3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 73

Part II

Primary Central Nervous System Tumors 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 74 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 75

3

Primary Cerebral Tumors

MITCHEL S. BERGER, STEVEN A. LEIBEL, JANET M. BRUNER, JONATHAN L. FINLAY, AND VICTOR A. LEVIN

The intracranial supratentorial compartment is the General symptoms are headache, gastrointestinal most common site for central nervous system (CNS) upset such as nausea and vomiting, personality tumors. Understanding tumors that arise within the changes, and slowing of psychomotor function. Be- supratentorial brain is, therefore, of the utmost im- cause the brain parenchyma does not have pain- portance. This chapter provides the reader with a rea- sensitive structures, headache has been attributed to sonably thorough understanding of (1) the symptoms local swelling and distortion of pain-sensitive nerve and signs of tumors that arise within the various endings associated with blood vessels, primarily in supratentorial locations; (2) the various approaches the meninges. Many tumors grow without headache to surgical intervention, from biopsy to cortical map- as a prominent symptom, but others rapidly lead to ping, for tumors in eloquent sites of the brain; headache either because of the tumor’s proximity to (3) the pathologic classification and grading schemes pain-sensitive fibers or due to its rapid growth and used around the world for the various supratentorial the achievement of a critical volume that causes com- tumors; (4) conventional as well as other radiation pression and displacement of brain. Under the lat- approaches such as brachytherapy, radiosurgery, ter circumstances, the onset and disappearance of three-dimensional conformal radiotherapy, and pro- headache correlate with changes in intracranial ton beam treatment for the tumors; and (5) the pressure. Headaches can vary in severity and qual- progress and limitations of chemotherapy used with ity; frequently they occur in the early morning hours radiation and as independent therapy. In addition, or upon first awakening. Sometimes patients com- some major and subtle differences between children plain of an uncomfortable feeling in the head rather and adult patients with primary brain tumors are than headache. noted. Therapeutic approaches to tumors in infants Gastrointestinal symptoms such as loss of appetite, are not covered in this chapter as they are rare and queasiness, nausea, and occasionally vomiting can too specialized for this text. occur in all patients but are most common in chil- dren and in patients harboring infratentorial rather than supratentorial tumors. Changes in personality, SYMPTOMS AND SIGNS mood, mental capacity, and concentration can be noted early or can be the only abnormalities ob- Neurologic symptoms and physical signs reflect the served. In general, patients with brain tumors tend to location of tumor rather than tumor histology. Symp- sleep longer at night and nap during the day. These toms are at two levels: general symptoms, associated symptoms are not unique to individuals with brain with increased intracranial pressure; and focal symp- tumors and can easily be confused with depression, toms, related to tumor location. neurasthenia, and other psychological problems.

75 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 76

76 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Focal symptoms can be episodic (seizures) or pro- the common occurrence of bilateral disease in frontal gressive. Seizures are important harbingers of brain lobe tumors that readily cross through the corpus cal- tumors; although only 10% of patients presenting with losum. However, in some cases, use of opposed lat- seizures are diagnosed with a brain tumor, the asso- eral field radiation portals can lead to bilateral tem- ciation increases with increasing patient age. Seizures poral lobe damage that can be devastating for the are a presenting symptom in approximately 20% of patient as it produces impairment of recent memory patients with supratentorial brain tumors. Rapidly and can lead to dementia. growing, infiltrative malignant gliomas are likely to Parietal lobe tumors affect sensory and perceptual produce complex partial motor or sensory seizures, functions more than motor functions, although mild although generalized grand mal seizures are also hemiparesis is sometimes seen with extensive pari- common. In patients with slowly growing astrocy- etal lobe tumors. Abnormalities may range from mild, tomas, gangliogliomas, and oligodendrogliomas, gen- and observable only by formal testing, to severe sen- eralized seizures may antedate the clinical diagnosis sory loss leading to hemianesthesia and/or other by months to years. The etiology of seizures associ- hemisensory abnormalities. In addition to homony- ated with these tumors is unclear, although experi- mous hemianopsia or visual inattention, involvement mental evidence implicates a depletion of gamma of the nondominant parietal lobe can lead to per- aminobutyric acid (GABA) and somatostatin- ceptual abnormalities, anosognosia, and an apraxia immunoreactive neurons in the adjacent, nontumor- for dressing oneself; dominant parietal lobe tumors infiltrated epileptogenic brain (Haglund et al., 1992). lead to alexia, dysgraphia, and other types of apraxia. Focal seizures that occur in patients older than 40 Occipital lobe tumors can produce contralateral years of age are indicative of a brain tumor until homonymous hemianopsia or visual aberrations that proven otherwise. take the form of imperception of color, object size, The distribution of infiltrative parenchymal tumors or object location. Bilateral occipital damage rarely in the brain has a direct relationship to the mass of occurs as a result of tumor invasion, but it can be the affected lobe or region. The most frequently in- produced in herniation syndromes and can lead to volved locations in the cerebrum are, in descending cortical blindness. order of frequency, frontal, parietal, temporal, and Thalamic and basal ganglia tumors can reach 3 to occipital lobes. Clinical patterns of tumor growth in 4 cm in diameter before the patient experiences the various brain locations are less stereotypic than symptoms, which can be nonspecific headaches re- those observed after strokes; nonetheless, under- sulting from hydrocephalus and increased intracra- standing the nature of the various syndromes that nial pressure secondary to trapping of the lateral horn present will help clinicians to better understand the of one of the ventricles. Patients can also present with effect of tumor growth in the CNS. contralateral sensory abnormalities detected only by Frontal lobe tumors can be asymptomatic or can testing for sensory extinction or, rarely, with a severe produce mild slowing of contralateral hand move- neuropathic pain syndrome. Some patients complain ments, contralateral spastic hemiplegia, marked ele- of intermittent paresthesias on the contralateral side; vation in mood or loss of initiative, and dysphasia (if these are at times so episodic that anticonvulsant the involved lobe is the dominant lobe). Bifrontal dis- drugs are prescribed. Contralateral intention tremor ease is, unfortunately, all too common and can cause and hemiballistic-like movement disorders are un- bilateral hemiparesis, spastic bulbar palsy, severe im- common. pairment of intellect, lability of mood, and dementia. Temporal lobe tumors can be clinically silent or can produce impairment of recent memory, homon- GENERAL SURGICAL METHODS ymous quadrantanopsia, auditory hallucinations, and AND TECHNIQUES even aggressive behavior. Involvement of the non- dominant temporal lobe can lead to minor percep- Biopsy Techniques tual problems and spatial disorientation. Dominant temporal lobe involvement can lead to dysnomia, im- A simple and readily available method that can be paired perception of verbal commands, and even full- used to obtain a biopsy specimen is with computer blown, fluent Wernicke-like aphasia. Bilateral disease tomography (CT) or magnetic resonance imaging involving both temporal lobes is rare compared with (MRI) guidance. After a localizing CT or MRI scan is 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 77

Primary Cerebral Tumors 77

done, a small twist drill opening is made in the skull sue, regardless of the type of frame used, and this and a biopsy needle is placed into it and imaged to method is the standard against which all other meth- ensure correct placement; a tumor sample is then ods of obtaining small tissue samples should be com- taken and a repeat scan performed to make sure no pared. hemorrhage has occurred. A comparison of the free- hand method with one using a stereotactic appara- Stereotactic-Guided tus, found no significant difference in morbidity and Volumetric Resections mortality between the two procedures (Wen et al., 1993). Kelly and colleagues (1982, 1983) developed an in- An alternative biopsy approach involves the use of novative technique that coupled imaging with com- ultrasonography via a burr hole (Berger, 1986; Enz- puter-assisted stereotactic resection of tumors. Re- mann et al., 1984; Tsutsumi et al., 1989). This has constructed tumor sections based on CT and MRI been performed very accurately on lesions greater results are displayed to the surgeon on a video ter- than 7 to 10 mm and provides immediate feedback minal during surgery, allowing for precise, stereo- after the procedure to ensure that a hemorrhage has tactic laser vaporization of any intracranial target. It not occurred. In a study comparing this method to became apparent early in their experience that this CT-guided stereotactic techniques, the diagnostic approach was more beneficial for patients with cir- yield rate was found to be comparable for CT-guided cumscribed lesions (i.e., metastasis, pilocytic astro- and ultrasound-guided approaches (94% versus cytoma, and so forth) than for those with more infil- 91%) (Di Lorenzo et al., 1991). Both methods re- trative glial tumors (Kelly et al., 1986). The morbidity sulted in a similar number of complications. The ul- associated with this procedure is acceptable; how- trasound biopsy method, however, had a shorter op- ever, the outcome for malignant glial tumors was dis- erative time and was significantly less costly. appointing. Survival data were not significantly dif- With the advent of CT-coupled stereotactic frames ferent compared with those from conventional radical in the late 1970s and early 1980s, surgeons had the resections (Kelly, 1988). capability of obtaining tissue specimens with mil- The limitations of all surgical approaches have limeter accuracy. The components of the most com- more to do with the infiltrative nature of gliomas than monly used apparatus, the Brown-Roberts-Wells with surgical technique. Stereotactic biopsy samples frame, included a base ring that was fixed to the skull from brain adjacent to the contrast-enhanced and hy- and a localizing ring with nine graphite rods to allow podense areas in high- and low-grade gliomas dem- lesions to be referenced in three dimensions (Heil- onstrated isolated tumor cells well beyond the bulk brun et al., 1983). An arc guidance system fit into the of the image-defined tumor mass (Kelly et al., 1987; base ring allows any two points in three-dimensional Kelly, 1993). Extending the resection into these ar- space to be traversed (e.g., entry and target points). eas without consideration for functional white matter This represented a tremendous step forward in neu- tracts will result in unacceptable morbidity, empha- rosurgical instrumentation and precision. Experience sizing the need to consider in the overall treatment has shown that tumor biopsy accuracy has been great- plan the infiltrated brain adjacent to the main tumor est when directed to the tumor center and the im- nidus. mediately surrounding contrast-enhancing tissue. The diagnostic accuracy for grade and type of lesion may Frameless Navigational be enhanced by including cytologic squash prepara- Resection Devices tions with the histology (Cappabianca et al., 1991). Comparing histologic findings from lesions that The next generation of image-based computerized lo- were biopsied and subsequently resected, the dis- calization for tumor resection will not use frames at- crepancy rate was greatest with astrocytic tumors in tached to the skull for three-dimensional reference terms of their grade (Chandrasoma et al., 1989). This of an intracranial target. The primary components of is clearly a problem when small samples are obtained, any contemporary navigational system are registering especially when differentiating between low-grade the surgical target with respect to surrounding struc- gliomas and reactive gliosis (Taratuto et al., 1991). tures and physical space, interacting with a localiza- Notwithstanding that, the stereotactic biopsy tech- tion device, integrating real-time data, and inter- nique is the most accurate method for obtaining tis- facing with a computer. Contemporary frameless 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 78

78 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

navigation systems include ultrasonic digitizer sys- mal brain) and color (slightly paler than white mat- tems, magnetic field digitizers, multijointed encoder ter). Most malignant gliomas have a very soft and arms, infrared flash systems, and robotic systems often necrotic grayish appearance; characteristic (Zakhary et al., 1999). The majority of these newer thrombosed veins are almost always seen with systems use a localizing arm that initializes and cali- glioblastoma. These tumors are usually highly vascu- brates fiducial markers that are attached to the pa- lar with vascularization corresponding to the con- tient’s head during the preoperative scan and at the trast-enhancing rim on the imaging studies. time of surgery. Changes in the position of the local- Following the opening of the dura, the cortex is in- izing arm are updated using acoustic transit times be- spected for expanded gyri and red “arterialized” tween the sound sources and the fiducial markers veins, which are pathognomonic for malignant (Roberts et al., 1986; Barnett et al., 1993); other sys- gliomas secondary to reduced oxygen extraction with tems use mechanical sensors (Watanabe et al., 1987) increased blood flow (Fig. 3–1). Before the resec- or light emitters. tion begins, it is helpful to use ultrasound localiza- Regardless of the system employed, surgeons will tion to determine the overall tumor size, depth, and ultimately use this technology to preoperatively plan underlying cystic structures. Tumor volumes, as seen incisions and bone flaps as well as to guide the ini- on both CT and MRI scans, closely correspond to tial phases of the resection. Shifting of the brain con- high- and low-grade gliomas that have been previ- tents will necessarily limit the utility of these methods ously unresected and untreated (LeRoux et al., 1989, when intra-axial tumors are resected because local- 1993). Once these are operated on and radiated, glio- ization is based on findings from the preoperative sis accumulates and increases the echogenic back- scan. This will not, however, be a factor during com- ground, which tends to overestimate the true size of plex skull base surgery. the tumor. Intraoperative verification of tumor and of the tran- sition zone between the tumor and the adjacent tu- Intraoperative Imaging Techniques mor-infiltrated brain is best achieved with serial Unlike frameless and frame-based systems that are frozen sections or smear preparations (Reyes et al., limited by their reliance on preoperative imaging, 1991). This is often a very time-consuming process both intraoperative CT and MR scanning provide in- and is complicated by the need to distinguish reac- traoperative updates of data sets for navigational sys- tive astrocytes from infiltrating tumor cells. Alterna- tems. Intraoperative re-registration of target anatomy tive ways to intraoperatively document the extent of eliminates the problem of brain shift that may be tumor removal involve imaging with dedicated CT and caused by resection or brain retractors and allows MRI, laser activation of hematoporphyrin (Perria et the surgeon to more precisely achieve resection con- al., 1988), fluorescent dyes (Poon et al., 1992), in- trol and to modify the preplanned surgical approach, travenous (IV) indocyanine green (Hansen et al., if necessary. Use of intraoperative MR requires MR- 1993), and IV fluorescein with ultraviolet photoacti- compatible instruments (i.e., titanium or ceramic) to vation (Moore, 1947). minimize artifact. Surgical instruments can be tracked with the use of light-emitting diode sensors to pro- Functional Mapping-Guided vide image guidance during movements and interac- Tumor Resection tive feedback on corresponding images (Black et al., 1997; Tronnier et al., 1997; Steinmeier et al., 1998). As radical resective surgery becomes more com- monplace in modern neurosurgical practice and teaching, the risks of functional morbidity will cer- Intraoperative Localization of tainly be the rate-limiting factor for most surgeons. Tumor and Margins Over the past several years, functional mapping of For neurosurgeons, gross visualization for consis- cortical and subcortical regions using both intraop- tency and color has been the sine qua non used to erative and extraoperative techniques has become an distinguish normal brain from tumor at the time of indispensable adjunct to avoid morbidity while per- surgery. Low-grade glial tumors differ from their ma- forming wide, radical tumor resections in eloquent lignant counterparts in both texture (firmer than nor- brain areas. For example, in the dominant cerebral 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 79

Primary Cerebral Tumors 79

Figure 3–1. Intraoperative photograph demonstrating swollen gyri infiltrated with a glioblastoma multiforme. Tumor margins are depicted as A, D, and C. A large, arterialized vein is seen draining the tumor.

hemisphere, language testing for reading, speech, and Because seizures associated with low-grade naming must be done before tumors that involve the gliomas are often medically refractory, neurosurgical posterior frontal, anterior parietal, and temporal approaches have been adapted from epilepsy surgery lobes are resected. Preoperative deficits in these lan- to provide better seizure control. Intraoperative map- guage functions could be due either to swelling or to ping of seizure foci using electrocorticography will tumor infiltration into essential language sites; a pre- readily identify epileptogenic areas and, combined operative trial of high-dose dexamethasone will usu- with the functional brain map, will allow for removal ally distinguish between the two causes. The motor of these regions without associated deficits. cortex is located within 3 to 5 cm behind the coro- Not all patients will be good candidates for func- nal suture superiorly and a similar distance posterior tional mapping resections. Preoperative planning be- to the outer border of the sphenoid wing. The region gins with a thorough neurologic assessment to decide is flanked posteriorly by the primary somatosensory whether the patient is a candidate for mapping. If a cortex and near the vertex by the supplementary mo- dense hemiparesis is present despite administration tor area anteriorly. The presence of any tumor on ei- of steroids, it is unlikely that intraoperative stimula- ther side of the motor cortex should dictate that stim- tion will elicit motor responses. In that setting, so- ulation mapping be done to identify the cortical motor matosensory evoked responses may be used to iden- neurons and also their descending motor tracts in the tify the central sulcus by documenting a phase subcortical white matter. These include the corona reversal potential across this sulcus (Woolsey et al., radiata, internal capsule, cerebral peduncles, and the 1979). Because the motor cortex in young children corticospinal pathways to the brain stem and spinal is often unexcitable by direct stimulation mapping, cord. Tumors involving the insula, thalamus, and ba- evoked potentials should also be available for this sal ganglia often abut the descending motor tracts, particular patient population (Goldring and Gregorie, which may be readily stimulated. 1984). Language function is assessed by having the 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 80

80 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Figure 3–2. T2-weighted axial image with arrows pointing to the precentral sulcus immediately anterior to the motor cortex.

patient count to 10 and stick out his or her tongue, the insular triangle, respectively, which mark the which verifies that Broca’s area and the inferior face combined sensorimotor (i.e., Rolandic) cortex. motor cortex are intact. The patient is then shown a With the patient awake and the cortex exposed, the series of picture slides with common objects to name. sensory and motor cortex is easily stimulated using A baseline naming error rate greater than 25% will currents as low as 2 mA and usually not greater than prevent functional mapping from providing reliable 6 mA. The current is produced with a constant cur- information. rent generator, which elicits a train of biphasic square The MRI scan is used to identify the motor cortex wave pulses (frequency, 60 Hz; duration per phase, by localizing two mirror image lines on both sides of 1.25 msec) via a bipolar electrode. Patients who are the midline that represent the central sulcus (Berger, asleep will require higher currents (i.e., 6 to 16 mA 1990; Berger et al., 1990b). This is best seen on the maximum). Using multichannel electromyographic high T2-weighted axial images (Fig. 3–2). Sagittal and recordings in addition to visual observation of motor far-sagittal scans may be used to identify the marginal activity results in greater sensitivity, allowing the use sulcus and an imaginary line drawn from the back of of lower stimulation levels and facilitating detection 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 81

Primary Cerebral Tumors 81

Figure 3–3. T2-weighted axial image demonstrating a hyperintense glioma infiltrating the face motor cortex.

of stimulation-induced seizure activity (Yingling et al., cortex was mapped to evoke orofacial movements in 1999). A current greater than 16 mA has never been addition to finger and wrist flexion and extension (Fig. necessary to evoke sensory or motor responses. At 3–4). The face motor cortex was resected in its en- this point, cold Ringer’s lactate solution should be tirety, and the tumor was removed until subcortical immediately available for irrigation of the stimulated stimulation demonstrated hand movements (Figs. cortex if a focal motor seizure develops. The best 3–5 and 3–6). The patient had a left-sided facial management of intraoperative stimulation-induced droop, which cleared in 3 weeks, and brief hand focal motor seizures is rapid cortical irrigation at the weakness, which lasted a few days. stimulation site with ice-cold Ringer’s solution, which Following localization of Broca’s area based on will abruptly stop the seizure activity originating from stimulation-induced counting arrest, naming is tested the irritated cortex without using short-acting barbi- as a language measure that best predicts postopera- turates (Sartorius and Berger, 1998). The current tive deficits (Penfield and Roberts, 1959). Before test- should be elevated in 1 and 2 mA increments for ing it is essential to determine the optimal stimula- awake and asleep patients, respectively. When oper- tion current based on recording after discharge ating near the vertex, the leg motor cortex will be hid- potentials from the cortex following bipolar stimula- den along the falx. Thus, a strip electrode should be tion. This is done to ensure that subclinical seizure inserted between the midline cortex and the falx to activity is not the cause of speech dysfunction during evoke stimulation-induced responses of the leg and the mapping. The current to be used will vary be- foot. The same or a slightly higher current may be tween 2 and 8 mA and should be adjusted to 1 mA used to stimulate subcortical motor tracts without below the current that causes after-discharge poten- concern for current spread, which remains limited to tials. A wide surface area is tested after sterile num- within 2 to 4 mm of the bipolar electrode contacts. bered tickets are placed on the cortex for documen- Figure 3–3 demonstrates the utility of mapping the tation purposes. Approximately 15 to 25 sites are cortex and the underlying white matter in a patient tested, with each site being stimulated at least three with an infiltrative astrocytic glioma involving the face times. Errors in naming take the form of hesitation motor cortex in the nondominant hemisphere. The or complete anomia. Hesitation in naming is not con- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 82

82 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Figure 3–4. Intraoperative photograph following stimulation mapping of the face (2, 3, 4) and hand (1) motor cortex. Tumor margins are A–D.

sidered critical, whereas difficulty with naming is a guage sites could be found. A companion study to this critical function and denotes an essential language evaluation of temporal lobe language in patients with site. gliomas in the same region showed a number of find- In a series of 117 patients who had language test- ings that again emphasized that language localization ing when undergoing operation on the left, dominant cannot be determined anatomically but must rely on hemisphere, essential language sites were randomly stimulation mapping (Haglund et al., 1993). In that identified throughout that hemisphere, with the heav- study, we failed to identify a language function in the iest concentration of sites located around the peri- inferior temporal gyrus, whereas the superior tem- sylvian cortex (Ojemann et al., 1989). At least 67% poral gyrus was nearly twice as likely as the middle had more than one essential language site, and in temporal gyrus to have stimulation-induced language 16% of the population tested no temporoparietal lan- errors. Contrary to accepted neurosurgical teaching, 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 83

Figure 3–5. Subcortical hand motor fibers (6, 7) are seen following resection of the overlying cortex. A indicates a tumor margin.

Figure 3–6. Postoperative T1-weighted axial image showing the resection cavity.

83 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 84

84 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

essential naming sites were found in the anterior tem- vere dysnomia, which cleared within 4 to 6 weeks fol- poral lobe (i.e., first 3 cm) in nearly 15% of patients lowing surgery, and his neurologic functions returned with temporal lobe gliomas. It was also learned from to baseline. this study that resecting the tumor within 7 mm of an essential language site results in a permanent nam- ing deficit 40% of the time. However, if the resection PATHOLOGY OF DIFFUSE is further than 1 cm from the essential naming cor- INFILTRATIVE GLIOMAS tex, no permanent deficits will result. Knowing this, we are now able to avoid permanent morbidity while The most common tumors of the cranial–spinal axis removing tumors from the dominant cerebral hemi- are those that infiltrate or displace the brain paren- sphere when language is mapped before resection. chyma of the intracranial supratentorial compart- An example of the utility of speech mapping dur- ment. Of these tumors, histologically, the most com- ing tumor resection is shown in Figure 3–7. In this mon belong to the glioma family of tumors. case, the patient had a brief history of seizure activ- ity accompanied by postictal confusion and naming General Features errors, which resolved after each seizure. A MRI scan revealed a non–contrast-enhancing tumor in the pos- The astrocytic gliomas, the most common class of terior parietal lobe on the left side. Intraoperative supratentorial glial neoplasms, are derived from and speech mapping demonstrated repetitive errors in have an appearance simulating normal astrocytes. As- naming at three essential sites between 1 and 3 cm trocytes occur most commonly in the white matter, in front of the tumor nidus (Fig. 3–8). The lesion was but are also found in the cerebral cortex. Astrocy- resected completely using intraoperative ultrasound tomas are diffusely infiltrating neoplasms of the cere- guidance, and the resection cavity did not come closer bral hemispheres, and there is evidence that their than 1 cm to the nearest essential language site (Figs. prognosis and survival depend on the tumor’s histo- 3–9 and 3–10). Postoperatively, the patient had a se- logic grade (Burger et al., 1985; Fulling and Nelson,

Figure 3–7. T2-weighted axial image showing a hyperintense glioma involving the angular and supramarginal gyri of the domi- nant hemisphere. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 85

Primary Cerebral Tumors 85

Figure 3–8. Intraoperative photograph of the tumor and adjacent brain following awake stimulation mapping of the cortex. Repet- itive errors in naming were documented in numbers 33, 34, and 35. Numbers 24 and 25 overlie the center of the tumor.

1984; Kim et al., 1991). All grades of diffuse infiltra- Astrocytic neoplasms exhibit a stereotypic range of tive cerebral astrocytoma should be considered ma- features that become more prominent with increas- lignant, even if only by virtue of their location within ing grades of anaplastic malignancy. In general, the the enclosed bony skull, because almost none of these features we use to grade tumor malignancy are those tumors is cured by surgical excision alone. that permit us to diagnose the tumors. These features Astrocytes are supporting glial cells and generally include degree of cellularity, nuclear and cytoplasmic have small round nuclei with inconspicuous mi- pleomorphism of individual cells, mitotic activity, vas- cronucleoli. Under normal circumstances in routinely cular changes such as small vessel proliferation or stained histologic sections, astrocytes do not have vis- vascular mural cell proliferation, and tumor necro- ible cytoplasm, but ultrastructural studies reveal short sis. None of these features in isolation is specifically cytoplasmic processes containing intermediate fila- diagnostic of malignancy, but the combined spectrum ments. These filaments are known, through im- is used for both diagnosis and tumor grading. Other munohistochemical studies, to be composed of glial histologic features that suggest a diagnosis of tumor fibrillary acidic protein, a specific type of astrocytic and tend to suggest increasing malignancy, but are protein (Fig. 3–11). Tumors can be confirmed as as- not statistically significant in studies of histologic trocytomas by specific immunostaining of tumor cells grading, include tumor invasion of the cerebral cor- for glial fibrillary acidic protein. Astrocytes can react tex, microcystic areas (indirectly correlated with to brain injury by increasing the amount of their cy- grade), presence of tumor gemistocytes, and pial or toplasm and the complement of filaments filling it. subpial invasion by tumor cells. Their cytoplasmic processes become elongated and Most astrocytomas are derived from fibrillary as- thickened, their cell bodies become enlarged, and trocytes of white matter and arise in the subcortical their cytoplasm is then visible in routine sections. white matter. Low-grade astrocytomas show an in- These enlarged, reactive astrocytes are called gemis- creased cellularity over normal white matter, and the tocytes. cell nuclei are slightly enlarged. No mitotic activity is 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 86

Figure 3–9. Post-resection photograph with essential language sites preserved.

Figure 3–10. Postoperative T1-weighted sagittal images of the resection cavity.

86 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 87

Primary Cerebral Tumors 87

Figure 3–11. Reactive supporting astrocytes. Supporting reactive astrocytes have voluminous cytoplasm filled with dark-staining immunoreactive filaments of glial fibrillary acidic protein. They send foot processes to surround blood vessels (V) and form part of the blood–brain barrier. Immunoperoxidase for glial fibrillary acidic protein with aminoethylcarbazole-hematoxylin. Original magnification, 200.

seen, vessels are not abnormal, and necrosis is not toma (Krouwer et al., 1991), although designation as identified. Small nucleoli may become more promi- a gemistocytic astrocytoma requires that approxi- nent, but the nuclei themselves are regular in outline mately 60% of the tumor cells be gemistocytes. without much pleomorphism. Microcysts may be a The distinguishing histologic features of glioblas- distinctive feature and may contain proteinaceous toma, the highest malignant grade of astrocytoma, are fluid. Astrocytomas derived from the protoplasmic as- the presence of vascular mural cell proliferation or trocytes of the cerebral cortex are relatively filament- coagulative tumor necrosis (Burger and Green, 1987; poor and arise within the cortex itself. These tumor Kleihues and Cavenee, 2000). Necrosis may be ac- cells have short, delicate processes and poor im- companied by pseudopalisading of neoplastic cells munoreactivity for glial fibrillary acidic protein. around it, but evidence of palisading is not required As cellularity and pleomorphism, either nuclear or by most neuropathologists to make the diagnosis of cytoplasmic, increase, the tumor becomes more glioblastoma (Burger et al., 1985). anaplastic. Anaplastic astrocytomas may have moder- In addition to specific histologic features of ma- ate degrees of all these features. In addition, mitotic lignancy grades, astrocytomas may have a widely vari- activity is present. Anaplastic astrocytomas are more able histology from one microscopic field to another likely to show cortical invasion or subpial accumula- within the same tumor, a feature known as hetero- tion of neoplastic cells. Any astrocytoma with gemis- geneity. Tumor heterogeneity is thought to arise from tocytic change involving greater than about 20% of a predominance of various transformed clones of as- the cells should be graded as an anaplastic astrocy- trocytes in different regions of the tumor during its 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 88

88 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

growth. As the name multiforme suggests, glioblas- Necrosis was used to divide the intermediate-type as- tomas are especially noted for their variable histol- trocytoma from the glioblastoma. Although the three ogy. systems defined histologic grade according to the presence of cellular anaplasia, the systems of Svien and Kernohan determined tumor grade according to Approaches to Grading Gliomas the proportion of normal tissue remaining mixed with Whereas for most tumors it is recognized that higher the invading tumor and on the type of invading edge grades of malignancy are associated with a poorer of tumor into normal tissue. Because of the infiltra- patient prognosis, the grading criteria for gliomas in tive growth pattern of gliomas, this feature is notori- most grading schemes are subjective. Within a single ously difficult to distinguish, especially in well- grade, the prognosis for tumors in individual patients differentiated tumors. Also, it has subsequently been may be difficult to predict on the basis of grade alone considered that the presence of even a single highly (i.e., astrocytomas with similar histologic features malignant focus of glioblastoma in an otherwise may behave in widely disparate clinical fashions). Any lower-grade astrocytoma portends a grave prognosis. scheme of histologic grading has two main goals: the Data from the original publications (Svien et al., tumor grade must predict behavior, and the grading 1949; Kernohan et al., 1949) showed no clinical dis- criteria must be sufficiently objective and defined to tinction between grades II and III astrocytomas, minimize variation among observers and to maximize whereas there was a clear survival difference among reproducibility (Fulling and Nelson, 1984). Numer- the three groups of patients in Ringertz’s study. Thus, ous grading systems have been used for astrocytomas, the three-tiered grading schemes were popularized. most of them utilizing three or four grades of malig- Three-tiered grading systems have been shown to nancy (Burger et al., 1985, 1991; Fulling and Nelson, be closely correlated with the clinical prognosis for 1984; Svien et al., 1949; Kernohan et al., 1949; astrocytomas and are widely used today in diagnos- Ringertz, 1950; Nelson et al., 1983; Burger and tic neuropathology. These grading systems include Green, 1987; Daumas-Duport et al., 1988b; Kleihues those of Ringertz, with its modifications (Ringertz, and Cavenee, 2000). Histologic criteria used to de- 1950; Fulling and Nelson, 1984; Nelson et al., 1983; fine these grades vary somewhat for different schemes Burger et al., 1985; Burger and Green, 1987), and a (Table 3–1). Increasingly, more malignant tumors scheme widely called the St. Anne-Mayo system show a continuum of histologic features that permits (Daumas-Duport et al., 1988b). The St. Anne-Mayo their classification into grades. scheme designates four histologic features to be used The first systems for grading gliomas were pre- in grading and nominally has four grades of astrocy- sented at about the same time (Svien et al., 1949; Ker- toma. Their grade 1 tumor, however, is exceedingly nohan et al., 1949; Ringertz, 1950). Svien and Ker- rare (0.25% in one series), and the other three nohan each used a four-grade system (designated as grades produce three distinct survival curves; thus, grades 1 through 4) that was patterned after the his- this scheme should also be considered as a three- tologic grading system for epithelial neoplasms used tiered scale (Kim et al., 1991). at the Mayo Clinic. Ringertz’s system had three grades Histologic features used to grade astrocytomas in- of astrocytoma malignancy: astrocytoma, intermedi- clude increasing cellularity; microcysts (which sug- ate-type astrocytoma, and glioblastoma multiforme. gest an improved prognosis); nuclear and cellular

Table 3–1. Comparison of Astrocytoma Grading Systems Modified Ringertz St. Anne-Mayo, 1988 WHO 2000 Diffuse Astrocytoma (low grade) Astrocytoma, grades 1 and 2 astrocytoma (II) Anaplastic Astrocytoma, grade 3 astrocytoma (III) Anaplastic astrocytoma Glioblastoma Astrocytoma, grade 4 multiforme (IV) Glioblastoma multiforme 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 89

Primary Cerebral Tumors 89

Figure 3–12. Histologic features used to grade astrocytoma malignancy. (A) High cellularity, microvascular proliferation (V), and frequent mitotic figures (arrows) are features of an anaplastic astrocytoma. Original magnification, 200. (B) Focus of co- agulation necrosis (N) with peripheral pseudopalisading of neoplastic cells that also show marked nuclear and cytoplasmic pleo- morphism. These are features of a glioblastoma multiforme. Original magnification, 100. Both are hematoxylin and eosin stained.

pleomorphism; vascular mural cell proliferation; mi- among individual pathologists who use the St. Anne- totic activity; and coagulative tumor necrosis (Fig. Mayo system is reported to be as high as 94% (Kim 3–12). The St. Anne-Mayo system rates cellular pleo- et al., 1991). That single study examined 251 astro- morphism, mitotic activity, vascular endothelial pro- cytoma cases and suggested, because of survival data, liferation, and necrosis within a standardized and ob- that the presence of necrosis should cause a tumor jective numeric scale. When present, each feature to immediately be classified as grade 4, even using receives one scoring point, and the grades are based their system. In this discussion, the terms low-grade solely on the total numeric score. Concordance astrocytoma, anaplastic astrocytoma, and glioblas- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 90

90 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

toma are used. In most cases it can be assumed that growing tumors, most neuropathologists and neuro- the glioblastoma is primarily derived from astrocytes, surgeons do not consider them benign because of even though its morphology may be so bizarre as to their invasive quality and their location within the con- obscure such recognition, and sometimes it follows fines of the bony calvarium. These astrocytomas are an earlier diagnosis of oligodendroglioma. rarely cured because they cannot be completely ex- The most recent World Health Organization cised and because their ability to expand without (WHO) classification also incorporates grading of damage to the host is limited by the skull. Thus, pa- nervous system neoplasms (Kleihues and Cavenee, tients may die from recurrent astrocytomas, some- 2000). The WHO criteria and general grading for as- times even lower grade ones, because of associated trocytomas are analogous to the St. Anne-Mayo sys- increased intracranial pressure or invasion of vital tem, but the terms used are diffuse astrocytoma (low CNS structures. grade), anaplastic astrocytoma, and glioblastoma. Diffuse fibrillary astrocytomas are the most com- They are considered as WHO grades II, III, or IV, re- mon morphologic subtype and usually arise in the spectively (Daumas-Duport et al., 1988b; Kleihues white matter, which is the location of their normal and Cavenee, 2000). cellular counterparts. The lobar distribution of these Previous grading schemes for oligodendrogliomas tumors is similar to that of the amount of white mat- are not as well correlated with patient survival as are ter present in each brain lobe, with a higher inci- those for astrocytomas (Smith et al., 1983; Burger et dence in the frontal regions (Burger et al., 1991). al., 1987; Shaw et al., 1992; Mork et al., 1985; Kros Grossly, these tumors are slightly discolored yellow et al., 1988, 1990). Well-differentiated oligoden- or gray and have indistinct margins with the sur- drogliomas are considered as WHO grade II neo- rounding brain. The usually discrete border between plasms, whereas anaplastic oligodendrogliomas are cerebral white matter and gray matter may be blurred WHO grade III (Kleihues and Cavenee, 2000). Grad- by the tumor. The tumor consistency is variably re- ing oligodendrogliomas using WHO criteria has been ported as firm, almost rubbery, or soft and gelati- shown to be clinically significant (Dehghani et al., nous. This character may depend on the degree of 1998). Only these two grades of malignancy are ac- fibrillarity of the individual cells within the neoplasm. cepted by the WHO. Many features used to grade Histologically, low-grade astrocytomas show in- oligodendrogliomas are similar to those for astrocy- creased cellularity compared with normal brain tis- tomas: cellularity, pleomorphism, mitotic activity, vas- sue and have mild or moderate nuclear pleomor- cular changes, and necrosis (Smith et al., 1983; phism. The increase in cellularity may be only slight, Burger et al., 1987). Sometimes lower grade oligo- producing a challenge for the diagnostic pathologist, dendrogliomas may have microcysts as well (Mork et and may simulate the slightly increased cellularity of al., 1986). Oligodendrogliomas of all histologic reactive astrocytosis, a repair process. However, re- grades tend to infiltrate the cortex readily to form active astrocytic cells show generally more abundant clusters of neoplastic cells in the subpial region and and luxuriant cytoplasmic processes than do neo- around neurons and blood vessels. plastic astrocytes, with a more even cellular distribu- Grading systems for the other types of gliomas are tion and smaller, darker nuclei. The nuclei of astro- much less well defined. For ependymomas, most au- cytoma cells are enlarged and show more prominent thors have found little correlation between postoper- chromatin granules. Microcysts may be a useful fea- ative survival and tumor grade (Schiffer et al., 1991; ture for making a differential diagnosis with gliosis Fokes and Earle, 1969; Ross and Rubinstein, 1989; because this feature is rare in gliosis but common in Chiu et al., 1992). low-grade astrocytomas. Astrocytoma cytoplasmic processes may be evident in routine sections but can be better visualized after immunohistochemical stain- Specific Tumor Characteristics ing for glial fibrillary acidic protein. Other features of anaplasia, such as mitotic activity, vascular prolifer- Low-Grade Astrocytoma ative changes, and necrosis, are absent. Microcalci- Diffuse low-grade astrocytomas are WHO grade II fications can be present in as many as 15% of astro- (Kleihues and Cavenee, 2000). Even though low- cytomas. Although these tumors are diffusely invasive grade astrocytomas might be considered slow- into the surrounding brain, their invasion is largely 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 91

Primary Cerebral Tumors 91

limited to white matter. Invasion of the cerebral cor- havior, and some markers of cell proliferation have tex with perineuronal satellitosis and subpial accu- been used in an attempt to predict prognosis more mulation suggests a more aggressive astrocytoma or accurately. The most used markers in this area have an oligodendroglioma. been antibodies to bromodeoxyuridine (BrdU) and Astrocytomas, especially low-grade examples, may Ki-67. The cellular incorporation of BrdU is a spe- occur as mixed tumors with neuronal or other glial cific marker of the DNA synthesis phase of the cell components. Mixtures with oligodendroglioma are cycle, whereas the Ki-67 antibody labels an antigen especially common, but other elements should be that is present in all phases of the cell cycle except present in significant proportions to justify a diagno- G0. Both antibodies can be identified by immunohis- sis of mixed glioma. tochemical staining in paraffin-embedded tissue sec- tions. In longer use, the BrdU labeling index was found to correlate with tumor grade and prognosis Anaplastic Astrocytoma in all grades of astrocytoma and oligodendroglioma The gross appearance of anaplastic astrocytomas is (Hoshino et al., 1993; Prados et al., 1998b; Lamborn similar to that of the lower grade tumors, except that et al., 1999), but not within the glioblastoma multi- the anaplastic astrocytomas have a softer consistency forme histology (Ritter et al., 1994). The BrdU la- because they usually lack extreme fibrillarity. Diffuse beling index appears to be an age-independent pre- infiltration remains a feature, without evidence of ne- dictor of survival only in low-grade and anaplastic crosis in the tumor, although the relatively greater dif- astrocytomas (Ito et al., 1994). ference with surrounding normal brain tissue may Because BrdU must be introduced into the tumor falsely suggest a more discrete lesion. by injection into the patient before surgery, the The histologic features of anaplastic astrocytomas MIB-1 antibody to the Ki-67 antigen is far more pop- are similar to those of low-grade astrocytomas but ular today. The MIB-1 antibody can be used in paraf- these features are more abundant and exaggerated. fin tissue sections and appears reliable for labeling These tumors are WHO grade III (Kleihues and Cave- index prognostications (Key et al., 1993; Davis et al., nee, 2000). Cellularity is more increased, as are nu- 1995) in a manner similar to BrdU. Some studies sug- clear and cellular pleomorphism. These features may gest that astrocytomas with MIB-1 labeling indices of be extreme, with back-to-back cells and bizarre, hy- 5% or 7.5% are associated with shorter survival perchromatic nuclei. Cytoplasm may be scanty, with times and an anaplastic histologic grade (Jaros et al., nuclear lobation and enlargement indicating anapla- 1992; Montine et al., 1994). sia. Alternatively, the abundant eosinophilic cyto- plasm of gemistocytes may be prominent, with Glioblastoma relatively small and uniform nuclei. Gemistocytic as- trocytic foci generally occur in tumors with other Glioblastoma, also known as glioblastoma multi- more usual fibrillary areas. The presence of more forme, is the glioma with the highest grade of malig- than 20% of gemistocytes in a glial neoplasm is a poor nancy, WHO grade IV (Kleihues and Cavenee, 2000). prognostic sign (Krouwer et al., 1991); thus, gemis- It represents 15% to 23% of intracranial tumors and tocytic astrocytomas should be considered anaplas- about 50% to 60% of astrocytomas. Most examples tic. Mitotic activity is easily recognized in most are generally considered to arise from astrocytes be- anaplastic astrocytomas but inexplicably may be ab- cause glial fibrillary acidic protein can be identified sent in the gemistocytic areas. Despite this seeming in the cell cytoplasm. Some examples, however, ap- anomaly, gemistocytic cells commonly transform to parently arise from other glial lineages, such as oligo- more highly anaplastic small cells. dendrocytes. Glioblastoma is the most frequently oc- The range of anaplasia in this grade is broad, with curring astrocytoma. Autopsy and serial biopsy some examples showing low cellularity and pleo- studies have shown that some astrocytomas progress morphism with a few mitotic figures and others be- through the grades of malignancy with transforma- ing highly cellular and pleomorphic with frequent mi- tion from low-grade to anaplastic astrocytoma to toses, lacking only the necrosis required for a glioblastoma (Muller et al., 1977). But, because some histologic diagnosis of glioblastoma. For this reason, examples of glioblastoma appear to arise rapidly in it is useful to have a more objective indicator of be- otherwise normal patients and are recognized when 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 92

92 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

they are small, it is thought that this variety of glioblas- phologic cell types can be confirmed by a positive toma can also arise directly from malignant transfor- glial fibrillary acidic protein immunostain. mation of astrocyte precursor cells without passing The regional distribution of glioblastomas is sim- through the lower grades of malignancy (Kleihues and ilar to that of other astrocytomas, although these neo- Ohgaki, 1997, 1999). plasms occur relatively frequently in the brain stem Tumor necrosis is the characteristic gross feature and spinal cord in younger patients. If the tumors that distinguishes glioblastoma from anaplastic as- gain access to the ventricular system or the sub- trocytoma (Fulling and Nelson, 1984; Nelson et al., arachnoid space, these pathways become avenues for 1983; Burger et al., 1985; Burger and Green, 1987). wide dissemination. Another microscopic feature that is distinctive and di- agnostic is the presence of proliferative vascular Gliosarcoma changes within the tumor. These changes may occur in the endothelial cells (vascular endothelial hyper- Gliosarcoma is a variant of glioblastoma in which a plasia or proliferation) or in the cells of the vessel distinct component of sarcoma is admixed with the wall itself (vascular mural cell proliferation). Both glioma. The frequency of gliosarcoma in glioblastoma types of change are sometimes considered together is reported to be from 2% to 8% (Morantz et al., as microvascular proliferation. Glioblastomas may 1976; Meis et al., 1991). In some cases, this change show evidence of older or more recent hemorrhage, is recognized at recurrence of an anaplastic astrocy- and their cellularity is usually extremely high. The in- toma or glioblastoma and may also be seen de novo. dividual cells may be small, with a high nuclear/cy- The sarcoma element may be of any histologic type, toplasmic ratio, or very large and bizarre, with abun- with fibrosarcoma being the usual one. Other types dant eosinophilic cytoplasm. With cell proliferation include malignant fibrous histiocytoma, osteosar- labeling studies, the small cells are the more prolif- coma, rhabdomyosarcoma, and chondrosarcoma. erative ones, and these small cell glioblastomas have Currently, the sarcomatous element is thought to be a more aggressive course, with even shorter patient derived from a mesenchymal cell associated with the survival times than those of highly pleomorphic or vascular adventitia. Differentiation into endothelium, better differentiated astrocytes (Burger and Green, smooth muscle, or pericytes can occur (Miller et al., 1987). Some glioblastomas may be so highly cellular 1991; Haddad et al., 1991). The prognosis of gliosar- that the population of small anaplastic cells simulates coma is similar to that of glioblastoma. primitive neuroectodermal tumors such as the medul- loblastoma. These same small cells may appear to Oligodendroglial Tumors condense in rows around areas of tumor necrosis, forming the characteristic pseudopalisades. They also Oligodendrogliomas, like astrocytomas, mimic the have a propensity to infiltrate the brain extensively, histology of their presumed cell of origin. They also spreading even to distant locations and giving the ap- arise primarily in the white matter but tend to infil- pearance of a multifocal glioma. Although some ex- trate the cerebral cortex more than do astrocytomas amples are truly multifocal (i.e., arising in multiple of a similar grade of malignancy. Like astrocytomas, simultaneous primary sites), many of these multifo- grading schemes of histologic malignancy have been cal tumors show a histologic connection when the used for oligodendrogliomas, but these correlate less whole brain is examined at autopsy. well with prognosis than those used for astrocytomas. The histologic morphology of glioblastoma can be Many of the histologic features used to grade oligo- highly variable, conferring the name multiforme. dendrogliomas are similar to those used for astrocy- Some tumors are largely spindled and simulate a fi- tomas: cellularity, pleomorphism, mitotic activity, brosarcoma. Others show cytoplasmic lipidization or vascular changes, and necrosis. Lower grade oligo- epithelioid structures that imitate squamous or glan- dendrogliomas may have microcysts. Oligoden- dular patterns with positive keratin immunoreactivity drogliomas of all histologic grades tend to infiltrate (Rosenblum et al., 1991; Gherardi et al., 1986; Gal- the cortex readily and to form clusters of neoplastic loway and Roessmann, 1986). A myxoid or chondroid cells in the subpial region, around neurons, and type of appearance has also been seen (Kepes et al., around blood vessels. In general, the cells of oligo- 1984). The astrocytic origin of these diverse mor- dendrogliomas have round, regular nuclei and dis- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 93

Primary Cerebral Tumors 93

Figure 3–13. Low-grade oligodendroglioma. Proliferation of cells with round nuclei and cleared cytoplasm (“fried egg” cells). The oligodendroglioma cells are more numerous than would be seen in normal brain tissue, but neuropil is present between the neoplastic cells. Small, thin-walled capillaries (arrows) are also a feature. Hematoxylin and eosin. Original magnification, 200.

tinct cytoplasmic borders with clearing of the cyto- be useful if these tumors are nonimmunoreactive (as plasm (Fig. 3–13). Their cytologic appearance has might be expected) for glial fibrillary acidic protein. been compared to “fried eggs.” The neoplastic cells With increasing anaplasia, oligodendrogliomas can do not have fibrillary cytoplasmic processes. Another become highly cellular and pleomorphic, approach- fairly distinctive and diagnostically helpful feature is ing an appearance of glioblastoma multiforme with the vascular pattern of oligodendrogliomas, referred the presence of necrosis. Although it is correct to clas- to as “chicken wire” vessels. The blood vessels divide sify these as anaplastic oligodendrogliomas, some the tumor into discrete lobules. For this reason and would use the term glioblastoma once necrosis because of the not uncommonly seen discrete mar- is identified in any high-grade glial neoplasm. One gin between oligodendrogliomas and adjacent white justification for separating anaplastic oligoden- matter, some are mistaken for metastatic carcinomas. drogliomas from astrocytic glioblastomas is the Variants of the usual clear cells are recognized. Some slightly better prognosis of the former, even in this oligodendrogliomas can have cells with a small rim highest grade of malignancy (Ludwig et al., 1986; De- of eosinophilic cytoplasm. These cells are distinctive hghani et al., 1998). Some authors have reported that and are called mini-gemistocytes. Despite the cyto- an MIB-1 labeling index of 3% to 5% predicts a plasm, there are few or no cell processes, and these worse prognosis in oligodendrogliomas (Heegard et mini-gemistocytes are believed to be true oligoden- al., 1995; Kros et al., 1996; Coons et al., 1997; Cairn- drocytes rather than astrocytes (Herpers and Budka, cross et al., 1998; Dehghani et al., 1998). 1984). As might be expected, this rim of eosinophilic cy- Oligoastrocytomas toplasm may contain intermediate filaments ultra- structurally (Kros et al., 1992), and it shows positive Many, if not most, oligodendrogliomas occur with a immunoreactivity for glial fibrillary acidic protein. In regional or intimate cellular mixture of astrocytoma. addition, most oligodendrocytes show positive nu- For the diagnosis of mixed glioma, the proportion of clear and cytoplasmic reactivity for S-100 protein, a each should be substantial, but authors have differ- characteristic that can help in distinguishing them ing opinions with respect to exact numbers; usually from other clear cell tumors in the brain and that can a mixture with a range from 10% to 25% of the mi- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 94

94 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

nor element is used to diagnose a mixed glioma. and frequent mitoses. Vascular mural cell prolifera- Oligoastrocytomas and anaplastic oligoastrocytomas tion and necrosis do not seem to be predictive of poor correspond to WHO grade II or grade III, respectively prognosis in ependymomas. (Kleihues and Cavenee, 2000). Histologic features of anaplasia may be present in either component and will affect the prognosis adversely. Such features in- PATHOLOGY OF NONINFILTRATIVE clude marked cellular pleomorphism, high cellular- GLIOMAS ity, and a high mitotic rate. Microvascular prolifera- tion and necrosis may also be seen. Prognosis and The following sections of this chapter discuss the very response to therapy have not been shown to depend low-grade, noninfiltrating gliomas and other low- on the proportion of the oligodendroglial versus the grade infiltrating gliomas whose prognosis, while astrocytic component of the tumor (Shaw et al., quite good for brain tumor patients, is less favorable 1994). than that of patients with noninfiltrating gliomas. Most low-grade glioma patients are children, adolescents, Supratentorial Ependymomas or young adults; therefore, special consideration with respect to surgery, perioperative care, radiotherapy, Ependymomas are tumors that arise from the ependy- and chemotherapy is discussed. mal cells of the ventricles. Although more common In addition to the diffuse supratentorial astrocy- in the posterior fossa of children, they are also well- tomas that are histologically graded according to the recognized supratentorial gliomas. In contrast to the above scheme, special variants of astrocytoma exist posterior fossa examples, supratentorial ependymo- that do not conform to these grades and that occur mas often arise with no direct connection to the cere- in specific clinical situations. These include the juve- bral ventricles (Burger et al., 1991; Bigner et al., nile pilocytic astrocytoma, which is the most common 1998). They have a distinctive, almost epithelial, ap- tumor in this group; the pleomorphic xanthoastro- pearance and are commonly calcified. One diagnos- cytoma; the desmoplastic astrocytoma; and the tic feature is the presence of true ependymal rosettes, subependymal giant cell astrocytoma of tuberous but these are rare. A feature that serves to suggest sclerosis. Other more uncommon variants of mixed the diagnosis is the alternating appearance of zones neuronal–glial and mature neuronal tumors include with fairly uniform round nuclei alternating with the desmoplastic infantile ganglioglioma, the dysem- eosinophilic anuclear zones. These anuclear zones bryoplastic neuroepithelial tumor, the central neuro- occur most commonly around blood vessels and rep- cytoma, and the ganglioglioma. resent ependymal cell cytoplasmic processes. True ependymal rosettes may be impossible to find with- out resorting to ultrastructural studies. Characteristics of Specific Tumors Ependymomas may have some immunoreactivity to Juvenile Pilocytic Astrocytoma glial fibrillary acidic protein and may also react with antikeratin antibodies. Other ultrastructural features The most common locations of supratentorial juve- that attest to these tumors’ histologic origin are a nile pilocytic astrocytomas, considered as WHO grade columnar cell shape, a definite polar orientation with I, are the cerebral hemispheres followed by, in order respect to blood vessels, basal lamina near the ves- of decreasing frequency, optic pathways, hypothala- sel, and abundant microvilli on the cell surface dis- mus, and thalamus (Clark et al., 1985; Sutton, 1987). tant from the vessel. True cilia may be recognized but Juvenile pilocytic astrocytomas of the cerebral hemi- may be difficult to find. Many supratentorial ependy- spheres are usually cystic, with a tumor nidus pre- momas have regions with clear cells, simulating a senting as a mural nodule (Palma and Guidetti, 1985; mixture of oligodendroglial cells. However, these Tomita et al., 1986). On occasion, the tumor is clear cells are thought to be a variant of ependymal arranged in a plaque-like fashion at the perimeter of cells. Ependymomas are classified with WHO grade II the cyst, which is brightly enhanced on CT following tumors. Anaplastic ependymomas are WHO grade III the administration of IV contrast agents (Maiuri, (Kleihues and Cavenee, 2000). Ominous features that 1988). It should be noted, however, that not all cys- suggest a higher grade include very high cellularity tic hemispheric lesions, with or without the appear- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 95

Primary Cerebral Tumors 95

ance of a mural nodule, are astrocytic in nature. The parenchymal pilocytic astrocytomas are commonly clinical differential diagnosis of these tumors includes described as cystic with a mural nodule. This nodule ependymoma, hemangioblastoma, neuroblastoma, has a fleshy, dark red appearance grossly and is dis- meningioma, and primitive neuroectodermal tumor. tinguished histologically by its biphasic nature. It ap- The vast majority of patients with juvenile pilocytic pears as an irregular alternating pattern of pilocytic astrocytomas develop symptoms in the first three areas and looser microcystic areas (Fig. 3–14). The decades of life, in particular between the ages of 10 microcysts contain eosinophilic, acellular, proteina- and 25 years (Clark et al., 1985; Garcia and Fulling, ceous material. The cells in the looser areas are more 1985; Schisano et al., 1963). The duration of symp- stellate, with multiple short cell processes and round toms before surgery is usually 3 to 5 years. The most nuclei. The pilocytic areas are commonly located common symptoms of cerebral hemisphere tumors around blood vessels. The numerous blood vessels are related to mass effect and location and include can show thickening and hyalinization of their walls, headaches, nausea and vomiting, weakness, visual with endothelial cell hypertrophy or proliferation, disturbances, and seizures. Tumors of the hypothal- simulating the vascular mural cell proliferation seen amus and thalamus can cause precocious puberty, in malignant astrocytomas of the cerebral hemi- “diencephalic syndrome” (Scott and Mickle, 1987), spheres. However, in the pilocytic astrocytomas of and hemiparesis, respectively. Children less than 1 to children, these vascular changes do not portend an 2 years of age may also present with macrocephaly. ominous prognosis. Other areas of these tumors may This finding of an enlarged head in an infant or new- show cells with clear cytoplasm, resembling oligo- born who also is failing to thrive indicates a tumor in dendroglioma. Individual nuclei of the pilocytes can the hypothalamic area until proven otherwise (Wisoff be enlarged, hyperchromatic, and pleomorphic, but et al., 1990). Obstructive hydrocephalus results from these nuclear features again are not associated with a tumor, usually of a pilocytic phenotype, occurring a worse behavior. It is thought that this nuclear pleo- in the posterior optic pathway, optic–hypothalamic morphism represents a degenerative change. The pi- region, or the anterior thalamus (Nishio et al., 1993). locytic astrocytomas with the biphasic pattern have Anatomically, the tumor occludes the anterior third been termed juvenile pilocytic astrocytomas to dis- ventricle, blocking either the aqueduct of Sylvius or tinguish them from the diffuse astrocytomas having a the foramen of Monro. Anteriorly located optic path- pilocytic growth pattern that occurs in the cerebral way tumors are rarely associated with hydrocephalus. hemispheres of adults or in the brain stem. These dif- Astrocytomas of the third ventricle region, optic fuse pilocytic astrocytomas are typically graded ac- apparatus, and cerebellum in children have a similar cording to the usual criteria for malignant astrocy- histologic appearance and biologic behavior. Their tomas and behave as such. singular common feature is the presence of elon- Pilocytic astrocytomas of the optic apparatus and gated, spindled astrocytes, the so-called pilocytes third ventricle region are less commonly biphasic or “hair-like” cells (Burger et al., 1991). These than are the hemispheric parenchymal examples, but cells have spindle-shaped nuclei and long, thin, have the distinctive pilocytes and lack mitotic activity eosinophilic cytoplasmic processes that extend from (Fig. 3–14). These tumors distort the usual organi- the cells in a bipolar array. Another usual histologic zational pattern of the optic nerves and chiasm, ex- feature is the presence of Rosenthal fibers within the panding and then obliterating the septate structure. tumor or at its periphery and the interface with sur- There is an increase in glial cellularity but little pleo- rounding brain. These distinctive eosinophilic beaded morphism. An exuberant growth of connective tissue structures are derived from astrocytic processes. Ul- is also a common feature, and the tumors can grow trastructurally, the eosinophilic material is electron into the leptomeninges of the optic structures. Rosen- dense, and residual intermediate filaments can be dis- thal fibers are a common finding. Metastatic spread tinguished at its periphery. Immunocytochemically, in the cerebrospinal fluid has been described, even studies for glial fibrillary acidic protein usually show in the absence of increased histologic anaplasia peripheral positive reactivity in Rosenthal fibers, with (Obana et al., 1991). a largely negative reaction in the center. Pilocytic astrocytomas show strong cytoplasmic In addition to this common feature, tumors of the glial fibrillary acidic protein immunoreactivity in the separate locations have some distinctive features. The pilocytic cells with diffuse reactivity in the stellate 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 96

96 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Figure 3–14. Low-grade pilocytic astrocytoma. (A) The “juvenile” type shows a biphasic solid and microcystic (M) growth pat- tern. Hematoxylin and eosin. Original magnification, 200. (B) The diffuse type of pilocytic astrocytoma shows elongated “hair- like” cells without the microcystic component. No mitotic activity is present. Hematoxylin and eosin. Original magnification, 300.

cells. Vascular structures fail to stain with glial fibril- young adults afflicted with this syndrome are typically lary acidic protein, but can be delineated with a Mas- cognitively retarded and have seizures that are often son trichrome, reticulin, or Factor VIII immunostain. refractory to medical care and dermatologic mani- These three methods label the perivascular collagen, festations of facial adenoma sebaceum. Cortical connective tissue fibers, or endothelial cells, respec- hamartomas are the rule and are associated with tively. periventricular calcified nodules (Pinto-Lord et al., 1986). These latter lesions are typically located near the foramen of Monro and histologically are Tuberous Sclerosis (Bourneville’s Disease) subependymal giant cell astrocytomas (McLaurin and Tuberous sclerosis is a phakomatosis associated with Towbin, 1986), a lesion that has not been docu- supratentorial noninfiltrative gliomas. Children and mented in patients without tuberous sclerosis (Shep- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 97

Primary Cerebral Tumors 97

herd et al., 1991). If the subependymal nodule grows, to its pleomorphic and giant cell appearance. None- it will block the foramen of Monro and cause ob- theless, cases have been reported with a rapidly pro- structive hydrocephalus. Often this is a very indolent gressive tumor growth resulting in death (Weldon- process with a gradual onset of symptoms. Linne et al., 1983). Xanthoastrocytomas, like other gliomas, have been found in patients with neurofi- bromatosis (Ozek et al., 1993). Subependymal Giant Cell Astrocytoma The pleomorphic xanthoastrocytoma is a distinc- tive astrocytoma variant known for its extreme histo- The subependymal giant cell astrocytoma is a tumor logic pleomorphism in the absence of significant mi- of uncertain histogenesis that is usually included with totic activity and necrosis (Kepes et al., 1979; Kepes, astrocytomas. It is classified as WHO grade I (Klei- 1993). The importance of recognizing this variant is hues and Cavenee, 2000). The degree of histologic that it should not be mistaken for glioblastoma mul- pleomorphism of the cells may suggest a diagnosis of tiforme. This superficially located tumor of early glioblastoma multiforme, but mitotic activity and ne- adulthood is usually considered benign and is surgi- crosis are lacking. The subependymal giant cell as- cally curable. The individual tumor cells, in addition trocytoma generally occurs in the setting of tuberous to showing marked pleomorphism tending even to- sclerosis and is located, as its name suggests, in the ward the bizarre, have a high content of lipid, re- subependymal region, two features that may be help- vealed in frozen sections by staining for neutral fat ful in suggesting the correct diagnosis. However, spo- with oil red O. The astrocytic nature of these foamy, radic examples in patients without a history of tuber- lipidized cells is revealed by positive cytoplasmic im- ous sclerosis do occur (Boesel et al., 1979). The munoreactivity for glial fibrillary acidic protein (Grant tumor cells are large and bizarre, and the tumor is and Gallagher, 1986). The finding of mitotic activity highly cellular. Abundant eosinophilic cytoplasm is a or necrosis or the occurrence of this neoplasm in pa- usual feature. The nuclei are very enlarged and have tients beyond the third decade of life should raise cau- prominent nucleoli, even suggesting a neuronal ori- tion about the diagnosis and prognosis. These exam- gin (Nakamura and Becker, 1983). Ultrastructurally, ples may be classified as WHO grade III (Giannini et intermediate filaments are numerous, as are long, al., 1999). Some examples have been shown to trans- thickened cytoplasmic processes. Immunohisto- form to glioblastoma or to behave in a malignant fash- chemical studies, however, failed to reveal a defined ion (Weldon-Linne et al., 1983). histogenesis for this neoplasm, as tumor cells may mark poorly for both glial and neuronal antigens or may mark for both. This tumor is usually considered benign and is curable by surgical excision alone Desmoplastic Cerebral Astrocytoma (Shepherd et al., 1991). The desmoplastic infantile astrocytoma is another dis- tinctive variant that occurs in a stereotypical clinical situation. It corresponds to WHO grade I (Kleihues Pleomorphic Xanthoastrocytoma and Cavenee, 2000). This tumor is generally one of Another tumor found to involve the cerebral hemi- infancy and childhood (VandenBerg, 1993; Louis et spheres in children and young adults is the pleo- al., 1992; de Chadarevian et al., 1990). A few exam- morphic xanthoastrocytoma. This lesion is almost al- ples in adults have been described. The singular his- ways superficial in location and often abuts a tologic feature is the extreme spindling of astrocytes leptomeningeal surface (Kepes et al., 1979). The tem- within an extensive collagen matrix. The tumor may poral lobe is most commonly involved, followed in be superficially located in the hemisphere, and it has frequency by the parietal lobe. In our experience, been postulated that the collagen arises from growth seizures are the most common form of presentation, of the tumor into the pia with collagen proliferation and the duration of symptoms may extend up to sev- from that source. These tumors may even resemble eral years before the surgical diagnosis. Although usu- a sarcoma, but immunohistochemical studies positive ally described as a benign and indolent lesion and for glial fibrillary acidic protein again reveal their as- classified as WHO grade II, the clinical course and trocytic nature. The prognosis for these tumors is outcome of this tumor are controversial because of considered favorable with complete excision, but they the difficulty in arriving at a histologic diagnosis due occur very infrequently. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 98

98 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

A tumor having similar histology but also exhibit- seizures (i.e., complex partial epilepsy, with or with- ing primitive small cells and neuronal differentiation out secondary generalization). Often patients with occurs in very young children, generally under 2 years these tumors present with medically refractory of age (VandenBerg et al., 1987; VandenBerg, 1993). epilepsy that requires intraoperative electrocorticog- These are the desmoplastic infantile gangliogliomas, raphy (see below) to improve seizure control which are discussed below. (Pilcher et al., 1993). The ganglioglioma is a biphasic tumor that has a mature but atypical neuronal component mixed with Mixed Neuronal–Glial Tumors a neoplastic glial component. Gangliomas may be Mixed neuronal–glial tumors include a broad histo- WHO grade I or II (Kleihues and Cavenee, 2000). The logic spectrum, ranging from the typical gangli- tumor’s histologic appearance is quite characteristic, oglioma, with its complement of atypical but well- with a prominent vascular septa dividing it into a lob- differentiated neuronal elements, to the more com- ular pattern. Mature lymphocytes can be a feature of plex examples of desmoplastic infantile ganglioglioma the vascular component. The glial component is usu- and dysembryoplastic neuroepithelial tumor. ally astrocytic and low grade, with rare or absent mitotic figures (Haddad et al., 1992). The glial mor- Dysembryoplastic Neuroepithelial Tumor. The phology may suggest a pilocytic astrocytoma. Inter- dysembryoplastic neuroepithelial tumor is found ex- mixed with the astrocytic component is a population clusively in patients who undergo cortical resections of atypical neurons, which are manifested in their size for intractable epilepsy (Daumas-Duport et al., and shape, arrangement, and neurohistology. Multi- 1988a). This tumor is described as showing multiple nucleated forms may be seen. Because large, gemis- intracortical nodules of glial proliferative elements tocytic astrocytes may simulate atypical neurons with combined with a dysplastic cerebral cortex and a their abundant eosinophilic cytoplasm, special im- columnar orientation of a “specific glioneuronal el- munohistochemical markers of neuronal differentia- ement” composed of oligodendroglia, neurons, and tion, such as synaptophysin or neurofilament protein, capillaries (Daumas-Duport, 1993). The astrocytic may be required to confirm the diagnosis (Miller et component is generally pilocytic and forms nodules al., 1993). that may be microcystic. Mitotic activity is not seen, and necrosis is not a feature of this tumor. These are Desmoplastic Infantile Ganglioglioma. The des- considered WHO grade I (Kleihues and Cavenee, moplastic infantile ganglioglioma is a distinctive and 2000). The tumor’s intracortical location and nodu- rare tumor that occurs in infants, generally within the lar architecture, with associated cortical dysplasia, first 2 years of life (VandenBerg, 1993). These tu- distinguish it from the typical ganglioglioma. After re- mors are very similar to the superficial cerebral as- section for epilepsy treatment, recurrence or pro- trocytoma with dural attachment described by gression of dysembryoplastic neuroepithelial tumor Taratuto et al. (1984) and tend to be located in the has not been demonstrated. frontal and parietal regions. These tumors may be Dysembryoplastic neuroepithelial tumors have a large, with a prominent cystic or multicystic appear- predilection for the temporal lobe in young patients ance, and generally have a superficial hemispheric lo- who have difficult-to-control seizure activity. These tu- cation. They have a predominant desmoplastic char- mors are well circumscribed grossly and on diag- acter, resulting in a firm gross consistency, and often nostic imaging studies (Koeller and Dillon, 1992). have a large amount of proteinaceous fluid. Despite They are typically very indolent lesions with a long this, they infiltrate into brain. The desmoplasia con- natural history (Daumas-Duport, 1993). sists of admixed fibrous collagen and glial elements. The astrocytic component may be moderately pleo- Ganglioglioma. Gangliogliomas usually have a long morphic but is identified by its positive cytoplasmic duration of symptoms, sometimes up to 15 to 20 immunoreactivity for glial fibrillary acidic protein. years, before definitive clinical diagnosis (Kalyan- The neuronal elements range from mature neurons Raman and Oliverio, 1987; Silver et al., 1991). Gan- to primitive small cells and can be recognized by im- gliogliomas have a predilection for the temporal and munohistochemical stains for synaptophysin or neu- frontal lobes and almost always are associated with rofilament protein. The neuronal elements are usu- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 99

Primary Cerebral Tumors 99

ally seen in the less desmoplastic areas of the tumor postoperative performance deficits, alteration of con- and may be of variable amounts. Mitoses and necro- sciousness and personality change (Laws et al., 1985; sis are more likely to be found in association with the Piepmeier, 1987; Medbery et al., 1988; Shaw et al., small cell component. Despite their cellular hetero- 1989; North et al., 1990), and more extensive sur- geneity and primitive small cell component, these tu- gery (Leibel et al., 1975; Laws et al., 1985; Shaw et mors are associated with a prolonged survival pro- al., 1989; Soffietti et al., 1989; North et al., 1990) are vided that the tumor is maximally resected. They are associated with an improved outcome. Patients with WHO grade I (Kleihues and Cavenee, 2000). mixed oligoastrocytomas have a better prognosis than do those with pure astrocytomas (Shaw et al., 1991). Oligodendrogliomas, however, have a significantly Central Neurocytoma better long-term survival than mixed oligoastrocy- Neurocytomas are neoplasms of small mature neu- tomas and astrocytomas (Shaw et al., 1997). While rons that may arise within the ventricles (intraven- sometimes included with grade II astrocytomas, in tricular or central neurocytoma) or in the brain pa- general, survival of patients with gemistocytic astro- renchyma (cerebral neurocytoma) (Burger et al., cytoma is similar to those with anaplastic astrocy- 1991; Hassoun et al., 1993; Hessler et al., 1992). The tomas. classic central neurocytoma is a supratentorial, often calcified, mass within the lateral ventricle. It may arise from the septum pellucidum (Kim et al., 1992). SURGICAL TREATMENT AND ITS IMPACT Histologically, this neoplasm is characterized by a diffuse proliferation of small, clear cells with round The belief that most gliomas are infiltrative into ad- nuclei and finely punctate chromatin. Fibrillary zones jacent brain has been strengthened in recent years by may be interspersed and may help to distinguish this studies based on diagnostic imaging and neuroimage- tumor from oligodendroglioma. The tumor is not guided stereotactic biopsies of these tumors and their anaplastic, mitotic figures are rare, and necrosis is surrounding periphery. This, along with technical ad- unusual. It corresponds to WHO grade II. The diag- vances in neurosurgical instrumentation allowing for nostic feature is evidence of neuronal differentiation, precise tissue sampling using image-coupled refer- either immunohistochemically or ultrastructurally ence frames, has had a significant effect on the role (Hessler et al., 1992). Tumor cells are positive for of surgery in managing the patient with a glioma. the neuronal markers synaptophysin and neurofila- In a study correlating results of CT scans with re- ment and negative for glial fibrillary acidic protein. sults of postmortem examination for extent of disease Electron microscopy reveals dense-cored granules in patients with glioblastoma, Burger and colleagues within the cell bodies (Hassoun et al., 1993). These (1988) found tumor beyond the contrast-enhancing diagnostic studies can distinguish neurocytomas from rim at autopsy in most cases that had shown hypo- ependymomas and oligodendrogliomas. With com- dense regions on the scan. However, there was a de- plete removal, neurocytomas seem to behave in a be- lay of several weeks between the imaging study and nign fashion and generally do not recur (Kim et al., death; also the cause of death was the actual tumor 1992). in fewer than 50% of cases. Furthermore, the authors stated that the “methodology to identify neoplastic cells (in adjacent brain) has significant limitations.” OUTCOME VARIABLES Daumas-Duport et al. (1987) optimized tissue fix- ation methods to be able to define in greater detail Several patient and tumor characteristics influence cells that appeared to be neoplastic and had been ob- outcome in those with astrocytomas. The most im- tained stereotactically from peripheral glioma zones. portant prognostic variables are patient age at diag- In addition, this group implanted corresponding nosis, functional status, completeness of resection, specimens into nude mice, and the implants formed and tumor histology. Young age (Leibel et al., 1975; tumors composed of astrocytes. There was no attempt Marsa et al., 1975; Laws et al., 1985; Piepmeier, made to correlate the regions from where these iso- 1987; Medbery et al., 1988; Shaw et al., 1989; North lated tumor cells were taken and patterns of recur- et al., 1990; Whitton and Bloom, 1990), absence of rence in patients on the basis of follow-up imaging 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 100

100 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

studies. Making a qualitative interpretation of neo- It is generally believed that the survival rate is plastic versus reactive astrocytes suffers from the lack higher for patients with gross total tumor resections of specific tumor cell markers. Extending beyond the compared with those with biopsy only. Unfortunately, CT era, biopsy material obtained from hyperintense, no prospective randomized study has ever been done T2-weighted signal regions seen on MR imaging stud- to evaluate the relationship between amount of tumor ies also verified isolated tumor cells outside the con- removed and subsequent influence on recurrence trast-enhancing rim in high-grade glial tumors (Kelly, patterns and survival. Most studies are retrospective. 1993). Reports have been based on the surgeon’s impres- Despite the knowledge base surgeons have ac- sion at the time of surgery, appearance of the quired about the histologic limits of infiltrative postoperative CT or MRI images, or measurement of gliomas, clinical studies have supported a localized perpendicular tumor diameters as seen on contrast- pattern of tumor recurrence. Reports by Liang et al. enhanced neuroimages. In addition, several studies (1991) and Hochberg and Pruitt (1980) documented include pediatric patients and evaluate survival char- recurrent disease within 2 cm of the contrast- acteristics for a combination of adult and pediatric enhancing rim in 85% to 90% of their patients. These patients, without taking into account that age is the two studies were done 10 years apart, and even with strongest prognostic indicator within a histologic sub- improvements in imaging, the conclusions were the group. This approach results in longer survival times same. Similarly, in another study Gaspar et al. (1992) and better overall outcomes for series incorporating found tumor recurrence in all cases within 4 cm of younger patients. the contrast margin and not beyond this imaging line Another common methodologic problem arises of demarcation. when various histologies with different natural histo- ries are combined and evaluated together without ma- nipulating the data to separate individual histologic Low-Grade Infiltrative Gliomas subtypes. In several studies, pilocytic or gemistocytic Controversy continues to exist with regard to extent astrocytomas are included in the study population. of tumor removal and outcome in patients with supra- These heterogeneous populations make the data tentorial low-grade glial tumors, which, in most cir- harder to interpret, and conflicting results are often cumstances, involve the cerebral hemispheres. Sev- obtained in studies evaluating the efficiency of a par- eral points regarding the prognostic effect of tumor ticular treatment option. Currently, there are only a removal remain unclear, and management strategies few hemispheric low-grade glioma series with more are almost always based on class III evidence (i.e., than 75 patients that exclude the pediatric age group evidence provided by expert opinions, nonrandom- as well as gemistocytic and pilocytic histologies. All ized historic controls, or case reports) (Bampoe and of these studies with relatively homogeneous patient Bernstein, 1999; Eddy, 1990). A review by the Guide- populations show a statistically significant advantage lines and Outlines Committee of the American Asso- in terms of survival for low-grade glioma patients who ciation of Neurological Surgeons (1998), based on receive the most extensive resections (Leighton et al., data published between 1966 to 1994 on low-grade 1997; Philippon et al., 1993; Rajan et al., 1993; Soffi- gliomas showed that the only available standard for etti et al., 1989; van Veelen et al., 1998). None of adults with suspected or known supratentorial these studies, however, offers class I evidence (ran- nonoptic pathway low-grade glioma is to obtain tis- domized controlled clinical trial). sue diagnosis before active treatment. Several critical Of additional importance in interpreting surgical issues (e.g., the need for biopsy before observation results is the fact that a large number of low-grade and the effects of observation, resection, and radio- gliomas recur with a more malignant tumor grade therapy on outcome) remained unclear. There were (Afra et al., 1978; Laws et al., 1985; North et al., 1990; no standards or guidelines that could be supported Piepmeier et al., 1987). by class I evidence (randomized controlled clinical We previously analyzed our database, using a com- trial) or class II evidence (case–control or cohort puterized image analysis technique (Duong et al., studies). Therefore, all current management strate- 1992) that yields a quantitative, volumetric determi- gies were considered as practice options based on nation of preoperative and postoperative tumor vol- class III evidence. umes, including percent of resection, to ascertain 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 101

Primary Cerebral Tumors 101

whether these parameters influence recurrence pat- grade appearance on the initial imaging studies terns. By measurement we found a mean preopera- (Recht et al., 1992). Although controversy will con- tive tumor volume of 26.6 cc (based on T2-weighted tinue about biopsy versus radical resection as well as MR signal) and found no tumor recurrence (mean about the timing of surgery, such data as are pre- follow-up, 42 months) when the preoperative volume sented in this section show a trend toward an in- was 10 cc (Berger et al., 1994). The likelihood of creasing risk of early and malignant recurrence when recurrence was greater and the time to tumor pro- a conservative approach is taken once the presump- gression shorter when the volume increased (e.g., tive diagnosis of low-grade glioma is made. 41% recurrence rate at a median 30 months for pa- Very few data exist on what role extensive surgery tients with tumors 30 cc). Other studies have found plays in determining the outcomes of patients with radical resection to favorably influence outcome for oligodendroglioma. Survival was positively influenced patients who have a low-grade glioma. Data on the by total tumor resection in some studies (Chin et al., amount of resection were based on the operative re- 1980; Shaw et al., 1992; Varma et al., 1983; Mork et port or on the qualitative assessment of the postop- al., 1985; Celli et al., 1994; Schiffer et al., 1997). This erative scans, without an attempt to use volumetric contrasts with studies by Sun and colleagues (1988) analysis (Guthrie and Laws, 1990; Laws et al., 1985; and others (Kros et al., 1994; Daumas-Duport et al., North et al., 1990). These data are similar to those 1997) in which survival was not affected by the ex- reported by the Japanese Brain Tumor Registry dur- tent of resection. In one analysis, radiotherapy did ing the past 25 years (1992), which include 5 year not affect outcome following total tumor removal, survival rates of 48% for biopsy and subtotal resec- implying that, unlike other infiltrative low-grade tions of 50% or less versus a 70% survival for those gliomas, oligodendroglioma may be more amenable with greater than 90% resections (Brain Tumor Reg- to a complete tumor resection because of its some- istry of Japan, 1992). what circumscribed nature associated with macro- These data also parallel studies that evaluated post- scopically distinct margins (Lindegaard et al., 1987). operative tumor volume and survival time for high- As with all low-grade gliomas, the surgical strategy grade gliomas (Levin et al., 1980b). Using volumet- should favor radical tumor removal when at all fea- ric measurements (based on tumor measurements in sible. 3 intersecting directions), it was found that when the Finally, because low-grade tumors tend to have an postoperative tumor volume was 10 cc the chance indolent growth pattern, they are often associated with of recurrence was 46% (median time to tumor pro- seizure activity that is infrequently not refractory to gression, 30 months) compared with a tumor volume medical care. This is especially true for oligoden- of 10 cc, which had a chance of recurrence of 15% drogliomas, gangliogliomas, and astrocytic gliomas (median, 50 months); there were no recurrences fol- (Arseni and Petrovici, 1971; Hirsch et al., 1989; lowing complete resection at a mean follow-up time Pilcher et al., 1993). A controversial surgical issue of 54 months. involves the role of intraoperative electrocorticogra- The timing of surgical intervention following a di- phy during tumor resection used to minimize post- agnosis based on imaging studies is also controver- operative seizures. For most patients with intermittent sial, with some studies recommending immediate seizures that are well controlled with antiepileptic intervention (Laws et al., 1985; Morantz, 1987; drugs, removing the tumor alone without seizure Weingart et al., 1991) and others advocating a less mapping is sufficient and results in good control of urgent approach (Cairncross and Laperriere, 1989; tumor-associated epilepsy (Cascino et al., 1990; Cas- Recht et al., 1992). One study that quantified the tu- cino, 1990; Franceschetti et al., 1990; Goldring et al., mor area as being less than or greater than 25 cc 1986; Hirsch et al., 1989). In some low-grade glioma found a significant difference in survival when patients patients who have intractable seizures, tumor removal with smaller tumors were operated on earlier by itself often will not diminish seizure activity (Awad (Shibamoto et al., 1993). Another retrospective re- et al., 1991; Cascino et al., 1990; Spencer et al., 1984; view of patients whose surgery was deferred until dis- Zentner et al., 1997). Our approach has been to use ease progressed demonstrated that in nearly 50% of electrocorticography during the tumor resection to the cases malignancy was documented at the time of identify separate seizure foci adjacent to the tumor the biopsy even though the lesions had a classic low- nidus (Berger et al., 1993). Of our adult patients, 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 102

102 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

88% became seizure free with (47%) or without section, especially when the tumor is deep or attached (41%) antiepileptic drugs. Even the few patients who to subcortical functional tracts. Solid tumors should had persistent seizure activity while taking medica- be aggressively resected, however, unless excessive tion benefited from surgery and had fewer seizures or unacceptable morbidity would result. The long- that were often less intense. Each child and adoles- term results without radiotherapy or chemotherapy cent in our series was seizure free and not taking are excellent following a complete resection. Never- medication except for two children who continued theless, as a precautionary measure we advocate care- therapy with antiepileptic drugs because they had a ful and routine follow up every 6 to 12 months for few seizures following surgery. These data are also the first several years after surgery. The 10 year sur- supported by other investigators who use electrocor- vival rate for a completely resected pilocytic hemi- ticography to control postoperative seizures in this spheric astrocytoma, verified with postoperative particular patient population (Awad et al., 1991; imaging studies, should approach 100% (Wallner et Drake et al., 1987; Gonzalez and Elvidge, 1962; Ras- al., 1988a; Shaw et al., 1989). This favorable out- mussen, 1975; Ribaric et al., 1991). come also pertains to adult patients with this diag- nosis (Garcia and Fulling, 1985). Although incom- pletely resected tumors may have a protracted course, Juvenile Pilocytic Astrocytomas and Other recurrence is likely over a period of several years, Noninfiltrative Gliomas necessitating routine follow up (Palma and Guidetti, The goal of surgical intervention in patients with 1985). cerebral hemispheric juvenile pilocytic astrocytomas Thalamic juvenile pilocytic astrocytomas tend to be is to evacuate the cyst contents and remove all con- well circumscribed and may be associated with a cyst trast-enhanced tissue documented on the preopera- (Wald et al., 1982). Anterior thalamic tumors are best tive CT or MRI studies. This includes the mural nod- resected via a parasagittal incision in front of the pre- ule and any or all parts of the cyst that are contrast motor cortex, thus entering the ventricle anteriorly. enhanced. Failure to remove the lesion while simply These lesions displace the internal capsule laterally, draining the cyst will uniformly result in re-accu- whereas posterior thalamic masses push the poste- mulation of the cyst contents. Several studies have rior limb of the internal capsule forward. Therefore, documented that the cyst wall is gliotic without neo- the best approach in the latter circumstance is plastic cells when it does not enhance with contrast through the posterior ventricle (i.e., the atrium). It administration (Maiuri, 1988; Tomita et al., 1986). is critical to localize the capsular motor fibers using If the cyst wall enhances, it must be completely re- subcortical stimulation methods to avoid morbidity sected to avoid tumor recurrence (Morota et al., while maximizing the extent of resection (Berger et 1990). The mural nodule is quite discrete and has al., 1990). An alternative surgical approach is to vol- a reddish-brown appearance. It is friable and mod- umetrically resect the pilocytic tumor using com- erately bloody. At the end of the resection, the adja- puter-assisted stereotactic methods (Lyons and Kelly, cent, noninfiltrated white matter should be clearly 1992). This technology is very accurate and effective, visible. As expected, there is a distinct microscopic but is expensive and is not available to most neuro- margin between the tumor and the contiguous white surgeons. matter (Garcia and Fulling, 1985). Because the cyst Because subependymal giant cell astrocytomas are is often in continuity with the ependymal surface of always located within the anterior horn of the lateral the ventricular system, we do not recommend, as do ventricle, the preferred approach to them depends on some authors, that the cyst be fenestrated into the the ventricular size. Ventricular dilatation is the rule ventricle (Palma et al., 1983; Mercuri et al., 1981). for symptomatic patients because of the relationship Following removal of the tumor, the cyst will not re- between the tumor and the foramen of Monro. There- cur. In addition, debris from the resection cavity may fore, in most circumstances a frontal transventricu- seep into the ventricle and cause a communicating lar approach is performed to gain the necessary ex- hydrocephalus. posure, taking advantage of the large ventricles. With solid juvenile pilocytic astrocytomas it is However, some surgeons still recommend a tran- somewhat more difficult to achieve a gross total re- scallosal exposure because there is less brain re- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 103

Primary Cerebral Tumors 103

traction and less potential risk of postoperative control of recurrence and seizures (VandenBerg, epilepsy, although this latter point is debatable 1993; Daumas-Duport, 1993). (McLaurin and Towbin, 1986). Regardless of ap- proach, virtually all surgeons agree that these lesions High-Grade Gliomas should only be removed when symptoms of ventric- ular obstruction exist. Unlike for low-grade gliomas, surgeons have been The supratentorial ependymoma and pleomorphic much more willing to use cytoreductive surgery for xanthoastrocytoma appear grossly as an encapsulated bulky malignant gliomas to alleviate mass effect as- mass distinct from the adjacent brain. However, the sociated with symptoms and signs of increased in- latter lesion tends to have more of a contiguous re- tracranial pressure. Moreover, areas of necrosis are lationship with the leptomeninges (Kepes et al., more amenable to surgical aspiration than is firm 1989), making its resection more difficult, especially white matter infiltrated with a low-grade glial tumor. when the tumor occurs within the sylvian fissure. With However, the idea that the extent of resection suffi- both lesions, a superficial (i.e., pial-based) attach- ciently influences survival of patients with malignant ment should alert the surgeon that these tumors are gliomas is still not universally accepted; thus, treat- circumscribed and amenable to complete resection. ment recommendations will vary among surgeons. One review of the literature, however, revealed that For example, in a retrospective study that analyzed even with complete resection of a pleomorphic xan- outcome after a limited (i.e., stereotactic) biopsy, the thoastrocytoma, there is a 30% chance of recurrence median survival times for patients with glioblastoma (Macaulay et al., 1993); thus, routine surveillance and anaplastic astrocytoma were each less than 30 with postoperative imaging studies is mandatory. weeks; if deep midline tumors were excluded, the me- While there is universal agreement among sur- dian survivals improved to 47 and 129 weeks, re- geons that cerebral hemispheric gangliogliomas spectively. In this study, the few patients who received should be completely resected (Silver et al., 1991; Ot- cytoreductive surgery exhibited no statistical differ- subo et al., 1992), controversy exists regarding the ence in outcome when biopsy was compared with a extent of resection as it pertains to optimal tumor more aggressive surgery, although in the latter situa- control. If the literature is reviewed with respect to tion no information was provided about the extent of seizure control and extent of surgical resection, it ap- resection (Coffey et al., 1988). Kelly (1990) demon- pears that a complete tumor resection results in a strated a survival advantage for those patients with 50% seizure-free condition, a rate not greatly dis- glioblastoma who underwent complete stereotacti- similar to that with incomplete tumor removal (Had- cally guided resection (48 weeks) over those patients dad et al., 1992). However, when intraoperative elec- who had a biopsy or a standard craniotomy and re- trocorticography is used to map epileptogenic foci so section (30 and 38 weeks, respectively). Comparing as to include them as part of the operative procedure, results from these two studies suggests that biopsy is the likelihood that patients with a gross total tumor not a substitute for a good tumor resection. resection would achieve complete seizure control was A number of radiographic studies have shown an 92% (Pilcher et al., 1993). This outcome is signifi- inverse relationship between postsurgical tumor vol- cantly better than that for removing only the tumor. ume and survival (Levin et al., 1980b; Wood et al., Even when the tumor was removed and the epileptic 1988). Volumetric analysis of 107 operations on pa- focus incompletely resected, a 95% reduction in post- tients with hemispheric glioblastoma multiforme operative seizures was demonstrated. Therefore, in- showed that patients who underwent a total tumor re- traoperative electrocorticography is indispensable for section with no residual disease had a significantly patients who have intractable epilepsy associated with longer time to tumor progression and survival (Keles gangliogliomas or other similar, yet less frequently et al., 1999). In this study, as volume of residual tu- occurring, lesions (e.g., desmoplastic infantile gan- mor increased, a shorter time to progression and a gliogliomas and dysembryoplastic neuroepithelial tu- longer survival were observed. Most cooperative mors). For the latter two noninfiltrative, circum- prospective multimodality trials show a survival ad- scribed tumor types, the goal of surgery is complete vantage for subtotal and gross total resection com- resection, which positively influences outcome for pared with biopsy. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 104

104 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

In addition to the concept that an aggressive re- tactic procedure should be directed at any contrast- section, as opposed to limited tumor removal, will enhancing tissue when present. Because of the in- positively influence outcome (Chang et al., 1983; tracranial hypertension associated with a large tumor Levin et al., 1980b; Walker et al., 1978; Dinapoli et mass, a biopsy may result in rapid neurologic dete- al., 1993), quality of life also becomes a critical is- rioration (Ross et al., 1991); therefore, the patient sue, which is influenced by surgery. It has been rec- should be observed in the intensive care unit ognized for quite some time that removal of a very overnight following biopsy. small tumor or biopsy specimen may result in ex- In summary, radical tumor resections performed cessive morbidity from swelling or hemorrhage under the proper conditions will usually preserve or (Fadul et al., 1988). Studies (Ammirati et al., 1987; improve a patient’s functional status and augment Ciric et al., 1987; Keles et al., 1999) have demon- survival for a longer period of time than when more strated that patients usually improve neurologically or limited tumor resections are performed (Mornex et remain the same with aggressive tumor removal. This al., 1993; Chandler et al., 1993; Kornblith et al., 1993; has also been our experience, and, combined with Kaplan, 1993). This applies not only to glioblastoma neurophysiologic cortical and subcortical mapping but to other anaplastic (including gemistocytic), as- methods, the likelihood of creating a permanent trocytic, and oligodendroglial tumors (Krouwer et al., deficit with current surgical techniques should be less 1991; Winger et al., 1989; Prados et al., 1992a). How- than 3% to 5%. ever, in the anaplastic subtypes, the advantage of sur- The role of radical resection for children with ma- gery must be considered to be strongly influenced by lignant supratentorial gliomas has not been as rigor- age at diagnosis because that is a known key prog- ously examined as for adults. In the few studies that nostic variable for malignant gliomas (Curran et al., address this issue without any reference to how ex- 1992a; Winger et al., 1989) in terms of tumor biol- tent of resection was determined, it appears that chil- ogy and in biasing the surgeon to be more aggres- dren who received a radical resection rather than a sive. biopsy survived longer (Allen et al., 1986; Artico et al., 1993). In a study evaluating extent of resection Ependymomas on survival conducted by the Childrens Cancer Group (CCG-945), the 5 year progression-free survival (PFS) Few reports are available that describe the rationale rates for anaplastic astrocytoma were 44% following for surgical resection of supratentorial ependymo- gross total resection and 22% for less radically re- mas. Similar to other tumors that arise within the ven- sected tumors; for glioblastoma the 5 year PFS rates tricle and extend into the cerebral hemispheres, sur- were 26% for gross total resection and only 4% for gical excision is designed to relieve mass effect and children with less radical resections (Wisoff et al., unblock the ventricular system. By itself, total tumor 1998). removal for ependymomas and mixed ependymoma Gliosarcoma continues to be a difficult problem astrocytomas will often result in long-term survival for the surgeon because of its highly invasive nature. without the need for radiation (Palma et al., 1993). Often, the lesion may appear somewhat circum- However, this is not the case with the malignant vari- scribed (Maiuri et al., 1990), but it is extremely vas- ant of this lesion, which requires focal radiation and cular and frequently invades the overlying dura sometimes chemotherapy to achieve similar results. (Morantz et al., 1976). Too few studies have been Unfortunately, most studies that comment on survival done to determine whether the extent of surgery in- in relation to extent of tumor removal do so without fluences survival for this tumor type, although the data separating supratentorial from fourth ventricular tu- will likely be similar to those for glioblastoma. mors. Nonetheless, there is a distinct survival advan- Gliomatosis cerebri is mentioned in this section, tage for a radical tumor resection in the majority of although it is an unusual finding. These very exten- series (Kovalic et al., 1993; Vanuytsel et al., 1992; sive, diffusely infiltrating gliomas usually present with Undjian and Marinov, 1990; Papadopoulos et al., symptoms and signs of increased intracranial pres- 1990). sure. A biopsy is mandatory and will suffice along with In a series from Boston that combined tumors in the MRI sequences (Spagnoli et al., 1987). There is both supratentorial and fourth ventricular locations, no indication for extensive resection, and a stereo- it is important to note a discrepancy between the op- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 105

Primary Cerebral Tumors 105

erative report and the postoperative imaging study gliomas; these studies defined important clinical pa- (i.e., overestimation in 33% of the cases) (Healey et rameters that predicted a good outcome (survival ad- al., 1991). This discrepancy will certainly change as vantage and an improved quality of life) following re- MRI replaces CT scans in the postoperative analyses operation. Prognostically significant factors included of future studies. In the Boston study, the survivals at duration of time from initial resection to tumor pro- 10 years after surgery were 75% for those patients gression, age, and Karnofsky performance status without radiographic evidence of residual tumor and (KPS) (Wilson, 1975; Young et al., 1981; Salcman et 0% for patients with residual disease. Two other se- al., 1982). Overall, patients with glioblastoma and ries, incorporating multivariate analyses, reported anaplastic astrocytoma may expect median survivals significantly longer PFS and overall survival rates for of 36 weeks and 88 weeks, respectively, following re- patients who underwent gross total resections operation and subsequent chemotherapy (Harsh et (Rousseau et al., 1993; Pollack et al., 1995). al., 1987). Somewhat better figures have been re- In the Italian Pediatric Neuro-oncology Group ported by Berger et al. (1992) when the interval of study, which included 92 patients, radical surgery was time between initial surgery and reoperation is long, the only prognostic factor found in multivariate anal- the KPS is 70 or greater, and the age of the patient at ysis to have a statistically significant effect in predict- the time of reoperation is less than 60 years. They ing both PFS and overall survival (Perilongo et al., found for glioblastoma a mean survival time of ap- 1997). Recently, in a multi-institutional study, in- proximately 70 weeks when the Karnofsky score be- cluding patients from 11 United States pediatric on- fore reoperation was 70 or better; for a small group cology centers, less than gross total resection was a of patients with anaplastic astrocytoma, the mean sur- significant adverse risk factor for event-free and over- vival time after reoperation was 135 weeks. The all survival (Horn et al., 1999). interval between initial surgery and diagnosis to re- Two studies dispute these findings, however. The operation was predictive: patients with disease pro- authors claim that, in their collective experience, rad- gression within 6 months of the initial operation had ical surgery is not prognostically important (Goldwein a mean survival of 40 weeks after reoperation versus et al., 1990; Rawlings et al., 1988). Only in the Gold- 150 weeks if the tumor did not recur until 1 year fol- wein et al. (1990) study were postoperative imaging lowing the original surgery. From these studies we scans obtained. Notwithstanding these two studies, conclude that patients who benefit most from reop- current neurosurgical management of ependymomas, eration with a malignant glioma are younger than 60 regardless of histology, involves achieving a radical years old, have a good functional status, and have a resection of the lesion without removing more than prolonged period of disease stability following the ini- 5 to 10 mm of adjacent white matter. The prevailing tial operation. consensus is that greater extent of resection seems to improve outcome and can be maximized by careful preoperative planning, a meticulous but aggressive RADIOTHERAPY AND surgical strategy, and excellent postoperative care CHEMOTHERAPY TREATMENT (Smyth et al., 2000).

Approaches for Low-Grade RATIONALE FOR REOPERATION Infiltrating Gliomas Radiotherapy With heightened interest in treating patients who have failed first-line therapy, the role of reoperation must Until recently, there have been no randomized trials be critically evaluated in terms of the benefits versus to clarify many of the issues surrounding the treat- risks. Clearly, for those patients who have mass ef- ment of low-grade astrocytomas; thus, therapeutic de- fect, cytoreductive surgery is an important therapeu- cisions generally have been based on information ob- tic modality that can be carried out if the patient is tained from retrospective reports. Some series do not to have a good quality of life during the retreatment separate patients with pilocytic astrocytomas from phase. Historically, a few important retrospective those with nonpilocytic tumors, although this dis- studies have analyzed patients who had malignant tinction is important in the therapeutic decision-mak- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 106

106 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Table 3–2. Supratentorial Astrocytoma: 5 and 10 Year Survival Rates No. of Survival Rate (%) Author Cases Treatment 5 Year 10 Year Levy and Elvidge (1956) 42 S 26 NA 45 S RT 36 NA Bouchard (1980) 105 S RT 49 NA Leibel et al. (1975) 35 S 23 15 49 S RT 35 24 Marsa et al. (1975) 40 S RT 40 20 Fazekas (1977) 18 S 22 NA 36 S RT 50 18 Laws et al. (1984) 461 S, S RT 36 20 167 S 34 NA 74 S RT 49 NA Garcia et al. (1985) 23 S 21 10 57 S RT 50 25 Medbery et al. (1988) 50 S RT 45 32 50 Gy 55 45 50 Gy 22 13 Shaw et al. (1989) 19 S 32 11 35 S RT 53 Gy 68 39 67 S RT 53 Gy 47 21 Whitton and Bloom (1990) 60 S RT 36 26 North et al. (1990) 77 S, S RT 55 43 52 45–59 Gy 66 NA McCormack et al. (1992) 53 S RT 64 48 Shibamoto et al. (1993) 71 S RT 54 (age 30) 33 Philippon et al. (1993) 179 S 65 NA S RT 55 NA Eyre et al. (1993) 54 S RT 50 30 Karim et al. (1996) 391 S RT (45 Gy) 58 NA S RT (59.4 Gy) 59 NA Leighton et al. (1997) 167 S, S RT 72 50 Shaw et al. (1998) 203 S RT (50.4 Gy) 73 NA S RT (64.8 Gy) 68 (p 0.57) NA Karim et al. (1998) 311 S immediate RT 63 (PFS 44%)* NA S delayed RT 66 (PFS 37%) NA S, surgery alone; S RT, surgery and postoperative irradiation; N, data not available; PFS, 5 year progression-free survival. *Difference significant, p 0.02.

ing process. Representative survival rates for supra- Median survival times in recent series range from tentorial astrocytomas treated with surgery or surgery 7.2 to 12.9 years, raising concerns over the value to- and radiation therapy are summarized in Table 3–2. day of the older literature in making treatment deci- The outcomes of patients diagnosed and treated in sions. The improved outcome appears to be related the era of modern neuroimaging are notably better to the earlier diagnosis of neurologically intact pa- than those reported in older studies when the condi- tients who exhibited only seizures at the time of di- tions for making a diagnosis were less sophisticated agnosis (Piepmeier, 1987; Cairncross and Laperriere, (Philippon et al., 1993). 1989; McCormack et al., 1992; Philippon et al., 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 107

Primary Cerebral Tumors 107

1993). As mentioned, survival is significantly affected Mercuri et al., 1981). In contrast, the outcomes by the age of the patient at diagnosis, presence of in adult patients after total or radical subtotal resec- seizures at presentation, KPS, as well as whether the tion have been found in some series to be similar to lesion enhances with IV contrast administration. Us- those of patients undergoing less extensive surgery ing a recursive partition analysis, Bauman et al. (Fazekas, 1977; Garcia et al., 1985; Shaw et al., (1999a) were able to distinguish four distinct prog- 1989). Thus, some authors have recommended post- nostic subgroups of adult, supratentorial, low-grade operative irradiation after complete resection in glioma (including oligodendroglioma) patients: adults (Garcia et al., 1985; Shaw et al., 1989, 1991), group 1 (70, age 40 years), median survival time whereas others suggest that radiation therapy be with- 1 year; group II (70, age 40 years, the tumor en- held until there is evidence of tumor recurrence hances on CT/MRI), median survival time 3.8 years; (Leibel et al., 1975; Morantz, 1987; Soffietti et al., group III (70, age 18 to 40 years or 70, age 1989; Wara, 1985). 40, the tumor does not enhance), median survival Postoperative irradiation appears to prolong the time 7.2 years; and group IV (70, age 18 to 40 survival of patients with incompletely excised supra- years), median survival time 10.7 years. tentorial astrocytomas (Fazekas, 1977; Garcia et al., Using a BrdU labeling index, Hoshino et al. 1985; Leibel et al., 1975; Levy and Elvidge, 1956; (1988) found that 10% (3/29) of patients with tu- Shaw et al., 1989). On average, the 5 and 10 year mors exhibiting a labeling index 1% died within a survival rates for patients who receive radiation ther- 3.5 year follow-up period, whereas 50% (9/18) of apy are 52% and 26%, respectively, compared with patients with a labeling index 1% died within the 26% and 12% for patients who do not (Table 3–2). same time interval. This concept has been corrobo- Leibel et al. (1975) reported the outcomes of 108 rated in subsequent studies as well (Ito et al., 1994; patients of all ages with incompletely resected lesions. Lamborn et al., 1999). These data emphasize that Of those, 71 patients received postoperative irradia- low-grade astrocytomas may vary considerably in tion and 37 did not. Only 6% of tumors were pilo- their biologic behavior even though they have simi- cytic, whereas 78% arose in the cerebral hemi- lar histologic appearances. Repeat biopsy specimens spheres. The age distribution and performance status and autopsy studies indicate that over time at least of the patients in the two groups were similar. The 5 50% of astrocytomas transform into more anaplas- year recurrence-free survival rate after incomplete re- tic lesions (Laws et al., 1985; Muller et al., 1977; section alone was 19%, whereas it was 46% after in- Rubinstein, 1972; Soffietti et al., 1989). Thus, complete resection and irradiation. For adult patients, whereas astrocytomas are generally considered to be the 5 year survival rate was 10% after surgery alone, benign, slow-growing tumors, in some cases they compared with 32% when radiation therapy was may be highly lethal. added. The survival of patients undergoing incom- Opinions differ regarding the need for postopera- plete resection and postoperative irradiation was su- tive irradiation when a complete surgical resection perior to that of patients treated by surgery alone at has been performed. The 5 year recurrence-free sur- all follow-up time intervals from 3 to 20 years. All pa- vival rates for patients with supratentorial astrocy- tients not receiving radiation therapy had died by 20 tomas or mixed oligoastrocytomas who undergo to- years, whereas 23% of those who did remained alive tal or radical subtotal tumor resection range from and free of disease. 52% to 95% (Fazekas, 1977; Garcia et al., 1985; In their series of adult patients with cerebral as- Hirsch et al., 1989; Leibel et al., 1975; Medbery et trocytomas, Garcia et al. (1985) reported an actuar- al., 1988; Shaw et al., 1989). The variations in out- ial 5 year survival rate of 21% after resection alone come reflect prognostic differences related to age, the compared with 50% after resection and postopera- inclusion of patients with radical subtotal resections, tive irradiation ( p 0.02). Shaw et al. (1989) and the reliance on retrospective evaluations of op- found that patients receiving high-dose postoperative erative reports to determine the completeness of re- irradiation had a significantly longer survival than section in the era before the availability of CT and those receiving low-dose irradiation or surgery alone. MRI studies. Because recurrences are infrequent in The 5 year survival rate was 68% for patients receiv- children whose astrocytomas have been completely ing a dose of at least 53 Gy, 47% for those receiving resected, postoperative irradiation is generally not less than 53 Gy, and 32% for those receiving none recommended (Hirsh et al., 1989; Nishio et al., 1989; ( p 0.04). 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 108

108 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Older patients appear to benefit most from post- tibody to BrdU incorporation into tumor cells operative irradiation (Garcia et al., 1985; Laws et al., (Hoshino et al., 1988) and the MIB-1 antibody to the 1985; Shaw et al., 1989). Shaw et al. (1989) found Ki-67 protein (Shibuya et al., 1993). The develop- that patients 35 years of age or older who received ment of such immunohistochemical and molecular postoperative irradiation had 5 and 10 year survival markers to better predict the prognosis for an indi- rates of 67% and 45%, respectively, compared with vidual patient may provide an opportunity for earlier 37% and 5% for patients treated with low-dose irra- intervention and improvement in outcome for the diation or surgery alone ( p 0.008). For patients prognostically more unfavorable subsets of patients. 34 years of age or less, the 5 year survival rate was Limited radiation fields are used in the treatment 70% with high-dose irradiation compared with 53% of low-grade astrocytomas. Recurrences are nearly with low-dose irradiation or surgery alone ( p always found at the original primary tumor site (Med- 0.69). bery et al., 1988; Shaw et al., 1989), and there is no The fact that low-grade astrocytomas are diag- difference in the survival distributions or patterns of nosed earlier in their natural history has raised ques- failure between patients receiving partial and those tions regarding whether radiotherapy should be ad- receiving whole-brain irradiation (Medbery et al., ministered immediately after surgery or be delayed 1988; North et al., 1990; Shaw et al., 1989). Fields until recurrence or progression has been demon- should encompass the T2-weighted MRI abnormality, strated. It is generally agreed that patients with in- which tends to be larger than the CT-defined le- tractable seizures or those with large, progressive, sion, with a margin of 1 to 2 cm (Kun, 1992; Shaw symptomatic, unresectable or incompletely resected et al., 1991; Karlsson et al., 1992). Complex three- tumors should undergo immediate radiotherapy. dimensional treatment plans are used whenever ap- However, radiotherapy is commonly deferred in pa- propriate to limit the high-dose volume and to mini- tients with medically controlled seizures who present mize the risk of long-term radiation sequelae (Ellen- with asymptomatic, indolent tumors (MacDonald, berg et al., 1987). 1994). Proponents of this approach argue that it is The optimal dose of radiation for astrocytomas is unclear whether early irradiation provides an out- not well defined. The standard dose for adult patients come advantage over delayed irradiation or whether is 54 Gy, administered in daily fractions of 1.8 to 2.0 such treatment delays or prevents tumor dedifferen- Gy. This dose level is relatively conservative so as to tiation (Cairncross and Laperriere, 1989; Whitten and decrease the risk of excessive treatment-related mor- Bloom, 1990). bidity (Bloom, 1982). At this dose level about 75% This issue was clarified in a clinical trial conducted of patients will improve neurologically, and the max- by the European Organization for Research and imum radiographic improvement occurs within a me- Treatment of Cancer (EORTC) and British Medical dian of 2.8 months (Bauman et al., 1999b). Two ran- Research Council Brain Tumour Working Party domized trials have shown that higher dose levels do (MRC). Patients with low-grade astrocytomas (65%), not improve patient outcome (at least at 5 years). In oligodendrogliomas (25%), or mixed tumors (10%) a trial conducted by the EORTC, patients were ran- were randomized to receive immediate postoperative domized to receive 45 Gy in 25 fractions or 59.4 Gy irradiation to a dose of 54 Gy or no further treatment in 33 fractions. No difference in survival was observed until there was evidence of disease progression. between the two dose levels. The 5 year survival rates Among those in the deferred treatment arm, 65% of were 58% for 45 Gy and 59% for 59.4 Gy. The PFS patients received subsequent radiotherapy, 19% un- rates were also similar (47% versus 50%, respec- derwent surgery and/or chemotherapy, and the re- tively) (Karim et al., 1996). mainder received only supportive care. A preliminary Similarly, a combined North Central Cancer Treat- analysis of the study demonstrated that immediate ir- ment Group (NCCTG), Radiation Therapy Oncology radiation significantly improved the 5 year PFS (44% Group (RTOG), and Eastern Cooperative Oncology versus 37%, p 0.02). However, there was no im- Group (ECOG) trial randomized adult patients with provement in overall 5 year survival (63% versus supratentorial astrocytomas to receive 50.4 Gy in 28 66%) (Karim et al., 1998). The outcomes of patients fractions or 64.8 Gy in 36 fractions. As in the EORTC with astrocytomas strongly correlate with the prolif- study, the 5 year survival rates were similar for the erative potential of the tumor as measured by an an- two dose levels studied: 73% for 50.4 Gy and 68% 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 109

Primary Cerebral Tumors 109

for 64.8 Gy (Shaw et al., 1998). An increase in func- achieved even in the absence of objective radio- tional sequelae (Kiebert et al., 1998) and radiation graphic responses; and (3) radiation therapy can be necrosis (Shaw et al., 1998) was observed in patients delayed for up to several years with the use of such treated in the high-dose arms of these studies. These therapeutic strategies, either alone or combined with data support the use of lower radiation levels for low- judicious surgical debulking. grade gliomas. The actual effect of chemotherapy on overall sur- The 4 year overall survival of children with cere- vival and on quality of survival will take many years bral hemispheric astrocytomas is 90% (Gajjar et al., to establish. At least for children with low-grade as- 1997). Radiation therapy is likely to lead to unac- trocytomas, it is hoped that to address this question ceptable sequelae in children younger than 3 to 5 the Pediatric Oncology Group and the Childrens Can- years of age. Therefore, in this age group, radiation cer Group (of North America) will conduct random- therapy is postponed for as long as possible provided ized trials of observation versus chemotherapy in that no significant neurologic deficits or changes in- children less than 5 years of age who have newly di- dicative of rapid tumor progression are present. Pe- agnosed low-grade astrocytomas. To date, prospec- riodic imaging studies are performed to monitor the tive trials of chemotherapy for astrocytoma in adults disease status. The radiation dose is reduced to 50 have been impossible to conduct because of the small Gy for children under 5 years of age. Management numbers of patients with low-grade astrocytoma decisions are also frequently individualized in older available for study. children with incompletely resected astrocytomas With increasing patient age, histologically low- who, compared with adults, have a generally better grade astrocytomas assume a biologic malignancy prognosis, a less pronounced survival improvement that leads many clinicians to treat those who present with postoperative irradiation, and a greater risk of with these tumors after age 45 years more aggres- late radiation sequelae (Hirsch et al., 1989; Mercuri sively than younger patients with higher dose radia- et al., 1981; Nishio et al., 1989; Yule et al., 2001). tion (60 Gy instead of 54 Gy) and to add chemo- therapy. Other clinicians adhere to the histologic criteria of malignancy and withhold chemotherapy Chemotherapy until tumor recurrence. Even more controversial is Low-grade astrocytomas account for 40% to 50% of the use of chemotherapy for those patients under the brain tumors in childhood. The use of chemotherapy age of 45 years. In a Southwest Oncology Group trial, in the management of children and adults with low- patients with incompletely excised low-grade gliomas grade astrocytomas is currently evolving. It is unclear were randomized to receive radiation therapy alone how great an effect chemotherapy will have on pa- (n 19; median age 36 years) or radiation therapy tients with low-grade astrocytomas as these tumors and CCNU (n 35; median age 39 years). The me- have a history of being relatively indolent, and un- dian survival time of the 54 evaluable patients was 4.4 certainty remains as to the timing of radiation ther- years with no difference between the two treatment apy as well as chemotherapy. Because of concern arms (Eyre et al., 1993). about the toxic effects of radiation in young children, Based on the observation of Piepmeier (1987) that several studies have focused on the use of chemo- patients with contrast-enhancing low-grade astrocy- therapy in an attempt to delay the need for radiation tomas have shorter survival times than those with therapy in young children with low-grade astrocy- non–contrast-enhancing low-grade astrocytomas tomas in general (Pons et al., 1992; Packer et al., (3.9 versus 7.8 years), Levin et al. (1995) treated a 1993) and optic/chiasmatic/hypothalamic astrocy- subset of the CT-based contrast-enhanced low-grade tomas in particular (Packer et al., 1988). These stud- glioma group with aggressive chemoradiation and ies have each used a weekly “back-bone” therapy of adjuvant chemotherapy. Twenty-two consecutive pa- IV vincristine, with the addition of actinomycin-D, or tients received BrdU during radiation therapy fol- etoposide, or carboplatin. Results have demonstrated lowed by combination chemotherapy with procar- that (1) objective radiographic responses can be ob- bazine, lomustine, and vincristine (PCV) for 1 year served in patients with either recurrent or newly di- after irradiation (Levin et al., 1995). Seventy percent agnosed low-grade astrocytomas; (2) prolonged dis- of patients survived nearly 7 years. As a result of this ease stabilization, often for several years, can be study, during the last 10 years, Levin et al. have treated 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 110

110 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

most histologically low-grade infiltrating astrocy- When the recurrence is either a low- or mid-grade tomas in adults younger than 45 years with three astrocytoma, chemotherapy is frequently beneficial, courses of lomustine and procarbazine followed by although cure at that point is not a likely outcome. limited-field radiation. Controversy surrounds this The drugs and schedules cited in the section on practice. anaplastic gliomas and in Tables 3–3 and 3–4 should One of the problems with adjuvant chemotherapy be referred to for chemotherapeutic approaches for programs for patients with low-grade astrocytoma is the patient with recurrent low-grade astrocytoma. that it is difficult to conduct randomized trials be- cause of the relative rarity of the tumor in the gen- eral population, the bias of clinicians with respect to Juvenile Pilocytic Astrocytomas its management, and the length of time required to conduct trials. When low-grade astrocytomas recur, Radiotherapy approximately 50% do so with the original low-grade Because the long-term survival of patients with supra- histology, whereas the rest recur as a more aggres- tentorial juvenile pilocytic astrocytomas approaches sive anaplastic astrocytoma or glioblastoma multi- 100% when a complete or near-complete (90%, forme (Afra et al., 1978; Laws et al., 1984; Muller et equivalent to 1 log of cells) surgical resection has al., 1977; Rubinstein, 1972; Soffietti et al., 1989). been performed, postoperative irradiation is not in-

Table 3–3. Single-Agent Chemotherapy for Recurrent and Progressive Supratentorial Glioblastoma Multiforme and Anaplastic Astrocytomas GBM, GBM, AA, AA, Treatment % R SD MTP (Weeks) % R SD MTP (Weeks) Carmustine (Levin, 1985; Wilson et al., 1976) 29 22* 64 22* PCNU (Levin et al., 1984) 33 8* 69 28* Procarbazine (Rodriquez et al., 1989) 27 30* 28 49* Diaziquone, 5 day bolus (Schold et al., 1984; 24* 24 Decker et al., 1985) Diaziquone, 5 day bolus (EORTC Brain Tumor 41 18* 44 18* Cooperative Group, 1985) Melphalan, oral (Chamberlain et al., 1988) 0 NA 7 NA Cisplatin (Bertolone et al., 1989)† 73 8* 83 12* Cisplatin (Sexauer et al., 1985)† 0NA7NA Carboplatin (Yung et al., 1991a) 40 20* 57 21* Eflornithine (Levin et al., 1992) 21 NA 44 48* Betaseron (Yung et al., 1991b) 51 18* 50 16* Cis-retinoic acid (Yung et al., 1993) 58 43* 48 25* Trans-retinoic acid (Kaba et al., 1997) 26 16* 71 18* Ifosfamide/mesna (Elliott et al., 1991) 0 NA 0 NA Temozolomide (Yung et al., 1999, 2000) 46 20* 58 41* Tamoxifen (Chamberlain and Kormanik, 1999) 63* 78 Irinotecan (Friedman et al., 1999) 71 NA 70 NA GMB, glioblastoma multiforme; AA, anaplastic astrocytoma; % R SD, percent of all treated patients who responded or had stable disease; MTP, median time to tumor progression; EORTC, European Oncology Radiation and Treatment Committee; NA, not available; PCNU, 1-(2-chloroethyl)- 3-(2,6-dioxo-1-piperidyl)-1-nitrosourea. *Report combined GBM and AA groups because histologies were not separated or there were too few patients in each group to separate activity by histology. †Childhood tumor data. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 111

Primary Cerebral Tumors 111

Table 3–4. Combination Chemotherapy for Recurrent and Progressive Supratentorial Glioblastoma Multiforme and Anaplastic Astrocytomas GBM, GBM AA AA, Treatment % R SD MTP (Weeks) % R SD MTP (Weeks) Procarbazine, lomustine, vincristine (Gutin et al., 1975; 45 15 65 27 Levin et al., 1980a) Carmustine, 5-fluorouracil, hydroxyurea, 55 23 71 46 6-mercaptopurine (Levin et al., 1986a) Elfornithine, carmustine (Prados et al., 1989) 30 8 57 76 Eflornithine, methyl bisguanylhydrazone (Levin et al., 1987) 72 49 6-Thioguanine, procarbazine, dibromodulcitol, 61 40 92 65 lomustine, 5-fluorouracil, hydroxyurea (Levin et al., 1993) Diaziquone, carmustine (Schold et al., 1987; 0–28 9 80 37 Yung et al., 1989) Diaziquone, procarbazine (Schold et al., 1987) 31 25 53 42 Cyclophosphamide, vincristine (Longee et al., 1990) 60 15 78 35 Carboplatin, 5-fluorouracil, procarbazine 32 20 57 36 (Flowers et al., 1993) Carboplatin, etoposide (Jeremic et al., 1992) 50 43 75 38 Mechlorethamine, vincristine, procarbazine 38 43 100 54 (Coyle et al., 1990) Etoposide, cisplatin (Buckner et al., 1990) 39 NA 15 NA 6-Thioguanine, procarbazine, CCNU, hydroxyurea 33 21 77 38 (Kyritsis et al., 1996) Nitrogen mustard, vincristine, procarbazine 4 11 18 16–19 (Galanis et al., 1998) Carboplatin, cis-retinoic acid (Kunschner et al., 1999) 52 30 GBM, glioblastoma multiforme; AA, anaplastic astrocytoma; % R SD, percent of all treated patients who responded or had stable disease; MTP, median time to tumor progression; NA, not available.

dicated (Garcia et al., 1985; Hirsch et al., 1989; Mer- Chemotherapy curi et al., 1981; Shaw et al., 1989; Wallner et al., 1988a). The efficacy of radiation therapy for incom- Chemotherapy in the treatment of juvenile pilocytic pletely resected pilocytic astrocytomas is not well es- tumors is generally limited to recurrent disease in pa- tablished (Shaw et al., 1989; Wallner et al., 1988a). tients previously treated with surgery and irradiation. Shaw et al. (1989) found that patients who under- Frequently, chemotherapy is initiated after reopera- went subtotal resection or biopsy and postoperative tion. The dearth and diversity of these tumors has led irradiation survived longer than those who did not, to anecdotal experience but no prospective studies. although the number of patients treated with surgery The most commonly used therapies have been with alone was small. The 5 year survival rate for patients cell-cycle phase nonspecific drugs such as alkylating who did not receive radiation therapy was 50% com- agents. In our experience, nitrosourea-based drug pared with 85% for those who received postoperative combinations have been capable of long-term pallia- irradiation ( p 0.08). Therefore, after subtotal re- tion. Therapies such as procarbazine, lomustine section the recommendation may be either immedi- (CCNU), and vincristine (Rodriguez et al., 1990; ate irradiation or close follow up, deferring treatment Petronio et al., 1991) and a combination of 6- until there is clinical or radiographic evidence of dis- thioguanine, dibromodulcitol, lomustine, procar- ease progression. bazine, and vincristine (Petronio et al., 1991; Cham- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 112

112 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

berlain and Levin, 1989) have been quite active. extent of surgery, and radiation dose identified six Based on the low level of activity of cisplatin and prognostic classes of malignant glioma patients with etoposide in infiltrative low-grade astrocytomas, these median survival times ranging from 4.6 to 58.6 are not the first choice for the treatment of recurrent months (Curran et al., 1993) (Table 3–5). noninfiltrating gliomas. The amount of tissue to include within the treat- ment volume has been a subject of considerable dis- cussion. Studies that compared cerebral angiography Malignant Astrocytomas and pneumoencephalography interpretations of tu- mor extent with autopsy findings demonstrated that Radiation Therapy malignant gliomas were usually more extensive than The importance of radiation therapy for malignant had been determined clinically (Concannon et al., gliomas was demonstrated in prospective clinical 1960; Kramer, 1969; Salazar et al., 1976). Although studies conducted by the Brain Tumor Cooperative patients included in these series had very advanced Group (BTCG) (formerly the Brain Tumor Study tumors and often died soon after diagnosis, the con- Group) (Walker et al., 1978, 1980) and the Scandi- clusion that treatment failure could be due to inade- navian Glioblastoma Study Group (Kristiansen et al., quate tumor coverage led to a recommendation that 1981). Patients who received postoperative irradia- the whole brain be irradiated (Kramer, 1969; Salazar tion had a longer survival time than those treated with and Rubin, 1976). either supportive care or chemotherapy alone. Other studies comparing CT and MRI scans with The practice of prescribing a radiation dose of 60 pathologic findings and patterns of failure after radi- Gy in conventional fractionation schemes is based on ation therapy support treating malignant gliomas with data from clinical studies and on limitations imposed limited-field rather than with whole-brain irradiation. by the radiation tolerance of normal brain tissue These investigations have shown that (1) malignant (Leibel and Sheline, 1991). A randomized trial con- gliomas are localized, and microscopic invasion of ducted by the Radiation Therapy Oncology Group the peritumoral brain tissue is limited at the time of (RTOG) and the Eastern Cooperative Oncology Group initial diagnosis (Burger et al., 1983); (2) only 1.1% (ECOG) compared 60 Gy whole-brain irradiation with to 7.3% of patients with glioblastoma and 4.4% of all 60 Gy delivered to the whole brain plus an additional patients with astrocytomas present with multifocal le- 10 Gy delivered to the tumor volume. No survival im- sions (Choucair et al., 1986; Kyritsis et al., 1992); provement was noted with the 70 Gy regimen (Chang (3) after initial treatment, most tumors recur at or et al., 1983; Nelson et al., 1988). Thus, it is custom- within 1 to 2 cm of their original location (Hochberg ary to treat anaplastic astrocytomas and glioblastoma and Pruitt, 1980; Wallner et al., 1989; Gaspar et al., multiforme to 60 Gy given in single daily fractions of 1992); (4) following resection of recurrent tumors, 1.8 to 2.0 Gy for 5 days per week (Leibel and She- second tumor recurrences develop at the original pri- line, 1987). mary tumor location (Massey and Wallner, 1990); With conventional radiation radiotherapeutic tech- and (5) cognitive impairment and frank dementia oc- niques the median survival time for patients with curred all too frequently in patients treated with glioblastoma multiforme is 9 to 10 months, whereas whole-brain radiation fields. On the other hand, in the 3 year survival rate is only 2% to 6% (Leibel et biopsy (Kelly et al., 1987) and autopsy (Burger et al., al., 1975, 1994; MRC Working Group, 2001). Less 1988) studies, isolated tumor cells have been found firm is the median survival time for patients with to infiltrate into edematous brain tissue surrounding anaplastic astrocytoma treated with irradiation only. the primary lesion. These observations have suggested Estimates of median survivals of 13 to 19 months have that radiation fields should extend to the periphery of been reported (Leibel et al., 1975; MRC Working CT-defined peritumoral hypodense regions or should Group Party, 2001). In a mixed group of high-grade encompass tissues that have a prolonged T2-weighted glioma patients treated between 1974 and 1989 with MRI signal. However, analyses of recurrence patterns radiation therapy and variable types and amounts of after conventional irradiation have not demonstrated chemotherapy on RTOG protocols, a recursive parti- a propensity for tumors to recur within edematous tion analysis based on age, histology, mental status, areas at a distance from the primary tumor site (Wall- duration of symptoms, neurologic functional status, ner et al., 1989). It is possible that all the observed 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 113

Primary Cerebral Tumors 113

Table 3–5. Outcomes of Patients with Malignant Astrocytic Gliomas According to Prognostic Class Based on a Recursive Partition Analysis of the RTOG Database of 1578 Patients Treated Between 1974 and 1989 Median Survival 2 Year Class Characteristics (Months) Survival (%) III AAF, age 50 years, normal mental status 58.6 76 II AAF, age 50 years, KPS 70–100, time from first 37.4 68 symptom to diagnosis 3 months AAF, age 50 years, abnormal mental status III or 17.9 35 GBM, age 50 years, KPS 90–100 GBM, age 50 years, KPS 90 or IV AAF, age 50 years, time from first symptom to 11.1 15 diagnosis 3 months or GBM, KPS 70–100, partial resection, “work” neurologic functional status GBM, age 50 years, KPS 70–100, partial resection, “home” or “hospital” neurologic functional status or V GBM, age 50 years, KPS 70–100, biopsy only, 8.9 6 received 54.4 Gy or GBM, age 50 years, KPS 70, normal mental status GBM, age 50 years, KPS 70–100, biopsy only, received 54.4 Gy VI or 4.6 4 GBM, age 50 years, KPS 70, abnormal mental status AAF, astrocytoma with atypical or anaplastic features; GBM, glioblastoma multiforme; KPS, Karnofsky performance status. Data from Curran et al. (1993).

isolated tumor cells were not tumorigenic because a surrounding edema may initially be included in the critical mass of tumor cells was not present. treatment volume (Leibel et al., 1991b). However, if As there is no apparent survival advantage to irra- the tumor is extensive and accompanied by a large diating the whole brain compared with irradiating volume of peritumoral edema, coverage of the entire more limited fields (Leibel and Sheline, 1990; Levin hypodense region would require treatment of a large et al., 1995) and because many patients who survive volume of brain tissue for what will probably be min- for prolonged periods after whole-brain irradiation imal or no therapeutic gain (Wallner, 1991). Radia- and chemotherapy develop significant treatment- tion portals should, when possible, be designed with related sequelae (Shapiro, 1986), partial brain field three-dimensional conformal techniques to spare, to irradiation is accepted as a standard approach in the the extent possible, the surrounding normal tissues. treatment of malignant gliomas. The fields are de- The response of malignant gliomas to standard ra- signed to encompass the perimeter of the contrast- diation therapy techniques is limited by their striking enhancing tumor with a 2.5 to 3.0 cm margin of tis- inherent radioresistance and the radiosensitivity of sue based on preoperative diagnostic imaging studies. the surrounding normal brain tissue. Accordingly, a If the tumor and associated edema are limited, the considerable amount of research has been directed 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 114

114 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

at improving the efficacy of radiotherapy. In addition tional fractionated irradiation to 60 Gy with carmus- to incorporating chemical radiation-response modi- tine (BCNU) given in both study arms. Hyperfrac- fiers and pursuing more effective chemotherapy pro- tionation did not lead to an improved outcome. The grams (see below), areas of investigation have in- median survival times for patients with glioblastoma cluded the use of altered fractionation schemes, dose multiforme in the two treatment arms were similar— escalation with interstitial brachytherapy, radio- 10.2 months with the hyperfractionated regimen and surgery, and three-dimensional conformal radiother- 11.2 months with conventional fractionation ( p apy and the application of heavy particle irradiation. 0.44). Similarly, for patients with anaplastic astrocy- Hyperfractionated irradiation differs from conven- toma the median survival times were 43.5 months tionally fractionated irradiation in that with the for- with hyperfractionation and 49.5 months with con- mer, two or more treatments are given daily with frac- ventional fractionation ( p 0.81) (Scott et al., tion sizes smaller than the usual dose fractions. The 1998). goal of this approach is to deliver a higher total ra- Another fractionation option, accelerated fraction- diation dose in the same overall time (6 to 6.5 weeks) ation, attempts to improve radiation-induced tumor as a conventional treatment schedule. Compared with cell kill in rapidly proliferating tumors by reducing rapidly proliferating glioma cells, normal neural tis- the length of time needed to complete the course of sues exhibit a slow rate or absence of cell division treatment (Fowler, 1990). Conventional size dose and have a greater capacity to repair sublethal radi- fractions (1.6 to 2.0 Gy) are given two or three times ation damage. The size of the radiation dose per frac- daily. This treatment schedule may also improve the tion has a predominant effect on the incidence of late therapeutic ratio by reducing tumor cell repopulation toxic effects in neural tissue (Leibel and Sheline, during treatment, thereby increasing the probability 1991). Therefore, the tolerance of these tissues to ra- of tumor control for a given dose level (Withers, diation should be improved by reducing the size of 1985). Several trials using accelerated regimens have the fractional dose (Nelson et al., 1986). Because been conducted, but none has shown a survival ben- rapidly proliferating tumor cells are less affected by efit over conventional irradiation for these regimens the reduction in fraction size, tumor control proba- (Simpson and Platts, 1976; Keim et al., 1987; bility should improve with hyperfractionation, Shenouda et al., 1991; Prados et al., 2001). These whereas late toxic effects should be equivalent to studies, however, indicate that the overall treatment those observed with conventional fractionation sched- time can be shortened, which may be especially ap- ules. The 6 to 8 hour interval between dose fractions propriate for patients with relatively short survival ex- allows the normal tissues to repair sublethal radia- pectancies (Simpson and Platts, 1976). tion damage. On the other hand, rapidly proliferating Because most gliomas are localized to a single area tumor cells progress into more radiosensitive phases of the brain (Choucair et al., 1986; Hochberg and of the cell cycle during the interval between fractions. Pruitt, 1980; Kyritsis et al., 1992), they should be con- Target cells for late sequelae proliferate slowly; thus, trollable if sufficiently high radiation doses can be de- for these tissues little cell cycle reassortment or “self- livered without damaging the surrounding normal sensitization” occurs during irradiation (Withers, brain tissue. Therefore, as a second strategy for 1985). improving the outcome of malignant gliomas, A randomized phase II dose-escalation study con- interstitial brachytherapy, radiosurgery, and three- ducted by the RTOG found that patients receiving 72 dimensional proton beam and photon radiation ther- Gy in 1.2 Gy fractions twice daily had a longer me- apy are being used as an adjunct to conventional ir- dian survival time (14.2 months) and a higher 18 radiation to augment the dose to the primary tumor. month survival rate (44%) than patients who received Interstitial brachytherapy has been extensively 81.6 Gy (11.7 months and 28%, respectively). The studied as an approach for increasing the tumor dose. inferior outcome in the 81.6 Gy arm of the study was Unlike conventional radiation therapy, the intratu- attributed in part to excess neurotoxicity. On the ba- moral placement of encapsulated radioactive sources sis of findings from a randomized phase II dose- results in maximal doses to the tumor while the sur- escalation study (Werner-Wasik et al., 1996), the rounding normal tissues receive considerably lower RTOG carried out a randomized trial comparing the radiation doses (Leibel et al., 1989a,b). 125-Iodine hyperfractionated irradiation to 72 Gy with conven- and 192-Iridium sources have most frequently been 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 115

Primary Cerebral Tumors 115

used in clinical practice, and stereotactic techniques doses of 110 to 120 Gy compared with patients treated have been devised for the placement of afterloading with conventional radiation therapy (alone or with catheters, which are removed after the prescribed chemotherapy) but who otherwise satisfied the crite- dose has been given. ria for implantation (Florell et al., 1992; Gutin et al., Well-circumscribed, peripheral supratentorial tu- 1991; Loeffler et al., 1990; Prados et al., 1992b; mors measuring up to 5 cm are best suited to receive Scharfen et al., 1992). Such comparisons, however, implantation. Patients with multifocal tumors or le- are subject to considerable patient selection bias sions with poorly defined borders, as well as those (Florell et al., 1992). with corpus callosum involvement and subependymal The brachytherapy approach was studied in three spread, and tumors of the cerebellum, midbrain, and randomized trials. The BTCG compared interstitial brain stem are not considered for this technique. To implantation (60 Gy at 10 Gy per day) preceding ex- be eligible for radiation implantation, the patient ternal irradiation (60.2 Gy at 1.72 Gy per fraction) should have good neurologic function and a KPS of and BCNU with external irradiation and BCNU alone. at least 70. Based on these criteria, approximately Implanted patients experienced a significant im- 33% of patients with newly diagnosed malignant provement in median survival (16 versus 13 months) gliomas are candidates for brachytherapy (Florell et and 18 month survival (47% versus 32%) compared al., 1992). with those who did not receive brachytherapy (Green The limiting aftereffect associated with interstitial at al., 1994). However, in another trial, Laperriere brachytherapy is focal peritumoral radiation injury to et al. (1998) found no difference in outcome of pa- the brain. The clinical and radiographic changes tients randomized to receive external beam irradia- caused by implanted radioactive sources may be in- tion alone to 50 Gy or external beam irradiation and distinguishable from signs of tumor progression. a 125I implant delivering a minimum tumor dose of Brachytherapy-induced neurologic deterioration and 60 Gy. The median survival time of patients receiv- steroid dependency may be effectively reversed by re- ing external beam irradiation and brachytherapy was section of the necrotic tissue, a step necessary in ap- 13.8 months compared with 13.2 months for those proximately 40% of implanted patients (Leibel et al., treated with external beam irradiation alone ( p 1991a). 0.49). Initially, interstitial implantation was used for re- A randomized trial testing the addition of intersti- treating previously irradiated patients who had re- tial microwave hyperthermia to the brachytherapy current malignant gliomas. When improvements in boost after external irradiation in newly diagnosed survival and quality of life were demonstrated (Gutin patients with glioblastoma multiforme was conducted et al., 1987; Leibel et al., 1989a), this treatment was at the University of California, San Francisco. Time to integrated into the primary management of patients progression and survival (median survival 85 versus with newly diagnosed malignant gliomas. As shown in 76 weeks; 2 year survival 31% versus 15%) were sig- Table 3–6, early phase II studies demonstrated sur- nificantly longer in patients receiving hyperthermia vival improvements in patients with glioblastoma mul- than in those treated with external beam irradiation tiforme who received a combination of external beam and brachytherapy alone ( p 0.045 and p 0.02, irradiation and interstitial brachytherapy to total respectively) (Sneed et al., 1998).

Table 3–6. Effect of 125I Implantation on Survival of Patients with Glioblastoma Multiforme Median Survival No. of Dose 1 Year 2 Years Study Cases Treatment (Gy) Months (%) (%) Florell et al. (1992) 68 Ext RT 60 14 55 12 Loeffler et al. (1990) 40 Ext RT 60 10 40 12 Loeffler et al. (1990) 35 Ext 125I 110 NR 87 57 Scharfen et al. (1992) 106 Ext 125I 110–120 22 86 29 Ext RT, external beam irradiation; Ext 125I, external beam irradiation plus 125I implantation; NR, not reached. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 116

116 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Stereotactic radiosurgery has also been used to in- months for ineligible patients ( p 0.07). A com- crease the tumor dose. Radiosurgery is a method of parison of these data with those reported by Shrieve focused, closed-skull external beam irradiation that et al. (1999) suggests that for glioblastoma multi- uses an imaging-compatible stereotactic device for forme a survival advantage of approximately 7 months precise target localization. The relationship between is conferred when radiosurgery is actually given. The the stereotactic coordinate system and the radiation benefit of radiosurgery diminishes when broader se- source allows accurate delivery of radiation to the tar- lection criteria are used (Mehta et al., 1994), whereas get volume (Larson et al., 1992). Radiosurgery may a better outcome is associated with lower pathologic be delivered using multiple cobalt beams (as in the grade, younger age, higher KPS, smaller tumor vol- “gamma knife”), modified linear accelerators, or ume, and unifocal tumors. (Larson et al., 1996) Al- charged particle beams. The radiation beams inter- though radiosurgery is biologically at the opposite sect at one (sometimes two or more) point(s) within end of the fractionation spectrum from brachyther- the skull after entering through numerous points or apy, the therapeutic ratios of these two modalities arcs distributed over the surface of the skull (Larson appear to be similar (Larson et al., 1993). Stereo- et al., 1990). This technique was designed to deliver tactic radiosurgery offers several advantages over a high radiation dose to an intracranial target in a brachytherapy. Radiosurgery is noninvasive, thus single session without delivering significant radiation avoiding the risks of hemorrhage, infection, and tu- to adjacent normal tissues. To maintain a steep dose mor seeding. Furthermore, prolonged hospitalization gradient at the edges of the field, the target volume is not required. Thus, for the most part, radiosurgery must be small. Furthermore, because the dose that has replaced brachytherapy for selected patients to can be safely administered in a single fraction is lim- augment the tumor dose. ited by the volume irradiated, the application of ra- Radiosurgery may also be used to re-treat patients diosurgery in effective single doses is restricted to le- with small, previously irradiated tumors. In one se- sions measuring approximately 4 cm or less in ries, the median survival time of 86 patients with re- diameter (Wilson et al., 1992). current glioblastoma multiforme was 10 months from Stereotactic radiosurgery techniques have been the time of radiosurgery, similar to the published ex- used to treat a variety of benign intracranial pro- perience with brachytherapy at recurrence (Shrieve cesses, most frequently small arteriovenous malfor- et al., 1995). mations as well as benign tumors such as pituitary Three-dimensional conformal photon radiation adenomas, acoustic neurinomas, and meningiomas. therapy is another method of treatment planning and The role of radiosurgery as a boost after conventional delivery designed to enhance the conformation of the radiation therapy for malignant gliomas is currently radiation dose to the target volume while maximally an area of active research (Loeffler et al., 1992). restricting the dose delivered to normal tissue out- Shrieve et al. (1999) used radiosurgery as a boost af- side the treatment volume. When applied to cerebral ter standard external beam irradiation (59.4 Gy) in tumors, conformal treatment planning techniques 78 patients with newly diagnosed glioblastoma mul- have permitted a 30% to 50% reduction in the vol- tiforme. Patients with discrete, geometrically spheri- ume of normal brain tissue irradiated at high doses cal lesions measuring 4 cm or less in diameter and (Thornton et al., 1991, 1992). This new approach to a KPS of 70 or higher were selected for radiosurgery. treatment planning may not only decrease the risk of The median tumor volume at the time of radiosurgery normal tissue injury but may allow higher than tra- was 9.3 cm3, and the median minimum tumor dose ditional radiation doses to be safely administered to was 12 Gy. The median survival time was 19.9 months, patients with malignant gliomas. In an ongoing study and the 1 and 2 year survival rates were 88.5% and at the University of Michigan, doses as high as 90 Gy 35.9%, respectively. Similar to brachytherapy, 50% of have been administered using three-dimensional the patients required reoperation to resect necrotic techniques (Lee et al., 1999; AS Lichter, personal tissue or recurrent tumor. Based on tumor size and communication, 2000). geometry and functional status criteria (Loeffler et al., Protons and helium ion beams, which feature 1992), only 9% of patients are eligible for this pro- sharp lateral beam edges and have a finite range in cedure (Curran et al., 1992b). The median survival tissue, may also be used to deliver higher radiation time for radiosurgery-eligible glioblastoma multi- doses to limited tumor volumes while keeping the forme patients was 12.5 months compared with 10.5 dose to neighboring critical structures at a safe level. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 117

Primary Cerebral Tumors 117

The depth of radiation penetration can be tailored to use of chemotherapy added to the survival benefit al- the tumor target by varying the energy of the beam or ready achieved by surgery and radiotherapy. Histori- by interposition of bolus material in the beam path. cally, the nitrosoureas (carmustine and lomustine) Proton beam arrangements designed with three- have been the most frequently studied agents (Chang dimensional treatment planning systems are being et al., 1983; Walker et al., 1978, 1980). A meta- used as boost therapy for glioblastoma multiforme analysis of the results of eight randomized trials re- (approximately 60 cc in volume) after conventional ported between 1976 and 1985 showed that the ad- radiation therapy (Suit and Urie, 1992). Fitzek et al. dition of a nitrosourea agent to radiation therapy pro- (1999) treated 23 patients with glioblastoma multi- vided a significant, albeit limited, benefit over forme using combined photon and proton beams to radiation therapy. The 1 year survival rate was in- administer a tumor dose of 90 cobalt gray equivalent creased by only 9% ( p 0.002) and the 2 year sur- (CGE) with an accelerated fractionation approach. vival rate by only 3.5% ( p 0.046) (Stenning et al., The median survival time was 20 months (a 5 to 11 1987). Patients younger than 60 years of age ( p month improvement compared with patients with 0.01) appear to benefit most from carmustine, comparable risk factors treated with conventional ra- whereas carmustine is less beneficial for older pa- diotherapy). The 2 and 3 year actuarial survival rates tients (Chang et al., 1983; Nelson et al., 1988). Based were 34% and 18%, respectively. Tumor regrowth, on these studies, the combination of conventional ir- demonstrated by histologic tissue examination, oc- radiation and carmustine became the standard treat- curred most commonly in regions that received dose ment regimen for malignant gliomas (Deutsch et al., levels of 70 CGE or less, whereas tumor was found in 1989). the 90 CGE dose volume in only one case. In an effort to improve on these data, a number of chemotherapy combinations were evaluated in clini- cal trials. The most widely used combination of pro- Chemotherapy with Irradiation carbazine, lomustine, and vincristine (PCV) was re- For patients with malignant astrocytomas, chemo- ported by Levin and colleagues (1990, 1995) in trials therapy has been given before, during, and after ra- where PCV was given after surgery and radiation ther- diotherapy to improve survival. In the past, the most apy. Tables 3–7 and 3–8 summarize some of the ad- common forms of chemotherapy have been cytotoxic juvant chemotherapy trials conducted and published drugs. The earliest question asked was whether the through 1999.

Table 3–7. Comparison of Survivals of Patients with Anaplastic Glioma 50% Survival 25% Survival Treatment No. (Weeks) (Weeks) WBRT, HU, carmustine 37 82 157 WBRT, HU, PCV (Levin et al., 1990) 36 157 NA (317) WBRT, mPCV (Jeremic et al., 1992) 31 148 NA LFRT, PCNU versus 50 74 204 LFRT, carmustine (Dinapoli et al., 1993) 46 88 208 BrdU, LFRT, PCV (Levin et al., 1995) 138 208 NA LFRT, 6TG, carmustine (Prados et al., 1998b) 110 284 NA WB/LFRT, carmustine 257 168 NA WB/LFRT, PCV (RTOG retrospective study 175 168 NA reported by Prados et al.,1999a) LFRT, PCV versus 92 ? NA LFRT, BrdU, PCV (Prados et al., 1999b) 97 ? NA WBRT, whole-brain radiotherapy; LFRT, limited-field radiation therapy; HU, hydroxyurea; PCV, procar- bazine, lomustine, vincristine; mPCV, dose-modified PCV; BrdU, bromodeoxyuridine; 6TG, 6-thioguanine; NA, not available. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 118

118 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Table 3–8. Comparison of Survival Among Studies that Evaluated Glioblastoma Patients as a Distinct Histology 50% Survival 25% Survival Treatment No. (Weeks) (Weeks) WBRT, HU, carmustine (Levin et al., 1990) 29 57 71 WBRT, HU, PCV (Levin et al., 1990) 31 50 94 WBRT, mPCV (Jeremic et al., 1992) 62 69 LFRT, PCNU versus 118 42 62 LFRT, carmustine (Dinapoli et al., 1993) 120 41 73 5FU, lomustine, LFRT, MISO, HU, PCB, VCR, 64 50 NA BCNU, 5 FU (Levin et al., 1986b) BrdU, LFRT, PCV (Phillips, et al., 1991) 160 62 104 CARBO, AFLFRT, PCV (Levin et al, 1995) 74 55 91 LFRT and PCV versus 134 60 82 LFRT and PCV, DFMO (Levin et al., 2000b) 138 58 78 BrdU, AFLRT, PCV (Groves et al., 1999) 70 57 79 AFLRT versus 231 40 LFRT versus 37 AFLRT and DFMO during RT versus 42 LFRT and DFMO during RT (Prados et al., 2001) 34 LFRT versus 226 39 57 LFRT and PCV (MRC Working Group Party, 2001) 223 40 61 WBRT, whole-brain radiotherapy; LFRT, limited-field radiation therapy; HU, hydroxyurea; PCV, procarbazine, lomustine, vincristine; mPCV, dose- modified PCV; MISO, misonidazole; BrdU, bromodeoxyuridine; CARBO, carboplatin; DFMO, difluoromethylornithine (eflornithine); AFLRT, accel- erated fractionated radiation therapy; VCR, vincristine; NA, not available.

The two variables having the most impact on che- There is no doubt that patients with anaplastic motherapy studies are age at diagnosis and tumor his- gliomas respond better to chemotherapy than those tology. As an approximation, survival appears to be with glioblastoma (Levin et al., 2001). In addition, inversely correlated to age and grade of malignancy: there are data that show that adjuvant (after sur- young patients survive longer than the elderly, and gery and radiotherapy) combination chemotherapy patients with low-grade gliomas live longer than those of procarbazine, CCNU, and vincristine (PCV) or with high-grade gliomas. In a randomized trial con- 6-thioguanine and BCNU leads to longer survival for ducted by the Childrens Cancer Group, children re- patients with anaplastic astrocytoma and other ceiving radiation therapy and the combination of lo- anaplastic gliomas (see Tables 3–7 and 3–8) than mustine (CCNU), vincristine, and prednisone had a 5 does monotherapy with adjuvant nitrosourea or pro- year event-free survival rate of 46% compared with carbazine. The median survival in three studies ranges 17% for children who received radiation therapy between 3 and 5 years (Levin et al., 1990, 1995; Pra- alone ( p 0.026). When patients who underwent dos et al., 1998a). at least a partial resection were separated by pathol- In addition to pursuing more effective chemother- ogy groups, chemotherapy was found to be beneficial apy programs, investigational research efforts are to children with glioblastoma multiforme (5 year directed at improving the therapeutic efficacy of ex- event-free survival, p 0.011; 5 year survival, p ternal irradiation. One strategy has focused on tech- 0.044) but not to those with anaplastic astrocytomas. niques designed to selectively enhance tumor cell This is comparable to our experience with adults with killing using methods to overcome tumor cell hy- glioblastoma for whom 5 year survival is closer poxia, halogenated pyrimidine analogue radiation to 5%. sensitizers, and modified radiation time–dose frac- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 119

Primary Cerebral Tumors 119

tionation schedules such as hyperfractionation and therapy during irradiation. The data from that study accelerated fractionation. were, however, inconclusive (Prados et al., 1998b). Hypoxia protects cells against the effects of radia- Disappointing was the recent closure of a multigroup tion. Hypoxic but viable cells may be found in regions open-label randomized phase III trial in newly diag- of coagulation necrosis that are present in glioblas- nosed patients with anaplastic glioma comparing ra- toma multiforme (Nelson et al., 1986). Even when diotherapy plus adjuvant PCV chemotherapy with or only 2% to 3% of such resistant cells are present, the without BrdU given as a 96 hour infusion each week radiation dose required to completely eradicate a tu- of radiotherapy. The study was closed early because mor may double (Dische, 1991). Various methods to a conditional power analysis indicated that even with modify the effects of hypoxia have been studied, in- an additional 12 months of accrual and follow up the cluding irradiation under hyperbaric oxygen condi- probability of detecting a difference between the two tions (Chang, 1977) and combining radiation ther- arms was less than 0.01% (Prados et al., 1999b). The apy with electron-affinic hypoxic cell radiation final report of this study will be available in 2002. sensitizers such as misonidazole (Bleehen, 1990). To date, neither of these approaches has resulted in an improvement in outcome over that produced by con- Chemotherapy of Malignant Astrocytomas ventional radiation therapy alone. Treatment with of Childhood high-linear energy transfer radiations (i.e., neutrons and heavy ions) that are less dependent on oxygen Although the major effect of chemotherapy on the out- tension than conventional (low-linear energy trans- comes of adult patients with malignant astrocytomas fer) photon irradiation has also not improved sur- has, to date, been limited to anaplastic gliomas, ther- vival rates (Castro et al., 1982; Laramore et al., 1988). apeutic trials during the last 15 years in children with The halogenated pyrimidine analogues are radia- these tumors have demonstrated some benefit, even tion sensitizers that are incorporated into rapidly di- for those with glioblastoma multiforme. This finding viding cells undergoing DNA synthesis because of suggests that high-grade astrocytomas have a some- their similarity to the DNA precursor thymidine. When what different biology in adults and children. the analogue is substituted for thymidine in the DNA In 1976, the Childrens Cancer Group embarked chain, the cell becomes more susceptible to radia- on a randomized trial in children newly diagnosed tion injury (Djordjevic and Szybalski, 1960). These with high-grade astrocytomas in which involved-field analogues appear to enhance the effects of a given irradiation alone was compared with involved-field ir- dose of radiation by a factor of 1.5 to 2 (Mitchell et radiation plus adjuvant chemotherapy (vincristine, al., 1983). Tumors located in the brain are ideally CCNU, and prednisone) for a duration of approxi- suited for halogenated pyrimidine analogue ra- mately 1 year. Although the number of fully evaluable diosensitization because the clonogenic cells undergo patients accrued to this randomized study was small division more rapidly and, therefore, incorporate (n 58), an advantage for adjuvant chemotherapy more drug into DNA than do the surrounding normal was observed for patients with glioblastoma multi- neural tissues. Two analogues, BrdU and iodo- forme: those receiving radiation alone died of pro- deoxyuridine, have been evaluated in patients with gressive disease, whereas 49% of patients receiving malignant gliomas. The initial impression from these the adjuvant chemotherapy remained free of tumor studies was that BrdU with radiation therapy pro- progression beyond 5 years. Because of the small duced survival improvement for patients who had numbers of patients enrolled in the study, no advan- anaplastic astrocytomas (Levin et al., 1995) but not tage for chemotherapy could be statistically con- for those with glioblastoma multiforme (Goffman et firmed in patients with anaplastic astrocytoma despite al., 1992; Phillips et al., 1991; Groves et al., 1999). a 20% 5 year PFS rate for children with radiation Unfortunately, recent publications bring this con- alone versus 45% for children with radiation plus clusion into question. First, a retrospective analysis chemotherapy. The most powerful factor predictive of RTOG and Northern California Oncology Group of outcome, in addition to the use of chemotherapy, (NCOG) databases since 1974 suggested that there was extent of resection (Sposto et al., 1989). might be a subset of relatively younger glioblastoma Based on results from the above study, in 1985 the patients who were most likely to benefit from BrdU Childrens Cancer Group embarked on a second ran- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 120

120 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

domized therapeutic trial in which the regimen of 16 patients with recurrent high-grade astrocytomas CCNU and PCV from the earlier study served as the (Miser et al., 1991). Ifosfamide alone demonstrated “standard therapy” arm and was compared with the no activity in eight patients with recurrent astrocy- “eight drugs in one day” regimen, a chemotherapeu- toma (Chastagner et al., 1993). Partial and minor re- tic regimen that actually incorporated seven drugs sponse rates varying from 0% to 38% have been re- (vincristine, CCNU, procarbazine, hydroxyurea, cis- ported in children with recurrent high-grade platin, cytosine arabinoside, and either cyclophos- astrocytomas treated with carboplatin (Allen et al., phamide or dacarbazine) in addition to methylpred- 1987) or cisplatin (Bertolone et al., 1989). The orig- nisolone for antiedema and antiemetic effects. In inal experience with the “8-in-1” regimen showed 1991, when the study closed, 185 children had been overall response rates of 22% to 44% in 27 patients entered and randomized to a treatment group, and with recurrent high-grade astrocytoma treated in an additional 40 children younger than 3 years of age three studies. In the Childrens Cancer Group proto- were nonrandomly assigned to receive the experi- col 945 discussed above, however, the partial re- mental treatment in addition to radiation. Results sponse rate was less than 10% in 80 children (Fin- showed no benefit for the more intensive experi- lay et al., 1994). Clearly, these chemotherapeutic mental therapy compared with control arm therapy; agents, alone or in combination and at conventional nevertheless, the study reaffirmed the benefit of ad- or dose-intensive strategies, have a minimal ability to juvant chemotherapy for childhood high-grade astro- produce significant tumor shrinkage in high-grade as- cytomas. Again, the prognostic impact of extent of re- trocytomas in children. section is observed in this study, with virtually all An approach initiated in the mid-1980s for chil- patients who underwent a biopsy or a partial resec- dren with recurrent brain tumors was the use of mye- tion having died of disease progression (Finlay et al., loablative doses of chemotherapy followed by bone 1995). marrow rescue with the patient’s own previously cry- Of children younger than 3 years treated with the opreserved marrow. One initial regimen utilized “8-in-1” chemotherapy regimen with the intent of de- thiotepa plus etoposide (Finlay et al., 1990); this reg- laying or avoiding irradiation, 35% were alive with- imen demonstrated a 20% 2 year disease PFS rate out disease progression beyond 3 years after diagno- among patients with recurrent high-grade astrocy- sis, a finding similar to that for older children. This tomas. Among patients so treated following radical more favorable outcome in young children may re- surgical resection and who had no gross evidence of flect an idiosyncratic biologic behavior of these tu- residual disease or had achieved a complete response mors to certain cytotoxic chemotherapy agents in very to the myeloablative chemotherapy, approximately young children. Certainly the time to disease pro- 33% were alive beyond 2 years post-diagnosis. Be- gression is much shorter for younger children (less cause of these early results, the regimens were sub- than 6 months) than for older ones. Young children sequently intensified for patients with recurrent tu- also display a more dramatic radiographic response mor, and high-dose carboplatin was added (Finlay, to chemotherapy than older children, indicating the 1992b). Additionally, children with newly diagnosed promising potential of chemotherapeutic strategies anaplastic astrocytoma or glioblastoma multiforme without irradiation for the treatment of younger chil- were treated with high-dose chemotherapy (thiotepa, dren with high-grade astrocytomas. These children etoposide, and carmustine) followed by marrow res- have traditionally fared poorly with any therapy and cue and subsequent involved-field irradiation; this are particularly vulnerable to the devastating conse- was the basis for a Childrens Cancer Group multi- quences of cranial irradiation (Geyer et al., 1995). center trial in North America for such children and Results of phase II trials evaluating new agents or young adults between the ages of 3 and 25 years (CCG- older drugs used in new, more dose-intensive strate- 9921; Groves et al., 1999). gies have been most encouraging for the medul- loblastoma/primitive neuroectodermal family of tu- Chemotherapy for Malignant Astrocytomas mors but far less so for children or adults with in Adults high-grade astrocytomas. In trials a decade ago, ifos- famide and etoposide in combination produced a 6% Chemotherapy for glioblastoma multiforme is, for the response rate (with no complete responses) among most part, a frustrating experience for both the physi- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 121

Primary Cerebral Tumors 121

cian and the patient. During the last 25 years, che- time to tumor progression of 20 to 30 weeks in 30% motherapy has provided small gains measured in to 40% of patients. For anaplastic astrocytomas, a added months rather than years of life. Most agents “good” median time to progression is 30 to 50 weeks found to be active in phase II clinical trials produce in 40% to 60% of patients. short-duration responses of 3 to 9 months in patients Two recent publications have addressed response with progressive or recurrent glioblastoma. Response covariates and study approaches that address the is- rates vary from 0% to 70%, depending on the crite- sue of durability of treatment benefit. In the first, the ria used to define response. For anaplastic astrocy- authors analyzed eight consecutive phase II trials that tomas, the gains have been much greater: Response included 225 glioblastoma and 150 anaplastic astro- rates of 25% to 90% have been reported, with me- cytoma patients (Wong et al., 1999). The data are dian times to progression of 6 to 15 months de- summarized in Table 3–9. Based on their data, the pending on the drugs and drug combinations used. investigators propose that phase II studies could uti- Because the most important measure of the activ- lize PFS at 6 months for glioblastoma as an outcome ity of a chemotherapuetic drug or regimen is time to measure and be quite comfortable with a new ther- treatment failure (when the tumor grows), what is apy that substantially exceeded the 15% observed in needed to assess the value of therapy are precise, re- their study. On the other hand, for anaplastic astro- liable criteria to define and determine tumor pro- cytoma patients the PFS at 6 months was 31%, and gression. Even though clinical considerations are im- this group of patients responds longer to chemo- portant in good patient management, clinical stability therapy than glioblastoma patients. This suggests that or deterioration alone is insufficient to guide treat- PFS at 9 months or 1 year would be a better study ment. endpoint for anaplastic astrocytoma. This approach Over the years, a variety of methods have been used is addressed in a recent paper by Hess et al. (1999). to assess response. These have included electroen- Excluding studies utilizing radiosurgical ap- cephalography, radionuclide brain scans, CT scans, proaches, the best published results today for the and, most recently, MRI. Of these, MRI is the most treatment of glioblastoma by surgery, radiation, and informative method, providing precise anatomic lo- chemotherapy have included chemotherapy during calization as well as information on tumor growth, radiation therapy to potentiate radiation followed by associated edema and mass effect, hydrocephalus, the PCV combination after radiation. Table 3–8 sum- brain atrophy, radiation damage, leptomeningeal marizes current data. There certainly have been no spread of tumor, and intracranial (or tumoral) bleed- giant steps made in the success of any one therapy. ing. This is important to keep in mind because many of Tables 3–3 and 3–4, respectively, summarize sin- the approaches being proposed today have public ap- gle-agent and combination phase II chemotherapy peal because they utilize twenty-first century technol- trial results for many, but not all, published trials. It ogy: gene therapy, radiosurgery, the gamma knife, is our belief that an effective treatment must be bone marrow transplantation, osmotic opening of the durable in at least a subset of patients. Response alone blood–brain barrier to improve drug access to the is insufficient as a measure of benefit unless it can be brain and tumor, and toxin immunoconjugates. Per- maintained. For single agents, a “good” therapy for haps in the future one or more of these approaches glioblastoma multiforme should produce a median will have a place in the therapy for malignant gliomas,

Table 3–9. Outcome Analysis of Phase II Studies for Recurrent Malignant Gliomas CR CR, PR, and MR, PFS PFS OS Variable No. PR and SD 6 Months 1 Year 1 Year Anaplastic astrocytoma 150 14% 48% 31% 20% 47% Glioblastoma multiforme 225 6% 33% 15% 8% 21% CR, complete response; PR, partial response; MR, minor response; SD, stable disease; PFS, progression- free survival; OS, overall survival. Data from Wong et al. (1999). 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 122

122 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

but currently their benefits are inconsistent and for post-treatment natural history make it difficult to eval- the most part problematic. One must be wary of re- uate the effect of radiation therapy on these tumors ferring patients for costly and marginally effective (Sheline, 1983). Representative survival data for treatments without sound scientific principles under- oligodendrogliomas are summarized in Table 3–10. lying the proposed treatment plan. Ten year survival rates vary, ranging from 8% to 59%. Another concern is that current treatments add lit- In contemporary series, median survival times of as tle to the survival or quality of life of older patients long as 16 years have been reported (Olson et al., (Grant et al., 1995), particularly those older than 60 2000). years. All clinical trials to date have found precipi- The role of postoperative irradiation for patients tous declines in survival after 60 to 65 years of age. with oligodendrogliomas is controversial, and con- This is not to say that surgery should not be carried clusions regarding its value are contradictory. Some out aggressively in these patients, but rather to inject authors recommend immediate postoperative irradi- caution about expectations from radiation and che- ation for patients with incompletely resected lesions motherapy. Cytotoxic chemotherapy especially is not (Chin et al., 1980; Griffin et al., 1992; Lindegaard et likely to improve survival for older patients. al., 1987; Wallner et al., 1988b), whereas others have For anaplastic gliomas, the story is quite different. been unable to show that postoperative irradiation is Although advanced age is an important variable that of benefit (Bullard et al., 1987; Sun et al., 1988). It negatively influences survival, chemotherapy can in- has also been suggested that radiation therapy be de- crease survival substantially. Treatment advances are ferred until there is evidence of tumor progression in part due to (1) drugs such as the nitrosoureas or recurrence (Reedy et al., 1983; Olson et al., 2000) (CCNU, carmustine), procarbazine, carboplatin, or that only patients with anaplastic tumors or mixed temozolomide, and cis-retinoic acid; and (2) some oligoastrocytomas should receive radiation treatment. drug combination schedules. Table 3–7 summarizes Caution should be exercised when drawing firm some of the gains made in adjuvant chemotherapy for conclusions from retrospective studies. For example, anaplastic gliomas. Survival gains have increased most retrospective studies comparing surgery alone from 1.5 to 5 years during the last decade. It is likely with surgery plus radiation therapy do not contain that temozolomide (Temodar), an alkylating agent analyses to ensure that the distribution of patients in approved by the U.S. Federal Drug Administration in the two treatment groups are comparable with re- August 1999, will create new opportunities for drug spect to prognostic variables such as age (Bullard et combination therapy against malignant gliomas as al., 1987), completeness of resection (Mork et al., well as medulloblastomas. In appears to be well tol- 1985; Whitton and Bloom, 1990), neurologic signs erated with less morbidity than many of the drugs cur- and symptoms (Bullard et al., 1987; Griffin et al., rently used to treat brain tumors. 1992), and histopathologic features (Bullard et al., The future will likely be characterized by slow in- 1987; Griffin et al., 1992; Mork et al., 1986). Fur- creases in the length of patient survival unless there thermore, treatment selection factors are either not is a research breakthrough that will provide a new stated or are unknown (Lindegaard et al., 1987), and chemotherapeutic approach. Tumor-specific and se- changes in the quality of surgery, radiation therapy, lective therapies at the genetic and chemical levels and patient management over the long time intervals will be required for any truly meaningful progress in spanned by the studies are not taken into considera- lengthening survival times. Achieving this task will re- tion. In addition, carefully performed pathologic stud- quire major research commitments and a good deal ies suggest that certain histologic and cytologic fea- of luck. tures have a significant effect on prognosis (Mork et al., 1986; Ludwig et al., 1986), making it important that pathologic material be independently reviewed. Oligodendrogliomas and Taken together, the data shown in Table 3–10 sug- Oligoastrocytomas gest that there may be a benefit to radiation therapy for oligodendrogliomas during the first 5 years after Radiation Therapy treatment, but this effect appears to diminish over The infrequent occurrence of oligodendrogliomas as time. Gannett et al. (1994) found a significant im- well as their variable and often long prediagnosis and provement in survival with postoperative irradiation. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 123

Primary Cerebral Tumors 123

Table 3–10. 5 Year and 10 Year Survival Rates for Patients with Oligodendroglioma Survival Rate (%) Study No. of Cases Treatment 5 Year 10 Year Marsa et al. (1975) 14 S RT 74 36 Chin et al. (1980) 11 S 82 — 24 S RT 100 — Reedy et al. (1983) 21 S 67 46* 27 S RT 63 51* Lindegaard et al. (1987) 62 S 27 12 108 S RT 36 8 Bullard et al. (1987) 34 S 48† 17† 37 S RT 60 15† Wallner et al. (1988b) 11 S 55† 18 14 S RT | 45 Gy 78† 56 Sun et al. (1988) 16 S 31 16 30 S RT 57 43 Shaw et al. (1991) 8 S 25 25 26 S RT 50 Gy 39 20 29 S RT 50 Gy 62 31 Griffin et al. (1992) 14 S 45 13 27 S RT 78 43 S, surgery alone; S RT, surgery and postoperative irradiation; —, data not available. *Percent survival at 8 years. †Data estimated from graphs.

Patients treated with surgery alone had 5 and 10 year radiation conferred a 14% improvement in 5 year sur- survival rates of 51% and 36%, respectively, com- vival ( p 0.01) (Shimizu et al., 1993). The data pared with 83% and 46%, respectively, for patients presented in these studies suggest that more effective who received radiation therapy ( p 0.032). Lin- therapies are needed to improve the long-term out- degaard et al. (1987) reported that radiation therapy comes of patients with oligodendrogliomas. prolonged the median survival time of patients with As for low-grade astrocytomas, it is also difficult incompletely resected tumor (37 months with radio- to take a categorical position regarding the role of therapy versus 26 months for surgery alone; p radiation therapy in the treatment of oligoden- 0.0089) but did not influence the overall cure rate. drogliomas. Patients with completely resected low- The studies by Wallner et al. (1988b) and Shaw et al. grade oligodendrogliomas as well as those with small, (1992) suggested a survival advantage for patients asymptomatic (except for seizures controlled with an- with incompletely resected tumors who received at ticonvulsant medication), incompletely resected le- least 45 to 50 Gy. However, the advantage of irradi- sions can be observed, deferring radiotherapy until ation versus nonirradiation did not reach statistical the time of recurrence. Large, symptomatic, unre- significance. On the other hand, Bullard et al. (1987) sectable or incompletely resected tumors should re- found no improvement in the time to clinical deteri- ceive postoperative irradiation (Macdonald, 1994). oration, time to tumor recurrence, or median sur- Proliferation markers such as the MIB-1 labeling in- vival time with the addition of radiation therapy, al- dex may help to distinguish patients who require ag- though the percentage of patients surviving with gressive treatment from those who can merely un- radiation therapy was consistently higher than those dergo observation (Schiffer et al., 1997). surviving without radiation therapy during the first 5 Radiation therapy is given using fields that en- years after treatment. A meta-analysis on reports from compass the tumor volume with a 2 to 2.5 cm mar- the current literature concluded that postoperative ir- gin. A dose of 54 Gy in daily 1.8 Gy fractions is given 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 124

124 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Table 3–11. Effects of Histologic Classification Schemas on the “A to D” and the St. Anne-Mayo grading schemes. Median Survival Time for Patients with Oligodendrogliomas From the reports cited it can be concluded that pa- Histologic Type tients with low-grade tumors survive 108 months (9 Low Grade* High Grade† years) compared with 26 months (2.2 years) for Series (Months) (Months) those with high-grade tumors. In general, the patients Smith et al. (1983) 94 17 reported (Smith et al., 1983; Shaw et al., 1989, 1992) Kros et al. (1988) 113 15 were treated with surgery and irradiation; few were Shaw et al. (1992) 117 47 formally treated with chemotherapy. It is clear from Table 3–12, given the sensitivity of oligoden- Mean ( SD) 108 ( 12) 26 ( 18) drogliomas to chemotherapy, that the survival of Mean, mean of reported median survival duration; SD, standard de- anaplastic oligodendroglioma patients treated with viation. adjuvant chemotherapy (or chemotherapy at tumor *Smith type A and St. Anne-Mayo low grade. progression) will easily exceed the median 2.2 years. †Smith type D and St. Anne-Mayo high grade. Table 3–12 summarizes adjuvant chemotherapy (pre-radiation and/or post-radiation) with two dif- ferent dose schedules of procarbazine, CCNU, and vin- to adults, and that dose is reduced to 50.4 Gy for chil- cristine (PCV and slightly higher dose mPCV). The re- dren (Karlsson et al., 1992). The dose is increased sponse and stable disease rates are nearly 100%, and to 59.4 Gy with a reduced field for patients with the number of patients without progression at 2 years anaplastic oligodendrogliomas (Bauman and Cairn- range from 50% to 85%. PCV combinations have been cross, 2001). Complex three-dimensional techniques used widely in the treatment of recurrent oligoden- should be used for treatment planning and delivery droglioma and oligoastrocytoma tumors with com- when appropriate. PCV chemotherapy (see next sec- plete response rates of 12% to 33%, partial response tion) may be useful in reducing the size of large tu- rates of 40% to 50%, and stable disease rates of 19% mors before radiotherapy is begun (Streffer et al., to 31% (Cairncross et al., 1994; Soffietti et al., 1998; 2000). van den Bent et al., 1998). Taken together, the three studies yield PFS values of 31 months for complete response, 16 months for partial response, and 14 Chemotherapy for Oligodendrogliomas months for stable disease. As previously discussed, oligodendroglial tumors can There is little doubt that other chemotherapy ap- be categorized by histologic features that correlate proaches that utilize combinations of cytotoxic agents with survival. Table 3–11 summarizes survivals for such as carboplatin and temozolomide would be ex- low-grade and high-grade oligodendrogliomas using pected to be active against oligodendrogliomas. As a

Table 3–12. Initial treatment of Oligodendrogliomas, Anaplastic Oligodendrogliomas, and Oligoastrocytomas Series Treatment No. % R SD % 2 Year TTP Anaplastic oligodendroglioma Glass et al. (1992) mPCV RT 5 100 50* Kyritsis et al. (1993) PCV RT 8 100 62 Oligoastrocytoma Glass et al. (1992) mPCV RT 14 92 85† Kyritsis et al. (1993) PCV RT 12 100 75 PCV, procarbazine, lomustine, and vincristine; mPCV, different PCV schedule; RT, radiation therapy; TTP, time to tumor progression, % R SD, percent of all patients who respond or have stable disease. *Three pre-radiation PCV courses. †Three no radiation at evaluation; 10 pre-radiation PCV; 2 year survival projected because no patient has been evaluated long enough. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 125

Primary Cerebral Tumors 125

conservative approach, while we wait for newer drugs on these data, it seems most appropriate to treat chil- and better approaches to chemotherapy for oligo- dren as well as adults with a PCV combination until dendrogliomas, it is reasonable to advocate using a a better therapy comes along. As part of the initial combination of procarbazine, CCNU, and vincristine treatment, we normally advocate that a nitrosourea- or simply the combination of CCNU and procarbazine based chemotherapy regimen such as PCV be used for patients with recurrent tumors and to anticipate after limited-field irradiation. that adjuvant chemotherapy with surgery and radia- tion may also turn out to be efficacious. Ependymoma Radiation Therapy Chemotherapy for Mixed Oligoastrocytomas Most supratentorial ependymomas cannot be com- Although most high-grade astrocytic tumors in adults pletely excised because of their location and growth and children are either glioblastoma multiforme or characteristics (Marks and Adler, 1982; Kricheff et anaplastic astrocytoma, a few patients have mixed ma- al., 1964). Historically, 0% to 15% of patients with lignant gliomas. These are usually anaplastic astrocy- supratentorial tumors treated with surgery survived 5 toma with oligodendroglioma or, less commonly, years (Ringertz and Reymond, 1949; Mork and Lo- anaplastic astrocytoma and ependymoma. Only lim- ken, 1977). More recently, with improved surgical ited information on chemotherapy for children with techniques, survival gains have been realized. Palma these tumors is available. In the Childrens Cancer et al. (1993) reported on 20 patients between the Group protocol 945, 14 patients were considered to ages of 1 and 20 years. Of the 18 who survived sur- have mixed gliomas, with PFS and overall survival at gery, 12 (67%) were alive and disease free at a mean 5 years of 64% and 71%, respectively. follow-up of 12 years (range 5.2 to 21 years); there Adjuvant chemotherapy trials for adults with mixed were nine ependymoma, one anaplastic ependy- oligoastrocytomas are also listed in Table 3–12. As moma, and two subependymoma patients in this sur- with oligodendrogliomas, response to therapy in viving group. these tumors is high, with more of these patients than Radiation therapy significantly improves tumor those with anaplastic oligodendroglioma having no control and survival and is an accepted part of the tumor progression at 2 years (80% versus 56%). standard treatment for ependymomas. As shown in Most patients received the PCV combination. Based Table 3–13, 5 year survival rates for patients who re-

Table 3–13. 5 Year Survival Rates for Patients Receiving Radiation Therapy for Supratentorial Ependymomas Studies Cases Treatment 5 Year Survival (%) Ringertz and Reymond (1949) 13 S 15 Mork and Loken (1977) 14 S 0 S RT Phillips et al. (1964) 12 S RT 80 S RT 45 Gy Kim and Fayos (1977) 11 S RT 46 Bouchard (1980) 17 S RT 65 Marks and Adler (1982) 20 S RT 35 Pierre-Kahn et al. (1983) 15 S RT 51 Salazar et al. (1983) 20 S RT 35 Read (1984) 19 S RT 40 (adults) 17 (children) Goldwein et al. (1990) 18 S RT 35 Vanuytsel et al. (1992) 40 S RT 48 S, surgery alone; S RT, surgery and postoperative radiation therapy. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 126

126 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

ceived radiation therapy range from 30% to 80%. The ous local fields (Goldwein et al., 1990; Leibel and correlation between survival and histopathologic Sheline, 1987; Pierre-Kahn et al., 1983; Shaw et al., grade is controversial, although we found it to be an 1987; Wallner et al., 1986; Vanuytsel et al., 1992) to important predictor of tumor behavior and outcome fields encompassing the whole brain (Salazar et al., (Leibel and Sheline, 1987). 1983). However, the survival and local tumor con- Supratentorial tumors generally have a poorer trol rates achieved by partial brain irradiation (Gold- prognosis than their infratentorial counterparts, wein et al., 1990; Vanuytsel et al., 1992; Wallner et probably because a greater proportion of them are al., 1986) are comparable with those obtained with high grade (Chin et al., 1982) and because there is whole-brain irradiation (Salazar et al., 1983). As a tendency for larger volumes of residual disease to nearly all recurrences are limited to the original pri- be present after surgical resection at this location mary tumor site, whole-brain irradiation would not (Marks and Adler, 1982). Unlike other brain tumors, be expected to improve the chance of cure. Patients young children with ependymomas tend to have a are treated using a generous target volume, based on poorer outcome than older children or adults (Gar- operative and radiographic findings, to a dose of at rett and Simpson, 1983; Goldwein et al., 1990; Shaw least 54 Gy. Pretreatment myelography, spinal MRI, et al., 1987). This may reflect the practice of reduc- and CSF evaluation are not performed unless there is ing the dose of radiation for very young children evidence of ventricular involvement or signs of sub- (Goldwein et al., 1990, 1991; Marks and Adler, arachnoid metastases. 1982). In the past most authors recommended inclusion The amount of normal CNS tissue that should be of the entire craniospinal axis in the treatment of included in the treatment volume is a major area of anaplastic supratentorial ependymomas (Salazar et controversy. Differences in opinion are based on the al., 1983; Wallner et al., 1986; Vanuytsel et al., 1992), potential for ependymomas to spread into the ven- although some recommended whole-brain irradia- tricular system and to disseminate into the spinal sub- tion with an additional boost for high-grade supra- arachnoid space. In some reports, the inclusion of tentorial lesions located away from the CSF pathways ependymoblastomas, which are known for their (Pierre-Kahn et al., 1983; Goldwein et al., 1990) if propensity to disseminate throughout the CNS, tends there was no evidence of leptomeningeal spread. to overestimate the risk of seeding of the spinal sub- However, despite the apparent superiority of cran- arachnoid space (Goldwein et al., 1991; Ross and Ru- iospinal radiation therapy noted in some series binstein, 1989). Autopsy and clinical studies show (Salazar et al., 1983), the findings that (1) local re- that patients who develop intracranial and intraven- currence is the primary pattern of failure (Goldwein tricular spread also have recurrent or persistent dis- et al., 1991; Salazar et al., 1983; Marks and Adler, ease at the primary tumor site (Marks and Adler, 1982; Shaw et al., 1987; Vanuytsel et al., 1992; Mer- 1982; Salazar et al., 1983; Wallner et al., 1986). chant et al., 1997; McLaughlin et al., 1998; Schild et Spinal seeding is uncommon in supratentorial al., 1998), (2) spinal seeding is uncommon in the ependymomas, occurring in only 1.6% of cases in a absence of local recurrence (Leibel and Sheline, literature review by Vanuytsel and Brada (1991). In 1987; Marks and Adler, 1982; Vanuytsel and Brada, that review, no patient with high-grade supratentor- 1991; Schild et al., 1998), (3) the patterns of failure ial lesions developed spinal subarachnoid seeding, are similar in patients with high-grade tumors treated whereas the incidence of seeding in low-grade supra- with local fields or with craniospinal axis irradiation tentorial lesions was 2.7%. Relapse of the primary (Goldwein et al., 1991a; Vanuytsel et al., 1992; tumor had the greatest effect on the subsequent de- McLaughlin, 1998), and (4) prophylactic treatment velopment of spinal seeding, regardless of tumor may not prevent spinal metastases have led many in- grade or site. Spinal dissemination occurred in 3.3% vestigators to question whether the routine use of of patients with locally controlled tumors, whereas craniospinal or whole-brain irradiation leads to im- 9.5% with uncontrolled primary lesions developed proved survival (Goldwein et al., 1991; Vanuytsel and seeding ( p 0.05). The frequency of seeding was Brada, 1991; Vanuytsel et al., 1992). not affected by whether prophylactic spinal irradia- Merchant et al. (1997) reviewed the outcomes of tion was given. 28 pediatric patients with anaplastic ependymoma The treatment volumes recommended for low- treated either with or without craniospinal irradia- grade supratentorial ependymomas vary from gener- tion. The actuarial 5 and 10 year survival rates were 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 127

Primary Cerebral Tumors 127

56% and 38%, respectively. A benefit from cran- tion of anaplastic (or malignant) ependymomas than iospinal irradiation could not be demonstrated. All of the more common cellular (or low-grade) ependy- 19 patients who failed radiotherapy relapsed at the momas. In most series, supratentorial ependymomas primary site, and one of these also developed sub- occur nearly as frequently as infratentorial ependy- arachnoid dissemination. On the basis of these and momas. When supratentorial ependymomas recur, other data, craniospinal irradiation is generally not they almost always do so at the primary tumor site, recommended for patients with supratentorial regardless of pathology. anaplastic (high-grade) ependymomas unless evi- The value of chemotherapy for ependymomas in dence of leptomeningeal spread is pathologically or general and for supratentorial ependymomas in par- radiographically documented (Vanuytsel and Brada, ticular remains unclear (Puccetti and Finlay, 1992). 1991; Goldwein et al., 1991; Goldwein and Lefkowitz, No distinction has been made in phase II chemo- 1991; Merchant et al., 1997; McLaughlin, 1998). Al- therapy trials between supratentorial and infratento- though dose levels of 54 Gy have typically been used rial ependymomas. Table 3–14 summarizes therapies to treat anaplastic ependymomas, the administration for ependymomas in both locations. Single agents vary of 59.4 Gy with conformal techniques would appear in the quality of responses and the durability of re- to be warranted. sponses that they produce. Those yielding good re- Because the inability to eradicate the primary tu- sponses and longer duration of response activity are mor remains the single most important factor lead- carmustine (Levin et al., 1993), dibromodulcitol ing to treatment failure (Vanuytsel et al., 1992), more (Levin et al., 1984), and diaziquone (Ettinger et al., aggressive local therapies to improve local tumor 1990). Cisplatin and carboplatin appear less effec- control, in both low- and high-grade ependymomas, tive; more recent trials with ifosfamide have shown are being explored. These include the use of boosts similar activity. with stereotactic radiotherapy or conformal radio- Unfortunately, no studies to date have demon- therapy techniques as well as hyperfractionated radi- strated a survival advantage with single-agent chemo- ation schedules (McLaughlin, 1998; Schuller et al., therapy plus irradiation compared with irradiation 1999). alone in children or adults with newly diagnosed ependymomas (Goldwein and Lefkowitz, 1991). Levin and colleagues (2000a) reported on 17 patients with Chemotherapy malignant ependymoma treated with a combination Supratentorial ependymomas are uncommon tumors, of 6-thioguanine, procarbazine, dibromodulcitol, probably representing fewer than 30% of all ependy- CCNU, and vincristine (TPDCV) chemotherapy and momas. While firm data on age incidence for these craniospinal radiation. With the combined chemo- supratentorial tumors are unavailable, in our experi- therapy–radiotherapy protocol, 65% of patients ence these tumors are most often seen in the first two failed, with a median time to progression of 141 decades of life. Also, there may be a higher propor- weeks and a median survival of 42 months, and there

Table 3–14. Chemotherapy for Recurrent Ependymoma and Anaplastic Ependymoma Regardless of Location 50% TTP 25% TTP Series Drug No. % R SD (Months) (Months) Levin et al. (1984) Dibromodulcitol 12 75 16 20 Levin et al. (1987) Carmustine 14 78 13 24 Bertolone et al. (1989) Cisplatin 8 75 3.8 4.3 Goldwein et al. (1990) Various* 37 trials/16 patients 22 9 10 Gaynon et al. (1990) Carboplatin 14 28 14 NA Ettinger et al. (1990) Diaziquone 12 42 10 16 Prados et al. (1991) TPDCV 11 82 21.6 NA % R SD, percent of patients who responded or had stable disease; TTP, time to tumor progression; TPDCV, 6-thioguanine, procarbazine, di- bromodulcitol, lomustine, and vincristine (MD Prados and VA Levin, personal communication); NA, not attained yet. *Various combinations of vincristine, cisplatin, lomustine, procarbazine, etoposide, and ifosfamide combinations. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 128

128 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

were 8 patients alive at a median of 9 years after study Clinical Features entry. Primitive neuroectodermal tumors occur most com- Therapy for ependymomas relies first on surgery, monly in the first two decades of life, although they second on radiation, and third on chemotherapy. In are increasingly recognized in adults into the sixth most cases, failure of the first two approaches forces decade. Following the initial description of primary consideration of chemotherapy. Although we advo- cerebral neuroblastoma by Horten and Rubinstein cate no specific chemotherapy at this time, we believe (1976), a distinction in clinical behavior and out- that platinum-based therapies are not particularly come between that entity and supratentorial PNET active and should be avoided. Unfortunately, few (Hart and Earle, 1973) has been emphasized by some adjuvant chemotherapy studies of the treatment of authors but not by others. Both types of tumors dem- anaplastic ependymoma are ongoing at this time. onstrate similar clinical and diagnostic features, al- though the in vivo appearance of each may be some- what different. Either lesion may be cystic, PRIMITIVE NEUROECTODERMAL grayish-granular, sharply demarcated, and hemi- TUMORS spheric in location (Kosnik et al., 1978). In addition to being cystic, the cerebral neuroblastoma may also Pathology have a mural tumor nodule that is not seen in PNET. The latter lesion often shows gross areas of necrosis, The term primitive neuroectodermal tumor (PNET) which are not typical of the cerebral neuroblastoma. was originally used to describe small cell, predomi- Staging for extent of tumor is similar to that for in- nantly undifferentiated, tumors occurring in the cere- fratentorial medulloblastoma, although the yield of bral hemispheres of young patients (Hart and Earle, positive studies at initial diagnosis is lower. Patients 1973). Some tumors showed minor components of must undergo imaging of the entire neuraxis by brain glial or neuronal differentiation. The common histo- and spine MRI, as well as a diagnostic lumbar punc- logic feature of all of these tumors is, however, some ture within 2 to 3 weeks following surgery for cyto- component of immature, small embryonal-like cells. logic evaluation of CSF for malignant cells. Most have frequent mitotic activity and zones of ne- crosis. Subsequently, the term cerebral neuroblas- toma has been used for those morphologically sim- Surgery ilar tumors that show evidence of primitive or more advanced neuronal differentiation. The features of The goal of surgery is complete tumor resection neuronal differentiation include the presence of whenever feasible. Because of the tumor’s hemi- Homer-Wright fibrillary rosettes and large neuron- spheric location, it may be necessary to utilize phys- like cells with more abundant eosinophilic cytoplasm, iologic mapping to achieve that goal, as previously large round nuclei, and prominent nucleoli. Cerebral described. There are sparse data on how extent of re- neuroblastomas also tend to have an abundant com- section affects outcome. Although several reports of ponent of connective tissue and a desmoplastic ap- long-term survival appear to coincide with radical re- pearance. Small cell, undifferentiated tumors that section of either tumor type (Duffner et al., 1981; show astrocytic differentiation or both neuronal and Halperin et al., 1993; Gaffney et al., 1985), this is not astrocytic differentiation may be called PNET with glial always the case (Tomita et al., 1988). The patients or with divergent differentiation. who benefit most from total resection have cystic Most pathologists accept positive immunoreactiv- cerebral neuroblastomas (Berger et al., 1983). ity for glial fibrillary acidic protein (GFAP) in neo- Although PNET may occur in the first three decades plastic cells as evidence of astrocytic differentiation. of life (unlike neuroblastomas, which always affect Positive immunoreactivity for synaptophysin or for younger children), both lesions must be distinguished neurofilament protein confirms neuronal or neuro- from the adult central neurocytoma, which is an in- blastic components. Examples with definite epithelial traventricular, neuronally derived tumor originating ependymal rosettes are designated ependymoblas- from the septum pellucidum (Hassoun et al., 1993). tomas, or PNET with ependymal differentiation. Tu- Central neurocytomas are histologically distinct from mors arising in the pineal region are pineoblastomas. the neuroblastic and neuroectodermal tumors de- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 129

Primary Cerebral Tumors 129

scribed above. Complete resection of the central neu- is important to remember that almost 40% of the rocytoma is usually curative without further adjuvant body’s bone marrow resides in the spinal vertebrae therapy (Kim et al., 1992). and that, following craniospinal irradiation at the standard dose of 36 Gy, the patient’s ability to toler- ate even modest doses of chemotherapy is signifi- Radiation Therapy cantly impaired. For patients with pineoblastomas, The general consensus is that patients with PNET who are known to experience a very high rate of neu- should receive postoperative radiation therapy raxis dissemination at diagnosis and relapse, re- (Humphrey et al., 1981). Although radiation therapy stricting the irradiation field (especially in older chil- appears to increase survival time (Duffner et al., dren and adults) is probably inappropriate; however, 1981; Jenkin, 1981), the outcome is usually poor, in patients with other supratentorial PNET this may and most patients develop local or regional recur- be appropriate within the conduct of a formal clini- rences. Because of their propensity to spread cal trial. One other approach to permit intensifica- throughout the subarachnoid space (Kosnik et al., tion of chemotherapy is to deliver such treatment in 1978; Parker et al., 1975; Knapp et al., 1981; Ashwal a therapeutic window before the delivery of cran- et al., 1984; Humphrey et al., 1981), PNET are treated iospinal irradiation. with craniospinal axis irradiation (Jenkin, 1981; Cerebral neuroblastomas are biologically distinct Gaffney et al., 1985). The primary tumor is given 54 from other PNET. They tend to be less malignant, have to 55 Gy, and the remainder of the axis receives 36 a better outcome, and are less likely to disseminate Gy. The dose should be reduced for very young chil- throughout the craniospinal axis (Cohen and Duffner, dren (Gaffney et al., 1985). 1984). This tumor may present as a cystic lesion with Chemotherapy is usually a part of the treatment a peripheral nodule or as a solid mass, and the tu- program. Although primitive neuroectodermal tu- mor’s morphologic appearance is related to progno- mors are included in pediatric cooperative group sis. Berger et al. (1983) reported the outcomes of 11 protocols designed for high-risk medulloblastoma patients with cerebral neuroblastoma. Six patients patients, they are less radiocurable than medul- presented with cystic tumors, and five had solid le- loblastomas (Gaffney et al., 1985; Humphrey et al., sions. Gross total resection was achieved in only two 1981). In some series 1 year survival rates are as low patients, both with cystic tumors. All patients received as 10% (Kosnik et al., 1978), whereas others report postoperative irradiation. Seven of the 11 patients, all 5 year survival rates of 20% to 47% (Gaffney et al., treated with local irradiation to an average of 52 Gy, 1985; Humphrey et al., 1981; Paulino and Melian, remained alive with no evidence of tumor progres- 1999). The disparity in survival figures reflects the sion. None of the six patients with cystic tumors de- heterogeneity of malignancies that are classified un- veloped recurrent disease, whereas four of the five der the term primitive neuroectodermal tumors patients with solid tumors had a relapse at the pri- (Rubinstein, 1985). For example, in a series of 14 mary site within 8 to 31 months after treatment. The patients reported by Gaffney et al. (1985), the 3 year only patient with a solid lesion who did not have a tu- survival rate was 29%. None of the patients with tu- mor recurrence received adjuvant chemotherapy. Al- mors containing more than 90% undifferentiated el- though subarachnoid dissemination is found in au- ements were alive at 3 years, whereas 60% of those topsied cases (Horten and Rubinstein, 1976), this with less primitive tumors survived 3 years. pattern of spread does not represent a significant clin- An important question is raised by the relative in- ical problem. Thus, localized cerebral neuroblas- frequency of neuraxis dissemination at diagnosis of tomas are treated with limited-field irradiation to 54 supratentorial nonpineal region PNET compared with Gy. The craniospinal axis is included only if there is infratentorial medulloblastoma. Because the outcome evidence of tumor dissemination beyond the site of for patients with residual tumor in the absence of origin by imaging studies or CSF cytology. metastases remains so poor even in the face of full- dose craniospinal irradiation, one might argue that Chemotherapy for such patients restricting irradiation to involved fields might permit greater tolerance of dose intensi- Although the mainstay of therapy for PNET is surgery fication of chemotherapy with improved outcome. It and irradiation, young patient age, incomplete resec- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 130

130 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

tion, and recurrent tumor require the addition of che- PNET. Interestingly, greater than 40% of infratentor- motherapy. Several single-institution reports on ial tumors showed evidence of neuraxis dissemina- supratentorial PNET have indicated a poor prognosis tion at diagnosis (either positive CSF cytology and/or following surgical resection alone or combined with radiographic evidence of intracranial or spinal lep- postoperative irradiation; this reflects a high inci- tomeningeal dissemination on head CT scan or myel- dence of primary site recurrence as well as neuraxis ogram), whereas only 21% of patients with supra- dissemination throughout the leptomeninges, even in tentorial PNET showed such dissemination. The the face of prophylactic craniospinal irradiation. This outcome for patients with supratentorial PNET was is particularly true of pineoblastomas, which have a significantly affected by neuraxis dissemination; pa- high rate of neuraxis dissemination at both initial pre- tients without neuraxis dissemination had a 3 year sentation and subsequent recurrence. Patient age at PFS of 56%, whereas all patients with evidence of neu- onset of these tumors may also have a powerful prog- raxis dissemination at diagnosis ultimately relapsed. nostic importance; most studies indicate that very An additional effect of volume of residual tumor young children fare worse than older patients, with was observed in patients with nonpineal PNET treated higher rates of dissemination at diagnosis and recur- in the CCG-921 study; those with residual tumor vol- rence and a shorter time to recurrence. Chemother- ume less than 1.5 cm2 on postoperative scans had a apy alone is now generally employed for children 4 year survival of 40% 22% compared with younger than 3 years with supratentorial PNET. 13% 8% for those with greater residual tumor vol- The site of origin of PNET has been documented ume. Because of small patient numbers, however, this as an important prognostic indicator in a large phase difference did not attain statistical significance (Al- III trial conducted by the Childrens Cancer Group in bright et al., 1995). The even smaller patient num- North America (the CCG-921 study). In that study, bers with pineal PNET obviated any valid compar- both posterior fossa and supratentorial PNET were isons, but suggest that patients with bulky residual randomized to one of two chemotherapy-containing disease relapsed at a higher rate and earlier follow- regimens; all patients underwent maximal surgical re- ing diagnosis (Jakacki et al., 1995). Subsequent pe- section of the primary site tumor and were given post- diatric cooperative group trials attempted to improve operative craniospinal irradiation with a boost to the upon the poor results for patients with supratentor- primary tumor site. Chemotherapy with CCNU, vin- ial PNET with either significant residual tumor post- cristine, and prednisone was compared with the operatively and/or neuraxis dissemination. The stud- 8-in-1 regimen of CCNU, vincristine, cisplatin, hy- ies employed intensification of chemotherapy with droxyurea, cytosine arabinoside, cyclophosphamide, agents known to be efficacious against PNET (e.g., and methylprednisolone. The most recently published alkylating agents such as cyclophosphamide or ifos- reports from this study (Jakacki et al., 1995; Cohen famide at high doses, and cisplatin or carboplatin). et al., 1995; Zeltzer et al., 1999) demonstrate some Again, however, evidence of benefit over conventional surprising differences between patients with pineal chemotherapy is lacking. region PNET, other nonpineal region PNET, and in- Because the occurrence of supratentorial PNET is fratentorial PNET or medulloblastoma. For patients not uncommon within the first 3 years of life, and be- with pineal PNET, the 3 year PFS was 61% 13%, cause children with these tumors are particularly vul- whereas the corresponding 3 year PFS for patients nerable to the deleterious effects of whole-brain ir- with nonpineal supratentorial PNET was only 33% radiation on the developing brain, chemotherapy is 9%, a statistically significant difference. Considering now routinely used instead of irradiation either to both extent of resection and tumor stage, patients with avoid entirely or to at least delay the need for or re- supratentorial PNET have a poorer PFS and overall strict the volume and dose of radiation to the brain. survival rate than do patients with infratentorial PNET. In the Childrens Cancer Group CCG-921 study, chil- In the above cooperative group study, the in- dren younger than 18 months with medulloblas- fratentorial PNET were primarily medulloblastoma, toma/PNET were nonrandomly assigned to receive the whereas the supratentorial PNET carried other diag- 8-in-1 chemotherapy regimen with the intent of de- nostic labels; no differentiation was made between laying irradiation until the completion of at least 1 cerebral neuroblastomas and other supratentorial year of chemotherapy. In a publication reporting the 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 131

Primary Cerebral Tumors 131

CCG-921 study, all children younger than 18 months presence of leptomeningeal dissemination, surgical of age at diagnosis of pineoblastoma died following debulking of mass lesions is more problematic, al- tumor recurrence at a median of just 4 months from though at times it can contribute toward an improved diagnosis (Jakacki et al., 1995). level of palliation. The outcome for young children with nonpineal Postoperative chemotherapy approaches for pa- region supratentorial PNET is less clear. All nine chil- tients with tumor recurrence range from the use of dren aged 19 to 36 months treated with reduced-dose new investigational agents to myeloablative chemo- irradiation to the spine (23.4 Gy) and the primary therapy regimens with autologous hemopoietic stem site (45 Gy) as well as with chemotherapy relapsed cell rescue if the patient can be brought to a state of and died (Albright et al., 1995). However, those chil- minimal residual tumor either through surgical re- dren younger than 18 months of age treated with section of localized tumor or through reinduction 8-in-1 chemotherapy without irradiation achieved a chemotherapy. This intensive approach is particularly 55% 16% 3 year PFS, comparable with the sur- warranted for young children who have not received vival rate of children older than 36 months of age irradiation before recurrence and for the small pro- treated with full-dose irradiation and the same che- portion of patients for whom myeloablative chemo- motherapy regimen (Geyer et al., 1994). This un- therapy with autologous stem cell rescue followed by usually favorable outcome in a small number of pa- irradiation may afford cure of disease (Guruangan et tients (n 11) contrasts with other reports al., 1998). indicating equally poor outcomes for infants with pineal and nonpineal supratentorial PNET treated with chemotherapy alone (Duffner et al., 1993). CHOROID PLEXUS PAPILLOMAS To avoid irradiation for children younger than 6 AND CARCINOMAS years of age with supratentorial PNET and for chil- dren younger than 3 years of age with PNET in any Pathology location, the “Head Start I” chemotherapy regimen was developed (Mason et al., 1998). This regimen Benign or malignant proliferations of choroid plexus uses five cycles of intensive induction chemotherapy epithelium are termed papillomas or carcinomas, (cisplatin, vincristine, high-dose cyclophosphamide, respectively. Both are rare, occurring most often in and etoposide) followed by a single cycle of mye- the first decade of life (Matson and Crofton, 1960; loablative chemotherapy (thiotepa, etoposide, and Bigner et al., 1998). Congenital examples have been carboplatin) with autologous hemopoietic stem cell described (Tomita and Naidich, 1987). The papil- rescue. The most recent update (Golomb et al., 1999) loma is usually cured surgically (McGirr et al., 1988). indicates a 4 year PFS of almost 50% for patients with These tumors may arise at any choroid plexus site, supratentorial PNET, with avoidance of irradiation for but the lateral and third ventricles are more common more than 80% of patients. The more recent proto- sites in children, whereas the fourth ventricle is more col, “Head Start II,” being conducted in a multina- common in adults (Burger et al., 1991; Bigner et al., tional setting, employs the same regimen as Head Start 1998). Symptoms are usually referable to hydro- I but with intensification in induction using high-dose cephalus, a sequela either of ventricular outflow ob- methotrexate for patients with neuraxis dissemination struction by the mass or of active oversecretion of CSF at diagnosis and increasing the age of eligibility for by the proliferating choroid plexus epithelium, or patients with supratentorial tumors to 10 years. both (Boyd and Steinbok, 1987; Eisenberg et al., For recurrences of supratentorial PNET, the sub- 1974). sequent course of management must be tailored to The diagnosis of choroid plexus papilloma hinges the individual patient, depending on the extent of dis- on the presence of an unusually large mass of cho- ease at recurrence, prior exposure to irradiation roid plexus tissue and some cellular crowding and and/or chemotherapy, the patient’s general condition, pseudostratification (Burger et al., 1991) with almost and specific organ function at time of recurrence. For no mitotic activity. The histologic features are re- those with localized recurrence, radical surgical re- markably similar to those of normal choroid plexus section should be considered as a first step. In the epithelium. Foci of bone, cartilage, or oncocytic 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 132

132 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

change may be seen (Bonnin et al., 1987). Even his- massive ventriculomegaly that often results. In the tologically benign lesions have been reported to seed past, the mortality from surgery approached 20% to the subarachnoid space, but they do not invade the 30% as a result of excessive bleeding (Hawkins, cerebral parenchyma (Bigner et al., 1998). 1980), and neurologic morbidity was unacceptably Choroid plexus carcinomas are highly anaplastic, high. malignant, and invasive neoplasms (Lewis, 1967; The approach to the lesion depends on the posi- Packer et al., 1992; Paulus and Janisch, 1990). They tion it occupies in the ventricle. Lateral ventricular tu- appear as poorly differentiated carcinomas, with ul- mors are best approached through the posterior tem- trastructural features of adenocarcinomas, such as poral lobe, inferior to the middle gyrus (dominant cilia, microvilli, and well-formed cytoplasmic junc- hemisphere), and via the middle to superior gyrus in tions (Burger et al., 1991; Bigner et al., 1998). The the nondominant hemisphere (Spallone et al., 1990). differential diagnoses are intraventricular ependy- Tumors situated in the third ventricle may be ap- moma and metastatic carcinoma. The distinction proached by a transcortical (superior frontal gyrus) from ependymoma can usually be made by identify- incision or through the transcallosal route (Schijman ing the neoplastic stroma as either glial (ependy- et al., 1990). The tumor must be mobilized to find moma) or mesenchymal (choroid plexus neoplasm). the arterial pedicle from the anterior choroidal artery Immunostaining for glial fibrillary acidic protein is in the case of a lateral ventricular lesion entering into usually strongly and diffusely positive in an ependy- its inferomedial surface (Raimondi and Gutierrez, moma but focal in a choroid plexus tumor (Coffin et 1975). However, this approach is not always possi- al., 1986; Mannoji and Becker, 1988; Miettenen et ble due to the massive size of the tumor. In this set- al., 1986; Bigner et al., 1998). In contrast, cytoker- ting, the tumor must be removed piecemeal with bipo- atins are present in choroid plexus tumors as carci- lar coagulation (Tomita and Naidich, 1987) as noembryonic antigen may be present in carcinomas, opposed to an en bloc delivery, which is the ideal sur- whereas both agents are focal or absent in ependy- gical approach. In neonates and infants, the blood momas (Coffin et al., 1986; McComb and Burger, volume may range between 250 and 500 ml, thus re- 1983). quiring massive transfusions to replete the rapid The role of transthyretin (prealbumin) immuno- bleeding that often occurs with piecemeal removal. staining in making a diagnosis is still controversial Tumor in the third ventricle will protrude through (Herbert et al., 1990; Paulus and Janisch, 1990). Al- the foramen of Monro (Fig. 3–15) and derives its vas- though this substance is produced in normal choroid cular pedicle from the roof of the third ventricle. It plexus and immunoreactivity is seen in most papillo- may be necessary to split both fornices or to maneu- mas, the extent and intensity of reactivity decrease ver under the lateral ventricle choroid plexus to en- with increasing anaplasia (Matsushima et al., 1988). ter the mid to posterior third ventricle. Venous The presence of transthyretin immunoreactivity is, drainage also becomes a problem because of mas- however, not entirely specific for choroid plexus ep- sive distention due to increased tumor blood volume ithelium as other systemic (metastatic) neoplasms and the fragility of the draining veins in a young child. may also react (Albrecht et al., 1991). Once the tumor is removed, a drain should be left in In adults, the rarity of choroid plexus neoplasms place for 2 to 3 days to remove all bloody CSF and mandates that metastatic carcinoma be the preferred tumor debris. In most circumstances, a larger sub- diagnosis unless it is absolutely and rigorously elim- dural effusion will develop, although some surgeons inated (Gottschalk et al., 1993; Bigner et al., 1998). claim that this can be obviated by sealing the corti- cal opening with fibrin glue (Boyd and Steinbok, 1987). The treatment of choice for this problem if it Surgery is symptomatic or does not resolve within 3 to 6 Choroid plexus tumors present a difficult challenge months is to place a subdural–peritoneal shunt. to the surgeon because of their large size and enor- The etiology of the hydrocephalus is thought to be mous blood volume. These lesions are often found in secondary to overproduction of CSF (Eisenberg et al., the lateral and third ventricles during childhood and 1974), although the altered flow of CSF along with its in the fourth ventricle in adults. The goals of surgery elevated protein concentration may also be implicated include gross total resection and management of the (Rekate et al., 1985). Usually, removal of the tumor 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 133

Primary Cerebral Tumors 133

Figure 3–15. Choroid plexus papilloma (CPP) protruding through the foramen of Monro.

will reduce the hydrocephalus and the need for per- tients who received low-dose preoperative irradiation manent ventriculoperitoneal shunting, which is re- (approximately 30 Gy) demonstrated a significant re- quired in 20% or more of cases (Hawkins, 1980). duction in tumor size, obliteration of the tumor cap- Choroid plexus carcinomas should likewise be illary bed, and necrosis, although at this dose level completely resected if possible, as this has been more there was little change observed in neoplastic effective in determining long-term outcome than has choroidal cells (Carrea and Polak, 1977). Naguib et adjuvant therapy (Ellenbogen et al., 1989; St. Clair et al. (1981) reported a case of an inoperable cerebel- al., 1991). lopontine angle choroid plexus papilloma that exten- sively involved the adjacent mastoid bone. This pa- tient received 49.5 Gy in 32 treatment fractions over Radiation a period of 33 days. Within 16 months after comple- Because choroid plexus papillomas are often cured tion of radiation therapy, the mass was markedly re- by total surgical excision (McEvoy et al., 2000), not duced in size. Such anecdotal reports suggest that it much information exists about their response to ir- is reasonable to offer local irradiation to patients with radiation (Cohen and Duffner, 1984). The studies that inoperable or recurrent choroid plexus papillomas are available do not provide sufficient detail or data (Cohen and Duffner, 1984). to allow an assessment of the value of radiation ther- Radiation therapy may be beneficial for patients apy for these lesions. Pathology specimens from pa- with choroid plexus carcinomas (Ausman et al., 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 134

134 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

1984; Carpenter et al., 1982; Chow et al., 1999). Sub- tive chemotherapy may improve the ability to perform arachnoid tumor seeding has been identified in as a complete resection (Souweidane et al., 1999). How- many as 44% of cases (Ausman et al., 1984). There- ever, the outcomes for patients with incompletely re- fore consideration should be given to irradiating the sected or metastatic choroid plexus carcinoma is entire craniospinal axis; however, data to support this poor, even with the addition of radiation therapy approach are not available. Chow et al. (1999) rec- and/or conventional chemotherapy. Given that the ommend that patients with completely excised, local- carcinomas have shown sensitivity to chemotherapy, ized choroid plexus carcinomas be treated with ei- it is appropriate to advocate intensive chemotherapy ther adjunctive chemotherapy or limited-field regimens for children with newly diagnosed, incom- irradiation provided that their spinal MRI and CSF cy- pletely resected, and/or metastatic choroid plexus tology studies are negative. Because of potential in- carcinoma, with avoidance of irradiation, as most of tellectual and endocrinologic sequelae, radiotherapy these children will be younger than 6 years of age and is used sparingly for infants and young children. How- at greatest risk for tumor recurrence and profound ever, radiotherapy appears to be helpful to those also developmental and psychosocial dysfunction if sub- receiving chemotherapy who develop progressive dis- jected to cranial irradiation. In light of the impor- ease. Both chemotherapy and craniospinal axis irra- tance of radical surgical resection, any child with lo- diation may be indicated for patients with incom- calized but incompletely resected primary choroid pletely excised tumors and for those with evidence of plexus papilloma or carcinoma should be considered leptomeningeal spread (Chow et al., 1999). for reoperation to achieve gross total resection.

Chemotherapy REFERENCES

The role of chemotherapy in the management of cho- Afra D, Muller W, Benoist G, Schroder R. 1978. Supratentor- roid plexus tumors remains ill defined. Several phase ial recurrences of gliomas. Results of reoperations on as- II trials of chemotherapeutic agents in recurrent pe- trocytomas and oligodendrogliomas. Acta Neurochir diatric brain tumors have included occasional pa- (Wien) 43:217–227. tients with choroid plexus carcinomas, and objective Albrecht S, Rouah E, Becker LE, Bruner J. 1991. Transthyretin immunoreactivity in choroid plexus neoplasms and brain radiographic responses to such agents have been ob- metastases. Mod Pathol 4:610–614. served. A small number of reports have focused on Albright AL, Wisoff JH, Zeltzer PM, et al. 1995. Prognostic fac- the efficacy of chemotherapy for choroid plexus car- tors in children with supratentorial (nonpineal) primitive cinomas and have demonstrated objective radio- neuroectodermal tumors. A neurosurgical perspective from graphic responses to (1) a combination of cisplatin the Childrens Cancer Group. Pediatr Neurosurg 22:1–7. Allen JC, Bloom J, Ertel I, et al. 1986. Brain tumors in chil- and etoposide, (2) myeloablative doses of thiotepa dren: current cooperative and institutional chemotherapy and etoposide, (3) the 8-in-1 regimen ( J Finlay, trials in newly diagnosed and recurrent disease. Semin On- personal communications, 1993), and (4) cyclo- col 13:110–122. phosphamide-based therapies with or without intra- Allen JC, Walker R, Luks E, Jennings M, Barfoot S, Tan C. 1987. ventricular chemotherapy (VA Levin personal com- Carboplatin and recurrent childhood brain tumors. J Clin Oncol 5:459–463. munications, 1993; Souweidane et al., 1999). Ammirati M, Vick N, Liao Y, Ciric I, Mikhael M. 1987. Effect Even though chemotherapy has produced objec- of extent of surgical resection on survival and quality of life tive responses in patients with choroid plexus carci- in patients with supratentorial glioblastomas and anaplas- noma, it does not follow that such drugs have a clear tic astrocytomas. Neurosurgery 21:201–206. role in improving the outcome for newly diagnosed Arseni C, Petrovici IN. 1971. Epilepsy in temporal lobe tumors. Eur Neurol 5:201–214. tumors in children. It seems clear that the single most Artico M, Cervoni L, Celli P, Salvati M, Palma L. 1993. Supra- powerful prognostic factor of a favorable outcome for tentorial glioblastoma in children: a series of 27 surgically children with choroid plexus tumors is extent of sur- treated cases. Childs Nerv Syst 9:7–9. gical resection; children who undergo a successful Ashwal S, Hinshaw DB, Bedros A. 1984. CNS primitive neu- gross total resection of tumor at diagnosis have an roectodermal tumors of childhood. Med Pediatr Oncol 12:180–188. excellent survival; it is unclear whether such children Ausman JI, Schrontz C, Chason J, Knighton RS, Pak H, Patel S. require any additional therapy, either irradiation or 1984. Aggressive choroid plexus papilloma. Surg Neurol chemotherapy. Volumetric reduction with preopera- 22:472–476. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 135

Primary Cerebral Tumors 135

Awad IA, Rosenfeld J, Ahl J, Hahn JF, Luders H. 1991. In- hamartomas and tumors associated with tuberous sclero- tractable epilepsy and structural lesions of the brain: map- sis. Neurosurgery 4:410–417. ping, resection strategies and seizure outcome. Epilepsia Bonnin JM, Colon LE, Morawetz RB. 1987. Focal glial differ- 32:179–186. entiation and oncocytic transformation in choroid plexus Bampoe J, Bernstein M. 1999. The role of surgery in low grade papilloma. Acta Neuropathol (Berl) 72:277–280. gliomas. J Neurooncol 42:259–269. Bouchard J. 1980. Central nervous system. In: Fletcher GH Barnett GH, Kormos DW, Steiner CP, Weisenberger J. 1993. (ed), Textbook of Radiotherapy, 3rd ed. Philadelphia: Lea Use of a frameless, armless stereotactic wand for brain tu- & Febiger, p 444. mor localization with two-dimensional and three-dimen- Boyd MC, Steinbok P. 1987. Choroid plexus tumors: problems sional neuroimaging. Neurosurgery 33:674–678. in diagnosis and management. J Neurosurg 66:800–805. Bauman GS, Cairncross JG. 2001. Multidisciplinary manage- Brain Tumor Registry of Japan. 1992. Neurol Medico-Chir ment of adult anaplastic oligodendrogliomas and anaplas- 32:385–547. tic mixed oligo-astrocytomas. Semin Radiat Oncol Buckner JC, Brown LD, Cascino TL, et al. 1990. Phase II eval- 11:170–180. uation of infusional etoposide and cisplatin in patients with Bauman G, Lote K, Larson D, et al. 1999a. Pretreatment fac- recurrent astrocytoma. J Neurooncol 9:249–254. tors predict overall survival for patients with low-grade Bullard DE, Rawlings CE 3d, Phillips B, et al. 1987. Oligo- glioma: a recursive partitioning analysis. Int J Radiat Oncol dendroglioma. An analysis of the value of radiation therapy. Biol Phys 45:923–929. Cancer 60:2179–2188. Bauman G, Pahapill P, Macdonald D, Fisher B, Leighton C, Burger PC, Dubois PJ, Schold SC, et al. 1983. Computerized Cairncross G. 1999b. Low grade glioma: a measuring ra- tomographic and pathologic studies of the untreated, qui- diographic response to radiotherapy. Can J Neurol Sci escent, and recurrent glioblastoma multiforme. J Neurosurg 26:18–22. 58:159–169. Berger MS. 1986. Ultrasound-guided stereotaxic biopsy using Burger PC, Green SB. 1987. Patient age, histologic features, a new apparatus. J Neurosurg 65:550–554. and length of survival in patients with glioblastoma multi- Berger MS, Cohen W, Ojemann GA. 1990b. Correlation of mo- forme. Cancer 59:1617–1625. tor cortex brain mapping data with magnetic resonance Burger PC, Heinz ER, Shibata T, Kleihues P. 1988. Topographic imaging. J Neurosurg 72:383–387. anatomy and CT correlations in the untreated glioblastoma Berger MS, Deliganis AV, Dobbins J, Keles GE. 1994. The ef- multiforme. J Neurosurg 68:698–704. fect of extent of resection on recurrence in patients with Burger PC, Rawlings CE, Cox EB, McLendon RF, Schold SC, low grade cerebral hemisphere gliomas. Cancer Bullard DE. 1987. Clinicopathologic correlations in the 74:1784–1791. oligodendroglioma. Cancer 59:1345–1352. Berger MS, Edwards MS, Wara WM, Levin VA, Wilson CB. 1983. Burger PC, Scheithauer BW, Vogel FS. 1991. Surgical Pathol- Primary cerebral neuroblastoma. Long-term follow-up re- ogy of the Nervous System and Its Coverings, 3rd ed. New view and therapeutic guidelines. J Neurosurg 59:418–423. York: Churchill Livingstone, 737 pp. Berger MS, Ghatan S, Haglund MM, Dobbins J, Ojemann GA. Burger PC, Vogel FS, Green SB, Strike TA. 1985. Glioblastoma 1993. Low-grade gliomas associated with intractable multiforme and anaplastic astrocytoma. Pathologic criteria epilepsy: seizure outcome utilizing electrocorticography and prognostic implications. Cancer 56:1106–1111. during tumor resection. J Neurosurg 79:62–69. Cairncross JG, Laperriere NJ. 1989. Low-grade glioma. To treat Berger MS, Ojemann GA, Lettich E. 1990a. Neurophysiologi- or not to treat? Arch Neurol 46:1238–1239. cal monitoring during astrocytoma surgery. Neurosurg Clin Cairncross G, Macdonald D, Ludwin S, et al. 1994. Chemo- Am 1:65–80. therapy for anaplastic oligodendroglioma. National Cancer Berger MS, Tucker A, Spence A, Winn HR. 1992. Reoperation Institute of Canada Clinical Trials Group. J Clin Oncol for glioma. Clin Neurosurg 39:172–186. 12:2013–21. Bertolone SJ, Baum ES, Krivit W, Hammond GD. 1989. A phase Cairncross JG, Ueki K, Zlatescu MC, et al. 1998. Specific ge- II study of cisplatin therapy in recurrent childhood brain netic predictors of chemotherapeutic response and survival tumors. A report from the Childrens Cancer Study Group. in patients with anaplastic oligodendrogliomas. J Natl Can- J Neurooncol 7:5–11. cer Inst 90:1473–1479. Bigner DD, McLendon R, Bruner JM. 1998. Russell and Ru- Cappabianca P, Spaziante R, Caputi F, et al. 1991. Accuracy of binstein’s Pathology of Tumors of the Nervous System, 6th the analysis of multiple small fragments of glial tumors ob- ed. London: Edward Arnold, 757 pp. tained by stereotactic biopsy. Acta Cytol 35:505–511. Black PM, Moriarty T, Alexander E 3d, et al. 1997. Develop- Carpenter DB, Michelsen WJ, Hays AP. 1982. Carcinoma of the ment and implementation of intraoperative magnetic reso- choroid plexus. Case report. J Neurosurg 56:722–727. nance imaging and its neurosurgical applications. Neuro- Carrea R, Polak M. 1977. Preoperative radiotherapy in the surgery 41:831–845. management of posterior fossa choroid plexus papillomas. Bleehen NM. 1990. Studies of high grade cerebral gliomas. Int Childs Brain 3:12–24. J Radiat Oncol Biol Phys 18:811–813. Cascino GD. 1990. Epilepsy and brain tumors: implications for Bloom HJ. 1982. Intracranial tumors: response and resistance treatment. Epilepsia 31:S37–S44. to therapeutic endeavors, 1970–1980. Int J Radiat Oncol Cascino GD, Kelly PJ, Hirschorn KA, Marsh WR, Sharbrough Biol Phys 8:1083–1113. FW. 1990. Stereotactic resection of intra-axial cerebral le- Boesel CP, Paulson GW, Kosnik EJ, Earle KM. 1979. Brain sions in partial epilepsy. Mayo Clin Proc 65:1053–1060. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 136

136 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Castro JR, Saunders WM, Tobias CA, et al. 1982. Treatment of Cohen BH, Zeltzer PM, Boyett JM, et al. 1995. Prognostic fac- cancer with heavy charged particles. Int J Radiat Oncol Biol tors and treatment results for supratentorial primitive neu- Phys 8:2191–2198. roectodermal tumors in children using radiation and che- Celli P, Nofrone I, Palma L, Cantore G, Fortuna A. 1994. Cere- motherapy: a Childrens Cancer Group randomized trial. J bral oligodendroglioma: prognostic factors and life history. Clin Oncol 13:1687–1696. Neurosurgery 35:1018–1035. Cohen ME, Duffner PK. 1984. Miscellaneous: primitive neu- Chamberlain MC, Kormanik PA. 1999. Salvage chemotherapy roectodermal tumors, oligodendrogliomas, choroid plexus with tamoxifen for recurrent anaplastic astrocytomas. Arch papillomas. In: Cohen ME, Duffner PK (eds) Brain Tumors Neurol 56:703–708. in Children: Principles of Diagnosis and Treatment. New Chamberlain MC, Levin VA. 1989. Chemotherapeutic treatment York: Raven Press, p 273. of the diencephalic syndrome. A case report. Cancer Concannon JP, Kramer S, Berry R. 1960. The extent of in- 63:1681–1684. tracranial gliomas at autopsy and its relationship to tech- Chamberlain MC, Prados MD, Silver P, Levin VA. 1988. A phase niques used in radiation therapy of brain tumors. Am J I/II study of 24 hour intravenous AZQ in recurrent primary Roentgen 84:99–107. brain tumors. J Neurooncol 6:319–323. Coons SW, Johnson PC, Pearl DK. 1997. The prognostic sig- Chandler KL, Prados MD, Malec M, Wilson CB. 1993. Long- nificance of Ki-67 labeling indices for oligodendrogliomas. term survival in patients with glioblastoma multiforme. Neu- Neurosurgery 41:878–884. rosurgery 32:716–720. Coyle T, Baptista J, Winfield J, et al. 1990. Mechlorethamine, Chandrasoma PT, Smith MM, Apuzzo ML. 1989. Stereotactic vincristine, and procarbazine chemotherapy for recurrent biopsy in the diagnosis of brain masses: comparison of re- high-grade glioma in adults: a phase II study. J Clin Oncol sults of biopsy and resected surgical specimen. Neuro- 8:2014–2018. surgery 24:160–165. Curran WJ, Scott CB, Horton J, et al. 1992a. Does extent of Chang CH. 1977. Hyperbaric oxygen and radiation therapy in surgery influence outcome for astrocytoma with atypical or the management of glioblastoma. Natl Cancer Inst Monogr anaplastic foci (AAF)? A report from three Radiation Ther- 46:163–169. apy Oncology Group (RTOG) trials. J Neurooncol 12:219– Chang CH, Horton J, Schoenfeld D, et al. 1983. Comparison 227. of postoperative radiotherapy and combined postoperative Curran WJ, Scott CB, Horton J, et al. 1993. Recursive parti- radiotherapy and chemotherapy in the multidisciplinary tioning analysis of prognostic factors in three Radiation management of malignant gliomas. A joint Radiation Ther- Therapy Oncology Group malignant glioma trials. J Natl Can- apy Oncology Group and Eastern Cooperative Oncology cer Inst 85:704–710. Group study. Cancer 52:997–1007. Curran WJ, Scott CB, Nelson JS, et al. 1992b. Survival com- Chastagner P, Sommelet-Olive D, Kalifa C, et al. 1993. Phase parison of radiosurgery eligible and ineligible malignant II study of ifosfamide in childhood brain tumors: a report glioma patients treated with hyperfractionated radiation by the French Society of Pediatric Oncology (SFOP). Med therapy (RT) and BCNU: a report of Radiation Therapy On- Pediatr Oncol 21:49–53. cology Group (RTOG) 83-02. Int J Radiat Oncol Biol Phys Chin HW, Hazel JJ, Kim TH, Webster JH. 1980. Oligoden- 24(suppl 1):A29. drogliomas. I. A clinical study of cerebral oligoden- Daumas-Duport C. 1993. Dysembryoplastic neuroepithelial tu- drogliomas. Cancer 45:1458–1466. mours. Brain Pathol 3:283–295. Chin HW, Maruyama Y, Markesbery W, Young AB. 1982. In- Daumas-Duport C, Scheithauer BW, Chodkeiwicz JP, Laws ER, tracranial ependymoma. Results of radiotherapy at the Uni- Vedrenne C. 1988a. Dysembryoplastic neuroepithelial tu- versity of Kentucky. Cancer 49:2276–2280. mor: a surgically curable tumor of young patients with in- Chiu JK, Woo SY, Ater J, et al. 1992. Intracranial ependymoma tractable partial seizures. Report of thirty-nine cases. Neu- in children: analysis of prognostic factors. J Neurooncol rosurgery 23:545–556. 13:283–290. Daumas-Duport C, Scheithauer BW, Kelly PJ. 1987. A histo- Choucair AK, Levin VA, Gutin PH, et al. 1986. Development of logic and cytologic method for the spatial definition of multiple lesions during radiation therapy and chemother- gliomas. Mayo Clin Proc 62:435–449. apy in patients with gliomas. J Neurosurg 65:654–658. Daumas-Duport C, Scheithauer BW, O’Fallon J, Kelly P. 1988b. Chow E, Reardon DA, Shah AB, et al. 1999 Pediatric choroid Grading of astrocytomas. A simple and reproducible plexus neoplasms. Int J Radiat Oncol Biol Phys 44:249–254. method. Cancer 62:2152–2165. Ciric I, Ammirati M, Vick N, Mikhael M. 1987. Supratentorial Daumas-Duport C, Tucker ML, Kolles H, et al. 1997. Oligo- gliomas: surgical considerations and immediate postoper- dendrogliomas. Part II: a new grading system based on mor- ative results. Gross total resection versus partial resection. phological and imaging criteria. J Neurooncol 34:61–78. Neurosurgery 21:21–26. Davis RL, Onda K, Shubuya M, Lamborn K, Hoshino T. 1995. Clark GB, Henry JM, McKeever PE. 1985. Cerebral pilocytic Proliferation markers in gliomas: a comparison of BUDR, astrocytoma. Cancer 56:1128–1133. KI-67, and MIB-1. J Neurooncol 24:9–12. Coffey RJ, Lunsford LD, Taylor FH. 1988. Survival after stereo- de Chadarevian JP, Pattisapu JV, Faerber EN. 1990. Desmo- tactic biopsy of malignant gliomas. Neurosurgery 22:465– plastic cerebral astrocytoma of infancy. Light microscopy, 473. immunocytochemistry, and ultrastructure. Cancer 66:173– Coffin CM, Wick MR, Braun JT, Dehner LP. 1986. Choroid 179. plexus neoplasms. Clinicopathologic and immunohisto- Decker DA, Al Sarraf M, Kresge C, Austin D, Wilner HI. 1985. chemical studies. Am J Surg Pathol 10:394–404. Phase II study of aziridinylbenzoquinone (AZQ: NSC- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 137

Primary Cerebral Tumors 137

182986) in the treatment of malignant gliomas recurrent rial malignant brain gliomas—a phase II study. Eur J Can- after radiation. Preliminary report. J Neurooncol 3:19–21. cer Clin Oncol 21:143–146. Dehghani F, Schachenmayr W, Laun A, Krof HW. 1998. Prog- Ettinger LJ, Ru N, Krailo M, Ruccione KS, Krivit W, Hammond nostic implication of histopathological, immunohistochem- GD. 1990. A phase II study of diaziquone in children with ical and clinical features of oligodendrogliomas: a study of recurrent or progressive primary brain tumors: a report 89 cases. Acta Neuropathol (Berl) 95:493–504. from the Childrens Cancer Study Group. J Neurooncol Deutsch M, Green SB, Strike TA, et al. 1989. Results of a 9:69–76. randomized trial comparing BCNU plus radiotherapy, Eyre HJ, Crowley JJ, Townsend JJ, et al. 1993. A randomized streptozotocin plus radiotherapy, BCNU plus hyper- trial of radiotherapy versus radiotherapy plus CCNU for in- frationated radiotherapy, and BCNU following misonida- completely resected low-grade gliomas: a Southwest On- zole plus radiotherapy in the postoperative treatment of cology Group study. J Neurosurg 78:909–914. malignant glioma. Int J Radiat Oncol Biol Phys 16:1389– Fadul C, Wood J, Thaler H, Galicich J, Patterson RH, Posner 1396. JB. 1988. Morbidity and mortality of craniotomy for exci- Di Lorenzo N, Esposito V, Lunardi P, Delfini R, Fortuna A, Can- sion of supratentorial gliomas. Neurology 38:1374–1379. tore G. 1991. A comparison of computerized tomography- Fazekas JT. 1977. Treatment of grades I and II brain astrocy- guided stereotactic and ultrasound-guided techniques for tomas. The role of radiotherapy. Int J Radiat Oncol Biol brain biopsy. J Neurosurg 75:763–765. Phys 2:661–666. Dinapoli RP, Brown LD, Arusell RM, et al. 1993. Phase III Finlay JL. 1992b. High-dose chemotherapy followed by bone comparative evaluation of PCNU and carmustine combined marrow “rescue” for recurrent brain tumors. In: Packer RJ, with radiation therapy for high-grade glioma. J Clin Oncol Bleyer WA, Pochedly C (eds), Pediatric Neuro-oncology: 11:1316–1321. New Trends in Clinical Research. New York: Harwood Aca- Dische S. 1991. A review of hypoxic cell radiosensitization. Int demic Publishers, pp 278–297. J Radiat Oncol Biol Phys 20:147–152. Finlay JL, August C, Packer R, et al. 1990. High-dose multi- Djordjevic B, Szybalski W. 1960. Genetics of human cell lines. agent chemotherapy followed by bone marrow “rescue” for III. Incorporation of 5-bromo and 5-iododeoxyuridine into malignant astrocytomas of childhood and adolescence. the deoxyribonucleic acid of human cells and its effect on J Neurooncol 9:239–248. radiation sensitivity. J Exp Med 112:509–531. Finlay JL, Boyett JM, Yates AJ, et al. 1995. Randomized phase Drake J, Hoffman HJ, Kobayashi J, Hwang P, Becker LE. 1987. III trial in childhood high-grade astrocytoma comparing Surgical management of children with temporal lobe vincristine, lomustine, and prednisone with the eight-drugs- epilepsy and mass lesions. Neurosurgery 21:792–797. in-1-day regimen. Childrens Cancer Group. J Clin Oncol 13: Duffner PK, Cohen ME, Heffner RR, Freeman AI. 1981. Prim- 112–123. itive neuroectodermal tumors of childhood. An approach Finlay JL, Geyer JR, Turski PA, et al. 1994. Pre-irradiation che- to therapy. J Neurosurg 55:376–381. motherapy in children with high-grade astrocytoma: tumor Duffner PK, Horowitz ME, Krischer JP, et al. 1993. Postoper- response to two cycles of the “8-drug-in-1-day” regimen. A ative chemotherapy and delayed radiation in children less Childrens Cancer Group study CCG-945. J Neurooncol than three years of age with malignant brain tumors. N Engl 21:255–265. J Med 328:1725–1731. Fitzek MM, Thornton AF, Rabinov JD, et al. 1999. Accelerated Duong DH, Rostomily RC, Haynor DR, Keles GE, Berger MS. fractionation proton/photon irradiation to 90 cobalt gray 1992. Measurement of tumor resection volumes from com- equivalent for glioblastoma multiforme: results of a phase puterized images. Technical note. J Neurosurg 77:151–154. II prospective trial. J Neurosurg 91:251–260. Eddy DM. 1990. Designing a practice policy. Standards, guide- Florell RC, Macdonald DR, Irish WD, et al. 1992. Selection lines, and options. JAMA 263:3077–3084. bias, survival, and brachytherapy for glioma. J Neurosurg Eisenberg HM, McComb JG, Lorenzo AV. 1974. Cerebrospinal 76:179–183. fluid overproduction and hydrocephalus associated with Flowers A, Gleason MJ, Levin VA, et al. 1993. Combination che- choroid plexus papilloma. J Neurosurg 40:381–385. motherapy with carboplatin, 5-fluorouracil (5FU) and pro- Ellenberg L, McComb JG, Siegel SE, Stowe S. 1987. Factors af- carbazine (PCB) for recurrent malignant gliomas. Proc Am fecting intellectual outcome in pediatric brain tumor pa- Soc Clin Oncol 12:180. tients. Neurosurgery 21:638–644. Fokes EC, Earle KM. 1969. Ependymomas: clinical and patho- Ellenbogen RG, Winston KR, Kupsky WJ. 1989. Tumors of the logical aspects. J Neurosurg 30:585–594. choroid plexus in children. Neurosurgery 25:327–335. Fowler JF. 1990. How worthwhile are short schedules in ra- Elliott TE, Buckner JC, Cascino TL, Levitt R, O’Fallon JR, Schei- diotherapy? A series of exploratory calculations. Radiother thauer BW. 1991. Phase II study of ifosfamide with mesna Oncol 18:165–181. in adult patients with recurrent diffuse astrocytoma. J Neu- Franceschetti S, Binelli S, Casazza M, et al. 1990. Influence of rooncol 10:27–30. surgery and antiepileptic drugs on seizures symptomatic of Enzmann DR, Irvin KM, Marshall WH, Silverberg GD, Britt RH, cerebral tumours. Acta Neurochir (Wien) 103:47–51. Hanbery JW. 1984. Intraoperative sonography through a Friedman HS, Petros WP, Friedman AH, et al. 1999. Irinote- burr hole: guide for brain biopsy. AJNR Am J Neuroradiol can therapy in adults with recurrent or progressive malig- 5:243–246. nant glioma. J Clin Oncol 17:1516–1525. EORTC Brain Tumour Cooperative Group. 1985. Effect of Fulling KH, Nelson JS. 1984. Cerebral astrocytic neoplasms in AZQ (1,4-cyclohexadiene-1,4-diacarbamic acid-2,5-bis(1- the adult: contribution of histologic examination to the as- aziridinyl)-3,6-dioxodiethylester) in recurring supratento- sessment of prognosis. Semin Diagn Pathol 1:152–163. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 138

138 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Gaffney CC, Sloane JP, Bradley NJ, Bloom HJ. 1985. Primitive in patients presenting with a chronic seizure disorder. Clin neuroectodermal tumors of the cerebrum. Pathology and Neurosurg 33:15–42. treatment. J Neurooncol 3:23–33. Goldwein JW, Corn BW, Finlay JL, Packer RJ, Rorke LB, Gajjar A, Sanford RA, Heideman R, et al. 1997. Low-grade as- Schut L. 1991. Is craniospinal irradiation required to cure trocytoma: a decade of experience at St. Jude Children’s Re- children with malignant (anaplastic) intracranial ependy- search Hospital. J Clin Oncol 15:2792–2799. momas? Cancer 67:2766–2771. Galanis E, Buckner JC, Burch PA, et al. 1998. Phase II trial of Goldwein JW, Leahy JM, Packer RJ, et al. 1990. Intracranial nitrogen mustard, vincristine, and procarbazine in patients ependymomas in children. Int J Radiat Oncol Biol Phys with recurrent glioma: North Central Cancer Treatment 19:1497–1502. Group results. J Clin Oncol 16:2953–2958. Goldwein JW, Lefkowitz I. 1991. Advances in ependymomas: Galloway PG, Roessmann U. 1986. Anaplastic astrocytoma radiation and chemotherapy. In: Packer RJ, Bleyer WA, mimicking metastatic carcinoma. Am J Surg Pathol 10:728– Pochedly C (eds), Pediatric Neuro-oncology: New Trends 732. in Clinical Research. New York: Harwood Academic Pub- Gannett DE, Wisbeck WM, Silbergeld DL, Berger MS. 1994. lishers, pp 242–253. The role of postoperative irradiation in the treatment of Golomb L, Gardner S, Sapp M, et al. 1999. Chemotherapy only oligodendroglioma. Int J Radiat Oncol Biol Phys 30:567– for young children with newly-diagnosed malignant brain 573. tumors: an update on the “Head Start” protocol. Med Pe- Garcia DM, Fulling KH. 1985. Juvenile pilocytic astrocytoma diatr Oncol 33:201. of the cerebrum in adults. A distinctive neoplasm with fa- Gonzalez D, Elvidge AR. 1962. On the occurrence of epilepsy vorable prognosis. J Neurosurg 63:382–386. caused by astrocytoma of the cerebral hemispheres. J Neu- Garcia DM, Fulling KH, Marks JE. 1985. The value of radia- rosurg 19:470–482. tion therapy in addition to surgery for astrocytomas of the Gottschalk J, Jautzke G, Paulus W, Goebel S, Cervos-Navarro adult cerebrum. Cancer 55:919–927. J. 1993. The use of immunomorphology to differentiate cho- Garrett PG, Simpson WJ. 1983. Ependymomas: results of ra- roid plexus tumors from metastatic carcinomas. Cancer diation treatment. Int J Radiat Oncol Biol Phys 9:1121– 72:1343–1349. 1124. Grant JW, Gallagher PJ. 1986. Pleomorphic xanthoastrocy- Gaspar LE, Fisher BJ, Macdonald DR, et al. 1992. Supratento- toma: Immunohistochemical methods for differentiation rial malignant glioma: patterns of recurrence and implica- from fibrous histiocytomas with similar morphology. Am J tions for external beam local treatment. Int J Radiat Oncol Sug Pathol 10:336–341. Biol Phys 24:55–57. Grant R, Liang BC, Page MA, Crane DL, Greenberg HS, Gaynon PS, Ettinger LJ, Baum ES, Siegel SE, Krailo MD, Ham- Junck L. 1995. Age influences chemotherapy response in mond GD. 1990. Carboplatin in childhood brain tumors. A astrocytomas. Neurology 45:929–933. Children’s Cancer Study Group phase II trial. Cancer Green SB, Shapiro WR, Burger PC, et al. 1994. A randomized 66:2465–2469. trial of interstitial radiotherapy (RT) for newly diagnosed Geyer JR, Finlay JL, Boyett JM, et al. 1995. Survival of infants malignant glioma: Brain Tumor Cooperative Group (BTCG) with malignant astrocytomas: a report from the Childrens trial 8701. Proc Am Soc Clin Oncol 13:174. Cancer Group. Cancer 75:1045–1050. Griffin BR, Silbergeld DL, Berger MS, Wisbeck W, Gannett DE. Geyer JR, Zeltzer PM, Boyett JM, et al. 1994. Survival of infants 1992. Oligodendrogliomas: postoperative radiotherapy in- with primitive neuroectodermal tumors or malignant creases survival. Int J Radiat Oncol Biol Phys 24(suppl ependymomas of the CNS treated with eight drugs in 1 day: 1):142. a report from the Childrens Cancer Group. J Clin Oncol Grovas AC, Boyett JM, Lindsley K, Rosenblum M, Yates AJ, Fin- 12:1607–1615. lay JL. 1999. Regimen related toxicity of myeloablative che- Gherardi R, Baudrimont M, Nguyen JP, et al. 1986. Monstro- motherapy with BCNU, thiotepa and etoposide followed by cellular heavily lipidized malignant glioma. Acta Neu- autologous stem cell rescue for children. Med Pediatr On- ropathol (Berl) 69:28–32. col 33:83–87. Giannini C, Scheithauer BW, Burger PC, et al. 1999. Pleo- Groves MD, Maor MH, Meyers C, et al. 1999. Kyritsis AP, morphic xanthoastrocytoma: what do we really know about Jaeckle KA, Yung WKA, Sawaya RE, Hess K, Bruner JM, Pe- it? Cancer 85:2033–2045. terson P, Levin VA. A phase II trial of high-dose bromod- Glass J, Hochberg FH, Gruber ML, Louis DN, Smith D, Rattner eoxyuridine with accelerated fractionation radiotherapy fol- B. 1992. The treatment of oligodendrogliomas and mixed lowed by procarbazine, lomustine, and vincristine for oligodendroglioma-astrocytomas with PCV chemotherapy. glioblastoma multiforme. Int J Radiat Oncol Biol Phys. J Neurosurg 76:741–745. 45:127–135. Goffman TE, Dachowski LJ, Bobo H, et al. 1992. Long-term Guidelines and Outcomes Committee of the AANS. 1998. Prac- follow-up on National Cancer Institute phase I/II study of tice parameters in adults with suspected or known supra- glioblastoma multiforme treated with iododeoxyuridine and tentorial nonoptic pathway low-grade glioma. Neurosurg hyperfractionated irradiation. J Clin Oncol 10:264–268. Focus 4, Article 10. Goldring S, Gregorie EM. 1984. Surgical management of Guruangan S, Dunkel IJ, Goldman S, et al. 1998. Myeloabla- epilepsy using epidural recordings to localize the seizure tive chemotherapy with autologous bone marrow rescue in focus. Review of 100 cases. J Neurosurg 60:457–466. young children with recurrent malignant brain tumors. Goldring S, Rich KM, Picker S. 1986. Experience with gliomas J Clin Oncol 16:2486–2493. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 139

Primary Cerebral Tumors 139

Guthrie BL, Laws ER. 1990. Supratentorial low-grade gliomas. Herpers MJ, Budka H. 1984. Glial fibrillary acidic protein Neurosurg Clin North Am 1:37–48. (GFAP) in oligodendroglial tumors: gliofibrillary oligoden- Gutin PH, Leibel SA, Wara WM, et al. 1987. Recurrent malig- droglioma and transitional oligoastrocytoma as subtypes of nant gliomas: survival following interstitial brachytherapy oligodendroglioma. Acta Neuropathol (Berl) 64:265–272. with high-activity iodine-125 sources. J Neurosurg 67:864– Hess KR, Wong ET, Jaeckle KA, et al. 1999. Response and pro- 873. gression in recurrent malignant glioma. Neuro-Oncology Gutin PH, Prados MD, Phillips TL, et al. 1991. External irra- 1:282–288. diation followed by an interstitial high activity iodine-125 Hessler RB, Lopes MB, Frankfurter A, Reidy J, VandenBerg SR. implant “boost” in the initial treatment of malignant 1992. Cytoskeletal immunohistochemistry of central neu- gliomas: NCOG study 6G-82-2. Int J Radiat Oncol Biol Phys rocytomas. Am J Surg Pathol 16:1031–1038. 21:601–606. Hirsch JF, Sainte Rose C, Pierre-Khan A, Pfister A, Hoppe- Gutin PH, Wilson CB, Kumar AR, et al. 1975. Phase II study of Hirsch E. 1989. Benign astrocytic and oligodendrocytic tu- procarbazine, CCNU, and vincristine combination chemo- mors of the cerebral hemispheres in children. J Neurosurg therapy in the treatment of malignant brain tumors. Cancer 70:568–572. 35:1398–1404. Hochberg FH, Pruitt A. 1980. Assumptions in the radiotherapy Haddad SF, Moore SA, Menezes AH, VanGilder JC. 1992. Gan- of glioblastoma. Neurology 30:907–911. glioglioma: 13 years of experience. Neurosurgery 31:171–178. Horn B, Heideman R, Geyer R, et al. 1999. A multi-institutional Haddad SF, Moore SA, Schelper RL, Goeken J. 1991. Smooth retrospective study of intracranial ependymoma in children: muscle cells can comprise the sarcomatous component of identification of risk factors. J Pediatr Hematol Oncol gliosarcomas. J Neuropathol Exp Neurol 50:291A. 21:203–211. Haglund MM, Berger MS, Kunkel DD, Franck JE, Ghatan S, Horten BC, Rubinstein LJ. 1976. Primary cerebral neuroblas- Ojemann GA. 1992. Changes in gamma-aminobutyric acid toma. A clinicopathological study of 35 cases. Brain and somatostatin in epileptic cortex associated with low- 99:735–756. grade gliomas. J Neurosurg 77:209–216. Hoshino T, Ahn D, Prados MD, Lamborn K, Wilson CB. 1993. Haglund MM, Berger MS, Shamseldin M, Lettich E, Ojemann Prognostic significance of the proliferative potential of in- GA. 1993. Cortical localization of temporal lobe language tracranial gliomas measured by bromodeoxyuridine label- sites in patients with gliomas. Neurosurgery 34:567–576. ing. Int J Cancer 53:550–555. Halperin EC, Friedman HS, Schold SC, et al. 1993. Surgery, Hoshino T, Rodriguez LA, Cho KG, et al. 1988. Prognostic im- hyperfractionated craniospinal irradiation, and adjuvant plications of the proliferative potential of low-grade astro- chemotherapy in the management of supratentorial embry- cytomas. J Neurosurg 69:839–842. onal neuroepithelial neoplasms in children. Surg Neurol Humphrey GB, Dehner LP, Kaplan RJ, et al. 1981. Overview 40:278–283. on the management of primitive neuroectodermal tumors. Hansen DA, Spence AM, Carski T, Berger MS. 1993. Indocya- In: Humphrey GB, Dehner LP (eds), Pediatric Oncology I. nine green (ICG) staining and demarcation of tumor mar- The Hague: M Nijhoff, p 289. gins in a rat glioma model. Surg Neurol 40:451–456. Ito S, Chandler KL, Prados MD, Lamborn K, Wynne J, Malec Harsh GR, Levin VA, Gutin PH, Seager M, Silver P, Wilson CB. MK, Wilson CB, Davis RL, Hoshino T. 1994. Proliferative 1987. Reoperation for recurrent glioblastoma and anaplas- potential and prognostic evaluation of low-grade astrocy- tic astrocytoma. Neurosurgery 21:615–621. tomas. J Neurooncol 19:1–9. Hart MN, Earle KM. 1973. Primitive neuroectodermal tumors Jakacki RI, Zeltzer PM, Boyett JM, et al. 1995. Survival and of the brain in children. Cancer 32:890–897. prognostic factors following radiation and/or chemother- Hassoun J, Soylemezoglu F, Gambarelli D, Figarella-Branger apy for primitive neuroectodermal tumors of the pineal re- D, von Ammon K, Kleihues P. 1993. Central neurocytoma: gion in infants and children: a report of the Childrens Can- a synopsis of clinical and histological features. Brain Pathol cer Group. J Clin Oncol 13:1377–1383. 3:297–306. Jaros E, Perry RH, Adam L, et al. 1992. Prognostic implica- Hawkins JC 3d. 1980. Treatment of choroid plexus papillomas tions of p53 protein, epidermal growth factor receptor, in children: a brief analysis of twenty years’ experience. and Ki-67 labeling in brain tumours. Br J Cancer 66:373– Neurosurgery 6:380–384. 385. Healey EA, Barnes PD, Kupsky WJ, et al. 1991. The prognos- Jenkin D. 1981. Primitive neuroectodermal tumor. In: tic significance of postoperative residual tumor in ependy- Humphrey GB, Dehner LP (eds), Pediatric Oncology I. The moma. Neurosurgery 28:666–672. Hague: M Nijhoff, p 243. Heegard S, Sommer HM, Broholm H, Broendstrup O. 1995. Jeremic B, Grujicic D, Jevremovic S, et al. 1992. Carboplatin Proliferating cell nuclear antigen and Ki-67 immunohisto- and etoposide chemotherapy regimen for recurrent malig- chemistry of oligodendrogliomas with special reference to nant glioma: a phase II study. J Clin Oncol 10:1074–1077. prognosis. Cancer 76:1809–1813. Kaba SE, Kyritsis AP, Conrad C, et al. 1997. The treatment of Heilbrun MP, Roberts TS, Apuzzo ML, Wells TH, Sabshin JK. recurrent cerebral gliomas with all-trans-retinoic acid 1983. Preliminary experience with Brown-Roberts-Wells (tretinoin). J Neurooncol 34:145–151. (BRW) computerized tomography stereotaxic guidance sys- Kalyan-Raman UP, Olivero WC. 1987. Ganglioglioma: a cor- tem. J Neurosurg 59:217–222. relative clinicopathological and radiological study of ten Herbert J, Cavallaro T, Dwork AJ. 1990. A marker for primary surgically treated cases with follow-up. Neurosurgery choroid plexus neoplasms. Am J Pathol 136:1317–1325. 20:428–433. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 140

140 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Kaplan RS. 1993. Supratentorial malignant gliomas: risk pat- Kernohan JW, Mabon RF, Svien HJ, Adson AW. 1949. A sim- terns and therapy. J Natl Cancer Inst 85:690–691. plified classification of the gliomas. Proc Staff Meet Mayo Karim AB, Cornu P, Bleehan N, et al. 1998. Immediate post- Clin 24:71–75. operative radiotherapy in low grade glioma improves pro- Key G, Becker MH, Baron B, et al. 1993. New Ki-67–equiva- gression free survival but not overall survival: preliminary lent murine monoclonal antibodies (MIB 1–3) generated results of an EORTC/MRC randomized phase III trial. Proc against bacterially expressed parts of the Ki-67 cDNA con- Am Soc Clin Oncol 17:400A. taining three 62 base pair repetitive elements encoding for Karim AB, Maat B, Hatlevoll R, et al. 1996. A randomized trial the Ki-67 epitope. Lab Invest 68:629–636. on dose–response in radiation therapy of low-grade cere- Kiebert GM, Curran D, Aaronson NK, et al. 1998. Quality of bral glioma: European Organization for Research and Treat- life after radiation therapy of cerebral low-grade gliomas of ment of Cancer (EORTC) Study 22844. Int J Radiat Oncol the adult: results of a randomised phase III trial on dose Biol Phys 36:549–556. response (EORTC trial 22844). EORTC Radiotherapy Co- Karlsson UL, Leibel SA, Wallner K, et al. 1992. Brain. In: Perez operative Group. Eur J Cancer 34:1902–1909. CA, Brady LW (eds), Principles and Practice of Radiation Kim DG, Chi JG, Park SH, et al. 1992. Intraventricular neuro- Oncology, 2nd ed. Philadelphia: JB Lippincott, p 515. cytoma: clinicopathological analysis of seven cases. J Neu- Keim H, Potthoff PC, Schmidt K, Schiebusch M, Neiss A, Trott rosurg 76:759–765. KR. 1987. Survival and quality of life after continuous ac- Kim YH, Fayos JV. 1977. Intracranial ependymomas. Radiol- celerated radiotherapy of glioblastomas. Radiother Oncol ogy 124:805–808. 9:21–26. Kim TS, Halliday AL, Hedley-Whyte ET, Convery K. 1991. Cor- Keles GE, Anderson B, Berger MS. 1999. The effect of extent relates of survival and the Daumas-Duport grading system of resection on time to tumor progression and survival in for astrocytomas. J Neurosurg 74:27–37. patients with glioblastoma multiforme of the cerebral hemi- Kleihues P, Cavenee WK. 2000. Pathology and Genetics of Tu- sphere. Surg Neurol 52:371–379. mors of the Nervous System. Lyon: IARC Press. Kelly PJ. 1988. Volumetric stereotactic surgical resection of Kleihues P, Ohgaki H. 1997. Genetics of glioma progression intra-axial brain mass lesions. Mayo Clin Proc 63:1186– and the definition of primary and secondary glioblastoma. 1198. Brain Pathol 7:1131–1136. Kelly PJ. 1990. Image-directed tumor resection. In: Rosenblum Kleihues P, Ohgaki H. 1999. Primary and secondary glioblas- ML (ed), Neurosurgery Clinics of North America. The Role toma: from concept to clinical diagnosis. Neuro-Oncology of Surgery in Brain Tumor Management, vol 1. Philadel- 1:44–51. phia: WB Saunders, pp 81–95. Knapp J, Francini D, Van Eys J, Cangir A. 1981. Primitive neu- Kelly PJ. 1993. Computed tomography and histologic limits in roectodermal tumors of brain in childhood: literature re- glial neoplasms: tumor types and selection for volumetric view and the M. D. Anderson experience. In: Humphrey GB, resection. Surg Neurol 39:458–465. Dehner LP (eds), Pediatric Oncology I. The Hague: M Nij- Kelly PJ, Alker GJ, Goerss S. 1982. Computer-assisted stereo- hoff, p 215. tactic microsurgery for the treatment of intracranial neo- Koeller KK, Dillon WP. 1992. Dysembryoplastic neuroepithe- plasms. Neurosurgery 10:324–331. lial tumors: MR appearance. AJNR Am J Neuroradiol Kelly PJ, Daumas-Duport C, Kispert DB, Kall BA, Scheithauer 13:1319–1325. BW, Illig JJ. 1987. Imaging-based stereotaxic serial biop- Kornblith PK, Welch WC, Bradley MK. 1993. The future of ther- sies in untreated intracranial glial neoplasms. J Neurosurg apy for glioblastoma. Surg Neurol 39:538–543. 66:865–874. Kosnik EJ, Boesel CP, Bay J, Sayers MP. 1978. Primitive neu- Kelly PJ, Kall B, Goerss S, Alker GJ. 1983. Precision resection roectodermal tumors of the central nervous system in chil- of intra-axial CNS lesions by CT-based stereotactic cran- dren. J Neurosurg 48:741–746. iotomy and computer monitored CO2 laser. Acta Neurochir Kovalic JJ, Flaris N, Grigsby PW, Pirkowski M, Simpson JR, (Wein) 68:1–9. Roth KA. 1993. Intracranial ependymoma long term out- Kelly PJ, Kall BA, Goerss S, Earnest F. 1986. Computer-assisted come, patterns of failure. J Neurooncol 15:125–131. stereotaxic laser resection of intra-axial brain neoplasms. Kramer S. 1969. Tumor extent as a determining factor in ra- J Neurosurg 64:427–439. diotherapy of glioblastomas. Acta Radiol Ther Phys Biol Kepes JJ. 1993. Pleomorphic xanthoastrocytoma. The birth of 8:111–117. a diagnosis and concept. Brain Pathol 3:269–274. Kricheff II, Becker M, Schneck SA, Taveras JM. 1964. In- Kepes JJ, Rubinstein LJ, Ansbacher L, Schreiber DJ. 1989. tracranial ependymomas. A study of survival in 65 cases Histopathological features of recurrent pleomorphic xan- treated by surgery and irradiation. Am J Roentgenol thoastrocytomas: further corroboration of the glial nature 91:167–175. of this neoplasm. A study of 3 cases. Acta Neuropathol Kristiansen K, Hagen S, Kollevold T, et al. 1981. Combined mo- (Berl) 78:585–593. dality therapy of operated astrocytomas grade III and IV. Kepes JJ, Rubinstein LJ, Chiang H. 1984. The role of astrocytes Confirmation of the value of postoperative irradiation and in the formation of cartilage in gliomas. An immunohisto- lack of potentiation of bleomycin on survival time: a chemical study of four cases. Am J Pathol 117:471–483. prospective multicenter trial of the Scandinavian Glioblas- Kepes JJ, Rubinstein LJ, Eng LF. 1979. Pleomorphic xanthoas- toma Study Group. Cancer 47:649–652. trocytoma: a distinctive meningocerebral glioma of young Kros JM, de Jong AA, van der Kwast TH. 1992. Ultrastructural subjects with relatively favorable prognosis. A study of 12 characterization of transitional cells in oligodendrogliomas. cases. Cancer 44:1839–1852. J Neuropathol Exp Neurol 51:186–193. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 141

Primary Cerebral Tumors 141

Kros JM, Hop WC, Godschalk JJ, Krishnadath KK. 1996. Prog- gical management of low-grade astrocytoma of the cerebral nostic value of the proliferation-related antigen Ki-67 in hemispheres. J Neurosurg 61:665–673. oligodendrogliomas. Cancer 78:1107–1113. Lee SW, Fraass BA, Marsh LH, et al. 1999. Patterns of failure Kros JM, Pieterman H, van Eden CG, Avezaat CJ. 1994. Oligo- following high-dose 3-D conformal radiotherapy for high- dendroglioma: the Rotterdam-Dijkzigt experience. Neuro- grade astrocytomas: a quantitative dosimetric study. Int J surgery 34:959–966. Radiat Oncol Biol Phys 43:79–88. Kros JM, Troost D, Van Eden CG, van der Werf AJ, Uylings HB. Leibel SA. 1990. Teleradiotherapy: methods and expectations. 1988. Oligodendroglioma. A comparison of two grading In Apuzzo MLJ (ed), Malignant Cerebral Glioma. Park systems. Cancer 61:2251–2259. Ridge, IL: American Association of Neurological Surgeons, Kros JM, Van Eden CG, Stefanko SZ, Waayer-Van Batenburg M, p 159. van der Kwast TH. 1990. Prognostic implications of glial Leibel SA, Gutin, PH, Davis RL. 1991a. Factors affecting radi- fibrillary acidic protein containing cell types in oligoden- ation injury after interstitial brachytherapy for brain tumors. drogliomas. Cancer 66:1204–1212. In: Gutin PH, Leibel SA, Sheline GE (eds), Radiation Injury Krouwer HG, Davis RL, Silver P, Prados M. 1991. Gemistocytic to the Nervous System. New York: Raven Press, p 257. astrocytomas: a reappraisal. J Neurosurg 74:399–406. Leibel SA, Gutin PH, Sneed PK, et al. 1989a. Interstitial irra- Kun LE. 1992. Brain tumors in children. In: Perez CA, Brady diation for the treatment of primary and metastatic brain LW (eds), Principles and Practice of Radiation Oncology, tumors. Principles Practice Oncol Updates 3:1–11. 2nd ed. Philadelphia: JB Lippincott, p 417. Leibel SA, Gutin PH, Wara WM, et al. 1989b. Survival and qual- Kunschner LJ, Yung WK, Levin VA, Jaeckle KA. 1999. Carbo- ity of life after interstitial implantation of removable high- platin and 13-cis-retinoic acid for recurrent glioblastoma activity iodine-125 sources for the treatment of patients with multiforme. Neuro-Oncology 1:320. recurrent malignant gliomas. Int J Radiat Oncol Biol Phys Kyritsis AP, Yung WK, Bruner J, Gleason MJ, Levin VA. 1993. 17:1129–1139. The treatment of anaplastic oligodendrogliomas and mixed Leibel SA, Scott CB, Loeffler JS. 1994. Contemporary ap- gliomas. Neurosurgery 32:365–371. proaches to the treatment of malignant gliomas with radi- Kyritsis AP, Yung WK, Jaeckle KA, et al. 1996. Combination of ation therapy. Semin Oncol 21:198–219. 6-thioguanine, procarbazine, lomustine, and hydroxyurea Leibel SA, Scott CB, Pajak TF. 1991b. The management of ma- for patients with recurrent malignant gliomas. Neurosurgery lignant gliomas with radiation therapy: therapeutic results 39:921–926. and research strategies. Semin Radiat Oncol 1:32–49. Kyritsis AP, Yung WK, Leeds NE, Bruner J, Gleason MJ, Levin Leibel SA, Sheline GE. 1987. Radiation therapy for neoplasms VA. 1992. Multifocal cerebral gliomas associated with sec- of the brain. J Neurosurg 66:1–22. ondary malignancies [letter]. Lancet 339:1229–1230. Leibel SA, Sheline GE. 1990. Radiotherapy in the treatment of Lamborn KR, Prados MD, Kaplan SB, Davis RL. 1999. Final re- cerebral astrocytomas. In: Thomas DGT (ed), Neuro- port on the University of California San Francisco experi- oncology: Primary Malignant Brain Tumours. Baltimore: ence with bromodeoxyuridine labeling index as a prog- Johns Hopkins University Press, p 193. nostic factor for the survival of glioma patients. Cancer Leibel SA, Sheline GE. 1991. Tolerance of the brain and spinal 85:925–935. cord to conventional irradiation. In: Gutin PH, Leibel SA, Laperriere NJ, Leung PM, McKenzie S, et al. 1998. Random- Sheline GE (eds), Radiation Injury to the Nervous System. ized study of brachytherapy in the initial management of pa- New York: Raven Press, p 239. tients with malignant astrocytoma. Int J Radiat Oncol Biol Leibel SA, Sheline GE, Wara WM, Boldrey EB, Nielsen SL. 1975. Phys 41:1005–1011. The role of radiation therapy in the treatment of astrocy- Laramore GE, Diener-West M, Griffin TW, et al. 1988. Ran- tomas. Cancer 35:1551–1557. domized neutron dose searching study for malignant Leighton C, Fisher B, Bauman G, et al. 1997. Supratentorial gliomas of the brain: results of an RTOG study. Radiation low-grade glioma in adults: an analysis of prognostic fac- Therapy Oncology Group. Int J Radiat Oncol Biol Phys tors and timing of radiation. J Clin Oncol 15:1294–1301. 14:1093–1102. LeRoux PD, Berger MS, Ojemann GA, Wang K, Mack LA. 1989. Larson DA, Flickinger JC, Loeffler JS. 1993. The radiobiology Correlation of intraoperative ultrasound tumor volumes and of radiosurgery. Int J Radiat Oncol Biol Phys 25:557–561. margins with preoperative computed tomography scans. An Larson DA, Gutin PH, Leibel SA, Phillips TL, Sneed PK, Wara intraoperative method to enhance tumor resection. J Neu- WM. 1990. Stereotaxic irradiation of brain tumors. Cancer rosurg 71:691–698. 65:792–799. LeRoux PD, Winter TC, Berger MS, Mack LA, Wang K, Elliott Larson DA, Gutin PH, McDermott M, et al. 1996. Gamma knife P. 1994. A comparison between preoperative magnetic res- for glioma: selection factors and survival. Int J Radiat On- onance scans and intraoperative ultrasound tumor volumes col Biol Phys 36:1045–1053. and margins. J Clin Ultrasound 22:29–36. Larson DA, Wasserman TH, Dryzmala DE, Simpson JR. 1992. Levin VA. 1985. Chemotherapy of primary brain tumors. In: Stereotactic External-beam irradiation. In: Perez CA, Brady Vick NA, Bigner DD (eds), Neuro-Oncology, Neurologic LW (eds), Principles and Practice of Radiation Oncology, Clinics of North America, 3d ed. Philadelphia: WB Saun- 2d ed. Philadelphia: JB Lippincott, p 530. ders, p 855. Laws ER, Taylor WF, Bergstralh EJ, Okazaki H, Clifton MB. Levin VA, Prados MD, Yung WK, Gleason MJ, Ictech S, Ma- 1985. The neurosurgical management of low-grade astro- lec M. 1992. Treatment of recurrent gliomas with eflor- cytoma. Clin Neurosurg 33:575–588. nithine. J Natl Cancer Inst 84:1432–1437. Laws ER, Taylor WF, Clifton MB, Okazaki H. 1984. Neurosur- Levin VA, Chamberlain MC, Prados MD, et al. 1987. Phase I–II 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 142

142 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

study of eflornithine and mitoguazone combined in the ysis of clinicopathological features, radiotherapy, and sur- treatment of recurrent primary brain tumors. Cancer Treat vival in 170 cases of oligodendroglioma. J Neurosurg Rep 71:459–464. 67:224–230. Levin VA, Edwards MS, Wright DC, et al. 1980a. Modified pro- Loeffler JS, Alexander E 3d, Shea WM, et al. 1992. Radiosurgery carbazine, CCNU, and vincristine (PCV 3) combination che- as part of the initial management of patients with malignant motherapy in the treatment of malignant brain tumors. Can- gliomas. J Clin Oncol 10:1379–1385. cer Treat Rep 64:237–244. Loeffler JS, Alexander E 3d, Wen PY, et al. 1990. Results of Levin VA, Hoffman WF, Heilbron DC, Norman D. 1980b. Prog- stereotactic brachytherapy used in the initial management nostic significance of the pretreatment CT scan on time to of patients with glioblastoma. J Natl Cancer Inst 82:1918– progression for patients with malignant gliomas. J Neuro- 1921. surg 52:642–647. Longee DC, Friedman HS, Albright RE, et al. 1990. Treatment Levin VA, Lamborn K, Wara W, et al. 2000a. Phase II study of of patients with recurrent gliomas with cyclophosphamide 6-thioguanine, procarbazine, dibromodulcitol, lomustine, and vincristine. J Neurosurg 72:583–588. and vincristine chemotherapy with radiotherapy for treat- Louis DN, von Deimling A, Dickersin GR, Dooling EC, Seizinger ing malignant gliomas in children. Neuro-Oncology BR. 1992. Desmoplastic cerebral astrocytomas of infancy: 2:22–28. a histopathological, immunohistochemical, ultrastructural, Levin VA, Leibel SA, Gutin PH. 2001. Neoplasms of the central and molecular genetic study. Hum Pathol 23:1402–1409. nervous system. In: DeVita VT Jr, Hellman S, Rosenberg SA Ludwig CL, Smith MT, Godfrey AD, Armbrustmacher VW. 1986. (eds), Cancer: Principles and Practice of Oncology, 6th ed. A clinicopathological study of 323 patients with oligoden- Philadelphia: Lippincott-Raven, pp 2022–2082. drogliomas. Ann Neurol 19:15–21. Levin VA, Phuphanich S, Liu HC, et al. 1986a. Phase II study Lyons MK, Kelly PJ. 1992. Computer-assisted stereotactic bi- of combined carmustine, 5-fluorouracil, hydroxyurea, and opsy and volumetric resection of thalamic pilocytic astro- 6-mercaptopurine (BFHM) for the treatment of malignant cytomas. Report of 23 cases. Stereotact Funct Neurosurg gliomas. Cancer Treat Rep 70:1271–1274. 59:100–104. Levin VA, Prados MR, Wara WM, et al. 1995. Radiation ther- Macaulay RJ, Jay V, Hoffman HJ, Becker LE. 1993. Increased apy with bromodeoxyuridine chemotherapy followed by mitotic activity as a negative prognostic indicator in pleo- procarbazine, lomustine, and vincristine for the treatment morphic xanthoastrocytoma. Case report. J Neurosurg of anaplastic gliomas. Int J Radiat Oncol Biol Phys 79:761–768. 32:75–83. MacDonald DR. 1994. Low-grade gliomas, mixed gliomas, and Levin VA, Resser K, McGrath L, Vestnys P, Nutik S, Wilson CB. oligodendrogliomas. Semin Oncol 21:236–248. 1984. PCNU treatment for recurrent malignant gliomas. Maiuri F. 1988. Cysts with mural tumor nodules in the cere- Cancer Treat Rep 68: 969–973. bral hemispheres. Neurosurgery 22:703–706. Levin VA, Silver P, Hannigan J, et al. 1990. Superiority of post- Maiuri F, Stella L, Benvenuti D, Giamundo A, Pettinato G. 1990. radiotherapy adjuvant chemotherapy with CCNU, procar- Cerebral gliosarcomas: correlation of computed tomo- bazine, and vincristine (PCV) over BCNU for anaplastic graphic findings, surgical aspect, pathological features, and gliomas: NCOG 6G61 final report. Int J Radiat Oncol Biol prognosis. Neurosurgery 26:261–267. Phys 18:321–324. Mannoji H, Becker LE. 1988. Ependymal and choroid plexus Levin VA, Thall PF, Maor MH, et al. 1995. Phase II study of tumors. Cytokeratin and GFAP expression. Cancer accelerated fractionation radiation therapy with carboplatin 61:1377–1385. followed by PCV chemotherapy for the treatment of glioblas- Marks JE, Adler SJ. 1982. A comparative study of ependymo- toma multiforme. Int J Radiat Oncol Biol Phys 33:357–364. mas by site of origin. Int J Radiat Oncol Biol Phys 8:37–43. Levin VA, Uhm JH, Jaeckle KA, et al. 2000b. Phase III Marsa GW, Goffinet DR, Rubinstein LJ, Bagshaw MA. 1975. randomized study of post-radiotherapy chemotherapy Megavoltage irradiation in the treatment of gliomas of the with alpha-difluoromethylornithine-procarbazine, N-(2- brain and spinal cord. Cancer 36:1681–1689. chloroethyl)-N-cyclohexyl-N-nitrosurea, vincristine (DFMO- Mason WP, Grovas A, Halpern S, et al. 1998. Intensive che- PCV) versus PCV for glioblastoma multiforme. Clin Cancer motherapy and bone marrow rescue for young children Res 6:3878–3884. with newly diagnosed malignant brain tumors. J Clin Oncol Levin VA, Wara WM, Davis RL, et al. 1986b. Northern Califor- 16:210–221. nia Oncology Group protocol 6G91: response to treatment Massey V, Wallner KE. 1990. Patterns of second recurrence of with radiation therapy and seven-drug chemotherapy in pa- malignant astrocytomas. Int J Radiat Oncol Biol Phys tients with glioblastoma multiforme. Cancer Treat Rep 18:395–398. 70:739–744. Matson DD, Crofton FDL. 1960. Papilloma of the choroid Levy LF, Elvidge AR. 1956. Astrocytoma of the brain and spinal plexus in childhood. J Neurosurg 17:1002–1027. cord—a review of 176 cases, 1940–1949. J Neurosurg Matsushima T, Inoue T, Takeshita I, Fukui M, Iwaki T, Kita- 13:413–443. moto T. 1988. Choroid plexus papillomas: an immunohis- Lewis P. 1967. Carcinoma of the choroid plexus. Brain tochemical study with particular reference to the coex- 90:177–186. pression of prealbumin. Neurosurgery 23:384–389. Liang BC, Thornton AF, Sandler HM, Greenberger HS. 1991. McComb R, Burger PC. 1983. Choroid plexus carcinoma. Re- Malignant astrocytomas: focal tumor recurrence after focal port of a case with immunohistochemical and ultrastruc- external beam radiation therapy. J Neurosurg 75:559–563. tural observations. Cancer 51:470–475. Lindegaard KF, Mork SJ, Eide GE, et al. 1987. Statistical anal- McCormack BM, Miller DC, Budzilovich GN, Voorhees GJ, Ran- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 143

Primary Cerebral Tumors 143

sohoff J. 1992. Treatment and survival of low-grade astro- Morantz RA, Feigin R, Ransohoff J 3d. 1976. Clinical and patho- cytoma in adults—1977–1988. Neurosurgery 31:636–642. logical study of 24 cases of gliosarcoma. J Neurosurg McEvoy AW, Harding BN, Phipps KP, et al. 2000. Management 45:398–408. of choroid plexus tumours in children: 20 years experience Mork SJ, Halvorsen TB, Lindegaard KF, Eide GE. 1986. Oligo- at a single neurosurgical centre. Pediatr Neurosurg dendroglioma. Histologic evaluation and prognosis. J Neu- 32:192–199. ropathol Exp Neurol 45:65–78. McGirr SJ, Ebersold MJ, Scheithauer BW, Quast LM, Shaw EG. Mork SJ, Lindegaard KF, Halvorsen TB, et al. 1985. Oligoden- 1988. Choroid plexus papillomas: long-term follow-up re- droglioma: incidence and biological behavior in a defined sults in a surgically treated series. J Neurosurg 69:843–849. population. J Neurosurg 63:881–889. McLaughlin MP, Marcus RB, Buatti JM, et al. 1998. Ependy- Mork SJ, Loken AC. 1977. Ependymoma: a follow-up study of moma: results, prognostic factors and treatment recom- 101 cases. Cancer 40:907–915. mendations. Int J Radiat Oncol Biol Phys 40:845–850. Mornex F, Nayel H, Taillandier L. 1993. Radiation therapy for McLaurin RL, Towbin RB. 1986. Tuberous sclerosis: diagnos- malignant astrocytomas in adults. Radiother Oncol tic and surgical considerations. Pediatr Neurosci 12:43–48. 27:181–192. Medbery CA 3d, Straus KL, Steinberg SM, Cotelingam JD, Fisher Morota N, Sakamoto K, Kobayashi N, Hashimoto K. 1990. Re- WS. 1988. Low-grade astrocytomas: treatment results and current low-grade glioma in children with special reference prognostic variables. Int J Radiat Oncol Biol Phys to computed tomography findings and pathological 15:837–841. changes. Childs Nerv Syst 6:155–160. Mehta MP, Masciopinto J, et al. 1994. Stereotactic radiosurgery MRC Working Group Party. 2001. Randomized trial of pro- for glioblastoma multiforme: report of a prospective study carbazine, lomustine, and vincristine in the adjuvant treat- evaluating prognostic factors and analyzing long-term sur- ment of high-grade astrocytoma: a Medical Research Coun- vival advantage. Int J Radiat Oncol Biol Phys 30:541–549. cil trial. J Clin Oncol 19:509–518. Meis JM, Martz KL, Nelson JS. 1991. Mixed glioblastoma mul- Muller W, Afra D, Schroder R. 1977. Supratentorial recur- tiforme and sarcoma. A clinicopathologic study of 26 Ra- rences of gliomas. Morphological studies in relation to time diation Therapy Oncology Group cases. Cancer 67:2342– intervals with astrocytomas. Acta Neurochir (Wien) 2349. 37:75–91. Merchant TE, Haida T, Wang MH, Finlay JL, Leibel SA. 1997. Naguib MG, Chou SN, Mastri A. 1981. Radiation therapy of a Anaplastic ependymoma: treatment of pediatric patients choroid plexus papilloma of the cerebellopontine angle with or without craniospinal radiation therapy. J Neurosurg wtih bone involvenment. J Neurosurg 52:245–247. 86:943–949. Nakamura Y, Becker LE. 1983. Subependymal giant cell tu- Mercuri S, Russo A, Palma L. 1981. Hemispheric supratento- mor: astrocytic or neuronal? Acta Neuropathol (Berl) rial astrocytomas in children. Long-term results in 29 cases. 60:271–277. J Neurosurg 55:170–173. Nelson DF, Diener-West M, Horton J, Chang CH, Schoenfeld Miettinen M, Clark R, Virtanen I. 1986. Intermediate filament D, Nelson JS. 1988. Combined modality approach to treat- proteins in choroid plexus and ependyma and their tumors. ment of malignant gliomas—re-evaluation of RTOG Am J Pathol 123:231–240. 7401/ECOG 1374 with long-term follow-up. A joint study of Miller DC, Lang FF, Epstein FJ. 1993. Central nervous system the Radiation Therapy Oncology Group and the Eastern Co- gangliogliomas. Part 1: pathology. J Neurosurg 79:859– operative Oncology Group. NCI Monogr 6:279–284. 866. Nelson DF, Urtasun RC, Saunders WM, Gutin PH, Sheline GE. Miller LL, Ostrow PT, Chau R. 1991. Characterization of 1986. Recent and current investigations of radiation ther- gliosarcomas by image analysis of superimposed serial sec- apy of malignant gliomas. Semin Oncol 13:46–55. tions. J Neuropathol Exp Neurol 50:365A. Nelson JS, Tsukada Y, Schoenfeld D, Fulling K, Lamarche J, Miser J, Krailo M, Smithson W Belasco J, Ortega J, Hammond Peress N. 1983. Necrosis as a prognostic criterion in ma- GD. 1989. Treatment of children with recurrent brain tu- lignant supratentorial, astrocytic gliomas. Cancer 52:550– mors with ifosphamide, etoposide and mesna. Results of a 554. phase II Childrens Cancer Study Group trial. Proc Am Soc Nishio S, Takeshita I, Fujii K, Fukui M. 1989. Supratentorial Clin Oncol 8:A328. astrocytic tumours of childhood: a clinicopathological study Mitchell JB, Kinsella TJ, Russo A, et al. 1983. Radiosensitiza- of 41 cases. Acta Neurochir (Wien) 101:3–8. tion of hematopoietic precursor cells (CFUc) in glioblas- Nishio S, Takeshita I, Fujiwara S, Fukui M. 1993. Optico-hy- toma patients receiving intermittent intravenous infusions pothalamic glioma: an analysis of 16 cases. Childs Nerv Syst of bromodeoxyuridine (BUdR). Int J Radiat Oncol Biol Phys 9:334–338. 9:457–463. North CA, North RB, Epstein JA, Piantadosi S, Wharam MD. Montine TJ, Vandersteenhoven JJ, Aguzzi A, et al. 1994. Prog- 1990. Low-grade cerebral astrocytomas. Survival and qual- nostic significance of Ki-67 proliferation index in supra- ity of life after radiation therapy. Cancer 66:6–14. tentorial fibrillary astrocytic neoplasms. Neurosurgery Obana WG, Cogen PH, Davis RL, Edwards MS. 1991. Metasta- 34:674–678. tic juvenile pilocytic astrocytoma. Case report. J Neurosurg Moore GE. 1947. Fluorescein as an agent in the differentia- 75:972–975. tion of normal and malignant tissues. Science 106:130– Ojemann G, Ojemann J, Lettich E, Berger M. 1989. Cortical 131. language localization in left, dominant hemisphere. An elec- Morantz RA. 1987. Radiation therapy in the treatment of cere- trical stimulation mapping investigation in 117 patients. bral astrocytoma. Neurosurgery 20:975–982. J Neurosurg 17:316–326. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 144

144 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Olson JD, Riedel E, DeAngelis LM. 2000. Long-term outcome erations in tumors affecting the central nervous system: of low-grade oligodendroglioma and mixed glioma. Neu- ependymomas. Radiology 83:98–105. rology 54:1442–1448. Piepmeier JM. 1987. Observations on the current treatment of Otsubo H, Hoffman HJ, Humphries RP, et al. 1992. Detection low-grade astrocytic tumors of the cerebral hemisheres. and management of gangliogliomas in children. Surg Neu- J Neurosurg 67:177–181. rol 38:371–378. Pierre-Kahn A, Hirsch JF, Roux FX, Renier D, Sainte-Rose C. Ozek MM, Sav A, Pamir MN, Ozer AF, Ozek E, Erzen C. 1993. 1983. Intracranial ependymomas in childhood. Survival and Pleomorphic xanthoastrocytoma associated with von Reck- functional results of 47 cases. Childs Brain 10:145–156. linghausen neurofibromatosis. Childs Nerv Syst 9:39–42. Pilcher WH, Silbergeld DL, Berger MS, Ojemann GA. 1993. In- Packer RJ, Lange B, Ater J, et al. 1993. Carboplatin and vin- traoperative electrocorticography during tumor resection: cristine for progressive, recurrent and newly diagnosed impact on seizure outcome in patients with gangliogliomas. low-grade gliomas of childhood. J Clin Oncol 11:850–856. J Neurosurg 78:891–902. Packer RJ, Perilongo G, Johnson D, et al. 1992. Choroid plexus Pinto-Lord MC, Abroms IF, Smith TW. 1986. Hyperdense cere- carcinoma of childhood. Cancer 69:580–585. bral lesion in childhood tuberous sclerosis: computed to- Packer RJ, Sutton L, Bilaniuk L, et al. 1988. Treatment of chi- mographic demonstration and neuropathologic analysis. asmatic/hypothalamic gliomas of childhood with chemo- Pediatr Neurol 2:245–248. therapy: an update. Ann Neurol 23:79–85. Pollack IF, Gerszten PC, Martinez AJ, et al. 1995. Intracranial Palma L, Celli P, Cantore G. 1993. Supratentorial ependymo- ependymomas of childhood: long-term outcome and prog- mas of the first two decades of life. Long-term follow-up of nostic factors. Neurosurgery 37:655–667. 20 cases (including two subependymomas). Neurosurgery Pons M, Finlay JL, Walker RW, Puccetti D, Packer RJ, McEl- 32:169–175. wain M. 1992. Chemotherapy with vincristine (VCR) and Palma L, Guidetti B. 1985. Cystic pilocytic astrocytomas of the etoposide (VP-16) in children with low-grade astrocytoma. cerebral hemispheres. Surgical experience with 51 cases J Neurooncol 14:151–158. and long-term results. J Neurosurg 62:811–815. Poon WS, Schomacker KT, Deutsch TF, Martuza RL. 1992. Palma L, Russo A, Mercuri S. 1983. Cystic cerebral astrocy- Laser-induced fluorescence: experimental intraoperative tomas in infancy and childhood: long-term results. Childs delineation of tumor resection margins. J Neurosurg 76: Brain 10:79–91. 679–686. Papadopoulos DP, Giri S, Evans RG. 1990. Prognostic factors Prados MD, Gutin PH, Phillips TL, et al. 1992a. Highly anaplas- and management of intracranial ependymomas. Anticancer tic astrocytoma: a review of 357 patients treated between Res 10:689–692. 1977 and 1989. Int J Radiat Oncol Biol Phys 23:3–8. Parker JC, Mortara, RH, McCloskey JJ. 1975. Biological be- Prados MD, Gutin PH, Phillips TL, et al. 1992b. Interstitial havior of the primitive neuroectodermal tumors: significant brachytherapy for newly diagnosed patients with malignant supratentorial childhood gliomas. Surg Neurol 4:383–388. gliomas: the UCSF experience. Int J Radiat Oncol Biol Phys Paulino AC, Melian E. 1999. Medulloblastoma and supraten- 24:593–597. torial primitive neuroectodermal tumors: an institutional Prados MD, Larson DA, Lamborn K, et al. 1998a. Radiation experience. Cancer 86:142–148. therapy and hydroxyurea followed by the combination of Paulus W, Janisch W. 1990. Clinicopathologic correlations in 6-thioguanine and BCNU for the treatment of primary ma- epithelial choroid plexus neoplasms: a study of 52 cases. lignant brain tumors. Int J Radiat Oncol Biol Phys 40:57– Acta Neuropathol (Berl) 80:635–641. 63. Penfield W, Roberts L. 1959. Speech and Brain Mechanisms. Prados M, Rodriguez L, Chamberlain M, Silver P, Levin V. Princeton, NJ: Princeton University Press, 286 pp. 1989. Treatment of recurrent gliomas with 1,3-bis(2- Perilongo G, Massimino M, Sotti G, et al. 1997. Analyses of chloroethyl)-1-nitrosourea and alpha-difluoromethylor- prognostic factors in a retrospective review of 92 children nithine. Neurosurgery 24:806–809. with ependymoma: Italian Pediatric Neuro-oncology Group. Prados MD, Scott C, Curran WJ, Nelson DF, Leibel S, Kramer Med Pediatr Oncol 29:79–85. S. 1999a. Procarbazine, lomustine, and vincristine (PCV) Perria C, Carai M, Falzoi A, et al. 1988. Photodynamic therapy chemotherapy for anaplastic astrocytoma: a retrospective of malignant brain tumors: clinical results of, difficulties review of Radiation Therapy Oncology Group protocols with, questions about, and future prospects for the neuro- comparing survival with carmustine or PCV adjuvant che- surgical applications. Neurosurgery 23:557–563. motherapy. J Clin Oncol 17:3389–3395. Petronio J, Edwards MS, Prados M, et al. 1991. Management Prados MD, Scott CB, Rotman M, et al. 1998b. Influence of of chiasmal and hypothalamic gliomas of infancy and child- bromodeoxyuridine radiosensitization on malignant glioma hood with chemotherapy. J Neurosurg 74:701–708. patient survival: a retrospective comparison of survival data Philippon JH, Clemenceau SH, Fauchon FH, Foncin JF. 1993. from the Northern California Oncology Group (NCOG) and Supratentorial low-grade astrocytomas in adults. Neuro- Radiation Therapy Oncology Group trials (RTOG) for surgery 32:554–559. glioblastoma multiforme and anaplastic astrocytoma. Int J Phillips TL, Levin VA, Ahn DK, et al. 1991. Evaluation of bro- Radiat Oncol Biol Phys 40:653–659. modeoxyuridine in glioblastoma multiforme: a Northern Prados MD, Scott C, Sandler H, Buckner JC, et al. 1999b. A Cancer Center Group Phase II study. Int J Radiat Oncol Biol phase 3 randomized study of radiotherapy plus procar- Phys 21:709–714. bazine, CCNU, and vincristine (PCV) with or without BUdR Phillips TL, Sheline GE, Boldrey E. 1964. Therapeutic consid- for the treatment of anaplastic astrocytoma: a preliminary 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 145

Primary Cerebral Tumors 145

report of RTOG 9404. Int J Radiat Oncol Biol Phys relation of malignant ependymomas with postoperative sur- 45:1109–1115. vival. J Neurosurg 70:31–36. Prados MD, Wara WM, Sneed PK, et al. 2001. Phase III trial Ross IB, Robitaille Y, Villemure JG, Tampieri D. 1991. Diag- of accelerated hyperfractionation with or without di- nosis and management of gliomatosis cerebri: recent fluromethylornithine (DFMO) versus standard fractionated trends. Surg Neurol 36:431–440. radiotherapy with or without DFMO for newly diagnosed Rousseau P, Habrand JL, Sarrazin D, et al. 1993. Treatment of patients with glioblastoma multiforme. Int J Radiat Oncol intracranial ependymomas of children: review of a 15-year Biol Phys 49:71–77. experience. Int J Radiat Oncol Biol Phys 28:381–386. Puccetti D, Finlay JL. 1992. Current treatment of intracranial Rubinstein LJ. 1972. Tumors of the Central Nervous System. ependymomas in children. Crit Rev Neurosurg 2:93–107. Atlas of Tumor Pathology, Series 2, Fascicle 6. Washington, Raimondi AJ, Gutierrez FA. 1975. Diagnosis and surgical treat- DC: Armed Forces Institute of Pathology, p 1. ment of choroid plexus papillomas. Childs Brain 1:81–115. Rubinstein LJ. 1985. Embryonal central neuroepithelial tumors Rajan B, Pickuth D, Ashley S, et al. 1993. The management of and their differentiating potential. A cytogenetic view of a histologically unverified presumed cerebral gliomas with complex neuro-oncological problem. J Neurosurg 62:795– radiotherapy. Int J Radiat Oncol Biol Phys 28:405–413. 805. Rasmussen TB. 1975. Surgery of epilepsy associated with brain Salazar OM, Castro-Vita H, VanHoutte P, Rubin P, Aygun C. tumors. Adv Neurol 8:227–239. 1983. Improved survival in cases of intracranial ependy- Rawlings CE III, Giangaspero F, Burger PC, et al. 1988. Ependy- moma after radiation therapy. Late report and recommen- momas: a clinicopathologic study. Surg Neurol 29:271–281. dations. J Neurosurg 59:652–659. Read G. 1984. The treatment of ependymoma of the brain and Salazar OM, Rubin P. 1976. The spread of glioblastoma mul- spinal canal by radiotherapy: a report of 79 cases. Clin Ra- tiforme as a determining factor in the radiation treatment diol 35:163–166. volume. Int J Radiat Oncol Biol Phys 1:627–637. Recht LD, Lew R, Smith TW. 1992. Suspected low-grade Salazar OM, Rubin P, McDonald JV, Feldstein ML. 1976. Pat- glioma: is deferring treatment safe? Ann Neurol 31:431– terns of failure in intracranial astrocytomas after irradia- 436. tion: Analysis of dose and field factors. Am J Roentgenol Reedy DP, Bay JW, Hahn JF. 1983. Role of radiation therapy 126:279–292. in the treatment of cerebral oligodendroglioma: an analy- Salcman M, Kaplan RS, Durken TB, Abdo H, Montgomery E. sis of 57 cases and a literature review. Neurosurgery 1982. Effect of age and reoperation on survival in the com- 13:499–503. bined modality treatment of malignant astrocytoma. Neu- Rekate HL, Erwood S, Brodkey JA, et al. 1985. Etiology of ven- rosurgery 10:454–463. triculomegaly in choroid plexus papilloma. Pediatr Neu- Sartorius CJ, Berger MS. 1988. Rapid termination of intraop- rosci 12:196–201. erative stimulation-evoked seizures with application of cold Reyes MG, Homsi MF, McDonald LW, Glick RP. 1991. Imprints, Ringer’s lactate to the cortex. Technical note. J Neurosurg smears and frozen sections of brain tumors. Neurosurgery 88:349–351. 29:575–579. Scharfen CO, Sneed PK, Wara WM, et al. 1992. High activity Ribaric I, Nagulic M, Djurovic B. 1991. Surgical treatment of iodine-125 interstitial implant for gliomas. Int J Radiat On- epilepsy: our experiences with 34 children. Childs Nerv Syst col Biol Phys 24:583–591. 7:402–404. Schiffer D, Chio A, Cravioto H, et al. 1991. Ependymoma: In- Ringertz N. 1950. “Grading” of gliomas. Acta Pathol Microbiol ternal correlations among pathologic signs: the anaplastic Scand 27:51–64. variant. Neurosurgery 29:206–210. Ringertz N, Reymond A. 1949. Ependymomas and choroid Schiffer D, Dutto A, Cavalla P, et al. 1997. Bosone I, Chio A, plexus papillomas. J Neuropathol Exp Neurol 8:355–380. Villani R, Bellotti C (eds), Prognostic Factors in Oligoden- Ritter AM, Sawaya R, Hess KR, Levin VA, Bruner JM. 1994. droglioma. Can J Neurol Sci 24:313–319. Prognostic significance of bromodeoxyuridine labeling in Schijman E, Monges J, Raimondi AJ, Tomita T. 1990. Choroid primary and recurrent glioblastoma multiforme. Neuro- plexus papillomas of the III ventricle in childhood. The di- surgery 35:192–198. agnosis and surgical management. Childs Nerv Syst Roberts DW, Strohbehn JW, Hatch JF, Murray W, Kettenberger 6:331–334. H. 1986. A frameless stereotaxic integration of computer- Schild SE, Nisi K, Scheithauer BW, et al. 1998. The results of ized tomographic imaging and the operating microscope. radiotherapy for ependymomas: the Mayo Clinic experi- J Neurosurg 65:545–549. ence. Int J Radiat Oncol Biol Phys 42:953–958. Rodriguez LA, Edwards MS, Levin VA. 1990. Management of Schisano G, Tovi D, Nordenstam H. 1963. Spongioblastoma hypothalamic gliomas in children: an analysis of 33 cases. polare of the cerebral hemisphere. J Neurosurg 20:241– Neurosurgery 26:242–247. 251. Rodriguez LA, Prados M, Silver P, Levin VA. 1989. Re-evalua- Schold SC, Friedman HS, Bjornsson TD, Falletta JM. 1984. tion of procarbazine for the treatment of recurrent malig- Treatment of patients with recurrent primary brain tumors nant CNS tumors. Cancer, 64:2420–2423. with AZQ. Neurology 34:615–619. Rosenblum MK, Erlandson RA, Budzilovich GN. 1991. The Schold SC, Mahaley MS, Vick NA, et al. 1987. Phase II diaz- lipid-rich epitheliod glioblastoma. Am J Surg Pathol iquone-based chemotherapy trials in patients with anaplas- 15:925–934. tic supratentorial astrocytic neoplasms. J Clin Oncol 5:464– Ross GW, Rubinstein LJ. 1989. Lack of histopathological cor- 471. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 146

146 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Schuller P, Schafer U, Micke O, Willich N. 1999. Radiother- Shimizu KT, Tran LM, Mark RJ, Selch MT. 1993. Management apy for intracranial and spinal ependymomas. A retrospec- of oligodendrogliomas. Radiology 186:569–572. tive analysis. Strahlenther Onkol 175:105–111. Shrieve DC, Alexander E 3d, Black PM, et al. 1999. Treatment Scott CB, Curran W, Yung WK, et al. 1998. Long term results of patients with primary glioblastoma multiforme with stan- of RTOG 9006: randomized trial of hyperfractionated ra- dard postoperative radiotherapy and radiosurgical boost: diotherapy (RT) to 72 Gy & carmustine vs standard RT & prognostic factors and long-term outcome. J Neurosurg carmustine for malignant glioma patients with emphasis on 90:72–77. anaplastic astrocytoma patients. Proc Am Soc Clin Oncol Shrieve DC, Alexander E, Wen PY, et al. 1995. Comparison of 17:40A. stereotactic radiosurgery and brachytherapy in the treat- Scott EW, Mickle JP. 1987. Pediatric diencephalic gliomas— ment of recurrent glioblastoma multiforme. Neurosurgery a review of 18 cases. Pediatr Neurosci 13:225–232. 36:275–282. Sexauer CL, Khan A, Burger PC, et al. 1985. Cisplatin in re- Silver JM, Rawlings CE 3d, Rossitch E Zeidman SM, Friedman current pediatric brain tumors. A POG phase II study. A Pe- AH. 1991. Ganglioglioma: a clinical study with long-term diatric Oncology Group Study. Cancer 56:1497–1501. follow-up. Surg Neurol 35:261–266. Shapiro WR. 1986. Therapy of adult malignant brain tumors: Simpson WJ, Platts ME. 1976. Fractionation study in the treat- what have the clinical trials taught us? Semin Oncol ment of glioblastoma multiforme. Int J Radiat Oncol Biol 13:38–45. Phys 1:639–644. Shaw E, Arusell R, Scheithauer B, et al. 1998. A prospective Smith MT, Ludwig CL, Godfrey AD, Armbrustmacher VW. 1983. randomized trial of low- versus high-dose radiation ther- Grading of oligodendrogliomas. Cancer 52:2107–2114. apy in adults with supratentorial low-grade glioma: initial Smyth MD, Horn BN, Russo C, Berger MS. 2000. Intracranial report of a NCCG-RTOG-ECOG study. Proc Am Soc Clin On- ependymomas of childhood: current management strate- col 17:401A. gies. Pediatr Neurosurg 33:138–150. Shaw EG, Daumas-Duport C, Scheithauer BW, et al. 1989. Ra- Sneed PK, Stauffer PR, McDermott MW, et al. 1998. Survival diation therapy in the management of low-grade supraten- benefit of hyperthermia in a prospective randomized trial torial astrocytomas. J Neurosurg 70:853–861. of brachytherapy boost / hyperthermia for glioblas- Shaw EG, Evans RG, Scheithauer BW, Istrup DM, Earle JD. toma multiforme. Int J Radiat Oncol Biol Phys 40:287–295. 1987. Postoperative radiotherapy of intracranial ependy- Soffietti R, Chio A, Giordana MT, Vasario E, Schiffer D. 1989. moma in pediatric and adult patients. Int J Radiat Oncol Prognostic factors in well-differentiated cerebral astrocy- Biol Phys 13:1457–1462. tomas in the adult. Neurosurgery 24:686–692. Shaw EG, Scheithauer BW, O’Fallon JR. 1991. Management of Soffietti R, Ruda R, Bradac GB, Schiffer D. 1998. PCV chemo- supratentorial low-grade gliomas. Oncology (Huntingt) therapy for recurrent oligodendrogliomas and oligoastro- 7:94–111. cytomas. Neurosurgery 43:1066–1073. Shaw EG, Scheithauer BW, O’Fallon JR. 1997. Supratentorial Souweidane MM, Johnson JH, Lis E. 1999. Volumetric reduc- gliomas: a comparative study by grade and histologic type. tion of a choroid plexus carcinoma using preoperative che- J Neurooncol 31:273–278. motherapy. J Neurooncol 43:167–171. Shaw EG, Scheithauer BW, O’Fallon JR, Davis DH. 1994. Mixed Spagnoli MV, Grossman RI, Packer RJ, et al. 1987. Magnetic oligoastrocytomas: a survival and prognostic factor analy- resonance imaging determination of gliomatosis cerebri. sis. Neurosurgery 34:577–582. Neuroradiology 29:15–18. Shaw EG, Scheithauer BW, O’Fallon JR Tazelaar HD, Davis DH. Spallone A, Pastore FS, Guiffre R, Giudetti B. 1990. Choroid 1992. Oligodendrogliomas. The Mayo Clinic experience. plexus papillomas in infancy and childhood. Childs Nerv J Neurosurg 76:428–434. Syst 6:71–74. Sheline GE. 1983. Radiotherapy of adult primary cerebral neo- Spencer DD, Spencer SS, Mattson RH, Williamson PD. 1984. plasms. In: Walker MD (ed), Oncology of the Nervous Sys- Intracerebral masses in patients with intractable partial tem. Boston: M Nijhoff, p 223. epilepsy. Neurology 34:432–436. Shenouda G, Souhami J, Freeman C, Hazel J, Lehnert S, Joseph Sposto R, Ertel IJ, Jenkin RD, et al. 1989. The effectiveness of L. 1991. Accelerated fractionation for high-grade cerebral chemotherapy for treatment of high-grade astrocytoma in astrocytomas. Preliminary treatment results. Cancer children; results of a randomized trial. A report from the 67:2247–2252. Childrens Cancer Study Group. J Neurooncol 7:165–177. Shepherd CW, Scheithauer BW, Gomez MR, Altermatt HJ, Katz- St Clair SK, Humphreys RP, Pillay PK, Hoffman HJ, Blaser SI, mann JA. 1991. Subependymal giant cell astrocytoma: a Becker LE. 1991. Current management of choroid plexus clincial, pathological, and flow cytometric study. Neuro- carcinoma in children. Pediatr Neurosurg 17:225–233. surgery 28:864–868. Steinmeier R, Fahlbusch O, Ganslandt O, et al. 1998. Intraop- Shibamoto Y, Kitakabu Y, Takahashi M, et al. 1993. Supra- erative magnetic resonance imaging with the magnetom tentorial low-grade astrocytoma. Correlation of computed open scanner: concepts, neurosurgical indications, and tomography findings with effect of radiotherapy and prog- procedures: a preliminary report. Neurosurgery 43:739– nostic variables. Cancer 72:190–195. 748. Shibuya M, Ito S, Miwa T, Davis RL, Wilson CB, Hoshino T. Stenning SP, Freedman LS, Bleehen NM. 1987. An overview of 1993. Proliferative potential of brain tumors. Analyses with published results from randomized studies of nitrosoureas Ki-67 and anti-DNA polymerase alpha monoclonal anti- in primary high grade malignant glioma [letter to editor]. bodies, bromodeoxyuridine labeling, and nuclear organizer Br J Cancer 56:89–90. region counts. Cancer 71:199–206. Streffer J, Schabet M, Bamberg M, et al. 2000. A role for preir- 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 147

Primary Cerebral Tumors 147

radiation PCV chemotherapy for oligodendroglial brain tu- irradiation in localized intracranial ependymoma. Int J Ra- mors. J Neurol 247:297–302. diat Oncol Biol Phys 21:825–830. Suit H, Urie M. 1992. Proton beams in radiation therapy. J Natl van Veelen ML, Avezaat CJ, Kros JM, van Putten W, Vecht C. Cancer Inst 84:155–164. 1998. Supratentorial low grade astrocytoma: prognostic Sun ZM, Genka S, Shitara N, Akanuma A, Takakura K. 1988. factors, dedifferentiation, and the isuue of early versus late Factors possibly influencing the prognosis of oligoden- surgery. J Neurol Neurosurg Psychiatry 64:581–587. droglioma. Neurosurgery 22:886–891. Varma RR, Crumrine PK, Bergman I, et al. 1983. Childhood Sutton LN. 1987. Current management of low-grade astrocy- oligdendrogliomas presenting with seizures and low- tomas of childhood. Pediatr Neurosci 13:98–107. density lesions on computed tomography. Neurology Svien HJ, Mabon RF, Kernohan JW, Adson AW. 1949. Astrocy- 33:806–808. tomas. Proc Staff Meet Mayo Clinic 24:54–63. Wald SL, Fogelson H, McLaurin RL. 1982. Cystic thalamic Taratuto AL, Monges J, Lylyk P, Leiguarda R. 1984. Superficial gliomas. Childs Brain 9:381–393. cerebral astrocytoma attached to dura. Report of six cases Walker MD, Alexander E, Hunt WE, et al. 1978. Evaluation of in infants. Cancer 54:2505–2512. BCNU and/or radiotherapy in the treatment of anaplastic Taratuto AL, Sevlever G, Piccardo P. 1991. Clues and pitfalls gliomas. A cooperative clinical trial. J Neurosurg in stereotactic biopsy of the central nervous system. Arch 49:333–343. Pathol Lab Med 115:596–602. Walker MD, Green SB, Byar DP, et al. 1980. Randomized com- Thornton AF, Hegarty TJ, Ten Haken RK, et al. 1991. Three- parisons of radiotherapy and nitrosoureas for the treatment dimensional treatment planning of astrocytomas: a dosi- of malignant glioma after surgery. N Engl J Med 303:1323– metric study of cerebral irradiation. Int J Radiat Oncol Biol 1329. Phys 20:1309–1315. Wallner KE. 1991. Radiation treatment planning for malignant Thornton AF, Sandler HM, Ten Haken RK, et al. 1992. The astrocytomas. Semin Radiat Oncol 1:17–22. clinical utility of magnetic resonance imaging in 3-dimen- Wallner KE, Galicich JH, Krol G, Arbit E, Malkin MG. 1989. sional treatment planning of brain neoplasms. Int J Radiat Patterns of failure following treatment for glioblastoma mul- Oncol Biol Phys 24:767–775. tiforme and anaplastic astrocytoma. Int J Radiat Oncol Biol Tomita T, McLone DG, Naidich TP. 1986. Mural tumors with Phys 16:1405–1409. cysts in the cerebral hemispheres of children. Neurosurgery Wallner KE, Gonzales MF, Edwards MS, Wara WM, Sheline GE. 19:998–1005. 1988a. Treatment results of juvenile pilocytic astrocytoma. Tomita T, McLone DG, Yasue M. 1988. Cerebral primitive neu- J Neurosurg 69:171–176. roectodermal tumors in childhood. J Neurooncol 6:233– Wallner KE, Gonzales M, Sheline GE. 1988b. Treatment of 243. oligodendrogliomas with or without postoperative irradia- Tomita T, Naidich TP. 1987. Successful resection of choroid tion. J Neurosurg 68:684–688. plexus papillomas diagnosed at birth: Report of two cases. Wallner KE, Wara WM, Sheline GE, Davis RL. 1986. Intracra- Neurosurgery 20:774–779. nial ependymomas: results of treatment with partial or Tronnier VM, Wirtz CR, Knauth M, et al. 1997. Intraoperative whole brain irradiation without spinal irradiation. Int J Ra- diagnostic and interventional magnetic resonance imaging diat Oncol Biol Phys 12:1937–1941. in neurosurgery. Neurosurgery 40:891–902. Wara WM. 1985. Radiation therapy for brain tumors. Cancer Tsutsumi Y, Andoh Y, Sakaguchi J. 1989. A new ultrasound- 55:2291–2295. guided brain biopsy technique through a burr hole. Tech- Watanabe E, Watanabe T, Manaka S, Mayanagi Y, Takakura K. nical note. Acta Neurochir (Wien) 96:72–75. 1987. Three-dimensional digitizer (neuronavigator): new Undjian S, Marinov M. 1990. Intracranial ependymomas in equipment for computed tomography-guided stereotaxic children. Childs Nerv Syst 6:131–134. surgery. Surg Neurol 27:543–547. van den Bent MJ, Kros JM, Heimans JJ, et al. 1998. Response Weingart J, Olivi A, Brem H. 1991. Supratentorial low-grade rate and prognostic factors of recurrent oligodendroglioma astrocytomas in adults. Neurosurg Q 1:141–159. treated with procarbazine, CCNU, and vincristine chemo- Weldon-Linne CM, Victor TA, Groothius DR, Vick NA. 1983. therapy. Dutch Neuro-oncology Group. Neurology Pleomorphic xanthoastrocytoma. Ultrastructural and im- 51:1140–1145. munohistochemical study of a case with a rapidly fatal out- VandenBerg SR. 1993. Desmoplastic infantile ganglioglioma come following surgery. Cancer 52:2055–2063. and desmoplastic cerebral astrocytoma of infancy. Brain Wen DY, Hall WA, Miller DA, Seljeskog EL, Maxwell RE. 1993. Pathol 3:275–281. Targeted brain biopsy: a comparison of freehand computed VandenBerg SR, May EE, Rubinstein LJ, et al. 1987. Desmo- tomography-guided and stereotactic techniques. Neuro- plastic supratentorial neuroepithelial tumors of infancy with surgery 32:407–413. divergent differentiation potential (“desmoplastic infantile Werner-Wasik M, Scott CB, Nelson DF, et al. 1996. Final re- gangliogliomas”). Report on 11 cases of a distinctive em- port of a phase I/II trial of hyperfractionated and acceler- bryonal tumor with favorable prognosis. J Neurosurg ated hyperfractionated radiation therapy and carmustine for 66:58–71. adults with supratentorial malignant gliomas. Radiation Vanuytsel LJ, Bessell EM, Ashley SE, Bloom HJ, Brada M. 1992. Therapy Oncology Group Study 83-02. Cancer 77:1535– Intracranial ependymoma: long-term results of a policy of 1543. surgery and radiotherapy. Int J Radiat Oncol Biol Phys Whitton AC, Bloom HJ. 1990. Low grade glioma of the cere- 23:313–319. bral hemispheres in adults: a retrospective analysis of 88 Vanuytsel L, Brada M. 1991. The role of prophylactic spinal cases. Int J Radiat Oncol Biol Phys 18:783–786. 3601_e03(part2)_p73-148 2/15/02 4:02 PM Page 148

148 PRIMARY CENTRAL NERVOUS SYSTEM TUMORS

Wilson CB. 1975. Reoperation for primary tumors. Semin On- Young B, Oldfield EH, Markesbery WR, et al. 1981. Reopera- col 2:19–20. tion for glioblastoma. J Neurosurg 55:917–921. Wilson CB, Gutin PH, Boldrey EB, Drafts D, Levin VA, Enot KJ. Yule SM, Hide TA, Cranney M, Simpson E, Barrett A. 2001. 1976. Single-agent chemotherapy of brain tumors. A five- Low grade astrocytomas in the West of Scotland 1987–96: year review. Arch Neurol 33:739–744. treatment, outcome and cognitive functioning. Arch Dis Wilson CB, Larson DA, Gutin PH. 1992. Radiosurgery: a new Child 84:61–64. application? J Clin Oncol 10:1373–1374. Yung WK, Albright RE, Olson J, et al. 2000. A phase II study Winger MJ, Macdonald DR, Cairncross JG. 1989. Supratento- of temozolomide vs. procarbazine in patients with glioblas- rial anaplastic gliomas in adults. The prognostic importance toma multiforme at first relapse. Br J Cancer 83:588–593. of extent of resection and prior low-grade glioma. J Neu- Yung WK, Harris MI, Bruner JM, Feun LG. 1989. Intravenous rosurg 71:487–493. BCNU and AZQ in patients with recurrent malignant gliomas. Winger MJ, Macdonald DR, Schold SC, Cairncross JG. 1989. J Neurooncol 7:237–240. Selection bias in clinical trials of anaplastic glioma. Ann Yung WK, Mechtler L, Gleason MJ. 1991a. Intravenous carbo- Neurol 26:531–534. platin for recurrent malignant gliomas: a phase II study. Wisoff JH, Abbott R, Epstein F. 1990. Surgical management of J Clin Oncol 9:860–864. exophytic chiasmatic-hypothalamic tumors of childhood. Yung WK, Prados MD, Levin VA, et al. 1991b. Intravenous re- J Neurosurg 73:661–667. combinant interferon-beta in patients with recurrent ma- Wisoff JH, Boyett JM, Berger MS, et al. 1998. Current neuro- lignant gliomas: a phase I/II study. J Clin Oncol 9:1945– surgical management and the impact of the extent of re- 1949. section in the treatment of malignant gliomas of childhood: Yung WK, Prados MD, Yaya-Tur R, et al. 1999. Multicenter a report of the Children’s Cancer Group trial no. CCG-945. phase II trial of temozolomide in patients with anaplastic J Neurosurg 89:52–59. astrocytoma or anaplastic oligoastrocytoma at first re- Withers HR. 1985. Biologic basis for altered fractionation lapse. Temodal Brain Tumor Group. J Clin Oncol 17: schemes. Cancer 55:2086–2095. 2762–2771. Wong ET, Hess KR, Gleason MJ, et al. 1999. Outcomes and Yung WK, Simaga M, Levin VA. 1993. 13-cis-retinoic acid: a prognostic factors in recurrent glioma patients enrolled new and potentially effective agent for recurrent malig- onto phase II clinical trials. J Clin Oncol 17:2572–2578. nant astrocytomas. Proc Ann Meet Am Soc Clin Oncol 12: Wood JR, Green SB, Shapiro WR. 1988. The prognostic im- A497. portance of tumor size in malignant gliomas: a computed Zakhary R, Keles GE, Berger MS. 1999. Intraoperative imag- tomographic scan study by the Brain Tumor Cooperative ing techniques in the treatment of brain tumors. Curr Opin Group. J Clin Oncol 6:338–343. Oncol 11:152–156. Woolsey CN, Erickson TC, Gilson WE. 1979. Localization in so- Zeltzer PM, Boyett JM, Finlay JL, et al. 1999. Metastasis stage, matic sensory and motor areas of human cerebral cortex adjuvant treatment, and residual tumor are prognostic fac- as determined by direct recording of evoked potentials and tors for medulloblastoma in children: conclusions from the electrical stimulation. J Neurosurg 51:476–506. Children’s Cancer Group 921 randomized phase III study. Yingling CD, Ojemann S, Dodson B, Harrington MJ, Berger MS. J Clin Oncol 17: 832–845. 1999. Identification of motor pathways during tumor sur- Zentner J, Hufnagel A, Wolf HK, et al. 1997. Surgical treatment gery facilitated by multichannel electromyographic record- of neoplasms associated with medically intractable epilepsy. ing. J Neurosurg 91:922–927. Neurosurgery 41:378–387.