in the actual shunt parameters. The Codman system avoid seeking help unless they are quite ill or their anxiety too great, (Codman & Shurtleff, Inc., Raynham, MA), in particular, is pretty particularly if experience tells them that a minor adjustment of the finicky and radiographic confirmation is needed to truly prove that the shunt is all that is needed. For such individuals, the feeling that they intended change was effected. This technique has the advantage of can help themselves must be very reassuring. It is not surprising that sparing the neurosurgeon endless visits for reprogramming. On the their responses to the questionnaire were generally positive. We all other hand, it still has the risk that an inadvertent change to a high- recognize that such patients must be selected and educated very care- pressure valve could result in death. It is also potentially quite expen- fully, as the authors have done in this study. sive, and it is unclear who would pay for the home programmers. This Paul H. Chapman is an idea that deserves further investigation, but a placebo controlled Boston, Massachusetts trial would be most helpful, if it could be conducted in an ethical way.

his is a Phase 1 safety study of the use of home shunt programming. Leslie N. Sutton TIt is an interesting idea that proved to be safe in a small number of Philadelphia, Pennsylvania

“Sugar Ray” Robinson (1921–1989) and Carmen Basilio (1927–) in the 10th round of their classic middle-weight championship match at Chicago Stadium, 1958. CLINICOPATHOLOGICAL STUDY

EXPRESSION OF HYPOXIA-INDUCING FACTOR-1α AND ENDOGLIN IN INTIMAL HYPERPLASIA OF THE MIDDLE CEREBRAL ARTERY OF PATIENTS WITH Yasushi Takagi, M.D., Ph.D. Department of Neurosurgery, MOYAMOYA DISEASE Kyoto University Graduate School of Medicine, Kyoto, Japan OBJECTIVE: Moyamoya disease (MMD) is a cerebrovascular occlusive disease characterized by progressive stenosis or occlusion at the distal ends of the bilateral internal Ken-ichiro Kikuta, M.D., Ph.D. arteries. In MMD, intimal hyperplasia was previously reported to be found in autopsy Department of Neurosurgery, Kyoto University samples. In this study focusing on the mechanism of remodeling of the intracranial arte- Graduate School of Medicine, rial walls of patients with MMD, we surgically collected tiny pieces of the wall of the Kyoto, Japan middle cerebral artery (MCA) from patients with MMD and analyzed them using histological and immunohistochemical methods. Kazuhiko Nozaki, M.D., Ph.D. METHODS: Twelve patients underwent surgical procedures for treatment of standard Department of Neurosurgery, Kyoto University indications of MMD at Kyoto University Hospital. Specimens of MCA were obtained Graduate School of Medicine, from MMD patients during the surgical procedures. Nine MCA samples were also Kyoto, Japan obtained in the same way from control patients. The samples were analyzed by immunohistochemical methods. Motoaki Fujimoto, M.D. Department of Neurosurgery, RESULTS: MCA specimens from MMD patients had a thicker intima than those from Kyoto University the control group. In MMD samples, the immunoreactivity indicating hypoxia- Graduate School of Medicine, inducing factor-1α was higher in the endothelium and intima; endoglin expression was Kyoto, Japan also higher in the endothelium. No vascular endothelial growth factor immunoreactivity was detectable in the MMD samples. In addition, transforming growth factor-β3 Junya Hayashi, M.D., Ph.D. immunoreactivity was also detected and was co-localized with that of hypoxia-induc- Department of Neurosurgery, α Kyoto University ing factor-1 and endoglin, mainly in the endothelium. Graduate School of Medicine, CONCLUSION: Our results indicate that the MCA specimens from MMD patients had Kyoto, Japan thicker intimal walls than the specimens from control patients. In addition, hypoxia- α Hirotoshi Imamura, M.D. inducing factor-1 and endoglin were overexpressed in the intima of the MCA of MMD Department of Neurosurgery, patients. Kyoto University KEY WORDS: Endoglin, Hypoxia-inducing factor-1α, Intimal hyperplasia, Moyamoya disease Graduate School of Medicine, Kyoto, Japan Neurosurgery 60:338–345, 2007 DOI: 10.1227/01.NEU.0000249275.87310.FF www.neurosurgery-online.com

Nobuo Hashimoto, M.D., Ph.D. Department of Neurosurgery, oyamoya disease (MMD) is a cere- Histopathological investigations on autopsy Kyoto University brovascular occlusive disease charac- subjects have demonstrated that the main vas- Graduate School of Medicine, terized by progressive stenosis or cular lesion in moyamoya disease is stenosis Kyoto, Japan M occlusion at the distal ends of bilateral internal or occlusion caused by a fibrocellular intimal arteries (29). The unusual vascular network at thickening with a multilayered elastic lamina Reprint requests: Yasushi Takagi, M.D., Ph.D., the base of the brain (moyamoya vessels) is and some lipid deposits (11, 29, 33). In addi- Department of Neurosurgery, considered to represent collateral channels tion, a decrease in the number of medial Kyoto University formed as a result of progressive ischemic smooth muscle cells was found (11, 33). As for Graduate School of Medicine, changes in the brain (29). The etiology of the the treatment of MMD, superficial temporal Kawahara-cho 54, Shogoin, disease is undefined. The findings that the inci- artery-middle cerebral artery (STA-MCA) Sakyo, Kyoto 606-8507, Japan. Email: [email protected] dence of the disease is highest in, but not con- bypass surgery or indirect revascularization is fined to, the Japanese population and that the usually performed (12, 14, 17). During this Received, January 5, 2006. condition is often familial suggest the involve- surgery, the middle cerebral artery (MCA) of Accepted, October 3, 2006. ment of a genetic factor in its pathogenesis (29). MMD patients is observed to have a type of

338 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HYPOXIA-INDUCING FACTOR-1␣ AND ENDOGLIN IN MOYAMOYA INTIMA

arterial wall different from that bearing atherosclerotic lesions A (17, 34). Several studies on extracranial vessels have been reported previously (2, 4, 8, 10, 28). However, few reports are available on intracranial arteries from MMD patients other than those based on autopsy specimens (11, 33). In this study, we collected tiny pieces of MCA walls from patients with MMD and analyzed their histopathological features. In addition, we analyzed these vessel walls immunohistochemically to explore the mechanism of remodeling of the intracranial arterial walls in the MMD patients.

PATIENTS AND METHODS B Patients Twelve patients underwent surgical procedures for treatment of standard indications of moyamoya disease at Kyoto University Hospital in Kyoto, Japan. Specimens were obtained from the patients during the surgical procedures (Fig. 1). Clinical data on the patients are summarized in Table 1. Twelve control MCA samples were also obtained in the same manner from the control subjects described in Table 1. Nine patients FIGURE 1. A, surgical view of STA-MCA bypass for the treatment of MMD. with internal carotid artery or MCA occlusion underwent STA- The operator (YT or KK) performed arteriotomy with microscissors (arrow MCA bypass surgery in the chronic stage. Three patients with indicates collected tissue). B, samples were fixed in 10% formalin and viewed aneurysms also underwent STA-MCA bypass surgery when under a microscope. the parent artery was occluded. This study was performed under the guidelines provided by the ethics committee of the a Kyoto University School of Medicine. TABLE 1. Summary of patients MMD Control Sample Preparation No. of patients 12 12 During STA-MCA bypass surgery, an 11–0 nylon monofilament Age (yr) 37 Ϯ 8 55 Ϯ 20 was passed around the wall of the recipient artery. The vessel was Sex (men/women) 2/10 10/2 then pulled up by lifting the monofilament with forceps, and the Type of onset operator (YT or KK) performed arteriotomy with microscissors TIA 11 7 (Fig. 1). Then, STA-MCA end-to-side anastomosis was per- Infarction 14 formed. In some experiments, a control sample was obtained Hemorrhage 00 from a 27-year-old woman with an arteriovenous malformation. Mass effect 01 All specimens were fixed in 10% formalin overnight and Disease ICA-O 8 embedded in paraffin the next day. The specimens were stored MCA-O 1 at room temperature. In each case, multiple, sequential, 6-µm Aneurysm 3 thick tissue sections cut from paraffin blocks were deparaf- Samples finized in xylene, rehydrated, and prepared for immunohisto- M4 12 10 chemical studies. M2 02

a Antibodies MMD, moyamoya disease; TIA, transient ischemic attack; ICA-O, internal cerebral artery occlusion; MCA-O, middle cerebral artery occlusion. Data Anti-hypoxia-inducible factor-1α (HIF-1α; 1:100; are expressed as mean Ϯ standard deviation. NeoMarkers, Fremont, CA), anti-endoglin, (CD105; 1:30; DAKO, Glostrup, Denmark), anti-vascular endothelial growth factor (VEGF; 1:100; NeoMarkers), and anti-transforming Immunohistochemical Analysis growth factor (TGF)-β3 (1:50; SantaCruz Biotechnology, Santa The sections were washed for 5 minutes with 0.01 mol/L Cruz, CA) were used as the primary antibodies in this study. phosphate-buffered saline (PBS; pH, 7.2), followed by a As secondary antibodies, fluorescein isothiocyanate- 15-minute incubation with 10 µg/ml proteinase K. After being conjugated anti-mouse immunoglobulin G and cyanogen3- blocked with 3% H2O2, the sections were preincubated with conjugated anti-rabbit immunoglobulin G (1:200; Jackson normal goat serum (diluted 1:50) and then incubated overnight Immunoresearch, West Grove, PA) were used. at 4ЊC with the desired primary antibody. After three 15-minute

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 339 TAKAGI ET AL.

rinses with PBS, the sections were incubated for 30 minutes A B with the anti-mouse Envision/horseradish peroxidase system (Dako). After three more rinses with PBS, the sections underwent color development for 5 minutes at room temperature in a substrate medium containing 0.05% 3,3-diaminobenzidine and 0.02% H2O2 in Tris-hydrochloride buffer (pH, 7.6). The specificity of the staining was confirmed by the absence of specific staining when nonimmune rabbit immunoglobulin G was substituted for the primary antibody. The sections were analyzed under a BX51 fluorescent microscope (Olympus Optical C D Co., Tokyo, Japan) or a Fluoview FV300 laser confocal microscope (Olympus Optical Co.).

Immunofluorescence Double-staining The sections were washed for 5 minutes with 0.01 mol/L PBS (pH, 7.2) and then incubated for 15 minutes with 10 µg/ml proteinase K. The sections were preincubated with skim milk Њ (diluted 1:50) and were then incubated overnight at 4 C with FIGURE 2. A and B, hematoxylin and eosin-stained specimens from the MCA the desired primary antibodies. After three 15-minute rinses of a patient with MMD (A) and a control patient (B). C and D, histological with PBS, they were incubated for 30 minutes with the fluores- findings assessed by interference differential microscopy of the MCA of a cent secondary antibodies. After three additional rinses with patient with MMD (C) and a control patient (D). The arrows indicate the PBS, the sections were observed under a BX51 fluorescence internal elastic lamina. Scale bar, 150 µm (C); original magnification, ϫ100 microscope (Olympus Optical Co.). (A and B) and ϫ200 (C and D).

Immunohistochemical Analysis Under a BX51 fluorescence microscope (Olympus Optical TABLE 2. Intimal hyperplasia of middle cerebral arteries of a Co.), the histological images were captured with a computer. patients with moyamoya disease Then, intimal thickness and the number of immunopositive MMD Control cells were analyzed by use of an Image-Pro image-analyzing No. of patients 12 12 system (Media Cybernetics, Silver Spring, MD). Using an Intimal thickness (␮m) 38 Ϯ 19b 11 Ϯ 5 image analysis system, the immunoreactivity was assessed by a Ϯ two observers (YT, KK) in a blind manner. The two observers MMD, moyamoya disease. Data are expressed as mean standard deviation. set their own result about the percentage of immunopositive b P ϭ 0.0006. cells in each sample, and the mean score was recorded.

Single-photon Emission Computed Tomography 11 Ϯ 5 µm; all data are expressed as the mean Ϯ standard deviation; P ϭ 0.0006) (Fig. 2, Table 2). The mean values of rCBF and Cerebral blood flow (CBF) and cerebrovascular reserve rCVR in the MMD patients were 35 Ϯ 5 ml/100g/minute and (CVR) were assessed as previously described (35). The entire –9 Ϯ 9%, respectively. In control patients with internal carotid territory of the MCA was the region of interest for evaluation of artery or MCA occlusion (n ϭ 9), those values were 26 Ϯ 4 the regional CBF (rCBF). ml/100g/minute and –8 Ϯ 7%, respectively. Statistical Analysis Expression of HIF-1α Immunoreactivity As for the results including intimal thickness, the percentage Next, we performed immunohistochemical studies on the of immunopositive cells, rCBF, and regional CVR (rCVR), the MCA from MMD patients by using antibody specific for HIF- Mann-Whitney and Fisher’s exact tests were used for statistical 1α. For the MMD specimens, 66 Ϯ 22% of the endothelial and analysis (StatView; SAS Institute, Cary, NC). A P value less 33 Ϯ 17% of the intimal cells demonstrated anti-HIF-1α than 0.05 was considered statistically significant. immunoreactivity (Fig. 3, Table 3). On the contrary, for control samples, the respective values were 21 Ϯ 12% and 14 Ϯ 7% RESULTS (Fig. 3, Table 3). (All data are expressed as the mean Ϯ standard deviation; P Ͻ 0.0001 and 0.006, respectively.) Thickness of the Intima Using the collected specimens, we measured the thickness of Expression of Endoglin Immunoreactivity the intima in hematoxylin and eosin-stained sections (Fig. 2). Next, we performed immunohistochemical analysis of The mean thickness of the MMD intima was significantly greater MMD specimens using antibodies specifically recognizing Ϯ µ than that of the control specimens (MMD, 38 19 m; control, VEGF and endoglin. VEGF immunoreactivity was not

340 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HYPOXIA-INDUCING FACTOR-1␣ AND ENDOGLIN IN MOYAMOYA INTIMA

AB C detected in MMD samples (Fig. 4). Immunoreactivity indicating endoglin was detected in 67 Ϯ 27% of the endothelial cells in the MMD specimens (Fig. 4, Table 3). For the control specimens, this value was 23 Ϯ 15% (Fig. 4, Table 3; P ϭ 0.008). On the DE F contrary, 5 Ϯ 8% of the intimal cells in the MMD specimens and 3 Ϯ 5% in the control samples showed endoglin immunoreactivity (Table 3; P ϭ 0.6).

Colocalization of TGF-β3 Immunoreactivity with FIGURE 3. Immunohistochemical analysis of HIF-1α. A–C, cells that tested immunopositive for HIF-1α are detected in HIF-1α and Endoglin the endothelium and intima of the MCA specimens of patients with MMD. Arrows indicate the internal elastic lamina. Immunoreactivities D and E, in this control specimen, a smaller number of such cells are detected. Arrows indicate the internal elastic lam- To identify the nature of the ina. F, no positive staining is seen in the negative control samples in which case the primary antibody was omitted. cells immunoreactive with anti-HIF-1α and anti-endoglin antibodies, we performed double-staining and observed the ␣ cells under a ﬂuorescence microscope. The results of double- TABLE 3. Intimal expression of hypoxia-inducing factor-1 and α Β endoglin in middle cerebral arteries with moyamoya diseasea staining for HIF-1 and TGF- 3 indicated that the double- labeled cells were detected mainly in the endothelium (Fig. 5, HIF-1␣ Endoglin A–C and G–I). Moreover, we also performed double-staining Endothelium Intima Endothelium Intima for endoglin and TGF-Β3. The cells in the endothelium with MMD (n ϭ 12) 66 Ϯ 22% 33 Ϯ 17% 67 Ϯ 27% 5 Ϯ 8% immunoreactivity indicating endoglin were also labeled with Control (n ϭ 12) 21 Ϯ 12% 14 Ϯ 7% 23 Ϯ 15% 3 Ϯ 5% anti-TGF-Β3 (Fig. 5, D–F and J–L). P value Ͼ0.0001b 0.006b 0.008b 0.6

a HIF-1␣, hypoxia-inducing factor 1 ␣; MMD, moyamoya disease. Data are Correlation of the Results of the CBF Study with expressed as mean Ϯ standard deviation. HIF-1α and Endoglin Expression in the MMD Patients b P values less than 0.05 were considered signiﬁcant. To analyze a correlation of CBF and CVR with immunohistochemical results, we assessed rCBF and rCVR in the MMD

AB CFIGURE 4. Immunohistochemical analysis of VEGF and endoglin. A and B, immunoreactivity indicating that VEGF is not detected in the intima and the media of the MCA of patients with MMD. C, in a positive control specimen (cerebral arteriovenous malformation), positive immunoreactions for VEGF are detected. D and E, immunoreactivity indicating that DE Fendoglin is detected in the endothelium of the MCA from the patients with MMD. F, in this control specimen, no such cells are detected in the endothelium. Arrows indicate the endothelium. Original magnification, ϫ100 (A–C) and ϫ200 (D–F).

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 341 TAKAGI ET AL.

A B C

D E F

GHI JKL

FIGURE 5. Fluorescent microscopical assessment by double immunofluorescence staining to characterize the cells immunoreactive for HIF-1α, endoglin, and TGF-β3 in MMD specimens. A–C, results for double-staining for HIF-1α (A, red) and TGF-β3 (B, green). HIF-1α-containing cells are also positive for TGF- β3 (C, yellow). D–F, results for double-staining for endoglin (D, red) and TGF-β3 (E, green). Endoglin is co-localized with TGF-β3, as indicated in the merged image (F, yellow). G–I, results for double staining for HIF-1α (G, red) and TGF-β3 (H, green). HIF-1α-containing cells are also positive for TGF-β3 (I, yellow; arrow indicates a co-localized cell). J–L, results for double-staining for endoglin (J, red) and TGF-β3 (K, green). Endoglin is co-localized with TGF-β3, as indicated in the merged image (L, yellow; arrow indicates a co-localized cell). Scale bar, 50 µm (L). Original magnification, ϫ100 (D–F), ϫ200 (A–C), and ϫ400 (G–L). patients (Table 4). The patients with more than 60% positive HIF- 1α immunoreactivity in the intima showed significant rCVR TABLE 4. Cerebral blood flow studies and up-regulation of reduction compared with those with less than 60% immunoreac- hypoxia-inducing factor-1␣ and endoglin in moyamoya diseasea Ϯ Ϯ ϭ tivity (–14 7% and 1 7%, respectively; P 0.02). HIF-1␣ Endoglin rCBF rCVR (%) rCBF rCVR (%) DISCUSSION % Positive In the present study, we obtained data on surgically sampled Յ60 34 Ϯ 7 Ϫ14 Ϯ 7 32 Ϯ 5 Ϫ5 Ϯ 8 MCA specimens from patients with MMD. On histopathologi- Ͼ60 37 Ϯ 41 Ϯ 7 33 Ϯ 5 Ϫ10 Ϯ 8 cal observations, the MMD specimens had a thicker intima P value 0.4 0.02b 0.9 0.5 than the controls. Furthermore, the immunoreactivity indicat- a ␣ ␣ α HIF-1 , hypoxia-inducing factor 1 ; rCBF, regional cerebral blood flow; ing the presence of HIF-1 and endoglin was higher in MMD rCVR, regional cerebrovascular reserve. specimens than in the controls. b P values less than 0.05 were considered significant. MMD is characterized by angiographic findings of intracranial carotid artery stenosis and occlusions as well as a fine network of vessels at the base of the brain (29). As concluded

342 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HYPOXIA-INDUCING FACTOR-1␣ AND ENDOGLIN IN MOYAMOYA INTIMA

from inspection of postmortem specimens, the pathological endoglin in cardiovascular development and vascular remod- changes in the stenotic and occluded vessels are intimal thick- eling has been learned from several experimental approaches. ening, hyperplasia, and irregularities in the internal elastic The function of endoglin in vascular morphogenesis was lamina (11, 13). These changes in intracranial arteries and sec- demonstrated in knockout mice, the embryos of which die at 10 ondary angiogenesis are considered characteristic of MMD to 11.5 days because of vascular and cardiac abnormalities (1, 7, (9). The molecular mechanism underlying MMD has been 20). In the central nervous system, endoglin expression is up- investigated previously but has not been fully clarified. Our regulated in arteriovenous malformations and cavernous group and others reported basic fibroblast growth factor to be hemangiomas (6, 27). involved in the pathogenesis of MMD, as basic fibroblast Recently, evidence has been obtained indicating that HIF-1␣ growth factor expression was elevated in the cerebrospinal also seems to regulate TGF-␤3 transcription (9, 22). Two other fluid and superficial temporal artery (STA) of MMD patients isoforms of TGF-␤, TGF-␤1 and TGF-␤2, have also been identi- (10, 26, 32). In addition, Hojo et al. (8) recognized an elevated fied. Although TGF-␤ isoforms are structurally similar, they dif- level of TGF-␤ in the cerebrospinal fluid, as well as TGF-␤ fer in their ability to bind receptors (9, 22). In human umbilical expression in the STA, of patients with MMD. Yamamoto et al. cord vein endothelial cells, the dimeric membrane glycoprotein (37) also showed TGF-␤ to play a role in MMD by elastin syn- endoglin coexists with TGF-␤ receptors I and II and binds TGF- thesis. These results indicate the accumulation of elastin via ␤1 and -␤3 with high affinity but does not bind TGF-␤2 (9, 22). the TGF-Β pathway in the intimal thickening and suggest that The role of TGF-␤ family members in vascular development is this accumulation is responsible for the intimal thickening undoubtedly complex, as is the cascade of events that lead to seen in MMD (37). Besides basic fibroblast growth factor, the development of a blood vessel. One might conclude that the other growth factors such as platelet-derived growth factor ratio of TGF-␤ signals via these receptors determines whether and hepatocyte growth factor also play a role in MMD (2, 24, or not TGF-␤ stimulates or inhibits angiogenesis (8, 19). 32). Moreover, intercellular adhesion molecule-1, vascular cell As for the significance of this study, we emphasize that ours adhesion molecule-1, and elastin levels were previously is the first to study use of MCA samples collected during sur- reported to be increased in the cerebrospinal fluid of MMD gery. As formerly mentioned, studies on growth factors in cere- patients (28). brospinal fluid and STA samples have already been published In this study, we also analyzed the expression of HIF-1α and (2, 3, 4, 8, 10, 18, 26, 28, 32, 36). However, patients with MMD endoglin using immunohistochemistry. Our group studied the present systemic arterial lesions, including those of the STA or role of TGF-β in MMD (8). We have searched for the up- and renal artery (4, 31). Considering that the main characteristic of downstream molecules of TGF-␤. Recently, evidence has been MMD is intracranial stenosis, studies using intracranial arterial obtained indicating that HIF-1 regulates TGF-␤3 transcription samples are the most important. In this study, the control (19, 21). In MMD, a reduction in CBF and hypoxia was often patients are significantly older than the MMD patients. In observed. HIF-1α is one of the major factors involved tissue- carotid arteries, intimal hyperplasia often occurs in older oxygen homeostasis; thus, we selected HIF-1α in this study. patients with arteriosclerosis. In the present study, intimal Endoglin is known to modulate cellular responses to TGF-␤. As hyperplasia is more obvious in the MMD patients. Intimal a downstream molecule of TGF-␤16, we analyzed endoglin. hyperplasia in these patients may not be owing only to age. Thus, we selected HIF-1α and endoglin, as they are molecules Considering the differences in angiogenic markers between associated with TGF-␤. We analyzed the expression of a novel control and MMD patients, CBF reduction is not the only cause member of the TGF-␤ family, TGF-␤3, in MMD. HIF-1α, which of this difference because the mean rCBF value was lower in the is a transcriptional activator involved in tissue-oxygen homeo- control specimens than in the MMD patients. As shown in Table stasis (15, 22), is a heterodimeric protein formed by inducible 4, CVR reduction in MMD is one of the origins for HIF-1α HIF-1α and constitutively expressed HIF-1α proteins. It binds induction. In addition, growth factors such as fibroblast growth to hypoxia-responsive elements to activate transcription of factor are reported to be elevated in the cerebrospinal fluid of genes related to iron, energy, and matrix metabolism, to vascu- MMD patients (10). With the presence of fibroblast growth fac- lar regulation, and to cell survival (15, 22). These types of genes tor, HIF-1α promotes the proliferation of smooth muscle cells include VEGF and VEGF receptor 1, TGF-␤ isoforms, heme (25, 30). In addition, hepatocyte growth factor, which is also ele- oxygenase-1, inducible nitric oxide synthase, and plasminogen vated in the cerebrospinal fluid of these patients, induces HIF- activator inhibitor-1 (15, 22). Among them, VEGF is one of main 1α (5, 18). These growth factors may play a role in HIF-1α target molecules of HIF-1␣ (15). Thus, we first analyzed the induction and proliferative responses in MMD (25). expression of VEGF in MMD specimens but did not detect any In this study, we used M4 samples. In MMD disease, pro- VEGF expression in them. Endoglin is a component of the gressive occlusion occurs in the circle of Willis arteries, but is TGF-␤ receptor complex expressed mainly on the surface of thought to occur rarely in the cortical M3 and M4. However, endothelial cells (7, 16, 20). Endoglin binds to members of the according to previous reports using autopsy specimens, corti- TGF-␤ superfamily, including TGF-␤1, TGF-␤3, activin A, cal MCA branches also showed intimal hyperplasia (33). These bone morphogenic protein-2, and bone morphogenic protein-7, reports, as well as our study, indicated that MMD affects not in the presence of the signaling receptors and modulates cellu- only the area around the terminal portion of internal carotid lar responses to TGF-␤1 (7, 16, 20). An important role for artery, but also the cortical branches of the MCA.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 343 TAKAGI ET AL.

Ultimately, a mechanistic insight into MMD pathophysiology 16. Li C, Hampson IN, Hampson L, Kumar P, Bernabeu C, Kumar S: CD105 may have to wait for the results of genetics studies. We know antagonizes the inhibitory signaling of transforming growth factor beta1 on the importance of genetic study and recognize that our study is human vascular endothelial cells. FASEB J 14:55–64, 2000. 17. Miyamoto S, Nagata I, Hashimoto N, Kikuchi H: Direct anastomotic bypass limited. Descriptive studies such as ours are open to bias, and for cerebrovascular moyamoya disease. Neurol Med Chir (Tokyo) 38 [Suppl]: the molecules studied are still somewhat arbitrary. Our group 294–296, 1998. has an ongoing project to study the genetics of MMD. 18. Nanba R, Kuroda S, Ishikawa T, Houkin K, Iwasaki Y: Increased expression However, only 10 to 20% of MMD patients have a defined of hepatocyte growth factor in cerebrospinal fluid and intracranial artery in genetic background. Thus, genetic approach cannot completely moyamoya disease. Stroke 35:2837–2842, 2004. 19. Nishi H, Nakada T, Hokamura M, Osakabe Y, Itokazu O, Huang LE, Isaka K: clarify the pathophysiology of MMD and, therefore, observa- Hypoxia-inducible factor-1 transactivates transforming growth factor-beta3 in tional studies are also important. trophoblast. Endocrinology 145:4113–4118, 2004. In summary, we showed that intimal hyperplasia occurred in 20. Qu R, Silver MM, Letarte M: Distribution of endoglin in early human devel- the MCA intima of patients with MMD. In addition, HIF-1α opment reveals high levels on endocardial cushion tissue mesenchyme dur- and endoglin were overexpressed in the intima. This study and ing valve formation. Cell Tissue Res 292:333–343, 1998. 21. Schaffer L, Scheid A, Spielmann P, Breymann C, Zimmermann R, Meuli M, future analysis using intracranial MCA specimens have the Gassmann M, Marti HH, Wenger RH: Oxygen-regulated expression of TGF- possibility to clarify the etiology of MMD. beta 3, a growth factor involved in trophoblast differentiation. Placenta 24:941–950, 2003. 22. Scheid A, Wenger RH, Schaffer L, Camenisch I, Distler O, Ferenc A, Cristina REFERENCES H, Ryan HE, Johnson RS, Wagner KF, Stauffer UG, Bauer C, Gassmann M, Meuli M: Physiologically low oxygen concentrations in fetal skin regulate 1. Attisano L, Wrana JL: Smads as transcriptional co-modulators. Curr Opin hypoxia-inducible factor 1 and transforming growth factor-beta3. FASEB J Cell Biol 12:235–243, 2000. 16:411–413, 2002. 2. Aoyagi M, Fukai N, Sakamoto H, Shinkai T, Matsushima Y, Yamamoto M, 23. Schultz K, Fanburg BL, Beasley D: Hypoxia and hypoxia-inducible factor- Yamamoto K: Altered cellular responses to serum mitogens, including 1alpha promote growth factor-induced proliferation of human vascular smooth platelet-derived growth factor, in cultured smooth muscle cells derived from muscle cells. Am J Physiol Heart Circ Physiol 290:H2528–H2534, 2006. arteries of patients with moyamoya disease. J Cell Physiol 147:191–198, 1991. 24. Scott RM, Smith JL, Robertson RL, Madsen JR, Soriano SG, Rockoff MA: 3. Aoyagi M, Fukai N, Yamamoto M, Matsushima Y, Yamamoto K: Development Long-term outcome in children with moyamoya syndrome after cranial of intimal thickening in superficial temporal arteries in patients with moy- revascularization by pial synangiosis. J Neurosurg 100 [Pediatrics Suppl 2]: amoya disease. Clin Neurol Neurosurg 99 [Suppl 2]:S213–S217, 1997. 142–149, 2004. 4. Aoyagi M, Fukai N, Yamamoto M, Nakagawa K, Matsushima Y, Yamamoto K: Early development of intimal thickening in superficial temporal arteries in 25. Shih SC, Claffey KP: Role of AP-1 and HIF-1 transcription factors in TGF-beta patients with moyamoya disease. Stroke 27:1750–1754, 1996. activation of VEGF expression. Growth Factors 19:19–34, 2001. 5. Bos R, van Diest PJ, de Jong JS, van der Groep P, van der Valk P, van der Wall 26. Suzui H, Hoshimaru M, Takahashi JA, Kikuchi H, Fukumoto M, Ohta M, Itoh E: Hypoxia-inducible factor-1alpha is associated with angiogenesis, and N, Hatanaka M: Immunohistochemical reactions for fibroblast growth factor expression of bFGF, PDGF-BB, and EGFR in invasive breast cancer. receptor in arteries of patients with moyamoya disease. Neurosurgery Histopathology 46:31–36, 2005. 35:20–25, 1994. 6. Bourdeau A, Cymerman U, Paquet ME, Meschino W, McKinnon WC, 27. Sure U, Freman S, Bozinov O, Benes L, Siegel AM, Bertalanffy H: Biological Guttmacher AE, Becker L, Letarte M: Endoglin expression is reduced in normal activity of adult cavernous malformations: A study of 56 patients. J Neurosurg vessels but still detectable in arteriovenous malformations of patients with 102:342–347, 2005. hereditary hemorrhagic telangiectasia type 1. Am J Pathol 156:911–923, 2000. 28. Soriano SG, Cowan DB, Proctor MR, Scott RM: Levels of soluble adhesion 7. Grisanti S, Canbek S, Kaiserling E, Adam A, Lafaut B, Gelisken F, Szurman P, molecules are elevated in the cerebrospinal fluid of children with moyamoya Henke-Fahle S, Oficjalska-Mlynczak J, Bartz-Schmidt KU: Expression of syndrome. Neurosurgery 50:544–549, 2002. endoglin in choroidal neovascularization. Exp Eye Res 78:207–213, 2004. 29. Suzuki J, Kodama N: Moyamoya disease–a review. Stroke 14:104–109, 1983. 8. Hojo M, Hoshimaru M, Miyamoto S, Taki W, Nagata I, Asahi M, Matsuura N, 30. Tacchini L, De Ponti C, Matteucci E, Follis R, Desiderio MA: Hepatocyte Ishizaki R, Kikuchi H, Hashimoto N: Role of transforming growth factor-beta1 growth factor-activated NF-kappaB regulates HIF-1 activity and ODC expres- in the pathogenesis of moyamoya disease. J Neurosurg 89:623–629, 1998. sion, implicated in survival, differently in different carcinoma cell lines. 9. Holifield JS, Arlen AM, Runyan RB, Tomanek RJ: TGF-beta1, -beta2 and Carcinogenesis 25:2089–2100, 2004. -beta3 cooperate to facilitate tubulogenesis in the explanted quail heart. J Vasc 31. Takagi Y, Hashimoto N, Goto Y: Haemodynamic ischaemia in paediatric moy- Res 41:491–498, 2004. amoya disease associated with renovascular hypertension. Acta Neurochir 10. Hoshimaru M, Takahashi JA, Kikuchi H, Nagata I, Hatanaka M: Possible (Wien) 139:257–258, 1997. roles of basic fibroblast growth factor in the pathogenesis of moyamoya dis- 32. Takahashi A, Sawamura Y, Houkin K, Kamiyama H, Abe H: The cere- ease: An immunohistochemical study. J Neurosurg 75:267–270, 1991. brospinal fluid in patients with moyamoya disease (spontaneous occlusion of 11. Hosoda Y, Ikeda E, Hirose S: Histopathological studies on spontaneous occlu- the circle of Willis) contains high level of basic fibroblast growth factor. sion of the circle of Willis (cerebrovascular moyamoya disease). Clin Neurol Neurosci Lett 160:214–216, 1993. Neurosurg 99 [Suppl 2]:S203–S208, 1997. 33. Takekawa Y, Umezawa T, Ueno Y, Sawada T, Kobayashi M: Pathological and 12. Houkin K, Kuroda S, Ishikawa T, Abe H: Neovascularization (angiogenesis) immunohistochemical findings of an autopsy case of adult moyamoya dis- after revascularization in moyamoya disease. Which technique is most useful ease. Neuropathology 24:236–242, 2004. for moyamoya disease? Acta Neurochir (Wien) 142:269–276, 2000. 34. Touho H: A simple surgical technique of direct anastomosis for treatment of 13. Ikeda E, Maruyama I, Hosoda Y: Expression of thrombomodulin in patients moyamoya disease: Technical note. Surg Neurol 62:366–368, 2004. with spontaneous occlusion of the circle of Willis. Stroke 24:657–660, 1993. 35. Toyoda H, Nishizawa S, Shiozaki T, Ueno M, Konishi J: A simplified double- 14. Karasawa J, Kikuchi H, Furuse S, Kawamura J, Sakaki T: Treatment of moya- injection method to quantify cerebral blood flow and vascular reserve using moya disease with STA-MCA anastomosis. J Neurosurg 49:679–688, 1978. iodine-123 IMP-SPECT. Ann Nucl Med 16:127–135, 2002. 15. Lario S, Mendes D, Bescos M, Inigo P, Campos B, Alvarez R, Alcaraz A, 36. Yamamoto M, Aoyagi M, Fukai N, Matsushima Y, Yamamoto K: Increase in Rivera-Fillat F, Campistol JM: Expression of transforming growth factor-beta1 prostaglandin E(2) production by interleukin-1beta in arterial smooth muscle and hypoxia-inducible factor-1alpha in an experimental model of kidney cells derived from patients with moyamoya disease. Circ Res 85:912–918, transplantation. Transplantation 75:1647–1654, 2003. 1999.

344 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HYPOXIA-INDUCING FACTOR-1␣ AND ENDOGLIN IN MOYAMOYA INTIMA

37. Yamamoto M, Aoyagi M, Tajima S, Wachi H, Fukai N, Matsushima Y, serum levels of the angiogenic proteins VEGF, basic fibroblast growth Yamamoto K: Increase in elastin gene expression and protein synthesis in factor, and metalloproteinase were significantly elevated in patients arterial smooth muscle cells derived from patients with moyamoya disease. with MMD. There could be several reasons for that discrepancy. First, Stroke 28:1733–1738, 1997. it is conceivable that HIF-1 upregulates the expression of soluble VEGF. This diffusible signal induces neovascularization in areas at risk. Secondly, it is possible that VEGF is not expressed in well developed COMMENTS and stable MMD vessels. The same would be true for more distal MCA branches typically harvested for direct vascularization. Soluble VEGF, athology of moyamoya disease (MMD) shows intimal thickening however, could be responsible for intimal hyperplasia in the intracra- owing to cellular and extracellular matrix proliferation. The inter- P nial circulation of MMD patients. The authors also explored an alterna- nal elastic lamina is fragmented and seems to be separated into layers, tive pathway. Endoglin is a component of TGF-β and was first there may be lipid deposits in the intima, and the tunica media smooth described as a candidate gene for hereditary hemorrhagic teleangiecta- muscle is decreased. These changes are observed at the end stage of the sia. It is known to be important in vascular development and remod- disease. Of course, they must be the cause of the angiographic narrow- eling. ing observed, but the exact makeup of the changes over time and their The authors are to be congratulated for this extremely novel histo- etiology are unknown. Studies such as this provide only very small logical and immunocytochemical study on intracranial arterial samples pieces of the puzzle. Whether or not the changes observed are a cause from MMD patients undergoing surgery. This is an important contribu- of the disease or consequences observed at the end stages cannot be tion to the literature, and further studies should be pursued to obtain determined from data such as this. This is a difficult problem for dis- better insight into molecular mechanisms underlying this debilitating eases such as MMD because there is no animal model, patients are disease. rare, and it is impossible to obtain biochemical and molecular information over the course of the disease. Some of the methods could be Raphael Guzman improved by using additional immunohistochemical controls and Gary K. Steinberg stereological counting methods, but this is a waste of time because it Stanford, California will not change the results or add any important new information beyond that already presented. his study aimed to clarify the pathogenesis of MMD through an immunohistochemical analysis of human MCA samples. The R. Loch Macdonald T authors compared 12 M4 samples obtained at the time of superficial Chicago, Illinois temporal artery-MCA bypass in MMD patients with those obtained from patients with other causes of MCA or internal carotid occlusion. akagi et al. studied potential pathophysiological events leading to They report that the MMD samples showed a thicker intima, higher intimal hyperplasia of the middle cerebral artery (MCA) of patients T staining for HIF-α in the intima and endothelium, and higher staining with MMD. The authors immunohistochemically analyzed MCA spec- for endoglin in the endothelium compared with the controls. imens in 12 MMD patients undergoing direct revascularization for the Although the results are interesting and add to the literature, their expression of hypoxia-inducible factor-1 HIF-1α, endoglin, vascular use in elucidating the mechanism of MMD is limited, owing in part to endothelial growth factor (VEGF), and transforming growth factor-β methodological problems with the study. First, the control patients rep- and compared these with MCA samples from 12 control subjects. They resent a markedly different population than the MMD patients in that found that MCA specimens from MMD patients had marked intimal they are an average 18 years older and most of them are the opposite hyperplasia compared with those from the control patients. Endothelial sex. It is possible that these disparities alone could have led to the dif- expression of HIF-1α and endoglin was higher in samples from MMD ferences in staining. Furthermore, the methods section mentions that patients and the immunoreactive cells were colocalized with trans- control samples in some experiments were obtained from a 27-year forming growth factor-β. There was, however, no VEGF immunoreac- old woman with an arteriovenous malformation. However, this patient tivity in samples from MMD patients. They conclude that intimal is not accounted for in the summary of cases reported in Table 1. hyperplasia in MMD patients occurred through the HIF-1α and Secondly, the samples analyzed were from the fourth portion of the endoglin pathway. MCA. Although obtaining more proximal portions is impractical in HIF-1 is a well-known heterodimeric transcriptional factor consisting humans, these are the portions that are classically implicated in the of an inducible HIF-1α subunit and a constitutive HIF-1β subunit. It pathogenesis of MMD. Extrapolating M4 findings to more proximal has been shown that HIF-1α expression is exponentially induced as the segments, and thus to the mechanism of the disease, may not be valid. oxygen concentration of cells decreases. HIF-1 targets genes leading to The authors should be commended for their work on human tissue, the transcriptional activation of several dozen genes such as erythro- an endeavor which is often fraught with challenges. The results of this poietin, glucose transporter 1, glycolytic enzymes (aldolase A and eno- descriptive study, however, are circumstantial suggestions that do not lase 1), and VEGF. Recent experimental work has shown that intimal definitively implicate HIF-1α or endoglin as players in the pathogene- hyperplasia can be induced by VEGF, and it is known that VEGF is a sis of MMD. We hope this initial work encourages this group and oth- potent angiogenic factor. In rat models of chronic cerebral hypoperfu- ers to further evaluate the specific pathways that may be involved in sion, administration of VEGF combined with vasoreconstructive sur- the mechanism of this disease. gery significantly increased capillary density in the brain. In the current report, the investigators did not find expression of VEGF despite ele- Lance S. Governale vated HIF-1. A recent neuropathological case report also failed to Robert M. Friedlander demonstrate VEGF immunoreactivity in MMD vessels. We found that Boston, Massachusetts

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 345 EXPERIMENTAL STUDIES

MULTIPLE DIFFERENTIATION POTENTIALS OF NEONATAL DURA MATER-DERIVED CELLS

Ioana A. Peptan, D.D.S., M.S. OBJECTIVE: The involvement of the dura mater in calvarial development and bone Department of Bioengineering, healing lead to a hypothesis that progenitor cells with multiple differentiation potentials University of Illinois at Chicago, exist within this tissue. The present study investigated the differentiation potentials of dura Chicago, Illinois mater-derived cells by driving them into several cell-restricted lineages. Liu Hong, M.D., Ph.D. METHODS: Dissected dura mater tissue of neonatal rats was washed, ﬁnely minced, and Departments of Orthodontics enzymatically digested. The harvested cells were exposed to different differentiation and Bioengineering, (osteogenic, adipogenic, and chondrogenic) and basic media. University of Illinois at Chicago, Chicago, Illinois RESULTS: At deﬁned time points, dura mater-derived cells were observed to differentiate into osteoblastic, adipoblastic, and chondroblastic cells, evidenced by specific Carla A. Evans, D.D.S., D.M.Sc. biochemical staining. In addition, gene expressions of osteogenesis (alkaline phos- Department of Orthodontics, phatase, osteocalcin, and osteopontin), chondrogenesis (collagen Type II and aggre- University of Illinois at Chicago, can core protein) and adipogenesis (peroxisome proliferator activated receptor γ-2) Chicago, Illinois were up-regulated in the differentiated dura mater-derived cells, confirmed by poly- Reprint requests: merase chain reaction. Liu Hong, M.D., Ph.D., CONCLUSION: Preliminarily, it was concluded that a subpopulation of multiple poten- Department of Orthodontics, Stem Cell and Craniofacial tial mesenchymal cells exists in neonatal dura mater, which explains the function of Tissue Engineering Laboratory, the dura mater on neurocranium development and calvarial bone healing. University of Illinois at Chicago, 801 South Paulina Street, MC 841, KEY WORDS: Differentiation potential, Dura mater-derived cell, In vitro, Mesenchymal Chicago, IL 60612-7211. Email: [email protected] Neurosurgery 60:346–352, 2007 DOI: 10.1227/01.NEU.0000249278.72063.59 www.neurosurgery-online.com

Received, March 29, 2006. Accepted, October 5, 2006. ranial bone formation occurs either by paracrine signaling (23). Anomalies in neuro- intramembranous or by endochondral cranium formation are often associated with Cprocesses involving mesenchymal pro- congenital malformations involving central genitor cell differentiation, synthesis, and min- nervous system development (6). Moreover, the eralization of the extracellular matrix. Cranial osteogenic properties of the dura are thought to vault formation takes place through intramem- contribute to the regenerative capacity of neu- branous ossiﬁcation of the surface mesenchyme rocranium bone defects. The capacity for osteo- surrounding the developing brain, which is conduction and osteoinduction is unique to presumably achieved by direct transformation immature or young dura mater. Subtotal of mesenchymal stem cells into osteoblasts. The calveriectomy with preservation of the dura dura mater interposing between the cerebral mater in children younger than 2 years of age hemispheres and calvarium plays an important and immature animals is associated with com- role in calvarial morphogenesis and bone heal- plete bone regeneration (7, 11). It is assumed ing. The functions of the dura mater differ that the bone defects are repaired by osteochon- depending on the anatomic sites. For example, droblastic differentiation of the dura cells, by cranial suture-associated dura has a ruling the local action of dura-derived growth factors function on suture growth and maintenance, on mesenchymal cells, or both (16). whereas squamous dura mater participates in The complex role of the dura mater in cal- neurocranium development and bone healing. varial bone formation, bone healing, and cra- Osteoblasts cocultured with sutural dura can nial suture function leads to the assumption of affect osteoblast differentiation, suggesting that the existence of mesenchymal progenitor cells the regional dura mater underlying cranial within this tissue. We hypothesized that a sub- sutures regulates cranial suture fate through population of mesenchymal cells within the

346 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com NEONATAL DURA MATER-DERIVED CELLS

dura mater can be differentiated into several cell lineages. To Osteogenic differentiation medium consisted of basic medium this end, cells isolated from neonatal rat dura were used to supplemented with 100 nmol/L dexamethasone, 10 mmol/L identify their various differentiation potentials under specific β-glycerophosphate, and 0.05 mmol/L ascorbic acid-2-phosphate, stimulations. which has been used extensively to differentiate mesenchymal stem cells into osteoblasts (2, 18). Cells cultured in basic medium MATERIALS AND METHODS served as controls. At each defined time point, cultured cells were collected and subjected to analysis of cell proliferation and Cell Isolation and Culture osteogenic differentiation. Osteogenic gene expressions were All animals used in the present study were under the super- identified by polymerase chain reaction (PCR) analysis. vision of the University of Illinois at Chicago’s Animal Care Deoxyribonucleic Acid Measurement Committee. Dura mater cells were isolated from approximately 80 2-day-old, neonatal Sprague-Dawley rats. After sacrifice, the On Days 7, 14, and 21, cells cultured in six-well plates from calvarium of each rat was removed, and the squamous dura both osteoinductive and control groups were washed twice with mater that covers only the endocranial surface of the calvarium PBS and then suspended and lysed in 1% Triton-X100 solution was collected. All harvesting tissue procedures were performed (Sigma Chemical Co., St. Louis, MO). The collected cells subse- under the dissecting microscope. Squamous dura mater tissue quently were homogenized using sonication (Dismembrator from all rats was washed three times in phosphate-buffered Model 100; Fisher Scientific, Pittsburgh, PA). The deoxyribonu- saline (PBS) to remove contaminated tissues and blood. Finely cleic acid (DNA) content in the cells was determined by a fluo- minced tissues were digested with 0.075 wt% collagenase Type rometric assay using a spectrofluorometer and a DNA quantifi- I (Worthington, Lakewood, NJ) for 45 minutes at 37ЊC, with cation kit (Hoechst 33258; BioRad, Hercules, CA). The intermittent shaking. The concentration of collagenase used to fluorescent optical density of each sample was measured at an collect adipose-derived cells was duplicated from previous excitation wavelength of 360 nmol/L and an emission wave- studies (25). The digested tissue solution was then neutralized length of 460 nmol/L. The amount of DNA in each sample was and filtered through a cell strainer with a pore size of 70 µm to determined by using a prepared standard curve (19). DNA remove undigested tissues. The collected cells were cultured in results were expressed as micrograms per milliliter. basic cell culture medium consisting of Dulbecco’s Modified Eagle’s Medium (GibcoBRL, Carlsbad, CA) supplemented with Assessment of Alkaline Phosphatase Activity 10% fetal bovine serum and 1% antibiotic/antimycotic in a 37ЊC and Mineralization incubator with 5% CO2. Culture medium was exchanged every The cultured cells were fixed in 10% paraformaldehyde and third day. After reaching 70 to 80% confluence, cells were disso- incubated with 120 mmol/L Tris buffer (pH 8.4) containing 0.9 ciated with 0.25 wt% trypsin/ethylenediamine tetra-acetic acid mmol/L Naphtol AS-MX Phosphate (Sigma Chemical Co.) and (GibcoBRL) and subcultured for cell differentiation studies. 1.8 mmol/L Fast Red TR (Sigma Chemical Co.). Naphtol AS-MX Phosphate was solubilized with N,N-dimethylformamide Osteogenesis (Fisher Scientific) before dilution with Tris buffer. After 45 min- First-passage dura mater-derived cells (DMDCs) were subcul- utes at room temperature, the cultures were washed with deion- tured in six-well plates at a density of 5 105 cells/well and ized water. Mineral deposition of extracellular matrix was exposed to osteogenic differentiation medium for up to 3 weeks. reflected by the presence of black nodules using von Kossa

TABLE 1. Reverse transcriptase polymerase chain reaction primers for osteogenic, adipogenic, and chondrogenic differentiation markers Gene Sequence Osteopontin (OP) Sense: CCTCCTGTCTCCCGGTGAAA Antisense: AAACTCGTGGCTCTGATGTT Osteocalcin (OC) Sense: AGGACCCTCTCTCTGCTCAC Antisense: AACGGTGGTGCCATAGATGC Alkaline phosphatase (ALP) Sense: TCCATGGTGGATTATGCTCA Antisense: TTCTGTTCCTGCTCGAGGTT Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) Sense: TGAACGGGAAGCTCACTGG Antisense: TCCACCACCCTGTTGCTGTA Peroxisome proliferator activated receptor -2 (PPAR -2) Sense: AAACTCTGGGAGATCCTCCT Antisense: TCTTGTGAACGGGATGTCTT Aggregan core protein Sense: AGGATGGCTTCCACCAGTGT Antisense: CATAAAAGACCTCACCCTCC Collagen Type II Sense: CTCAAGTCGCTGAACAACC Antisense: CTATGTCCACACCAAATTCC

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 347 PEPTAN ET AL.

AB10 mmol/L of insulin, and 5 mmol/L of isobutyl-methylxan- thine for 1 week (8). The medium was exchanged three times a week. After 1 week, the adipogenic differentiated cells and control group cells were evaluated by biochemical and molecular analysis. Oil-O-Red Staining The cells were fixed in 10% CDparaformaldehyde at room temperature. Then, the cells were incubated in 2% (w/v) Oil-O-Red reagent (Sigma Chemical Co., St. Louis, MO) for 7 minutes. The excess stain was removed with deionized water followed by counter- staining with hematoxylin. Intracellular lipid accumulation, a marker of adipogenesis, stained in red. Gene expression EFof peroxisome proliferator- activated receptor (PPAR) γ-2 was investigated to further support adipogenesis. Chondrogenesis Chondrogenic differentiation was induced using the high-density “micromass” culture technique. Ten-µL drops of cellular suspension (5 107 cells/ml) were placed in petri dishes and incubated for FIGURE 1. Osteogenesis of fibroblast-like cells isolated from dura mater tissue. Under osteogenic stimulation, dura mater Њ cells showed intense positive reaction to ALP and von Kossa staining at 1 (A), 2 (B), and 3 (C, arrows) weeks. The con- 2 hours at 37 C and 5% CO2. trol group cells exposed to basic medium showed a mild positive reaction to ALP and a negative reaction to von Kossa The dishes were then filled staining at 1 (D), 2 (E, arrows), and 3 (F) weeks. Original magnification, 10. with basic medium and incubated overnight. The next day, staining. The fixed cells were incubated in 2.5% (w/v) silver chondrogenic supplements consisting of 10 ng/ml transform- nitrate (Sigma Chemical Co.) solution for 30 minutes in sunlight. ing growth factor-β1, 6.25 µg/ml insulin, 6.25 µg/ml transfer- The excess silver nitrate was washed away with deionized rine, and 0.1 µmol/L dexamethasone were added to basic water. To quantify alkaline phosphatase (ALP) activity and cal- medium (4, 21). On defined days, cell masses were collected cium content, cultured cells were washed twice with PBS and for histological evaluation. Gene expressions of collagen Type then suspended and lysed in 0.5 ml of 1% Triton-X100 solution. II and aggregan core protein were used to evaluate chondroge- The collected cells were subsequently homogenized using son- nesis of DMDCs. Total ribonucleic acid (RNA) was extracted ication. ALP and calcium content of solution were measured by from cells subcultured for 3 weeks in chondrogenic and basic spectrophotometry with a colorimetric kit, according to the medium using TRIzol reagent (GibcoBRL). PCR was used to manufacturer’s instructions (Sigma Chemical Co.). identify chondrogenic gene expression. Adipogenesis Alcian Blue Staining The first passage DMDCs were subcultured at a density of Alcian blue staining is specific for visualizing sulfated pro- 20,000 cells/cm2 in adipogenic medium that consists of basic teoglycans during chondrogenesis (16). Before staining, the cel- medium supplemented with 50 nmol/L of dexamethasone, lular masses were washed twice with PBS and fixed in 10%

348 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com NEONATAL DURA MATER-DERIVED CELLS

A Statistical Analysis All quantitative data were expressed as mean standard deviation. Student’s t tests were used to compare the extent of osteogenic differentiation between treatment and control groups. P values less than 0.05 were considered signiﬁcant.

RESULTS B Osteogenic Potentials Fibroblast-like cells were isolated from neonatal rat dura mater tissue. DMDCs were proliferated by in vitro culture in basic medium. After exposure to osteogenic differentiation medium, DMDCs formed an extensive network of dense, multilayered nodules with high ALP activities and min- FIGURE 2. Quantitative assessment of osteogenic differentiation potential. The proliferation rate of DMDCs exposed to eral deposition. ALP activity osteogenic medium decreased with time compared with that of the control group (A). In the osteogenic differentiation and mineralization content group, ALP activity and calcium content (Ca) increased during a 3-week period (B) and were significantly greater in the increased as experimental osteogenic differentiation group than the control group (B, C). PCR detection of messenger RNA for osteocalcin (OC), times were extended (Fig. 1, ALP, osteopontin (OP), and GAPDH was performed after 3 weeks of differentiation. The PCR products were analyzed A–C). In contrast, the control on an ethidium bromide-stained agarose gel (D). Asterisk, P 0.05; double asterisk, P 0.01. group cells showed mild positive reaction to ALP staining paraformaldehyde. The samples were embedded in paraffin but no mineralization (Fig. 1, D–F). Quantitative assessment of and cut into 5-µm sections. The slides were incubated for cell proliferation revealed that the proliferation rate of DMDCs 30 minutes with 1% (w/v) Alcian blue (Sigma Chemical Co.) in slowed after exposure to osteogenic differentiation medium, 0.1 N HCl (pH 1.0) and washed with 0.1 N HCl for 5 minutes as compared with control cells exposed to basic medium (Fig. to remove excess stain. 2A). ALP activity and calcium content of DMDCs exposed to osteogenic differentiation increased with time and were signif- PCR Analyses icantly higher than controls (Fig. 2, B and C). Osteogenic gene Total RNA was extracted from subcultured cells using TRIzol expression revealed that, although ALP and osteopontin were reagent according to the manufacturer’s protocol. Approxi- expressed in control and treatment groups, osteocalcin was mately 5 µg of total RNA was converted to single-stranded com- only expressed in the osteogenic differentiation group (Fig. 2D). plementary DNA (cDNA) using a commercial cDNA synthesis kit (GIBCO Life Sciences). Aliquots of the cDNA were amplified Adipogenic Potentials with Ampli-Taq DNA polymerase (PerkinElmer, Norwalk, CT) After 1 week of incubation with adipogenic differentiation for ALP, osteocalcin, osteopontin, collagen Type II, aggrecan medium, a few cells morphologically resembled adipocytes, core protein, and PPARγ-2 genes. Thirty-five cycles were used exhibiting a rounded appearance. Under a higher magnification for all genes, each consisting of 45 seconds of denaturation at of inverted microscopy, intracellular lipid vesicles were observed 95ЊC, 45 seconds of annealing at 60ЊC, and 1 minute of polymer- (Fig. 3A). A positive reaction of Oil-O-Red staining demonstrated ization at 72ЊC, followed by a final 10 minutes extension at 72ЊC. accumulations of lipid droplets in monolayer-cultured DMDCs The housekeeping gene, glyceraldehyde-3-phosphate dehydrogenase under adipogenic differentiation (Fig. 3B). In contrast, control (GAPDH) was used as a control for RNA loading samples. PCR cells exposed to the nondifferentiation medium retained their products were analyzed electrophoretically with an Agilent 2100 fibroblast-like spindle shapes, and no positive reaction to Oil-O- bioanalyzer (Agilent Technologies, Palo Alto, CA) on 2.0% Red staining was found (Fig. 3C). PPARγ-2 gene, a specific adi- agarose gel and visualized by staining with ethidium bromide. pogenic marker, was expressed at 1 week of adipogenic differen- The primer sequences used in this study for PCR analysis are tiation in the treatment group. No band was observed in the summarized in Table 1. control group exposed to basic medium (Fig. 3D).

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 349 PEPTAN ET AL.

A A

FIGURE 4. Under chondrogenic stimulation, DMDCs formed spherical pellets that stained positively with Alcian blue staining (A, arrows). PCR detection of mRNAs for collagen Type II, aggregan core protein, and GAPDH was performed after 3 weeks of differentiation. The PCR products were analyzed on an ethidium bromide-stained agarose gel (B).

C Chondrogenic Potentials DMDCs cultured under micromass conditions formed cell pellets associated with a well-organized extracellular matrix rich in sulfated proteoglycans and collagen Type II. Up to 3 weeks, the DMDCs pellets were positive for Alcian blue staining, indicating the presence of sulfated proteoglycans (Fig. 4A). Collagen Type II and aggregan core protein, genes of speciﬁc chondrogenic markers, were up-regulated after 3 weeks of differentiation (Fig. 4B). The control group failed to express any chondrogenic markers (Fig. 4B).

DISCUSSION

Cell lineage studies using embryo cultures or short-term labeling techniques (β-galactosidase, dioctadecylindocar- bocyanine, and X-gal staining) elucidated neural crest cell migration pathways (10, 17, 20). These studies demonstrated that meninges (including dura mater) originate from neural FIGURE 3. DMDCs exposed to adipogenic medium for 1 week demonstrated crest cells. During craniofacial development, neural crest cells intracellular lipid accumulation (A), which stained positively with Oil-O-Red migrate ventrolaterally as they populate the branchial arches. (B). Cells cultured in basic medium lacked intracellular lipid accumulation These ectodermally derived cells are multipotent stem cells (C). Original magniﬁcation, 40. PCR detection of mRNAs for PPARγ-2 and GAPDH was performed after 1 week of differentiation. The PCR products that contribute signiﬁcantly to the formation of mesenchymal were analyzed on an ethidium bromide-stained agarose gel (D). structures in the head and neck (1, 14). In this study, we examined the multilineage potential of a putative ectodermal cell population obtained from rat neonatal dura mater. After dissimilar differentiation stimulations, osteogenic, adipogenic, and chondrogenic potentials of the cells were demonstrated by his-

350 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com NEONATAL DURA MATER-DERIVED CELLS

tological and biochemical techniques. Although the capacity of nary understanding of the mechanism of function of dura cell dura mater tissue to form bone and cartilage ectopically was phenotypes in this study is useful in further investigations for previously reported (24), this study is the first one to our craniosynostosis treatment and craniofacial bone healing. To knowledge to identify mesenchymal cells with multiple differ- further confirm DMDC characteristics as those of mesenchy- entiation potentials within neonatal squamous dura mater. It mal stem cells, self-renewing capability of the DMDCs and preliminarily explains the function of the dura mater on neuro- differentiation potentials of DMDCs to other mesodermal lin- cranium development and calvarial bone healing. eages (myogenesis) must be identified. Adult dura is demon- The dura mater has been demonstrated to play a critical role strated to have reduced bone formation capabilities as com- in calvarial morphogenesis. Early embryonic skeletogenesis pared with young dura (7). The mechanism might be requires intimate interactions between surface epithelium and answered by quantitative comparison studies of multiple the dura mater. Mehrara et al. (12) showed that the dura mater potential cell populations among different ages. In addition, underlying the developing calvarial bone strongly expressed the role of these pluripotent cells in neonatal dura need to be β transforming growth factor- 1 and fibroblast growth factor-2 further characterized in the future. messenger RNA (mRNA) from embryonic until neonatal age, and that was attributed to cranial vault bone formation. REFERENCES Moreover, they found that calvarial osteoblasts located near the endocranial surface and in contact with the developing 1. Bronner-Fraser M: Rostrocaudal differences within the somites confer seg- dura were strongly stained for transforming growth factor-β1 mental pattern to trunk neural crest migration. Curr Top Dev Biol 47:279–296, and fibroblast growth factor-2. The dura mater has also been 2000. shown to manipulate calvarial bone healing. Uddstromer and 2. Cheng SL, Yang JW, Rifas L, Zhang SF, Avioli LV: Differentiation of human Ritsila (22) showed that neonatal dura posses a stronger bone marrow osteogenic stromal cells in vitro: Induction of the osteoblast phenotype by dexamethasone. Endocrinology 134:277–286, 1994. osteogenic capacity than periosteum during the healing of 3. De Bari C, Dell’Accio F, Tylzanowski P, Luyten FP: Multipotent mesenchymal membraneous cranial bone defects. Warren et al. (23) demon- stem cells from adult human synovial membrane. Arthritis Rheum strated that the underlying dura of coronal and sagittal sutures 44:1928–1942, 2001. can affect osteoblast differentiation, indicating that regional 4. Denker AE, Nicoll SB, Tuan RS: Formation of cartilage-like spheroids by dura mater regulates the fate of cranial sutures through micromass cultures of murine C3H10T1/2 cells upon treatment with transforming growth factor-beta 1. Differentiation 59:25–34, 1995. paracrine signaling. In addition, Yu et al. (24) found that differ- 5. De Ugarte DA, Morizono K, Elbarbary A, Alfonso Z, Zuk PA, Zhu M, Dragoo ent regional dura have different functions and capabilities of JL, Ashjian P, Thomas B, Benhaim P, Chen I, Fraser J, Hedrick MH: differentiation, indicating that squamous dura might contain Comparison of multi-lineage cells from human adipose tissue and bone mar- different cell phenotypes from sutural dura. This suggests that row. Cells Tissues Organs 174:101–109, 2003. the dura mater of young animals can turn into bone and carti- 6. Garol JD, Fields HW Jr, Metzner L, Kokich VG: The craniofacial skeleton in anencephalic human fetuses. II. Calvarium. Teratology 17:67–73, 1978. lage formation when transplanted in epithelial pouches. The 7. Hobar PC, Schreiber JS, McCarthy JG, Thomas PA: The role of the dura in cra- present study demonstrates that DMDCs possess multiple dif- nial bone regeneration in the immature animal. Plast Reconstr Surg 92: ferentiation potentials under specific differentiation conditions, 405–410, 1993. which are similar to those of mesenchymal stem cells. 8. Hong L, Peptan I, Clark P, Mao JJ: Ex vivo adipose tissue engineering by The existence of mesenchymal stem cells with multiple dif- human marrow stromal cell seeded gelatin sponge. Ann Biomed Eng ferentiation potentials has been demonstrated in bone mar- 33:511–517, 2005. 9. Jankowski RJ, Deasy BM, Huard J: Muscle-derived stem cells. Gene Ther row, trabecular bone, adipose tissue, synovium, skeletal mus- 9:642–647, 2002. cle, and dental pulp (3, 5, 9, 13, 15, 25). They act as reservoirs 10. Jiang X, Iseki S, Maxson RE, Sucov HM, Morriss-Kay GM: Tissue origins and of reparative cells, ready to differentiate and mobilize in interactions in the mammalian skull vault. Dev Biol 241:106–116, 2002. response to growth and wound signals or pathological condi- 11. Mabbutt LW, Kokich VG: Calvarial and sutural re-development following tions. However, the population of these progenitor cells craniectomy in the neonatal rabbit. J Anat 129:413–422, 1979. 12. Mehrara BJ, Most D, Chang J, Bresnick S, Turk A, Schendel SA, Gittes GK, reduces with age. Similarly, adult DMDCs were found in a Longaker MT: Basic fibroblast growth factor and transforming growth factor parallel study of our laboratory to lose the characteristics of beta-1 expression in the developing dura mater correlates with calvarial bone multiple differentiation capabilities but maintain their formation. Plast Reconstr Surg 104:435–444, 1999. osteogenic potentials (data not shown) There are not available 13. Miura M, Gronthos S, Zhao M, Lu B, Fisher LW, Robey PG, Shi S: SHED: Stem specific markers for the mesenchymal stem cells. The basic cells from human exfoliated deciduous teeth. Proc Natl Acad Sci U S A criteria for identifying stem cells are multiple differentiation 100:5807–5812, 2003. 14. Noden DM: Craniofacial development: New views on old problems. Anat potentials and the ability to self-renew. In the present study, Rec 208:1–13, 1984. neonatal DMDCs demonstrated osteogenic, chondrogenic, 15. Noth U, Osyczka AM, Tuli R, Hickok NJ, Danielson KG, Tuan RS: and adipogenic differentiation, as illustrated by the presence Multilineage mesenchymal differentiation potential of human trabecular of specific cell lineage markers: osteocalcin, collagen Type II, bone-derived cells. J Orthop Res 20:1060–1069, 2002. aggregan core protein, and PPARγ-2. The plasticity of the 16. Ogle RC, Tholpady SS, McGlynn KA, Ogle RA: Regulation of cranial suture morphogenesis. Cells Tissues Organs 176:54–66, 2004. commitment to different cell lineages under specific stimuli is 17. Osumi-Yamashita N, Ninomiya Y, Doi H, Eto K: The contribution of both a primordial feature of mesenchymal stem cells. Because of forebrain and midbrain crest cells to the mesenchyme in the frontonasal mass the role of dura in cranial suture development, the prelimi- of mouse embryos. Dev Biol 164:409–419, 1996.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 351 PEPTAN ET AL.

18. Pittenger MF, Mackay AM, Beck SC, Jaiswal RK, Douglas R, Mosca JD, of cells with multiple differentiation potential remain unclear. It may Moorman MA, Simonetti DW, Craig S, Marshak DR: Multilineage potential turn out that the existence of these cells plays a role in neurocranial of adult human mesenchymal stem cells. Science 284:143–147, 1999. development and calvarial bone healing, as the authors suggest. 19. Rengarajan K, Cristol SM, Mehta M, Nickerson JM: Quantifying DNA concen- However, there is no conclusive evidence of this in the present study. trations using fluorometry: A comparison of fluorophores. Mol Vis 8:416–421, Alternatively, these cells may not play any important role. In this study, 2002. as with others involving stem cells in other tissue, the gap between 20. Serbedzija GN, Bronner-Fraser M, Fraser SE: Vital dye analysis of cranial neural crest cell migration in the mouse embryo. Development 116:297–307, existence and functional role can be difficult to bridge. Nevertheless, 1992. the results of this study are interesting and the role of dura mater- 21. Tacchetti C, Tavella S, Dozin B, Quarto R, Robino G, Cancedda R: Cell con- derived cells deserves further study. densation in chondrogenic differentiation. Exp Cell Res 200:26–33, 1992. 22. Uddstromer L, Ritsila V: Healing of membranous and long bone defects. An Charles Y. Liu experimental study in growing rabbits. Scand J Plast Reconstr Surg 13: Los Angeles, California 281–287, 1979. 23. Warren SM, Greenwald JA, Nacamuli RP, Fong KD, Song HJ, Fang TD, Mathy JA, Longaker MT: Regional dura mater differentially regulates osteoblast his interesting study provides the reader with a background on dural gene expression. J Craniofac Surg 14:363–370, 2003. Tbiology and cranial development. There are some questions related 24. Yu JC, McClintock JS, Gannon F, Gao XX, Mobasser JP, Sharawy M: Regional to the authors’ methods, including the effects of differentiation agents or differences of dura osteoinduction: Squamous dura induces osteogenesis, collagenase. It is not likely that the dura will prove to be a source for cel- sutural dura induces chondrogenesis and osteogenesis. Plast Reconstr Surg lular repair. The clinical relevance of the work is somewhat questionable 100:23–31, 1997. and has potential ramifications for craniosynostosis surgery. 25. Zuk PA, Zhu M, Mizuno H, Huang J, Futrell JW, Katz AJ, Benhaim P, Lorenz HP, Hedrick MH: Multilineage cells from human adipose tissue: Implications Douglas Kondziolka for cell-based therapies. Tissue Eng 7:211–228, 2001. Pittsburgh, Pennsylvania

Acknowledgments he authors have demonstrated the existence of progenitor cells We thank the laboratory of Brenda Russell, Ph.D., (University of Illinois at within the neonatal rat squamous dura mater and have shown Chicago, Department of Physiology and Biophysics) for providing animal tissue. T their multiple differentiation potential using osteogenic, adipogenic, This research was supported by a Biomedical Engineering Research Grant from the Whitaker Foundation (RG-02–0701), the Cleft Palate Foundation, and the and chondrogenic stimulation in vitro. Interestingly, similar tissue from March of Dimes Foundation (RG-01–0075) (LH). adult rats is reported to lose this capacity and maintain only an osteogenic potential. To further evaluate the properties of these progenitor cells, it will be COMMENTS interesting to learn more about their potential in vivo, e.g., after allo- genic transplantation into adequate osteogenic, adipogenic, and chon- n this report, the authors provide evidence that neonatal dura mater- drogenic niches. Iderived cells have multiple differentiation potentials. They found that dura mater-derived cells can be driven into speciﬁc cell-restricted Håvard Ølstørn lineages by exposure to various differentiation media. Although these Iver A. Langmoen ﬁndings are important and interesting, the implications of the presence Oslo, Norway

Historic 12th round knockout sequence from the ﬁrst Joe Louis (1914–1981) (left)—Max Schmeling (1905–2005) bout, Yankee Stadium, New York, June 19, 1936. EXPERIMENTAL STUDIES

INTERACTION BETWEEN KRIT1 AND MALCAVERNIN: IMPLICATIONS FOR THE PATHOGENESIS OF CEREBRAL CAVERNOUS MALFORMATIONS

Jun Zhang, Ph.D. OBJECTIVE: Cerebral cavernous malformations (CCM) are a relatively common auto- Department of Neurological Surgery, somal dominant disorder leading to the formation of vascular malformations in the Johns Hopkins University nervous system. Mutations in krit1 and malcavernin, the proteins encoded by the genes School of Medicine, Baltimore, Maryland at the CCM1 and CCM2 loci, respectively, are responsible for the majority of CCMs. Similar to integrin cytoplasmic domain-associated protein-1α, a known krit1 interac- Daniele Rigamonti, M.D. tor, malcavernin is a phosphotyrosine binding protein. We report here that krit1 also Department of Neurological Surgery, interacts with malcavernin. Johns Hopkins University School of Medicine, METHODS: We used two-hybrid analysis, in vivo coimmunoprecipitation, and epitope Baltimore, Maryland mapping to explore the interaction between krit1 and malcavernin. Immunocytochemistry was used to study the cellular localization of these proteins. Harry C. Dietz, M.D. RESULTS: We demonstrate that malcavernin independently binds to two of the three Howard Hughes Medical Institute, and NPXY (asparagine, proline, undetermined/variable amino acid, and tyrosine) motifs in The Institute of Genetic Medicine, Johns Hopkins University, krit1. By immunocytochemistry, malcavernin protein is cytoplasmic at steady state, but School of Medicine, shuttles between the nucleus and cytoplasm, despite lacking either a nuclear localiza- Baltimore, Maryland tion signal or a nuclear export signal in its sequence. CONCLUSION: These data suggest that krit1 interacts with malcavernin through its Richard E. Clatterbuck, M.D., Ph.D. NPXY motifs and may shuttle it through the nucleus via its nuclear localization signal Department of Neurological Surgery, and nuclear export signals, thereby regulating its cellular function. Johns Hopkins University KEY WORDS: CCM1, CCM2, Cerebral cavernous malformations School of Medicine, Baltimore, Maryland Neurosurgery 60:353–359, 2007 DOI: 10.1227/01.NEU.0000249268.11074.83 www.neurosurgery-online.com

Reprint requests: Richard E. Clatterbuck, M.D., Ph.D., Hattiesburg Clinic, 415 South 28th Avenue, erebral cavernous malformations krit1 and malcavernin, respectively, both with Hattiesburg, MS 39401. (CCM) are found in the nervous sys- unknown functions (6, 7). The majority of Email: [email protected] Ctem in 0.5% of the population and rep- mutations lead to premature termination resent up to 10% of all vascular malforma- codons, likely leading to nonsense-mediated Received, September 16, 2005. tions (4, 8). Common complications include ribonucleic acid decay, making haploinsufﬁ- Accepted, October 5, 2006. headache, seizure, and hemorrhagic stroke. ciency or a ”somatic-second hit” model attrac- Little is known regarding the etiology and tive candidates for the pathogenic mechanisms pathogenesis of CCMs. Lesions are lined with (5–7, 10, 11, 14). endothelium but lack other structural ele- There are three asparagine, proline, undeter- ments of normal vessel walls, and loss of tight mined/variable amino acid, and tyrosine junctions between adjacent endothelial cells (NPXY) motifs in krit1 (Residues 191–194, suggests a loss of cellular communication and 231–234, and 250–253). We previously reported adhesion (2). that the ﬁrst NPXY motif (Residues 191–194) CCM is an autosomal dominant condition interacts with integrin cytoplasmic domain- characterized by the presence of multiple associated protein-1α (icap1α), a modulator of CCMs (9). It is a genetically heterogeneous dis- β1 integrin signal transduction, suggesting a order with at least three loci on chromosomes role for krit1 in β1 integrin-mediated angiogen- 7q (CCM1), 7p (CCM2), and 3q (CCM3) (3). esis (13, 15). We report here that krit1 interacts The genes corresponding to the CCM1 and with malcavernin through its second and third CCM2 loci have been defined and encode NPXY motifs (Residues 231–234 and 250–253).

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 353 ZHANG ET AL.

MATERIALS AND METHODS and KMUT7R (CTGGAGCTCCTAGACCAGCGTAT GGAGC- TATTCCCACTTTATCTAC) were used to create the single Plasmid Construction and Yeast Two-hybrid Analysis mutation, N250A, and the compound mutation, N250A and F253A, in the single expression construct, respectively. The All polymerase chain reaction (PCR) amplifications were per- double compound mutation N231A, F234A and N250A, F253A formed with Platinum Pfx DNA Polymerase (Invitrogen Corp., in a single expression construct was also subsequently cloned. Carlsbad, CA). Both GAL4 binding domain and activation domain fusion constructs were assembled by cloning PCR- Mutagenesis Assay amplified complementary deoxyribonucleic acid fragments into vectors pGBKT7 and pGADT7, respectively (Clontech pGADT7-malcavernin and each of the pGBKT7-krit1 mutants Laboratories, Inc., Mountain View, CA). PCR products using and wild-type construct were co-transformed and evaluated in a standard liquid phase β-galactosidase assay (Clontech malcavernin primers MGC1F (GATATGGAAGAG- β GAGGGCAAGAA) and MGC1R (GCTGAGTCCTGGTCC Laboratories, Inc.). -galactosidase activity reveals the degree of ATGCT) were amplified from human brain. The different frag- interaction between malcavernin and each form of krit1. Results ments of krit1 were created from pGADT7-Krit1 with primers, shown for each experiment reflect the performance of three inde- including KRIT02F (ATGGGAAATCCAGAAAACATAG) and pendent assays for each of three independent transformants. HK504R (GAGATTGTGCA TGACGTTCA) (krit1: 1–171); Tissue Culture and Transfections KRIT02F and K5-32R (CTAATGTAGTGAGTTTTCT GTCTGA) (krit1: 1–207); THK2F (TACACCATGGGCTATAGTGCACTA- COS-7 and HeLa cells were cultured in Dulbecco’s modified GAA) and KRIT5R (GAGTAACAGTTACTTCTCTTTC) (krit1: Eagle’s medium supplemented with 10% fetal calf serum and an 208–736); THK2F and KC32R (TT CTTCCCAGTTGTTTTG) antibiotic-antimycotic mixture (Invitrogen Corp.). Transfections (krit1: 208–406); THK2F and KC33R (TATCTACC CGATTTG- were carried out with Lipofectamine 2000 reagent (Invitrogen TATACT) (krit1: 208–245); THK2F and KC34R (TACAGGTAT Corp.) for COS-7 and Hela Monster reagent (Invitrogen Corp.) for CTGCTTTCTCT) (krit1: 208–228); KC33F (TATA- HeLa cells, as described by the manufacturer. CAAATCGGGTAGATAAAG) and KC32R (krit1: 239–406); KC33F and KC34R (krit1: 239–286); KC34F (GCTCCA GAC- Mammalian Cell Expression and TACTCAAAAATC) and KC32R (krit1: 256–406); KC34F and Coimmunoprecipitation KC31R (TCAC CACTGTCGTTCCTTGT) (krit1: 256–286), and All in vivo coimmunoprecipitation (co-IP) experiments were FERM1F (GCTGCAAAAT TGTTGAAGGAAG) and KRIT5R performed as described (13). Immunoblotting with a rabbit (krit1: 407–736). All mammalian expression constructs were polyclonal antibody to hemagglutinin (Clontech Laboratories, fashioned by subcloning the appropriate PCR products into Inc.), a mouse monoclonal antibody (Covance Research pcDNA3.1/V5/HIS-TOPO (Invitrogen Corp.). Primer pair Products, Inc., Princeton, NJ), or anti-V5 (Invitrogen Corp.) was MGCKOZ1F (CCACCATGGAAGAGGAGG GCAAGAA) and performed as specified by the manufacturer. MGC2R (TGCTGAGTCCTGG TCCAT) were used for the C- terminal V5-tagged malcavernin construct (malcavernin: 1–444- Subcellular Localization and Fluorescence Microscopy V5). All other constructs were previously described (13). Yeast HeLa cells were grown on coverslips in 35-mm wells at library screening using full-length krit1 as bait was performed 3 ϫ 103 cells/coverslip and were transfected with 3 µg of as described (13). expression construct. Twenty-four hours after transfection, 5 ng/ml of leptomycin (Sigma Chemical Co., St. Louis, MO) was Plasmid Mutagenesis used to treat cells for 3 hours; and anti-V5, anti-HA, or anti- cyclin b (Santa Cruz Biochemicals, Santa Cruz, CA) was Point mutations in krit1 were created from pGBKT7-Krit1 applied. Images were acquired with the Nikon PCM 2000 con- using the QuickChange Gold site-directed mutagenesis kit focal laser scanning microscope (Nikon, Inc., Melville, NY) (Stratagene, La Jolla, CA). Primers KMUT4F (GAAAGCAGAT- using COMPIX software (Intel Corp., Santa Clara, CA). ACC TGTATTTACGCCCCTTTGTTTGGATCAGATC) and KMUT4R (GATCTGATCC AAACAAAGGGGCGTAAAT- Statistical Analysis ACAGGTATCTGCTTTC), and, subsequently, primers KMUT5F (GCAGATACCTGTATTT ACGCCCCTTTGGCTG- One-way analysis of variance was used to detect the differ- β GATCAGATCTTCAG) and KMUT5R (CTGAAGATCTGA ences in the mean values of the -galactosidase activity among TCCAGCCAAAGGGGCGTAAATACAGGT ATCTGC) were constructs. All pair-wise comparisons were performed with a used to create the single mutation, N231A, and the compound Bonferroni t test. mutation, N231A and F234A, in the single expression construct, respectively. Primers KMUT6F (CGGGTAGATAAAGTG- RESULTS GTAATAGCTCCATACTTTGGTCTAGGAGC) and KMUT6R (GCTCCTAGACCAAAGTATGGAGCTATTACCACTTT ATC- Krit1 Interacts with Malcavernin TACCCG), and subsequently, primers KMUT7F (GTAGATA Among the positive clones in two-hybrid screening, those AAGTGGTAATAGCTCCATACG CTGGTCTAGGAGCTCCAG) encoding malcavernin were the second most frequently

354 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com INTERACTION BETWEEN KRIT1 AND MALCAVERNIN

α observed after icap1 (data not shown). To verify this interac- A tion, the malcavernin constructs were purified and reintro- duced with full-length krit1 fusion constructs. Similar to icap1α, malcavernin is a phosphotyrosine binding (PTB) protein (7). To further explore putative interactions, yeast strains were co-transformed with constructs encoding full-length krit1, the integrin β1 cytoplasmic domain (Residues 778–798), full- length icap1α, a centrally deleted icap1 isoform (icap1β) and full-length malcavernin. Full-length krit1 shows evidence of an interaction with malcavernin. Colonies could be observed within 48 hours under high-stringency conditions (compared with that of krit1 and icap1α in less than 24 h). In contrast to icap1α, malcavernin does not interact with β1 integrin (Fig. 1).

Krit1 and Malcavernin Coimmunoprecipitate in Vivo B In vivo co-IPs of interacting complexes were performed. Both malcavernin and icap1α, a known krit1 interactor, pull down krit1 (Fig. 1), indicating interaction between krit1 and malcavernin. HA-tagged full-length krit1, HK5 (C-terminally truncated krit1 containing only the N-terminal NPXY) and THK (N- terminally truncated krit1 containing only the two C-terminal NPXY motifs) fusion constructs were co-transfected with a V5- tagged malcavernin fusion construct. The Co-IP results show that malcavernin can be immunoprecipitated with full-length krit1 and THK but not with HK5, indicating the importance of the second and third NPXY motifs (Residues 231–234 and 250–253) in krit1 for this interaction (Fig. 2).

NPXY Motifs in Krit1-Malcavernin Interaction NPXY is an amino acid motif known to be important in PTB interactions. Two-hybrid analysis was used to map malcavernin- binding sites on krit1. Our results show that the two shortest fragments of krit1 that can independently bind malcavernin are FIGURE 1. A, two-hybrid analysis for krit1 interactors. Yeast was co- residues 229 to 245, covering the second NPXY motif, and transformed with the indicated GAL4 binding domain (GAL4BD; rows) and residues 239 to 255, covering the third NPXY motif (Fig. 3). We GAL4 activation domain (GAL4AD; columns) fusion constructs. Growth of next measured the interaction between malcavernin and the all colonies in ϪTrp/ϪLeu medium indicates that all transformations were different fragments of krit1 in a β-galactosidase activity assay to successful. Growth in selective ϪHis/ϪAde medium indicates high-strin- evaluate the relative importance of the second and third NPXY gency interaction between the indicated molecules. Full-length krit1 interacts β motifs in krit1 in malcavernin binding (Table 1). We found sig- with malcavernin and itself. Both krit1 and the cytoplasmic tail of 1 integrin α β nificantly increased β-galactosidase activity with the fragments interact with icap1 but not icap1 . The robust interaction between SV40 large T antigen (SV40-T) and p53 is shown as a positive control. The lack of of krit1 containing either the second or third NPXY motif Ͻ β any interaction with lamin C is shown as a negative control. Because of inher- (P 0.001). The -galactosidase activity of a fragment contain- ent variation in this assay, we consistently observed that the interaction of ing the third NPXY motif with several ankrin repeats was GAL4AD- cytoplasmic tail of β1 integrin fusion construct and GAL4BD- almost three times higher than that of a fragment containing the icap1α fusion construct was weak, and much longer incubation times were third NPXY motif but without the ankrin repeats (P Ͻ 0.001), needed for the appearance of a colony. B, co-IP of krit1 and malcavernin. Cells illustrating the importance of the sequence flanking the NPXY were co-transfected with constructs encoding the indicated epitope-tagged motifs in this interaction. The relatively lower β-galactosidase molecules. Immunoprecipitation of interacting complexes was performed with activity of a fragment containing the second NPXY motif alone an antibody directed against V5 (V5-IP, upper and middle panels) and HA may be explained by this missing flanking sequence. (HA-IP, lower panel). After sodium dodecyl sulfate polyacrylamide gel elec- We consequently evaluated whether or not expression of trophoresis, the resulting blot was probed with an antibody directed against V5 (α-V5, middle panel) and HA (α-HA, upper and lower panels). Positive mutant forms of krit1 could diminish the interaction between α β signals indicate interaction between krit1 and either icap1 or malcavernin krit1 and malcavernin in a -galactosidase activity assay. Point (upper panel). Expression of V5-tagged constructs was confirmed by prob- mutants N231A (KMUT4) and N250A (KMUT6), compound ing with V5 (middle panel). Expression of HA-tagged constructs was con- mutants N231A/F234A (KMUT5) and N250A/F253A (KMUT7), firmed by IP and probing with HA (lower panel). Lack of any interactions and the double compound mutant N231A/F234A/N250A/ between icap1β and krit1 served as a negative control.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 355 ZHANG ET AL.

TABLE 1. ␤-galactosidase activity in krit1 truncation mutantsa Krit1 fragment Growth ␤-Gal unit K1 (residues 1–171) ϩ 1.187 Ϯ 0.122 K2 (residues 1–207) ϩ 1.073 Ϯ 0.127 K3 (residues 208–406) ϩϩϩϩ 9.807 Ϯ 0.669 K4 (residues 407–736) ϩϩ 2.987 Ϯ 0.350 K5 (residues 208–245) ϩϩϩϩ 1.247 Ϯ 0.095 K6 (residues 208–228) ϩ 0.940 Ϯ 0.122 K7 (residues 239–406) ϩϩϩϩ 6.587 Ϯ 0.378 K8 (residues 239–286) ϩϩϩϩ 2.887 Ϯ 0.140 K9 (residues 256–406) ϩ 0.967 Ϯ 0.012 Negative control 1b ϩ 0.740 Ϯ 0.020 Negative control 2b ϩ 0.770 Ϯ 0.042 Positive controlb ϩϩϩϩ 6.100 Ϯ 0.225

a The second and third NPXY motifs in krit1 are important for malcavernin binding. ␤-galactosidase activity was determined for yeast strains harboring malcavernin and different fragments of the krit1 expression construct in selective –His/–Ade medium. Growth scores that reflect the stringency of the interactions were determined by time of initial appearance and the robustness FIGURE 2. Evidence for interaction between N-terminally truncated krit1 and of colony growth. ␤-galactosidase activity (␤-gal unit) of the fragments malcavernin. In co-IP of V5-tagged malcavernin with HA-tagged krit1, or N- containing either the second or third NPXY motif in krit1 was consistently terminally truncated krit1, THK-transfected COS7 cell lysates were pulled and significantly higher than that observed after expression of any negative controls (P Ͻ 0.001). ␤-galactosidase activity of the fragment K5 containing down with antibodies against HA (upper and middle panels) and antibod- the second NPXY motif was significantly higher than that observed after ies against V5 (lower panel). Western blots were probed with antibodies expression of the two negative controls (P Ͻ 0.001) and two fragments that against HA (middle panel) and V5 (upper and lower panels). Positive sig- did not harbor either the second or third NPXY (K6, K9) (P Ͻ 0.001). ␤- nals indicate interaction between malcavernin and either krit1 or N-terminally galactosidase activity of the fragment K5 was also higher than that observed truncated krit1 (upper panel). Expression of HA-tagged constructs was con- after expression of two N-terminal fragments of krit1, K1 and K2; but did not firmed by probing with HA (middle panel). Expression of V5-tagged malcav- reach significance. These N-terminal fragments of krit1, K1, and K2, ernin was confirmed by co-IP with anti-V5 and probed with anti-V5 (lower demonstrated ␤-gal activity greater than the negative controls (P Ͻ 0.001), panel). Lack of any interaction is seen between malcavernin and C-terminally although no interaction with malcavernin was found by Western blot or truncated krit1. colony formation in the yeast two-hybrid analysis. The difference is significant compared with any negative controls, but not significant compared with the other two fragments that did not harbor the second or third NPXY motif. The higher ␤-galactosidase activity of fragment K4 containing the FERM F253A (KMUT8) were assayed for interaction with malcavernin domain in the C-terminus of krit1 was expected as this construct shows (Table 2). The β-galactosidase activity in krit1 point mutants, very high intrinsic activity (P Ͻ 0.001), with nonspecific binding as observed KMUT4 or KMUT6 was decreased by approximately 20 to 25% in the yeast two-hybrid experiments. ␤-galactosidase activity of fragment K7 compared with wild-type krit1 (P Ͻ 0.001). The β-galactosi- containing the third NPXY motif with the ankrin repeats was significantly higher than that of fragment K8 containing the third NPXY motif without the dase activity in compound mutant KMUT5 was significantly ankrin repeats (P Ͻ 0.001), suggesting the contribution of some element decreased compared with either wild-type krit1 or its relative in the ankrin repeats to this interaction. point mutant (KMUT4) (P Ͻ 0.001). The β-galactosidase activ- b The control interactions indicated in the table are negative control 1, ity in the double compound mutant, KMUT8, was further sig- between fusions GAL4BD-Lamin C and GAL4AD-large T antigen; negative nificantly decreased compared with either wild-type krit1 or control 2, between GAL4BD-Lamin C and GAL4AD-malcavernin; and positive control, between GAL4BD-p53 and GAL4AD-large T antigen. either of the compound mutants (KMUT5 or KMUT7) (P Ͻ 0.001). These data suggest that krit1 binds to malcavernin via these two NPXY motifs. Furthermore, we consistently observed that the β-galactosidase activity in the mutants involving the nuclear export signal (NES). Malcavernin shows the same cel- second NPXY motif (KMUT4/KMUT5) was always smaller lular distribution as krit1 in the absence of a recognizable than when mutations involved the third NPXY motif NLS and NES (Fig. 4). (KMUT6/KMUT7).

Cellular Localization of Malcavernin and Krit1 DISCUSSION Immunolocalization shows that krit1 is predominantly We have previously reported the interaction between icap1α cytoplasmic at steady state, but shuttles between the nucleus and krit1 (13). Here, we report the interaction of malcavernin, and the cytoplasm, as suggested by nuclear accumulation in the CCM2 gene product, and krit1. This interaction is demon- the presence of leptomycin B, an inhibitor of nuclear export. strated by yeast two-hybrid analysis and conﬁrmed by in vivo Krit1 has both a nuclear localization signal (NLS) and a co-IP in transfected cultured mammalian cells. The NPXY motif

356 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com INTERACTION BETWEEN KRIT1 AND MALCAVERNIN

interaction with malcavernin, but not the first NPXY motif that binds to icap1α. The redundancy of NPXY motifs in krit1 for malcavernin- binding may exist for several reasons. It may be that the location of the second and third NPXY motifs in the middle of the krit1 primary sequence may not be easily accessed or simultaneously available for binding. The tertiary structure of krit1 may be altered by the binding of other proteins altering the availability of either or both of these NPXY motifs. Another possible explanation is that the molecular mass of malcavernin (49 kD) is more than twice that of icap1α (22kD) and that sufficient interaction for binding might require an additional NPXY motif. Finally, this interaction may be crucially vital for normal cellular signaling, and redundancy may exist to ensure it. Icap1α also binds an NPXY motif in the cytoplasmic tail of β1 integrin. Malcavernin does not bind to the cytoplasmic tail of β1 integrin or to the initial NPXY motif in krit1, emphasizing the specificity of these NPXY motifs in these distinct PTB interactions. Interestingly, malcavernin shows the same cellular distribution patterns as krit1 and is predominantly cytoplasmic FIGURE 3. A, epitope mapping for malcavernin-binding domain in krit1. A, two-hybrid analysis to determine which of at steady state, but shuttles the three NPXY motifs in the krit1 is required for interaction with malcavernin. Yeast were co-transformed with the indi- through the nucleus despite Ϫ Ϫ cated GAL4BD (rows) and GAL4AD (columns) fusion constructs. Growth of all colonies in Trp/ Leu medium indi- absence of a recognizable NLS cates that all transformations were successful. Because GAL4BD-malcavernin has some intrinsic nonspecific activity, and NES. Icap1α has a puta- GAL4BD portions of krit1 were used to screen GAL4AD-malcavernin (left panel). Growth in selective ϪHis/ϪAde tive NLS, but lacks an NES. medium indicates high stringency interaction between the indicated molecules. The robust interaction between SV40-T α and p53 served as a positive control. The lack of any interaction with lamin C served as a negative control (right panel). Icap1 also seems to shuttle B, two-hybrid data are summarized in this diagram. The two smallest fragments of krit1 interacting with malcavernin were through the nucleus (unpub- identified, one (residues 229–245) covering the second NPXY motif (pY) (residues 231–234), and another (residues lished results). Because both 239–255) covering the third NPXY motif (residues 250–253), each able to independently interact with malcavernin. icap1α and malcavernin lack FLHK, full length human krit1; ANK, ankrin; FERM, protein 4.1, elvin, radixin, moesin. the necessary sequences to shuttle through the nuclear is a well-known binding substrate for PTB proteins; both compartment and because both interact with krit1 (a protein icap1α and malcavernin contain PTB domains (1, 7). There are with both a strong NLS and NES), we hypothesize that krit1 three NPXY motifs in krit1. We have previously shown that shuttles malcavernin and icap1α through the nucleus, thereby icap1α binds to the first NPXY motif in krit1 (13). We demon- exerting regulatory influences on signal transduction pathways strate that the second and third NPXY motifs are crucial for the mediated by these molecules.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 357 ZHANG ET AL.

and subcellular compartmentalization that results from this interaction may be important in the regulation of cellular signaling involved in the pathogenesis of CCMs. The goal of research aimed at understanding the cellular signaling abnormalities resulting from recently identiﬁed genetic mutations associated with CCMs is the development of effective molecular interventions to prevent the development of these vascular malformations. Understanding such molecular lesions may allow for the development of small molecule regulators of crucial signal transduction pathways that revert the abnormal phenotype of patients with cavernous malformations preventing hemorrhage, seizures, and the need for surgical resection.

REFERENCES FIGURE 4. Cellular localization of krit1 and malcavernin. Krit1 is seen pre- 1. Chang DD, Wong C, Smith H, Liu J: Icap-1: A novel β1 integrin cytoplasmic dominantly of the cytoplasm of cells but accumulates in the nucleus after domain-associated protein, binds to a conserved and functionally important blockade of nuclear export by leptomycin (A). Malcavernin (B) shows exactly npxy sequence motif of beta1 integrin. J Cell Biol 138:1149–1157, 1997. the same pattern as krit1. Cyclin b, known to accumulate in the nucleus after 2. Clatterbuck RE, Eberhart CG, Crain BJ, Rigamonti D: Ultrastructural and leptomycin treatment, serves as a positive control (C). immunocytochemical evidence that an incompetent blood-brain barrier is related to the pathophysiology of cavernous malformations. J Neurol Neurosurg Psychiatry 71:188–192, 2001. 3. Craig HD, Gunel M, Cepeda O, Johnson EW, Ptacek L, Steinberg GK, Ogilvy TABLE 1. ␤-galactosidase activity in krit1 point mutantsa CS, Berg MJ, Crawford SC, Scott RM, Steichen-Gersdorf E, Sabroe R, Kennedy ␤-Gal (% CT, Mettler G, Beis MJ, Fryer A, Awad IA, Lifton RP: Multilocus linkage Krit1 mutation Growth wild type) identifies two new loci for a mendelian form of stroke, cerebral cavernous malformation, at 7p15–13 and 3q25.2–27. Hum Mol Genet 7:1851–1858, 1998. ϩϩϩϩ Wild type 100 4. Del Curling O Jr, Kelly DL Jr, Elster AD, Craven TE: An analysis of the natu- Kmut4 (N231A) ϩϩϩ 76.2 ral history of cavernous angiomas. J Neurosurg 75:702–708, 1991. Kmut5 (N231A/F234A) ϩϩ 59.4 5. Eerola I, Plate KH, Spiegel R, Boon LM, Mulliken JB, Vikkula M: Krit1 is Kmut6 (N250A) ϩϩϩ 79.4 mutated in hyperkeratotic cutaneous capillary-venous malformation associ- ϩϩ ated with cerebral capillary malformation. Hum Mol Genet 9:1351–1355, 2000. Kmut7 (N250A/F253A) 76.9 6. Laberge-le Couteulx S, Jung HH, Labauge P, Houtteville JP, Lescoat C, Kmut8 (N231A/F234A/N250A/F253A) ϩ 41.6 Cecillon M, Marechal E, Joutel A, Bach JF, Tournier-Lasserve E: Truncating mutations in ccm1, encoding krit1, cause hereditary cavernous angiomas. a Both the second and third NPXY motifs in krit1 are important for malcavernin Nat Genet 23:189–193, 1999. binding. ␤-galactosidase (␤-gal) activity in this table is expressed as percentage 7. Liquori CL, Berg MJ, Siegel AM, Huang E, Zawistowski JS, Stoffer T, Verlaan of wild-type activity. ␤-galactosidase activity of wild-type krit1 was consistently D, Balogun F, Hughes L, Leedom TP, Plummer NW, Cannella M, Maglione V, and significantly higher than that observed after expression of any mutant Squitieri F, Johnson EW, Rouleau GA, Ptacek L, Marchuk DA: Mutations in a forms of krit1 (P Ͻ 0.001); likewise, ␤-galactosidase activity of the point gene encoding a novel protein containing a phosphotyrosine-binding domain mutant form of krit1, Kmut4 (with a single altered amino acid in the second cause type 2 cerebral cavernous malformations. Am J Hum Genet NPXY motif) was consistently and significantly higher than that observed after 73:1459–1464, 2003. expression of the compound mutant form Kmut5 (altering two amino acids 8. Otten P, Pizzolato GP, Rilliet B, Berney J: 131 cases of cavernous angioma (cav- in the second NPXY motif) (P Ͻ 0.001). Although the difference between ernomas) of the CNS, discovered by retrospective analysis of 24,535 autopsies Kmut6 and the compound mutant Kmut7 (affecting one or two amino acids [in French]. Neurochirurgie 35:82–83, 128–131, 1989. in the third NPXY motif, respectively) was not statistically significant at this 9. Rigamonti D, Hadley MN, Drayer BP, Johnson PC, Hoenig-Rigamonti K, incubation, statistically significant differences were observed at longer Knight JT, Spetzler RF: Cerebral cavernous malformations. Incidence and incubation times (data not shown). Similarly, ␤-galactosidase activity of familial occurrence. N Engl J Med 319:343–347, 1988. compound mutant forms of krit1 (Kmut5, Kmut7) were consistently and 10. Sahoo T, Johnson EW, Thomas JW, Kuehl PM, Jones TL, Dokken CG, significantly higher than that observed after expression of the double Touchman JW, Gallione CJ, Lee-Lin SQ, Kosofsky B, Kurth JH, Louis DN, compound mutant forms of Kmut8 (P Ͻ 0.001), suggesting that both the Mettler G, Morrison L, Gil-Nagel A, Rich SS, Zabramski JM, Boguski MS, second and third NPXY motifs in krit1 together are essential for optimal Green ED, Marchuk DA: Mutations in the gene encoding Krit1, a krev- malcavernin binding. 1/rap1a binding protein, cause cerebral cavernous malformations (ccm1). Hum Mol Genet 8:2325–2333, 1999. 11. Verlaan DJ, Davenport WJ, Stefan H, Sure U, Siegel AM, Rouleau GA: Zawistowski et al. (12) have recently published the same Cerebral cavernous malformations: Mutations in Krit1. Neurology 58: 853–857, 2002. results demonstrating the interaction of krit1 (CCM1) and mal- 12. Zawistowski JS, Stalheim L, Uhlik MT, Abell AN, Ancrile BB, Johnson GL, cavernin (CCM2) and mapping their interaction site using sim- Marchuk DA: CCM1 and CCM2 protein interactions in cell signaling: ilar molecular techniques. They also demonstrated nuclear shut- Implications for cerebral cavernous malformations pathogenesis. Hum Mol tling of complexes containing krit1 and malcavernin. Our data Genet 14:2521–2531, 2005. confirm these results in a series of similar but not identical 13. Zhang J, Clatterbuck RE, Rigamonti D, Chang DD, Dietz HC: Interaction between krit1 and icap1alpha infers perturbation of integrin beta1-mediated experiments, providing independent validation of their find- angiogenesis in the pathogenesis of cerebral cavernous malformation. Hum ings. These data support the concept that these proteins interact Mol Genet 10:2953–2960, 2001.

358 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com INTERACTION BETWEEN KRIT1 AND MALCAVERNIN

14. Zhang J, Clatterbuck RE, Rigamonti D, Dietz HC: Mutations in Krit1 in famil- published similar results showing that these two proteins interact and ial cerebral cavernous malformations. Neurosurgery 46:1272–1279, 2000. mapping their interaction site (1). Both of these novel studies help 15. Zhang J, Clatterbuck RE, Rigamonti D, Dietz HC: Cloning of the murine advance our understanding of CCM pathophysiology as well as impor- Krit1 cDNA reveals novel mammalian 5’ coding exons. Genomics 70:392–395, tant details of intracellular signaling pathways that may be relevant to 2000. CCM formation and angiogenesis in general. Further research will determine whether or not this type of molecular genetic knowledge Acknowledgments will ultimately lead to new treatments or identify prognostic markers We thank Junji Chen for valuable discussions and technical assistance. This for patients with CCM. work was supported by the Center for Inherited Neurovascular Disease (www.cind.org) (DR), the Salisbury Family Foundation (DR), the Fondazione Michael E. Kelly Agarini (DR), and the Howard Hughes Medical Institute (HCD). Gary K. Steinberg Stanford, California COMMENTS 1. Zawistowski JS, Stalheim L, Uhlik MT, Abell AN, Ancrile BB, Johnson GL, hang et al. demonstrate that malcavernin independently binds to Marchuk DA: CCM1 and CCM2 protein interactions in cell signaling: Ztwo of the three asparagine, praline, undetermined/variable amino Implications for cerebral cavernous malformations pathogenisis. Hum Mol acid, and tyrosine motifs in krit1. Furthermore, malcavernin location is Genet 14:2521–2531, 2005. analyzed by immunohistochemistry. These data suggest that krit1 interacts with malcavernin through its asparagine, praline, undetermined/ his is an excellent study that confirms the recently published results variable amino acid, and tyrosine motifs. This study is interesting and Tfrom another group (1), which also describe the same interaction of includes important observations on the formation of cerebral cavernous malcavernin and krit1, absent interaction with the truncated protein, malformations. The authors present in vitro data and the results and cellular colocalization of the two CCM gene products for CCM2 obtained from two-hybrid analysis. In tissue culture experiments, they and CCM1. This is consistent with the important notion of a common use COS-7 and HeLa cells that are not suited for the study of the cen- macromolecuar complex involving both proteins and resulting in CCM tral nervous system. However, these data support the hypothesis that disease when either component of the complex is perturbed. these proteins interact and that the subcellular compartmentalization Understanding this interaction, other components of this disease path- that results from this interaction may be important in the regulation of way, and its function in the normal state will be essential to better elu- cellular signaling involved in the pathogenesis of cerebral cavernous cidate mechanisms and variants of CCM disease and, ultimately, its malformations (CCMs). Recently, Zawistowski et al. (1) have published prevention or modification. It will also be essential to determine any the same results demonstrating the interaction of krit1 and malcav- relationship of this complex to the CCM3 gene product, PDCD10, ernin using similar molecular techniques. These results help us under- which must also be linked to the same emerging disease pathway. stand the pathogenesis of CCMs. Although the observation reported in this study confirms a previous report by another group, this does not diminish the credit deserved by Yasushi Takagi the authors for their well planned and executed experiments. Such Nobuo Hashimoto independent confirmation is essential to the scientific validity of a con- Kyoto, Japan cept or hypothesis, and its impact is often as critical as that of the earlier reported work.

1. Zawistowski JS, Stalheim L, Uhlik MT, Abell AN, Ancrile BB, Johnson GL, Issam A. Awad Marchuk DA: CCM1 and CCM2 protein interactions in cell signaling: Evanston, Illinois Implications for cerebral cavernous malformations pathogenisis. Hum Mol Genet 14:2521–2531, 2005. 1. Zawistowski JS, Stalheim L, Uhlik MT, Abell AN, Ancrile BB, Johnson GL, he results of this study elucidate novel molecular interactions of Marchuk DA: CCM1 and CCM2 protein interactions in cell signaling: Tkrit1 and malcavernin, the proteins encoded by the genes of the Implications for cerebral cavernous malformations pathogenisis. Hum Mol CCM1 and CCM2 loci. As the authors mention, another group recently Genet 14:2521–2531, 2005.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 359 EXPERIMENTAL STUDIES Sarah C. Jost, M.D. Department of Neurosurgery, Washington University, School of Medicine, St. Louis, Missouri John E. Wanebo, M.D. Department of Neurosurgery, Uniformed Services University of the IN VIVO IMAGING IN A MURINE MODEL Health Sciences National Naval Medical Center OF GLIOBLASTOMA Bethesda, Maryland

Sheng-Kwei Song, Ph.D. OBJECTIVE: To use in vivo imaging methods in mice to quantify intracranial glioma Department of Radiology, Washington University, growth, to correlate images and histopathological findings, to explore tumor marker School of Medicine, specificity, to assess effects on cortical function, and to monitor effects of chemother- St. Louis, Missouri apy. Michael R. Chicoine, M.D. METHODS: Mice with DBT glioma cell tumors implanted intracranially were imaged Department of Neurosurgery, Washington University, serially with a 4.7-T small-animal magnetic resonance imaging (MRI) scanner. MRI School of Medicine, tumor volumes were measured and correlated with postmortem histological findings. St. Louis, Missouri Different nonspecific and specific positron emission tomography radiopharmaceuti- -Keith M. Rich, M.D. cals, [18F]2-fluoro-2-deoxy-D-glucose, [18F]3؅-deoxy-3؅-fluorothymidine, or [11C]RHM Department of Neurosurgery, σ Washington University, I, a 2-receptor ligand, were visualized with microPET (CTI-Concorde MicroSystems School of Medicine, LLC, Knoxville, TN). Intrinsic optical signals were imaged serially during contralateral St. Louis, Missouri whisker stimulation to study the impact of tumor growth on cortical function. Other Thomas A. Woolsey, M.D. groups of mice were imaged serially with MRI after one or two doses of the antimitotic .(Department of Neurosurgery, N,N؅-bis(2-chloroethyl)-N-nitrosourea (BCNU Washington University, School of Medicine, RESULTS: MRI and histological tumor volumes were highly correlated (r 2 ϭ 0.85). St. Louis, Missouri Significant binding of [11C]RHM-I was observed in growing tumors. Over time, tumors Jason S. Lewis, Ph.D. reduced and displaced (P Յ 0.001) whisker-activated intrinsic optical signals but did not Department of Radiology and change intrinsic optical signals in the contralateral hemisphere. Tumor growth was Washington University, School of Medicine, delayed 7 days after a single dose of BCNU and 18 days after two doses of BCNU. St. Louis, Missouri Mean tumor volume 15 days after DBT implantation was significantly smaller for treated Robert H. Mach, Ph.D. mice (1- and 2-dose BCNU) compared with controls (P ϭ 0.0026). Department of Radiology and CONCLUSION: Mouse MRI, positron emission tomography, and optical imaging pro- Washington University, School of Medicine, vide quantitative and qualitative in vivo assessments of intracranial tumors that corre- St. Louis, Missouri late directly with tumor histological findings. The combined imaging approach pro- Jinbin Xu, Ph.D. vides powerful multimodality assessments of tumor progression, effects on brain function, Department of Radiology, and responses to therapy. Washington University, School of Medicine, KEY WORDS: Glioma, Magnetic resonance imaging, Optical imaging, Positron emission tomography, St. Louis, Missouri Therapeutic response Joel R. Garbow, Ph.D. Neurosurgery 60:360–371, 2007 DOI: 10.1227/01.NEU.0000249264.80579.37 www.neurosurgery-online.com Departments of Radiology and Chemistry, Washington University, School of Medicine, St. Louis, Missouri lioblastoma (GBM) is the most common The evaluation of animal models requires The Alvin J. Siteman Cancer Center primary malignant neoplasm of the quantitative methods to assess the efficacy of (JSL, RHM, JRG) Gadult brain (62, 65). Treatment outcomes possible therapeutic interventions. Small- Reprint requests: are poor: median survivals are only 1 year (19). animal imaging, including magnetic resonance Joel R. Garbow, Ph.D., Department of Radiology, GBMs are highly invasive and tumor cells are imaging (MRI), positron emission tomography Washington University, found in areas of the brain at significant dis- (PET), and optical methods, can follow tumor School of Medicine, tances from the bulk tumor, making surgical progression in vivo (39). Combination of these 4525 Scott Avenue, Campus Box 8227, resection especially challenging (34, 54, 60). The methods will allow quantification of intracra- St. Louis, MO 63110. limitations of current treatments make the devel- nial tumor growth and monitoring tumor biol- Email: [email protected] opment of new therapies imperative. Such ogy and will permit longitudinal assessment Received, March 29, 2006. advances depend, in part, on animal models that of the effects of therapeutic interventions in Accepted, September 18, 2006. accurately mimic human glioblastomas (28). vivo. Complementing this approach are meth-

360 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MONITORING BRAIN TUMORS IN VIVO WITH SMALL ANIMAL IMAGING

ods that assess function, including imaging intrinsic optical blue dye, and counted with a hemocytometer. The cells were signals (IOSs). recentrifuged and suspended in serum-free DMEM to a con- Visualizing the in vivo progress of tumor growth has been a centration of 80,000 or 100,000 DBT cells/µl and cooled on ice limiting factor in the characterization of many tumor models. before injection. Small-animal MRI has offered a potential solution to this challenging problem (1, 5–7, 10, 44, 52, 55, 67, 78). Longitudinal Animal Surgery MRI studies provide a practical approach for identifying Seventy-four adult BALB/c mice weighing approximately intracranial glioma tumors and following tumor progression in 25 g were studied. All animal protocols were approved by small animals (1, 6, 28, 44, 57). Even when symptomatic with the Washington University Division of Comparative Medicine brain tumors, mice generally tolerate serial MRI studies (6, 44). and met or exceeded American Association for the Recent work has shown that calculations of intracranial tumor Accreditation of Laboratory Animal Care International and volumes from MRI correlate well with postmortem tumor vol- National Institutes of Health standards. The mice were anes- umes (16, 55). Success in imaging murine brain tumors using thetized with intraperitoneal ketamine (25 mg/kg ), xylazine clinical MRI scanners (7, 48, 67) and in imaging multiple mice (5 mg/kg), and acepromazine (2.5 mg/kg) before intracranial in a single experiment have also been reported (5, 78). DBT cell implantation. After a midline scalp incision, a cran- Although early in vivo studies of brain tumors in mice used iotomy was made over the cortex with a 1-mm cutting burr. subcutaneously implanted tumor cells (32, 37, 46), more recent Mice were secured in a stereotactic frame and 5 µl DBT tumor studies have used cultured tumor cells implanted directly in cell suspension were aspirated into a Hamilton syringe the brain (1, 3, 13, 14, 44, 45, 51, 57). Here, we used implanted attached to the frame which was then passed into the brain. DBT cells, cells that are similar to human GBMs in their aggres- For the optical imaging studies, the whisker-barrel cortex sive growth pattern, histopathological features, and immunos- (first somatosensory cortex) was targeted using cranium and taining (1, 15, 57, 76). vascular landmarks, and the syringe was advanced to 750 µm In this study, small-animal MRI scanning was used to char- below the cortical surface. 2.0 ϫ 105 tumor cells were injected acterize longitudinal DBT tumor growth quantitatively in mice. during a 3-minute time period. For the BCNU treatment, PET PET was used to assess biologically relevant markers of GBMs, experiments, and some of the volumetric and MRI studies, including metabolism, mitoses, and tumor-specific σ recep- 2 2.5 ϫ 105 tumor cells were injected into the striatum at a site tors in vivo. IOS was used to evaluate the effects of GBM on 2 mm lateral and 2 mm posterior to the bregma and 4 mm sensory function of the brain in response to whisker stimula- deep to the cortical surface. After the needle was removed, the tion. Finally, the efficacy of therapeutic interventions on glioma craniotomy was sealed with bone wax and the scalp was growth in mice with single and multiple doses of N,N؅-bis(2- closed with super glue or a metal staple that was later chloroethyl)-N-nitrosourea (BCNU) was measured. removed before MRI examination. After surgery, mice were housed in nonbarrier animal facilities for the duration of the MATERIALS AND METHODS imaging experiments. Although animals with large brain tumors are more likely to die under anesthesia than healthy Mouse DBT Cell Line animals, mice were successfully and routinely imaged at mul- The DBT glioblastoma cell line used in these experiments tiple time points throughout this study. was donated by Dr. Michael Lai, Department of Microbiology, University of Southern California. DBT cells frozen in culture MRI medium with 5% dimethyl sulfoxide were thawed, washed Images were collected in an Oxford Instruments 4.7-T magnet with magnesium-free Hank’s balanced salt solution, and grown (33 cm, clear bore; Oxford, United Kingdom) equipped with 15-cm in Dulbecco’s modified Eagle’s medium (DMEM) with 15% inner diameter, actively shielded gradient coils (maximum gradi- heat-inactivated fetal calf serum, 0.2 mmol/L glutamine, ent, 18 G/cm; rise time, 100 µs). The magnet/gradients were inter- 50 mg/ml neomycin, 100 mg/ml penicillin, and 100 mg/ml faced with a Varian INOVA console (Palo Alto, CA), and data streptomycin. The cells were plated in T-75 culture flasks for were collected using a 1.5-cm outer diameter surface coil (receive) Њ incubation in a 5% CO2 humidified atmosphere at 37 C and and a 9-cm inner diameter Helmholtz coil (transmit). Before the used at low passages. imaging experiments, mice were anesthetized with isoflurane/O2 (4% [v/v]) and maintained on isoflurane/O2 (1.5% [v/v]) Preparation of Cells for Injection throughout the experiments. Feeding medium was aspirated from the culture flasks and Initially, data were collected using T2- and diffusion- 3 ml of 0.5% trypsin-ethylenediamine tetra-acetic acid were weighted, spin-echo, multislice imaging sequences with a added during a 1-minute time period. After the cells were sus- 1.5 ϫ 1.5-cm2 coronal field of view (T2-weighted: repetition pended, trypsin was inactivated by adding 7 ml of DMEM with time [TR], 3 s; echo time [TE], 50 ms; slice thickness, 0.5 mm; fetal calf serum. The suspended cells were centrifuged at diffusion-weighted imaging: TR, 1.5 s; TE, 42 ms; slice thick- 250 ϫ g for 10 minutes at 4ЊC and resuspended in 2 ml of ness, 1 mm). In later imaging studies, mice were injected DMEM. An aliquot of cells was aspirated, stained with trypan intraperitoneally with 500 µl Omniscan (Gadodiamide, GE

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 361 JOST ET AL.

Healthcare, Little Chalfont, United Kingdom) contrast agent, RHM-I (24 h apart) then again the following week with only diluted 1:10 in sterile saline, 15 minutes before being placed in [11C]RHM-I. For [18F]FDG imaging, animals were fasted the magnet. T1-weighted, gradient-echo multislice images overnight, injected via a tail-vein catheter with approximately (coronal and horizontal views) were collected (TR, 0.125 s; TE, 0.5 mCi of [18F]FDG, and imaged at 1 hour after injection 0.0025 s; field of view [coronal], 1.5 ϫ 1.5 cm2; field of view (1 ϫ 10 min frame). For [18F]FLT imaging, animals were [horizontal], 3 ϫ 3cm2; slice thickness, 0.5 mm). In addition, injected with approximately 0.25 mCi [18F]FLT and imaged at T2-weighted, multislice spin-echo coronal images were col- 1 hour after injection (1 ϫ 10 min frame). For [11C]RHM-I lected for each animal (TR, 1.5 s; TE, 0.05 s; field of view, imaging, animals were injected with 0.4 to 0.5 mCi of the 11C- 1.5 ϫ 1.5 cm2; thickness, 0.5 mm). T2-weighted images were labeled compound and were imaged at 30 minutes and 1 hour used when it was difficult to see tumor margins. Tumor vol- after injection (1 ϫ 15 min frame). Following PET imaging umes were measured by manual outlining tumors using protocols, selected mice were imaged for coregistration on Varian’s ImageBrowser software (Varian NMR Instruments, the microCAT. Inc., Palo Alto, CA) or post hoc with the public domain program NIH Image (available at http://rsb.info.nih.gov/nih-image). Ligand Specificity To compare data from different experimental groups, tumor The tritiated compound [3H]RHM-1 was synthesized by volumes were first normalized by dividing individual values in American Radiolabeled Chemicals, Inc. (St. Louis, MO) via a particular case by the sum of all values for that case. These O-alkylation of the corresponding phenol precursor (66); chem- normalized values were used to compute means and standard ical purity was greater than 99% and the specific activity of the deviations or ranges (for BCNU, ϫ2). Computed values and radioligand was 80 Ci/mmol. Membrane homogenates were variations then were scaled by calculating the ratio of the final prepared from approximately 4-g DBT tumors that were largest values in a series to one (1) and multiplying all values in removed from previously implanted mice and frozen immedi- the series by this scaling factor. None of these procedures ately on dry ice and stored at Ϫ80ЊC, thawed slowly on ice, and changes the exponents of the fitted curves. homogenized at 4ЊC with a Potter-Elvehjem tissue grinder (Wheaton Science Products, Millville, NJ) at a concentration of PET 1 g tissue/ml of 50 mmol/L Tris-HCl at pH 8.0. The crude Images were collected on either a microPET-Focus-220 or a membrane homogenate was then transferred to a 50-ml cen- microPET-Focus-120 (CTI-Concorde MicroSystems LLC; trifuge tube and resuspended to a concentration of 0.2 g of tis- Knoxville, TN) tomograph (63). Coregistration of the PET images sue/ml of 50 mmol/L Tris-HCl. Additional homogenization was achieved in combination with a microCAT-II camera (Imtek was accomplished using an Ultra-Turrax T8 polython homog- Inc., Knoxville, TN) for high-resolution computed tomographic enizer (IKA Works, Inc., Wilmington, NC). The final (CT) images. Mice were anesthetized with 1 to 2% isoflurane homogenate then was centrifuged for 10 minutes at 1000 ϫ g, before scanning and were immobilized in the supine position in the pellet was discarded, and the supernatant was mixed by a custom cradle. Two mice were imaged side by side in the same vortexing. Aliquots were stored at Ϫ80ЊC until use. The protein bed position at all time points. Image registration between concentration of the suspension was determined using the DC microCT and PET images was accomplished using a landmark protein assay (Bio-Rad, Hercules, CA) and averaged approxi- registration technique and AMIRA image display software mately 10 mg protein/ml stock solution. (AMIRA; TGS, Inc., San Diego, CA). The registration method Approximately 350 µg of membrane homogenates were proceeds by rigid transformation of the microCT images from diluted with 50 mmol/L Tris-HCl buffer, pH 8.0, and incubated landmarks provided by fiducial markers directly attached to the with [3H]RHM-1 in a total volume of 150 µl at 25ЊC in 96-well animal bed. This coregistration method allows for accurate reg- polypropylene plates (Fisher Scientific, Pittsburgh, PA). The istration of CT and PET images to within Ϯ0.5 mm. concentrations of the radioligand ranged from 0.1 to 8 nM. After incubation for 60 minutes, the reactions were terminated Analysis of PET Images by the addition of 150 µl of cold wash buffer (10 mmol/L Tris- Mice were imaged at different time points during tumor HCl, 150 mmol/L NaCl, pH 7.4, at 4ЊC) with a 96-channel development with [18F]2-fluoro-2-deoxy-D-glucose ([18F]FDG) transfer pipette (Fisher Scientific), and the samples were har- to monitor tumor metabolism, 18F-3؅-deoxy-3؅-fluo- vested and filtered rapidly to 96-well fiberglass filter plates (61 ,42) rothymidine ([18F]FLT) to measure thymidine kinase-1 activ- (Millipore, Billerica, MA) presoaked with 100 µl 50 mmol/L 11 σ 11 ity (58, 59), and a C-labeled 2 receptor ligand ([ C]RHM-I) Tris-HCl buffer, pH 8.0, for 1 hour. Each filter was washed three (66) to evaluate proliferation. [18F]FDG was prepared using times with 200 µl of ice-cold wash buffer. A Wallac 1450 the Coincidence Technologies [18F]FDG synthesis module (GE MicroBeta liquid scintillation counter (Perkin Elmer, Boston, Healthcare). [18F]FLT (26) and [11C]RHM-I were produced MA) was used to quantitate the bound radioactivity (77). according to published protocols (66). Two groups of animals Nonspecific binding was determined from samples that con- were studied. The first group of animals (n ϭ 8) was imaged tained 10 µM RHM-1. The equilibrium dissociation constant 18 with [ F]FDG on a weekly basis over a 3-week period after (Kd) and maximum number of binding sites (Bmax) were deter- DBT cell injection. The second group (n ϭ 4) was imaged dur- mined by a linear regression analysis of the transformed data ing active tumor development with both [18F]FLT and [11C]- using the method of Scatchard (53).

362 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MONITORING BRAIN TUMORS IN VIVO WITH SMALL ANIMAL IMAGING

Optical Imaging of Functional Responses toxylin and eosin and were correlated with MRI data at similar Functional optical imaging experiments were performed at levels (Fig. 1). In a second subset of mice, seven animals with two different time points at least 1 week apart in 23 animals tumor cells injected into the barrel cortex were imaged by MRI (between 6 and 26 d after tumor injection). Details of the imaging and perfusion on Day 15 before they were sacrificed. These system and analysis are provided elsewhere (20, 22). Illumination brains were cut parallel to the pia and were stained for was provided by focused (Edmund Scientific, Tonawanda, NY) cytochrome oxidase (57). Histological tumor volumes from filtered light (520–560 nm) from a 150-W, 21-V EKE light bulb digital microscopic images were measured with NIH Image. (Opti-Quip, Highland Mills, NY) with a regulated power supply. Whisker barrels and related images of the cortical surface were The voltage was stabilized by placing 10 0.29-F capacitors in par- reconstructed by standard techniques (57). allel with the power supply. The image of the brain surface was magnified ϫ0.75 to ϫ3 with a Nikon SMZ-U dissecting micro- Evaluation of Therapeutic Response with MRI scope (Nikon Corp., Melleville, NY) while focusing on surface Twenty-four animals were randomized into three groups of vessels through the cranium and captured with a charge-cou- eight after tumor injection. Seven surviving mice in the first pled device camera (72S; Hamamatsu Corp., Bridgewater, NJ) at group were not treated (vide supra). Animals in the second 30 Hz. These 640 ϫ 480-pixel images were stored with a real-time group were treated with a single intraperitoneal dose of BCNU capture card (AG-5; Scion Corporation, Frederick, MD) in a (10 mg/kg) on postoperative Day 7, and those in the third Power Macintosh G3 Computer (Apple Computer, Inc., group were treated with two intraperitoneal doses of BCNU Cupertino, CA) running NIH Image 1.62, optimized by modify- (10 mg/kg) given on postinjection Days 7 and 14. All untreated ing the source code. The IOS was averaged from 20 trials of 3 sec- animals (n ϭ 7) and those with 1 dose of BCNU (n ϭ 8) under- onds of stimulation, and the prestimulation signal was subtracted went MRI evaluation on postoperative Days 5, 8, 12, and 15. to yield optical images. Both the normal and tumor-injected hemi- Of the eight animals treated with a second dose of BCNU, four spheres of each animal were imaged during a single recording survived to be imaged on postinjection Days 12, 15, 19, and 22. session. For two of these animals, data also were collected 26, 29, and 33 days after implantation. IOS Response Assessment IOS images from control and tumor-involved cortices were RESULTS rank-ordered in a blinded analysis. Color prints of all 78 IOS images (46 from 23 tumor-injected hemispheres and 32 from 16 All 74 animals showed tumor growth by MRI. Hydro- uninjected hemispheres from a subset of these animals) were cephalus, brain herniation (related to tumor location and size), printed at the same magnification and coded. These were shuf- and BCNU toxicity contributed to animal death. Furthermore, fled as a deck of playing cards and three members of the labo- animals with large tumors occasionally died when anesthetized ratory not participating in the experiment were asked to sort for MRI, PET, and optical imaging experiments. the images according to color, active (hot) to inactive (cold). The The results reported are from 51 mice that were studied at ranks were assigned to each image, and these were converted least twice by imaging or with a combination of several meth- to percentile rankings (hottest, 100; coldest, 0) for each observer. ods, that is, MRI and histological analysis. Tumors were visu- The 3rd percentile values for each image were averaged. The alized consistently either with a T1-weighted MRI sequence percentile scores were then sorted by condition and the time of with gadolinium contrast agent (n ϭ 19) or combined T2/DWI imaging. The average ranks of the first control images were spin-echo protocols (n ϭ 32). Tumor volumes measured from compared with the second control images and the ranks of the terminal MRI scans correlated significantly with those from the first tumor images were compared with the second tumor same animals measured from histological findings (Fig. 1) images by Student’s t test. The barrel fields and vascular pat- (r2 ϭ 0.85, data not shown). terns reconstructed from the histological findings were aligned Tumor-related hemorrhage and hydrocephalus were clear in carefully to these images (22). T2-weighted spin-echo images. Figure 2 shows images from one of three animals displaying moderate hydrocephalus Histological Analysis and Comparison before significant tumor growth. of MRI and Histological Volume PET imaging experiments with either [18F]FDG or [18F]FLT Fourteen animals were sacrificed and their brains were har- failed to show consistent, differential uptake in the region of vested and stained for correlation with MRI findings. In one the tumor. In eight mice studied with [18F]FDG 2 weeks after subset of mice, seven out of eight animals with tumor injections injection of DBT cells, [18F]FDG uptake in tumor was indistin- in the striatum survived and were imaged serially 5, 8, 12, and guishable from that in normal brain. Only five mice survived to 15 days later. After the last day of imaging, they were sacrificed 3 weeks; it was possible to delineate the tumor in just one of with sodium pentobarbital. Brains were removed from the mice these animals. In the [18F]FLT studies, although the PET images and were placed in 4% paraformaldehyde with 30% sucrose for were consistent with accumulation of [18F]FLT in the tumor 24 to 48 hours at 4ЊC. The specimens were removed and cut in (two out of four mice), interpretation of the images was con- the coronal plane. Fixed specimens were stained with hema- founded by the uptake of the tracer in bone (58).

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 363 JOST ET AL.

A AB

FIGURE 2. The same anatomic slice plane from contrast-enhanced, T1- weighted gradient-echo image (A) and T2-weighted spin-echo image (B) of a mouse 8 days after injection of DBT cells. Hydrocephalus was more evident in T2-weighted spin-echo images and was often seen before tumor was detected. B A

FIGURE 1. A, coronal T2-weighted, spin-echo MRI scan of a mouse just before sacriﬁce 15 days after injecting DBT cells into the right hemisphere. A mass effect, bright tumor, and hemorrhage are evident. B, photomicrograph of a coronal section showing stained tumor and hemorrhage corresponding to the area indicated by the rectangle in the MRI scan in A. Tumor cell inﬁltration at the margins and mass effect are obvious (hematoxylin and eosin). C, higher power image of B showing the rectangular region outlined in A and B.

As an alternative, a microPET strategy to assess cell prolifer- 3 σ FIGURE 3. Scatchard analysis of [ H]RHM-1 binding to the 2 receptors in 11 11 σ ation in vivo using [ C]RHM-I, a C- 2 receptor tracer, was membrane homogenates from DBT cells. A, Scatchard plot showing saturation σ used in four mice. The efficacy of RHM-I as a probe for 2 binding indicating the speciﬁc bound, total bound, and nonspeciﬁc bound receptors was first established in DBT cells by standard protein. B, Scatchard plot showing Kd and Bmax values. Scatchard methods (53). Direct saturation binding studies were carried out using [3H]RHM-1 with membrane homogenates of

364 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MONITORING BRAIN TUMORS IN VIVO WITH SMALL ANIMAL IMAGING

AB AB

CD CD

E FIGURE 5. Similar anatomical slices from contrast-enhanced, T1-weighted multislice MRI scans of an untreated mouse collected 5 days (A), 8 days FIGURE 4. A and C, two slices from contrast-enhanced, T1-weighted MRI (B), 12 days (C), and 15 days (D) scans. B and D, matched microPET/CT images of the cell proliferation marker after injection of DBT cells show 11 σ rapid tumor growth. The tumor in D [ C]RHM-I, a 2 receptor ligand. In this mouse 15 days after injection of DBT cells, localized tracer binding in microPET largely coincides with tumor is hemorrhagic. E, reproduction of D extent seen with MRI. with the tumor outlined manually.

DBT mouse brain tumor xenografts (Fig. 3). The Kd and Bmax shows a plot of tumor volumes versus time for untreated mice values of the receptor-radioligand binding of [3H]RHM-1 were injected with DBT cells in the striatum (therapeutic controls; 0.76 nM and 1190 fmol/mg protein. The high affinity and low 2.5 ϫ 105 cells) and the somatosensory/barrel cortex (IOS con- nonspecific binding indicate RHM-1 is a valuable probe for trols; 2 ϫ 105 cells). Corresponding tumor growth curves after σ evaluating the 2 receptor status of solid DBT tumors. In vivo normalization for differences in tumor volumes are shown in 11 σ uptake of the C- 2 receptor tracer was seen in tumors in four Figure 6B. Growth-rate exponents for therapeutic and IOS of four mice, and localization was confirmed by the coregistra- control mice, computed from measurement of tumor volumes tion of the PET and CT findings. (Image fusion is accurate on postoperative Days 7 and 15, are 0.264 Ϯ 0.085 and within Ϯ0.5 mm using the landmark fiducial method). The res- 0.223 Ϯ 0.023, respectively (mean Ϯ standard deviation; olution of the 11C PET images is limited by the positron range P ϭ 0.36, not significant). of the nuclide (38) and the microPET camera resolution Whisker-stimulation-evoked IOS in the barrel cortex of the (∼1.7 mm) (63). Although lower than the resolution of the cor- injected and contralateral hemispheres (with and without responding MRI scans, the microPET images in Figure 4 show glioma) of DBT-injected animals was recorded at two different that the region of increased tracer uptake closely matches the times. There was no statistical difference between the percentile region identified as tumor in the MRI scan. ranks of IOS taken on two occasions at least 1 week apart (54th Serial MRI measurements characterized tumor growth qual- percentile Ϯ 30 vs. 49th percentile Ϯ 28; P ϭ 0.58, not signifi- itatively and quantitatively (Fig. 5) in untreated (Fig. 6) and cant) in the uninjected hemispheres of 16 animals. However, in BCNU-treated mice (Fig. 7). For both untreated groups, tumor the 23 animals injected with DBT cells that were imaged twice, growth kinetics were consistent. Tumor volumes were calcu- there was significant reduction of IOS signal related to the bulk lated after manual tracing of all image slices containing tumor tumor (64th percentile Ϯ 26 vs. 38th percentile Ϯ 26; P Ͻ 0.001; in each mouse brain at each time point (e.g., Fig. 5E). Figure 6A Fig. 8), as confirmed by postmortem histological examination.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 365 JOST ET AL.

The effects of BCNU in slowing tumor growth were obvious from comparison of serial MRI scans of animals with and without BCNU treatment. Tumor growth kinetics and mean tumor volumes for four different groups of animals (two controls; two treated) are shown in Figure 7. A single dose of BCNU delayed the exponential phase of tumor growth by approximately 7 days, and two doses of BCNU delayed exponential growth by approximately 18 days. Tumors in all animals eventually returned to growth rates approaching those of untreated controls. The difference in tumor volume at 15 days after injection in controls versus the combined group of treated mice (one and two doses of BCNU) was statistically signiﬁcant (P ϭ 0.0026).

DISCUSSION

Progress in adjunctive treatment of human glioblastoma during the past 25 years has been slow (19). Clearly, better understanding of tumor microenvironment and its impact on tumor biology, brain function, and treatment effectiveness is central to developing new therapeutic strategies. Treatments that limit tumor cell proliferation, restrict metastases, increase apoptosis, or specifically target tumor cells with novel agents are logical targets for study. Small-animal imaging increases the power of such investigations (32, 37, 46). In the current study, we injected a mouse tumor cell line intracranially, and tumor cells grew in the microenvironment of the normal brain. The combination of imaging techniques described here allows longitudinal, in vivo assessment of tumor growth, brain function, and tumor biology. The power of small-animal MRI scans for measuring the effect of therapeutic interventions has been previously demonstrated (12, 36, 52). The present study is novel because a combination of FIGURE 6. Tumor volumes measured in two different groups of mice 1 and 2 Ϯ imaging methods was used to evaluate gliomas in mice. weeks after injection of DBT cells. A, graph showing averages ( standard DBT cells have morphological features similar to human deviation) demonstrating that tumors in the striatum are larger (therapeutic glioblastoma cells, are motile, are glial fibrillary acidic protein controls, 2.5 ϫ 105 DBT cells injected) than those in somatosensory cortex (IOS controls, 2 ϫ 105 DBT cells). B, graph showing that after normalization positive, and develop multicellular colonies when suspended (see Materials and Methods), the data from the two groups correlate. in agar. They have been characterized in a previous in vivo study evaluating rat C6 cells and DBT cells injected into barrel cortex of rats and mice, respectively (57). In the present work, as in previous studies, intracranial neoplasms developed in all animals with invasion away from bulk tumor (not shown). Invasion follows arteries, the principal pathway for the dis- semination of glioma cells (4, 24, 43, 57). Although previous studies have demonstrated correlation between MRI results and histological measures of tumor volume (16), such a correlation had not been established previously for the DBT glioblastoma model. This study demonstrated a significant, direct correlation between MRI results and histological measurements of tumor volumes. The potential for PET as an imaging tool to detect specific molecular markers or receptors for brain tumors in vivo is significant because changes seen with anatomical imaging are not 11 σ 11 necessarily tumor specific. The C-labeled 2 tracer [ C]RHM- I is under development as a PET radiotracer to image prolifer- FIGURE 7. Relative tumor volumes in the different control groups plotted for ation in solid tumors (40, 66, 75). Vilner and Bowen (68) and σ comparison with data from the two different treatment paradigms. Error bars, Vilner et al. (69) reported a high density of 2 receptors in a Ϯ standard deviation, except for BCNU 2ϫ, which indicate data ranges. broad panel of human and rodent tumors, including human

366 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MONITORING BRAIN TUMORS IN VIVO WITH SMALL ANIMAL IMAGING

FIGURE 8. IOSs from a control and DBT-injected hemisphere. A, surface of controls. The projected size of the tumor (purple) was estimated from growth the intact left hemisphere (projected as the right for comparison) of the mouse rates (see Fig. 6). The DBT cell injection site is indicated (asterisk). The ori- brain through the cranium. The scale bar applies to all panels. The compass entation is the same as in C. E, surface of the right hemisphere 26 days after indicates medial (m) and anterior (a) in this and in panels C and E. B, after injecting DBT cells. Brain swelling is evident. The center of the tumor out- contralateral whisker stimulation, a vigorous response (warmer colors indicate lined postmortem is marked (asterisk). The arrowhead points to the proxi- bigger changes) is evoked. Postmortem histological analysis localized the sig- mal MCA. F, whisker activation of the brain is reduced and conﬁned to some nal over stimulated barrels in layer IV. Selected barrels (A1, C3, and E5) are whisker barrels posterolateral to the tumor. On postmortem histological analy- labeled; the red lines indicate the principal branches of the MCA. The orien- sis, the pattern of the whisker barrels was distorted by the tumor (asterisk, tation is the same as in A. C, surface of the right hemisphere 7 days after DBT dashed line). The orientation is the same as in E. MCA, a lateral branch of cells were injected (asterisk). The arrowhead indicates the proximal MCA. the middle cerebral artery. D, whisker stimulation does not activate as much cortex as vigorously as in

(U-138MG glioblastoma) and rat (C6 glioma) brain tumors, brain. We did not observe a significant increase in [18F]FDG σ whereas Al-Nabulsi et al. (2) measured the density of 2 recep- uptake in the regions of mouse brain with DBT tumors, except tors in 9L glioma cells to be approximately 300,000 in one mouse at a late stage of tumor growth. Radiolabeled receptors/cell. In this study, microPET scans performed using nucleosides, such as [18F]FLT, which measure the salvage path- [11C]RHM-I showed considerable uptake of tracer in the region way of DNA synthesis, are under investigation as potential of the brain correlating with the location of the tumor in all four markers for cell proliferation (47, 58, 59). In this study, although mice studied. [11C]RHM-I delineates tumors without accumu- it was possible to see accumulation of [18F]FLT in regions con- lation of the tracer in normal brain (as with [18F]FDG) or bone taining tumor in two of four mice, the interpretation of the PET (as with [18F]FLT). The excellent qualitative agreement of the image was confounded by [18F]FLT uptake in the cranium. σ regions of tracer uptake (PET) and tumor growth (MRI) empha- Although definitive studies documenting that 2 receptor sizes the power of combining these imaging methods to char- binding correlates quantitatively with histological measures of acterize brain tumors. cell proliferation (e.g., Ki-67 or bromodeoxyuridine labeling [18F]FDG PET has been used widely in the study of brain studies) should be carried out, the present encouraging results σ tumors and was, therefore, important to evaluate in the present suggest that microPET imaging with 2 tracers will have useful- study. Clinical applications of PET imaging to human glioblas- ness for in vivo monitoring of specific tumor markers statically 18 σ toma therapy principally use [ F]FDG PET to distinguish and in response to treatment paradigms. Because 2 receptors recurrent tumor from radiation necrosis, a distinction that is also are localized in brain regions, such as the hippocampus, difficult to make with CT or MRI examination. However, caution is appropriate in attributing all nervous system labeling [18F]FDG PET lacks specificity for GBM (or any other tumor with this radiopharmaceutical to proliferation (8, 27). type). Increased uptake is confounded by other processes such Serial MRI measurements provide accurate, in vivo assess- as seizures (70) and the high resting metabolism of normal ment of tumor growth. In this study, brain tumor growth was

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 367 JOST ET AL.

measured for two different groups of untreated, control mice. and accurate measure of therapeutic efficacy than previous The difference in number of cells injected resulted in different standard models. tumor volumes in these two sets of animals. However, normal- The aim of this proof of principle study was to demonstrate ized tumor volume versus time curves for these two sets of that these imaging methods can be used with relative efficiency control mice were virtually superimposed, indicating nearly to test the effects of agents known to limit tumor growth and identical growth rates in untreated animals that are independ- function. Future studies will evaluate the effects of more cur- ent of injection site. rent and investigational therapeutic interventions. With these The reduction of optical signal elicited by cortical activation collective imaging tools, it should be possible to detect early within tumor-involved functional cortex is consistent with neu- effects and to study mechanisms of novel interventions, to ronal dysfunction in patients with tumors in eloquent cortex determine dose-response relationships, and to quantitate the (17). In these instances, it is critical to understand the mecha- impact of possible therapies on tumor growth with better sta- nisms involved in loss of function so that therapies can be tistical power and using fewer animals. developed that promote recovery of brain function while elim- The same technologies used to assess tumor growth, func- inating the tumor. Progressive loss of whisker-stimulated IOS tion, and biology in vivo in a simple DBT cell model can be was observed in this study as gliomas enlarged. We filtered applied to animal models that more directly reduplicate the light at 520 to 560 nm and primarily evaluated hemoglobin genotypic and histological features of human brain tumors. concentration related to changes in local vascular volume and These methods can be used to assess the efficacy of novel ther- hematocrit produced by blood flow. Several factors may con- apeutic agents and the impact of current treatments on biology tribute to IOS changes with glioma growth. These have not yet and brain function and to test novel imaging methods that offer been delineated but clearly include local redistribution of blood direct translation to humans. Combining imaging methods flow from brain to tumor; compression of neural tissues lead- offers a strategy to characterize human tumors that powerfully ing to embarrassed metabolism; functional disconnection from augments current genetic and biological studies. other cortex, thalamus, and descending targets; systemic In humans, MRI is used predominately to evaluate tumor responses to tumors with paraneoplastic effects; and, poten- volume and localization of function, or both, before surgical tially, compounds released by tumors that affect nervous func- and other therapeutic interventions. Investigational studies tion (18, 49). suggest, however, that newer MRI technologies, including dif- In models of stroke, IOS has been used to identify specific fusion methods (11, 31, 50) and dynamic contrast enhancement brain regions functionally to guide studies designed to deter- (10, 25, 41, 64), will provide important insights into the molec- mine the extent of pathological characteristics and their impact ular aspects of neuro-oncology. Imaging techniques that focus on brain function (9, 71–73). The relatively late appearance of not just on the quantitative measurement of tumor volume, symptoms (e.g., dysfunction, seizures) with many brain tumors but also on the delineation of tumor and tissue biology and the suggests compensatory neuroplasticity. Like the studies of the in vivo effects of therapeutic intervention, offer great promise. stroke model, IOS can guide studies to determine the impact of Such research will facilitate longitudinal in vivo assessment of tumors on brain structure and function. The ability to assess the tumor biology, tumor growth, and their interactions with the function of the whisker barrel cortex in vivo with optical imag- nervous system in vivo. These applications will be directly ing (20, 21, 33) and functional MRI (79) makes this model ideal dependent on the integration of MRI, PET, and optical imaging for evaluating alterations in neurological function in response technologies described in this article. Twenty-five years ago, to tumor growth and invasion. Optical imaging of rat and Shapiro et al. (56) suggested that the subcutaneous and human gliomas using contrast agents has demonstrated prom- intracranial implantation of brain tumor cells into nude mice ise in identifying tumor margins for resection (29, 30). would contribute to revolutionizing the field of neuro- The effects of BCNU treatment on tumor growth were eval- oncology. We think that the development of small-animal uated. BCNU is highly lipophilic and freely crosses the blood- molecular and functional imaging techniques will lead to the brain barrier to reach high concentrations in cerebrospinal fulfillment of Shapiro et al.’s vision, ultimately to the benefit of fluid (23, 74). It is a nitrosourea compound that nonspecifi- patients with glioblastoma. cally inhibits deoxyribonucleic acid, ribonucleic acid, and protein synthesis. BCNU is an established cytotoxic agent that has been used in the treatment of glioblastoma in humans. REFERENCES BCNU at standard human doses (5–10 mg/kg) did not show 1. Adzamli K, Yablonskiy DA, Chicoine MR, Won EK, Galen KP, Zahner MC, a significant effect in earlier small-animal research (32, 35, 37), Woolsey TA, Ackerman JJ: Albumin-binding MR blood pool agents as MRI but higher doses have been reported to be effective (46). Serial contrast agents in an intracranial mouse glioma model. Magn Reson Med imaging was a sensitive measure of the therapeutic impact of 49:586–590, 2003. a 10-mg/kg dose of BCNU in our mouse GBM model. A sin- 2. Al-Nabulsi I, Mach RH, Wang LM, Wallen CA, Keng PC, Sten K, Childers SR, gle BCNU dose of 10 mg/kg, administered 1 week after Wheeler KT: Effect of ploidy, recruitment, environmental factors, and tamox- ifen treatment on the expression of sigma-2 receptors in proliferating and qui- tumor implantation, significantly slowed tumor growth escent tumour cells. Br J Cancer 81:925–933, 1999. (Fig. 7). This indicates that MRI and other imaging methods, 3. Benedetti S, Bruzzone MG, Pollo B, DiMeco F, Magrassi L, Pirola B, Cirenei combined with an intracranial model, offer a more sensitive N, Colombo MP, Finocchiaro G: Eradication of rat malignant gliomas by

368 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MONITORING BRAIN TUMORS IN VIVO WITH SMALL ANIMAL IMAGING

:retroviral mediated in vivo delivery of the interleukin 4 gene. Cancer Res 26. Grierson JR, Shields AF: Radiosynthesis of 3؅-deoxy-3؅-[18F]fluorothymidine 59:645–652, 1999. [18F]FLT for imaging of cellular proliferation in vivo. Nuc Med Biol 4. Bernstein JJ, Woodard CA: Glioblastoma cells do not intravasate blood ves- 27:143–156, 2000. sels. Neurosurgery 36:124–132, 1995. 27. Gundlach AL, Largent BL, Snyder SH: Autoradiographic localization of 5. Bock NA, Konyer NB, Henkelman RM: Multiple-mouse MRI. Magn Reson sigma receptor binding sites in guinea pig and rat central nervous system Med 49:158–167, 2003. with (+)3h-3-(3-hydroxyphenyl)-n-(1-propyl)piperidine. J Neurosci 6. Bock NA, Zadeh G, Davidson LM, Qian B, Sled JG, Guha A, Henkelman 6:1757–1770, 1986. RM: High-resolution longitudinal screening with magnetic resonance imag- 28. Gutmann DH, Baker SJ, Giovannini M, Garbow J, Weiss W: Mouse models of ing on a murine brain cancer model. Neoplasia 5:546–554, 2003. human cancer consortium symposium on nervous system tumors. Cancer 7. Brockmann MA, Ulmer S, Leppert J, Nadrowitz R, Wuestenberg R, Nolte I, Res 63:3001–3004, 2003. Petersen D, Groden C, Giese A, Gottschalk S: Analysis of mouse brain using 29. Haglund MM, Berger MS, Hochman DW: Enhanced optical imaging of a clinical 1.5T scanner and standard small loop surface coil. Brain Res human gliomas and tumor margins. Neurosurgery 38:308–317, 1996. 1068:138–142, 2006. 30. Haglund MM, Hochman DW, Spence AM, Berger MS: Enhanced optical 8. Cao J, Kulkarni SS, Husbands SM, Bowen WD, Williams W, Kopajtic T, Katz imaging of rat gliomas and tumor margins. Neurosurgery 35:930–941, 1994. JL, George C, Newman AH: Dual probes for the dopamine transporter and 31. Hall DE, Moffat BA, Stojanovska J, Johnson TD, Li Z, Hamstra DA, sigma1 receptors: Novel piperazinyl alkyl-bis(4’-fluorophenyl)amine ana- Rehemtulla A, Chenevert TL, Carter J, Pietronigo D, Ross BD: Therapeutic efficacy of DTI-015 using diffusion magnetic resonance imaging as an early logues as potential cocaine-abuse therapeutic agents. J Med Chem surrogate marker. Clin Cancer Res 10:7852–7859, 2004. 46:2589–2598, 2003. 32. Hanley ML, Elion GB, Colvin OM, Modrich PL, Keir S, Adams DJ, Bigner DD, 9. Carmichael ST, Wei L, Rovainen CM, Woolsey TA: New patterns of intracor- Friedman HS: Therapeutic efficacy of vinorelbine against pediatric and adult tical projections after focal cortical stroke. Neurobiol Dis 8:910–922, 2001. central nervous system tumors. Cancer Chemother Pharmacol 42:479–482, 10. Cha S, Johnson G, Wadghiri YZ, Jin O, Babb J, Zagzag D, Turnbull DH: 1998. Dynamic, contrast-enhanced perfusion MRI in mouse gliomas: Correlation 33. Hodge CJ Jr, Stevens RT, Newman H, Merola J, Chu C: Identification of func- with histopathology. Magn Reson Med 49:848–855, 2003. tioning cortex using cortical optical imaging. Neurosurgery 41:1137–1145, 11. Chenevert TL, McKeever PE, Ross BD: Monitoring early response of experi- 1997. mental brain tumors to therapy using diffusion magnetic resonance imaging. 34. Kelly PJ, Daumas-Duport C, Kispert DB, Kall BA, Scheithauer BW, Illig JJ: Clin Cancer Res 3:1457–1466, 1997. Image based stereotaxic serial biopsies in untreated intracranial glial neo- 12. Chenevert TL, Stegman LD, Taylor JM, Robertson PL, Greenberg HS, plasms. J Neurosurg 66:865–874, 1987. Rehemtulla A, Ross BD: Diffusion magnetic resonance imaging: An early 35. Khil MS, Kolozsvary A, Apple M, Kim JH: Increased tumor cures using com- surrogate marker of therapeutic efficacy in brain tumors. J Natl Cancer Inst bined radiosurgery and BCNU in the treatment of 9l glioma in the rat brain. 92:2029–2036, 2000. Int J Radiat Oncol Biol Phys 47:511–516, 2000. 13. Chicoine MR, Silbergeld DL: Assessment of brain tumor cell motility in vivo 36. Kim B, Chenevert TL, Ross BD: Growth kinetics and treatment response of and in vitro. J Neurosurg 82:615–622, 1995. the intracerebral rat 9L brain tumor model: A quantitative in vivo study using 14. Chicoine MR, Silbergeld DL: Invading C6 glioma cells maintain tumorgeni- magnetic resonance imaging. Clin Cancer Res 1:643–650, 1995. city. J Neurosurg 83:665–671, 1995. 37. Kreklau EL, Pollok KE, Bailey BJ, Liu N, Hartwell JR, Williams DA, Erickson 15. Chicoine MR, Won EK, Zahner M: Intratumoral injection of lipopolysaccha- LC: Hematopoietic expression of O(6)-methylguanine DNA methyltransferase- ride causes regression of subcutaneously implanted mouse glioblastoma mul- P140K allows intensive treatment of human glioma xenografts with combina- tiforme. Neurosurgery 48:607–615, 2001. tion O(6)-benzylguanine and 1, 3-bis-(2-chloroethyl)-1-nitrosourea. Mol 16. Cornelissen B, Kersemans V, Jans L, Staelens J, Oltenfreiter R, Thonissen T, Cancer Ther 2:1321–1329, 2003. Achten E, Slegers G: Comparison between 1 T MRI and non-MRI based vol- 38. Laforest R, Rowland DJ, Welch MJ: MicroPET imaging with nonconventional umetry in inoculated tumours in mice. Br J Radiol 78:338–342, 2005. isotopes. IEEE Trans Nucl Sci 49:2119–2126, 2002. 17. Cushing H: Surgery of the head, in Keen WW (ed): Surgery: Its Principles and 39. Lewis JS, Achilefu S, Garbow JR, Laforest R, Welch MJ: Small animal imaging: Practice. Philadelphia, W.B. Saunders, 1919, pp 17–276. Current technologies and perspectives for oncological imaging. Eur J Cancer 18. Darnell RB, Posner JB: Paraneoplastic syndromes involving the nervous sys- 38:2173–2188, 2002. tem. N Engl J Med 349:1543–1554, 2003. 40. Mach RH, Smith CR, Al-Nabulsi I, Whirrett BR, Childer SR, Wheeler KT: 19. Davis FG, Freels S, Grutsch J, Barlas S, Brem S: Survival rates in patients Sigma 2 receptors as potential biomarkers of proliferation in breast cancer. with primary malignant brain tumors stratified by patient age and tumor his- Cancer Res 57:156–161, 1997. tological type: An analysis based on Surveillance, Epidemiology, and End 41. Maxwell RJ, Wilson J, Prise VE, Vojnovic B, Rustin GJ, Lodge MA, Tozer GM: Results (SEER) data, 1973–1991. J Neurosurg 88:1–10, 1998. Evaluation of the anti-vascular effects of combretastatin in rodent tumours by 20. Dowling JL, Henegar MM, Liu D, Rovainen CM, Woolsey TA: Rapid optical dynamic contrast enhanced MRI. NMR Biomed 15:89–98, 2002. imaging of whisker responses in the rat barrel cortex. J Neurosci Methods 42. Nabi HA, Zubeldia JM: Clinical applications of 18F-FDG in oncology. J Nucl 66:113–122, 1996. Med Technol 30:3–9, 2002. 21. Erinjeri J, Lui D, Woolsey TA: Timing of intrinsic optical and pial vessel diam- 43. Nagano N, Sasaki H, Aoyagi M, Hirakawa K: Invasion of experimental rat eter changes during activation of mouse barrel cortex. Soc Neurosci Abs brain tumor: Early morphological changes following microinjections of C6 28:632, 1998. glioma cells. Acta Neuropathol 86:117–125, 1993. 22. Erinjeri JP, Woolsey TA: Spatial integration of vascular changes with neural 44. Nelson AL, Algon SA, Munasinghe J, Graves O, Goumnerova L, Burstein D, activity in mouse cortex. J Cereb Blood Flow Metab 22:353–360, 2002. Pomeroy SL, Kim JY: Magnetic resonance imaging of patched heterozygous 23. Fewer D, Wilson CB, Boldrey EB, Enot KJ, Powell MR: The chemotherapy of and xenografted mouse brain tumors. J Neurooncol 62:259–267, 2003. brain tumors. Clinical experience with carmustine (BCNU) and vincristine. 45. Peiper RO: Defined human cellular systems in the study of glioma develop- JAMA 222:549–552, 1972. ment. Front Biosci 8:S19–S27, 2003. 24. Giese A, Westphal M: Glioma invasion in the central nervous system. 46. Plowman J, Waud WR, Koutsoukos AD, Rubinstein LV, Moore TD, Grever Neurosurgery 39:235–252, 1996. MR: Preclinical antitumor activity of temozolomide in mice: Efficacy against 25. Gossmann A, Helbich TH, Kuriyama N, Ostrowitzki S, Roberts TP, Shames human brain tumor xenografts and synergism with 1, 3-bis(2-chloroethyl)-1- DM, van Bruggen N, Wendland MF, Israel MA, Brasch RC: Dynamic contrast- nitrosourea. Cancer Res 54:3793–3799, 1994. enhanced magnetic resonance imaging as a surrogate marker of tumor 47. Ponde DE, Oyama N, Welch MJ, McCarthy TM: Synthesis of fluoro thymidine response to anti-angiogenic therapy in a xenograft model of glioblastoma ([F-18]FLT as radiotracer of tumors. Abstr Pap Am Chem Soc 223:081–Nucl multiforme. J Magn Reson Imaging 15:233–240, 2002. Part 2, 2002.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 369 JOST ET AL.

48. Poptani H, Duvvuri U, Miller CG, Mancuso A, Chargundla S, Fraser NW, Mechanisms for Neuromodulation and Neuroprotection. Ann Arbor, NPP Books, Glickson JD, Leigh JS, Reddy R: T1rho imaging of murine brain tumors at 4 1992, pp 341–353. T. Acad Radiol 8:42–47, 2001. 69. Vilner BJ, John CS, Bowen WD: Sigma-1 and sigma-2 receptors are expressed 49. Posner JB: Immunology of paraneoplastic syndromes: Overview. Ann N Y in a wide variety of human and rodent tumour cell lines. Cancer Res Acad Sci 998:178–186, 2003. 55:408–413, 1995. 50. Ross BD, Moffat BA, Lawrence TS, Mukherji SK, Gebarski SS, Quint DJ, 70. Weber W, Bartenstein P, Gross MW, Kinzel D, Daschner H, Feldmann HJ, Johnson TD, Junck L, Robertson PL, Muraszko KM, Dong Q, Meyer CR, Reidel G, Ziegler SI, Lumenta C, Molls M, Schwaiger M: Fluorine-18-FDG Bland PH, McConville P, Geng H, Rehemtulla A, Chenevert TL: Evaluation of PET and iodine-123-IMT SPECT in the evaluation of brain tumors. J Nucl cancer therapy using diffusion magnetic resonance imaging. Mol Cancer Med 38:802–808, 1997. Ther 2:581–587, 2003. 71. Wei L, Erinjeri J, Zeng JS, Rovainen C, Woolsey TA: Correlated optical imaging, electrical activity and 2DG reveal functional plasticity after focal ischemia 51. Rubin JB, Kung AL, Klein RS, Chan JA, Sun Y, Schmidt K, Kieran MW, Luster in rat whisker barrel cortex. Soc Neurosci Abstr 26:1033, 2000. AD, Segal RA: A small-molecule antagonist of CXCR4 inhibits intracranial 72. Wei L, Erinjeri JP, Rovainen CM, Woolsey TA: Functional recovery after local growth of primary brain tumors. Proc Natl Acad Sci U S A 100:13513–13518, ischemia revealed by online optical imaging. Soc Neurosci Abstr 23:1578, 1999. 2003. 73. Wei L, Rovainen CM, Woolsey TA: Ministrokes in rat barrel cortex. Stroke 52. Rustamzadeh E, Hall WA, Todhunter DA, Low WC, Liu H, Panoskaltsis- 26:1459–1462, 1995. Mortari A, Vallera DA: Intracranial therapy of glioblastoma with the fusion 74. Weingart J, Brem H: Biology and therapy of glial tumors. Curr Opin Neurol protein DTIL13 in immunodeficient mice. Int J Cancer 118:2594–2601, 2006. Neurosurg 5:808–812, 1992. 53. Scatchard G: The attractions of proteins for small molecules and ions. Ann 75. Wheeler KT, Wang LM, Wallen CA, Childers SR, Cline JM, Keng PC, Mach N Y Acad Sci 51:662–672, 1949. RH: Sigma-2 receptors as a biomarker of proliferation in solid tumours. Br J 54. Scherer HJ: The forms of growth in gliomas and their practical significance. Cancer 82:1223–1232, 2000. Brain 63:1–35, 1940. 76. Won EK, Zahner MC, Chicoine MR, Grant EA, Gore P: Analysis of the anti- 55. Schmidt KF, Ziu M, Schmidt NO, Vaghasia P, Cargioli TG, Doshi S, Albert MS, tumoral mechanisms of lipopolysaccharide against glioblastoma multiforme. Black PM, Carroll RS, Sun Y: Volume reconstruction techniques improve the Anticancer Drugs 14:457–466, 2003. correlation between histological and in vivo tumor volume measurements in 77. Xu J, Tu Z, Jones LA, Vangveravong S, Wheeler KT, Mach RH: [(3)H]N-[4-(3, mouse models of human gliomas. J Neurooncol 68:207–215, 2004. 4-dihydro-6, 7-dimethoxyisoquinolin-2(1H)-yl)butyl]-2-methoxy-5-methyl- 56. Shapiro WR, Basler GA, Chernik NL, Posner JB: Human brain tumor trans- benzamide: A novel sigma-2 receptor probe. Eur J Pharmacol 525:8–17, 2005. plantation into nude mice. J Natl Cancer Inst 62:447–453, 1979. 78. Xu S, Gade TP, Matei C, Zakian K, Alfieri AA, Hu X, Holland EC, 57. Sherburn EW, Wanebo JE, Kim P, Song SK, Chicoine MR, Woolsey TA: Soghomonian S, Tjuvajev J, Ballon D, Koutcher JA: In vivo multiple-mouse Glilomas in rodent whisker barrel cortex: A new tumor model. J Neurosurg imaging at 1.5 T. Magn Reson Med 49:551–557, 2003. 91:814–821, 1999. 79. Yang X, Hyder F, Shulman RG: Functional MRI BOLD signals coincides with 58. Shields AF, Grierson JR, Dohmen BM, Machulla HJ, Stayanoff JC, Lawhorn- electrical activity in the rat. Magn Reson Med 38:874–877, 1997. Crews JM, Obradovich JE, Muzik O, Mangner TJ: Imaging proliferation in vivo with [F-18]FLT and positron emission tomography. Nat Med Acknowledgments 4:1334–1336, 1998. Supported by National Institutes of Health grants NS28781, NS49048, and 59. Shields AF, Grierson JR, Stayanoff JC, Lawhorn-Crews JM, Obradovich J, CA102869; a National Cancer Institute/National Institutes of Health Small Muzik O, Mangner T: [F-18] FLT can be used to image cell proliferation in Animal Imaging Resource Program (SAIRP) grant (R24 CA83060); the Alvin J. vivo. J Nucl Med 39:228, 1998. Siteman Cancer Center at Washington University in St. Louis, a National Cancer 60. Silbergeld DL, Chicoine MR: Isolation and characterization of human malignant Institute Comprehensive Cancer Center (P30 CA91842); an award from the glioma cells from histologically normal brain. J Neurosurg 86:525–531, 1997. Spastic Paralysis Foundation of the Illinois-Eastern Iowa District of the Kiwanis 61. Som P, Atkins HL, Bandoypadhyay D, Fowler JS, MacGregor RR, Matsui K, International; and, in part, by generous donations of the family and friends of Oster ZH, Sacker DF, Shiue CY, Turner H, Wan CN, Wolf AP, Zabinski SV: A Deborah Forbush. We thank Michael Zahner for preparation of the DBT cells; fluorinated glucose analog, 2-fluoro-2-deoxy-D-glucose (F-18): Nontoxic Joseph P. Erinjeri, M.D., Ph.D., for IOS technical innovations; Scott P. Leary, tracer for rapid tumor detection. J Nucl Med 21:670–675, 1980. M.D., for histological reconstructions; Kathryn Diekmann, Brad Miller, Ph.D., 62. Surawicz TS, Davis F, Freels S, Laws ER Jr, Menck HR: Brain tumor survival: and John Christensen for evaluating IOS images; Douglas J. Rowland, Ph.D., and the technical support of John Engelbach, Nicole Fettig, and Jerrel Rutlin in the Results from the National Cancer Data Base. J Neurooncol 40:151–160, 1998. Washington University small-animal PET laboratory; and Zhude Tu, Ph.D., and 63. Tai YC, Ruangma A, Rowland DJ, Siegel S, Newport DF Chow PL, Laforest R: Datta Ponde, Ph.D., for synthesizing [11C]RHM-I and [18F]FLT. Performance evaluation of the microPET Focus: A third-generation microPET scanner dedicated to animal imaging. J Nucl Med 46:455–463, 2005. 64. Tofts PS, Brix G, Buckley DL, Evelhoch JL, Henderson E, Knopp MV, Larsson COMMENTS HB, Lee TY, Mayr NA, Parker GJ, Port RE, Taylor J, Weisskoff RM: Estimating kinetic parameters from dynamic contrast-enhanced T1-weighted MRI of a his study describes serial multimodality in vivo imaging of diffusable tracer: Standardized quantities and symbols. J Magn Reson Tglioblastoma-bearing mice by magnetic resonance imaging (MRI), Imaging 10:223–232, 1999. positron-emission tomography (PET), and optical imaging. The authors 65. Tooth HH: Some observations on the growth and survival period of intracra- conclude that tumor volumetric measurements calculated from MRI nial tumours, based on the records of 500 cases, with special reference to the scans correlate closely with tumor volumes measured by histological pathology of the gliomata. Brain 35:61–108, 1912. sectioning, that significant binding of 11C-RHM-I occurs in growing 66. Tu Z, Dence CS, Ponde DE, Jones L, Wheeler KT, Welch MJ, Mach MH: tumors, and that intrinsic optical signals induced by whisker stimula- Carbon-11 labeled sigma-2 receptor ligands for imaging breast cancer. Nucl tion are reduced and displaced over time by a growing tumor. Med Biol 32:423–430, 2005. The authors address an important topic. New research into tumor 67. van Furth WR, Laughlin S, Taylor MD, Salhia B, Mainprize M, Henkelman M, Cusimano MD, Ackerley C, Rutka JT: Imaging of murine brain tumors using molecular pathways is providing a plethora of new drugs which must a 1.5 tesla clinical MRI scanner. Can J Neurol Sci 30:326–332, 2003. be assessed in animal models before the initiation of human clinical tri- 68. Vilner BJ, Bowen WD: Characterization of sigma-like binding properties of als. The traditional method of assessing chemotherapeutic efficacy in NB41A3, S-20Y, and N1E-115 neuroblastoma, C6 glioma, and NG 108-15 animal models has been by euthanasia and examination of histological neuroblastoma-glioma hybrid cells: Further evidence for sigma-2 receptors, in sections. The advent of noninvasive imaging methods that permit serial Kamenka JM, Domino EF (eds): Multiple Sigma and PCP Receptor Ligands: imaging of individual animals over time has the potential of providing

370 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com MONITORING BRAIN TUMORS IN VIVO WITH SMALL ANIMAL IMAGING

a far more efﬁcient method of drug development and testing than tra- ment of tumor progression, effect on normal brain function, and ditional methods can provide. tumor regression in response to treatment. Previous work has been The primary weakness of this report is that it describes little that is limited with respect to these multiple factors of tumor biology. In new. Many groups have already reported serial MRI and PET scanning this work, Jost et al. have taken the next logical step and made a of tumor-bearing mice, both as a tumor grows and as it undergoes potentially significant contribution to the advancement of neuro- treatment; many groups have also reported volumetric measurements oncology by integrating a multidimensional approach. Tumor proof these tumors in conjunction with MRI. The whisker stimulation/ gression was assessed by serial MRI measurements as well as PET optical imaging paradigm reported in this study has already been imaging specifically targeting cell proliferation by using the Û2- described by the authors previously, and it is not clear what it offers receptor ligand 11C-RHM-I. Likewise, intrinsic optical signal meas- over the other imaging methods used in this report. The authors’ tumor urements were serially imaged to assess the effect of tumor growth on model, which involves implantation of tumor cells, is also not state-of- brain function. The overall results are very promising, even though the-art; genetically engineered mouse models offer a far richer environ- the MRI technique used was very basic and the images at the rela- ment in which to develop and test therapeutic agents and to investigate tively low field strength of 4.7 T were only acceptable. This work basic questions about tumor biology than the approach described here. points to further investigations at significantly higher field strengths (9.4 T studies have been safely performed in humans), particularly Elizabeth Bullitt with MRI techniques being studied in vivo in humans, such as diffu- Allan Friedman sion tensor imaging and susceptibility-weighted techniques, includ- Durham, North Carolina ing MRI perfusion and permeability imaging. n vivo imaging of small animal tumor models continues to be an Paul E. Kim Iimportant relative limitation in studying the behavior of tumors Neuroradiologist with multidimensional relevancy; principally, the longitudinal assess- Los Angeles, California

Joe Louis (right) defends his heavyweight title against Tami Mauriello at Yankee Stadium, September 18, 1946. Louis defeated Mauriello with a knockout two minutes and nine seconds into the ﬁrst round. EXPERIMENTAL STUDIES

Michael R. Chicoine, M.D. THE IN VIVO ANTITUMORAL EFFECTS OF Department of Neurosurgery, Washington University School of Medicine, LIPOPOLYSACCHARIDE AGAINST GLIOBLASTOMA Saint Louis, Missouri MULTIFORME ARE MEDIATED IN PART BY Michael Zahner, B.S. TOLL-LIKE RECEPTOR 4 Department of Neurosurgery, Washington University School of Medicine, OBJECTIVE: Toll-like receptor 4 (Tlr-4) mediates many biological effects of lipopolysac- Saint Louis, Missouri charide (LPS), which has antitumoral effects on glioblastoma both in vivo and in vitro. Eun Kyung Won, M.D. However, the precise role of Tlr-4 in these antitumoral effects remains unknown. Department of Neurosurgery, METHODS: The role of Tlr-4 in the antitumoral effect of LPS on glioblastomas was Washington University assessed in wild-type BALB/c mice and in Tlr-4 knockout (KO) BALB/c mice. Mice were School of Medicine, Saint Louis, Missouri implanted with DBT glioblastoma cells intracranially or subcutaneously, were treated with intratumoral LPS, and were assessed by histopathological examination for degrees Ricky R. Kalra, B.A. of tumor progression and inflammation. Flow cytometry and Western blotting with anti- Department of Neurosurgery, bodies to the Tlr-4 receptor and flow cytometry to the related CD14 moiety were per- Washington University formed to quantitate the expression levels of these two receptors by glioblastoma cells. School of Medicine, Saint Louis, Missouri RESULTS: For subcutaneous tumors, LPS caused near complete tumor elimination in wild-type mice, but only a 50% reduction in Tlr-4 KO mice. For mice implanted with Tetsuya Kitamura, D.D.S. intracranial glioblastomas, LPS increased survival times modestly in wild-type mice, Department of Radiation Oncology, but showed no benefit in the Tlr-4 KO mice. There were no histological differences Washington University among wild-type and Tlr-4 KO mice, except for tumor size. In both models, an early School of Medicine, Saint Louis, Missouri neutrophilic and later macrophage-rich inflammatory infiltrate were seen after LPS administration. Quantitative flow cytometry and Western blotting showed no Tlr-4 Arie Perry, M.D. receptor or CD14 expression in murine and human glioblastoma cells in vitro, and Department of Pathology, Western blotting suggested that Tlr-4 effects are mediated by nontumoral elements such Washington University as microglia and inflammatory cells. School of Medicine, Saint Louis, Missouri CONCLUSION: LPS-induced antitumoral effects on glioblastoma multiforme are mediated, in part, by the Tlr-4 receptor. Further understanding of this process may lead to Ryuji Higashikubo, Ph.D. novel treatment strategies for this uniformly fatal disease. Department of Radiation Oncology, KEY WORDS: Glioblastoma multiforme, Lipopolysaccharide, Mice, Toll-like receptor 4 Washington University School of Medicine, Neurosurgery 60:372–381, 2007 DOI: 10.1227/01.NEU.0000249280.61761.2E www.neurosurgery-online.com Saint Louis, Missouri

The Alvin J. Siteman Cancer Center, Washington University School of Medicine lioblastoma multiforme (GBM) re- 1990s (20, 29, 33), there has been an explosion (MRC, MZ, EKW, RRK, TK, AP, RH) mains among the most deadly of all of research showing the role of the toll-like cancers. Mean patient survival, even receptors (TLRs), the associated signal trans- Reprint requests: G with aggressive treatments, including surgery, duction pathways, and the downstream Michael R. Chicoine, M.D., κ Department of Neurosurgery, radiation therapy, and chemotherapy, is typi- nuclear factor -B transcription factors in var- Washington University cally 12 months or less (15, 17). Previous in ious immune-mediated processes (1, 7, 12, 20, School of Medicine, vivo investigations have demonstrated pow- 21, 23, 37). The TLRs are a class of evolutionar- 660 South Euclid, erful antitumoral effects of lipopolysaccharide ily conserved transmembrane receptors Campus Box 8057, belonging to the interleukin-1 receptor/TLR St. Louis, MO 63110. (LPS) on GBM, mediated in part by the tumor- Email: [email protected] bearing host’s immune system (6, 38). superfamily that has been described recently However, the mechanisms of this antitumoral (1, 4, 7, 12, 39). The toll-like nomenclature Received, February 24, 2005. response have not been fully elucidated. Since relates to the homology of these molecules to Accepted, October 4, 2006. its ﬁrst descriptions in vertebrates in the late the toll protein that mediates various physio-

372 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com TOLL-LIKE RECEPTOR 4 AND LIPOPOLYSACCHARIDE ANTITUMORAL EFFECTS

logical and developmental processes in the fruit fly, Drosophila magnesium-free Hunk’s balanced salt solution, and then melanogaster (7, 37, 39). Tlr-4, in conjunction with the CD14 detaching the cells from the flask with exposure to 0.25% molecule, has high specificity for LPS binding and initiates a trypsin-ethylenediamine tetra-acetic acid (EDTA; Gibco BRL cascade of downstream events mediated by nuclear factor κ-B, Life Technologies). Trypsin-EDTA was neutralized with feeding which, in turn, mediates a host of transcriptional functions (1, medium containing FBS, and the suspension was centrifuged at 7, 13, 24, 29, 37, 39). Limited studies have demonstrated a role 1200 rpm for 10 minutes. Supernatant was removed, the pellet for Tlr-4 in anticancer immunity in head and neck cancers. resuspended in feeding medium, and the cells replated at OK-432 is a streptococcal agent that has been used for im- 1 ϫ 105 cells per flask. munotherapy of head and neck cancer, among other malignan- cies, and OK-PSA is a lipoteichoic acid-related molecule iso- Maintenance of the J774a.1 and RAW 264.7 Cell Line lated from OK-432. In vivo studies in Tlr-4–deficient mice have The J774A.1 cell line (American Type Culture Collection shown that the antitumoral effects of OK-432 and OK-PSA are [ATCC], Manassas, VA) is a murine macrophage cell line mediated in part by Tlr-4 (25, 26). The role of the TLRs, partic- derived from a female BALB/c mouse and is active in ularly Tlr-4, in the antitumoral effects of LPS on GBM has not antibody-dependent phagocytosis (31). The RAW 264.7 cell been evaluated previously. We present studies showing that line (ATCC) is a murine macrophage cell line established from Tlr-4 plays an important in vivo role in the antitumoral effects a tumor induced by the Abelson murine leukemia virus in a of LPS on GBM in mouse models. We originally hypothesized male BALB/c mouse (32). For our quantitative flow cytometry that LPS may be acting directly on implanted glioblastoma cells studies of Tlr-4 and CD14, both cell lines served as positive through Tlr-4 or CD14 receptor pathways, or both. However, the controls; the J774A.1 line was also used as a positive control quantitative flow cytometry data of human and murine for the Tlr-4 Western blotting. Both cell lines were grown in glioblastoma cells in vitro and Western blot analysis of murine high-glucose DMEM with 10% FBS, 0.2 mM glutamine, glioblastoma cells in vivo with and without LPS treatments 50 µg/ml neomycin, 100 µg/ml penicillin, and 100 µg/ml showed no evidence of human or mouse Tlr-4 expressions, and streptomycin and plated in 100 mm tissue culture dishes for flow cytometry data confirmed no CD14 expression, suggesting incubation in a 5% CO humidified atmosphere at 37ЊC. On our original hypothesis to be incorrect. Therefore, other cell 2 reaching confluency, passage of the cells was accomplished by types in the tumor-bearing host must bear Tlr-4, which would removing old feeding media, washing the cells with calcium account for the diminished antitumoral response to LPS in and magnesium-free Hunk’s balanced salt solution, adding Tlr-4 knockout (KO) mice. fresh feeding media, then scraping the cells off the dish using a sterile cell scraper. The dislodged cells then were dispensed MATERIALS AND METHODS into new tissue culture dishes at a subcultivation ratio of 1:3 to 1:6. Media was either replaced or added two to three times Maintenance of DBT Glioblastoma Cells in Culture weekly until cells reached confluency. The delayed brain tumor (DBT) cell line was established from a tumor induced in an adult mouse by intracerebral injec- Maintenance of the U-87 MG Cell Line tion of the Rous sarcoma virus (16). Tumors generated after The U-87 MG cell line (ATCC) is a human glioblastoma cell intracranial or subcutaneous implantation of the DBT cells are line derived from a 44-year-old Caucasian woman. This is a analogous to human glioblastomas in terms of their aggressive hypodiploid human cell line with the modal chromosome 44 growth pattern, histopathological characteristics, and occurring in 48% of cells. This is one of a number of cell lines immunoreactivity for glial fibrillary acidic protein (6). DBT derived from malignant gliomas by Ponten and Macintyre (30) glioma cells (previously donated to our laboratory by from 1966 through 1969. The original cultures were established Dr. Michael M.C. Lai, Department of Molecular Microbiology as explants on grid-supported lens paper or gelatin foam with and Immunology, University of Southern California) were used Eagle’s minimum essential medium and 10% bovine calf serum for subcutaneous and intracranial implantations in Tlr-4 KO as the culture fluid. Trypsinization of the outgrowth of cells mice (The Jackson Laboratory, Bar Harbor, ME) and in appro- attached to the vessel floor with subsequent transfer to stan- priately matched BALB/c wild-type mice (The Jackson dard vessels in growth medium permitted cell line develop- Laboratory). As previously described (6, 35, 38), DBT cells were ment. A culture at passage 108 was deposited at ATCC by grown in Dulbecco’s modified Eagle’s medium (DMEM; Gibco Ponten in 1973. For our quantitative flow cytometry studies of BRL Life Technologies, Rockville, MD) with 15% fetal bovine Tlr-4 and CD14, U-87 MG cells were grown in Eagle’s minimal serum (FBS; Gemini Bioproducts, Calabasas, CA), 0.2 mM glu- essential medium with Earle’s balanced salt solution and tamine (Gibco BRL Life Technologies), 50 µg/ml neomycin 2 mmol/L L-glutamine that is modified to contain 1.0 mmol/L (Sigma-Aldrich Corp., St. Louis, MO), 100 µg/ml penicillin, sodium pyruvate and 0.1 mmol/L nonessential amino acids and 100 µg/ml streptomycin and plated in culture flasks for (Mediatech, Inc., Herndon, VA). In addition, 10% FBS, Њ incubation in a 5% CO2 humidified atmosphere at 37 C. On 50 µg/ml neomycin, 100 µg/ml penicillin, and 100 µg/ml reaching confluency, cell passage was accomplished by remov- streptomycin were added to the medium. Cells were plated in ing feeding medium, washing attached cells with calcium and culture flasks for incubation in a 5% CO2 humidified atmos-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 373 CHICOINE ET AL.

phere at 37ЊC. On reaching confluency, passage of the cells was depth of 4 mm into the brain using a Hamilton syringe accomplished by removing feeding medium, washing attached (Hamilton Company, Reno, NV). The animal was removed from cells with calcium and magnesium-free Hunk’s balanced salt the stereotactic frame and the scalp was closed with Autoclips solution, and then detaching the cells from the flask with expo- (Becton Dickinson & Company, Sparks, MD). After recovery sure to 0.25% trypsin-EDTA. Trypsin-EDTA was neutralized from anesthesia, mice were returned to cages to resume normal with feeding medium containing FBS, and the suspension was activity and diet ad libitum. Mice were observed daily to the centrifuged at 1200 rpm for 10 minutes. The supernatant was end point of death or a moribund state, at which time they were removed and the pellet resuspended in feeding medium at a sacrificed using Euthasol. Survival in days from the time of subcultivation ratio of 1:2 to 1:5. tumor cell implantation was recorded for each animal implanted with an intracranial DBT tumor. Differences in sur- Preparation and Care of Animals vival were compared between controls and the various LPS χ2 Mice used in this study were cared for according to the guide- treatment groups using the log-rank test. lines of the American Association for Accreditation of Laboratory Animal Care as approved by the Division of Comparative Intratumoral LPS Treatments for Mice Implanted Medicine at Washington University. For animal procedures, Intracranially with DBT Glioblastomas including implantation of DBT cells and injection of LPS, general On Day 0, while under anesthesia and in a stereotactic frame anesthesia was induced using intraperitoneal administration of as mentioned previously, mice were injected intracranially with a rodent cocktail (87 mg/Kg ketamine and 13 mg/Kg xylazine). 100 µg LPS via the same craniotomy site in the skull. The scalp was closed and the mice recovered as described previously. Subcutaneous Implantation of DBT Cells in Tlr-4 KO and Wild-Type BALB/C Mice Preparation of Tumor Specimens with and without General anesthesia was induced for female Tlr-4 KO BALB/c LPS Treatment for Histopathological Evaluation mice (The Jackson Laboratory) or corresponding wild-type At predetermined intervals after subcutaneous implantation female BALB/c mice as described previously. The flank region of of DBT cells, subsets of LPS-treated and control mice were sac- each mouse was prepared with 95% ethanol, and 2 million DBT rificed using intraperitoneal Euthasol as mentioned previously. mouse glioblastoma cells in 200 ml serum-free DMEM were Subcutaneous tumors were collected and weighed. Each tumor injected subcutaneously. On Day 21, mice were sacrificed by then was postfixed with 4% paraformaldehyde in PBS at 4ЊC for intraperitoneal administrations of 0.01 ml Euthasol (Delmarva at least 24 hours, followed by placement in 30% sucrose in PBS Laboratory Inc., Des Moines, IA). Tumors were harvested and at 4ЊC for at least 24 hours. Fixed specimens were embedded in mean tumor mass was compared between treatment and control Tissue Tek OCT compound (Miles, Elkhart, IN) and frozen on groups of mice implanted with subcutaneous DBT tumors. dry ice before being cut into 10- to 20-µm thick sections with a cryostat. Sections were fixed for 24 hours on frosted microscope Intratumoral LPS Treatments for Mice Implanted slides. Specimens were analyzed for general histopathological Subcutaneously with DBT Glioblastomas features, with particular emphasis on inflammatory infiltrates. Mice implanted with subcutaneous DBT glioblastomas were treated with intratumoral LPS (Sigma; 400 µg Escherichia coli Hematoxylin and Eosin Staining serotype O55:B5) on Days 7 and 14. Control Tlr-4 KO and con- Sections of specimens on frosted microscope slides were trol wild-type mice did not receive LPS, but rather were fixed serially in Pro-Par Clearant (Anatech, Ltd., Battle Creek, injected intratumorally with phosphate-buffered saline (PBS). MI) for 9 minutes, absolute alcohol for 2 minutes, 95% ethanol Tumors were harvested on Day 21 and mean tumor masses in for 1 minute, and 70% ethanol for 1 minute, followed by a grams were compared using unpaired t tests between LPS- 1-minute rinse in deionized water. Specimens were then placed treated mice and controls for both Tlr-4 KO mice and BALB/c in Harris’ acidified hematoxylin (Anatech, Ltd.) for 2 minutes, wild-type mice. followed by a brief tap water rinse, and were placed in acidic alcohol for 1 minute, followed by another brief rinse with tap Intracranial Implantation of DBT Cells in Tlr-4 KO water. The slides then were placed in 70% ethanol for 1 minute and Wild-Type BALB/c Mice and in eosin (Anatech, Ltd.) for 1.5 minutes, after which serial Female Tlr-4 KO mice and corresponding wild-type female rinses with 95% ethanol (1 min), 100% ethanol (3 min), and BALB/c mice were anesthetized as described previously. The Pro-Par Clearant (3 min) were completed. The slides were heads were prepped with 95% ethanol, a 1-cm midline longitu- mounted with Permount (Fisher, St. Louis, MO), and cover- dinal incision was made, and the mouse was fixed in a stereo- slips were applied. tactic frame (Stoelting Co., Wood Dale, IL). A right paramedian craniotomy was made (2 mm lateral and 2 mm posterior to Statistical Analyses bregma) using a dental drill with a 2-mm bit (Foredom Statistical analyses were conducted using SAS software (SAS Electronic Co., Bethel, CT). Five hundred thousand DBT cells Institute, Cary, NC). Data derived from mice with subcutaneous suspended in 9 µl DMEM were injected over 30 seconds to a DBT tumors treated with LPS were assessed using a one-way

374 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com TOLL-LIKE RECEPTOR 4 AND LIPOPOLYSACCHARIDE ANTITUMORAL EFFECTS

TABLE 1. Summary of results for subcutaneous implantationa No. of Mean tumor mass Compared Compared Compared Compared Group specimens (g) at 21 d (range) with Group 1 with Group 2 with Group 3 with Group 4 1: BALB/C wild-type mice, 10 2.15 (0.98–3.53) NA P ϭ 0.0135 P ϭ 0.9130 P ϭ 0.0514 no treatment 2: BALB/C wild-type mice, 6 0.21 (0–1.08) NA P ϭ 0.0058 P ϭ 0.0520 IT LPS d7 and 14 3: Tlr-4 knockout mice, 9 2.15 (0.67–3.38) NA P ϭ 0.1174 no treatment 4: Tlr-4 knockout mice, 8 1.09 (0.20–2.64) NA IT LPS d7 and 14

a NA, not applicable; IT, intratumoral; LPS, lipopolysaccharide; Tlr-4, toll-like receptor 4. analysis of variance and post hoc tests using Tukey’s studen- for each cell line tested was resuspended in cold PBS, incu- tized range test. Survival in days from the time of intracranial bated with PE-labeled monoclonal antibody to human or tumor implantation was analyzed using Kaplan-Meier product- mouse Tlr-4 (eBioscience, San Diego, CA) or FITC-labeled mon- limit survival estimates. Differences in survival were compared oclonal antibody to CD14 (Pharmingen, San Diego, CA) for between controls and the various LPS treatment groups using 1 hour at 4ЊC, washed once in PBS, spun down for 5 minutes at the log-rank χ2 test. 1200 rpm, and ﬁnally resuspended in 1 ml cold PBS.

In Vitro Glioblastoma Cells Labeled with Antibodies to Flow Cytometry Technique Human Tlr-4, Mouse Tlr-4, and CD14 Flow cytometric analyses of FITC-CD14–labeled and PE- Flow cytometry of glioblastoma cells incubated with phyco- human Tlr-4–labeled cells (or PE-mouse Tlr-4 labeled) were erythrin (PE)-labeled antibody to Tlr-4 or fluorescein isothio- conducted using a Becton Dickinson FACS 440 flow cytometer cyanate (FITC)-labeled antibody to CD14 was used to quantita- (BD Biosciences Immunocytometry Systems, San Jose, CA). tively assess the presence of human Tlr-4, mouse Tlr-4, and CD14 on glioblastoma cells and macrophages before and after LPS treatments. Assays for the mouse cell lines, DBT, RAW, and J774, were performed with rat antimouse antibodies to Tlr-4 and CD14, respectively, and assays for the human glioblastomas U87 were performed with mouse antihuman antibodies to Tlr-4 and CD14, respectively. One million cells of each cell line tested were plated on 100-mm tissue culture dishes 1 day before flow evaluation. After 24 hours, each cell line was exposed to either 100 or 200 µg/ml LPS in serum-free media for approximately 30 minutes. Controls were exposed to an equivalent amount of sterile PBS in serum- free media for the same amount of time. After 30 minutes, the cells were washed, trypsinized, and spun down for 5 minutes at FIGURE 1. Diagram summarizing the methodology and results for subcutaneous DBT glioblastoma implantations 1200 rpm. Then, each cell pellet in Tlr-4 KO mice and wild-type BALB/c mice and LPS treatments.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 375 CHICOINE ET AL.

FIGURE 2. Bar graph showing mean tumor masses for mice subcutaneously B implanted with DBT glioblastomas, including wild-type BALB/c mice with or without LPS treatments and Tlr-4 KO mice with or without LPS treatments (see text and Table 1 for details). The P value for BALB/c control mice versus BALB/c + LPS mice was 0.0135 and for Tlr-4 KO control mice versus Tlr-4 KO + LPS mice was 0.1174. For BALB/c control mice versus Tlr-4 KO control mice, the P value was 0.9130.

The system was interfaced to a CICERO/CYCLOPS data acquisition system (Cytomation, Inc., Ft. Collins, CO) for data storage and analysis. Both fluorochromes were excited with 300 mW at 488 nm of an argon ion laser. FITC fluorescence was detected through a 530-nm band-pass filter, whereas PE fluo- FIGURE 3. A, Kaplan-Meier survival plot for female BALB/c mice intracra- rescence was detected through a 580-nm band-pass filter. nially implanted with 5 ϫ 105 DBT glioblastoma cells mixed with 100 µg LPS A minimum of 20,000 events were acquired in the list mode for or PBS on Day 0. Median survival was 19 days (mean, 20.8 d; range, 12–34 each sample, and fluorescence distributions of live cells were d) for the five mice treated with 100 µg LPS versus 10 days (mean, 11.2 d; obtained by gating. range, 9–23 d) for the 13 control mice. Log-rank analysis of Kaplan-Meier plots showed statistically significant prolonged survival for the LPS-treated Western Blot Technique mice compared with control mice (P ϭ 0.0102). B, Kaplan-Meier survival plot for female Tlr-4 KO BALB/c mice intracranially implanted with 5 ϫ 105 DBT Western blot analysis of disaggregated specimens from glioblastoma cells mixed with 100 µg LPS or PBS on Day 0. Median survival implanted tumors in Tlr-4 KO BALB/c mice and wild-type was 14 days (mean, 16.4 d; range, 11–28 d) for the 10 Tlr-4 KO mice treated BALB/c mice with or without LPS treatments were conducted with 100 µg LPS versus 12 days (mean, 13.8 d; range, 10–21 d) for the nine using a standard sodium dodecyl sulfate-polyacrylamide gel control Tlr-4 KO mice treated with PBS. Log-rank analysis of Kaplan-Meier electrophoresis technique. The blot was incubated with an anti- plots showed no increase in survival for the LPS-treated Tlr-4 KO mice com- Tlr-4 antibody (Santa Cruz Biotechnology Inc., Santa Cruz, pared with Tlr-4 KO control mice (P ϭ 0.3277). CA). A horseradish peroxidase conjugated antigoat polyclonal (Santa Cruz Biotechnology Inc.) was used as the secondary Intracranial DBT Glioblastomas are More Resistant to the antibody. The prepared blot was visualized with the Antitumoral Effects of LPS When Implanted in Tlr-4 KO SuperSignal West Femto Maximum Sensitivity Substrate Mice than When Implanted in Wild-type BALB/c Mice (Pierce Biotechnology, Inc., Rockford, IL). Intracranial LPS caused a modest increase in survival for mice implanted with intracranial DBT cells in wild-type RESULTS BALB/c mice (Fig. 3A), but did not increase the survival of Tlr-4 KO BALB/c mice bearing intracranial DBT cells (Fig. 3B). Subcutaneous DBT Glioblastomas are More Resistant to These results indicate that the Tlr-4 KO BALB/c mice bearing the Antitumoral Effects of LPS When Implanted in Tlr-4 intracranially implanted DBT cells are more resistant to the KO Mice than When Implanted in Wild-type BALB/c Mice antitumoral effects of LPS than wild-type BALB/c mice bearing In the subcutaneous model, LPS caused near complete tumor intracranially implanted DBT cells. elimination in wild-type BALB/c mice; however, in Tlr-4 KO BALB/c mice, only a 50% tumor reduction was achieved Histopathological Examination (Figs. 1 and 2, Table 1). These results indicate that the Tlr-4 KO Microscopic examination of subcutaneously implanted DBT mice bearing subcutaneously implanted DBT cells are more glioblastomas with or without LPS treatments showed no clear resistant to the antitumoral effects of LPS than wild-type differences between wild-type BALB/c and Tlr-4 KO BALB/c BALB/c mice bearing subcutaneous implanted DBT cells. mice, except that the tumors tended to be larger in the Tlr-4 KO

376 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com TOLL-LIKE RECEPTOR 4 AND LIPOPOLYSACCHARIDE ANTITUMORAL EFFECTS

FIGURE 4. Photomicrographs of subcutaneous DBT tumors stained with inflammatory response 2 days after intratumoral injection of 400 µg LPS in hematoxylin and eosin at ϫ200 magnification. A–C, photomicrograph of sub- wild-type BALB/c mouse (E) and Tlr-4 KO mouse (F) implanted with subcu- cutaneous tumor stained with hematoxylin and eosin 7 days after implanta- taneous tumors. In both the wild-type and Tlr-4 KO mice, most of the speci- tion of 2 ϫ 106 DBT cells in BALB/c mice (A and B) and Tlr-4 KO BALB/c men consisted of inflamed granulation tissue rich in neutrophils. Small nests mice (C) without treatment (A) or after treatment (B and C) with PBS. The of neoplastic cells comprise a minority of the lesion (F, lower left). hypercellular histological results of these control tumors are identical when Histologically, the two tumors were similar, although the wild-type mice had comparing specimens from the wild-type BALB/c mice with specimens from much smaller tumors. G and H, photomicrographs of histologically similar the Tlr-4 KO BALB/c mice. C contains several entrapped skeletal muscle subcutaneous tumors encountered in wild-type BALB/c (G) and Tlr-4 KO (H) fibers from the abdominal wall at the site of tumor growth. D, photomicrograph mice 7 days after intratumoral injection of 400 µg LPS, although the former of similar subcutaneous tumor in Tlr-4 KO mouse 14 days after implantation typically were smaller. The inflammatory component consisted predominantly of DBT cells. The tumor demonstrated pseudopalisading necrosis characteris- of macrophages and lymphocytes (G), and the tumor had characteristic foci of tic of glioblastoma multiforme. E and F, photomicrographs demonstrating the pseudopalisading necrosis (H). mice (Fig. 4, A–D). Both subcutaneously and intracranially (U87) glioblastoma cells in vitro (Fig. 5). The murine implanted DBT tumors showed histopathological features of macrophage cell lines J774A.1 and RAW 264.7 served as Tlr-4– glioblastoma, including hypercellularity and pseudopalisading and CD14-positive controls. necrosis. There was a peritumoral neutrophilic infiltrate within a few days after LPS treatment, followed by a macrophage-rich Western Blot Analysis Shows No Expression of infiltrate 1 week after LPS administration (Fig. 4, E–H). Tlr-4 in Murine DBT Glioblastoma Cells Implanted Subcutaneously Flow Cytometry Analysis Shows No Expression of Western blotting did not show detectable Tlr-4 receptor in Human Tlr-4, Mouse Tlr-4, or CD14 in Murine the disaggregated tumors harvested from Tlr-4 KO and wild- and Human Glioblastoma Cells type mice previously implanted with DBT glioblastoma cells Quantitative flow cytometry did not show detectable Tlr-4 (Fig. 6). The murine macrophage cell line J774A.1 served as the receptor or the CD14 moiety for either murine (DBT) or human Tlr-4–positive control.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 377 CHICOINE ET AL.

ABC

DEF

FIGURE 5. A–D, graphs of flow cytometry data from four cell lines labeled with FITC-tagged CD14 antibodies. RAW macrophages (A), DBT mouse glioblastoma cells (B), U87 human glioblastoma cells (C), and G J774 macrophages (D) were incubated with the FITC-labeled CD14 antibody. After labeling in vitro, flow cytometry was performed as described in the Materials and Methods section. For each graph, the y axis represents the cell counts and the x axis represents the labeling intensity. The red peaks (vertical lines) represent the untreated controls (no LPS), the green peaks (angled lines from the lower left to the upper right) represent the 100 µg/ml LPS-treated cells, and the brown peaks (angled lines from the upper left to the lower right) represent the 200 µg/ml LPS-treated cells. The latter dose, 200 mg/ml, was tested only in the DBT and U87 cells. The rightward shift of the peaks for the two macrophage cell types indicates the presence of the CD14. The leftward position of the peaks for the two glioblastoma cell lines indicates the absence of expression of CD14. E–G, graphs of flow cytometry data from similar experiments performed using a PE- labeled Tlr-4 antibody using the J774 macrophage cell line (E), a human glioblastoma cell line (F), and one murine glioblastoma cell line (G). The red peaks (vertical lines) represent the untreated controls (no LPS and no Tlr-4 antibody), the green peaks (angled lines from the lower left to the upper right) represent untreated cells (no LPS) labeled with Tlr-4 antibody, and the brown peaks (angled lines from the upper left to the lower right) represent the 100 µg/ml LPS-treated cells. The rightward shift with the macrophage cell line indicates labeling of the Tlr-4 in these cell lines and the absence of a rightward shift of the peaks. In the two panels of glioblastoma cell cultures, the lack of a shift indicates no expression in these cell types. Note that neither the Tlr-4 nor the CD14 expression was affected by exposure in vitro to LPS, except for the CD14 expression in the J774 macrophage cells, which was increased by LPS exposure.

DISCUSSION Since their description just a few years ago, the TLRs have been characterized in considerable detail and are known to play a crit- Survival for patients diagnosed with GBM remains poor ical role in the immunological responses to bacteria and other even after maximal treatment; therefore, novel treatment strate- pathogens (1, 7, 12, 21, 23, 29, 33, 37). In particular, Tlr-4 is known gies are needed. We previously demonstrated that LPS leads to to mediate many of the effects of LPS on macrophages and other dramatic regression of malignant gliomas implanted subcuta- immune cell types. Before this article, the role of Tlr-4 in the anti- neously in mice (6, 38), but the mechanisms responsible for tumoral effects of LPS on glioblastoma had yet to be described. this antitumoral effect have not been well deﬁned. The current As in our previous studies, the antitumoral effect of LPS on study shows that the antitumoral effect of LPS against subcu- glioblastoma was more effective for tumors implanted subcutaneously and intracranially implanted DBT glioblastoma cells taneously compared with those implanted intracranially. This is attenuated in Tlr-4 KO BALB/c mice compared with the anti- effect has been ascribed to the immune privilege of the central tumoral effect of LPS against similar tumors implanted in cor- nervous system. There are other possible explanations for the responding BALB/c wild-type mice. Therefore, these studies differences in responses to LPS for glioblastomas implanted in indicate that the antitumoral effect of LPS against glioblastoma the brain compared with those implanted subcutaneously. The is mediated, at least in part, by Tlr-4. central nervous system is also unique in its absence of a lym-

378 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com TOLL-LIKE RECEPTOR 4 AND LIPOPOLYSACCHARIDE ANTITUMORAL EFFECTS

diminished antitumoral response to LPS in the Tlr-4 KO mice. This is in agreement with other investigations that have reported the presence of Tlr-4 activity in the central nervous system (8), particularly in microglia (18). These same investigators also reported the absence of Tlr-4 expression in astrocytes and oligodendroglial cells (18), which agrees with our flow cytometry and Western blotting studies, given that the glioblastoma cells likely originated in astrocytic or oligodendroglial progenitor cells. The antitumoral effect of LPS on glioblastoma being mediated by Tlr-4 further supports the concept that the tumor-bearing host’s immune system regulates the antitumoral effects of LPS. The role of the immune system in this response has been shown FIGURE 6. Western blots of Tlr-4 receptor and α-actin expression in disaggre- in our previous studies in which the antitumoral effect of LPS gated specimens harvested from Tlr-4 KO and wild-type BALB/c mice after was attenuated in immunodeficient mice (38). Other investiga- implantation of DBT glioblastoma cells. Lane 1 represents tumor specimen tors have also implicated T-cell immunity (3, 27), production of from Tlr-4 KO mice implanted with DBT glioblastoma cells without LPS treat- tumor necrosis factor (2, 9, 10, 14, 19, 27, 28), interferons (11, 27), ment. Lane 2 represents tumor specimen from Tlr-4 KO mice implanted with and interleukins (2, 10, 11, 14, 19, 27) in the antitumoral effects DBT glioblastoma cells and treated with LPS. Lane 3 represents tumor speci- of LPS. Further analysis of the role of Tlr-4–expressing cells in men from wild-type BALB/c mice implanted with DBT glioblastoma cells with- the immunological response of the tumor-bearing host may pro- out LPS treatment. Lane 4 represents disaggregated spleen specimen from vide further insights into immunotherapy strategies for patients wild-type BALB/c mice without LPS. Lanes 1, 2, and 3 have no expression for Tlr-4 receptor. Lane 4 shows expression of the Tlr-4 receptor in the spleen. Lane with malignant gliomas. In addition to immune-mediated 5 represents the J774A.1 murine macrophage line, which shows a strong Tlr-4 effects, LPS has been shown to have many other biological expression. Lane 6 represents disaggregated tumor specimen from wild-type effects, including proapoptotic effects (34, 36), antiapoptotic BALB/c mice implanted with DBT glioblastoma cells and treated with LPS. effects (5, 36), and effects on cellular migration and motility Lane 7 represents a disaggregated spleen specimen from wild-type BALB/c (22). This cascade of complex and multifaceted effects of LPS mice treated with LPS. Lane 8 represents disaggregated spleen specimen from may work in conjunction with and independent of the Tlr-4 Tlr-4 KO mice without LPS. Lanes 6 and 8 show no expression for the Tlr-4 immune-mediated effects of LPS. Further understanding of all receptors, whereas the wild-type spleen extract in Lane 7 shows expression of of the different mechanisms involved in the antitumoral effects the Tlr-4 receptor. The Western blot shows how in vivo DBT glioblastoma cells of LPS on glioblastoma multiforme is needed. with and without LPS treatment in both Tlr-4 KO and wild-type mice did not There were no histopathological differences in the tumors express Tlr-4 receptors; however, the macrophage cell line as well as the wild- type BALB/c spleen extracts showed expression of Tlr-4 with and without LPS from the Tlr-4 KO BALB/c mice compared with the wild-type treatment. Each lane additionally was probed for α-actin expression as a con- BALB/c mice, with the exception of the finding of a more trol. Lanes 1 through 8 showed expression of α-actin. robust LPS effect in the wild-type mice, manifested by smaller tumor sizes. The early neutrophilic and later macrophage-rich infiltrates in both types of mice indicate that the absence of phatic system and in the presence of the blood-brain barrier. Tlr-4 does not completely eliminate the immune-mediated anti- Mice are much less tolerant of LPS administered directly tumoral effect of LPS, but in some way does reduce its inten- intracranially in comparison with LPS administered subcuta- sity. Further studies will be needed to improve understanding neously or by other routes. This further points to the unique of the immune modulation of the antitumoral effect of LPS on and somewhat fragile nature of the central nervous system and glioblastoma by Tlr-4. may account for the less robust antitumoral effect of LPS for In summary, we have further elucidated the nature of the anti- intracranial tumors. Despite the limitations in the central nerv- tumoral mechanisms of LPS on glioblastoma and have shown ous system, a statistically significant increase in survival, albeit that the recently described Tlr-4 receptor plays an important role modest, was achieved in the wild-type BALB/c mice. The in this process. The mechanisms by which Tlr-4 mediates the absence of this increase in survival in the Tlr-4 KO BALB/c antitumoral effects of LPS on glioblastoma are in need of further mice indicates that Tlr-4 factors in the LPS-induced antitumoral investigation. Nonetheless, these results do suggest that the rap- response, even in the unique immunological environment of idly expanding topic of the TLRs may have applicability in the the central nervous system. We originally hypothesized that field of neuro-oncology and may provide novel treatment strate- LPS may be acting directly on implanted glioblastoma cells gies for patients with malignant gliomas and other cancers. through Tlr-4 or CD14 receptor pathways, or both. However, the quantitative flow cytometry data of glioblastoma cells in vitro and Western blot analysis of glioblastoma cells in vivo CONCLUSION with and without LPS treatments showed no evidence of Tlr-4, and flow cytometry confirmed no CD14 expression, suggesting This study provides the first evidence that Tlr-4 plays a role that this hypothesis was incorrect. Therefore, other cell types in in the immune-mediated antitumoral effects of LPS on GBM in the tumor-bearing host must bear Tlr-4, which accounts for the a mouse model. Further analysis of these mechanisms will pro-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 379 CHICOINE ET AL.

vide clues as to how these innate antitumoral mechanisms may 22. Mohan K, Cordeiro E, Vaci M, McMaster C, Issekutz TB: CXCR3 is required for be harnessed most efficiently to treat patients with GBM and migration to dermal inflammation by normal and in vivo activated T cells: other cancers of the central nervous system. Differential requirements by CD4 and CD8 memory subsets. Eur J Immunol 35:1702–1711, 2005. 23. Monick MM, Powers L, Butler N, Yarovinsky T, Hunninghake GW: Interaction of matrix with integrin receptors is required for optimal LPS-induced MAP REFERENCES kinase activation. Am J Physiol Lung Cell Mol Physiol 283:L390–L402, 2002. 24. Muroi M, Ohnishi T, Tanamoto K: Regions of the mouse CD14 molecule 1. Barton GM, Medzhitov R: Toll-like signaling pathways. Science required for toll-like receptor2- and 4-mediated activation of NF-Kappa B. 300:1524–1525, 2003. J Biol Chem 277:42372–42379, 2002. 2. Ben-Nathan D, Padgett DA, Loria RM: Androstenediol and dehydro- 25. Okamoto M, Oshikawa T, Ohe G, Nishikawa H, Furuichi S, Tano T, Moriya Y, epiandrosterone protest mice against lethal bacterial infections and Saito M, Sato M: Severe impairment of anti-cancer effect of lipoteichoic acid- lipopolysaccharide toxicity. J Med Microbiol 48:425–431, 1999. related molecule isolated from a penicillin-killed Streptococcus pyogenes in 3. Bowles AP Jr, Perkins E: Long-term remission of malignant brain tumors toll-like receptor 4-deficient mice. Int Immunopharmacol 1:1789–1795, 2001. after intracranial infection: A report of four cases. Neurosurgery 44:636–643, 26. Okamoto M, Oshikawa T, Tano T, Ohe G, Furuichi S, Nishikawa H, Ahmed 1999. SU, Akashi S, Miyake K, Takeuchi O, Akira S, Moriya Y, Matsubara S, Ryoma 4. Byrd-Leifer CA, Block EF, Takeda K, Akira S, Ding A: The role of MyD88 and Y, Saito M, Sato M: Involvement of Toll-like receptor 4 signaling in interferon- TLR4 in the LPS-mimetic activity of Taxol. Eur J Immunol 31:2448–2457, 2001. gamma production and antitumor effect by streptococcal agent OK-432. J 5. Chao W, Shen Y, Zhu X, Zhao H, Novikov M, Schmidt U, Rosenzweig A: Natl Cancer Inst 95:316–326, 2003. Lipopolysaccharide improves cardiomyocyte survival and function after 27. Old LJ: Tumor necrosis factor. Sci Am 258:59–75, 1988. serum deprivation. J Biol Chem 280:21997–2005, 2005. 28. Otto F, Schmid P, Mackesen A, Wehr U, Seiz A, Braun M, Galanos C, 6. Chicoine MR, Won EK, Zahner M: Intratumoral injection of lipopolysaccha- Mertlesmann R, Englehardt R: Phase II trial of intravenous endotoxin in ride causes regression of subcutaneously implanted mouse DBT glioblas- patients with colorectal and non-small cell lung cancer. Eur J Cancer toma multiforme. Neurosurgery 48:607–614, 2001. 32A:1712–1718, 1996. 7. Dunne A, O’Neill LA: The interleukin-1 receptor/Toll-like receptor superfam- 29. Poltorak A, He X, Smirnova I, Liu MY, Van Huffel C, Du X, Birdwell D, Alejos ily: Signal transduction during inflammation and host defense. Sci STKE E, Silva M, Galanos C, Freudenberg M, Ricciardi-Castagnoli P, Layton B, 2003:re3, 2003. Beutler B: Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: 8. Eklind S, Mallard C, Leverin A, Gilland E, Blomgren K, Mattsby-Baltzer I, Mutations in Tlr4 gene. Science 282:2085–2088, 1998. Hagberg H: Bacterial endotoxin sensitizes the immature brain to hypoxic- 30. Ponten J, Macintyre EH: Long term culture of normal and neoplastic human ischaemic injury. Eur J Neurosci 13:1101–1106, 2001. glia. Acta Pathol Microbiol Scand 74:465–486, 1968. 9. Engelhardt B, Conley FK, Kilshaw PJ, Butcher EC: Lymphocytes infiltrating 31. Ralph P, Nakoinz I: Phagocytosis and cytolysis by a macrophage tumour and the CNS during inflammation display a distinctive phenotype and bind its cloned cell line. Nature 257:393–394, 1975. toVCAM-1 but not to MAdCAM-1. Int Immunol 7:481–491, 1995. 32. Raschke WC, Baird S, Ralph P, Nakoinz I: Functional macrophage cell lines 10. Engelhardt R, Mackensen A, Galanos C, Andreesen R: Biological response to transformed by Abelson leukemia virus. Cell 15:261–267, 1978. intravenously administered endotoxin in patients with advanced cancer. 33. Rock FL, Hardiman G, Timans JC, Kastelein RA, Bazan JF: A family of human J Biol Response Modif 9:480–491, 1990. receptors structurally related to Drosophila Toll. Proc Natl Acad Sci U S A 11. Engelhardt R, Otto F, Mackensen A, Mertelsmann R, Galanos C: Endotoxin 95:588–593, 1998. (Salmonella abortus equi) in cancer patients. Clinical and immunological 34. Schildberg FA, Schulz S, Dombrowski F, Minor T: Cyclic AMP alleviates findings. Prog Clin Biol Res 392:253–261, 1995. endoplasmic stress and programmed cell death induced by lipopolysaccha- 12. Ewbank JJ, Gorvel J, Vivier E: A widely conserved smell of danger. Sci STKE rides in human endothelial cells. Cell Tissue Res 320:91–98, 2005. 2003:pe9, 2003. 35. Sherburn EW, Kim P, Wanebo JA, Song S, Chicoine MR, Woolsey TA: Gliomas 13. Flo TH, Ryan L, Latz E, Takeuchi O, Monks BG, Lien E, Halaas O, Akira S, in the rodent barrel cortex: A novel tumor model. J Neurosurg 91:814–821, Skjak-Braek G, Golenbock DT, Espevik T: Involvement of Toll-like Receptor 1999. (TLR) 2 and TLR4 in cell activation by mannuronic acid polymers. J Biol 36. Sherwin C, Fern R: Acute lipopolysaccharide-mediated injury in neonatal Chem 277:35489–35495, 2002. white matter glia: Role of TNF-alpha, IL-1beta, and calcium. J Immunol 14. Goto S, Sakai A, Kera J, Suma Y, Soma G, Takeuchi S: Intradermal adminis- 175:155–161, 2005. tration of lipopolysaccharide in treatment of human cancer. Cancer Immunol 37. Suzuki N, Suzuki S, Duncan GS, Millar DG, Wada T, Mirtsos C, Takada H, Immunother 42:255–261, 1996. Wakeham A, Itie A, Li S, Penninger JM, Wesche H, Ohashi PS, Mak TW, Yeh 15. Kleihues P, Cavenee WK (eds): World Health Organization Classification of WC: Severe impairment of interleukin-1 and Toll-like receptor signaling in Tumours-Pathology and Genetics—Tumors of the Nervous System. Lyon, France, mice lacking IRAK-4. Nature 416:750–756, 2002. IARC Press, 2000, pp 9–52. 38. Won EK, Zahner MC, Grant EA, Gore P, Chicoine MR: Analysis of the anti- 16. Kumanishi T: Brain tumors induced with Rous sarcoma virus, Scmidt-Ruppin tumoral mechanisms of lipopolysaccharide against glioblastoma multiforme. strain. I. Induction of brain tumors in adult mice with Rous chicken sarcoma Anticancer Drugs 14:457–466, 2003. cells. Jpn J Exp Med 37:461–474, 1967. 39. Xu Y, Tao X, Shen B, Horng T, Medzhitov R, Manley JL, Tong L: Structural 17. Landis SH, Murray T, Bolden S, Wingo PA: Cancer statistics, 1999. CA Cancer basis for signal transduction by the Toll/interleukin-1 receptor domains. J Clin 49:8–31, 1999. Nature 408:111–115, 2000. 18. Lehnardt S, Lachance C, Patrizi S, Lefebvre S, Follett PL, Jensen FE, Rosenberg PA, Volpe JJ, Vartanian T: The toll-like receptor is necessary for lipopolysaccharide-induced oligodendrocyte injury in the CNS. J Neurosci Acknowledgments 22:2478–2486, 2002. We thank Ralph G. Dacey, Jr., M.D., Department of Neurosurgery, Washington 19. Mackensen A, Galanos C, Engelhardt R: Treatment of cancer patients with University, School of Medicine, for providing the resources for creating and endotoxin induces release of endogenous cytokines. Pathobiology maintaining our neuro-oncology research laboratory; Betsy Grant, Ph.D., 59:264–267, 1991. Division of Biostatistics, Washington University School of Medicine, for per- 20. Medzhitov R, Preston-Hurlburt P, Janeway CA Jr: A human homologue of the forming statistical analyses; Michael M. C. Lai, Ph.D., Department of Molecular Drosophila Toll protein signals activation of adaptive immunity. Nature Microbiology and Immunology, University of Southern California, for his dona- 388:394–397, 1997. tion of the DBT glioblastoma cell line; Sherry Boeckelmann for assistance with 21. Medzhitov R, Janeway CA Jr: Self-defense: The fruit fly style. Proc Natl Acad manuscript preparation; and Claire Kramer and Karen Dodson, B.S., for edito- Sci U S A 95:429–430, 1998. rial assistance.

380 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com TOLL-LIKE RECEPTOR 4 AND LIPOPOLYSACCHARIDE ANTITUMORAL EFFECTS

COMMENTS glioblastoma regression. Further research elucidating these other pathways must be conducted; the findings reported in the current study his experimental study supports the role of toll-like receptor 4 provide a strong base for pursuing such work. (TLR-4) in mediating lipopolysaccharide (LPS) antitumoral effects T Deborah T. Blumenthal in wild-type and TLR-4 knockout mice models implanted with Neurologist intracranial and subcutaneous glioblastoma. The TLR-4 knockout mod- Salt Lake City, Utah els showed less benefit from LPS than the wild-type models, suggest- Philip H. Gutin ing a significant role for TLR-4 in vivo. For the subcutaneous tumors in New York, New York knockout models, LPS had some effect on tumor regression but not as significant as that seen in the wild types. In vitro studies did not sup- he authors present some very interesting work that, unfortunately, port such a role for TLR-4 when studied with glioblastoma cell lines, only suggests that LPS mediation by TLR-4 is an unlikely strategy suggesting that the receptor requires environmental factors for its full T for immune-mediated antitumor activity. The research is well done effect, as seen in the in vivo models. and the results are candidly presented. The modest improvement in the This study is an important contribution to the scientific literature and wild-type model with intracranial tumor has many causes. Further adds to our understanding of the complex cascades involved in tumor investigation into the reasons seems to be a long run for a short slide. growth and apoptosis. The experiments are well done with clear results Regardless, immunotherapy offers an exciting opportunity for glioma that support the hypothesis of the important, but not absolute, role of therapy. These data are important to direct future investigation. TLR-4 in mediating the antitumoral effect of LPS. However, as there was still some effect of LPS in the knockout models, there are obviously Joseph M. Piepmeier other pathways in addition to TLR-4 that mediate the effect of LPS on New Haven, Connecticut

The Pugilist, 1st century B.C. bronze (courtesy of Museo Nazionale Romano, photo by Araldo De Luca). LEGACIES

HISTORY OF SPINE BIOMECHANICS: PART I— THE PRE-GRECO-ROMAN, GRECO-ROMAN, AND MEDIEVAL ROOTS OF SPINE BIOMECHANICS

Sait Naderi, M.D. THE ROOTS OF spine biomechanics reside in the Antiquity and the Medieval and Department of Neurosurgery, Renaissance periods. A review of historical treatises reveals detailed information regard- Yeditepe University ing this often historically neglected discipline. Ancient medical, philosophical, and School of Medicine, Istanbul, Turkey physical documents were reviewed, as they pertained to the historical foundation of spine biomechanics. These included medical case reports and observations of nature and Niteen Andalkar, M.D. motion by ancient philosophers and scientists. These documents heavily influenced Department of Neurosurgery, the portion of the scientific literature that we now regard as “spine biomechanics” up Cleveland Clinic Spine Institute, through the Renaissance. The focus of Part I of this two-part series is placed on the Cleveland Clinic Foundation, Cleveland, Ohio ancient and medieval biomechanics-related literature and on associated literature that influenced the development of the field of modern spine biomechanics. Edward C. Benzel, M.D. KEY WORDS: Biomechanics, Greco-Roman period, History, Medieval period, Pre-Greco-Roman period, Department of Neurosurgery, Spine Cleveland Clinic Spine Institute, Cleveland Clinic Foundation, Neurosurgery 60:382–391, 2007 DOI: 10.1227/01.NEU.0000249276.94933.8D www.neurosurgery-online.com Cleveland, Ohio

Reprint requests: Edward C. Benzel, M.D., pine biomechanics is the physical science Egyptian and Indian origin. Many papyruses Department of Neurosurgery, that forms a substantial portion of the from the ancient Egyptians have been discov- Cleveland Clinic Spine Institute, foundation of modern spine surgery. As a ered. Among them is the Edwin Smith papyrus Cleveland Clinic Foundation, S 9500 Euclid Avenue, S-80, discipline, spine biomechanics has enjoyed sig- (2600–2200 B.C.), which is of great importance Cleveland, OH 44195. nificant growth during the past two centuries. from the perspective of spine biomechanics. A Email: [email protected] Its roots, however, reside in the Antiquity and copy of a copy, this papyrus is assumed to the Medieval and Renaissance periods. An have been written by Imhotep, the physician Received, February 20, 2006. assessment of these historical roots indicates for the court of Pharaoh Zoster of the Third Accepted, October 13, 2006. that an impressive understanding of the essen- Dynasty. It was sold to Edwin Smith in 1869 tials and art of scientific investigation existed and donated to the New York Historical in these periods. With such a strong scientific Society in 1906. The papyrus was not inter- background based in the physical, rather than preted for many years; however, in the 1930s, biological, sciences, it is appropriate to review its transcription was published for the first the true origins of this discipline. The portrayal time (Fig. 2) (5). In 1992, the American of the origins of spine biomechanics presented Association of Neurological Surgeons repub- herein has been drawn from medical case his- lished the Edwin Smith papyrus (23). In this tories, medical treatises, and treatises concern- papyrus, 48 cases of trauma were reported, ing kinesiology and mechanics. Of note, the including six cases of spinal trauma. Unfor- historical periods have been variably defined. tunately, the portion containing thoracic and Therefore, they have been arbitrarily chosen lumbar spine trauma was missing. This pap- and depicted as overlapping. These overlap- yrus reported vertebral dislocations (wenekh) ping time periods are summarized in Figure 1. and burst fractures (sehem). It also presented evidence in support of mechanisms of injury PRE-GRECO-ROMAN (i.e., falling on one’s head [axial loading] (ANCIENT) PERIOD resulting in a burst fracture [5]). The papyrus focused on the neurological The first documentation of literature related consequences of spinal fractures, emphasizing to spine biomechanics was drawn from pre- that fracture-dislocations are associated with a Greco-Roman case histories, including those of poor prognosis. Although the Egyptians

382 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART I

FIGURE 1. Timeline from the pre-Greco-Roman period through the Renaissance.

The use of spinal traction was reported in the ancient Indian epic Srimad Bhagwat Mahapuranam, which is estimated to have been written between 3500 and 1800 B.C. In this epic, one story tells of Lord Krishna correcting the hunchback of one of his devotees. This is the ﬁrst known report addressing the correction of spinal deformity. Kumar (12) translated the original Sankrit verse as follows: To shower the fruits of his blessings, Happy Lord Krishna decided to straighten Kubja, Who was deformed from three places. He pressed her foot by his foot, Held her chin by two ﬁngers and pulled her up. By touch and pull of Lord, She became a beautiful straight woman. (22, p 653) Srimad Bhagwat Mahapu- ranam is, therefore, the oldest known reference to the clinical application of a biomechanical principle (i.e., axial traction) to optimize the care of a patient with a spinal disorder.

FIGURE 2. The Edwin Smith papyrus was transcribed and published in the 1930s. GRECO-ROMAN PERIOD understood the importance of splinting for long bone fractures and probably had a fundamental understanding of the under- Ancient Greek medicine lying biomechanical principles, the use of such splints was not can be divided into two peri- addressed in the Edwin Smith papyrus. ods. The ﬁrst period, known Egyptian physicians, however, were aware of the spinal col- as the mythological period, umn. Much of the anatomic nomenclature terminology used began with the Trojan War today was derived from these ancient physicians’ perspective and lasted until the of the human body. For example, a derivative of the term Hippocratic period. The sec- “spinal column,” the “djet column,” was used in ancient Egypt ond period was the scientiﬁc (Fig. 3) (13). period, which began during FIGURE 3. A derivative of the “spinal column,” the djet column was The only other potential sources of information regarding the time of Hippocrates and ended with Paulus of Aegina. known to the ancient Egyptian physi- the ancient roots of biomechanics are of Indian and, possibly, cians (from, Lang JK, Kolenda H: Chinese origin. Unfortunately, there is no known bona fide The Mythological Period First appearance and sense of the term information that documents ancient Chinese interest in spinal “spinal column” in ancient Egypt. disorders. However, some information does exist regarding the The mythological period Historical vignette. J Neurosurg 97 interest of ancient Indians in this arena (12). was associated with much [Suppl 1]: 152–155, 2002 [13]).

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 383 NADERI ET AL.

mysticism and the recognition of a number of gods, including those of medicine and health. The predominant health centers of this period were known as Asclepions. These were founded for Aesculapius, the god of health. There were many Asclepions in the major Greek settlements of Titan, Trika, Rhodes, Kos, Epidauros, Athens, Alexandria, Tiber, and Pergamon. The infrastructure of Asclepions resembled modern hospitals. Treatment modalities in the Asclepions included hydrotherapy, physiotherapy, hygienic rule, diet, well-known drug therapies, and minor surgical procedures. A priest- physician was responsible for examinations and treatments. Interestingly, pregnant women and patients with severe, untreatable disease were not accepted into Asclepions. It is assumed, but not proven, that minor spinal disorders were treated in these facilities. Ultimately, Asclepions were destroyed by the Christians after Christianity achieved dominance in Rome in the second century A.D. Although some important philosophers (e.g., Thales of Miletus [625–585 B.C], Anaximenes of Miletus [556–440 B.C.], Pythagoras of Samos [580–489 B.C.], and Alcamaeon of Crotona [circa. 500 B.C.]) recorded their thoughts in the field of medicine, it seems that there is no documentation suggesting a contribution to kinesiology and biomechanics. The only information regarding the spine was reported by Empedocles, a famous physician and philosopher in the 5th century B.C. He FIGURE 4. Among his other accomplishments, Hippocrates studied the anatomy and pathology of the spine. thought that the vertebrae were initially unified, forming a rigid spine, and that this solid osseous column subsequently muscles, permitting him to describe the normal curvatures of broke down (segmented) into pieces as a result of movements the spine. of the body (1, 6). He defined tuberculous, spondylitis, posttraumatic kyphosis, scoliosis, spinal dislocation, and the spinous process fracture. Scientific Period Hippocrates was the first physician to address the relationship Modern scientific thought was first observed with between spinal tuberculosis and gibbus formation. According Hippocrates, who freed the art of medicine from the “influ- to Hippocrates, a spinous process fracture was not dangerous. ence” of supernatural spirits. Hippocrates was born on the However, he observed that fractures of the vertebral body were Island of Ceos, Greece, where he studied medicine and became of greater clinical significance (15). a priest-physician in Ceos Asclepion. Hippocrates founded an He described two frames for reduction of the dislocated open air school after he became prominent in the field of med- spine and associated deformities, including the Hippocratic icine. He wrote many treatises on a variety of medical issues. ladder and the Hippocratic board (Figs. 5 and 6) (14). He rec- Unfortunately, his original treatises were lost. They were, how- ommended simultaneous traction of the spine and the manual ever, published in 12 books known as Corpus Hippocraticum. It application of focal pressure over the kyphotic area: is thought that the books on prognostics and aphorisms were “But the physicians, or some person who is strong, and not authored by Hippocrates and the remaining books were writ- uninstructed, should apply the palm of the hand to the ten by other physicians from his school. Paul Emile Littre hump, and then, having laid the other hand upon the for- (1801–1881), a French physician and a master of language and mer, he should make pressure, attending whether this medical history, translated the 12 books, which included dis- force should be applied directly downward, or toward the cussions of spine pathologies and treatments, into French. head, or toward the hips . . . and there is nothing to pre- Therefore, information concerning spinal biomechanics in vent a person from placing a foot on the hump, and sup- ancient times can be drawn from the translated version of porting his weight on it, and making gentle pressure ...” Corpus Hippocraticum. (22, p 659). Hippocrates (460–377 B.C.) If this maneuver did not work, Hipppocrates recommended an alternative procedure: Hippocrates (Fig. 4) focused on the anatomy and pathology of the spine (6, 15, 23). As an anatomist, his contributions ”But the most powerful of the mechanical means is this; if were not remarkable. However, he realized that the spine was the hole in the wall, or in the piece of wood fastened into held together by means of intervertebral discs, ligaments, and the ground, be made as much below the man’s back as

384 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART I

FIGURE 5. The Hippocratic ladder was one method used for the reduction of FIGURE 6. The Hippocratic board was one method used for the reduction of the dislocated spine and its associated deformities. the dislocated spine and its associated deformities.

may be judged proper, and if a board, made of lime-tree, or clavicle. The C2 and the great vertebra (which corresponds to any wood, and not too narrow, be put into the hole, then either C1 or C7) were in this group. The second group included a rag, folded several times or a small leather cushion, the thoracic spine; the third group consisted of the five verte- should be laid on the hump ...when matters are thus brae between the chest and the pelvis. Hippocrates evidently adjusted, one person, or two if necessary, must press down did not consider the sacrum and coccyx to be part of the spine. at the end of the board, while others at the same time make However, he portrayed them as vertebrae when he described extension and counter-extension along the body, as for- the normal spinal curves. merly described” (22). Hippocrates used the term “ithioscoliosis” to indicate that the spine is straight in the coronal plane, but curved in the Although Hippocrates practiced spinal manipulation, he sagittal plane (15). He observed that the cervical and lumbar warned against the use of succussion to achieve spinal reduc- lordoses were normal or natural and that the sacrum arches tion. Succession was a practice in which a patient was secured dorsally and forms a cavity harboring the bladder and the to a ladder in an upside down position. He was then suddenly rectum. According to Hippocrates, the purpose of the spine is released, only to be abruptly suspended by the attached ropes. to maintain the erect posture and to form the shape of the It was thought that the rapid deceleration aided in achieving human body. spinal realignment. Hippocrates stated, Hippocrates categorized spinal disorders into five groups: “Wherefore succussion on a ladder has never straight- 1) traumatic and nontraumatic (tuberculosis, congenital, age- ened anybody, as far as I know, but it is principally prac- related-degenerative, those secondary to bilateral dislocation of ticed by those physicians who seek to astonish the mob- the pelvis, and rarely in epileptic patients) kyphosis; 2) scolio- for to such persons these things appears wonderful. For sis; 3) burst fractures; 4) vertebral dislocations; and 5) fractures example, if they see a man suspended or thrown down, of the spinous processes (15). or the like; and they always extol such practices, and According to Hippocrates, the most common cause of non- never give themselves any concern whatever may result traumatic kyphosis is spinal tuberculosis. He classified tubercu- from the experiment, whether bad or good. But the physi- lous kyphosis into two main groups: those above the level of cians who follow such practices, as far as I have known attachment of the diaphragm and those below it. He observed them, are all stupid” (22). that spinal deformity is more prominent in patients before puberty and that, after puberty, the course of the disease is Hippocrates emphasized the importance of acquiring knowl- more benign. edge regarding spinal anatomy: “One should first get a knowledge of the structure of the spine; for this is also requisite for “When hump-back occurs in children before the body has many diseases” (10). completed its growth, the legs and arms attain full size, He classified the spinal vertebrae into three groups. The first but the body will now grow correspondingly at the spine; group consisted of the vertebrae lying above the level of the these parts are defective . . . when curvature comes on in

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 385 NADERI ET AL.

persons whose bodily growth is complete, its occurence produces an apparent crisis in disease then present. In time, however, some of the same symptoms found in the younger patients show themselves to a greater or lesser degree, but in general they are all less malignant” (10). Hippocrates observed that traumatic kyphosis was commonly caused by falling on the shoulder or buttock. He addressed scoliosis associated with tuberculosis and also described a case of cervical scoliosis. He observed that burst fractures (sizis) resulted from vertical loading of the vertebra. Vertebral dislocation, both ventral and dorsal, were recorded and reported by Hippocrates. Ventral dislocation was observed to be secondary to the falling of a heavy object on the spine or by falling from a height. Dorsal dislocation was not commonly observed and was noted to be associated with serious abdominal injury, often resulting from a high fall. Hippocrates observed that dorsal dislocation was usually fatal. He identified fractures of the spinous process and observed that they healed rapidly and usually without neurological deficit. Hippocrates recommended diet and extension for the treatment of scoliosis. He favored the early management of dislocations because reduction was accomplished easily if attempted FIGURE 7. Aristotle was the most prominent research scientist in ancient before swelling occurred (15). Greece. In addition to Hippocrates, this period was represented by several other unique individuals, including Socrates, Plato, and At that time, Greek culture, with a trend toward the investiga- Aristotle. The most important aspect of this period was the tion of the human body and mind, was at its peak. The Greeks’ human inquiry into pathological processes and nature in gen- enthusiasm for athletics, sports, and gymnastics was part of eral. Socrates taught that we could not begin to understand their philosophy of developing the human being as a whole to the world around us until we understood our own nature (17). optimize functional capacity. Aristotle founded the first philosophy school in Assos (near Troy and Çanakkale, Turkey) Plato (427–347 B.C.) (Fig. 8) and was considered to be the most prominent philoso- Plato, a member of the Athenian aristocracy and 51 years pher of the time (11). junior to Socrates, began the philosophical inquiries that have influenced the disciplines of philosophy, psychology, logic, and politics. He thought that mathematics, a system of pure ideas, was the best tool for the pursuit of knowledge. Plato’s conceptualization of mathematics as the life force of science created the necessary womb for the birth and growth of the science of mechanics and spine biomechanics. Of note, however, he also thought that divine intervention contributed to the creation of the flexible spine (17). Herophilus of Chalcedon (335–280 B.C.) Herophilus was educated under the influence of Hippocratic philosophy in the Ceos Asclepion by the Greek physician Praxagoras. He made significant contributions to the understanding of anatomy, particularly human neuroanatomy. How- ever, there is no mention of spine anatomy and biomechancs in his works. Nevertheless, he was aware of spinal cord injuries and suggested that direct surgical intervention on the spinal column should be avoided (1, 9, 20). Aristotle (384–322 B.C.) The most prominent research scientist in ancient Greece was FIGURE 8. Map showing the location of the the first philosophy school in Aristotle (Fig. 7). He was born 7 years after Hippocrates’ death. Assos (arrow) (near Troy and Çanakkale, Turkey).

386 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART I

Aristotle, the universal and consummate scientist, left his mark in practically every major realm of science that was known to man during this period, including physics, mathematics, chemistry, botany, zoology, physiology, and psychology. He had an extraordinary talent for observation and should perhaps be considered the first biomechanist. He viewed the animal’s body as a mechanical system and wrote the first book on the subject, De Motu Animalium, or On the Movement of Animals (4, 7, 8, 11, 17, 21). His treatises, Parts of Animals, Movements of Animals, and Progression of Animals described the action of the muscles and subjected them to geometric analysis for the first time. He recorded some practical observations: “…the animal that makes its change of position by pressing against that which is beneath it…hence atheletes jump further if they have weights in their hands than if they have not; and runners run faster if they swing their arms, for in the extension of the arms there is a kind of leaning upon the hands and wrists” (3). In the same treatise, he described the action of muscles and used geometrical analysis. Aristotle defined the act of “muscular flexion” as a change from a straight line to an angle and noted that without this “flexion,” there could not be forward progression, such as walking and swimming. This implied, for the first time, the thought or conceptualization of transforma- FIGURE 9. Archimedes studied the laws of leverage and problems related to tion of rotatory into translational motion (21). determining the center of gravity. Aristotle’s discussion regarding the problems of pushing a boat under various conditions was, in essence, a precursor of Newton’s three laws of motion. For his time, Aristotle demon- during this era. Among those figures, only Archimedes is of strated a remarkable understanding of the role of the center of significance from a biomechanics perspective. gravity, the laws of motion, and leverage (4, 21). The writings of Aristotle mark the philosophical birth of Archimedes (287–212 B.C.) kinesiology, making Aristotle the father of this field. Although Archimedes (Fig. 9) was born approximately 50 years after Aristotle was a scientist, he performed no actual experiments the death of Aristotle. He was credited for his studies relating (4, 11). His numerous writings are the results of pure, logical to the hydrostatic principles associated with floating bodies. analysis. Aristotle obviously appreciated the existence of the Archimedes studied the laws of leverage and problems related “center of gravity” in his writings: “The reason why an animal to determining the center of gravity. Hence, his treatises on the which is to walk erect must be both a biped and also have the latter have been described as the foundation of theoretical upper part of its body lighter and the parts situated beneath mechanics. Before the time of Christ, he was the last of the these heavier is obvious; for only if it were so constituted ancient Greeks to excel in the field of mechanics. Archimedes would it be able to carry itself easily” (3). He stated that, identified our most fundamental scientific tools, deductive rea- because of this fact, a man’s legs are stronger than his upper soning, and mathematical analysis (4). limbs and that children have difficulty walking because the upper part of their body is large in proportion to the lower Aulus (Aurelius) Cornelius Celsus (25 B.C.–50 A.D.) half. These findings can be related to the gravitational moments There is controversy regarding the actual role of Aulus effective at the axes of rotation of the joints of the weight- Cornelius Celsus. Although some think he was a surgeon or bearing limbs (4). Although his studies were not directly physician, he would probably be more accurately classified as related to the spine, they were the first to discuss human kine- a medical encyclopedist. He wrote on observations in the field siology, and, in part, biomechanics. of spine surgery. Celsus reported that cervical spine injury may In the third century B.C., the center of Greek civilization and be associated with difficulty breathing and death; he also medicine shifted from the old Greek settlements toward the reported that lower spine injury may cause paraparesis and new Egyptian city of Alexandria. In the cultural melting pot of urinary incontinence. His recommendations included immo- Alexandria, Greek science produced some of its greatest bilization and external stabilization, but not surgery (9). achievements. Conversely, Eastern mysticism gained greater influence on Greek thinking. Archimedes (287–212 B.C.), Euclid, Galen of Pergamon (130–200 A.D.) Praxagoras, Herophilus of Chalcedon, and Erasistratus of Ceos Galen of Pergamon (Fig. 10), another ancient Greek physi- (330–250 B.C.) are among the important figures of Alexandria cian, was born in Pergamon, but later moved to Rome and

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 387 NADERI ET AL.

FIGURE 11. In the Medieval period, Paulus of Aeginata collected ancient data in his seven-volume encyclopedia.

Galen classified spinal disorders as kyphosis, lordosis, scoliosis, and succussion. He addressed, in detail, tuberculosis of the spine and spinal injury. He analyzed the mechanism and FIGURE 10. Galen’s work as an official surgeon of the gladiators led to his the consequence of spine trauma and noted that spinal injuries being considered the “father of sports medicine.” usually follow a fall from a height or from a violent force applied to the vertebrae (16). The direction of the trauma affects the type of the injury. Galen recommended the use of the became the physician to the emperor, Marcus Aurelius. He Hippocratic board for traumatic deformities and the Hippo- is, therefore, considered by some to be a physician of the cratic ladder for kyphotic deformities. Unlike Hippocrates, Roman period. Galen worked as a surgeon and anatomist Galen was more liberal in advocating surgery to remove bone and was the founder of experimental physiology and embry- fragments pressing on the spine (9). ology. He described the muscular system as a “unified but complex organ of locomotion” and demonstrated the phys- Oribasius (325–400 A.D.) iological relationship between the nervous and muscular Oribasius, a Byzantine physician, added a bar to the systems (16). Hippocratic reduction device and used it for reduction of both Galen focused on the anatomy of animals and extrapolated spinal traumas and deformities. his findings to the human anatomy. His anatomic doctrines affected medicine for more than 1200 years, until the studies of Vesalius altered knowledge. It is of note that he worked as BYZANTINE PERIOD an official surgeon of the gladiators in the amphitheaters, leading to his being remembered as the “father of sports Paulus of Aegina (625–690 A.D.) medicine.” He confirmed the observations of Imhotep and Although Paulus of Aegina (Fig. 11) lived during the Hippocrates regarding the neurological sequences of cervical Byzantine period (330–1453 A.D.), he is recognized as a repre- trauma (2, 16, 24). sentative of ancient medicine because he collected doctrines of Galen made important contributions in anatomy and disor- the antique period in his seven-volume encyclopedia (Fig. 12). ders of spine. According to Galen, if the spine was a single, Paulus of Aegina not only used the Hippocratic bed, but he rigid bone, it would be invulnerable but also inflexible like a also worked with a red-hot iron (a cauterization device) to treat statue. In that case, man would have been deprived of motion, painful maladies. He was credited with performing the first which is a vital feature of life. On the other hand, a spine con- known laminectomy in a case of spinal fracture resulting in sisting of many small parts would be more flexible, yet the spinal cord compression and emphasized the use of orthoses in unavoidable consequence of this flexibility would have been its spinal trauma (9). vulnerability. The number of the existing vertebrae is ideal as it allows the spine to bend in a circular, rather than angular, man- Avicenna (981–1037 A.D.) ner, thus avoiding injury of the spinal cord (16). He mentioned The Medieval age consists of two distinct periods, the Dark that the vertebrae are bound ventrally and articulated dorsally. Age and the Reconstruction Era. During the Dark Age, the The ventral parts provide for motion and the dorsal parts Church was dominant in the West, which had a tremendous ensure stability and prevent excessive extension. effect on scientific studies. During this age, most of the classics

388 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART I

FIGURE 12. The cover of Paulus of Aeginata’s encyclopedia. were translated into Arabic, Syrian, and Persian. Islamic physicians recorded some of their observations and experiences regarding the spine, spinal disorders, and spine anatomy. Avicenna (Fig. 13) was one of the most prolific contributors in this era. In his treatise, Al-Qanun fi al-Tibb, or The Canons of Medicine, he wrote eight chapters on the functional anatomy of FIGURE 13. A physician and philosopher in the Medieval period, Avicenna the spine. contributed to medical literature, including eight chapters on the functional In his treatise, Avicenna sought an explanation for the anatomy of the spine. anatomic features of each spinal region. He emphasized that the shape and the size of any given vertebra is determined by royal suburb of Cordoba, in southern Spain. Because he was a its regional function. Therefore, he classified the spine into seg- prominent surgeon, he taught many students and treated many ments similar to those known today, such as cervical, thoracic, patients from all over Europe. He was an inventor of surgical lumbar, sacral, and coccyx. He then described anatomic fea- instruments and described surgical techniques for the treat- tures of elements of vertebra in any region, particularly regard- ment of spine disorders, such as sciatica, low back pain, scolio- ing the anatomy of C2, T12, and the number of sacral vertebrae. sis, and spinal dislocations. He used cauterization for painful However, despite the aforementioned anatomic errors, maladies and described a device for reduction of dislocated Avicenna described the biomechanical features of the verte- vertebrae (2). His treatise provided medical foundations for brae and the spine almost correctly. He described flexion, exten- both Arabic and European medicine. His techniques were used sion, and lateral bending aspects of the motion segments, as and reported by the Turkish surgeon, Serefeddin Sabuncuoglu well as the coupling phenomenon of the thoracolumbar spine. (1386–1470). In the 15th century, Sabuncuoglu published these The most interesting writings of Avicenna were on the bio- works in his illustrated surgical atlas, Cerrahiyetul Haniye (18). mechanics of the craniovertebral junction. Avicenna described In this text, he described the treatment of spinal dislocations. the characteristics of the atlas and the axis. He reported that the Loss of sphincter tone and motor function below the level of cranial-atlas motion segment was responsible for lateral bend- injury was categorized as a complete dislocation. Less severe ing, whereas the C0–C2 segment makes anteroposterior motion injuries were described as partial. The reduction of partial dis- possible. According to Avicenna, the odontoid process has two locations was performed by placing the patient in the prone functions: 1) it is a protector of the high cervical spinal cord and position on a wooden frame. The chest and knees were tied 2) it prevents the displacement of the thinner first cervical ver- with ropes. The physician used both hands to apply manual tebra. During anteroposterior and lateral movements of the pressure over the dislocated segment until reduction was head, the C1 and C2 vertebrae act as a single bone and move achieved. together (19). He also described the reduction technique for spinal trauma cases (2, 12). CONCLUSION

Albucasis (936–1013 A.D.) The foundations of modern spine biomechanics were laid in Albucasis, also known as Al-Zahrawi, was undoubtedly the the ancient and medieval periods by physicians, philosophers, greatest surgeon of the Middle Ages. He was born in Zahra, the and scientists. Although each of these contributions, in and of

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 389 NADERI ET AL.

themselves, often did not yield a true biomechanical break- undertaking. The understanding of spine biomechanics plays a vital through, each of them either enhanced the knowledge base of part in the management of spinal disorders. These two articles describ- the era or laid the groundwork for further investigation and ing the history of biomechanics provide a good foundation for the observation, setting the stage for a more accelerated acquisition overall understanding of spine biomechanics. of knowledge in the upcoming centuries. Volker K.H. Sonntag Phoenix, Arizona

REFERENCES he authors present a comprehensive overview of the history of spine biomechanics. Dating back more than 2000 years, they have 1. Acar F, Naderi S, Güvençer M, Türe U, Arda MN: Herophilus of Chalcedon: T detailed numerous vignettes which allow the reader to recognize that A pioneer of neuroscience. Neurosurgery 56:861–867, 2005. 2. Albertsone CD, Naderi S, Benzel EC: History of spine surgery, in Benzel EC many clinical problems present today also had to be tackled in the (ed): Spine Surgery. Techniques, Complication Avoidance, and Management. early years of spine biomechanics. Furthermore, the primitive methods Philadelphia, Elsevier Churchill Livingstone, 2005, ed 2, pp 1–21. available to healthcare providers at that time are described. The lack of 3. Aristotle: Forster ES (transl). Parts of Animals. Movement of Animals. Progression current spinal treatments obviously made successful management of of Animals. Cambridge, Harvard University Press, 1937. these afflictions much more difficult. Nonetheless, our predecessors 4. Braun GL: Kinesiology: From Aristotle to the twentieth century. Res Q managed to effectively care for their patients in the great majority of 12:164–173, 1941. cases. The authors are to be lauded for specifically tracking down these 5. Breasted JH: The Edwin Smith Surgical Papyrus. Chicago, University of Chicago historic references and bringing them together to tell a cohesive story. Press, 1930, pp 430–466. 6. Çitak G, Naderi S: History of spine biomechanics, in Naderi S (ed): Essentials Robert F. Heary of Spine Biomechanics [in Turkish]. Izmir, Publications of Spine and Peripheral Newark, New Jersey Nerve Surgery Section of the Turkish Neurosurgical Society, 2004, pp 1–8. 7. Dugas R: A History of Mechanics. New York, Dover Publications, 1988. he authors have put together an extensive review of the early med- 8. Fung YC: Biomechanics: Mechanical Properties of Living Tissues. New York, ical and surgical writings in an effort to trace the heritage of spinal Springer Verlag, 1993, pp 1–13. T mechanics. Looking at spinal mechanics and the literature this would 9. Goodrich JT: History of spine surgery in the ancient and medieval worlds. Neurosurg Focus 16:E2, 2004. seem to be a most daunting task! The authors begin with the pre-Greco- 10. Hippocrates: On Joints, in Capps E, Page TE, Ruse WH (eds): Hippocrates: The Roman period and continue up through the medieval period. By focus- Loeb Classical Library. London, W. Heinemann, 1927, vol 3, pp 200–397. ing on personalities of the time and their writings, the authors have 11. Hirt S: What is kinesiology? A historical review. Phys Ther Rev 35:419–426, been able to ferret out some pearls of history. The task is not an easy 1955. one, as the literature from this era is sometimes quite scarce and some- 12. Kumar K: Did the modern concept of axial traction to correct scoliosis exist times confusing in translation. For the individual interested in the his- in prehistoric times? J Neurol Orthop Med Surg 8:309–310, 1987. tory of spinal mechanics, this article is an obligatory read. 13. Lang JK, Kolenda H: First appearance and sense of the term “spinal column” in ancient Egypt. Historical vignette. J Neurosurg 97 [Suppl 1]:152–155, 2002. James T. Goodrich 14. Loeser JD: History of skeletal traction in the treatment of cervical spine Bronx, New York injuries. J Neurosurg 33:54–59, 1970. 15. Marketos SG, Skiadas P: Hippocrates. The father of spine surgery. Spine he first in a two-part series, this article summarizes the origins of 24:1381–1387, 1999. our knowledge and understanding of spinal disorders. While read- 16. Marketos SG, Skiadas PK: Galen. A pioneer of spine research. Spine T 24:2358–2362, 1999. ing this article, one might be surprised by the ancient origins of many 17. Martin RB: A genealogy of biomechanics. http://asb-biomech.or/ of the concepts applied in modern spine management. As Bilroth, a history/biomech/. Accessed March 20, 2005. famous surgeon, once said “Only the man who is familiar with the art 18. Naderi S, Acar F, Arda MN: History of spinal disorders and Cerrahiyetul and science of the past is competent in its progress in the future.” Haniye (Imperial Surgery): A review of a Turkish treatise written by In this first part of the series, some historical figures and epics lack Serefeddin Sabuncuoglu in the 15th century. Historical vignette. J Neurosurg major contribution to our understanding of spine biomechanics. 96 [Suppl 3]:352–356, 2002. However, they remain entertaining, especially in a period in which 19. Naderi S, Acar F, Mertol T, Arda MN: Functional anatomy of the spine by mythology and medicine were interrelated. The authors mention how Avicenna in his eleventh century treatise Al-Qanun fi al-Tibb (The canons of the use of axial traction in Indian civilization transforms a kyphosed medicine): Neurosurgery 52:1449–1454, 2003. 20. Naderi S, Ture U, Pait TG: History of the spinal cord localization. Neurosurg woman into a beauty but omit to make reference to Homer’s Odyssey Focus 16:E15, 2004. in which several cervical injuries were described. 21. Rasch PJ: Notes toward a history of kinesiology. I. J Am Osteopath Assoc Although the Western world was plunged into darkness during the 57:572–574, 1958. medieval period, some important European figures thrived in this envi- 22. Sanan A, Rengachary SS: The history of spinal biomechanics. Neurosurgery ronment, especially in Italy where cadaveric dissections were permit- 39:657–669, 1996. ted on a limited basis. One such man was Theodoric of Bologna 23. Wilkins RH: Neurosurgical Classics I. Park Ridge, American Association of (1205–1298) who described in detail the examination of a patient expe- Neurological Surgeons, 1992, pp 1–5. riencing a cervical injury and the nonsurgical treatment of his cervical 24. Wiltse LL: The history of spinal disorders, in Frymoyer JW (ed): The Adult dislocation by reduction and stabilization. Spine. Principles and Practice. Philadelphia, Lippincott-Raven, 1997, pp 3–40. It is interesting to note that most of the pathologies reported during this ancient and medieval period were of traumatic lesions and infec- COMMENTS tious spinal deformities. Rarely were degenerative diseases described. Paul Khoueir he authors extensively review the history of biomechanics in two Michael Y. Wang Tparts. For the most part, they have succeeded with this ambitious Los Angeles, California

390 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART I

he authors provide an illuminating orientation to the accomplish- make possible, I ﬁrmly believe that without proper orientation to physi- Tments of Old World scientists and philosophers. These philosophers ology, anatomy, and philosophy, we are depriving our patients of the and scientists made invaluable contributions to the medical sciences best treatment possible. I enjoyed reading this article immensely and am under the most extreme and challenging conditions. It is regrettable that grateful for the many new insights it contained. some neurosurgeons today believe that the only means of achieving an efﬁcient surgery is through technology. While recognizing the impor- Yucel Kanpolat tance of technology and the extraordinary medical advances that they Ankara, Turkey

World Heavyweight champion Muhammad Ali’s (1942 –) knockout victory over Sonny Liston (1932–1971) happened less than two minutes into the ﬁrst round of their rematch on May 25, 1965 in Lewiston, Maine.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 391 LEGACIES

HISTORY OF SPINE BIOMECHANICS: PART II— FROM THE RENAISSANCE TO THE 20TH CENTURY

Sait Naderi, M.D. SPINE BIOMECHANICS PROVIDE the foundation for the disciplines of spine medicine Department of Neurosurgery, and spine surgery. Although modern spine biomechanics emerged during the second Yeditepe University half of the last century, it has many ancient, medieval, and post-Renaissance roots. In School of Medicine, Istanbul, Turkey Part I of this series, the ancient and medieval roots of spine biomechanics were reviewed. In Part II, the effects of post-Renaissance scientists on the development of modern spine Niteen Andalkar, M.D. biomechanics, as well as the studies on gait, bone, and muscles performed before the Department of Neurosurgery, 20th century, are reviewed. Subsequently, war-related studies performed in the 20th Cleveland Clinic Spine Institute, century contributed to the formation of modern biomechanics. The first biomechanics- Cleveland Clinic Foundation, Cleveland, Ohio related organizations and scientific publications did not emerge until the second half of the 20th century. These events provided the final bricks in the foundation that facil- Edward C. Benzel, M.D. itated the emergence of modern spine biomechanics research. Department of Neurosurgery, KEY WORDS: Biomechanics, History, Spine Cleveland Clinic Spine Institute, Cleveland Clinic Foundation, Neurosurgery 60:392–404, 2007 DOI: 10.1227/01.NEU.0000249263.80579.F9 www.neurosurgery-online.com Cleveland, Ohio

Reprint requests: Edward C. Benzel, M.D., he most important findings and advance- world to the works of the ancient civilizations. Department of Neurosurgery, ments regarding spine biomechanics These manuscripts encouraged the study of Cleveland Clinic Spine Institute, were defined during the second half of pre-Renaissance works. This, in turn, encour- Cleveland Clinic Foundation, T the 20th century. The progress enjoyed during aged and facilitated the efforts of the Renais- 9500 Euclid Avenue, S-80, Cleveland, OH 44195. the latter half of the 20th century had its roots sance scientists. This ultimately led to a scien- Email: [email protected] in the ancient age, the Renaissance, and the tific revolution during the Renaissance. post-Renaissance era. This is particularly true The scientists of the 17th century studied Received, February 20, 2006. of the latter part of the 17th century, also mathematics, mechanics, and occasionally bio- Accepted, October 13, 2006. known as the Century of Scientific Revolution, mechanics. The 18th century was the century and the 19th and 20th centuries. As noted in in which gait was predominantly studied. Part I of this series, the eras and periods have Many studies on muscles were performed as been arbitrarily chosen and are depicted as well (9, 27). Many more studies on bone overlapping. The significant contributors are mechanics were performed in the 19th century. portrayed for reference purposes in Figure 1. The scientists of the 19th century sought to The most significant contributions of ancient relate bone trabecular architecture to its and medieval scientists to the discipline of biomechanical, load-bearing attributes. During mechanics included an awareness of the nor- this era, it was commonly thought that bony mal and pathological anatomy of the human architecture responded anatomically to stress spine, a heightened knowledge base regarding (10, 48, 50, 61, 73, 74, 80). This process resulted traumatic and non-traumatic spine deformi- in the definition of Wolff’s law. These studies ties, and an understanding of the gait patterns and subsequent studies in the 20th century of mammals (46). It is notable that the majority contributed to the development of biomechan- of these aforementioned contributions were ics as a new discipline. derived from the physicians and philosophers of the ancient age. The contributions by THE RENAISSANCE medieval scientists were of less importance and, importantly, they often represented the reintro- It is commonly thought that modern scien- duction of concepts and knowledge drawn tific thought was born in the Renaissance, a from ancient manuscripts. Translations of period during which religious influence dimin- ancient manuscripts from Arabic to Western ished. Scientists began to probe and discover languages opened the door for the rest of the the wonders of the human body and other sci-

392 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART II

paintings actually functioned. da Vinci completed more than 750 drawings on the basis of his anatomic dissections on 10 cadavers (Figs. 3–5). In these meticulous drawings, he applied his knowledge of mechanical principles to the study of human anatomy by concentrating on dynamic illustrations of joints, muscles, bones, ligaments, tendons, and cartilage. Leonardo da Vinci was the first to accurately describe the human adult S-shape spinal posture with its curvatures, articulations, and vertebrae (notably, with the number of vertebrae portrayed accurately). He emphasized the contribution of muscles to cervical spine stability and described a method by which FIGURE 1. Timeline from the Renaissance period through the 19th century. the spine provided stability to the human body. He wrote “You will first make the spine of the neck with its ten- entific phenomena, rather than focusing on religious issues. It dons like the mast of a ship with its side-riggings (trans- is notable that the Renaissance initially provided few signifi- verse or spinous processes), this being without the head. cant, direct contributions to the sciences, compared with the Then make the head with its tendons (muscles that can substantial contributions to the arts made during this era. provide active force of effort) which (attached to the side There were essentially two relatively distinct periods of sci- riggings) gives it (the head) its movement on its fulcrum entific development before and during the Renaissance. The (spinal joints)” (68, p 660). first period began during the second half of the 11th century and peaked in the 13th century. The second period began in He stated “nature cannot give the power of movement to ani- the 16th century. Whereas the first period was characterized mals without mechanical means” (68, p 660). He seems be the by the translation of ancient manuscripts from Latin or Arabic first to understand the principles of lever systems, as applied to languages, the second period was characterized by studies in human motion. da Vinci analyzed the mechanics of walking, the arts (27). both up- and downhill, as well as rising from the seated position. The only major scientific contributions during the beginning However, delightful as his notebooks are to explore, they of the Renaissance were those of Leonardo da Vinci. The main were personal works and remained unpublished for centuries. topic of discussion during the 16th century was that of method- As a result, the brilliant recordings of his daydreams had little ology. Sir Francis Bacon (1561–1626) was the first scientist to criticize scholarly traditions. Rene Descartes (1596–1650) presented and introduced very valuable mathematical data and theory. At the beginning of the 17th century, there were no important advances in physics other than the development of the magnet. After this timeframe, however, the combination of mathematics and scientific experiments led to progress in physics. The first half of the 18th century was, for the most part, non-productive. During the second half of this century, however, many studies were performed by Leonard Euler (1707–1783), Joseph Lagrange (1736–1813), and Pier Simon Laplace (1749–1827). They set the stage for modern scientific and biomechanical thought. A discussion of the major scientists and contributors during this era is in order. Leonardo Da Vinci (1452–1519) Leonardo da Vinci (Fig. 2) was an artist, engineer, and scientist who contributed substantially to the understanding of biomechanics. Born in 1452, da Vinci became famous as an artist, but worked and functioned predominantly as an engineer. He made major contributions to the study of mechanics in the course of pursuing his numerous engineering projects and innovations. He had an understanding of the components of force vectors, friction coefficients, and the acceleration of falling objects. da Vinci also demonstrated an understanding of Newton’s third law (27, 68). He studied muscles because he wanted to understand how FIGURE 2. Leonardo da Vinci contributed substantially to the understand- the human physical specimens that he portrayed in his ing of biomechanics.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 393 NADERI ET AL.

FIGURE 5. Illustration by da Vinci depicting the relationship between the cervical roots and the cervical spine.

FIGURE 3. Illustration by da Vinci depicting the human spine and paraspinal muscles.

FIGURE 6. The frontispiece of Vesalius’ The Human Corporis Fabric, which was the ﬁrst book on modern anatomy.

scientiﬁc impact. Hence, his studies and observations had little effect on the scientiﬁc literature during his lifetime (51). Andreas Vesalius (1514–1564) The development of the study of anatomy contributed to the understanding of spine biomechanics. Andreas Vesalius produced an entire series of anatomic drawings (or plates) in “De Humani Corporis Fabrica” (Fig. 6). Many consider this work, pub- FIGURE 4. Illustration by da Vinci depicting the human spine. lished in 1543, to have heralded the era of modern medicine.

394 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART II

Galileo Galilei (1564–1642) According to the eminent scientific writer Gribbin, Galileo Galilei is the person who most deserves the title of “first scientist” (27). He not only applied what is essentially modern scientific methodology to his work, but he also confidently laid down the ground rules for others to follow. Galileo Galilei was born 21 years after the death of Copernicus. The strength of his powerful, irascible per- sonality dominated the scientific world of his time. Thus, he became the great animat- ing spirit of the scientific revolution that followed the Renaissance. Galileo Galilei was a student of medicine before he became famous as a physicist. Galileo’s fame was so great and his lectures in FIGURE 7. Andreas Vesalius, pioneer of modern anatomy. Padua were so popular that his influence on biomechan- A review of this book reveals that, like da Vinci, he also accu- ics went far beyond his per- rately described the nuances of spine anatomy in great detail (5). sonal contributions. He was Vesalius (Fig. 7) described the intervertebral disc, he termed particularly aware of the the spine the “dorsum” (backbone) (Fig. 8), and he observed mechanical aspects of bone that the spine provides the route of passage for the spinal cord, structure (51). In Discourses as did Galen and Avicenna. According to Vesalius, the spine on Two New Sciences (1638), was defined as the “keel of the body, composed of 34 bones Galileo noted the following: (vertebrae). The neck has seven bones…by means of the first of ”The mass of animals in- these bones, we move the head diectly forward and backward. crease disproportionately By the use of the second vertebra (to which a prominent to their size, and their process resembling a canine tooth is attached) we turn the bones must consequently head…” (5). Although such statements regarding the biome- also disproportionately chanics of the upper cervical spine were not new and had been FIGURE 8. Illustration of the spine increase in girth, adapting recorded by Avicenna in the 11th century, Vesalius described from The Human Corporis Fabric to load-bearing rather than partitions of the spinal column and foramina in detail. (back bone of Vesalius). mere size” (51). It is notable that Vesalius did not address the cervical and the “The bending strength of a lumbar curves of the spine. As was the case with many scien- tubular structure, such as a bone, is increased relative to its tific discoveries in the past, Andreas Vesalius built upon the weight by making it hollow and increasing its diameter” (51). rediscovery of the work of Galen. The studies of da Vinci and “Marine animals can be larger than terrestrial animals because Vesalius are indicative of the importance of a methodological the water’s buoyancy relieves their bones of weight bearing dissection by specialists in scientific studies. The notion that responsibilities” (51). human anatomy was an objective discipline based on observa- He also stated that to preserve the strength of a tubular bone tion and well-defined scientific principles began to emanate while increasing its length three times, it is necessary to within the scientific community. da Vinci and Vesalius, both increase its thickness nine times. His work gave impetus to the men of modern science, were instrumental in bringing about study of mechanical events in mathematical terms, which, in these changes. Their focus on revealing anatomic secrets turn, provided a basis for the emergence of kinesiology as a sci- through the investigation of human cadavers was, henceforth, ence. Galileo showed that mathematics was the essential key to brought to the forefront of the scientific community. science, without which nature could not be properly under-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 395 NADERI ET AL.

stood. This outlook inspired Descartes, a great mathematician, Castelli arranged for Borelli to pursue the discipline of physiology. to teach a lectureship on mathematics in Messina. In a Rene Descartes (1596–1650) short time, he was recog- Rene Descartes was not a major contributor in the field of nized throughout Italy in the biomechanics; his thoughts, however, had an indirect impact on fields of mathematics, physi- the field. Descartes was one of the founders of mechanical phi- ology, physics, and astron- losophy. He was the mastermind behind the creation of the omy (63). He then became Cartesian coordinate system. This philosophy suggested that professor of mathematics in changes observed in the natural world should be explained Pisa where he met Marcello only in terms of motion and rearrangements of the parts of Malpighi (Fig. 10). Both were matter. Descartes’ mathematical theory of mechanics provided founding members of the the basis for the maturation of the science of mechanics in the short lived “Accademia del 18th century. In 1624, Descartes published the first paper Cimento.” Around this time, devoted to physiology, L’homme. This treatise emphasized the Borelli also studied anatomy. FIGURE 10. Marcello Malpighi, an anatomist associate of Borelli. theory that movement was coordinated through the nervous The association between system. Descartes’ application of mechanics to humans Malpighi and Borelli resulted included the belief that humans were soul-containing organic in many new works. Malpighi stimulated Borelli’s interest in machines running on auto pilot (27). living beings, whereas Borelli stimulted Malpighi to investi- In 1675, Descartes stated in “Tractus de Homine et de Forma- gate the manner in which living systems work (51, 52). tione Fœtus” (“A Treatise on Humans and the Formation of Malpighi recalled, “What progress I made in philosophizing the Foetus”) that “all of animal physiology could be ex- stems from Borelli.” Borelli states this about Malpighi: “I plained by mechanics.” His following statement is well- worked hard dissecting living animals at his home and observ- known: “the body is a machine (lever is machine) made by ing their parts to satisfy his keen curiosity” (50). the hand of God!” (68, p 661). Borelli applied the principles of “Equilibrium of Rotation” and “Equilibrium of Translation” to spinal biomechanical Giovanni Alfonso Borelli (1608–1679) analysis. One of the most important mechanical features of Giovanni Alfonso Borelli (Fig. 9) was the founder of the con- animal (and human) motility observed by Borelli was that cept of iatrophysics (i.e., medical physics). Born in Naples, Italy, muscles act via short lever arms. Therefore, a joint transmit of on January 28, 1608, he studied mathematics in Rome and was force that is “n” order of magnitude greater than the weight of a student of Benedetto Castelli, who was a student of Galilei. the load applied (or lifted). Borelli essentially, and appropriately, discredited older concepts of muscle action that implied that long lever arms were required to allow weak muscles to move heavy objects. His book, De Motu Animalium, or On the Movements of Animals, published in 1680 shortly after his death, was the first in the field of biomechanics (Fig. 11). The first part of this book contained studies of external motions of the musculoskeletal system, whereas the second part contained studies on internal motions, such as muscle physiology and blood circulation. He defined his purpose in an introductory statement as follows: “Animals are bodies and their vital operations are either movements or actions which require movements. But bodies and movements are the subject of mathematics. Such a scientific approach is exactly geometry. Similarly, the oper- artions of animals are carried out using instruments and mechanical means such as sales, levers, pulleys, winding- drums, nails, spirals, etc.” (7). This book provided many calculations regarding spine biomechanics, such as Borelli’s calculation of forces on spine musculature and intervertebral discs (Fig. 12). He noted that the spine had to be “stable, much like the hull of a ship” (6), and also noted that the vertebrae were flat and wide to prevent dis- FIGURE 9. Giovanni Alfonso Borelli, known as the father of biomechanics. locations and to provide stability.

396 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART II

FIGURE 11. The cover of Borelli’s book, De motu Animalium, which was the ﬁrst comprehensive book on biomechanics. FIGURE 13. Illustration demonstrating Borelli’s load-sharing concept from De motu Animalium.

third of the resistance of the intervetebral disc is equal to 826 pounds. The muscular forces are equal to 413 pounds and the forces exerted by the disc are equal to 1239 pounds” (7). These calculations revealed that Borelli was aware of the load-sharing concept in spine biomechanics (Fig. 13). Borelli knew that for adequate flexibility of any animal, the spine must be divided into multiple segments by articulations. He noted that the intervertebral discs play an important role in spine biomechanics. According to Borelli, the intervertebral disc is a viscoelastic substance and functions as a cushion preventing attrition of the bone. He also noted that fibers comprising the intervertebral disc are much stronger than those in muscle. Therefore, he reported that the majority of the spinal load is borne by the intervertebral discs, with a much smaller portion borne by the spinal musculature. Borelli was the first to experimentally determine the position of the human center of gravity (50). He used a wooden plank and trihedral pyramidal system to precisely balance a person (Fig. 14), observing a point located between the pelvis and the buttocks. The validity of this technique was confirmed approximately 200 years later by Braune and Fisher (11) in frozen FIGURE 12. Borelli’s illustration depicting the spine, muscles, and intervertebral discs from De motu Animalium. cadavers. After Borelli, there is little sign of biomechanical study in Borelli calculated the effect of a load borne in the neck. For the literature until the latter half of the 19th century. Due to his example, he noted, “If the spine of a stevedore is bent and sup- early and substantial contributions, Borelli is widely recog- ports a load of 120 pounds carried on the neck, the force exerted nized as the father of biomechanics (62, 63). by nature in the intervertebral discs and in the extensor muscles of the spine is equal to 25,585 pounds. The force exerted by the Robert Hooke (1635–1703) muscle alone is not less than 6404 pounds. . . . Therefore, the sum According to Hooke’s law, it is estimated that no solid is per- of muscular forces which control the fifth lumbar vertebra and a fectly rigid. When several external forces act on a solid at rest

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 397 NADERI ET AL.

Newton is also credited with the first observations regarding the parallelogram of force, based on his observation that a moving body affected by two independent forces acting simultaneously moved along a diagonal equal to the vector sum of the forces acting independently (63). Newton’s laws of motion, along with his idea of the parallelogram of force, can be readily applied to spine biomechanics. Leonard Euler (1707–1783) Leonard Euler is another important individual who studied mathematics, astronomy, physics, and biomechanics. In 1736, FIGURE 14. Illustration of the method used by Borelli to experimentally Euler published a systematic introduction to mechanics in determine the human center of gravity from De motu Animalium. Mechanica sive motus scientia analytice exposita, or Mechanics or motion explained with analytical science. He stated that the human and the resultant net force is zero, the solid remains at rest. spine carries compressive loads like a column and that such Hooke’s law expresses that for small displacements, the size of loads may lead to instability or failure. He studied mathematic the deformation is proportional to the deforming force. This models to derive these findings (27, 68). law is of significant importance when one considers the forces Thomas Young (1773–1829) applied to the spine by a spine instrumentation construct (as well as the response of the construct to these forces). For larger Thomas Young studied the formation of the human voice displacements, however, the neutral zone is exceeded and the and identified it as resulting from vibrations. He connected elastic limit is reached. This is the point at which the force this process with the elasticity of materials. He improved on departs from the linear relationship between the size of defor- Hooke’s law by providing a measure, Young’s modulus. mation and the deforming force. Exceeding the elastic limit Young’s modulus defines a proportionality between force and causes the solid to acquire a permanent set; if the external stretch or compression for different substances. This modulus forces are removed, the solid does not spring back to its unde- is a measure of the elastic properties of stretchable and com- formed configuration. The solid will ultimately fail if further pressible bodies. It is defined as the limit, for small strains, of forces are applied. This point is termed the point of failure (27). Apart from the definition of Hooke’s law, Hooke studied gravity and was the first scientist to use the term “cell.” Isaac Newton (1642–1727) Isaac Newton (Fig. 15) made many significant scientific contributions, but did not write specifically about biomechanics. His findings regarding calculus, laws of motion, and analytical portrayals of the hydraulic characteristics of viscous fluids, however, were critical regarding the emergence of biomechanics as a field of study. His publication of Philosophiae Naturalis Principia Mathematica in 1686 presented laws of motion that are used today to describe and define motion. These laws express the relationship between the forces and their effects (27, 64). Newton’s First Law of Motion (Law of Inertia) Every body continues in its state of rest, or of uniform motion, in a right line, unless it is compelled to change that state by forces impressed upon it. Newton’s Second Law of Motion (Law of Momentum) The change of motion is proportional to the motive force impressed and made in the direction of the right line in which that force is impressed. Newton’s Third Law of Motion (Law of Interaction) To every action there is always opposed an equal reaction; or, the mutual actions of two bodies upon each other are always FIGURE 15. Isaac Newton’s contributions were critical in the emergence of equal and directed to the contrary parts. the field of biomechanics.

398 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART II

the rate of change of stress with strain. Young’s modulus can be thorough analysis of gait from photographs taken simultane- determined experimentally from the slope of a stress-strain ously by four cameras (10). curve created during tensile or compressive tests conducted on a sample of the material. Young’s modulus is extensively Julius Wolff (1836–1902) and Wolff’s Law used in spine biomechanics today (27). The definition of Wolff’s law may be one of the most important events defining the field of biomechanics. Julius Wolff THE 19TH CENTURY established this law on the basis of his own work and the studies of earlier scientists. Therefore, it is necessary to address the A variety of scientific studies were performed during the studies supporting his work. 19th century; however, a limited number of studies were pub- In 1832, Marc Jean Bougery (1797–1849), Claude Bernard lished on muscle, nerve, and bone physiology. The majority of (1813–1878), and Nicolas Henri Jacob (1782–1871) raised the the publications resulted from the collaboration between engi- question of the relationship between the architecture and the neers and physicians. The Weber brothers, Christian Wilhelm mechanical function of bones and assumed that, in the neck of Braune, Otto Fischer, and Julius Wolff were among the scien- the human femur, there was a line of compression along which tists contributing to the field of biomechanics during this era. the trabeculae seemed to be particularly dense and strong (8). In 1838, F.O. Ward, a London anatomist, compared the archi- The Weber Brothers tecture of the femoral neck with that of a street lamp in a trian- Ernst Heinrich Weber (1795–1878), Wilhelm Eduard Weber gular wall-bracket in Outlines of Human Osteology. He reported (1804–1891), and Eduard Friedrich Wilhelm Weber (1806–1871) that the horizontal trabeculae in the bone were responding to espoused the idea that the human torso was maintained in the stress and the oblique trabeculae were responding to pressure erect position primarily via tension of the ligaments, with little (74). This was a unique study that addressed stresses on bone. or no muscular exertion. The Webers were the first to investi- During this era, engineers were studying and analyzing gate the reduction in the length of an individual muscle during stresses associated with railways, bridges, cranes, etc. Ward contraction and devoted much study to the role of bones as applied the findings of engineers of his time to biological sys- mechanical levers (75). They were also the first to describe, in tems, specifically bone. chronological detail, the movement of the center of gravity. The In 1867, Hermann von Meyer (1801–1869), an anatomist from modern concept of locomotion originated with the studies of Zürich and author of The Architecture of the Spongiosa, and Karl Borelli (7). Very little was accomplished in this field before the Culmann (1821–1881), an engineer from Germany, compared Webers’ publication of Die Mechanik Der Menschlichen Gehwerk- the stresses on the femoral neck and on a crane. They showed zeuge (Mechanics of the Human Gait) in 1836 (76). Their treatise, that the structure of the femoral neck, which supports the torso, which still stands as a classical work in the field, was based was mathematically equivalent to a crane (48). They discovered solely on observations. Nevertheless, it firmly established the a remarkable similarity between the trabecular architecture of mechanism of muscular action on a scientific basis. the proximal femur and the patterns of stress trajectories, calculated using “Graphical Statics,” a new methodology devel- Christian Wilhelm Braune (1831–1892) oped by Culmann (73). It is said that Culmann, on seeing a lon- and Otto Fischer (1861–1917) gitudinal section through the proximal end of the femur In 1891, the first three-dimensional mathematic analysis of prepared by von Meyer, exclaimed: “This is my crane!” (50). In human gait was conducted by Wilhelm Braune and his stu- 1881, Wilhelm Roux suggested that the formation and func- dent Otto Fischer. It was published in their book Der Gang des tional adaptation of trabecular architecture in bone is regulated Menschen (Human Gait) (10). Their major premise was that locally by cells that are governed by mechanical stimuli (66). In knowledge of the position of the center of gravity of the human 1883, Hugh Owen Thomas (1834–1891) mentioned, “Eccentric torso and of the body’s component parts was fundamental to forms, that cannot be altered in the dead body without rupture an understanding of the resistive forces that the muscles must or fracture can, during life, be altered by mechanical influence, overcome during movement. Braune and Fischer (11) per- as time and physiological action will command the part to the formed a very careful study of mass, volume, and the center of direction of the employed force” (70). mass of three adult male cadavers and their body segments. In 1890, a university clinic for orthopaedic surgery was The cadavers, each of which had committed suicide, were close opened in Berlin, Germany with Julius Wolff (Fig. 16), an to the average build of German soldiers of that period. To avoid orthopaedic surgeon, as its first director. Before beginning to fluid loss, the cadavers were kept frozen throughout the study. work as director of this department, Wolff was a pupil of The center of mass of each body segment was not estimated by Langenbeck, who suggested that Wolff’s doctoral thesis should the use of balance plates, as in the previously described stud- be on the experimental production of bone in animals. Apart ies, but rather by driving thin rods into the tissue and hanging from the aformentioned studies, he reviewed the works of the body segment from three axes. The intersection of three Belchior, Hunter, Duhamel, and Flourens, all of whom had pub- externally fixed planes, e.g., vertically through each of the axes lished on osteogenesis (48). Roentgenography was not yet avail- formed on the segment, corresponded to the center of mass. able. To analyze trabecular architecture, Wolff made thin sec- After these preliminary studies, they were able to provide a tions from bone. In 1892 he wrote his book, “Das Gesetz de

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 399 NADERI ET AL.

Transformation de Knochen,” or turies, with the aim of reorganizing gymnastics in the United The Law of Bone Remodeling, States (17, 42). which was the culmination of World Wars I and II resulted in many casualties and many the knowledge gained from disabled veterans requiring long-term care. Many war-related the aforementioned project. injuries occurred among airplane pilots. Both patient popula- Wolff thought that bones tions presented significant opportunity for biomechanical were formed by satisfying a analysis. Jules Amar (1879–1935), using force and motion meas- mathematical optimization urement techniques, was one of the first researchers to provide rule. Apparently, Culmann’s a biomechanical evaluation of the gait and task performance of findings provided a blueprint thousands of disabled veterans in France. His book, The Human for a design of bone for Wolff Motor, was published in France in 1914 and was translated into FIGURE 16. Julius Wolff, whose who described a trabecular English in 1920 (4, 17). “trajectoral hypothesis” forms the orientation in healthy and World War II paralleled the introduction of high performance basis of one of the most important ele- aircraft, which led to an interest in spine biomechanical testing. deformed bones. He attrib- ments of spine biomechanics. uted this orientation to the At that time, ejection seats of airplanes could not be handled assumed capacity of bone to form and adapt its architecture in manually and complications were evident because of the lack accordance with externally applied loads. He determined that of extraction of German pilots from airplanes during an emer- bony deformation led to changes in internal structure and sec- gency. A multitude of spinal column injuries resulted, leading ondary adaptive microarchitectural changes. He stated that Siegfried Ruff to perform spine biomechanics studies on this “every change in the form and function of a bone, or function subject (67). alone, is followed by specific definite changes in the internal Similar studies were performed to test the strength of the architecture and equally definite secondary changes in its exter- spinal column. To achieve good spinal posture at the moment nal configuration, in accordance with mathematical laws” (80). of ejection, Olof Perey tested ejection seats from Swedish J-21 From this work, Wolff concluded that trabecular morphology fighters in 1945, and the Martin-Baker aircraft company per- matches the stress trajectories. This conclusion is known as his formed tests in England in 1944 (15). “trajectorial hypothesis” and forms the basis of Wolff’s law. The United States Air Force initiated similar studies to design Wolff’s law has become one of most important elements of ejection seats for aircraft in 1945. Contemporary researchers at spine biomechanics. Wayne State University performed studies on the biomechani- The law formally states that “Every change in the form and cal aspects of the spinal column, while teams at Massachusetts function of a bone, or of function alone, is followed by specific General Hospital and Massachusetts Institute of Technology definite change in its internal architecture and equally definite studied the properties of the intervertebral disc (15). secondary changes in its external configuration, in accordance Contemporary clinical studies were performed as well. with mathematical laws.” Elsewhere, the law states that Friedrich Pauwels (1885–1980) (Fig. 17) and Nikolai A. “Structure is nothing else than the physical expression of func- Bernshtein (1896–1966) were among the scientists who system- tion…under pathological conditions the structure and form of atically studied musculoskeletal biomechanics in the first half the parts change according to the abnormal conditions of force of the 20th century (50, 61). transmission” (80). In addition to the aformentioned studies performed in Europe, some were per- THE EMERGENCE OF MODERN formed in the United States. BIOMECHANICS IN THE 20TH CENTURY A “myodynamics laboratory” was established within the Although, there were a limited number of studies published Department of Surgery of the in the field of biomechanics (21), the collaboration between University of Rochester engineers and physicians contributed to the understanding of School of Medicine and biomechanical principles and the publication of high quality, Dentistry in 1926 by Russell meaningful works. The biomechanical studies from the first Plato Schwartz (1894–1965). half of the 20th century focused predominantly on acquiring Dr. Schwartz’s intention was information regarding gymnastics education and sports activ- to devise mechanisms for the ities in schools, gait and musculoskeletal analyses performed accurate recording of human for general health problems, World War I- and II-related locomotion to establish injuries, and analysis of craniospinal trauma from motor vehi- norms for both normal and FIGURE 17. Friedrich Pauwels was cle accidents. Wilbur Bowen, Tuth Glassow, William Skarstrom, abnormal gait. Beginning in one of the most prominent scientists Gladys Scoth, Louise Alley, Arthur Steindler, Katharine Welles, the mid-1930s, research in the to study musculoskeletal biomechan- Marion Broer, and John Cooper were among the scientists who myodynamics laboratory ics in the first half of the 20th cen- focused on biomechanics in the late 19th and early 20th cen- was focused increasingly on tury.

400 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART II

the development of the “functional” principles of shoe design, cal spine specimens (C2–T1). Using four 0.8-mm steel balls with a continued perfection of gait recording instruments and inserted into each vertebra and quantitative stereoradiography, the development of such surgical tools as the mirrorscope. The he measured the three-dimensional relative motion between laboratory maintained its interest in the pure mechanics of vertebrae. He studied a total of 28 specimens and found no human locomotion and the application of these studies to the effect of age or extent of degeneration on motion (49). diagnosis and management of gait abnormalities, whether Hirsch and his contemporary researchers were, therefore, the caused by injury or congenital conditions. The establishment of pioneers of modern biomechanics (36). They carried out their this rudimentary biomechanics laboratory provided a vision studies in well established laboratories, and their fellows and a direction for studies on spine biomechanics (3). founded similar laboratories in their respective new institu- Similarly, in the biomechanics laboratory at Wayne State tions. This increased both the quantity and quality of University in Detroit, Michigan, Professor Herbert Richard biomechanics-related research in the 1950s and 60s (12, 36, 38, Lissner (1908–1965), an engineer, and Professor E. Stephen 39, 53–57, 65, 69). It also established the process of biomechan- Gurdjian (1900-1985), a neurosurgeon, initiated studies on head ics education and the proliferation of qualified bona fide injury and cranial fracture mechanisms in 1939 (King A, per- researchers and educators. This contribution of Hirsch, more sonal communication). Professor Lissner studied spine biome- than any other research endeavor, secured the future of the dis- chanics in the early 1950s (20, 28, 29, 31). Together Lissner and cipline of spine biomechanics. Gurdjian attempted to determine the effects of axial compression and transverse bending on lumbar disc herniation. This Finite Element Analysis represents one of the first true modern spine biomechanics A brief history of finite element analysis (FEA) was experiments. reported by Peter Widas (79). According to Widas, FEA was They also sought to determine the reason for thoracolumbar first developed in 1943 by Courant and Hilbert (14) who wedge fractures in pilots ejecting from disabled military air- used the Ritz method of numerical analysis and minimiza- craft. Lissner built a vertical accelerator in an elevator shaft at tion of variational calculus to obtain approximate solutions the school of medicine to duplicate this injury in cadavers to vibration systems. In 1956, a study published by Turner (King A, personal communication). et al. (71) established a broader definition of numerical These preliminary studies were followed by other studies in analysis. The study focused on the stiffness and deflection of the 1950s, including those of Virgin (72), Hirsch (35, 37), Hirsch complex structures (79). In the late 1950s, the continuum and Schajowicz (41), Hirsch and Nachemson (40), Evans (18, model of the spine was first developed within the aviation 19), Evans and Lissner (20), Higgins (34), Friberg and Hirsch industry to determine the relationship between emergency (22), Sylven et al. (69), and Werne (77). These works led to the pilot ejection and the risk for spinal injury (33). This model performance of the first studies of bending moment in the spine evolved over the years. with the load-deflection, energy-absorption, and bending- By the early 1970s, FEA was limited to expensive main- moment studies of Evans and Lissner (20). frame computers, which were generally owned by aeronau- One of the pioneers in this era was Carl Hirsch (1913–1973), tics, automotive, defense, and nuclear industry companies. an orthopaedic surgeon from Sweden (Fig. 18) who performed Since the rapid decline in the cost of computers and the phe- biomechanical studies on the knee, hip, and spine in the 1940s nomenal increase in computing power, FEA has attained (35). Hirsch’s studies captured the imagination of many scien- incredible precision. FEA techniques have been used with tists, and many surgeons and engineers visited his center in the increasing frequency, including use for spine biomechanics 1950s and 60s (2). Victor Frankel, George Galante, Augustus applications (25, 30, 32, 44, 59). Today, FEA is used for the bio- White, Wilson C. Hayes, and Albert B. Scultz were among the mechanical assessment of healthy and pathological spine American scientists who vis- states and for the testing of spine implants in many biome- ited Hirsch’s laboratory dur- chanics laboratories. ing this era (Panjabi MM, personal communication). Clinical Studies Lysell was probably the Besides biomechanical laboratory research, many physicians first modern researcher to applied the results of laboratory studies to the clinical arena. conduct a thorough in vitro The term “stability” had been defined on many occasions by study of cervical spine motion multiple authors (78). This led surgeons to develop scales and and patterns of motion. He scoring systems, as well as to define column concepts. The two- also provided a comprehen- column system for spine stability was defined in 1962 by Sir sive review of the literature Frank Holdsworth (43) and the three-column system by Francis in which he credits Weber Denis (16). The definition of tumor-related instability required (1827) with performing the the definition of six columns (47). In this vein, Edward C. first objective assessment of FIGURE 18. Carl Hirsch, ortho- Benzel (6) described a cube system for determining stability spinal motion. Lysell used paedic surgeon and one of the pio- that addressed the integrity of only the ventral column (verte- fresh whole cadaveric cervi- neers of modern biomechanics. bral body and intervertebral disc).

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 401 NADERI ET AL.

Biomechanical Books, Journals, and Organizations In summary, the progress and maturation of biomechanical In the mid-1960s, the American Society of Mechanical studies in the past two centuries has, in part, led to an increase Engineers published a collection of articles on spine biomechan- in the volume and quality of publications. War-related disabil- ics in a monograph edited by Y.C. Fung (23). In 1967, Byars, ities, general health problems, and automotive industry-related Contini, and Roberts edited an American Society of Mechanical injuries created a demand for a myriad of innovative and clin- Engineers monograph (13), including an introduction by Lissner ically relevant works. In turn, this process helped craft the com- with the provocative title “Biomechanics- What is it?” Many plex environment that the field of spine biomechanics enjoys notable biomechanics books have been published since the today. The rapid emergence of a growing number of laborato- 1960s. These books were followed by others (1, 6, 24, 26, 60). ries and organizations, along with the involvement of a large The first biomechanics journal, Journal of Biomechanics, and growing number of scientists, has resulted in remarkable advancements in the field. was established in 1967. Currently, many journals publish biomechanics-related manuscripts. In addition to the well- known neurosurgical and orthopedic journals, the following REFERENCES journals contain articles related to biomechanics: • Bone 1. Adams M, Bogduk N, Burton K, Dolan P: The Biomechanics of Back Pain. Edinburg, Churchill Livingstone, 2002. • Clinical Biomechanics 2. Anonymous: Carl Hirsch. J Bone Joint Surg Am 56:210–211, 1974. • Computer Methods and Programs in Biomedicine 3. Anonymous: History of the study of locomotion. http://www.guardin. • Electroencephalography and Clinical Neurophysiology curtin.edu.au/cga/history/enlightment.html. Accessed April 8, 2005. • Gait and Posture 4. Amar J: The Human Motor, or the Scientific Foundation of Labour and Industry. New York, G. Routledge, 1920. • Injury 5. Benini A, Bonar SK: Andreas Vesalius 1514–1564. Spine 21:1388–1393, 1996. • Journal of Applied Biomechanics 6. Benzel EC: Biomechanics of Spine Stabilization. Park Ridge, American • Journal of Back and Musculoskeletal Rehabilitation Association of Neurological Surgeons, 2001. • Journal of Biomechanical Engineering 7. Borelli GA: On the Movement of Animals. Berlin, Springer-Verlag, 1989. 8. Bougery MJ, Bernard C, Jacob NH: Textbook of Human Anatomy [in French]. • Journal of Biomechanics Paris, Guerin, 1832. • Journal of Electromyography and Kinesiology 9. Braun GL: Kinesiology: From Aristotle to the Twentieth Century. Res Quart • Journal of Human Movement Studies 12:163–173, 1941. • Journal of Sport Sciences 10. Braune W, Fischer O: Human gait: Trial on loaded and unloaded humans [in German] Saech Gesellsch Wissensch 21:153–322, 1895. • Mathematical Biosciences 11. Braune W, Fischer O: On the Center of Gravity of the Human Body [in German]. • Medical Engineering and Physics Berlin, Springer-Verlag, 1988. • Medicine and Science in Sports and Exercise 12. Brown T, Hansen RJ, Yorra AJ: Some mechanical tests on the lumbosacral The development of biomechanical laboratories and the spine with particular reference to the intervertebral discs; a preliminary focused study of biomechanics were energized by the organiza- report. J Bone Joint Surg 39A:1135–1164, 1957. 13. Byars EF, Contini R, Roberts VL: Biomechanics Monograph. New York, The tion of biomechanics-oriented conferences, the first of which American Society of Mechanical Engineers, 1967, pp 245. was the First International Seminar on Biomechanics, which 14. Courant R, Hilbert D: Methods of Mathematical Physics. New York, Interscience was organized by the Research Committee of the International Publishers, 1953, vol 1. Council of Sports and Physical Education in 1967 and held in 15. Crawford H: Survivable impact forces on human body constrained by full body harness. Report HSL/2003/09 prepared for health and safety executive, Zurich. Subsequent meetings have been held biannually. 2003. The International Society of Biomechanics was formed in 16. Denis F: The three column injury and its significance in the classification of 1973, the European Society of Biomechanics in 1976, the acute thoracolumbar spinal injuries. Spine 8:817–831, 1983. Canadian Society of Biomechanics in 1973, the American 17. Drewlinger DM: Biomechanics. Emergence of an academic discipline in the United States. Denton, Texas Woman’s University, 1996 (dissertation). Society of Biomechanics in 1977, and the Australia New 18. Evans FG: Studies in human biomechanics. New York Acad Sciences Zealand Society of Biomechanics in 1996. Other biomechanics- 63:586–615, 1955. related organizations include the following: 19. Evans FG: Stress and strain in bones. Their relation to fractures and osteogenesis. • British Association of Sport and Exercise Sciences, United Springfield, Charles C. Thomas, 1957. 20. Evans FG, Lissner HR: Biomechanical studies on the lumbar spine and pelvis. Kingdom J Bone Joint Surg Am 41:278–290, 1959. • Bulgarian Society of Biomechanics, Bulgaria 21. Fick R: Handbook of Joint Anatomy and Mechanics by Taking Muscle Movement • Chinese Society of Sports Biomechanics, China into Consideration [in German]. Jena, Fischer, 19ll. • Comisia de Biomecanica Inginerie si Informatica, Romania 22. Friberg S, Hirsch C: Anatomical and clinical studies on lumbar disc degeneration. Acta Orthop Scand 19:222–242, 1949. • Czech Society of Biomechanics, Czech Republic 23. Fung YC: Biomechanics: Its scope, history, and some problems of continuum • German Society of Biomechanics, Germany in physiology. App Mech Rev 21:1–20, 1968. • Japanese Society of Biomechanics, Japan 24. Fung YC: Biomechanics: Mechanical Properties of Living Tissues. New York, • Korean Society of Sport Biomechanics, Korea Springer-Verlag, 1993, pp 1–13. • Polish Society of Biomechanics, Poland 25. Goel VK, Gilbertson LG: Applications of the finite method to thoracolumbar spinal research—past, present, and future. Spine 20:1719–1727, 1995. • Russian Society of Biomechanics, Russia 26. Goel VK, Weinstein JN: Biomechanics of the Spine: Clinical and Surgical • Societ de Biom‚ Canique, France Perspective. Boca Raton, CRC Press, 1990.

402 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com HISTORY OF SPINE BIOMECHANICS: PART II

27. Gribbin J: Science. A history. London, Penguin books, 2003, pp 1453–2001. 59. Panjabi MM: Three-dimensional mathematical model for the study of the 28. Gurdjian ES, Lissner HR: Mechanism of head injury as studied by the cath- mechanics of the human vertebral column. J Biomech 6:671–680, 1973. ode ray oscilloscope: Preliminary report. J Neurosurg 1:393–399, 1944. 60. Panjabi MM, White AA: Biomechanics in the Musculoskeletal System. London, 29. Gurdjian ES, Webster JE, Lissner HR: Studies on skull fracture, with particu- Churchill Livingstone, 2001. lar reference to engineering factors. Am J Surg 78:736–742, 749–751, 1949. 61. Pauwels F: A Collective Treatise on Functional Anatomy of Locomotor Apparatus 30. Hakim NS, King AI: A three dimensional finite element dynamic response [in German]. Berlin, Springer-Verlag, 1965. analysis of a vertebra with experimental verification. J Biomech 12:277–292, 62. Pope MH: Giovanni Alfonso Borelli—the father of biomechanics. Spine 1979. 30:2350–2355, 2005. 31. Hardy WG, Lissner HR, Webster JE, Gurdjian ES: Repeated loading tests of 63. Provencher MT, Abdu WA: Giovanni Alfonso Borelli: “Father of spinal biome- the lumbar spine; a preliminary report. Surg Forum 9:690–695, 1958. chanics.” Spine 25:131–136, 2000. 32. Henzel JH: The human spinal column and upward ejection acceleration. An 64. Richardson JA: History of biomechanics and kinesiology. www.usd.edu/ appraisal of biodynamic implications. Wright-Patterson Air Force Base, US jarichar/Hist.html. Accessed March 25, 2005. Aerospace Medical Research Laboratory, Report No. AMRL-TR-66–233, 65. Roaf R: A study of the mechanics of spinal injuries. JBJS 42B:810–823, 1960. September 1967. 66. Roux W: Collective Treatise on Developmental Mechanics of the Organism [in 33. Hess JL, Lombard CV: Theoretical investigations of dynamic response of man German]. Leipzig, Wilhelm Engelmann, 1895. to high vertical accelerations. J Aviation Med 29:66, 1958. 67. Ruff S: Brief acceleration: Less than one second, in German Aviation Medicine, 34. Higgins LS: Studies on Vertebral Injuries Sustained during Aircrew Ejection. World War II. Washington, D.C., United States Government Printing Office, Washington, D.C., Office of Naval Research, Contract No. NONR-4675 (00), 1950, vol I, chapter VI-C, pp 584–597. May 1965. 68. Sanan A, Rengachary SS: The history of spinal biomechanics. Neurosurgery 35. Hirsch C: Studies on the mechanism of low-back pain. Acta Orthop 39:657–669, 1996. Scandinavica 20:261–274, 1951. 69. Sylven B, Paulson S, Hirsch C, Snellman O: Biophysical and physiological 36. Hirsch C: The pathogenesis of chondromalacia of the patella. Acta Chir investigations on cartilage and other mesenchymal tissues. II. The ultrastruc- ture of bovine and human nuclei pulposi. J Bone Joint Surg Am 33A:333–340, Scand Suppl 83:1–107, 1944. 1951. 37. Hirsch C: The reaction of intervertebral discs to compression forces. J Bone 70. Thomas HO: Contributions to Medicine and Surgery, Part II, The Principles of Joint Surg Am 37A:1188–1196, 1955. Treatment of Diseased Joints. London, H.K. Lewis, 1883. 38. Hirsch C, Nachemson A: Clinical observations on the spine in ejected pilots. 71. Turner MJ, Clough RW, Martin HC, Topp LJ: Stiffness and deflection analy- Acta Orthop Scand 31:135–145, 1961. sis of complex. structures, J Aero Sci 23:805–824, 1956. 39. Hirsch C, Nachemson A: Clinical observations on the spine in ejected pilots. 72. Virgin WJ: Experimental investigations into the physical properties of the Aeromed Acta 34:629–632, 1963. intervertebral disc. J Bone Joint Surg 33B:607–611, 1951. 40. Hirsch C, Nachemson A: New observations on the mechanical behavior of 73. von Meyer H: The architecture of the trabecula [in German]. Reichert und lumbar discs. Acta Orthop Scandinavica 23:254–283, 1954. Dubois-Reymonds Arch 34:615–628, 1867. 41. Hirsch C, Schajowicz F: Studies on structural changes in the lumbar annulus 74. Ward FO: Outlines of Human Osteology. London, Renshaw, 1838, ed 1. fibrosus. Acta Orthop Scand 22:184–231, 1953. 75. Weber EH: Anatomical and physiological tests on some systems of human 42. Hirt S: What is kinesiology? A historical review. Phys Ther Rev 35:419–426, spine mechanism [in German]. Arch Anat Physiol 1:240–271, 1827. 1955. 76. Weber W, Weber E: Mechanics of the Human Walking Apparatus [in German]. 43. Holdsworth F: Fractures, dislocations, and fracture-dislocations of the spine. Böttingen, Dietrich, 1836. J Bone Joint Surg Am 52:1534–1551, 1970. 77. Werne S: Studies in spontaneous atlas dislocation. Acta Orthop Scand Suppl 44. Huiskes R, Chao EY: A survey of finite element analysis in orthopedic biome- 23:1–150, 1957. chanics: The first decade. J Biomech 16:385–409, 1983. 78. White AA, Panjabi MM: Clinical Biomechanics of the Spine. Philadelphia, 45. Deleted in proof. Lippincott Williams and Wilkins, 1990 46. Knoeller SM, Seifried C: Historical perspective: History of spinal surgery. 79. Widas P: Introduction to finite element analysis. http://www.sv.vt.edu/ Spine 25:2838–2843, 2000. classes/MSE2094_NoteBook/97ClassProj/num/widas/ history.html. 47. Kostuik JP, Errico TJ, Gleason TF, Errico CC: Spinal stabilization of vertebral Accessed April 20, 2005. column tumors. Spine 13:250–256, 1988. 80. Wolff J: The Law of Bone Remodeling [in German]. Berlin, A Hirschwald, 1892. 48. Le Vay D: The history of orthopaedics. Licensed edition, Basel, Roches, 1990, pp 179–121. 49. Lysell E: Motion in the cervical spine. An experimental study on autopsy COMMENTS specimens. Acta Orthop Scand Suppl 123:1, 1969. 50. Maquet P: Iatrophysics to biomechanics. From Borelli (1608–1679) to Pauwels n the second part of this two-part series, the authors continued their (1885–1980). J Joint Bone Surg Br 74:335–339, 1992. Iexhaustive review of spinal mechanics. Beginning with Leonardo da 51. Martin RB: A genealogy of biomechanics. http://asb-biomech.or/history/ Vinci, the contributions of the Renaissance period are reviewed. da biomech/. Accessed March 20, 2005. Vinci not only studied the anatomy of the spine, but also conducted 52. Mow VC, Huiskes R: A brief history of science, and orthopaedic biomechan- important and relevant research on the subject of spine mechanics. The ics, in Mow VC, Huiskes R (eds): Basic Orthopaedic Biomechanics and illustrations depicting this work nicely enhance this article. It was inter- Mechanobiology. Philadelphia, Lippincot Williams and Wilkins, 2005, pp 1–28. esting to read about Andreas Vesalius, who fathered much of what we 53. Nachemson A: Measurement of intradiscal pressure. Acta Orthop Scand now call modern anatomy, but it seems that his understanding of the 28:269–289, 1959. spine was not very advanced, particularly when compared with the 54. Nachemson A: Some mechanical properties of the lumbar intervertebral discs. earlier work of Leonardo da Vinci. I found the section on Giovanni Bull Hosp Joint Dis 23:130–143, 1962. Borelli to be particularly enlightening, particularly the discussion on 55. Nachemson A, Morris J: Lumbar discometry. Lumbar intradiscal pressure the interrelationship to Malphighi. Borelli’s mathematical studies on measurements in vivo. Lancet 25:1140–1142, 1963. 56. Nachemson A: The influence of spinal movements on the lumbar intradiscal the lever arm effect of muscles were particularly brilliant; his book De pressure and on the tensile stresses in the annulus fibrosus. Acta Orthop Motu Animalium is justifiably considered the first book on biomechan- Scand 33:183–207, 1963. ics. The illustrations added to the article are particularly helpful in 57. Nachemson A: The possible importance of the psoas muscle for stabilization explaining his views. The authors continue with reviews of the writings of the lumbar spine. Acta Orthop Scand 39:47–57, 1968. by Hooke, Newton, and the Weber brothers, among others, and nicely 58. Deleted in proof. developed the historical theme of spinal biomechanics. There is a won-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 403 NADERI ET AL.

derful depth of reading material contained in this article. It is an exten- n the current age of science, knowledge, and information, there is a sive piece of writing that puts the history of this most interesting sub- Igreat risk of our routinely using every new design as an important ject, one which all neurosurgeons will enjoy, into print. technological contribution to patient health. In this new age, the average life span has increased, which has brought with it an increase in James T. Goodrich problems of the spine and degeneration. There are so many great dis- Bronx, New York cussions in spine biomechanics. Very few scientists question the necessity of these systems. The real value of spine biomechanics could be he authors extensively review the history of biomechanics in this better understood if evaluated in conjunction with human physiology two-part series. For the most part, the authors have succeeded with T and anatomy. These articles orient us regarding the value of anatomy this ambitious undertaking. The understanding of biomechanics of the in biomechanics by guiding us from the past to the future. Without a spine plays a vital part in the management of spinal disorders. These sound knowledge of the history of anatomy, we cannot understand two articles describing the history of biomechanics provide a good the present or our future. foundation for the overall understanding of spinal biomechanics. To my knowledge, da Vinci was a great artist and performed dissec- Volker K.H. Sonntag tions on more than 30 cadavers. However, he was influenced by the Phoenix, Arizona works of another preeminent anatomist, Marc Antonio della Torre, as well as by Luca Pacioli, who was renowned for his knowledge of math- s in the first part of this series, this article describing the post- ematics and geometry. Just after da Vinci’s contact with Pacioli, he was ARenaissance era summarizes an interesting period of invention and noted as saying, “There can be no certainty unless one can apply one scientific progress. From Leonardo da Vinci to Julius Wolff, several sig- of the mathematical sciences.” nificant scientific thinkers of this period contributed to many areas other than medicine or biomechanics, including astronomy, physics, mathe- Yucel Kanpolat matics, and art. By studying and theorizing on these various fields, some Ankara, Turkey of the historical figures described in this article serendipitously laid down the foundation of the understanding of spine biomechanics as we he authors have provided a comprehensive overview of the history know it today. These scientists include Sir Isaac Newton, the father of cal- Tof spinal biomechanics. They have detailed numerous vignettes culus and the laws of motion, and Thomas Young and Robert Hooke, which allow the reader to recognize clinical problems present today who described how these forces interact with solid materials. Although that also had to be tackled in the past centuries. Furthermore, the these contributions did not primarily involve the spine, without them authors discuss the basic methods available to the healthcare providers our ability to understand the effect of our surgical constructs on spine at that time. The lack of technological advancements of the current biomechanics would be greatly limited. During the past century, spinal treatments obviously made successful management of these progress in spine biomechanics was remarkable. The experience of two afflictions much more difficult. Nonetheless, our predecessors man- World Wars has certainly accelerated our understanding in this field. The aged to effectively care for their patients in the great majority of cases. authors are to be commended for an excellent review of this time period. The authors are to be lauded for tracking down these historic references and bringing them together to tell a cohesive story. Paul Khoueir Michael Y. Wang Robert F. Heary Los Angeles, California Newark, New Jersey

Second and ﬁnal meeting between Joe Louis (standing) and Max Schmeling at Yankee Stadium, June 22, 1936. Louis scored the second quickest knockout in heavyweight title ﬁght history which came at two minutes and four seconds in the opening round.

404 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com COVER COMMENT

THE WAR OF THE GODS

Neurosurgery 60:405–412, 2007 DOI: 10.1227/01.NEU.0000255344.41206.A1

ORIGINS years of inactivity, he quickly fought his way to a title fight against Frazier with a third-round technical knockout (TKO) of he epic trilogy of fights between Joe Frazier and Muham- Jerry Quarry on October 26, 1970 and a 15th-round TKO of mad Ali embodies what was, perhaps, the most famous Oscar Bonavena on December 7, 1970. Frazier had defended his Trivalry in boxing history. Shaped by political and racial title with a knockout of Bob Foster on November 18, 1970 and drama, personal betrayal, and revenge, these bouts nonetheless subsequently signed to fight Ali in the first ever battle of unde- managed to capture the undeniable courage, pride, and deter- feated champions. mination of these two men. Beginning with an intensely hyped Born in Beaufort, South Carolina as the youngest of 12 chil- title fight in New York City, it would end with what may be dren in an economically struggling family, Joe Frazier would boxing’s most brutal match. often walk smaller schoolmates home and protect them from Born to a middle class family in Louisville, Kentucky, bullies in exchange for lunch money or extra food. Like Ali, he Muhammad Ali began boxing at the age of 12 after a local won an Olympic gold medal in the 1964 Tokyo Games, compet- policeman invited him to train at his gym. He became a suc- ing in the heavyweight category. He began his professional cessful amateur boxer while still in high school and won the career with a first-round knockout of Woody Goss and remained Olympic gold medal in boxing as a light heavyweight at the undefeated until his claim of the heavyweight title in 1968 (3, 8). age of 18. He won his first professional fight in 1960 and went Frazier had sympathized with Ali’s situation and had lob- on to defeat a succession of high-profile fighters. After defeat- bied for his return to boxing, even lending him money when Ali ing Sonny Liston in a six-round bout, Ali claimed his first was unable to fight. Upon Ali’s return and with the heavy- heavyweight title in 1965, shouting, “I am the greatest” to the weight title on the line, favors were forgotten. Motivated by crowd when Liston stayed in his corner with a shoulder injury either a desire to promote the fight or an attempt to use psycho- as the bell rang for the seventh round (17). logical warfare to lower Frazier’s morale, Ali’s prefight taunting In 1967, the World Boxing Association stripped Ali of his of Frazier became personal. Ali appeared at Frazier’s training title and boxing license after he refused to be drafted into the camp, sneering that he was too ugly to be champion, saying military. That same year, he was charged and convicted of vio- “Joe Frazier is so ugly. His mother told me that, when Joe was lating the Selective Service Act and was sentenced to 5 years in a little boy, every time he cried, the tears would stop, turn prison (9). Having joined the Nation of Islam in 1964 (9), Ali around, and go down the back of his head” (9). Ali goaded him cited religious prohibitions (17); he was also an outspoken as “Uncle Tom” and “The White Man’s Champion” (Fig. 1). objector to the Vietnam War and famously stated in 1966, “I Frazier, who even received death threats before the fight, felt ain’t got no quarrel with those Viet Cong” (9). Over the next 3.5 betrayed by the insults. As told by George Foreman, years, Ali appealed the conviction and sued for the reinstate- Joe told me why he had that hate for Ali. Muhammad ment of his boxing license. He partially supported himself by was calling him an Uncle Tom. Kids would go to school speaking at opposition rallies on college campuses. In 1970, and taunt his children, and they’d come home and his the New York State Supreme Court ruled that Ali had been wife would hear about it. What bothered Joe was that unjustly denied a boxing license; a Supreme Court decision every morning he’d get up really early, when it was reversed his conviction late the following year (9). dark, to get the roadwork in. He always wore this big Joe Frazier had ascended to the heavyweight championship hood over his head when he ran. And he said, “Man, I after Muhammad Ali’s forced exit from boxing. With Ali don’t want my wife thinking I’m peeping into people’s stripped of his title, the heavyweight division was thrown into windows.” The point is, at the time, Joe didn’t get what a state of disarray. In a move that would presage the fractured an Uncle Tom was. He hated Muhammad because he alphabet soup of contemporary boxing championships, the thought Ali was calling him a Peepin’ Tom. If someone World Heavyweight Championship was split into two parts. would have explained to Joe what an Uncle Tom was, The New York State Athletic Commission recognized Frazier as he might not have ever hated Ali. heavyweight champion with his defeat of Buster Mathis in 1968. After knocking out World Boxing Association champion ALI–FRAZIER I: THE FIGHT OF THE CENTURY Jimmy Ellis in 1970, Frazier’s claim to the championship title was universally recognized. The heavyweight title fight between Ali and Frazier took With his boxing license restored, Ali returned to the ring, pro- place at Madison Square Garden with a purse of $2.5 million claiming himself the “People’s Champion” (7). Following 3 each (13). Ali was 29 years old with a 31-0 record and 25 knock-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 405 BAIRD AND LEVY

Sammy Davis, Jr., and Hugh Hefner, as well as former champions Gene Tunney and Jack Dempsey. Woody Allen, Burt Lancaster, and Norman Mailer were also in attendance (13). The fight lasted all 15 rounds, with Ali prevailing in the first three. Frazier began to dominate in the fourth round, pinning Ali against the ropes while landing powerful body blows. Lying on the ropes, Ali called out to the spectators “noooo contest” (13). In the 11th round, Frazier landed his famous left hook and sent Ali stumbling backwards across the ring; in the 15th round, he landed another that lifted Ali off his feet and made him land flat on his back, only the third time Ali had been knocked down in his career. Ali’s doctor, Ferdie Pacheco, said, “[Ali] was up almost as fast as he went down, it was incredible. Not only could he take a punch; that night he was the most courageous fighter I’d ever seen. He was going to get up if he was dead. If Frazier had killed him, he’d have gotten up” (13). After 15 rounds, Frazier won the fight unanimously (Figs. 3–5) (17). It was Ali’s first defeat, but he accepted it gracefully, saying, FIGURE 1. Before the first fight: Muhammad Ali taunting Joe Frazier at “Just lost a fight, that’s all…The world goes on. You’ll all be Frazier’s gym. writing about something else soon. I had my day. You lose, you don’t shoot yourself” (13). He moved on to defeat a string outs. Frazier, at the pinnacle of his career, was 27 years old and of top heavyweights until suffering a second defeat against had a record of 26-0 with 23 knockouts. Intense media coverage Ken Norton in 1973. Norton lost a rematch 6 months later in a precluded the night. Promoted as “The Fight,” this bout has controversial split decision, giving Ali the chance to challenge come to be called “The Fight of the Century.” The title reflects Frazier again. the political tensions surrounding the fight as well as the public’s excitement and anticipation for the event. America had just ALI–FRAZIER II emerged from the turbulent 1960s and the fight represented the Ali and Frazier’s second meeting in the ring took place on current rift in the country. Ali had come to symbolize antiestab- January 28, 1974 at Madison Square Garden. Although consid- lishment and, to many, embodied the defiant spirit of the anti- ered the least exciting of the Frazier-Ali Trilogy, and certainly not war movement. Others held him in contempt as a brash, draft- nearly as dramatic as the first and third matches, the second dodging, antiwar, radical Muslim. Frazier, who read the Bible fight was still well publicized and emotions were tense. Frazier and was known to enjoy singing, became the unwitting symbol had lost his heavyweight title in 1973 in Kingston, Jamaica after of the conservative pro-war sentiment (2, 3). an upsetting two-round match against George Foreman. With no Madison Square Garden was sold out a month before the title at stake, the purse was a mere $850,000 each. Ali, however, fight, with ringside seats selling for a record $150. On March 8, wanted to avenge his loss from their first match, and the winner 1971, more than 20,000 would have a chance to challenge Foreman for the title. Before spectators filled the stands and 300 million more people watched around the world. Selected from more than 1200 applications, more than 700 working press credentials were issued. Frank Sinatra was ringside taking photographs for Life magazine (Fig. 2). Dustin Hoffman and Diana Ross were thrown out of the press section before the start of the fight. Also present were such celebrities as Barbra FIGURE 2. Frank Sinatra’s cover photo- Streisand, Bill Cosby, graph for Life magazine. FIGURE 3. Ali–Frazier I: Frazier’s devastating left hook.

406 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com THE WAR OF THE GODS

Frazier down. Ali continued to dictate the pace of the fight until the seventh round when Frazier began to take control, landing more aggressive punches, including an overhand right that forced Ali into the ropes during the eighth round. Ali seemed to dominate the ninth round, but Frazier regained control during the tenth. During the last two rounds, Ali kept his distance in what Ferdie Pacheco termed a “fighting retreat” with Frazier in pursuit (2). Although a moderate amount of controversy followed the outcome, the judges scored the fight unanimously in favor of Ali With Frazier out of the way, Ali reclaimed the heavyweight title after a bloody knockout over George Foreman in Kinshasa, Zaire on October 30, 1974. The fight was Don King’s debut as a professional boxing promoter, and he managed to have both fighters sign contracts with him with the promise of a $5 million purse. Many have noted that Muhammad Ali’s most significant fights took place under the auspices of two of FIGURE 4. Ali–Frazier I: Ali’s left jab. the decade’s most notorious dictators: Mobutu Sese Seko of Zaire and Ferdinand E. Marcos of the Philippines.

ONLY IN AMERICA

The “Thrilla in Manila” put Don King at boxing’s pinnacle. A gifted promoter with an acute business sense, King organized some of the largest purses in the history of the sport, max- imized exposure and compensation for the athletes, and aggressively promoted boxing in general. His impact on boxing, however, has not been without controversy. King, raised in Cleveland, Ohio, had once considered becoming a lawyer. Later, to pay his tuition at Case Western Reserve University, he became a numbers runner, an occupation that became so lucrative that he left Case Western after only 1 year to pursue it on a full time basis. In 1972, King tried his hand in the boxing business, persuading Muhammad Ali to compete in a benefit exhibition to raise money for a Cleveland hospital. FIGURE 5. Ali–Frazier I: Frazier’s left hook sends Ali to the canvas for only The success of the exhibition launched King’s career in boxing the third time in his career to that date. promotion. In an interview with Benny Henderson, Jr., King stated, “I migrated to boxing promotion because Ali thought the fight, Ali and Frazier were invited to the studios of the that I’d be good at it and after thinking about it, I agreed” (12). American Broadcasting Company to review their first match The 1974 Ali-Foreman fight was Don King’s first major boxing with Howard Cosell. After the first fight, Frazier had been hos- promotion. He promised each boxer $5 million; however, for a pitalized with kidney damage for 3 weeks, whereas Ali had black promoter in the existing landscape of white managers, been observed overnight and released with a broken jaw. When owners, and promoters, financial backing became problematic. Frazier referred to Ali’s hospital stay during the interview, Ali King ultimately obtained backing from Mobutu Sese Seko, who began taunting Frazier, saying “Everybody knows I went to the agreed to sponsor the fight in order to generate positive public- hospital for 10 minutes. You were there for 3 weeks. You igno- ity for both himself and Zaire. rant Joe.” Cosell was stunned when Frazier pulled out his ear The controversy surrounding King’s influence in the boxing piece and proceeded to wrestle Ali to the floor. Both fighters world arose from his many legal problems as well as the public were fined $5000 for the brawl after audience members had to perception of his tactics. Implicated in various criminal investiga- separate them. tions, King was indicted on a number of occasions. One investi- Light on his feet and circling away from Frazier’s left hook, gation by the United States Federal Bureau of Investigation in Ali dominated the beginning of the second match. Ali was 1999 centered around alleged payoffs to the president of the landing more punches, including a jarring right cross that put International Boxing Federation to advance the rankings for Frazier against the ropes. Referee Tony Perez mistakenly King’s boxers. He even had a colorful past before his success thought he heard the bell, giving Frazier a chance to recover with boxing promotion. King had been cleared of killing a man and preventing a follow-up from Ali that might have sent trying to rob one of his gambling houses in 1954 when the shoot-

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 407 BAIRD AND LEVY

Not everyone’s goals or dreams will be the same, but they all require a commitment (12). Not all of King’s fighters have been impressed with his commitment. He has been known to arrive at a fight with one boxer and leave with the other, and numerous boxers have accused him of defrauding them of money. Ali himself sued King after his 1980 fight against Larry Holmes, saying he was owed $1.2 million. Holmes’ own outlook reflects King’s undeniable influence within the world of boxing, stating, “I make more money with Don King stealing from me than 100 percent from other promoters” (14). Despite these sentiments, King has commented, I stood up for [my boxers]. I cared about them, not only as a talent that could make us some money, but as people. I like to think I treated them like human beings and not just pieces of meat. Most of the boxers who are with me, have been with me for a long time. FIGURE 6. Ali–Frazier III: Past their prime? I’ve gotten them opportunities and in most cases made them world champions. I don’t believe in kick- ing someone to the curb because they lose a fight or ing was ruled a justifiable homicide. Thirteen years later, he was two. I like to try and dust them back off and get them convicted of second-degree murder after beating a man who another opportunity. Most of those who have left owed him $600 to death. After the charge was reduced to invariably come back. That makes me feel good (12). manslaughter, King served just less than 4 years in prison before Don King’s ostentatious persona extends from his infamous his release in 1971. King has estimated spending nearly $30 mil- hair to his penchant for over-the-top statements: lion defending himself in court (18). When I wake up, I thank God, then I try to organ- The aggressive business strategies used by King to control the ize my thoughts for the day because I don’t sleep too top heavyweight fighters have also drawn a great deal of criti- much. I used to call my guys early in the morning cism. King has contractually obligated contending fighters to be and say “a sleeper can’t get nothing but dreams!” I promoted by King in the future should they defeat one of his own don’t have time for reflection. Maybe someday I will, boxers. “He has the most brilliant business mind I have ever but there are too many things I want to do. I’m an encountered,” Seth Abraham, former president of HBO sports, advocate and I stand up for what I believe in. I’ll fight has said. “Don King is formidable in his sleep” (18). for what I believe in. I’d rather die standing up than King has put together some of the most electrifying boxing live on my knees (12). events in the history of the sport, including the 1974 classic His patriotism is interminable: “I not only preach patriot- “Rumble in the Jungle” between Muhammad Ali and then ism—I believe it—God bless America—the greatest nation in unbeaten George Foreman, an event watched by one billion the world.” Whether his presence has been a positive or neg- people on television. He also brought to the boxing fans ative influence on professional boxing, there is no doubt Don “Thrilla in Manila” (Ali–Frazier), “The Last Hurrah” King’s impact has been profound. Ali–Holmes, “He’s Back” (Tyson–McNeeley), “Finally” (Tyson– Holyfield I), “The Sounds and the Fury” (Tyson–Holyfield II), THE “THRILLA IN MANILA” and many other unforgettable bouts (12). Of his achievements King states, Frazier and Ali met for the third time on October 1, 1975 at the The secret to my success is very simple—hard work Araheta Coliseum in Quezon City, Philippines just outside and dedication. I’ve been No. 1 ever since I first Manila. The now infamous “Thrilla in Manila” is considered by started promoting over 30 years ago. That’s a long many to be the greatest fight in boxing history. The 14 devastat- time to stay on top. It wasn’t easy. I couldn’t walk in ing rounds have been vividly described in the epic poem, War of the front door of the corporate office; I had to break it the Gods, by James Tokley, Poet Laureate of Tampa, Florida (20). down. Your desire to remain No. 1 must always be The event was another fantastic Don King promotion. King driving you. You have to be creative, innovative and had convinced President Ferdinand Marcos and his wife have the guts to stand up. A man who has no wings— Imelda, who treated the country’s money as their own, to spon- he cannot fly—he shall remain on the earth. I want to sor the fight. Ali was promised a purse of $4 million plus 43% let everyone know how much harder it was and still of the gross takings (he received $6 million) and Frazier was to is for me. We have to make sure that we’re a step receive $3 million (Fig. 6). ahead of everyone and work 24-7 to accomplish that. Both men were past their prime and near the end of their For anyone to live their lives that way—through ded- career, each with one defeat over the other and a desire to set- ication and hard work—they can realize their dreams. tle the rivalry once and for all (Fig. 7). The animosity between

408 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com THE WAR OF THE GODS

Ali and Frazier reached its peak in the prefight build up. Once LEGACY: EDDIE FUTCH more, Ali escalated the hype with his derisive taunts, rhyming: Eddie Futch was one of the greatest trainers in boxing and It will be a Killer one of the sport’s true gentlemen. Known for his integrity as And a Chiller well as his ability to turn talent into mastery, Futch’s name has And a Thrilla become inseparable from the sport of boxing. He not only When I get the gorilla in Manila trained boxers to fight, he also taught them to train. Admired Ali went even further by interrupting Frazier to correct his and respected by colleagues from both corners of the ring, he grammar during interviews and carrying around a toy gorilla has received immeasurable recognition for his work, including that he would pull out to punch at press conferences. Frazier Manager of the Year (1975), Long and Meritorious Service was again furious and hurt by Ali’s taunts. He said, “Look at (1982), awards from the Boxing Writers Association of America my beautiful kids. How can I be a gorilla? I don’t want to knock as well as Trainer of the Year by the International Boxing Hall him out, I want to hurt him. If I knock him down I’ll stand of Fame (1991 and 1992). back, give him a chance to breathe. It’s his heart I want” (3). The son of a Mississippi sharecropper, Eddie Futch was born The stadium was crammed with 28,000 people in sweltering in Hillsboro, Mississippi in 1911 and retired from boxing at the heat reported to be between 95 and 110 degrees. President age of 87 in 1998. He died in 2001, 2 months after celebrating Marcos and First Lady Imelda were in attendance. Ali started his 90th birthday. Futch won the Detroit Athletic Asociation well, dominating the first rounds. The intensity of the fight Lightweight championship in 1932 and the Detroit Golden escalated and, with both fighters moving a little more slowly Gloves the following year. At the Brewster Recreation Center than they had in their prime, plenty of hits were landing (14). gym, Futch became friends with the local light heavyweight In the sixth round, Frazier sent Ali’s head spinning with a left sensation, Joe Louis. The future all-time great often asked Futch hook that Angelo Dundee rated as the hardest shot he had ever to spar with him, saying, “If I can hit you, I know I’m sharp” seen thrown. Ali is reported to have taken the hit, saying, (3). Despite a heart murmur ending any hopes for a profes- “They told me Joe Frazier was washed up,” to which Frazier sional career, Futch felt that boxing had enabled him to create responded, “They lied, they lied” (10). Frazier began to domi- a life for himself outside of the ghetto. “It has been my passion nate the fight with ferocious body blows that wore down his for the last six decades to help other young men make some- opponent, with only the ropes keeping Ali on his feet by the thing of themselves.” His decision to stop the Ali–Frazier fight end of the ninth round. Frazier returned to his corner after the in Manila generated considerable controvery, but his dedication bell, saying, “Damn, what’s keeping that [expletive] fool up?” to his fighters is reflected in his own explanation of his deci- Frazier’s pounding was relentless—Frazier would later say he sion. “[Joe’s] a good father and I want him to see his kids grow hit Ali with shots that would bring cities down. The 10th round up” (3). was almost Ali’s last, but one of Ali’s trainers, Bundini Brown, Futch worked with 22 world champions, including heavy- implored him to get up one more time (10, 14). weights Joe Frazier, Larry Holmes, Riddick Bowe, Ken Norton, The heat was suffocating, and the pace and brutality of the Michael Spinks, and Trevor Berbick; light heavyweights Bob fight were unparalleled for a heavyweight match. Ali’s second Foster and Montell Griffin; junior middleweight Mike wind hit during the 12th round, when he began relentlessly McCallum; lightweights Alexis Arguello, Virgil Hill, and Wayne attacking Frazier’s face, successfully swelling his eyes nearly McCullough; and welterweights Don Jordan, who became his shut. The 13th round was a disaster for Frazier, with Ali’s right first champion in 1959, and Marlon Starling. Riddick Bowe, who cross sending Frazier’s mouthpiece flying into the stands. Eddie became the heavyweight champion in an upset over Evander Futch, Frazier’s trainer, only allowed him return to the ring Holyfield in 1992, was introduced to South African President because he thought Ali might have exhausted himself trying to Nelson Mandela during a world tour. The first question finish Frazier. Ali spent the 14th round throwing approximately Mandela asked him was, “Where’s Eddie Futch?” (3). 30 punches into Frazier’s left eye. Frazier had long had poor Futch endeavored to bring the best out of every boxer he vision in his left eye, a fact he had been able to hide for most of worked with. Understanding that ego could be an impediment his career, but it was now completely gone and Ali was landing to greatness, he had a no nonsense approach when it came to right hand punches at will (3). When the round ended, the referee handling self-importance and bad behavior in his fighters. had to guide Frazier back to his corner because he could not see Early in his career, he threw Sugar Ray Robinson out of his it himself. Ali had collapsed back in his own corner, asking his Detroit gym for causing trouble; he also once told Marlon trainers to cut his gloves off. Whether he would have come back Starling, “Marlon, I’ve taught you all you know, but I haven’t out or not will never be known (10). Eddie Futch knew Frazier’s taught you all I know” (3). He would not hesitate to walk out vision was too impaired to avoid Ali’s punches and informed the of a training session if he felt his fighters were not giving 100% referee the fight was over. Frazier argued, saying, “But I want or were not responding to his teaching. him, boss.” Eddie Futch, in what has since come to be considered Futch may be best known for developing the strategy that one of sporting’s greatest acts of compassion, stopped Frazier brought Muhammad Ali his first two defeats. Training both from going back into the ring, saying, “Sit down, son, it’s over. Frazier and Norton for their victories over Ali, Futch used tac- No one will ever forget what you did here today” (4). tics that seemed to avoid Ali’s strengths while exploiting his

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 409 BAIRD AND LEVY

fights on the weekends. “We would go to Boston, Hyde Park, or Newton to get more experience, training in different gyms. When I was 6 or 7 years old, my dad had me getting up and doing road work” (6). His father had Freddie fight his older brother, Pepper, in the backyard. Whoever won was the favorite that week. The more Freddy warmed up to boxing, the easier life was. When Freddy eventually began to enjoy boxing, he became the best boxer in the Roach household. After the 1976 Boxing Nationals in Las Vegas, he and his father moved there, and he began a professional career. Eddie Futch became his trainer, promising Freddie’s father that he would take care of his son. Because Freddie didn’t have much money, Eddie did everything from training to managing Freddie’s career. “I had a job and worked my whole career. Eddie would get me fights, and he got me plenty because I was a white kid that fought hard and bled” (6). Freddie, at 122 pounds, fought 12 times that year and worked as a bus boy and dishwasher at the Golden Nugget. Soon, promoter Bob Arum signed him to a professional boxing contract and put him on one of the first shows ever aired by ESPN. Freddie fought with the likes of Bobby Chacon and FIGURE 7. Thrilla’ in Manila: A Don King production Hector Camacho. Boxing analyst Al Bernstein has said that Roach was one of the toughest fighters he ever worked with. weaknesses. Helping Frazier develop his bob-and-weave style, He was ranked as high as eighth in the world before a broken Futch advised him to keep Ali on the ropes and maximize the hand destroyed his punching power. He retired at the age of 27. body punches. He also directed Frazier to coax Ali into throw- Roach became a trainer one afternoon when he noted that ing uppercuts so that he could counter with his left hook. The former light heavyweight and cruiserweight champion Virgil strategy succeeded during their first fight. Frazier landed a crit- Hill was sparring with no one in his corner. Describing the ical left hook in the 11th round and knocked Ali down with moment, Freddie said, another in the 15th. Virgil Hill was getting a guy named James Shuler In a 1996 interview, Futch deplored the state of contemporary ready to fight Thomas Hearns. I was watching them heavyweight boxing, in which boxers are paid big money to spar and I noticed that, in between rounds nobody not fight each other. He stated, “It’s the worst thing that could was giving Virgil any water. I ended up going over happen in boxing . . . It reminds me of how the division used to there and helping his corner out, and I gave the guy be. [Joe] Louis wanted to keep improving on his work all of the some advice. Virgil ended up doing very well, so he time. He had 25 successful title defenses, when he wasn’t win- asked me to be his trainer. Eight months later, Virgil ning title fights—he boxed exhibitions. Many exhibitions, just fought for a world championship, and he won. That to keep busy, just to stay in the game and stay in the ring, just was my first world champion (6). to keep doing the things he could do. You don’t see that any- Roach began working alongside Eddie Futch and eventually more. Back in those days, we wanted to stay within reach of began training fighters on his own. Roach went on to train and fighting condition at all times...so if there was any break in produce 17 World Champions. He has also been honored as the card we could jump in . . .” (15). 2003 Trainer of the Year by the Boxing Writers of America and has been inducted into the World Boxing Hall of Fame, the FREDDIE ROACH New England Boxing Hall of Fame, and, most recently, the “We didn’t have a swingset in my backyard growing up, but California Boxing Hall of Fame. we had a boxing ring.” — Freddie Roach Freddie Roach, one of seven children in the Roach family, was AFTERMATH born in 1960 in Dedham, Massachusetts. His father, an arborist and a fighter, gave Freddy his first pair of gloves when he was 5 The “Thrilla in Manila” left both Ali and Frazier with severe years old. Rather than a swingset in the backyard, he had a box- injuries. However, despite scoring a TKO over Frazier and retain- ing ring. His mother was a boxing judge for 9 years (19). ing the heavyweight title, Ali probably experienced heavier dam- When he got home from school, Freddie would finish his age. After the fight, he said, “It was like death. The closest thing homework, run his paper route, and then train with his brother to dying I know.” Many have speculated that this single fight Monday through Thursday. Fridays were spent traveling to largely impacted the subsequent deterioration of Ali’s health.

410 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com THE WAR OF THE GODS

Ali stayed in the ring for another 10 fights but was not the ogize to me—he apologized to the paper. I’m still waiting for boxer he once was. After defeating Jimmy Young in 1976 by a him to say it to me.” Ali responded by saying, “if you see controversial decision, Ferdie Pacheco, his longtime boxing Frazier, you tell him he’s still a gorilla.” physician, said, “Ali was missing, he had no timing…He was Whatever animosity remains between the fighters, they will getting tired a lot sooner than usual. His reflexes were only 25 be forever linked to each other in sporting history. Neither to 30 percent of what they should be” (1). Pacheco was banned would have reached the place they currently hold in boxing from Ali’s entourage after advising him to retire. In 1979, Ali legend without the other, a sentiment echoed by Ali when, in a announced his retirement. He emerged for two comeback fights rare moment of humility, he said “I couldn’t have done what I against Larry Holmes and Trevor Berbick, both of which he did without him, and he couldn’t have done what he did with- lost, before permanently retiring in 1981. out me.” Ali’s trainer, Angelo Dundee, said of their fighting, Ali’s health was suspect during his last fights and has contin- “Their styles were just meant for each other. It is remarkable. No ued to deteriorate during his retirement (1). He was eventually matter where or when they fought, if you put them together diagnosed with Parkinson’s syndrome, thought by most to be you couldn’t have a bad fight. I think they both brought each pugilistic Parkinson’s syndrome caused by multiple concussive other to a higher level. They brought the best out of each other.” injuries to his brain. He has remained an active public figure and, “You get so tired. It takes so much out of you men- notably, lit the Olympic torch during the 1996 games in Atlanta. tally. It changes you. It makes you a little insane. I was Frazier retired after being knocked out by George Foreman thinkin’ what am I doing here with this beast of a in 1976. He attempted a comeback in 1981 that ended in a draw man? I must be crazy. Joe Frazier is the greatest fighter with Jumbo Cummings after 10 rounds. He now runs his own of all time next to me. That’s one hell of a man.” gym in Philadelphia, where he has helped to train two of his — Muhammad Ali own children (Marvis Frazier and Jackie Frazier-Lyde) in their boxing careers. “I hit the punches that would have knocked The rivalry has continued outside of the ring for more than cities down. He took ‘em and came back. He’s a 30 years, and Frazier remains incredibly bitter. He told Ali’s great champion.” biographer, Tom Hauser, — Joe Frazier Lissa C. Baird I hated Ali. God might not like me talking that way, Michael L. Levy but it’s in my heart...I still want to take him apart San Diego, California piece by piece and send him back to Jesus. He shook me in Manila. He won. But I sent him home worse than he came. Look at him now. He’s damaged goods. REFERENCES I know it; you know it. Everyone knows it; they just 1. Anderson D: For Ali, What Price the Thrilla in Manila? New York Times, don’t want to say it. God has shut him down. He can’t September 23, 1984. talk no more because he was saying the wrong things. 2. Axelrod M: Ali-Frazier II—Was the Decision Fair and Accurate? http:// He was always making fun of me. I’m the dummy. www.eastsideboxing.com/news/php?p=1813&more=1. Accessed January 2007. I’m the one getting hit in the head. Tell me now. Him 3. BBC Sports: Futch—Trainer Supreme. News.bbc.co.uk/sport1/hi/box- or me, which one talks worse now? He can’t talk no ing/593061.stm. Accessed January 2007. more and he still tries to make noise. I don’t care how 4. Callahan T: Fight One More Round. Time December 14, 1981. the world looks at him. I see him different, and I know 5. Conrad M: Sportslaw History: Don King. him better than anyone. Manila don’t matter no more. 6. Cooney B: More Than a Boxing Trainer: The Life and Times of Freddie Roach. Interview by Brad Cooney—December 17, 2006. http://www.ringsidereport. He’s finished and I’m still here (9). com/rsr/print.php?type=N&item_id=759. Accessed January 2007. When asked about Frazier’s bitter feelings, Eddie Futch 7. Frazier J, Berger P: Smokin’ Joe: The Autobiography of the Champ. New York, remarked, Macmillan, 1996. I’m so sorry that he hasn’t learned not to go public 8. Fussman C: What I’ve Learned: George Foreman. Esquire January 2004, p.88. with that attitude. He always resented Ali for the 9. Hauser T: The Lost Legacy of Muhammad Ali. Wilmington, Sports Media, 2005, pp 16, 17, 40, 93, 95, 182. things Ali used to say when he was making publicity 10. Hauser T: The Unforgiven. Guardian Unlimited. September 4, 2005. for their fights. He’s being bitter now... He’s been 11. Hauser T, Ali M: Muhammad Ali: His Life and Times. New York, Touchstone, bitter for 18 years. What does he think he’s gonna get 1991. out of being bitter? ...I fought it. I fought it all the 12. Henderson B: Only in America: Don King speaks to Dog House Boxing. Dog time when I was with him, and I see him often now House Boxing/DogHouseBoxing.com, June 28, 2004. 13. Kindred D: The first Ali-Frazier fight was also the best. The Sporting News and if it comes up, I always tell him I don’t want to February 17, 1999. talk about it (15). 14. Kram M: Ghosts of Manila: The Fateful Blood Feud Between Muhammad Ali and In 2001, Ali apologized for his insults, telling the New York Joe Frazier. New York, HarperCollins, 2002. Times, “I said a lot of things in the heat of the moment that I 15. Lamele D: You Could Learn A Lot From This Smurf—Eddie Futch Interview. The Cyber Boxing Zone, 1996. www.cyberboxingzone.com/boxing/futch/ shouldn’t have said. Called him names I shouldn’t have called htm. Accessed January 2007. him. I apologize for that. I’m sorry.” Joe Frazier initially 16. Mallozzi V: Fire Still Burns Inside Smokin’ Joe Frazier. New York Times, accepted the apology, but later told TV Guide, “He didn’t apol- October 18, 2006.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 411 BAIRD AND LEVY

17. Myers WD: The Greatest: Muhammad Ali. New York, Scholastic, 2001, pp 29, 20. Tokley J: The War of the Gods. http://www.prx.org/pieces/3434. Accessed 60. January 2007. 18. Puma M: Only in America. Espn.go.com/classic/biography/s/King-Don. html. www.sportslawnews.com/archive/history/DonKinghistory.htm, Accessed January 2007. Acknowledgments 19. Sandomir R: No Floating, No Stinging: Ali Extends Hand to Frazier. New We thank Freddie Roach, James Toney, Don King, Joe Frazier, George Foreman, York Times, March 15, 2001. and Benny Henderson, Jr. for their signiﬁcant contributions to this article.

The ﬁrst of three historic heavyweight boxing matches between Muhammad Ali (left) and Joe Frazier (Madison Square Garden, New York, March 8, 1971). Pictured is the famed left hook by Frazier that felled Ali in the 15th round. In “Fight of the Century,” the promotional name given to this ﬁrst meeting, Frazier was declared the winner by unanimous decision.

412 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com SUBMIT LISTINGS TO Joel D. MacDonald, M.D. University of Utah CALENDAR OF EVENTS Department of Neurosurgery 30 North 1900 East, 3B-409 SOM Salt Lake City, UT 84132-2303 Full contact information is available online only TEL: 801/581-6908 Listings for this section need to be submitted at least 3 months in advance. Please include a telephone number and/or a fax number. FAX: 801/581-4138 EMAIL: [email protected]

INTERNATIONAL / GENERAL NEUROSURGERY

APRIL 14–19, 2007 American Association of Neurological Washington, DC CONTACT: Patty Anderson, Director of Meetings Surgeons 2007 Annual Meeting TEL: 847/378-0500 EMAIL: [email protected] WEB SITE: www.AANS.org JUNE 21–23, 2007 12th EMN Annual Meeting–Euroacademia Fiuggi, Italy CONTACT: Cristina Tartaglia Multidisciplinaria Neurotraumatologica TEL: +39 051 765357 FAX: +39 051 765195 EMAIL: [email protected] WEB SITE: www.csrcongressi.com JUNE 28–30, 2007 3rd International Meeting Updates Arezzo, Italy CONTACT: Cristina Tartaglia in Neuro-oncology TEL: +39 051 765357 FAX: +39 051 765195 EMAIL: [email protected] WEB SITE: www.csrcongressi.com NOVEMBER 18–22, 2007 12th AACNS/13th WFNS Interim Meeting Nagoya, Japan CONTACT: Tetsuo Kanno EMAIL: [email protected] WEB SITE: www.aacns07.umin.ne.jp DECEMBER 18–22, 2007 12th AACNS/13th WFNS Interim Meeting Nagoya, Japan CONTACT: Tetsuo Kanno EMAIL: [email protected] WEB SITE: www.aacns07.umin.ne.jp NATIONAL AND REGIONAL MEETINGS

FEBRUARY 2–7, 2007 Richard Lende Winter Neurosurgery Snowbird, UT CONTACT: Lanette Dunbar Conference TEL: 801/581-6908 FAX: 801/581-4385 EMAIL: [email protected] WEB SITE: www.lendemeeting.com FEBRUARY 23–25, 2007 European Association of Neurosurgical Antaly, Turkey TEL: +90 312 440 56 00 Societies Winter Meeting FAX: +90 312 440 55 97 EMAIL: [email protected] WEB SITE: www.eanswinter2007.org FEBRUARY 24–28, 2007 20th Annual Neurosurgery in the Rockies Vail, CO CONTACT: Laura Hitchcock TEL: 303-332-4127 FAX: 303-315-1331 EMAIL: [email protected] WEB SITE: www.uchsc.edu/neurosurgery MARCH 2, 2007 Interurban Neurosurgical Society Chicago, IL CONTACT: Marlene Rakowski TEL: 715/542-3201 EMAIL: [email protected] MARCH 9–11, 2007 7th South Cone Society of Neurological Gramado, Brazil CONTACT: Apio Antunes, M.D., Ph.D. Surgeons Meeting TEL: 55/51 3311 8969 FAX: 55/51 3311 8969 WEB SITE: www.plenariumcongressos.com.br/ congressos/neurocirurgiadoconesul/ MARCH 14–17, 2007 Southern Neurosurgical Society, 58th Sea Island, GA CONTACT: Anil Nanda, M.D. Annual Meeting: Finesse, Finances, and TEL: 318/675-6404 Floods WEB SITE: www.southernneurosurgery.org JUNE 16–20, 2007 Rocky Mountain Neurosurgical Society Jackson, WY CONTACT: Joel D. MacDonald, M.D. TEL: 801/581-6908 FAX: 801/581-4385 Email: [email protected] Web Site: www.rmns.org JUNE 26–29, 2007 Society of University Neurosurgeons London, England CONTACT: Neil Kitchen, M.D., F.R.C.S. (SUN) Annual Meeting TEL: 020/7837 3611, x3152 FAX: 020/7676 2045 EMAIL: [email protected] WEB SITE: www.uclh.nhs.uk

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | E413 CALENDAR OF EVENTS

JULY 27–28, 2007 Pennsylvania Neurosurgical Society Hershey, PA CONTACT: Jessica Judy, Meeting Manager Annual Scientific Meeting TEL: 717/558-7850, x1463 FAX: 717/558-7841 EMAIL: [email protected] SPINE / PERIPHERAL NERVE FEBRUARY 1–4, 2007 AO ASIF Advanced Concepts in the Sun Valley, ID Contact: Spine Course Registrar Management of Spinal Disorders Tel: 800/619-1391, x 7574 Fax: 610/695-2420 Email: [email protected] Web Site: www.aona.org FEBRUARY 3–4, 2007 ONE Spine: Controversies in Spine Surgery Cancun, Mexico CONTACT: Val Broyles Tel: 630/681-1040 FAX: 630/682-5811 EMAIL: [email protected] WEB SITE: www.one-spine.org FEBRUARY 18–21, 2007 Current Concepts in Emerging Vail, CO CONTACT: Pfiedler Enterprises, Inc. Technologies in Spine Web Site: www.pfiedlerenterprises.com/ courses.htm FEBRUARY 23–24, 2007 Controversies in Contemporary Phoenix, AZ CONTACT: Intellyst Medical Communications Spine Surgery TEL: 720/748-8800, x225 FEBRUARY 25– The Winter Clinics for Cranial & Spinal Snowmass Village, CO TEL: 513/569-5354 MARCH 2, 2007 Surgery FAX: 513/569-5365 EMAIL: trosenberger@mayfieldclinic.com MARCH 1–3, 2007 6th Annual Symposium on Current Las Vegas, NV CONTACT: Barbara Stokes, CME Coordinator Concepts in Spinal Disorders TEL: 310/423-2935 FAX: 310/423-0309 EMAIL: [email protected] WEB SITE: www.csmc.edu/cme MARCH 7–10, 2007 23rd Annual Meeting of the AANS/CNS Phoenix, AZ CONTACT: John Hurlbert, M.D. Section on Disorders of the Spine and TEL: 847/240-2500 Peripheral Nerves FAX: 847/240-0804 EMAIL: [email protected] WEB SITE: www.spinesection.org MARCH 9–11, 2007 Principles and Treatment of Spinal Atlanta, GA CONTACT: AO North America Disorders for Residents and Fellows TEL: 800/769-1391 Course FAX: 610/695-2420 WEB SITE: www.aona.org MARCH 28–31, 2007 Preservation of Motion in the Spine Florida Keys, FL TEL: 888/577-4635 EMAIL: cmemrf@floridaortho.com WEB SITE: www.currentsolutions.info APRIL 28–29, 2007 Challenges and Complications in San Francisco, CA CONTACT: AO North America Complex Spine Surgery Symposium TEL: 800/769-1391 FAX: 610/695-2420 WEB SITE: www.aona.org MAY 1–4, 2007 Spine Arthroplasty Society’s 7th Annual Berlin, Germany CONTACT: Pascale M. Davis Meeting TEL: 33/6 18 41 69 89 EMAIL: [email protected] WEB SITE: www.sas7berlin.com/clear.gif MAY 4–10, 2007 Uniformed Services University of the Bethesda, MD CONTACT: Gene Evans Health Sciences and Walter Reed and TEL: 757/496-5742 National Naval Medical Centers 19th WEB SITE: www.bethesdaspine.com Annual International Bethesda Spine & Peripheral Nerve Workshop MAY 19–20, 2007 Controversies in Spine Trauma Symposium Naples, FL CONTACT: AO North America TEL: 800/769-1391 FAX: 610/695-2420 WEB SITE: www.aona.org MAY 30–JUNE 1, 2007 23rd Annual Meeting of the Cervical Spine Leuven, Belgium CONTACT: Jan Goffin, M.D., Ph.D. Research Society-European Section FAX: 32 16 34 42 85 EMAIL: [email protected] WEB SITE: www.csrs-leuven2007.be

E413 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CALENDAR OF EVENTS

JUNE 2–3, 2007 Spine Deformity Symposium Coronado, CA CONTACT: AO North America TEL: 800/769-1391 FAX: 610/695-2420 WEB SITE: www.aona.org JUNE 2–3, 2007 Degenerative Spine Symposium Coronado, CA CONTACT: AO North America “Principles and Emerging Technologies” TEL: 800/769-1391 FAX: 610/695-2420 WEB SITE: www.aona.org JULY 7–8, 2007 Challenges and Advances: Distraction Toronto, Canada Contact: Osteogenesis Tel: Fax: Web Site: JULY 11–14, 2007 14th International Meeting on Advanced Paradise Island CONTACT: Jill E. Smith, Medical Meetings Spinal Technologies (IMAST) (Nassau) Bahamas Assistant TEL: 630/681 1040 FAX: 630/682 5811 EMAIL: [email protected] WEB SITE: www.imastonline.org AUGUST 2–5, 2007 Advanced Concepts in the Management Ontario, Canada CONTACT: AO North America of Spinal Disorders TEL: 800/769-1391 FAX: 610/695-2420 WEB SITE: www.aona.org SEPTEMBER 27–29, 2007 Spine Tumor Symposium Vancouver, Canada CONTACT: AO North America TEL: 800/769-1391 FAX: 610/695-2420 WEB SITE: www.aona.org NOVEMBER 2–4, 2007 Principles and Treatment of Spinal Las Vegas, NV CONTACT: AO North America Disorders for Residents and Fellows TEL: 800/769-1391 Course FAX: 610/695-2420 WEB SITE: www.aona.org STEREOTACTIC / FUNCTIONAL

FEBRUARY 13–14, 2007 UCLA Shaped Beam Radiosurgery Los Angeles, CA CONTACT: UCLA Stereotactic Radiosurgery Tutorial Course (Basic) TEL: 310/267-5217 FAX: 310/794-1848 Email: [email protected] Web Site: http://neurosurgery.ucla.edu/ conferences FEBRUARY 17–18, 2007 22nd Annual Washington Bethesda, MD CONTACT: Rene Sutton Neuroradiology Course TEL: 202-782-2637 FAX: 202-782-5020 EMAIL: sutton@aﬁp.osd.mil WEB SITE: www.aﬁp.org/Departments/ edu/upcoming.htm FEBRUARY 23–25, 2007 First International Symposium on Orlando, FL CONTACT: Martha Tobin Stereotactic Body Radiation Therapy and TEL: 800/223/2273, x53449 Stereotactic Radiosurgery EMAIL: [email protected] WEB SITE: www.clevelandclinic.org/ neuroscience/professionals/cme MARCH 12–16, 2007 Principles and Practice of Gamma Pittsburgh, PA CONTACT: Charlene Baker Knife Radiosurgery TEL: 412/647-7744 EMAIL: [email protected] WEB SITE: www.neurosurgery.pitt.edu APRIL 24–25, 2007 UCLA Shaped Beam Radiosurgery Los Angeles, CA CONTACT: UCLA Stereotactic Radiosurgery Tutorial Course (Basic) TEL: 310/267-5217 FAX: 310/794-1848 EMAIL: [email protected] Web Site: http://neurosurgery.ucla.edu/ conferences MAY 14–18, 2007 Principles and Practice of Gamma Pittsburgh, PA CONTACT: Charlene Baker Knife Radiosurgery TEL: 412/647-7744 EMAIL: [email protected] WEB SITE: www.neurosurgery.pitt.edu

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | E414 CALENDAR OF EVENTS

MAY 16–18, 2007 2nd Anatolian Course of Interventional Ankara, Turkey CONTACT: Prof. Dr. Saruhan Cekirge Neuroradiology TEL: 90/312 305 11 88 FAX: 90/312 311 21 45 EMAIL: [email protected] WEB SITE: www.acinr.org JUNE 12–13, 2007 UCLA Shaped Beam Radiosurgery Los Angeles, CA TEL: 310/267-5217 Tutorial Course (Basic) FAX: 310/794-1848 EMAIL: [email protected] Web Site: http://neurosurgery.ucla.edu/ conferences JUNE 23–27, 2007 ISRS: International Stereotactic San Francisco, CA CONTACT: ISRS Headquarters Secretariat Radiosurgery Society TEL: 32/2 779 59 59 FAX: 32/2 779 59 60 EMAIL: [email protected] WEB SITE: www.ISRS2007.org JULY 9–13, 2007 Principles and Practice of Gamma Pittsburgh, PA CONTACT: Charlene Baker Knife Radiosurgery TEL: 412/647-7744 WEB SITE: www.neurosurgery.pitt.edu SEPTEMBER 24–28, 2007 Principles and Practice of Gamma Pittsburgh, PA CONTACT: Charlene Baker Knife Radiosurgery TEL: 412/647-7744 EMAIL: [email protected] WEB SITE: www.neurosurgery.pitt.edu NOVEMBER 12–16, 2007 Principles and Practice of Gamma Pittsburgh, PA CONTACT: Charlene Baker Knife Radiosurgery TEL: 412/647-7744 EMAIL: [email protected] WEB SITE: www.neurosurgery.pitt.edu TUMOR

FEBRUARY 19–23, 2007 45th Annual Dr. Kenneth M. Earle Bethesda, MD CONTACT: Rene Sutton Memorial Neuropathology Review TEL: 202/782-2637 FAX: 202/782-5020 EMAIL: sutton@aﬁp.osd.mil WEB SITE: www.aﬁp.org/Departments/edu/ upcoming.htm APRIL 13–14, 2007 AANS/CNS Section on Tumors Seventh Washington, DC CONTACT: Paula Nedza Biennial Symposium TEL: 847/378-0500 EMAIL: [email protected] WEB SITE: www.AANS.org

CEREBROVASCULAR

FEBRUARY 7–9, 2007 AANS/CNS Cerebrovascular Section and San Francisco, CA CONTACT: Nathalie Johnson American Society of Interventional & TEL: 847/378-0500 Therapeutic Neuroradiology EMAIL: [email protected] WEB SITE: www.AANS.org FEBRUARY 7–9, 2007 International Stroke Conference 2007 San Francisco, CA CONTACT: National Center WEB SITE: strokeconference.org MARCH 5–9, 2007 Microsurgery of Aneurysms: St. Louis, MO CONTACT: Practical Anatomy and Surgical Recent Advances Education TEL: 314/535-4000 FAX: 314/535-8214 WEB SITE: http://pawslab.slu.edu JULY 30, 2007– 4th Annual ASITN Course & Workshops Dana Point, CA CONTACT: ASITN AUGUST 3, 2007 WEB SITE: www.asitn.org TRAUMA / CRITICAL CARE

FEBRUARY 9–11, 2007 Eighth Annual Conference of the Indian New Delhi, India CONTACT: Dr. P. K. Bithal Society of Neuroanaesthesiology and TEL: 91/11-2659-4347 Critical Care EMAIL: [email protected] MARCH 24–25, 2007 Principles of Operative Treatment of Syracuse, NY CONTACT: AO North American Craniomaxillofacial Trauma and TEL: 800/769-1391 Reconstruction FAX: 610-695-2420 WEB SITE: www.aona.org

E414 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CALENDAR OF EVENTS

MINIMALLY INVASIVE NEUROSURGERY

APRIL 21–22, 2007 Principles of Operative Treatment of Cleveland, OH CONTACT: AO North American Craniomaxillofacial Trauma and TEL: 800/769-1391 Reconstruction FAX: 610/695-2420 WEB SITE: www.aona.org MAY 24–26, 2007 2nd Biennial International Vocational Vancouver, Canada TEL: 604/875 1775 Outcomes in Traumatic Brain Injury FAX: 604/682 1521 Conference EMAIL: [email protected] WEB SITE: www.tbicvancouver.com MAY 29, 2007 Surgical Management of Maxillofacial Jasper, Canada CONTACT: AO North America Trauma Canadian Association of Oral TEL: 800/769-1391 and Maxillofacial Surgeons (CAOMS) FAX: 610/695-2420 WEB SITE: www.aona.org SEPTEMBER 21–23, 2007 International Conference on Recent Tianjin, China CONTACT: Kui Lui, M.D., Ph.D. Advances in Neurotraumatology TEL: 86/22-23359875 FAX: 86/22-23359858 EMAIL: [email protected] AUGUST 4–5, 2007 Principles of Operative Treatment of Kansas City, MO CONTACT: AO North America Craniomaxillofacial Trauma and TEL: 800/769-1391 Reconstruction FAX: 610/695-2420 WEB SITE: www.aona.org AUGUST 18–19, 2007 Principles of Operative Treatment of Minneapolis, MN CONTACT: AO North America Craniomaxillofacial Trauma and TEL: 800/769-1391 Reconstruction FAX: 610/695-2420 WEB SITE: www.aona.org OCTOBER 6–7, 2007 Principles of Operative Treatment of Charlotte, NC CONTACT: AO North America Craniomaxillofacial Trauma and TEL: 800/769-1391 Reconstruction FAX: 610/695-2420 WEB SITE: www.aona.org OCTOBER 20–21, 2007 Principles of Operative Treatment of Dallas, TX CONTACT: AO North America Craniomaxillofacial Trauma and TEL: 800/769-1391 Reconstruction FAX: 610/695-2420 WEB SITE: www.aona.org NOVEMBER 3–4, 2007 Principles of Operative Treatment of Miami, FL CONTACT: AO North America Craniomaxillofacial Trauma and TEL: 800/769-1391 Reconstruction FAX: 610/695-2420 WEB SITE: www.aona.org DECEMBER 1–2, 2007 Principles of Operative Treatment of New Jersey or CONTACT: AO North America Craniomaxillofacial Trauma and New York (TBD) TEL: 800/769-1391 Reconstruction FAX: 610/695-2420 WEB SITE: www.aona.org CRANIAL BASE

MAY 2–6, 2007 Eighth Congress of European Skull Prague, Czech Republic CONTACT: Prof. Vladimir Benes, M.D., Ph.D. Base Society: Multimodality Management TEL: 420/973 202 951 FAX: 420/973 202 963 EMAIL: [email protected] WEB SITE: www.esbs2007.com MARCH 26–30, 2007 Hands-on Workshop in “Endoscope- Mainz, Germany CONTACT: Ms. Sandra Hoelle Assisted Keyhole Microneurosurgery” TEL: 49/7461 95 1132 FAX: 49/7461 95 2050 EMAIL: [email protected] WEB SITE: www.aesculap-neuro.com MAY 9–12, 2007 Neuroendoscopy 2007 Paris, France CONTACT: NEUROENDOSCOPY 2007/MCI FRANCE TEL: 01 53 85 82 53 FAX: 01 53 85 82 83 EMAIL: [email protected] WEB SITE: www.neuroendoscopy2007.com

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | E415 CALENDAR OF EVENTS

JUNE 27–30, 2007 ISCAS—11th Annual Conference of the Berlin, Germany CONTACT: Ms. Franziska Schweikert International Society for Computer Aided TEL: +49/7742 922 434 Surgery FAX: +49/7742 922 438 EMAIL: ofﬁ[email protected] WEB SITE: http://www.cars-int.org JULY 2–6, 2007 Hands-on Workshop in “Endoscope- Tuttlingen, Germany CONTACT: Ms. Sandra Hoelle Assisted Keyhole Microneurosurgery” TEL: 49/7461 95 1132 FAX: 49/7461 95 2050 EMAIL: [email protected] WEB SITE: www.aesculap-neuro.com SEPTEMBER 3–7, 2007 Hands-on Workshop in “Endoscope- Mainz, Germany CONTACT: Ms. Sandra Hoelle Assisted Keyhole Microneurosurgery” TEL: 49/7461 95 1132 FAX: 49/7461 95 2050 EMAIL: [email protected] WEB SITE: www.aesculap-neuro.com SPECIAL TOPICS

FEBRUARY 9–11, 2007 2nd International Symposium and 8th New Delhi, India CONTACT: Dr. P.K. Bithal Annual Conference of Indian Society TEL: 986/839-8201 of Neuroanesthesiology and Critical Care EMAIL: [email protected] Web Site: www.aiims.edu FEBRUARY 17–18, 2007 22nd Annual Washington Bethesda, MD CONTACT: Rene Sutton NEURORADIOLOGY COURSE TEL: 202-782-2637 FAX: 202-782-5020 EMAIL: sutton@aﬁp.osd.mil WEB SITE: www.aﬁp.org/Departments/edu/ upcoming.htm FEBRUARY 19–21, 2007 The 3rd Annual Update Symposium on Tel Aviv, Israel CONTACT: Secretariat: ISAS International Clinical Neurology and Neurophysiology Seminars TEL: 972/2-6520574 FAX: 972/2-6520558 EMAIL: [email protected] WEB SITE: www.neurophysiologysymposium. com FEBRUARY 24–25, 2007 Weekend Update: Interactive Review of Atlanta, GA CONTACT: Vanessa Garlisch Clinical Neurosurgery by Case TEL: 847/378-0500 Management EMAIL: [email protected] WEB SITE: www.AANS.org FEBRUARY 24–26, 2007 Advanced Symposium on Lake Tahoe, NV CONTACT: AO North American Craniomaxillofacial Reconstruction TEL: 800/769-1391 FAX: 610/695-2420 WEB SITE: www.aona.org FEBRUARY 26–28, 2007 4th Annual Summit on Building & Scottsdale, AZ CONTACT: WRG Research, Inc. Streamlining Neuroscience Centers TEL: 800/647-7600 of Excellence FAX: 781/939-2490 EMAIL: [email protected] Web Site: www.worldrg.com FEBRUARY 28– Advancements in Neuroscience for the Phoenix, AZ CONTACT: Barrow Neurological Institute MARCH 2, 2007 Practicing Neurosurgeon and Neurologist: WEB SITE: www.thebarrow.com/conferences 33rd Annual Barrow Symposium MARCH 16–18, 2007 7th Meeting of the Conosur Society of Gramado, Brazil CONTACT: Apio Antunes, M.D., Ph.D. Neurological Surgeons TEL: 55/51-3222-5760 FAX: 55/51-3222-5760 EMAIL: [email protected] WEB SITE: www.plenariumcongressos.com.br MARCH 17–20, 2007 International Symposium on Cognitive Tübingen, Germany TEL: 49/70712980325 Neurosurgery FAX: 49/7071294549 EMAIL: [email protected] Web Site: www.iscns.org MARCH 30–APRIL 1, 2007 Challenges and Advances: State of the Art Baltimore, MD CONTACT: AO North American Reconstruction in Microsurgery TEL: 800/769-1391 FAX: 610/695-2420 WEB SITE: www.aona.org

E415 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com CALENDAR OF EVENTS

APRIL 28–29, 2007 Fundamentals of Orbital and Mid-facial Rosemont, IL CONTACT: AO North America Reconstruction TEL: 800/769-1391 FAX: 610/695-2420 WEB SITE: www.aona.org MAY 6–JUNE 8, 2007 The Society of Neurological Surgeons San Francisco, CA WEB SITE: www.societyns.org Annual Meeting (Senior Society) MAY 20–22, 2007 Neurosurgery Review by Case Houston, TX CONTACT: Vanessa Garlisch Management: Oral Board Preparation TEL: 847/378-0600 EMAIL: [email protected] WEB SITE: www.AANS.org MAY 22–26, 2007 ABNS Oral Board Examination Houston, TX CONTACT: ABNS TEL: 715/441-6015 FAX: 715/794-0207 EMAIL: [email protected] WEB SITE: www.abns.org MAY 27–29, 2007 ABNS Oral Board Examination Houston, TX TEL: 715/441 6015 FAX: 715/794 0207 EMAIL: [email protected] WEB SITE: www.abns.org JUNE 9–14, 2007 ASNR 45th Annual Meeting & NER Chicago, IL CONTACT: ASNR 45th Annual Meeting Foundation Symposium 2007 TEL: 630/574-0220 FAX: 630/574-1740 EMAIL: [email protected] WEB SITE: www.asnr.org/2007 JUNE 15–16, 2007 5th Annual ASITN Practicum Chicago, IL CONTACT: ASITN TEL: 703/691-2272 WEB SITE: www.asitn.org JUNE 27–30, 2007 11th Annual Conference of the Berlin, Germany CONTACT: CARS Conference Office International Society for Computer TEL: 49/7742922434 Aided Surgery FAX: 49/7742922438 EMAIL: offi[email protected] WEB SITE: www.cars-int.org JUNE 27–30, 2007 Computer Assisted Radiology & Berlin, Germany TEL: 49/7742922434 Surgery—21st International Congress FAX: 49/7742922438 & Exhibition EMAIL: offi[email protected] WEB SITE: www.cars-int.org JULY 7–8, 2007 Challenges and Advances: Distraction Toronto, Canada CONTACT: AO North America Osteogenesis TEL: 800/769-1391 FAX: 610/695-2420 WEB SITE: www.aona.org SEPTEMBER 5–8, 2007 The British Scoliosis Society in Conjunction Edinburgh, Scotland CONTACT: Scoliosis Research Society with the Scoliosis Research Society TEL: 414/289-9107 42nd Annual Meeting FAX: 414/276-3349 EMAIL: [email protected] WEB SITE: www.srs.org SEPTEMBER 5–8, 2007 SRS 42nd Annual Meeting & One-day Edinburgh, Scotland CONTACT: Michael John McMaster Course WEB SITE: www.srs.org/meetings NOVEMBER 4-6, 2007 Neurosurgery Review by Case Houston, TX CONTACT: Vanessa Garlisch Management: Oral Board Preparation TEL: 847/378-0500 EMAIL: [email protected] WEB SITE: www.AANS.org

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | E416 THIS MONTH IN NEUROSURGERY

FEBRUARY 2007 Podcasts available at www.neurosurgery-online.com EDITOR’S CHOICES

FEATURED ON THE WEB

242 268 307 392 PATHOGENESIS OF INTRACRANIAL ANEURYSMS Other than imaging studies to screen high risk NO LONG-TERM EXCESS MORTALITY IN patients with temporal mediobasal tumors. individuals, there are currently no reliable 280 PATIENTS WITH RUPTURED DISTAL The authors described, amongst others, methods to identify intracranial aneurysms (IA) in ANTERIOR CEREBRAL ARTERY ANEURYSMS tumors restricted to the hippocampus, asymptomatic patients. Population studies have Relatively little is known about the long- amygdalum, or parahippocampus alone. determined risk factors such as IA size, location, term outcome of patients after treatment for More mesial location and smaller size were hypertension, smoking, and even gender. associated with benign pathology. Classical subarachnoid hemorrhage. In this study, Furthermore, there is evidence to suggest that the authors conducted a population-based approaches such as transsylvian or anterior analysis of 280 consecutive patients treated two-third lobe resection were used fre- genetic susceptibility plays an important role in IA for ruptured distal anterior cerebral artery quently. Most neurologic complications formation and rupture. Molecular genetic studies aneurysms at two neurosurgical centers were transient, with permanent complica- have already identified multiple locations serving Southern and Eastern Finland from tions being under 2%. However, post- throughout the human genome that are relevant. 1976 to 2003. Of these patients, 262 were operative visual field deficits were not un- Identification of these genes will ultimately treated with surgical clipping, 10 with coiling, common. This series demonstrates that improve the detection and treatment of this highly and 8 received no intervention. The median these tumors can be surgically treated with debilitating disease. [p. 213] follow-up was 9.6 years, and no patients were relative safety. [p. 285] lost to follow-up. Long-term excess mortality ESULTS OF ICROSURGICAL LIPPING was compared to the general Finnish popu- OBJECTIFYING WHEN TO HALT A BOXING R M C lation matched for age, sex, and calendar MATCH: A VIDEO ANALYSIS OF FATALITIES OF 50 HIGH-COMPLEXITY BASILAR time. The study showed no excess mortality Currently, the criteria used to halt boxing APEX ANEURYSMS after surviving 3 years from rupture. [p. 235] matches are quite subjective. Developing a Although the contemporary treatment of basilar standardized, objective method to determine apex aneurysms rarely involves microsurgical RADIOSURGERY OF CEREBRAL ARTERIOVENOUS an appropriate halting point in a contest clipping, those of high complexity frequently fail MALFORMATIONS IN THE PEDIATRIC-AGE GROUP could increase the safety of boxers. To this endovascular treatment. The authors report on the In this retrospective study, the authors report end, this study compared the number and on the results of 100 children with 103 types of punches landed in a typical pro- results of the microsurgical treatment of 50 high cerebral arteriovenous malformations (AVMs) fessional match to those that ended in complexity basilar apex aneurysms. They report a treated with linear accelerator radiosurgery. fatalities. Several statistically significant good outcome in 92% and moderate outcome in This represents one of the largest published differences were discovered between the 4% of patients using the pretemporal trans- series reported in the literature. Sixty-seven control group and matches that resulted in cavernous approach. Complete clipping was lesions were located in functional areas, and fatalities. However, the practical application achieved in 98% without any rebleeding during 30% were inoperable. Fifty patients had of these findings becomes problematic as the the follow-up period. This experience reintro- multimodal treatment with embolization matches become more competitive. [p. 307] duces microsurgery as a safe and more durable and/or surgery, before and/or after radio- treatment option for complex basilar apex surgery. Complete obliteration of AVMs was RISK FACTORS ASSOCIATED WITH aneurysms. [p. 242] achieved in 70%, with a mean delay to POST-CRANIOTOMY MENINGITIS obliteration of 33 months. The morbidity The risk factors associated with post-crani- rate was 5%, and the major predictive factor otomy meningitis has been studied in a HISTORY OF SPINE BIOMECHANICS: of obliteration was lesion volume, with cohort of 453 patients. The cohort included PART I—THE ANCIENT AND MEDIEVAL efficacy decreasing with increasing age. patients over 18 years of age undergoing ROOTS OF SPINE BIOMECHANICS This study confirms the safety and efficacy non-stereotactic craniotomies and surviving This article is the first of a two part series over 7 days. In total, there were 25 cases of of radiosurgery for treating cerebral AVM in focusing on the history of spine biomechanics. the pediatric population. [p. 268] meningitis. The presence of infection at other body sites and duration of surgery were In this part, the authors address the ancient and NEUROCOGNITIVE FUNCTION IN PATIENTS important predisposing factors. However, medieval roots of biomechanics. The paper WITH ONE TO THREE NEW BRAIN the duration of drainage was the single most reviews historical treatises and ancient METASTASES INITIALLY TREATED WITH important risk factor. The study concludes medical, philosophical, and physical docu- STEREOTACTIC RADIOSURGERY ALONE that continuous device-related postoperative ments. The information contained in these Patients and physicians considering stereo- communication of cerebrospinal fluid (CSF) historical documents largely comprised the tactic radiosurgery (SRS) alone for the treat- and the environment is the major risk factor body of literature until the Renaissance. ment of newly diagnosed brain metastases for post-craniotomy meningitis. [p. 317] [p. 382] may be concerned with the neurocognitive implications associated with this choice. LONG-TERM OUTCOME IN PATIENTS WITH Fifteen patients with one to three lesions SUSPECTED NORMAL PRESSURE HYDROCEPHALUS THE HISTORY OF SPINE BIOMECHANICS: were treated with initial SRS alone with Recently, many studies have reported PART II—THE POST-RENAISSANCE ERA omission of whole brain radiation therapy. relatively good outcomes after CSF shunting This is the second article in a two part series Assessment of neurocognitive function (NCF) for normal pressure hydrocephalus (NPH). focusing on the history of spine bio- was performed over time. Most patients had However, most of these studies had mechanics. In this part, the authors address impairment of their NCF before treatment. relatively short follow-up times. In the the studies performed by post-Renaissance After treatment, stable or improved NCF in the present prospective study, results at 6 scientists on biomechanics and, in part, areas of learning/memory, abstract reasoning, months and 5 years are reported in 75 on spine biomechanics. This review also consecutive patients with a mean age of and motor dexterity was observed in the covers studies on bone and muscle majority of five long-term survivors. [p. 277] 72.5 years. Good long-term results were found to be hampered mainly by comorbid- biomechanics in the 18th and 19th SURGERY FOR TEMPORAL MEDIOBASAL TUMORS: ities, in particular vascular and malignant centuries. In addition, special studies on EXPERIENCE BASED ONASERIES OF 235 CASES diseases, emphasizing the need for strict spine biomechanics are considered, This study examines the histology, symptoms, patient selection criteria. The study found along with biomechanics organiza- outcome, and complications associated with that 40–60% of the patients remained tions and publications in the last the surgical treatment of a large series of improved at long-term follow-up. [p. 327] century. [p. 392]

CHARLES Y. LIU, M.D., PH.D., CONTRIBUTOR N5 CHRONICLING THE TIMES IN THE INFORMATION AGE

On Lumbar Disc Herniation, SPORT and “We Told You So” Science Times Editorial

he results of the randomized trial and obser- pretations of Randomized Controlled Trials (RCTs) vational cohort from the National Institutes of surgical disease and to caution against the mis- Tof Health sponsored Spine Patient Outcomes guided conclusion of “equivalence of surgical and Research Trial (SPORT) were recently published non-surgical care.” This was predictable from the by Weinstein et al. in two papers in the Journal of study design. A report prepared by one of us (REH) the American Medical Association (JAMA 296: for the AANS/CNS Washington Committee in May 2441–2450, and 2451–2459, 2006). Results of sur- 2000 regarding the design of SPORT included the gically treated patients were superior in every meas- following statements: ure and at every end point in comparison to those “One problem is that of patient selection. treated non-surgically, and yet the intent-to-treat Primary care physicians will send patients analysis, the only statistically and methodologically with severe pain and radiographically docu- valid comparison, did not yield a significant differ- mented structural spine problems directly for ence among cases initially randomized to surgery neurosurgical or orthopedic evaluation and or medical therapy. After many years and millions treatment. Patients with equivocal findings are of dollars spent on this trial, what have we learned? more likely to be sent to a comprehensive Almost two thousand eligible patients with spine clinic. Even within the spine clinic pop- painful lumbar disc herniation were offered enroll- underwent surgery at 3 months and 45% had sur- ulation, the investigators estimate that they will ment in the study; 1244 ultimately enrolled, and gery within 2 years of enrollment. While analyzing be able to randomize only 15 to 40 percent of 501 agreed to randomization to either surgical treat- patients who actually underwent surgery, there were patients who meet study criteria. Those ment or a strict regimen of medical therapy. The “strong, statistically significant advantages . . . for patients who are evaluated but elect not to be remaining 743 patients who declined randomiza- surgery at all follow-up times.” involved in the randomized study will be fol- tion selected their own treatment mode after discus- In the observational study, group comparison lowed. It is likely that patients with more sion with their physicians and were enrolled in a was even murkier. The patients who underwent sur- severe symptoms and more impressive struc- parallel observational study. Outcomes were gery in the non-randomized arm of the trial had tural pathology will be triaged to surgical care. assessed using the same validated endpoints and at greater initial pain and disability than those who This will eliminate patients from the random- the same time points in the observational study and chose non-surgical intervention and also greater ized study who are most likely to respond to the randomized trial. A critical feature of the study, than those who agreed to randomization and then surgical intervention and raises the question which is ethically unavoidable but methodologi- chose surgery. Those who refused randomization as to whether or not the study population will cally treacherous, allowed crossover to surgery at and had non-operative treatment had less pain and be representative of lumbar surgery patients. If any time from the subgroup assigned to medical disability than those in the randomized trial who the patient in agony with a large free disc frag- therapy, and the refusal or delay of surgery in selected non-operative management. Adjusting for ment benefits more from surgery than the patients assigned to the surgical group. baseline differences, the analysis of the observa- patient with intermittent sciatica from a In the randomized trial, the two randomized tional study showed that, at 3 months and 1 and 2 bulging disc, and if the former are underrepre- groups had similar baseline characteristics of initial years, “[t]reatment effects were statistically signif- sented and the latter are overrepresented in the pain and disability, and other measures of disease icant in favor of surgery for the primary outcome study, the benefits of surgical intervention will severity. There were no incident cases of neuro- measures,” and there were significantly better self- be underestimated.” logic deterioration in either assigned group, and reported outcomes among patients undergoing sur- “There also are problems with the method- there were substantial improvements in both gery. The authors again noted that both groups had ology in regard to crossover patients. In order groups. The intent-to-treat analysis, representing significant improvements over the 2 year study to retain the benefits of randomization, the the methodologically valid comparison of the ran- period, and no one was harmed during the study, study is designed as an-intention-to-treat domized groups revealed slightly better outcome regardless of treatment received. analysis. Patients are considered members of scores for the overall groups assigned to surgery, This was the first multi-center, prospective, ran- the group to which they were randomized, but these did not reach statistical significance. domized trial of surgical versus non-surgical treat- even when they have crossed over to the other However, crossover was a major problem, preclud- ment of patients with disc herniation. And it treatment group. The investigators anticipate ing a true assessment of superiority of one modal- demonstrated better outcome in patients treated sur- that up to 25 percent of the patients originally ity over another. Only 50% of the 245 patients ran- gically but no difference among patients initially randomized to non-surgical therapy may cross domized to the surgery arm underwent surgery by randomized to surgery or non-operative manage- over to the surgical group. If this group of the 3 month timepoint, and 60% underwent sur- ment. What an anticlimactic outcome! patients then does well in long-term follow- gery by 2 years. Of the 256 patients assigned to In this editorial we use this trial as a paradigm to up, the benefit will be credited to the nonsur- receive nonoperative care, 30% crossed over and illustrate the potential shortcomings and misinter- gical treatment group. This design will maxi-

N6 | VOLUME 60 | NEUROSURGERY 2 | FEBRUARY 2007 www.neurosurgery-online.com CHRONICLING THE TIMES IN THE INFORMATION AGE

mize the benefits of nonsurgical treatment and does not solve this problem as all unblinded Surgical RCTs also suffer from problems with minimize the benefits of surgical care.” observers bring their biases to the evaluation. surgeon selection because the skill and experience A properly designed RCT must have four essen- Choosing a representative patient population in of the surgeon have profound effects on outcome. A tial components: (1) concurrent comparisons to surgical trials is also difficult to accomplish. study showing a benefit from surgery with an eliminate temporal bias; (2) objective observation Surgeons have an implicit contract with the patient exceptional group of surgeons will not apply to the of clear endpoints to eliminate physician and patient to offer the best care available. If surgeons do not likely outcomes of an individual surgeon who fails bias; (3) randomization to equalize the effects of believe that surgical and non-surgical treatments to match the expertise of surgeons in the study. unknown, confounding variables; and (4) a represen- are equally efficacious they will offer surgical treat- Similarly, a study showing no surgical benefit may tative, adequately sized patient population to reduce ment outside the trial to those patients they believe not be applicable if the study surgeons have out- the likelihood of chance errors. The ideal RCT is an are most likely to benefit. In this situation, only comes significantly worse than a surgeon with adequately powered, double-blinded study with patients less likely to benefit from surgery will be exceptional skill and experience. We have no rea- unambiguous endpoints. Unfortunately, few if any randomized, skewing the study to the detriment of son to believe that this was a problem in SPORT surgical RCTs approximate this ideal. surgical treatment. Let us consider potential SPORT but it has been a problem in many surgical trials. Surgical RCTs differ from the ideal RCT in patients with acute, severely symptomatic lumbar As SPORT so elegantly demonstrates, surgical important ways. Surgical trials are nearly always disc herniations, who have already failed six weeks RCTs require substantial expenditures of time, unblinded and patients may readily elect to cross of conservative therapy. Are these patients’ referring effort and money. Because of these expenses it is over from one treatment arm to another. To pre- physicians more likely to refer them directly to a difficult or impossible to repeat a trial, even if there serve the benefits of randomization, it is necessary surgeon for operative care or to a comprehensive are grave concerns about the validity of the study. to analyze patients in their assigned groups even if spine center for evaluation? For the subset of Despite what we believe were inherent biases they cross over to another treatment arm. If a large patients who are sent to the spine center, are the underestimating the efficacy of surgery, the percentage of patients cross over from one treat- evaluating surgeons more likely to recommend sur- $13,500,000 spent on the SPORT study has told us ment arm to the other, as was the case in SPORT, gery or enrollment in a randomized trial? For the that most patients with herniated lumbar discs who an intent-to-treat analysis loses its relevance. sub-subset of patients seen at the spine center and do not improve adequately following conservative It is also difficult in many surgical trials to define whose surgeons recommend enrollment in the trial, therapy can anticipate a high likelihood of immedi- clear endpoints. A surgical RCT does not eliminate are they more likely to enroll or opt for surgery ate and clinically meaningful pain relief with lum- bias if endpoints are ambiguous and neither the outside the trial? Finally, we end up with a sub- bar discectomy. What a surprise. patient nor the evaluator is blinded. Patients may sub-subset of the patients likely to benefit from sur- experience a substantial placebo effect with surgery who will be entered into the randomized trial. ROBERT E. HARBAUGH, MD gery and investigators may harbor a surgical or In short, patients most likely to benefit from surgery EPIDEMIOLOGY AND OUTCOMES SCIENCE non-surgical bias. Having someone other than the will be underrepresented and those least likely to ISSAM A. AWAD, MD operating surgeon evaluate patients postoperatively benefit will be overrepresented. ASSOCIATE EDITOR

Super Size Me: New Insights Into Caloric Intake and Lifespan

The authors report on resveratrol 3,5,4’-trihy- droxystilbene, found in some red wines, which emerged from a small-molecule screen for en- hancers of SIRT1, a member of the NAD+-dependent deacetylases implicated as mediators of the ben- eficial effects of caloric restriction. As recent studies had demonstrated the ability of resveratrol to improve health and extend survival in lower Liver size in SD, HC, and HCR mice. organisms, the authors investigated the effects of this compound on health and survival in mice fed a calories towards mice on a standard diet. Mice on standard diet (SD) or high calorie diet (HC) high calorie diets lived as long as their healthier (Figure 1). After 6 months, there was a clear trend counterparts if resveratrol was consumed; insulin towards increased survival in HC mice treated with sensitivity and organ pathology were identical. The resveratrol (HCR) that was not attributable to notion that obesity-related morbidity and mortality weight loss. Moreover, insulin sensitivity in HCR may be modified in higher organisms by small-mol- mice equaled that of SD mice as determined by α Source: http://calorielab.com/news/categories/obesity- ecule modulation of PGC-1 activity is sure to research-studies oral glucose tolerance test. Strikingly, organ pathol- enter the clinical arena sooner rather than later. ogy seen in HC mice such as hepatomegaly was Perhaps just as relevant to neurosurgery is the find- nterest in obesity research will continue to prevented by the administration of resveratrol ing in the paper that resveratrol improves motor intensify as current estimates suggest that over (Figure 2), and gene expression profiles of HCR coordination as assessed by Rotarod perform- 2 billion people worldwide are overweight. mice were shifted to those of SD mice. The authors ance—this finding adds to the growing body of lit- I also suggested that SIRT1 activity is modulated by While the notion of obesity-associated morbidities erature suggesting that resveratrol may enhance is well accepted and the salutary effects of caloric resveratrol: PGC-1α, a master transcriptional coac- neuronal function and protect diseased brain from restriction are understood, how overeaters may curb tivator in mitochondrial biogenesis positively regu- further damage, perhaps by modulating cellular their excesses and how caloric restriction extends lated by SIRT1, was significantly deacetylated in metabolism. lifespan are unclear. A recent report by Baur et al in resveratrol-fed mice. Nature (444: 337–342, 2006) demonstrates a Taken together, these data show that a small- IAN F. D UNN, MD small-molecule solution to the former and sheds molecule mimetic of caloric restriction is sufficient ROBERT M. FRIEDLANDER, MD, MA light on the latter question. to shift the physiology of mice consuming excess BASIC SCIENCE RESEARCH

NEUROSURGERY VOLUME 60 | NEUROSURGERY 2 | FEBRUARY 2007 | N7 NEUROSURGERY’S Science Times

Neurocytoma Is a Tumor of Adult Neuronal Interventions to Improve Progenitor Cells Bone Density in

entral neurocy- Adolescent Girls toma (CN) is a steoporosis remains a major public rare periventricu- C health problem for aging American lar neuronal tumor of women. However, problems related to young adults whose deri- O bone density frequently begin earlier in life. For vation, lineage potential, teenage girls, proper bone maturity is important and molecular regulation for a number of reasons. First, because bone have been mostly unex- mineral loss is typically cumulative in later plored. Its neuronal phe- adult life, reaching maximal bone mineral den- notype, along with its sity in early adulthood is critical in offsetting or typical presentation along delaying the onset of sequelae from osteoporo- the lateral ventricular sis. Second, lifestyle habits conducive or wall, suggests that neuro- destructive to maintaining proper bone health cytoma originates from are frequently begun during adolescence. subependymal neural Dysregulated IGF2 signaling in neurocytoma may regulate tumorigenesis. Debar and colleagues have recently published progenitor cells. In a the results of a prospective, randomized study to recent publication by Sim et al. in the The wnt signaling, as well as GDF8 (growth differenti- investigate the effects of lifestyle alterations on Journal of Neuroscience (26:12544–12555, ation factor 8), PDGF-D, and neuregulin (NRG2), bone mineral density in teenage girls (Arch 2006) the authors looked at CN cells and found were differentially overexpressed by CN, suggest- Pediatr Adolesc 160:1269–1276, 2006). The that they exhibited an antigenic profile typical of ing that CN is characterized by the concurrent over- YOUTH study was an NIH-funded initiative neuronal progenitor cells in vivo, yet in vitro the activation of these pathways, which may serve to conducted through the Kaiser Permanente CN cells generated neurospheres, divided in drive neurocytoma expansion while restricting Northwest HMO, and the authors were affiliated response to bFGF (basic fibroblast growth factor), tumor progenitor phenotype (Figure). This strategy with the University of Arizona, Oregon Health & activated the neuroepithelial enhancer of the nestin of comparing the gene expression of tumor cells to Science University, and Washington State gene, and gave rise to both neuron-like cells and that of the normal VZ-derived progenitors may pro- University. The study attempted to enroll 228 astrocytes. As such, the neuroblasts of which neu- vide a focused approach to identifying transcripts girls between the age of 14 and 16 with a below rocytoma is mostly comprised in vivo should be important to stem and progenitor cell oncogenesis. normal body mass index. Participants were ran- viewed either as the progeny of a neoplastic, Targeted treatment strategies for neurocytoma, domized to either behavioral intervention with GFAP+ neural stem cell persistent within the including HER2 receptor blockade via the blocking bimonthly group meetings, quarterly coaching tumor mass, or as progenitor cells still able to antibody Trastuzumab (Herceptin; Genentech, San telephone calls, and weekly self-monitoring regenerate the parental GFAP+ stem cells from Francisco, CA) may prove efficacious given the ele- designed to improve diet and increase physical which they themselves may have originated. vation of neuregulin signals seen in this study. activity, or to a control group. Total bone mineral They nicely compared CN gene expression with Similarly, the PDGF receptor is a potential target of density was measured by dual-energy x-ray that of normal adult ventricular zone (VZ) neuronal several multifunctional antineoplastics, such as ima- absorptiometry in both groups at the beginning progenitors, and significant overlap was noted. tinib (Gleevec; Novartis, Basel, Switzerland) and of the yearly thereafter. Monitored behavioral Marker analysis suggested that the gene expression SU11248 (sunitinib malate) (Fiedler et al., Blood outcomes included intake of calcium, vitamin pattern of CN was that of a proneuronal population. 105: 986–993, 2005), each of which includes the D, soda, and fruits and vegetables; high-impact The particular expression pattern of CN included PDGF tyrosine kinase among its principal targets. and strength-training physical activity; measures both IGF2 and several components of its signaling of strength and fitness; and biomarkers (osteo- pathway, whose sharp overexpression implicated JEFFREY P. G REENFIELD, MD, PHD calcin and naltrexone). dysregulated autocrine IGF2 signaling in CN onco- JOHN A. BOOCKVAR, MD Compared with the control subjects, those genesis. Both receptors and effectors of canonical STEM CELL RESEARCH undergoing regular interventions experienced significantly higher bone mineral density in the spine and trochanter regions during the first study year, Taming Thrombolytics: Limiting the Damage of tPA which was maintained during the second study year. The naltrexone biomarker demonstrated a he acute treatment of ischemic stroke has metalloproteinase-9 pathway of tPA is blocked in greater relative decrease in the intervention group been revolutionized by the possibility of tis- the endothelium of ischemic brain both in vivo and compared with the control group, with no signif- sue plasminogen activator (tPA) therapy. in vitro, and in their model, markedly reduced hem- icant changes in osteocalcin consistent with more T orrhagic complications. This study, which is still While this has made a significant impact on our bone building in the intervention group. ability to treat patients with acute ischemic strokes, presented only in the rodent model notes that Participants in the intervention group reported sig- the hemorrhagic complications of thrombolytic matrix metalloproteinase-9 breaks down the neu- nificantly greater consumption of calcium in both therapy in the setting of stroke remain a serious rovascular matrix and disturbs blood brain barrier. study years, vitamin D in the first year, and fruits complication limiting the safety of this therapy. In Therefore, selectively inhibiting this function of and vegetables in both years. There was no effect an animal study published in Nature Medicine tPA may increase safety when APC is given con- on soda consumption or target exercise rates. (12: 1278–1285, 2006), Cheng et al. demonstrate comitantly with tPA in the ischemic setting. The These encouraging results led the authors to the possibility of using activated protein C (APC) authors essentially attempted to separate tPA func- conclude that lifestyle-based interventions can to block tPA hemorrhagic complication. They rea- tions on fibrin and clot from those damaging effects improve dietary intake habits and increases in son that APC, a plasma serine protease with multi- on the blood brain barrier. This certainly suggests bone mineral density in teenage girls. Notably, ple capacities, including among them, effects on that by increasing our understanding of the full this is the first healthcare-based study to demon- systemic anticoagulation, but also anti-inflamma- range of actions of this therapeutic agent, we may strate a significant improvement in bone quality tion and antiapoptotic activities, vasoprotective be able to progressively tailor its effects for safety in adolescent girls. and efficacy. activities and neuroprotective activities, may work MICHAEL Y. WANG, MD to inhibit hemorrhagic activity of tPA. In this study, ROBERT J. DEMPSEY, MD SPINAL RESEARCH they demonstrate that the NF-kb dependent matrix TRANSLATIONAL RESEARCH

N8 | VOLUME 60 | NEUROSURGERY 2 | FEBRUARY 2007 www.neurosurgery-online.com Coming in the March Issue of... EUROSURGER N OFFICIAL JOURNAL OF THE CONGRESS OF NEUROLOGICAL SURGEONSY MARCH 2007 VOLUME 60 NUMBER 3

TOPIC REVIEW EXPERIMENTAL STUDIES Rebuilding Lost Hearing by Cell Transplantation Quantum Dots are Phagocytized by Macrophages and Co-localize Sekiya with Experimental Glioma Jackson/Toms

CLINICAL STUDIES p53 May Play an Orchestrating Role in Apoptotic Cell Death After Experimental Subarachnoid Hemorrhage Coiling Versus Clipping for the Treatment of Aneurysmal Subarachnoid Cahill/Zhang Hemorrhage: A Longitudinal Investigation into Cognitive Outcome Frazer Treatment of Traumatic Brain Injury with a Combination Therapy of Marrow Stromal Cells and Atorvastatin in Rats Embolization Before Radiosurgery Reduces the Obliteration Rate Mahmood of Arteriovenous Malformations Andrade-Souza/Schwartz SPECIAL ARTICLE Radiosurgery to Reduce the Risk of First Hemorrhage from Brain Arteriovenous Malformations Evolution of the Human Brain: Changing Brain Size and the Fossil Record Maruyama Park/Levy

Stereotactic Radiosurgery for Vestibular Schwannomas in Patients LEGACY with Neurofibromatosis Type 2: An Analysis of Tumor Control, Complications, and Hearing Preservation Rates Lancisi’s Nerves and the Seat of the Soul Mathieu/Kondziolka Di Ieva

Gamma Knife Radiosurgery in the Management of Malignant Melanoma Brain Metastases CASE REPORTS Mathieu/Kondziolka Astroblastoma Mimicking a Cavernous Malformation: Case Report Tumialan/Barrow Extradural Transcavernous Approach to Cavernous Sinus Hemangiomas: Series of Seven Cases and Review Cerebellar Cryptococcoma in a Patient with Undiagnosed Sarcoidosis: Suri Case Report Kanaly/Adamson In Vivo Study of Head Impacts in Football: A Comparison of National Collegiate Athletic Association Division I versus High School Impacts Giant, Partially Thrombosed Internal Carotid Artery Aneurysm Schnebel Develops after Remote Angiographically Negative Subarachnoid Hemorrhage: Case Report Evaluation and Treatment of Patients with Suspected Normal Pressure Kakarla/Spetzler Hydrocephalus on Long-term Warfarin Anticoagulation Therapy Goodwin/Williams Acute Presentation of Spinal Epidural Cavernous Angiomas: Case Report Caruso/Galarza Minimally Invasive Lumbar Spinal Decompression for the Elderly: Outcomes of 50 Patients Aged 75 Years and Older TECHNICAL CASE REPORTS Rosen/Fessler Adjunctive Rheolytic Thrombectomy for Central Venous Sinus Thrombosis: Technical Case Report METHODOLOGY IMPROVEMENT Kirsh/Fiorella

Color Intensity Projection of Digitally Subtracted Angiography for Hyperbaric Oxygen in the Treatment of a Radiosurgical Complication: Visualization of Brain Arteriovenous Malformations Technical Case Report Cover Lynn/Friedman

SPECIAL ARTICLE SPECIAL TECHNIQUE APPLICATION Litigation of Missed Cervical Spine Injuries in Patients Presenting CyberKnife Targeting the Pterygopalatine Ganglion for the Treatment with Blunt Traumatic Injury of Chronic Cluster Headache Lekovic/Harrington Lad/Chang/Adler

Entire Journal content and NEUROSURGERY Podcasts available online at www.neurosurgery-online.com