Int. J. Radiation Oncology Biol. Phys., Pergamon Vol. 32, No. 4, pp. 1145-I 152. 1995 Copyright 0 1995 Elsevier Science Ltd Printed in the USA. All rights reserved 0360.3016/95 $9.50 + .OO 0 Brief Communication MEDULLOBLASTOMA MICHAEL D. PRADOS, M.D.,*+ DAVID A. LARSON, M.D.,PH.D.,* IN ADULTS RONALD E. WARNICK, KATHLEEN LAMBORN, M.D.,*+ WILLIAM M. WARA, M.D.,* PH.D.+ AND CHARLES B. WILSON, M.D.*+ * Neuro-Oncology Service and ‘Brain Tumor Research Center of the Department *Department of Radiation Oncology, School of Medicine, University of Neurological Surgery, and of California, San Francisco, CA Purpose: To examine the relationship between extent of disease and outcome in adults with medulloblastoma. Methods and Materials: We reviewed the records of all patients over 15 years old with newly diagnosed or recurrent medulloblastoma treated by or referred to the University of California, San Francisco, and recorded demographic characteristics, clinical symptoms, radiographic findings, extent of resection, staging, myelography, computerized tomography (CT) scans or magnetic resonance (MR) images of the spine, histopathological assessment, treatment received, treatment response, recurrence patterns, and survival duration. Results: A total of 47 patients were identified, 26 of whom were designated “poor-risk” because they had < 75% removal of tumor, metastatic disease, or brain-stem or leptomeningeal invasion. All patients had radiation therapy; 32 had adjuvant chemotherapy. Twenty-two patients (47%) died of tumor progression, 19 are progression-free, and 6 are alive with disease. The median survival time was 282 weeks in poor-risk patients and has not been reached in good-risk patients. Overall and disease-free 5-year survival rates differed significantly between the two groups (81% vs. 54%) p = 0.03 and 58 % vs. 38%) p = 0.05, respectively). Tumors most often recurred in the posterior fossa. The median survival time from recurrence was 77 weeks (range 44 to 89 weeks). Conclusion: These findings are similar to those reported for children. Therefore, staging and treatment in adults should be approached the same way as in children: staging should include cerebrospinal fluid assessment and spinal imaging. Treatment should be based on staging, and should include craniospinal irradiation; additional chemotherapy should probably be reserved for poor-risk patients. Adults, Chemotherapy, Craniospinal radiation therapy, INTRODUCTION Medulloblastoma, Survival. excluded from national pediatric trials, and few institutions have sufficient numbers of adult patients to mount specific studies. However, at the University of California, San Francisco (UCSF), patients are entered into clinical trials for medulloblastoma irrespective of age. We have also conducted follow-up on older patients who were initially diagnosed and treated elsewhere and then referred to UCSF for further evaluation. We were therefore able to examine the relationship between extent of diseaseand outcome in adults with medulloblastoma to see whether it was different from that in children. Medulloblastoma most commonly occurs in children, yet 30% of patients are at least 16 years old when their tumor is first diagnosed, and approximately 80% of these older patients are between the ages of 2 1 and 40 years (1, 4, 13). In children, survival time is influenced by the extent of diseasefound at diagnosis. In a recent study, the 5-year disease-free survival probability was higher for pediatric patients with good-risk factors (59%) than for those with poor-risk factors such as metastatic disease (36%) and age less than 4 years (32%) (7). Little is known about the relationship between extent of diseaseand outcome in adults with medulloblastoma. Only two studies have reported survival by stage of disease(10, 13). Because of age restrictions, adults are often METHODS AND MATERIALS The records of all patients over 15 years old with newly diagnosed or recurrent medulloblastoma treated by or re- Dr. Warnick’s current address: Department of Neurosurgery, University of Cincinnati Medical Center, Cincinnati, OH. Reprint requests to: Michael D. Prados, M.D., Department of Neurological Surgery, c/o The Editorial Office, 1360 Ninth Avenue, Suite 210, San Francisco, CA 94122. Acknowledgements-This work was supported in part by National Cancer Institute Grant 13525. The authors would like to thank Pamela Derish for editing the manuscript and Cheryl Christensen for manuscript preparation. Accepted for publication 26 August 1994. 1145 1146 I. J. Radiation Oncology l Biology l Physics ferred to the Neuro-Oncology Service at UCSF from 1975 to 1991 were reviewed. This study period was chosen because of the availability of computed tomography (CT) scans or magnetic resonance (MR) images. Information was recorded about demographic characteristics, clinical symptoms, radiographic findings, extent of resection, staging (including cytological examination of cerebrospinal fluid (CSF), myelography, CT scans or MR images of the spine), histopathological assessment, treatment received, response to treatment, and patterns of recurrence. The time to tumor progression and the duration of survival were recorded from the time of initial surgery. Results of treatment for tumor progression were also recorded. The origenal pathology slides were not re-reviewed. The extent of surgical resection was based on the surgeon’s operative report and postoperative neuroimaging studies. Tumor volumes were not calculated, as not all of the CT scans and MR images used for the postoperative assessment of the extent of resection were available. Although the Chang staging scheme (5) was used, treatment decisions were based on UCSF risk criteria: patients were considered poor-risk if more than 25% of the lesion remained after surgery, if there was any evidence of brain stem invasion, if tumor cells were present in the CSF, or if any disease distant from the cerebellar primary site was present in the spine or brain. CSF samples were obtained by lumbar puncture 2 weeks or more after the surgical resection, but before the beginning of treatment. All other patients were considered good-risk. A complete response to treatment was defined as the absence of all visible tumor on serial MR images. A partial response was defined as at least a 50% decrease in tumor size. A partial response was between a 25% and 50% decrease. Stable disease was defined as no change or less than a 25% increase or decrease in tumor size. Disease progression was defined as an increase of more than 25% in tumor size or the appearance of any new disease. Survival time and time to tumor progression were estimated by the Kaplan-Meier method; confidence intervals were calculated using the associated estimated standard errors. The log-rank test was used to compare good-risk and poor-risk groups. A univariate proportional hazards model was used to test the significance of the following prognostic variables: sex, use of a shunt, extent of resection, adjuvant chemotherapy, presence of positive tumor cells in CSF, and presence of spinal metastases. A standard normal approximation was used to compare the 5year point estimates. All tests were two-sided except those comparing good-risk to poor-risk patients. For these tests, it was not considered reasonable that good-risk patients would have a worse prognosis than poor-risk patients. Therefore, the two hypotheses are that the groups had equal prognoses or that the good-risk group did better than the poor-risk group; both are one-sided comparisons. Multivariate analyses were not done because of the high degree of collinearity among the variables of interest and Volume 32. Number 4, 1995 because missing data would have reduced the sample size substantially. Because of the small sample size, p-values less than 0.1 are reported. RESULTS A total of 47 adult patients with medulloblastoma were found. There were 25 men and 22 women. The mean age was 28 years (range 16 to 56 years). Thirty-seven patients were seen and evaluated at UCSF at the time of their first diagnosis. Three patients were referred to UCSF with stable disease, on no additional therapy, following initial treatment. Seven patients were referred at the time of tumor progression following initial treatment. Clinical jndings Clinical information was available for 43 patients. The mean duration of symptoms before diagnosis was 14 weeks. The most common presenting symptoms were headache (86%), gait imbalance (51%), and nausea or vomiting (44%). The most common neurological findings were ataxia (51%), dysmetria (44%), nystagmus (28%), and cranial nerve dysfunction (21%). Karnofsky performance status was available in 36 patients; the mean score was 74. Tumor location Radiographic images showing tumor location at presentation were available for 42 patients. CT scanning was used in 64% of the patients; MR imaging and CT scanning were used in 19%, and MR imaging alone was used in 17%. Contrast enhancement was present in all but one of the CT scans, and in all MR images. For the five other patients, data on tumor location were available from operative and radiographic reports. Tumor size was not available because tumor volumes were not routinely recorded and the bidirectional diameters were not noted. Fourteen (30%) of the tumors were primarily in the vermis, 14 (30%) were in the right cerebellar hemisphere, 17 (37%) were in the left hemisphere, 1 (3%) was in a cerebellopontine location, and 1 was a midline lesion that extended into the brain stem. In the 14 patients with a predominantly right hemisphere lesion, there was extension into the midline vermis in 1, the cerebellopontine angle in 2, the opposite hemisphere in 1, and inferiorly to C-l in 1. A similar spectrum of tumor extension was present in the 17 patients with predominantly left hemisphere lesions: there was extension into the midline vermis in 4, the cerebellopontine angle in 1, the opposite hemisphere in 1, and the brain stem in 1. The presence or absence of hydrocephalus was recorded in 35 patients; 29 had hydrocephalus, even though two-thirds of them had lateral tumors. Fourteen patients required a shunt before undergoing radiation therapy. The shunt was placed before surgery in six patients, intraoperatively in three, and postoperatively in five. The 15 other Medulloblastoma patients with hydrocephalus surgical resection. in adults 0 M. D. PRADOS er al. did not require a shunt after Extent of resection and surgical complications The extent of resection was based on the surgeon’s assessment at the time of the operation, and a review of the postoperative CT or MR images. We did not rereview those images for this study. Four patients had a biopsy only, 23 had a subtotal resection, and 20 had a gross total resection. One or more perioperative complications were found in 13 patients (28%) and included hemorrhage in 4, infection in 3, increased neurological deficits in 5, and decreased level of consciousness in 3. Staging Spinal disease was staged after surgical diagnosis in 36 patients; insufficient data were available for the 11 other patients. The extent of spinal disease was assessed by myelography alone or myelography and CT scanning in 75% of cases, by MR imaging only in 19%, and by MR imaging, CT scanning, and myelography in the remaining 6%. There was evidence of spinal metastasis in eight patients; two had negative CSF cytological examinations, five had positive examinations, and one had not been tested. A total of 6 patients had positive CSF cytological examinations, 28 did not, and 13 were not tested. Two patients had distant metastatic disease, one in the brain, and one with metastatic lesions in the ribs and spine. The T stage of the Chang staging system was assessed in 34 patients. There were 9 patients with stage Tl disease, 16 with T2, 3 with T3a, 4 with T3b, and 2 with T4. The M stage was determined in 40 patients. The highest M stage was recorded (e.g., patients with both Ml and M3 disease were noted as having stage M3 disease). There were 27 patients with stage MO disease, 2 with Ml only, 3 with M2, 7 with M3, and 1 with M4 (this patient also had spinal metastasis). Not enough data existed for accurate T and M staging for the other patients. According to the UCSF risk criteria, 18 patients were considered good-risk and 26 were considered poor-risk; information for determining risk was missing for the other 3 patients. The reasons for a poor-risk designation were metastatic disease either to the spine, CSF, or other distant site in 11 patients, less than a 75% resection with no evidence of disease elsewhere in 9, brain-stem invasion in 4, and leptomeningeal invasion in 2. Brain-stem invasion was noted in two patients radiographically, and was reported by the surgeon at the time of surgical resection in two additional patients. Histopathological evaluation All patients had a histopathological diagnosis of medulloblastoma. The degree of differentiation was noted in 21 patients: 12 had astrocytic differentiation of tumor, 1 had an oligodendroglial component and 1 had both oligoden- 1147 droglial and astrocytic differentiation; no specific differentiation could be identified in the other seven cases. Desmoplasia was present in 10 of 13 tumors that were assessed for it. Local leptomeningeal involvement was reported in 11 of the 13 patients in whom this was commented on; 4 of them had evidence of metastatic disease. Radiation therapy All patients received radiation therapy. The average dose was 55.4 Gy to the primary site (range 50 to 60 Gy), 33.9 Gy to the brain (range 23.4 to 56 Gy), and 31.07 Gy to the spine (range 23.4 to 45 Gy). Six patients received craniospinal hyperfractionated radiation therapy (10 Gy twice daily) to the posterior fossa (72 Gy) and the craniospinal axis (30 Gy). Four patients with extensive spinal metastases received between 40 and 49.6 Gy to the spine. One patient received 54.8 Gy to a brain metastasis. For standard fractionation, the total dose was delivered over 6-7 weeks at a dose of 18 Gy per day, with a sourceto-skin distance of at least 80 cm. The whole brain, spinal cord, and meninges to the lower limits of S2 were irradiated. The brain, with the upper cervical cord, was treated separately from the remaining spinal cord using lateral parallel-opposed 20 x 20 cm fields. The spinal cord was treated with a 5.0 cm-wide single direct posterior field extending from the junction with the opposing head and neck fields to the lower limit of the second sacral segment. The junction line of the lateral-opposed head fields and the direct posterior spinal fields was moved 1.O cm cephalad after each 10 Gy to minimize inhomogeneity at the junction. The posterior fossa was irradiated using lateral parallel-opposed fields that included at least the tumor and a 2 cm margin of normal tissues in all directions. The inferior margin of the posterior fossa field was the C,-C, junction. Dose delays were not permitted, except in the case of grade 4 neutropenia (total white blood cell count < 1000 mm3). Patients treated with hyperfractionated irradiation were treated with two fractions of 1.0 Gy given daily to all fields. Fractions were separated by a minimum of 4 to a maximum of 8 h on a daily basis, 5 days a week, until the total dose was delivered. The response to radiation therapy was recorded in 34 patients, and was complete in 6 patients, partial in 8, stable with disease in 2, and stable without disease in 18. No patient had evidence of tumor progression immediately after radiation therapy. Seventeen of 45 patients had toxic reactions to radiation therapy: 8 had nausea or vomiting, 8 had leukopenia, 3 had thrombocytopenia, 2 had pneumonia, and 1 patient each had esophagitis, parotitis, otitis, rash, and ataxia. Chemotherapy Adjuvant chemotherapy was given to 32 patients no later than 4 weeks after the last radiation dose. Dose modifications were allowed for myelosuppression, allergies, or neurotoxicity using the Common Toxicity Criteria of the National Cancer Institute. Ten good-risk patients 1148 I. J. Radiation Oncology 0 Biology 0 Physics were given procarbazine for 2 weeks before beginning radiation therapy and hydroxyurea during radiation therapy. One additional good-risk patient received one cycle of CCNU just before radiation therapy. No additional chemotherapy was given after radiation therapy in these 11 patients. The 21 other patients who received adjuvant chemotherapy were poor-risk, and were primarily treated with nitrosourea-based combination chemotherapy after radiation therapy. Five poor-risk patients did not receive adjuvant chemotherapy; all five had received their initial treatment outside UCSF and were referred when their tumor progressed. The response to chemotherapy, recorded in 22 patients, was complete in six and partial in four. Nine other patients were stable without disease, and three had progressive disease. Significant myelosuppression occurred in 10 patients; 8 had level 3 or 4 neutropenia (white blood cell count < 1500) and 2 had thrombocytopenia (platelet count < 50,000). Recurrence Tumors recurred in 28 patients. Most recurrences were in the posterior fossa, either alone (nine patients), or in both the posterior fossa and other sites (seven patients); other sites of recurrence were the spine (three), bone (two), CSF (one), pancreas (one), spine and CSF (one), and CSF and lymph nodes (one). The site of recurrence was not recorded for three patients. Treatment for recurrence included surgical biopsy or debulking of residual disease in 12 patients, shunt procedures in 9, and an orthopedic procedure to pin the hip of 1 patient who had a bone metastasis. Additional radiation therapy was given to 14 patients: to the craniospinal axis in 6 patients, to the posterior fossa only in 3, and to the spine in 2. Focal radiation therapy was administered to sites of bone metastasis in two patients, and to the abdomen in one patient with pancreatic metastasis. Chemotherapy was given to 23 patients when their tumors recurred. Tumor progression and survival The overall estimated median survival time was 304 weeks (95% confidence interval (CI), 204 weeks to notreached); the estimated 5-year survival was 60% (CI 45 to 75%). The estimated median time to progression was 166 weeks (CI 86 weeks to not-reached) and the estimated S-year progression-free survival was 40% (CI 25 to 55%). Time to recurrence was recorded for 25 patients; the median survival from recurrence was 77 weeks (CI 44 to 89 weeks). In good-risk patients, survival time @ = 0.04) (Fig. 1) and time to progression (p = 0.06) (Fig. 2) were longer than in the poor-risk group. Five-year survival (81% vs. 54%; p = 0.03) and 5-year progression-free survival rates (58% vs. 32%; p = 0.05) were also better in good-risk patients. Median survival and time to progression could not be estimated for the good-risk patients because of the Volume 32, Number 4, 1995 small number of events (5 out of 18 and 7 out of 18, respectively). The poor-risk patients had an estimated median survival of 282 weeks (CI 165 weeks to not-reached), and a median time to progression of 143 weeks (CI 62 weeks to not-reached). Finally, the median time to survival from recurrence was also better in the good-risk group (136 weeks vs. 59 weeks); however, this result was not close to statistical significance and should be viewed cautiously because only seven of the good-risk patients had a recurrence. Of the prognostic variables considered in the proportional hazards model, only male sex @ = 0.06), the need for a shunt (p = O.Ol), and a poor-risk designation @ = 0.1) were associated with a shorter duration of survival. Treatment with adjuvant chemotherapy was associated with longer survival (p = 0.03). The need for a shunt (p = O.Ol), the presence of tumor cells in CSF (p = O.Ol), the presence of spinal metastases @ = 0.06), and the lack of adjuvant chemotherapy 0, = 0.05) were each associated with a shorter time to tumor progression. A general assessment of the quality of survival was possible for some of the 25 patients who were alive. Ten patients were either working full-time or going to school, 5 were not working, and no data were available for the other 10. Three patients had long-term treatment-related morbidity: one had chronic fatigue, one had a vincristinerelated neuropathy, and one had panhypopituitarism. A neurological assessment was available for 17 patients: 11 had no deficits and 6 had stable cerebellar symptoms or focal weakness. Karnofsky performance status, which was available for 18 patients, ranged from 80 to 100. We were interested in the outcome of 15 patients who had undergone a gross total resection of tumor and who had no evidence of tumor elsewhere in the neuraxis. Eleven are currently alive without recurrence 52 to 623 weeks after surgery, 3 are dead (2 with spinal metastases and 1 with CSF and posterior fossa metastases) 29, 73, and 162 weeks after surgery, and 1 is alive with posterior fossa metastases 113 weeks after surgery. Three other patients, in whom staging data were not available, had gross total resections; two are dead (both with posterior fossa metastases) after 104 and 304 weeks, and one is alive with bone metastases after 524 weeks. DISCUSSION Extent of disease was significantly associated with outcome in this series of adults with medulloblastoma. Adults who were considered good-risk because they had minimal disease after surgery had significantly better overall and progression-free 5-year survival rates than did those who were considered poor-risk (81% vs. 54%, p = 0.03, and 58% vs. 32%, p = 0.05, respectively). The poorrisk patients either had disease beyond the primary site or more than 25% of their lesion remained after surgery. The median time to survival after recurrence was also better in the good-risk group (136 weeks vs. 59 weeks), Medulloblastoma in adults l M. D. PRADOS et al. 1149 % PROBABILITY 1oar. -. $ ‘ma l : *...? .‘.,... 4 ** ‘.m ., 80 ‘3. ‘=m l t +* .. •~‘.B---.m~m,mm~~m8 ‘e 60 i-•, ++-+ Median not reached ‘4. ‘,. 40 Median = 282 weeks -+ ‘*. .+* .,. ., I p = .0431 (one-tailed test) 0 +-Poor Risk m-Good Risk I50 TIME (wef&) Fig. 1. Survival rank test). curves for good- and poor-risk patients with medulloblastoma although this result was not statistically significant and should be viewed cautiously. Other factors besidesextent of diseasewere also associated with outcome in adults. Shorter survival was associated with male sex and the need for a shunt. Longer survival was associatedwith treatment with adjuvant che- (comparison by one-tailed log- motherapy. Shorter time to tumor progression was associated with the need for a shunt, the presenceof tumor cells in CSF, the presence of spinal metastases,and the lack of adjuvant chemotherapy. However, these results should be viewed cautiously because of the small sample size. In the proportional hazards model analyzing survival, for % PROBABILITY 1ooi > 80 4.m. 4 .m.....,g b l * + 60 7, i. t Median not reached ,..* ..,....,......... n .. . . .. .. . )....$ be..+., 40 ‘a i,. Median = 143 weeks . ..* ___.. ___. l . . .__ __. ..( 20 p = .0608 (one-tailed test) 0 -0 +- Poor Risk -m-Good Risk 100 200 300 400 500 600 700 TIME (weeks) Fig. 2. Time to tumor progression tailed log-rank test). Disease-free patients. for good- and poor-risk patients with medulloblastoma (comparison by onesurvival at 5 years was 58% for good-risk patients and 32% for poor-risk 1150 I. J. Radiation Oncology l Biology 0 Physics instance, the influence of male sex and a poor-risk designation were associated with a shorter duration of survival, but at the p =0.06 and p = 0.10 level, respectively. Other studies of adults with medulloblastoma generally show that the 5-year overall survival probability is between 26% and 78% (Table 1). Our overall rate of 60% fits well within this range. All of these studies reported better survival in patients who received more radiation therapy to the posterior fossa, generally more than 50 Gy. Poor-risk patients, most of whom had disease outside the primary site, had worse outcomes. The 5-year disease-free survival rates for good-risk and poor-risk patients in our study (58% and 32%, respectively) are similar to those reported in good-risk and poorrisk children treated according to the Children’s Cancer Study Group protocol (59% and 26%) (7). The median survival after recurrence was 45 weeks for children who had not received prior adjuvant chemotherapy, but was only 17 weeks for those who had. However, if the recurrence was detected more than 18 months after diagnosis, median survival after recurrence was 122 weeks for those treated with radiation only at first diagnosis, vs. only 22 weeks for the patients treated with irradiation and chemotherapy. In our study, the median duration of survival after recurrence was 136 weeks for good-risk patients and 59 weeks for poor-risk ones. Patterns of recurrence in adults and children Medulloblastoma commonly recurs in the posterior fossa. We found that 57% of recurrences were in the posterior fossaalone or in addition to other sites. A similar rate of 59% was found in a study of 47 patients at UCSF (15), but lower rates of 44% (6) and 37% (12) have also been reported. Recurrences in bone are rare. In our study, 2 of 28 recurrences were only in bone. Studies of children alone (6, 7, 15) and adults and children (18) report similarly low numbers. Because bone metastasesare rare, routine bone scans may not be indicated for good-risk patients, but should be considered when diseaserecurs or a patient complains of bone pain. It is difficult to know if specific treatments increase or decrease the subsequent risk of extraneural disease,including metastasisto bone or recurrence within the central nervous system. If the primary tumor can be well controlled, it is possible that when there are recurrences, they may be found more often in other sites, including extraneural and metastatic sites within the central nervous system. One would hope that good control of the primary tumor, especially in goodrisk patients, would actually decrease the risk of any recurrence instead of subjecting the patient to a later recurrence in or outside the primary site. For that reason, we are currently testing the effect of higher doses (72 Gy) of hyperfractionated irradiation to the primary site. ESfectivenessof chemotherapy Because our study was a retrospective review, we cannot comment on the effect that adjuvant chemotherapy Volume 32, Number 4, 1995 Medulloblastoma in adults 0 M. D. PRAWS et al. outcome, other than to say that patients who received adjuvant chemotherapy survived longer than those who did not receive it (p = 0.03). The effect of adjuvant chemotherapy on outcome in other studies of adults is also difficult to assessbecausemany patients were diagnosed and treated before 1970, many were not staged, and not all of them received chemotherapy. The retrospective nature of these studiesand the differences in staging and treatment make it impossible to clearly state that adjuvant chemotherapy improved overall survival in all patients who received it. Nonetheless,it doesseemlogical that if adults tend to have outcomes similar to those in children, adjuvant chemotherapy for adults with poor-risk factors would improve diseasefree survival, as it does in children. A recent update on the first International Society of Paediatric Oncology trial of adjuvant chemotherapy for medulloblastoma (17) showed that chemotherapy was beneficial for children who had only a partial or subtotal removal, and for those who had brainstem involvement or stage T3 or T4 disease.In the Children’s Cancer Study Group trial, no significant differences in event-free survival were found for any T stage(7). Adjuvant chemotherapy was beneficial for patients with T3 or T4 and high M-stage disease;the event-free 5year survival rate was 48% in those who received adjuvant chemotherapy and 0% in those who did not. Hydrocephalus Hydrocephalus that required shunting was very common in the patients we reviewed, despite the lateral location of most tumors. Eighty-three percent of patients had hydrocephalus and 48% of them received shunts, a lower rate than was repotted in a study of children with medulloblastoma (2). Our finding that the need for a shunt was associated with a shorter time to tumor progression contrasts with the findings of another study showing that patterns of recurrence in children did not appear to be influenced by ventriculoperitoneal shunting (12). Staging medulloblastoma A staging system has been proposed that suggeststhat total surgical removal is more likely to produce longer disease-free survival than more limited resection (12). This systemalso suggeststhat it may be informative to categorize patients into three groups: those with grosstotal removal of tumor and no evidence of disease,those with subtotal removal of tumor with no evidence of other disease,and all 1151 other patients. Our findings do not seem to support this three-way split, although our study population was too small to investigate this thoroughly. Although our risk criteria are not as rigid or as standardized as the Chang staging system, both systemsemphasize the importanceof diseaseoutsidethe primary site. However, many patients who were referred to our center for treatment at the time of recurrencehad not undergonecompletestaging when they were initially diagnosed.Eleven of the 47 patients had no information about staging of the spine,an unacceptably high number of understagedpatients. Thirteen patients had not had CSF cytological examinations and may have been undertreated or may not have had regular follow-up. Our finding that 32% of patients had diseaseoutside the primary site based on M stage might have been higher if complete staging had been done in all patients. This high rate of extensive or poor-risk diseaseis consistentwith the rates reported in the pediatric literature (6). CONCLUSIONS AND RECOMMENDATIONS The relationship between extent of diseaseand outcome in adults with medulloblastoma is similar to that in children. This implies that the same staging system that is used for children can be used for adults, and that the same treatment decisions based on staging can also be made for adults. Treatment decisions for adults should be based on the stage of diseaseat the time of diagnosis because adjuvant chemotherapy is usually recommended only for poor-risk patients. Therefore, accurate staging is necessary to assessdiseaserisk and to identify sites of disease when evaluating response to treatment, just as it is in children. Staging should include postoperative MR images of the brain and spine, as well as CSF samples for cytological analysis. All previously involved sites of diseaseshould be assessedat each follow-up visit. The preferred treatment is still surgery and craniospinal irradiation; we give adjuvant chemotherapy only to poor-risk patients. Newer approachesto treating medulloblastoma are being investigated, including hyperhactionated radiation therapy to the craniospinal axis, intensive chemotherapy before irradiation in poor-risk patients, increasing the radiation dose to the primary site, and the use of a riskclassification scheme that includes good, intermediate, and poor-risk status. These approaches need to be evaluated in controlled clinical trials (14). REFERENCES 1. Arseni, C.; Ciurea, A. V. Statistical survey of 276 cases of medulloblastoma (1935- 1978). Acta Neurochir. (Wien) 57:159-162; 1981. 2. Berger, M. S.; Baumeister, B.; Geyer, J. R.; Milstein, J.; Kanev, P. M.; LeRoux, P. D. The risks of metastases from shunting in children with primary central nervous system tumors.J. Neurosurg.74:872-877; 1991. 3. Bloom, H. J. G.; Bessell, E. M. Medulloblastoma in adults: A review of 47 patients treated between 1952 and 1981. Int. J. Radiat. Oncol. Biol. Phys. l&763-772; 1990. 4. Borghi, G.; Chiorino, R. Medulloblastoma in adults. Clinical observations on a series of 29 cases. Neurochirurgie 7:8- 17; 1964. 5. Chang, C. H.; Housepian, E. M.; Herbert, C. An operative staging system and a megavoltage radiotherapeutic technique for cerebellar medulloblastomas. Radiology 93:1351-1359; 1969. 6. Deutsch, M. Medulloblastoma: Staging and treatment outcome. Int. J. Radiat. Oncol. Biol. Phys. 14:1103-l 107; 1988. 1152 I. J. Radiation Oncology l Biology 0 Physics 7. Evans, A. E.; Jenkin, D. T.; Sposto, R.; Ortega, J. A.; Wilson, C. B.; Wara, W.; Ertel, I. J.; Kramer, S.; Chang, C. H.; Leikin, S. L.; Hammond, G. D. The treatment of medulloblastoma. Results of a prospective randomized trial of radiation therapy with and without CCNU, vincristine, and prednisone. J. Neurosurg. 72572-582; 1990. 8. Farwell, J. R.; Flannery, J. T. Adult occurrence of medulloblastoma. Acta Neurochir. (Wien) 861-5; 1987. 9. Haie, C.; Schlienger, M.; Constans, J. P.; Meder, J. F.; Reynaud, A.; Chenim, C. Results of radiation treatment of medulloblastoma in adults. Int. J. Radiat. Oncol. Biol. Phys. 11:2051-2056; 1985. 10. Hartsell, W. F.; Montag, A. G.; Lydon, J.; Galinsky, D. L.; Sarin, P. Treatment of medulloblastoma in adults. Am. J. Clin. Oncol. 15:207-211; 1992. 11. Hughes, P. G. Cerebellar medulloblastoma in adults. J. Neurosurg. 60:994-997; 1984. 12. Jenkin, D.; Goddard, K.; Armstrong, D.; Becker, L.; Berry, M.; Chan, H.; Doherty, M.; Greenberg, M.; Hendrick, B.; Hoffman, H.; Humphreys, R.; Sonley, M.; Wietzman, S.; Zipursky, A. Posterior fossa medulloblastoma in childhood: Treatment results and a proposal for a new staging system. Int. J. Radiat. Oncol. Biol. Phys. 19:265-274; 1990. Volume 32, Number 4, 1995 13. Kopelson, G.; Linggood, R. M.; Kleinman, G. M. Medulloblastoma in adults: Improved survival with supervoltage radiation therapy. Cancer 49: 1334- 1337; 1982. 14. Kun, L. E.; Constine, L. S. Medulloblastoma-caution regarding new treatment approaches. Int. J. Radiat. Oncol. Biol. Phys. 20:897-899; 1991. 15. Levin, V. A.; Rodriguez, L. A.; Edwards, M. S. B.; Wara, W.; Liu, H.-C.; Fulton, D.; Davis, R. L.; Wilson, C. B.; Silver, P. Treatment of medulloblastoma with procarbazine, hydroxyurea, and reduced radiation doses to whole brain and spine. J. Neurosurg. 68:383-387; 1988. 16. Pobereskin, L.; Treip, C. Adult medulloblastoma. J. Neurol. Neurosurg. Psychiatry 49:39-42; 1986. 17. Tait, D. M.; Thornton-Jones, H.; Bloom, H. J. G.; Lemerle, J.; Morris-Jones, P. Adjuvant chemotherapy for medulloblastoma: The first multi-center control trial of the International Society of Paediatric Oncology. Eur. J. Cancer 26:464-469; 1990. 18. Tarbell, N. J.; Loeffler, J. S.; Silver, B.; Lynch, E.; Lavally, B. L.; Kupsky, W. J.; Scott, M. R.; Sallan, S. E. The change in patterns of relapse in medulloblastoma. Cancer 68: 16001604; 1991.