Gemistocytic astrocytomas: a reappraisal HENDRIKUS G. J. KROUWER, M.D., RICHARD L. DAVIS, M.D., PAMELA SILVER, B.A., AND MICHAEL PRADOS, M.D.

Transcription

1 J Neurosurg 74: , 1991 Gemistocytic astrocytomas: a reappraisal HENDRIKUS G. J. KROUWER, M.D., RICHARD L. DAVIS, M.D., PAMELA SILVER, B.A., AND MICHAEL PRADOS, M.D. Neuro-Oncology Service, Brain Tumor Research Center of the Department of Neurological Surgery, and Neuropathology Unit, Department of Pathology, School of Medicine, University of California, San Francisco, California v" Although gemistocytic astrocytomas are considered slow-growing astrocytomas, they often behave aggressively. To clarify the biological and clinical behavior of these rare tumors, the authors retrospectively identified 59 patients with gemistocytic astrocytoma whose tumors were diagnosed and treated between June, 1976, and July, Three patients who were lost to follow-up review were excluded, as were two whose original slides could not be obtained and three whose tumors were diagnosed at recurrence or at autopsy. The pathological material of the remaining 51 patients was reviewed using two sets of histological criteria. Thirteen patients (Group A) had "pure" gemistocytic astrocytoma, defined as a glial tumor with more than 60% gemistocytes/high-power field and a background of fibrillary astrocytes. Fifteen patients (Group B) had "mixed" gemistocytic astrocytoma, defined as a glial tumor with 20% to 60% gemistocytes/high-power field and a background of anaplastic astrocytes. Twenty-three tumors did not meet these criteria and were excluded from analysis. The median age of the patients was 48.5 years in Group A and 38.3 years in Group B (p < 0.05). In both groups, the median Karnofsky Performance Scale score was greater than 90%. All patients underwent surgical procedures (four total and 19 partial resections, and five biopsies) and postoperative radiation therapy. The majority also had interstitial brachytherapy, chemotherapy, or both. Ten patients had one reoperation for tumor recurrence and one had two reoperations; other treatments for recurrence included brachytherapy, chemotherapy, and repeat irradiation. All four patients who originally underwent gross total resection are still alive; all five who had a biopsy have died. There was no significant difference in median survival times between groups: weeks in Group A (range I0 to 310+ weeks) and weeks in Group B (range 31 to 460+ weeks). Analysis of all 28 patients showed a better prognosis for patients less than 50 years of age ( 185 vs. 36 weeks survival time; p < ), patients with preoperative symptoms lasting for more than 6 months (228.1 vs weeks survival time; p < 0.05), and patients with seizures as the first symptom (185.7 vs. 80 weeks survival time; p < 0.01). Survival time did not correlate with the presence of perivascular lymphocytic infiltration. The authors conclude that the presence of at least 20% gemistocytes in a glial neoplasm is a poor prognostic sign, irrespective of the pathological background. It is proposed that gemistocytic astrocytomas be classified with anaplastic astrocytomas and treated accordingly. KEY WORDS 9 gemistocytic astrocytoma 9 anaplastic astrocytoma 9 glioma 9 lymphocytic infiltration G EM STOCYTIC astrocytomas are rare tumors, accounting for 21.4% ~E and 24% 2s of all astrocytomas in the earliest studies and 9% to 19% in more recent surveys, j6'2z'25"48 The wide variation in incidence seems to reflect differences in the histological definition. The most specific definition of gemistocytic astrocytoma is given in the classification published by the World Health Organization (WHO): 56 "a tumour composed predominantly of large, plump astrocytes with abundant eosinophilic cytoplasm and one or more, usually eccentric, nuclei." Although gemistocytic astrocytomas are considered Grade 2 astrocytomas in the Kernohan-Sayre system, 23 they show a propensity for anaplastic transformation. 38 Data on their biological and clinical behavior are scarce but generally seem to support this observation. 1~ In an autoradiographic study of gliomas exposed to tritiated thymidine, however, Hoshino, et al., 2~ showed that gemistocytic astrocytes themselves have very little, if any, proliferative potential. These conflicting data led us to review all patients diagnosed with gemistocytic astrocytoma between June, 1976, and July, The purpose of this retrospective study was to reconsider the definition of these tumors J. Neurosurg. / Volume 74/March,

2 H. G. J. Krouwer, et al. FIG. 1. Photomicrographs of a paraffin-embedded specimen of"pure" gemistocytic astrocytoma from Case 11, Group A. Left. Section composed of sheets of gemistocytes (clearly > 60% of the tumor cells). There is a moderately dense lymphocytic infiltrate around a local small vessel. H & E, original magnification x 40. Right: Plastic-embedded specimen from the same tumor. Plump, round cytoplasmic masses are evident and fine processes can be seen in the background. H & E, original magnification x 100. and to correlate the clinical and histopathological findings with survival. Clinical Material and Methods A review of the data base of the Neuro-Oncology Service, University of California at San Francisco (UCSF), identified 59 patients with an initial diagnosis of gemistocytic astrocytoma made between June, 1976, and July, Eight patients were excluded: three who were lost to follow-up review, two whose tumors were diagnosed at recurrence, one whose tumor was found at autopsy, and two whose original slides could not be obtained. The slides of tumor specimens from the initial operation of the remaining 51 patients were reviewed by two of us (H.G.J.K. and R.LD.) using strictly applied criteria. "Pure" gemistocytic astrocytoma was defined as a glial neoplasm with mere than 60% gemistocytes in all high-power fields and a background of fibrillary astrocytes (Fig. 1). "Mixed" gemistocytic astrocytoma was defined as a glial neoplasm with 20% to 60% gemistocytes and a background of anaplastic astrocytes (moderately increased cellularity, at least focally; a high nuclear/cytoplasmic ratio; coarse nuclear chromatin; increased mitotic activity; and nuclear or cytoplasmic pleomorphism) (Fig. 2). Patients whose tumors did not meet these criteria were excluded from the analysis. The pathological material was also reviewed to identify perivascular lymphocytic infiltration. Lymphocytes were defined as small cells with round, dark-staining nuclei and little or no visible cytoplasm. No attempt was made to classify the degree of perivascular lymphocytic infiltration, and infiltrates near areas of necrosis were not taken into account. In addition, all slides of material obtained during subsequent operations for recurrent tumor were reviewed with special attention to features indicating malignant transformation. The patients were further analyzed regarding age, Karnofsky Performance Scale score, duration and type of preoperative signs and symptoms, extent of surgery, adjuvant therapy, treatment at recurrence, and duration of survival after the initial operation. If available, the bromodeoxyuridine labeling index was considered an indicator of proliferative potential) 7-~9 Tumor recurrence was assessed by neurological examination and contrast-enhanced computerized tomography? 6 Survival curves were calculated using the method of Kaplan and Meier. 2~ Statistical comparison between groups was performed using the Wilcoxon- Gehan test) 4 Histopathology Summary of Cases During the 13 years encompassed by this study, several pathologists made the diagnosis of gemistocytic astrocytoma in the total series of 59 patients, almost certainly using various definitions. Therefore, the diagnosis of gemistocytic astrocytoma was confirmed in only 28 patients: 13 with "pure" gemistocytic astrocytoma (Group A) and 15 with "mixed" gemistocytic astrocytoma (Group B). These 28 patients represent approximately 1.3% of all patients seen at the Neuro- Oncology Service since According to histological criteria currently used to classify malignant gliomas at UCSF (Table 1), 14 of the remaining 23 tumors were glioblastomas multiforme and seven were highly anaplastic astrocytomas. One tumor was reclassified as a subependymal giant-cell astrocytoma and one as a mixed protoplasmic/fibrillary astrocytoma. These 23 tumors were excluded from further analysis. Perivascular lymphocytic infiltration was identified in 15 (54%) of 28 tumors. The bromodeoxyuridine labeling index was available for one patient in Group A (< 1%) and three patients in Group B (1.1%, 6.5%, 400 J. Neurosurg. / Volume 74/March, 1991

3 Gemistocytic astrocytomas FIG. 2. Photomicrographs of a paraffin-embedded specimen of "mixed" gemistocytic astrocytoma from Case 2, Group B. Upper L@: Section showing an area of gemistocytic astrocytes. H & E. original magnification 25. Upper Right: Area of anaplastic astrocytes in the same tumor. H & E, original magnification 250. Lower. Paraffin-embedded specimen of recurrent tumor from the same patient shows high cellularity and necrosis. This tumor was therefore reclassified as glioblastoma multiforme. H & E, original magnification x 10. and 7.0%). All labeled cells were anaplastic astrocytcs. No labeling of gemistocytes was seen. Only one tumor showed malignant transformation to glioblastoma multiforme at reoperation (Fig. 2 lower). In this tumor, originally a "mixed" gemistocytic astrocytoma, all of the histological features we consider essential for the diagnosis of glioblastoma multiforme (Table l) were unequivocally present, and necrosis was seen as well. The other recurrent tumors showed the TABLE 1 Histological criteria fi?r diagnosrs ~! ghoblastoma mult~forme and high@ anapla.~tic astro~ttoma glioblastoma multiforme a glial neoplasm that is at least focally highly cellular nuclear pleomorphism cytoplasmic pleomorphism vascular endothelial proliferation highly anaplastic astrocytoma not a glioblastoma multiforme at least focally moderately to highly cellular presence of at least two of the following characteristics: high nuclear/cytoplasmic ratio coarse nuclear chromatin much mitotic activity nuclear pleomorphism cytoplasmic pleomorphism typical histological features of astrocytomas after radiation therapy or chemotherapy: focal necrosis, astrocytes with bizarre nuclei, and mural vascular hyalinization. At least focally, some malignant features were also retained (increased cellularity, a high nuclear/cytoplasmic ratio, coarse nuclear chromatin, and mitoses). In six patients, focally prominent gemistocytes were still present. Clinical Characteristics The clinical characteristics are summarized in Table 2. The male:female ratio was 3:1. The median age was significantly higher in Group A than in Group B (48.5 vs years; p < 0.05), The median score on the Karnofsky Performance Scale at presentation was 95% (range 70% to 100%). All tumors were supratentorial and evenly distributed between the left and right hemispheres; 50% of the tumors were frontal and 32% were parietal. The median duration of preoperative symptoms was 21.9 weeks. The initial signs and symptoms are summarized in Table 3. The majority of the patients in both groups presented with seizures. Headache, personality changes, and disturbances of speech, reading, or writing were also observed. Initial Treatment The initial treatment is summarized in Tables 4 and 5. All 28 patients underwent surgery followed by radiation therapy. Four patients underwent gross total resection of their tumors as determined from the operative report; 19 had subtotal resection, and five had biopsy only. Ten patients received whole-brain radia- J. Neurosurg. / Volume 74/March,

4 H. G. J. Krouwer, et al. TABLE 2 Clinical characteristics" m 28 patients wilh GA * Characteristics Group A ("pure" GA) Group B ("mixed" GA) Total Cases sex (M:F) 12:1 9:6 21:7 age (yrs) median range KPS (%) median range duration of preop symptoms (wks) median L range tumor location rt hemisphere It hemisphere frontal frontotemporal temporal parietal occipital 1 -- l * GA = gemistocytic astrocytoma; KPS = Karnofsky Performance Scale score. TABLE 3 Presenting signs and symptoms in 28 patients with GA * Symptom Group A Group B Total ("pure" GA) ("mixed" GA) Cases headache nausea/vomiting visual disturbances personality changes (confu sion, memory loss, apraxia) epileptic seizures (grand real, focal motor or sensory, partial complex) weakness paresthesias speech/reading/writing dif ficulties * GA = gemistocytic astrocytoma. tion therapy (four also received a tumor radiation boost) and 18 received focal irradiation to the tumor plus a 2- to 3-cm margin. High-activity 125I seeds were implanted in the tumor in five patients after external irradiation. The median dose of whole-brain radiation therapy was 50.4 Gy (range 44 to 60 Gy) and that of focal irradiation was 60.6 Gy (range 50.8 to 61.2 Gy). The dose of interstitial brachytherapy ranged from 49 to 60 Gy. Nineteen patients were diagnosed and treated at UCSF according to protocols used at the time of presentation (Tables 4 and 5). The remaining nine patients were initially treated elsewhere and were managed at UCSF for tumor recurrence. All nine patients underwent radiation therapy, and two subsequently received adjuvant chemotherapy (lomustine (CCNU) in one and carmustine (BCNU) in the other). TABLE 4 Treatment and survival of 13 patients with "pure" GA (Group A)* Case Age Initial Treatment at Survival No. (yrs), Operation Initial Recurrence (wks) Sex (resection) Treatment (resection, drugs) I 55, M biopsy RT, BUdR-PCV , M subtotal RT, HU , M subtotal RT, BUdR-PCV subtotal , M subtotal RT, HU , M biopsy RT, BUdR-PCV -- 48"t" 6 37, M subtotal RT, J2SI-PCV subtotal ( , M biopsy RT, ~251-PCV subtotal , F subtotal RT subtotal; NU; 308 Proc 9 61, M subtotal RT subtotal, NU; 110 AZQ 10 43, M biopsy RT, tz~i-pcv subtotal , M subtotal RT, HU, PCV IFNB 103~ 12 40, M grosstotal RT, BUdR-PCV ~ 13 3I,M subtotal RT, BCNU NU; RT, NU; 252 subtotal, NU, Proc * GA = gemistocytic astrocytoma; subtotal = subtotal resection; gross total = gross total resection; RT = radiation therapy; BCNU = carmustine; BUdR-PCV = bromodeoxyuridine as a parenterally administered radiosensitizer, followed by lomustine (CCNU), procarbazine, and vincristine; 1251-PCV = interstitial brachytherapy with ~2~I, followed by CCNU, procarbazine, and vincristine; HU = hydroxyurea; NU = nitrosourea-based chemotherapy; Proc = procarbazine; AZQ = aziridinylbenzoquinone; IFNB = interferon B; PCV = procarbazine, CCNU, and vincfistine. t Died of pulmonary embolism. Alive at time of analysis. Treatment at Recurrence Eighteen patients had recurrent tumors. Nine had one recurrence, seven had two recurrences, and two had three recurrences. All 29 recurrences were local; in three patients, a new adjacent lesion developed. Ten patients had one reoperation for tumor recurrence and one had two reoperations; this was followed by J2sI interstitial brachytherapy in one patient and chemotherapy in four others. Chemotherapy alone was given in 13 instances of recurrence, and external reirradiation was administered in two patients (concurrently with BCNU in one). Two patients refused therapy for their second recurrence. Survival Time The median survival duration was weeks in Group A, weeks in Group B, and weeks overall. (The slightly higher overall median survival is a mathematical artifact resulting from the linear extrapolation used to calculate median survival.) Kaplan- Meier 2t survival curves for all 28 patients and for Groups A and B separately are shown in Fig. 3. Patients who received multimodality therapy postoperatively survived longer than those who received radiation therapy only (144 vs. 124 weeks), but the difference was not statistically significant. Fifteen of the 18 patients with recurrent tumors have died. The median duration 402 J. Neurosurg. / Volume 74/March, 1991

5 Gemistocytic astrocytomas TABLE 5 Treatment and survival o.[15 patients with "mixed" (7.4 (Group B)* Case Age Initial Treatment at Survival No. (yrs), Operation Initial Recurrence (wks) Sex (resection) Treatment (resection, drugs) 1 46, F gross total RT, BUdR-PCV , M subtotal RT, BUdR-PCV subtotal, IFNB; 121 NU 3 29, M subtotal RT subtotal, ~2~I; 138 NU 4 73, M subtotal RT AZQ , M subtotal RT, HU, PCV , M grosstotal RT, BUdR-PCV , M subtotal RT subtotal, NU; 117I" 269 RT 8 40, M gross total RT, HU, Miso, 4601" BCNU, 5-FU, & Proc-VCR alternating 9 24, M subtotal 10 30, F subtotal RT, BUdR-PCV RT NU II 44, F subtotal RT, ':~I-PCV subtotal 109t 12 39, M biopsy RT NU , F subtotal RT, BUdR-PCV NU 155t 14 30, M subtotal RT, CCNU AZQ, 6-TG , F subtotal RT, ~251-PCV AZQ 106 * GA = gemistocytic astrocytoma; subtotal = subtotal resection; RT = radiation therapy; VCR = vincristine; BCNU = carmustine; CCNU = Iomustine; BUdR-PCV = bromodeoxyuridine as a parenterally administered radiosensitizer, followed by CCNU, procarbazine, and VCR: t251-pcv = interstitial brachytherapy with t25i, followed by CCNU, procarbazine, and VCR: HU = hydroxyurea; Miso = misonidazole; 5-FU = 5-fluorouracil; Proc = procarbazine; IFNB = interferon B; NU = nitrosourea-based chemotherapy; ~251 = interstitial brachytherapy; AZQ = aziridinylbenzoquinone; 6-TG = 6-thioguanine. Alive at time of analysis. of survival after the first recurrence was 26 weeks. Excluding six patients with insufficient follow-up time and one patient who died of pulmonary embolism, the 5-year survival rate was 23.8% overall (five of 21), 18.2% in Group A (two of 11, with one patient still alive at 310+ weeks), and 30% in Group B (three of 10, with one patient still alive at 460+ weeks). Age correlated inversely with period of survival. Median survival time was longer in patients younger than 50 years of age at diagnosis than in older patients (185 vs. 36 weeks; p < 0.001). Patients with preoperative symptoms for more than 6 months survived longer than those with a shorter duration of symptoms (median vs weeks; p < 0.05). Seizures as the first symptom were associated with a longer median survival time ( vs. 80 weeks, p < 0.01 ). There were no significant differences in median survival time between patients with and those without perivascular lymphocytic infiltration (135 vs. 139 weeks). One patient in Group A with a bromodeoxyuridine labeling index of less than 1% survived 60 weeks. Three patients in Group B with labeling indices of 1.1%, 6.5%, and 7.0% survived for 31, 41, and 106 weeks, respectively. FIG. 3. Kaplan-Meier survival curves for all 28 patients and for those in Group A ("pure" gemistocytic astrocytoma) and Group B ("mixed" gemistocytic astrocytoma). Numbers in parentheses represent the ratio of survivors (and one censored patient) to the total number of patients in each group. The extent of resection also influenced outcome. The four patients who underwent gross total resection were still alive at the time of analysis, 117, 231, 310, and 460 weeks postoperatively. Of the five patients who had a biopsy only, four died after 10, 53, 80, and 153 weeks, and the remaining patient died of a pulmonary embolism after 48 weeks. All patients who underwent gross total resection had frontal tumors; more eloquent areas were affected in the biopsied patients (four with a parietal and one with an occipital tumor). Discussion Histopathologieal Considerations In 1935, Elvidge, et at., 12 introduced gemistocytic astrocytomas as a subtype of astrocytoma, defining them as tumors consisting of "an almost pure culture of gemistocytes" (gemistocytic astrocytes). The term gemistocyte (from the Greek word gemistos, meaning "filled up") describes a large (15- to 40-urn), round-tooval cell with abundant cytoplasm. 2~ Gemistocytic astrocytes are not found in normal brain but generally result from pathological conditions affecting brain tissue, including infiltration by neoplastic astrocytomas. ~2 The origin and significance of these cells are uncertain. In an autoradiographic study of gliomas, Hoshino, et al., 2~ found no labeled gemistocytes in biopsy specimens and only small foci of labeled gemistocytes in autopsy specimens. They concluded that gemistocytic astrocytes cannot synthesize deoxyribonucleic acid (DNA), but occasionally incorporate tritiated thymidine in" their original state as fibrillary or protoplasmic astrocytes, thereafter undergoing transformation into gemistocytic astrocytes and entering the nonproliferating pool. According to these authors, gemistocytes merely reflect the intense proliferative activity of surrounding cells but are relatively inert themselves. This hypothesis, however, does not explain the frequent observation of aggressive growth in these tumors consisting predominantly of gemistocytes. 2~ J. Neurosurg. / Volume 74/March,

6 H. G. J. Krouwer, et al. The histopathological features of gemistocytic astrocytomas have not been well defined. Indeed, Kernohan and Sayre 23 included these tumors in Grade 1 (gemistocytic astrocytoma) as well as in Grade 2 (gem~istete cell astrocytoma) of their classification system. 23 The WHO 56 defines gemistocytic astrocytoma as a tumor composed "predominantly" of gemistocytic astrocytes; no further description is offered of the gemistocytic component or, surprisingly, of the astrocytic component. Glanzmann, et al., 16 attempted to explain the differences in survival times between patients with gemistocytic astrocytoma by distinguishing between "welldifferentiated" and "anaplastic" gemistocytic astrocytomas, but their criteria were not clearly defined. Foci of gemistocytes may be found in any type of astrocytoma, especially diffuse fibrillary astrocytomas and glioblastomas multiforme. ~238'4~ This makes sampling errors a crucial issue in the pathological diagnosis of brain tumors. Consequently, the diagnosis of "pure" gemistocytic astrocytoma in four biopsy specimens is debatable and might reflect foci of gemistocytes within a glioblastoma multiforme, for which the prognosis is accordingly worse. Owing to the lack of autopsy data, we cannot evaluate this possibility; however, exclusion of these four patients from the "pure" gemistocytic astrocytoma group did not significantly alter the median survival time (136.5 vs. 149 weeks for the remaining nine patients) and they were therefore included in the final analysis. Perivascular lymphocyte infiltration has been reported in 30% to 89% ofgliomas. 2'3"34-36"3~ An especially strong association has been noted between this feature and gemistocytic astrocytoma. Takeuchi and Barnard 5~ identified perivascular lymphocytic infiltration in 31 (28%) of 111 supratentorial astrocytomas and in 21 (62%) of 34 gemistocytic astrocytomas; however, their criteria for diagnosing gemistocytic astrocytomas were less strict than ours. Perivascular lymphocytic infiltration was present in 15 (54%) of 28 tumors in our series. Lymphocytic infiltration, especially perivascular lymphocytic infiltration, in gliomas has variously been found to improve the duration of survival, 3"4"9'29'~3 exert no influence on survival time, 6,7,37,43 and reduce the survival period. 39'5t In our series, there was no significant difference in median survival times between patients with and those without perivascular lymphocytic infiltration (135 vs. 139 weeks, respectively). Malignant transformation of lower-grade astrocytomas is reportedly frequent, 5'243~ occurring in as many as 65% of patients. 3~ Russell and Rubinstein 3~ noted conversion to glioblastoma multiforme in 80% of their cases of gemistoeytic astrocytoma. In our seties, however, only one of 12 recurrent tumors showed malignant transformation. This discrepancy may be explained by our lack of autopsy data and by our strict histopathological criteria for diagnosing glioblastoma multiforme, according to which the mere presence of necrosis after irradiation or chemotherapy does not necessarily indicate malignant transformation. In this study, the bromodeoxyuridine labeling index was obtained in only four tumors. In contrast to recent conclusions of Hoshino, et al., ~8 prognostic implications cannot be inferred from the results of labeling studies in these few cases. One patient whose "pure" gemistocytic astrocytoma had a bromodeoxyuridine labeling index of less than l% survived 60 weeks, while three patients whose "mixed" gemistocytic astrocytomas had labeling indices of 1.1%, 6.5%, and 7.0% survived 31, 41, and 106 weeks, respectively. All of the labeled cells were anaplastic astrocytes. The absence of bromodeoxyuridine-labeled gemistocytes in these specimens seems to support previous findings that the proliferative potential of gemistocytes is extremely low. ~5,2o The discrepancy between the behavior of individual gemistocytes and the frequently aggressive growth of gemistocytic astrocytomas is therefore still unexplained. Clinical Considerations Clinical data on patients with gemistocytic astrocytomas are scarce. ~~ ~,28 Elvidge, et al., ~2 noted the exclusively supratentorial occurrence of gemistocytic astrocytomas which we also found in our series. In other series, ~'28 the male-to-female ratio was approximately 3:2, which corresponds to the ratio for all astrocytomas. 4~ In our series, males predominated by a ratio of 3:1 overall and by 12:1 among patients with "pure" gemistocytic astrocytoma. The median age at diagnosis (42.4 years) was similar to that reported by other authors (38.2 years l~ and 40 years28). The median age of our patients with "pure" gemistocytic astrocytomas, however, was 48.5 years. Elvidge and Martinez-ColP ~ reported 18 patients with gemistocytic astrocytoma, presumably diagnosed according to the criteria of Elvidge, et al. ~2 Their mean survival time was 42.7 months; eight patients survived 4 years or longer and one lived 10 years 4 months. In the series of Levy and Elvidge, 28 the mean survival time was 31 months in 18 patients who had incomplete resection and 46 months in those who had complete resection. The mean survival time was 26 months for the 19 patients who died and 70 months for the five patients who were still alive. These data are difficult to interpret because postoperative radiation therapy was not used routinely. In a report on the role of radiation therapy in the treatment of astrocytomas, Leibel, et al., 25 found only one of 1 1 patients with gemistocytic astrocytoma who survived more than 5 years. In a similar report of 77 patients, Glanzmann, et al.,'6 described six "well-differentiated" and nine "anaplastic" gemistocytic astrocytomas. Three of the patients with well-differentiated gemistocytic astrocytomas lived 5 years and one lived 10 years; in contrast, no patient with anaplastic gemistocytic astrocytoma lived longer than 6 years. Several authors j38"4~'44 have mentioned the unfavorable clinical course of gemistocytic astrocytoma without providing 404 J. Neurosurg. / Volume 74~March, 1991

7 Gemistocytic astrocytomas specific survival data or numbers of patients. In a recent report of radiation therapy for low-grade supratentorial astrocytomas, Shaw, et at.,48 described 17 patients with gemistocytic astrocytoma, as defined by the WHO? 6 They found the gemistocytic subtype to be associated with shorter survival times but did not present survival data for this subgroup. Our efforts to clarify the biological and clinical behavior of gemistocytic astrocytomas are based on histopathological criteria that differentiate between "pure" gemistocytic astrocytoma (> 60% gemistocytes with a background of fibrillary astrocytes) and "mixed" gemistocytic astrocytoma (20% to 60% gemistocytes with a background of anaplastic astrocytes). Although both groups received similar treatment, there was little difference in median survival time (136.5 vs weeks); this was less than in patients with anaplastic astrocytomas treated with postoperative radiation therapy and adjuvant chemotherapy with CCNU, procarbazine, and vincristine (157 weeks) 27 and was similar to that in patients with atypical and anaplastic astrocytomas (28 and 36 months, respectively). 3~32 Moreover, the 5-year survival rate in our series was considerably less than in recent studies of patients with low-grade astrocytomas treated with postoperative radiation therapy (23.8% vs. 50% to 60%)234s49 The only significant difference in prognostieally important variables between groups was the somewhat younger median age of patients with "mixed" gemistoeytic astroeytomas which, if anything, would have a favorable influence on prognosis. Thus, despite its nonanaplastic morphology and the nonproliferative character of its predominant cellular component, "pure" gemistocytic astrocytomas in our series were as aggressive clinically as "mixed" gemistocytic astrocytomas, which have more anaplastic astrocytic features. The prognostic importance of age, duration of preoperative symptoms, and seizures at presentation among patients with gliomas has been well documented. 7'46-52 Our study confirms the favorable impact on survival time of the following features: age less than 50 years, preoperative symptoms lasting more than 6 months, and the occurrence of seizures as the initial symptom. Our findings also support the importance of the extent of surgical resection as a prognostic variable in glioma patients. ~'4~ Conclusions Our findings suggest that the presence of at least 20% gemistocytes in a glial neoplasm is an unfavorable prognostic sign, regardless of whether the surrounding background has more fibrillary or more anaplastic astrocytic components. We propose that such tumors be classified as anaplastic astrocytomas and be considered "high-grade" rather than "low-grade" astrocytomas. These tumors should be resected as extensively as feasible and be treated postoperatively with focal radiation therapy followed by nitrosourea-based chemotherapy. Acknowledgments We thank Susan Owen and Cheryl Christensen for manuscript preparation and Stephen Ordway for editorial assistance. References 1. Barnard RO: The pathology of brain turnouts, in Bleehen NM (ed): Turnouts of the Brain. Berlin: Springer-Verlag, 1986, pp Bertrand I, Mannen H: Etude des reactions vasculaires dans les astrocytomes. Rev Neurol 102:3-19, Boker DK, Kalff R, Gullotta F, etal: Mononuclear infiltrates in human intracranial tumors as a prognostic factor. Influence of preoperative steroid treatment. I. Glioblastoma. Clin Neuropathol 3: , Brooks WH, Markesbery WR, Gupta GD, etal: Relationship of lymphocyte invasion and survival of brain tumor patients. Ann Neurol 4: , Bucy PC, Gustafson WA: Structure, nature and classification of the cerebellar astrocytomas. Am J Cancer 35: , Burger PC, Vogel FS, Green SB, etal: Glioblastoma multiforme and anaplastic astrocytoma. Pathological criteria and prognostic implications. Cancer 56: l 1, Burger PC, Vollmer RT: Histologic factors of prognostic significance in the glioblastoma multiforme. Cancer 46: 1179-I 186, Chang CH, Horton J, Schoenfeld D, et al: Comparison of postoperative radiotherapy and combined postoperative radiotherapy and chemotherapy in the multidisciplinary management of malignant gliomas. Cancer 52: , Di Lorenzo N, Palma L, Nicole S: Lymphocytic infiltration in long-survival glioblastomas: possible host's resistance. Acta Neurochir 39:27-33, Elvidge AR: Long-term survival in the astrocytoma series. J Neurosurg 28: , Elvidge AR, Martinez-Coll A: Long-term follow-up of 106 cases of astrocytoma, J Neurosurg 13: , Elvidge AR, Penfield W, Cone W: The gliomas of the central nervous system. A study of two hundred and ten verified cases. Proe Assoe Res Nerv Ment Dis 16: , Garcia DM, Fulling KH, Marks JE: The value of radiation therapy in addition to surgery for astrocytomas of the adult cerebrum. Cancer 55: , Gehan EA: A generalized Wilcoxon test for comparing arbitrarily singly-censored data. Biometrika 52: , Gcrmano IM, Ito M, Cho KG, etal: Correlation of histopathological features and proliferative potential of gliomas. J Neurosurg 70: , Glanzmann CH, Peters J, Horst W, etal: Indikationen und Ergebnisse der Radiotherapie in der Behandlung yon Astrozytomen. Strahlentherapie 156: , Hoshino T, Nagashima T, Murovic JA, et al: In situ cell kinetics studies on human neuroectodermal tumors with bromodeoxyuridine labeling. J Neurosnrg 64: , Hoshino T, Prados M, Wilson CB, etal: Prognostic implications of the bromodeoxyuridine labeling index of human gliomas. J Neurosurg 71: , Hoshino T, Rodriguez LA, Cho KG, etal: Prognostic implications of the proliferative potential of low-grade astrocytomas. J Neurosurg 69: , 1988 J. Neurosurg. / Volume 74 /March,

8 H. G. J. Krouwer, et al. 20. Hoshino T, Wilson CB, Ellis WG: Gemistocytic astrocytes in gliomas: an autoradiographic study. J Neuropathol Exp Nenrol 34: , Kaplan EL, Meier P: Nonparametric estimation from incomplete observations, d Am Stat Assoc 53: , Katakura R, Yoshimoto T: Epidemiology and statistical analysis ofgliomas, in Suzuki J (ed): Treatment of Glioma. Tokyo: Springer-Verlag, 1988, pp Kernohan JW, Sayre GP: Tumors of the central nervous system, in: Atlas of Tumor Pathology, Fascicle 35. Washington, DC: Armed Forces Institute of Pathology, Laws ER, Taylor WF, Clifton MB, et al: Neurosurgical management of low-grade astrocytoma of the cerebral hemispheres, d Neurosurg 61: , Leibel SA, Sheline GE, Wara WM, et al: The role of radiation therapy in the treatment of astrocytomas. Caneer 35: , Levin VA, Crafts DC, Norman DM, et al: Criteria for evaluating patients undergoing chemotherapy for malignant brain tumors. 3 Neurosurg 47: , Levin VA, Silver P, Hannigan J, et al: Superiority of postradiotherapy adjuvant chemotherapy with CCNU, procarbazine, and vincristine (PCV) over BCNU for anaplastic gliomas: NCOG 6G61 final report. Int J Radiat Oncol Biol Phys 18: , Levy LF, Elvidge AR: Astrocytoma of the brain and spinal cord. A review of 176 cases, J Neurosurg 13: , Manoury R, Vedrenne C, Constans JP: Infiltrations lymphocytaires dans les gliomes humains. Neurochirnrgie 21: , Mfiller W, Afra D, Schrrder R: Supratentorial recurrences of gliomas. Morphological studies in relation to time intervals with astrocytomas. Acta Neurochir 37:75-91, Nelson DF, Nelson JS, Davis DR, et al: Survival and prognosis of patients with astrocytoma with atypical or anaplastic features. J Neurooncol 3:99-103, Nelson JS, Tsukada Y, Schoenfeld D, et al: Necrosis as a prognostic criterion in malignant supratentorial, astrocytic gliomas. Cancer 52: , Palma L, Di Lorenzo N, Guidetti B: Lymphocytic infiltrates in primary glioblastomas and recidivous gliomas. Incidence, fate, and relevance to prognosis in 228 operated cases, d Neurosurg 49:854-86l, Ridley A, Cavanagh JB: Lymphocytic infiltration in gliomas: evidence of possible host resistance. Brain 94: , Rossi ML, Cruz-Sanchez F, Hughes JT, et al: Mononuclear cell infiltrate and HLA-DR expression in low grade astrocytomas: an immunohistological study of 23 cases. Acta Neuropathol 76: , Rossi ML, Hughes JT, Esiri MM, et al: Immuno-histological study of mononuclear cell infiltrate in malignant gliomas. Acta Neuropathol 74: , Rossi ML, Jones NR, Candy E, et al: The mononuclear cell infiltrate compared with survival in high-grade astrocytomas. Acta Neuropathol 78: , Russell DS, Rubinstein L J: Pathology of Tumours of the Nervous System, ed 5. London: Williams & Wilkins, Safdari H, Hochberg FH, Richardson EP: Prognostic value of round cell (lymphocyte) infiltration in malignant gliomas. Surg Neurol 23: , Salcman M: Supratentorial gliomas: clinical features and surgical therapy, in Wilkins RH, Rengachary SS (eds): Neurosurgery. New York: McGraw-Hill, 1985, Vol 1, pp Scherer H J: Cerebral astrocytomas and their derivatives. Am J Cancer 40: , Scherer HJ: The forms of growth in gliomas and their practical significance. Brain 63:1-35, Schiffer D, Cavicchioli D, Giordana MT, et al: Analysis of some factors affecting survival in malignant gliomas. Tumori 65: , Schiffer D, Chio A, Giordana MT, et al: Prognostic value of histologic factors in adult cerebral astrocytoma. Cancer 61: , Schoenberg BS: The epidemiology of central nervous system tumors, in Walker MD (ed): Oncology of the Nervous System. Boston: Martinus Nijhoff, 1983, pp Scott GM, Gibberd FB: Epilepsy and other factors in the prognosis of gliomas. Acta Neurol Scand 61: , Shapiro WR, Green SB, Burger PC, et ah Randomized trial of three chemotherapy regimens and two radiotherapy regimens in postoperative treatment of malignant glioma. Brain Tumor Cooperative Group Trial J Neurosurg 71:1-9, Shaw EG, Daumas-Duport C, Scheithauer BW, et al: Radiation therapy in the management of low-grade supratentorial astrocytomas. J Neurosnrg 70: , Shaw EG, Scheithauer BW, Gilbertson DT, et al: Postoperative radiotherapy of supratentorial low-grade astrocytomas. Int J Radiat Oncol Biol Phys 16: , Takeuchi J, Barnard RO: Perivascular lymphocytic cuffing in astrocytomas. Acta Neuropathol 35: , Vaquero J, Coca S, Oya S, et al: Presence and significance of NK cells in glioblastomas. J Neurosurg 70: , t Walker MD, Green SB, Byar DP, el al: Randomized comparisons of radiotherapy and nitrosoureas for the treatment of malignant glioma after surgery. N Engl J Med 303: , Wilson CB: Reoperation for primary tumors. Semin Oncol 2:19-20, Winger MJ, Macdonald DR, Cairncross JG: Supratentorial anaplastic gliomas in adults. The prognostic importance of extent of resection and prior low-grade glioma. 3 Neurosurg 71: , Wood JR, Green SB, Shapiro WR: The prognostic importance of tumor size in malignant gliomas: a computed tomographic scan study by the Brain Tumor Cooperative Group. J Clin Oncol 6: , Z/ilch KJ: Histological Typing of Tumours of the Central Nervous System. International Classification of Tumours, No. 21. Geneva: World Health Organization, 1979, pp Manuscript received March 27, Accepted in final form August 13, This work was supported in part by Grant CA from the National Institutes of Health. Address reprint requests to: Michael Prados, M.D., Department of Neurological Surgery, c/o The Editorial Office, 1360 Ninth Avenue, Suite 2[0, San Francisco, California J. Neurosurg. / Volume 74 / March, 1991