FUNCTIONAL RADIOSURGERY

NONISOCENTRIC RADIOSURGICAL RHIZOTOMY FOR TRIGEMINAL NEURALGIA John R. Adler, Jr., M.D. Department of Neurosurgery, OBJECTIVE: Although stereotactic radiosurgery is an established procedure for treat- Stanford University Medical Center, ing trigeminal neuralgia (TN), the likelihood of a prompt and durable complete response Stanford, California is not assured. Moreover, the incidence of facial numbness remains a challenge. To Regina Bower, M.D. address these limitations, a new, more anatomic radiosurgical procedure was devel- oped that uses the CyberKnife (Accuray, Inc., Sunnyvale, CA) to lesion an elongated Department of Neurological Surgery, Mayo Clinic College of Medicine, segment of the retrogasserian cisternal portion of the trigeminal sensory root. Because Rochester, Minnesota the initial experience with this approach resulted in an unacceptably high incidence of facial numbness, a gradual dose and volume de- escalation was performed over sev- Gaurav Gupta, M.S.E. eral years. In this single- institution prospective study, we evaluated clinical outcomes Department of Neurosurgery, in a group of TN patients who underwent lesioning with seemingly optimized non- Stanford University Medical Center, Stanford, California isocentric radiosurgical parameters. METHODS: Forty- six patients with intractable idiopathic TN were treated between Michael Lim, M.D. January 2005 and June 2007. Eligible patients were either poor surgical candidates or had Department of Neurosurgery, failed previous microvascular decompression or destructive procedures. During a single Johns Hopkins School of Medicine, radiosurgical session, a 6-mm segment of the affected nerve was treated with a mean Baltimore, Maryland marginal prescription dose of 58.3 Gy and a mean maximal dose of 73.5 Gy. Monthly neurosurgical follow-up was performed until the patient became pain- free. Longer- term Allen Efron, M.D. follow-up was performed both in the clinic and over the telephone. Outcomes were Department of Neurosurgery, Ͼ Stanford University Medical Center, graded as excellent (pain- free and off medication), good ( 90% improvement while still Stanford, California on medication), fair (50–90% improvement), or poor (no change or worse). Facial numb- ness was assessed using the Barrow Neurological Institute Facial Numbness Scale score. Iris C. Gibbs, M.D. RESULTS: Symptoms disappeared completely in 39 patients (85%) after a mean latency Department of Radiation Oncology, of 5.2 weeks. In most of these patients, pain relief began within the first week. TN Stanford University Medical Center, Stanford, California recurred in a single patient after a pain-free interval of 7 months; all symptoms abated after a second radiosurgical procedure. Four additional patients underwent a repeat Steven D. Chang, M.D. rhizotomy after failing to respond adequately to the first operation. After a mean follow- Department of Neurosurgery, up period of 14.7 months, patient- reported outcomes were excellent in 33 patients Stanford University Medical Center, (72%), good in 11 patients (24%), and poor/no improvement in 2 patients (4%). Significant Stanford, California ipsilateral facial numbness (Grade III on the Barrow Neurological Institute Scale) was reported in 7 patients (15%). Scott G. Soltys, M.D. Department of Radiation Oncology, CONCLUSION: Optimized nonisocentric CyberKnife parameters for TN treatment Stanford University Medical Center, resulted in high rates of pain relief and a more acceptable incidence of facial numbness Stanford, California than reported previously. Longer follow-up periods will be required to establish whether or not the durability of symptom relief after lesioning an elongated segment of the Reprint requests: John R. Adler, Jr., M.D., trigeminal root is superior to isocentric radiosurgical rhizotomy. Department of Neurosurgery, KEY WORDS: CyberKnife, Nonisocentric, Radiosurgery, Rhizotomy, Trigeminal neuralgia Stanford University Medical Center, 300 Pasteur Drive, Neurosurgery 64:A84–A90, 2009 DOI: 10.1227/01.NEU.0000341631.49154.62 www.neurosurgery- online.com Stanford, CA 94305. Email: [email protected] he treatment of trigeminal neuralgia (TN) Received, June 13, 2008. ABBREVIATIONS: BNI, Barrow Neurological by radiosurgical rhizotomy was con- Accepted, October 27, 2008. Institute; CT, computed tomographic; Dmax, ceived by Lars Leksell (9). Over the past maximal dose; MRI, magnetic resonance imag- T Copyright © 2009 by the ing; TN, trigeminal neuralgia several decades, radiosurgical parameters for Congress of Neurological Surgeons treating this condition have been optimized,

A84 | VOLUME 64 | NUMBER 2 | FEBRUARY 2009 SUPPLEMENT www.neurosurgery-online.com CYBERKNIFE RHIZOTOMY FOR TRIGEMINAL NEURALGIA

especially for the gamma knife. By investigating a range of max- Stanford University Medical Center have been managed with a imal doses (Dmax), from 60 to 90 Gy (2, 8, 16), and a variety of relatively consistent radiosurgical lesion. In this study, we ana- lesion locations along the retrogasserian trigeminal root (from lyze the outcomes in this cohort of patients and compare them the root entry zone to just proximal to the semilunar ganglion) with outcomes in published radiosurgery series. (3, 5, 8, 12, 15, 16, 21), neurosurgeons are now able to maximize the likelihood of pain relief while minimizing the risk of facial PATIENTS AND METHODS dysesthesias or numbness. Recent outcome studies report that complete long- term pain relief can be achieved in 66 to 84% of Patients patients undergoing radiosurgical rhizotomy alone, whereas Patient data were reviewed under an Institutional Review Board- varying degrees of facial numbness are observed in 10 to 54% of approved protocol. All cases had idiopathic TN of several years’ dura- cases (13, 14, 16, 18). As a general rule, better pain relief appears tion; either pain symptoms were refractory to standard anticonvul- to correlate with higher rates of sensory loss (16, 19, 20). sants or the patient experienced severe drug- induced side effects. Cases Given the straightforward objective of radiosurgical rhizo- involving either multiple sclerosis or tumors were excluded from this tomy in patients with TN, it is unclear why outcomes vary so analysis, as was any patient with atypical facial pain. A total of 46 con- much among patients. The long latency to pain relief and its secutive TN patients who met these criteria underwent lesioning at poor durability in some patients are lingering shortcomings of Stanford University Medical Center by CyberKnife rhizotomy between the procedure. In contrast, immediate, long- lasting symptom January 2005 and June 2007 (Table 1). In this group were 17 men and 29 relief is routinely achieved in similar patients who undergo less women with a median age of 78 years (range, 40–94 years). The aver- expensive percutaneous radiofrequency rhizotomy (22). Perhaps age time from the earliest onset of TN symptoms to stereotactic radio- surgery was 8.6 years (range, 8 months–20 years). Eleven patients (24%) with these limitations in mind, some neurosurgeons have sought had failed previous surgical procedures (averaging 1.8 procedures per to improve the outcome after radiosurgery by increasing the patient) directed toward their trigeminal symptoms; previous proce- length of the irradiated nerve (1, 6, 11). However, this concept dures included microvascular decompression (14 patients) and rhizo- was discredited in a small randomized study by Flickinger et al. tomy by means of an open (3 patients), glycerol (5 patients), or gamma (6). In a multi- institutional study that used the gamma knife, 2 knife (3 patients) technique. Eleven patients (24%) reported either a adjacent 4-mm shots served to lengthen the segment of lesioned moderate or significant degree of facial numbness before CyberKnife trigeminal nerve. Because of a trend (albeit not statistically sig- rhizotomy, which, unfortunately, was not prospectively scored on the nificant) toward increased “complications” in the experimental Barrow Neurological Institute (BNI) Scale. group, the study was stopped short of complete patient accrual. The investigators concluded: “Increasing the treatment volume Radiosurgical Procedure to include a longer nerve length for trigeminal neuralgia radio- Radiosurgical trigeminal rhizotomy was performed with the surgery does not significantly improve pain relief but may CyberKnife radiosurgical system, using the technique reported previ- increase complications” (6, p 449). Although Flickinger et al. did not provide much qualitative detail, facial numbness and dyses- thesias were central to the early termination of this study. The TABLE 1. Patient characteristicsa outcomes of the study are surprising because they violate the Characteristic No. direct relation between target dose, target volume, and clinical Total patients 46 response that underlies much of radiosurgery. One would have anticipated that any increase in facial numbness would have Male/female 17/29 been accompanied by an improved rate and durability of pain Age (y) relief. It seems possible that chance may have contributed to the Median 78 paradoxical findings observed in the small Flickinger et al. series. Range 40–94 With the goal of enhancing the clinical response to radio- Time from TN onset to radiosurgery (y) surgery, in 2002, the first author (JRA) began to lesion a longer Mean 8.6 segment of the trigeminal sensory root using a relatively homogeneous radiation dose. Because of the capacity of the Range 8–20 CyberKnife (Accuray, Inc., Sunnyvale, CA) for nonisocentric irra- Right side/left side 29/17 diation, it is a straightforward process to conformally lesion an Prior surgical proceduresb elongated nonspherical target with greater dose uniformity, in Microvascular decompression 14 contrast to the procedure with instruments that treat only isocen- Open rhizotomy 3 trically. In the initial report detailing the results of 41 patients treated in this manner, 92.7% of patients experienced initial pain Glycerol rhizotomy 5 relief after a median of only 7 days postradiosurgery; pain relief Gamma knife 3 was maintained in 78% of patients at 11 months. However, facial Pre-CyberKnife facial numbness, no. (%) 11 (26%) numbness developed in 73% of patients (11). To address this shortcoming, dose and volume were gradually de- escalated. a TN, trigeminal neuralgia. b Since January 2005, appropriately selected TN patients at Eleven patients (24%) had failed procedures.

NEUROSURGERY VOLUME 64 | NUMBER 2 | FEBRUARY 2009 SUPPLEMENT | A85 ADLER ET AL.

ously (11). Both the CyberKnife Model G3 (400 monitor units/min) and follow- up occurred at monthly intervals to ensure a stable clinical the CyberKnife Model G4 (600 monitor units/min) were used over the status. Any relapse of pain or the development of facial numbness course of this study. To summarize, a customized Aquaplast mask was clinically re- evaluated in person. At each point of contact, (WFR/Aquaplast Corp., Wyckoff, NJ) was fashioned for each patient. patients were grouped into 1 of 4 categories: excellent (pain- free While immobilized in the mask, the entire head was imaged with thin- and off medication), good (Ͼ90% improvement while still on med- section (1.25-mm slice thickness) contrast computed tomographic (CT) ication), fair (50–90% improvement), or poor (no change or worsen- scanning; the first 15 cases in this series received a lumbar intrathecal ing pain). The presence and timing of facial numbness were also injection of iodinated contrast medium (8 mL of iohexol) before CT recorded and classified according to the BNI Facial Numbness Scale scanning. Patients were then imaged with fast imaging employing score (Table 2) (19). Five patients (11%) underwent repeat CyberKnife steady- state acquisition sequence magnetic resonance imaging (MRI) rhizotomy more than 6 months after the initial procedure. In 4 of using 3-dimensional volumetric acquisition and 1-mm slice thickness. these cases, there was inadequate clinical response to the first radio- These CT and MRI techniques reliably demonstrated the trigeminal surgical procedure, whereas in 1 patient there was a relapse in TN nerve in negative relief as it transited the prepontine cistern (4). Midway symptoms 5 months after the initial operation. Longer- term follow- through the current series, we documented the reliability and spatial up was done by mail and telephone. Follow- up data were main- fidelity of radiosurgical targeting performed with only T2-weighted tained in a prospective database approved by the Stanford Institu - MRI scanning (4). Once we were confident in the accuracy of MRI- tional Review Board. based targeting, the use of CT cisternography for this purpose was largely abandoned. On the CyberKnife treatment planning workstation, the MRI and RESULTS CT scans were fused. Standard segmentation tools were used on the fused data sets to delineate a 6-mm length of the retrogasserian Thirty- nine patients (85%) in this study had complete disap- trigeminal sensory root, as defined in the reconstructed sagittal plane pearance of their TN and were off medications at some point of the nerve. This target volume deliberately spared the proximal 2 to over the course of follow- up. At a mean follow- up duration of 3 mm of trigeminal dorsal root at the brainstem. Dose to the brain- 14.7 months (range, 6–29 months; median, 10.5 months), 34 stem and semilunar ganglia was restricted to the maximal extent pos- patients (72%) graded their pain relief as excellent, and 11 (24%) sible. Four of 5 repeat rhizotomies were performed with the same described it as good. Consequently, 96% of patients obtained treatment parameters as the original procedure; the first re- treated meaningful relief of TN symptoms after CyberKnife rhizotomy. patient received a lower dose (Dmax was only 60 Gy). The earlier Within this group were 4 patients whose TN did not improve patients in the current investigation were treated with a decreased 65- cm source- to- axis distance “trigeminal node set,” which provides an and 1 patient who, at 7 months, experienced recurrent pain. All effective smaller collimator diameter of approximately 6 mm at the of these patients were re- treated, and the eventual outcome for isocenter; the “7.5-mm” collimator of the CyberKnife is defined by these 5 salvage cases was generally excellent (2 patients) or convention at a nominal source- to- axis distance of 80 cm. After the good (2 patients). However, 1 of the 5 re- treated patients had a higher- output CyberKnife G4 model was installed at Stanford, poor outcome. Although, across the entire series, an improve- patients were treated using a 5-mm collimator and an 80-cm source- ment in symptoms was typically experienced within a few days to- axis distance. or weeks of radiosurgery, the full benefits were obtained after With increasing experience, treatment planning techniques have a mean latency of 5.2 weeks (ranging from immediately to 6 evolved that enable ever better dosimetry for CyberKnife trigeminal months after the procedure). rhizotomy. Some of the methods used in the current series include: 1) TN recurred in only a single patient, 7 months after radio- add a tuning structure to protect brainstem and ipsilateral temporal lobe; 2) angle tuning structures to “spill” dose into the ambient cistern surgery; 9 months after the first procedure, repeat CyberKnife as much as possible; 3) to maximize the resolution of dose calculation, radiosurgical rhizotomy rendered this patient pain- free. At the use the smallest calculation grid which covers the proximal brainstem, time of the last follow-up evaluation, 11 patients (24%) reported semilunar ganglion, and 7th and 8th cranial nerves; 4) slightly expand a significant improvement (3 patients) or even complete abate- the targeted nerve segment to improve dosimetric coverage; and 5) ment (8 patients) in TN symptoms but required, or elected to constrain the number of monitor units per beam to increase the total continue, anticonvulsants for controlling their pain; 3 patients number of treatment beams. refused to give up their medication only because of concern A mean marginal prescription dose of 58.3 Gy (range, 56–62 Gy) and that their pain might return. Meanwhile, after rhizotomy, TN Dmax of 73.5 Gy (range, 71.4–86.2 Gy) were used over the course of this symptoms were deemed to be worse in only 1 case. series. The average prescription isodose line was 79%, whereas the dose Among the 41 patients who had only a single radiosurgi- at the edge of the brainstem was kept to less than 50% of the Dmax. Figure 1 depicts a representative radiosurgical plan for trigeminal neu- cal rhizotomy, significant new ipsilateral facial numbness ralgia. In Figure 2, note the relatively homogeneous dose within the tar- was reported in 7 cases (17%). New- onset sensory loss typi- get volume that is achievable with a nonisocentric technique. cally involved the hemicranium and occurred an average of Radiosurgery was performed as an outpatient procedure. Treatment 7 months after the procedure. Interestingly, only 1 of 5 sessions lasted approximately 1 hour. At the end of the procedure, patients who underwent repeat radiosurgery (after a previ- patients were given 4 mg of dexamethasone. ous gamma knife or CyberKnife rhizotomy) developed sig- nificant subsequent facial numbness. At the time of the most Follow-up recent follow- up evaluation, facial sensation and dysesthesia Patients were followed at 1-month intervals until they were pain- were categorized in all patients according to BNI Scale score free, at which point anticonvulsants were tapered. Subsequent as follows: BNI I, 28 patients; BNI II, 11 patients; and BNI III,

A86 | VOLUME 64 | NUMBER 2 | FEBRUARY 2009 SUPPLEMENT www.neurosurgery-online.com CYBERKNIFE RHIZOTOMY FOR TRIGEMINAL NEURALGIA

FIGURE 1. Screen shot taken from the CyberKnife (Accuray, Inc., Sunnyvale, plan for trigeminal neuralgia and illustrating the relatively homogeneous dose CA) treatment planning workstation depicting a representative radiosurgical within the target volume that is achievable with a nonisocentric technique.

the last follow-up evaluation, there were no cases of anesthe- TABLE 2. Barrow Neurological Institute Facial Numbness Scale sia dolorosa. score and outcomes Twenty- eight patients (61%) reported the optimal result of No. of Grade Description being pain- free and off anticonvulsants, while retaining pre- patients treatment facial sensation. Six additional patients were pain- I No facial numbness 28 free but either elected to remain on reduced doses of anticonvul- II Mild facial numbness, not bothersome 11 sants or were unable to taper off such drugs completely without III Facial numbness, somewhat bothersome 7 experiencing some recurrent symptoms. Consequently, Cyber - Knife rhizotomy resulted in 74% of all patients experiencing IV Facial numbness, very bothersome 0 good to excellent pain relief without significant facial numbness. If one includes 9 additional patients who developed significant 7 patients (Table 2). Three of the 7 patients who reported facial numbness, 96% of all patients experienced complete or somewhat bothersome facial numbness treated these symp- near- complete relief of trigeminal neuralgia. It is especially note- toms with either neurontin (2 patients) or amitriptyline. It is worthy that among the 31 TN patients who had never under- important to note that 11 patients had some measure of gone any previous surgical procedure, 100% had either a good numbness before CyberKnife rhizotomy, and at the time of or excellent response to CyberKnife rhizotomy.

NEUROSURGERY VOLUME 64 | NUMBER 2 | FEBRUARY 2009 SUPPLEMENT | A87 ADLER ET AL.

the outcome with nonisocentric radiosurgical rhizotomy is at least competitive in terms of overall rate of response. Ideally, the current experience could provide a reasonable substrate for a randomized trial that compares the 2 procedures. The rate of numbness observed in the present series appears to be greater than first reported by several groups using an isocentric gamma knife technique (12, 17, 19). However, higher rates of facial numbness after gamma knife rhizotomy have also been reported, especially with larger and more effective dosing regimens (14, 16). Consequently, the overall incidence and characteristics of sensory loss experienced by some of the patients in the current CyberKnife study seem comparable to what has been reported previously with more conventional isocentric procedures. Although there are important technical differences, the rel- atively unremarkable incidence of facial sensory changes that we observed in the present series may challenge one of the key findings reported by Flickinger et al. (6). This randomized trial, which examined the usefulness of 2 adjacent 4-mm FIGURE 2. Sagittal magnetic resonance imaging reconstruction demon- gamma knife shots when lesioning the trigeminal nerve, strating relative dose homogeneity. needed to be terminated prematurely when patients in the experimental arm developed severe dysesthesias. The investi- gators concluded that a longer radiosurgical lesion merely DISCUSSION increased the likelihood of “complications” without altering the pain response. What remains unclear is why Flickinger The benefits of radiosurgical rhizotomy for TN using the stan- et al. did not encounter an increased rate of response among dard isocentric technique are, for many patients, insufficient. patients treated with a longer lesion of the trigeminal sensory Because of the latency of pain relief, overall rate of response, root. The failure to find such a response would appear to run and recurrence of TN symptoms, radiosurgical rhizotomy has counter to the experience described by our group (10, 23) and, limited appeal in the eyes of some neurosurgeons. In an attempt perhaps, the reports of other gamma knife series that found a to address the shortcomings of the standard isocentric technique, direct correlation between integral dose to the trigeminal nerve our group has sought, over several years, to refine a procedure and the likelihood of pain relief (16). Moreover, the findings of that uses nonisocentric radiosurgery to lesion an elongated seg- Flickinger et al. (6) would seem to be at odds with the widely ment of the retrogasserian portion of the trigeminal sensory root. observed dose- volume- response phenomenon seen through- Over time, treatment parameters have been optimized to out most of radiosurgery. Of note, the Flickinger et al. study enhance the pain response and reduce facial numbness. The pri- compared the standard 1-isocenter treatment with a solitary mary objective of the current prospective single- institution study experimental group in which the longer rhizotomy consisted was to characterize short- term outcomes with this technique. of 2 adjacent isocenters. By comparison, our group required The overall response rate observed in the current CyberKnife several years to slowly titrate a radiosurgical dose and volume series appears satisfactory. In fact, 100% of the patients who had for lesioning the trigeminal nerve, which is proving, in our not previously undergone a surgical procedure reported their experience, to be more optimal in terms of pain relief and outcome as either good (13%) or excellent (87%); 1 of these avoidance of facial numbness. Perhaps a wider investigation of patients needed to be treated twice to achieve this outcome. It is volumes and doses delivered with a 2-isocenter gamma knife reasonable to propose that this high rate of response is attribut- technique might reveal a combination that improves pain relief able, at least in part, to the longer length of trigeminal nerve without further risk of unpleasant facial numbness. Only fur- lesioned with the CyberKnife. ther study, and most probably a randomized trial, can deter- Although the methods used to score the response of TN to mine whether or not a longer trigeminal rhizotomy ultimately trigeminal rhizotomy are relatively consistent among most clinical has any clinical advantages. series, patient inclusion criteria, length of follow- up, and measures One of the primary rationales for modifying the standard of facial numbness and/or dysesthesia vary widely. Consequently, radiosurgical rhizotomy technique was to increase the durabil- it is difficult to directly compare those outcomes with those in the ity of pain relief. Because only 1 of 39 patients experienced a current study, and as a result, it cannot be determined at present relapse in their TN (7 months after stereotactic radiosurgery), whether or not irradiating a longer nerve segment is clinically the current experience is encouraging. Nevertheless, much beneficial. Nevertheless, crude comparisons with published radio- longer follow- up periods will be required to prove that lesion- surgical rhizotomy series that have used isocentric gamma knife ing a longer segment of nerve can significantly increase the (16, 18) and linear accelerator- based (21) procedures suggest that durability of the radiosurgical rhizotomy.

A88 | VOLUME 64 | NUMBER 2 | FEBRUARY 2009 SUPPLEMENT www.neurosurgery-online.com CYBERKNIFE RHIZOTOMY FOR TRIGEMINAL NEURALGIA

There are significant differences between the treatment plan- of facial sensation will occur in enough additional patients to ning parameters used throughout the current series and those alter this primary finding. Ultimately, it seems clear that numb- that have been reported in previous descriptions of CyberKnife ness is significantly less likely using the optimized treatment nonisocentric rhizotomy (23). In the study by Villavicencio et al. parameters of the present study than it was during the develop- (23), the length of lesioned nerve ranged from 3 to 12 mm, mental stages of CyberKnife rhizotomy for TN. whereas the marginal prescription ranged from 40 to 70 Gy The specific treatment parameters used in the current series and the Dmax ranged from 50 to 86.4 Gy. Moreover, the dose at to achieve a lower complication rate thus far do not appear to the edge of the brainstem was as high as 50% in that study. have compromised the short- term efficacy of radiosurgical rhi- These values contrast markedly with the more uniform 6-mm zotomy when a comparison is made to the Villavicencio et al. length of lesioned nerve used throughout the current investiga- series (23). The 72% rate of being pain- free and off medication tion. Meanwhile, the mean marginal prescription dose that we that is observed in the current investigation compares favor- now report ranged only between 56 and 62 Gy, and the Dmax ably with the 67% rate of excellent outcome described in the ranged between 71.4 and 86 Gy. Of note, there was only a sin- earlier study. Although the longer- term likelihood of TN recur- gle patient who received a dose of 86 Gy, and the next highest rence when both the length of nerve and dose have been opti- Dmax was 77 Gy. Finally, in the present series, we kept the mized remains unknown, the preliminary 2% rate of relapse we dose at the edge of the brainstem to less than 30% of Dmax. observe in the current study seems competitive with the 37% These sizeable differences in treatment planning parameters rate of recurrent TN noted by Villavicencio et al., albeit the relative to the series reported by Villavicencio et al. (23) would follow- up period for their study is longer, at 2 years. Clearly, seem to be responsible for the seemingly better outcome we additional follow- up is needed to compare the ultimate dura- now report. bility of these 2 procedures. Unlike the more defined and reproducible isocentric lesion It is worth pointing out that TN is somewhat unique among made with the gamma knife, treatment planning for CyberKnife pain syndromes in that pain is usually either present or absent. trigeminal rhizotomy introduces a number of novel yet highly As a consequence, the large majority of patients with complete critical treatment planning variables. As a consequence, treat- relief were easy to categorize as such. In contrast, the nature of ment planning techniques have evolved with ever greater expe- symptoms and outcome among patients with only partial pain rience, and as a result, the procedure (and dosimetry) used in relief was more complex and less conducive to the relatively the current investigation represents an improvement over what simplistic measuring tools used in this investigation. Moreover, was described in the study by Villavicencio et al. (23). Some of several of the patients in this series had coexisting facial pain the specific treatment planning tactics used in the current series syndromes that were not TN- like and were precipitated in part are described previously in Methods. In addition to designing by previous failed ablative procedures or microvascular better treatment plans, surgical experience has also been impor- decompression. Because some of these same patients also even- tant in better defining the targeted nerve segment. With imag- tually developed post- rhizotomy delayed facial numbness, it ing, the relevant anatomy can be ambiguous in some TN was often difficult to distinguish new sensory symptoms from patients, especially after a failed microvascular decompression. their preexisting pain condition. Therefore, one of the limita- Given all of the above factors, improved surgical technique may tions of the current investigation is that, unlike the recent study have been yet another factor in the somewhat different outcome by Régis et al. (18), there is not a more detailed analysis of our being reported here relative to that of Villavicencio et al. (23). patients’ ultimate facial sensation. Despite this shortcoming, The incidence and intensity of delayed numbness observed in the current investigation should provide a useful starting point the current series represents a substantial improvement over for future investigation. what was previously reported for CyberKnife rhizotomy (11, 23). The biological mechanism underlying the cessation of TN For example, the 17% rate of bothersome facial numbness (BNI symptoms after radiosurgical rhizotomy remains poorly III) that we describe in the current report is a significantly lower defined. Some investigators have concluded that the likelihood rate than the 28% observed by Villavicencio et al. (23), which also of a clinical response to radiosurgery correlates with the devel- included BNI scores of III (16%) and IV (12%). In addition, none opment of facial numbness (16, 19). This suggests that a com- of the patients in the current series developed other, even more mon mechanism underlies these 2 phenomena. However, their serious complications such as anesthesia dolorosa, trismus, mas- very different time course implies that distinct mechanisms ticator weakness, diplopia, decreased hearing, dry eye syndrome, underlie pain relief and facial numbness; TN pain abates rela- or paresis, all of which were described by Villavicencio et al. As tively rapidly after a radiosurgical rhizotomy, whereas facial a consequence, it appears that the dose- volume parameters being numbness is clearly a delayed event (6–9 months or more after used with the present nonisocentric technique make for a better radiosurgery). Meanwhile, both the latency of sensory changes clinical outcome. Given the reported latency between radio- and the appearance of enhancement on contrast MRI at the surgery and the onset of facial numbness, it is quite possible that, root entry zone, the latter being consistent with breakdown in with further follow- up, additional patients will develop sensory the blood- brain barrier, strongly suggest that facial numbness symptoms. However, because the mean overall follow- up dura- stems from injury to the microvasculature of the trigeminal tion is 14.7 months in the present study, whereas the mean time nerve and adjacent brainstem (7). If so, such a phenomenon to occurrence of numbness is 7 months, it seems unlikely that loss raises some intriguing possibilities for intervention. In particu-

NEUROSURGERY VOLUME 64 | NUMBER 2 | FEBRUARY 2009 SUPPLEMENT | A89 ADLER ET AL.

lar, the use of an intravascular free radical scavenger, like ami- improve trigeminal neuralgia radiosurgery? A prospective double- blind, ran- fostine (Ethyol; Medimmune, Gaithersburg, MD), at the time of domized study. Int J Radiat Oncol Biol Phys 51:449–454, 2001. radiosurgery might provide relative radioprotection to the 7. Gorgulho AA, De Salles AA: Impact of radiosurgery on the surgical treatment of trigeminal neuralgia. Surg Neurol 66:350–356, 2006. endothelium and limit the incidence of delayed microvascular 8. Kondziolka D, Lunsford LD, Flickinger JC, Young RF, Vermeulen S, Duma injury. Theoretically, such an agent would not alter the biolog- CM, Jacques DB, Rand RW, Regis J, Peragut JC, Manera L, Epstein MH, ical mechanism underlying the therapeutic benefits of radio- Lindquist C: Stereotactic radiosurgery for trigeminal neuralgia: A multiinsti- surgery. A randomized trial is being planned to investigate tutional study using the gamma unit. J Neurosurg 84:940–945, 1996. 9. Leksell L: Stereotaxic radiosurgery in trigeminal neuralgia. Acta Chir Scand such a possibility. 137:311–314, 1971. 10. Lim M, Cotrutz C, Romanelli P, Schaal D, Gibbs I, Chang SD, Adler JR: Stereotactic radiosurgery using CT cisternography and non- isocentric planning CONCLUSION for the treatment of trigeminal neuralgia. Comput Aided Surg 11:11–20, 2006. 11. Lim M, Villavicencio AT, Burneikiene S, Chang SD, Romanelli P, McNeely L, Compared with previous reports that describe the clinical McIntyre M, Thramann JJ, Adler JR: CyberKnife radiosurgery for idiopathic outcome after nonisocentric radiosurgical rhizotomy in patients trigeminal neuralgia. Neurosurg Focus 18:E9, 2005. with TN, the present investigation demonstrates both high 12. Maesawa S, Salame C, Flickinger JC, Pirris S, Kondziolka D, Lunsford LD: Clinical outcomes after stereotactic radiosurgery for idiopathic trigeminal rates of pain relief and an acceptable incidence of facial numb- neuralgia. J Neurosurg 94:14–20, 2001. ness. We are not able to determine whether or not this tech- 13. Massager N, Lorenzoni J, Devriendt D, Levivier M: Radiosurgery for trigem- nique can ultimately improve on either the rate of response or inal neuralgia. Prog Neurol Surg 20:235–243, 2007. the durability of symptom relief that can be achieved with the 14. Massager N, Murata N, Tamura M, Devriendt D, Levivier M, Régis J: Influence of nerve radiation dose in the incidence of trigeminal dysfunction standard isocentric radiosurgical rhizotomy. Further study is after trigeminal neuralgia radiosurgery. Neurosurgery 60:681–688, 2007. warranted, and, ideally, a randomized trial comparing isocen- 15. Nicol B, Regine WF, Courtney C, Meigooni A, Sanders M, Young B: Gamma tric and nonisocentric techniques should be conducted. knife radiosurgery using 90 Gy for trigeminal neuralgia. J Neurosurg 93 [Suppl 3]:152–154, 2000. 16. Pollock BE, Phuong LK, Foote RL, Stafford SL, Gorman DA: High- dose Disclosure trigeminal neuralgia radiosurgery associated with increased risk of trigemi- Funding for this investigation came from the Victoria and Gary Reed Fund for nal nerve dysfunction. Neurosurgery 49:58–64, 2001. Radiosurgical Research. John R. Adler, Jr., M.D., is a shareholder and a member 17. Régis J, Metellus P, Dufour H, Roche PH, Muracciole X, Pellet W, Grisoli F, of the board of directors of Accuray, Inc., the manufacturer of the CyberKnife Peragut JC: Long- term outcome after gamma knife surgery for secondary radiosurgical system. The other authors have no personal financial or institu- trigeminal neuralgia. J Neurosurg 95:199–205, 2001. tional interest in any of the drugs, materials, or devices described in this article. 18. Régis J, Metellus P, Hayashi M, Roussel P, Donnet A, Bille- Turc F: Prospective controlled trial of gamma knife surgery for essential trigeminal neuralgia. J Neurosurg 104:913–924, 2006. REFERENCES 19. Rogers CL, Shetter AG, Fiedler JA, Smith KA, Han PP, Speiser BL: Gamma knife radiosurgery for trigeminal neuralgia: The initial experience of The Barrow 1. Alpert TE, Chung CT, Mitchell LT, Hodge CJ, Montgomery CT, Bogart JA, Neurological Institute. Int J Radiat Oncol Biol Phys 47:1013–1019, 2000. Kim DY, Bassano DA, Hahn SS: Gamma knife surgery for trigeminal neural- 20. Shetter AG, Rogers CL, Ponce F, Fiedler JA, Smith K, Speiser BL: Gamma gia: Improved initial response with two isocenters and increasing dose. knife radiosurgery for recurrent trigeminal neuralgia. J Neurosurg 97 [Suppl J Neurosurg 102 [Suppl]:185–188, 2005. 5]:536–538, 2002. 2. Brisman R: Gamma knife surgery with a dose of 75 to 76.8 Gray for trigemi- 21. Smith ZA, De Salles AA, Frighetto L, Goss B, Lee SP, Selch M, Wallace RE, nal neuralgia. J Neurosurg 100:848–854, 2004. Cabatan- Awang C, Solberg T: Dedicated linear accelerator radiosurgery for 3. Brisman R, Mooij R: Gamma knife radiosurgery for trigeminal neuralgia: the treatment of trigeminal neuralgia. J Neurosurg 99:511–516, 2003. 22. Taha JM, Tew JM Jr, Buncher CR: A prospective 15-year follow up of 154 con- Dose- volume histograms of the brainstem and trigeminal nerve. J Neurosurg secutive patients with trigeminal neuralgia treated by percutaneous stereotac- 93 [Suppl 3]:155–158, 2000. tic radiofrequency thermal rhizotomy. J Neurosurg 83:989–993, 1995. 4. Chávez GD, De Salles AA, Solberg TD, Pedroso A, Espinoza D, Villablanca P: 23. Villavicencio AT, Lim M, Burneikiene S, Romanelli P, Adler JR, McNeely L, Three- dimensional fast imaging employing steady- state acquisition magnetic Chang SD, Fariselli L, McIntyre M, Bower R, Broggi G, Thramann JJ: resonance imaging for stereotactic radiosurgery of trigeminal neuralgia. CyberKnife radiosurgery for trigeminal neuralgia treatment: A preliminary Neurosurgery 56:E628, 2005. multicenter experience. Neurosurgery 62:647–655, 2008. 5. Cheuk AV, Chin LS, Petit JH, Herman JM, Fang HB, Regine WF: Gamma knife surgery for trigeminal neuralgia: Outcome, imaging, and brainstem correlates. Int J Radiat Oncol Biol Phys 60:537–541, 2004. Acknowledgment 6. Flickinger JC, Pollock BE, Kondziolka D, Phuong LK, Foote RL, Stafford SL, We acknowledge the helpful suggestions and diligent edits made by David Lunsford LD: Does increased nerve length within the treatment volume Schaal, Ph.D., during the preparation of this manuscript.

A90 | VOLUME 64 | NUMBER 2 | FEBRUARY 2009 SUPPLEMENT www.neurosurgery-online.com