Int. J. Radiation Oncology Biol. Phys., Vol. 30, No. 3, pp. 551-556, 1994 Copyright 0 1994 Elscvier Science Ltd Printed in the USA. All rights reserved
0360-3016/94 $6.00 + .OO
??Clinical Original Contribution
PATTERN OF RECURRENCE OF MEDULLOBLASTOMA AFTER LOW-DOSE CRANIOSPINAL RADIOTHERAPY
WILLIAM M. WARA, M.D.,* QUYNH-THU X. LE, M.D.,* PENNY K. SNEED, M.D.,* DAVID A. LARSON, M.D.,* MICHAEL D. PRADOS, M.D.,+ VICTOR A. LEVIN, M.D.,$
MICHAEL S. B. EDWARDS, M.D.+ AND MICHAEL D. WEIL, M.D.*
*Departments of Radiation Oncology and +Neurologic Surgery, University of California, San Francisco, CA 94 143 *Department of Neuro-Oncology, M. D. Anderson Cancer Center, Houston, TX 77030
Purpose: We retrospectively evaluated relapse of medullobhrstoma after low- or high-dose craniospinal radiotherapy, and after conventional or hyperfractionated posterior fossa irradiation. Methods and Materials: Ninety-two pediatric patients were treated postoperatively since 1970 at the University of California, San Francisco. Until 1989, we employed conventional fractionation with low (I 30 Gy) or high-dose craniospinal fields and low-dose (I 56 Gy) posterior fossa boosts. Recently, hyperfractionation delivered low- or high-dose to the craniospinal axis and high-dose to the posterior fossa. Most patients treated after 1979 received chemotherapy. Results: Median follow-up was 70 months. Five-year disease-free survival was 36% (22% for poor-risk vs. 59% for good-risk patients). Five-year overall survival was 52% (43% for poor vs. 68% for good-risk). Neither the dose to the posterior fossa nor the craniospinal axis was statistically related to recurrence. Failure in the posterior fossa occurred despite boosts greater than 56 Gy. Females, over the age of 6 years, had significantly better relapse-free survival than males of the same age. Six of the 54 patients who relapsed were long-term survivors. Conclusions: Low-dose craniospinal radiotherapy, where the majority of patients received chemotherapy, was not associated with increased failure. High-dose posterior fossa hyperfractionation did not improve control. Long-term survival was noted in a number of patients after relapse. We recommend 60 Gy or greater with conventional fractions to the primary area, and continued study of low-dose craniospinal irradiation with adjuvant chemotherapy.
Medulloblastoma, Craniospinal, Radiotherapy, Radiotherapy dosage, Hyperfractionation, Recurrence.
Reported survival for medulloblastoma has improved over the past several decades. Bloom reported a 32% survival rate with radiotherapy for a period between 1950 and 1964, during which time the operative mortality rate was 25% (1). Trials by the International Society of Pediatric Oncology (SIOP) and the Childrens Cancer Study Group (CCSG), which closed in 1979 and 198 1, respectively, added chemotherapy to high-dose radiotherapy of the posterior fossa and craniospinal axis (6, 14). Five-year survival was 53% and 65% with no statistically significant difference in survival between patients treated after surgery with radiotherapy alone or radiotherapy plus chemother- apy. However, Packer showed a significant improvement in survival in patients treated after 1982 with platinum-
based chemotherapy and radiotherapy ( 12). In those pa- tients with a median follow-up of 40 months, a five-year actuarial disease-free survival of 82% overall was reported (88% in poor-risk patients). Halberg most recently updated the results for treatment of adult and pediatric medullo- blastoma at the University of California, San Francisco (UCSF) (8). There was no difference in the rate of recur- rence in the craniospinal axis after reducing the dose to this area. Five-year actuarial survival for all patients was 73%.
We performed a retrospective analysis of the pattern of relapse of pediatric medulloblastoma patients in an attempt to discern a benefit of high-dose hyperfrac- tionated radiotherapy to the posterior fossa, and to reassess low-dose radiotherapy to the craniospinal axis. Most of the patients received adjuvant chemotherapy. The sites
Presented at the 35th Annual Meeting of ASTRO, New Or- leans, LA, October 1 l-l 5, 1993.
Reprint requests to: Michael D. Weil, M.D., Department of Radiation Oncology, Room L-75 (Box 0226), University ofCal- ifornia, San Francisco, CA 94 143-0226.
Acknowledgements-The authors thank Lea Conlu and Marion Thornton for manuscript preparation.
Accepted for publication 6 May 1994.
552 I. J. Radiation Oncology 0 Biology 0 Physics Volume 30, Number 3, 1994
of failure were categorized in relation to the extent of disease and the intensity of the dose of radiation admin- istered.
METHODS AND MATERIALS
We evaluated all patients with primary medulloblas- toma, I 21 years of age, treated with postoperative ra- diotherapy at UCSF between 1970 and 1992. Extent of surgical resection was assessed by the surgeons report and, more recently, by examination with magnetic resonance imaging (MRI). All patients had a pathologic diagnosis of medulloblastoma from UCSF. Most patients underwent spinal evaluation with myelogram (4), examination of the cerebrospinal fluid (CSF), MRI, or a combination of these tests. Risk stage was assigned based on the age at diagnosis, the type of surgical procedure (gross total resection vs. biopsy or subtotal resection), and evidence of tumor be- yond the primary area. We did not attempt to subdivide patients by the degree of subtotal resection. Patients were assigned to a good risk category if they were greater than 3 years of age, had undergone a gross total resection, and had negative evaluation of the craniospinal axis. Poor- risk patients were considered to be children 3 years old or less, or those who had undergone a subtotal resection, or patients found to have evidence of microscopic or gross disease outside of the posterior fossa.
diation at reduced craniospinal doses; 36 Gy to the whole brain and 25 Gy to the spine (Brain Tumor Research Center protocol 7922) (10). Between 1984 and 1992, postoperative chemotherapy was procarbazine, 1-(2- chloroethyl)-3-cyclohexyl- 1 -nitrosourea (CCNU), vin- cristine, 6-thioguanine (6-TG), and dibromodulcitol, fol- lowed by radiotherapy with hydroxyurea (Brain Tumor Research Center protocol 8422). Patients received 54 Gy to the posterior fossa, and 24 Gy to the craniospinal axis. After 1990, most poor-risk patients, over 3 years old, were treated with hyperfractionated radiotherapy followed by cisplatinum, CCNU, and vincristine (Brain Tumor Re- search Center protocol 9023). Patients were treated with 1 Gy twice daily, and received 72 Gy to the posterior fossa, and 30 Gy to the craniospinal axis.
Tumor volumes were localized on the simulation films with the aid of computed tomography (CT), beginning in the mid- 1970s and later with MR imaging. All patients were treated with linear accelerators using 4 or 6 MV photons. The brain was treated with opposed lateral fields and, except for several infants, patients were treated prone. The spinal field was treated directly from the posterior to the level of S2-S3 without a spade flair at the lower border of the portal. Divergence between the brain and spinal fields was accounted for by appropriate rotation of the collimator and angling of the treatment table towards the lateral field. The match line between the fields was shifted 1 cm every 10 Gy of treatment. Until 1989, ra- diation was delivered to the posterior fossa in single daily fractions of 1.8 Gy to 56 Gy or less (low dose). Since 1989, patients were treated at 1 Gy twice daily to a total dose of 72 Gy to the posterior fossa (high dose). Patients treated once a day received an average dose of 36 Gy to the craniospinal axis and those treated twice daily received 24-40 Gy. Six infants received radiation to the posterior fossa only.
No patients were lost to follow-up. After treatment, patients were followed with regular examination of the primary site and craniospinal axis using CT or MR scans and CSF cytology. Disease-free survival was measured from the start of radiotherapy until relapse; patients who expired without evidence of disease at the time of death were censored. Duration of survival was measured from the start of radiotherapy until last follow-up or death. Es- timates of disease-free survival and survival were com- puted using the method of Kaplan and Meier (9). Differ- ences among patient groups were evaluated by the Mantel- Haenzel log rank test (11). Univariate and multivariate analyses were performed with the Cox proportional haz- ards regression model (3) using statistical software.
Combinations of chemotherapeutic agents changed over time. Prior to 1979, adjuvant therapy was radiation only; 54 Gy to the posterior fossa, 40 Gy to the brain, and 36 Gy to the spine. Subsequently, procarbazine was given postoperatively, followed by hydroxyurea plus ra-
Ninety-two patients were treated from 1970- 1992 (Ta- ble 1). The mean age of the patients was 8 years and 27% were 3 years of age or less. Fifty patients were male and 42 were female. Eighty-seven percent had evaluation of the spine with a myelogram, MRI, and/or examination of the spinal fluid. Both radiographic and cytological studies of the craniospinal axis were performed in 69 of the 92 patients. Gross total resection was performed on over one-half of the group (53%), while 36% underwent a subtotal resection, and 11% had a biopsy only. Thirty- five of the 92 patients were assigned to the good-risk group and 57 to the poor-risk group. There were eight patients assigned to the poor-risk category solely on the basis of their age being 3 years old or less at the time of diagnosis. No significant change, in the distribution of good vs. poor- risk patients over the different treatment eras, was noted. Twenty-three patients, treated between 1970 and 1979, received postoperative radiation only. Subsequently, ad- juvant chemotherapy was used in 46 poor-risk patients and in 15 good-risk patients. Seventeen patients in the early 1980s were treated with procarbazine, followed by
Stata software (Computing Resource Center, Santa Monica, CA).
Medulloblastoma recurrence after craniospinal radiotherapy 0 W. M. WARA et a/. 553
Table 1. Patient characteristics
Age: 11 months-21 years (median 8 years) (27% 5 3 year-old)
Sex: 50 male/42 female Cranio-spinal 87% had MRI or myelogram or CSF
MRI = magnetic resonance imaging. CSF = cerebrospinal fluid.
hydroxyurea plus radiation at reduced craniospinal doses. Between 1984 and 1990, there were 27 patients who received postoperative chemotherapy (procarbazine, CCNU, vincristine, 6-TG, and dibromodulcitol), followed by radiotherapy with hydroxyurea. After 1990, most poor- risk patients were treated with hyperfractionated radio- therapy, followed by cisplatinum, CCNU, and vincristine.
Treatment of the posterior fossa with I 56 Gy was defined as low-dose therapy and > 56 Gy was consid- ered high dose. Radiotherapy doses ranged from 22- 56 Gy to the posterior fossa, O-54 Gy to the brain, and O-50 Gy to the spine for standard fractionation. Hyper- fractionation doses ranged from 60-72 Gy to the posterior fossa and 24-40 Gy to the craniospinal axis. Conventional fractionation given to the posterior fossa was, therefore, low dose, and the hyperfractionated posterior fossa treat- ments were high dose. Low (I 30 Gy) and high-dose (> 30 Gy) therapy to the craniospinal axis used conven- tional or hyperfractionated regimens. The median low dose was 54 Gy to the posterior fossa (with conventional fractions) and 25 Gy to the craniospinal axis. The median high-dose was 70 Gy to the posterior fossa ( via hyper- fractionation) and 40 Gy to the craniospinal axis. In pa- tients without evidence of relapse, the mean dose to the posterior fossa was 57.7 Gy. In patients who failed, the mean dose was 56.18 Gy. Mean dose to the brain in pa- tients who failed in that area was 28.16 vs. 3 1.60 Gy in patients who were without brain relapse. Patients who were disease-free in the spinal cord received a mean dose of 27.50 Gy as opposed to a mean dose of approximately 32.00 Gy for those who failed in the spine.
The median follow-up was 70 months and the median time to relapse was 22 months. The actuarial 5-year dis- ease-free survival was 36% (59% for good-risk patients and 22% for poor-risk patients: p = 0.002). There were relapses in 54 patients; 15 of these patients were considered good risk, and 39 were considered poor risk. Median sur- vival was 69 months. Five-year survival was 68% for the good-risk patients vs. 43% for the poor-risk patients (p = 0.015), and 52% overall (Fig. 1). Of the 54 patients who failed, 6 have survived for at least 5 years after relapse. Actuarial analysis of 5-year survival after recurrence was 25%.
Prognostic factors that were significant for disease-free survival by univariate analysis included stage (good risk vs. poor risk) (p = 0.002), sex (p = O.OlO), extent of sur- gical resection (p = 0.017), and whether or not adjuvant
chemotherapy was given (p = 0.046). Female gender and greater extent of resection were associated with improved disease-free survival. Of note, the use of chemotherapy was associated with decreased disease-free survival, but chemotherapy was more commonly given to poor-risk patients. Multivariate analysis, including stage, gender, and use of adjuvant chemotherapy, showed that stage and gender were significant determinants of disease-free sur- vival (p < 0.01 and a = 0.040, respectively) and che- motherapy was not. Thirty-six of 50 (72%) boys have re- lapsed and 18 of 42 (43%) girls have relapsed.
Prognostic factors that were significant for overall sur- vival by univariate analysis included stage (good risk vs. poor risk) (p = 0.015), sex (p = 0.013), and extent of surgical resection (p = 0.00 1). A multivariate analysis in- cluding stage and gender yielded p = 0.023 for gender and p = 0.027 for stage. When the analysis was performed including gender and extent of surgical resection, gender remained a significant determinant of survival (p = 0.040 for gender and p = 0.002 for surgery). Females over 6 years of age had a highly-significant survival advantage (p < 0.00 1): 82% for girls compared to 36% 5-year survival for boys. There was no significant difference in relapse rate or survival detected over the varying diagnostic and treatment eras.
We evaluated the sites of failure to try to correlate treatment with relapse patterns. In the good-risk group, only single sites of failure were noted. This is in distinction to the poor-risk group where patients relapsed concur- rently in multiple areas (Fig. 2). Multiple sites of relapse occurred in 11 of 39 (28%) poor-risk failures. Overall, there were 28 total failures in the posterior fossa, 18 of which were single failures (Table 2). Nineteen patients failed in the brain and 11...