- Email: [email protected]
Medulloblastoma: Long-term results for patients treated with definitive radiation therapy during the computed tomography era
Int. J. Radiation
Oncology
Biol.
Phys.. Vol. 36, No. I. pp. 29-3.5, 1996 Copyright 0 1996 Elsevier Science Inc. Printed 10 the USA. ALI rights reserved 0360.3016/96 $15.00 + .OO
PII: SO360-3016(96)00274-X
0 Clinical
Original
Contribution
MEDULLOBLASTOMA: WITH DEFINITIVE
LONG-TERM RESULTS FOR PATIENTS TREATED RADIATION THERAPY DURING THE COMPUTED TOMOGRAPHY ERA
THOMAS E. MERCHANT, D.O., PH.D.,* MING-HSIEN WANG, B .S .,* TONI HAIDA, B.A.,* KAREN L. LINDSLEY, M.D.,* JONATHAN FINLAY, M.B., CH.B.,’ IRA J. DUNKEL, M.D.,+ MARC K. ROSENBLUM, M.D.’ AND STEVEN A. LEIBEL, M.D.* *Departments
of Radiation
Oncology,
‘Pediatrics, and zPathology, Memorial New York, NY
Sloan-Kettering
Cancer Center,
Purpose: We performed a retrospective evaluation of the patterns of failure and outcome for medulloblastoma patients treated with craniospinal irradiation therapy during the conputed tomography (CT) era. Materials and Methods: The records of 100 patients treated at Memorial Sloan-Kettering Cancer Center between 1979 and 1994 were reviewed. CT scans or magnetic resonance imaging were used to guide surgical intervention and evaluate the extent of resection postoperatively. All patients were treated with conventional fractionation (1.8 Gy/day) and the majority received full-dose neuraxis radiation therapy and > 50 Gy to the primary site. Results: With a median follow-up of 100 months, the median, S-year, and IO-year actuarial overall survival for the entire group were 58 months, 50%) and 25 % , respectively. The median, 5- and lo-year actuarial disease-free survivals were 37 months, 41%, and 27%, respectively. Patients with localized disease (no evidence of disease beyond the primary site) had significantly improved overall @ < 0.02) and disease-free (p < 0.02) survivals compared to those with nonlocalized disease. For patients with localized disease, the 5- and lo-year overall survival rates were 59% and 31%) whereas the disease-free survivals were 49% and 31%) respectively. Diseasefree and overall survivals at similar intervals for patients with nonlocalized disease were 29% and 30% (5 years), and 29% and 20% (10 years), respectively. Sixty-four of 100 patients failed treatment. Local failure as any component of first failure occurred in 35% of patients or 55% (35 of 64) of all failures and as the only site of first failure in 14% or 22% (14 of 64) of all failures. For patients presenting with localized disease (n = 68), local failure as any component of first failure occurred in 32% (22 of 68) and in 18% (12 of 68) as the only site. A multivariate analysis showed that M stage was the only prognostic factor to influence overall survival. For diseasefree survival, M stage and the extent of resection were prognostic factors. Ventriculoperitoneal shunting and the use of chemotherapy were associated with a poor outcome; however, these results were confounded by the positive impact of chemotherapy in decreasing the risk of extraneural metastases and the use of these therapies in the more advanced patients. Conclusion: These long-term follow-up data represent one of the largest series of patients with complete followup who were treated with a consistent radiation therapy treatment policy during the CT era. Local failure in patients with localized disease, the persistent risk of late failures, treatment-related toxicity, and the ever-present risk of secondary malignancies demonstrate the limitations of standard therapies. Strategies used to increase the total dose to the primary site should be pursued along with other adjuvant therapies such as intensive chemotherapy.
INTRODUCTION Medulloblastoma remains one of the most intriguing tumors studied and treated by the multidisciplinary oncology team because of its unique clinical and pathologic characteristics and apparent radiocurability. Standard therapy consists of maximal resection compatible with good neurologic outcome and postoperative craniospinal radiation therapy with a boost to the primary site of disease in the posterior fossa using con-
ventional fractionation schemes. Although the treatment volumes and doses have changed little during the past 40 years, the treatment of medulloblastoma continues to evolve. Radiation doses are altered based on the age of the patient and the perceived risk for recurrence. Likewise, chemotherapy is used based on the age of the patient and the presence of factors prognostic for a high risk of failure, whether distant or local. The specific role of chemotherapy and the optimal regimen remain unclear.
Reprint requests to: Thomas E. Merchant, D.O., Ph.D., Department of Radiation Oncology. St. Jude Children’s Research
Hospital, 332 North Lauderdale, Memphis, Accepted for publication 18 May 1996.
29
TN 38105.2794.
30
I. J. Radiation Oncology
0 Biology
0 Physics
Table 1. Clinical characteristics for 100 medulloblastoma patients treated from 1979 to 1994 Clinical characteristics Age Sex (M:F) T stage Tl T2 T3 T4 CSF fluid cytology Positive Negative Unknown M stage (Chang) MO Ml M2 M3 Mx Initial imaging study CT MRI Spine imaging Myelogram MRI Unknown
n
%
14 yr (mean) (53:47)
(l-35 yr, range) (53%:47%)
3 39 47 11
3% 39% 47% 11%
25 70 5
25% 70% 5%
68 12 2 13 5
68% 12% 2% 13% 5%
80 20
80% 20%
76 18 6
76% 18% 6%
CSF = cerebrospinal fluid; CT = computed tomography; MRI = magnetic resonance imaging.
Alterations have been made in the treatment of medulloblastoma and new regimens have been recommended without the benefit of long-term follow-up data for patients treated with standard radiation therapy technique in the era of modern medical imaging and neurosurgical technique. To this end we have compiled complete, long-term follow-up data on 100 patients treated with conventional radiation therapy at our institution during the CT era. The clinical-pathologic characteristics and outcomes of these patients, all of whom were treated with radiation therapy in a similar manner, form
the basis of this report.
MATERIALS
AND METHODS
Patient data The records of 100 evaluable patients treated at Memorial Sloan-Kettering Cancer Center between 1 January 1 1979 and 31 December 31 1994 were reviewed. The information obtained from these records included the following patient and treatment characteristics: date of birth; sex; date and age at presentation; initial imaging study; cerebrospinal fluid (CSF) analysis; imaging study of spine; extent of resection based on operative reports and postoperative imaging; initial use of ventriculoperitoneal shunt; use and sequencingof radiation and chemotherapy; radiation dosesto the neuraxis, whole-brain, and posterior
Volume 36, Number 1, 1996
fossa; and additional use of surgery, radiation therapy, or chemotherapy in failure. The frequencies of the categoric variables and the mean values of the continuous variables are presented in Table 1 and Table 2. Surgery A total of 97 patients underwent an attempted resection and 3 patients were only biopsied. Resections were performed to the extent compatible with good neurologic outcome. A gross-total resection was achieved in 43% and a subtotal resection in 54% basedon the surgeon’soperative impression and postoperative neuroimaging. Ventriculoperitoneal shunting was performed as an initial part of the management in 40% of the cases.
Pretreatment
evaluation
The initial evaluation included X-ray CT scanning or magnetic resonance imaging (MRI). For patients treated in our study, 80% had a CT performed following the onset of symptoms and the remainder had MRI. Contrast myelography was performed in 76% of patients, spinal MRI in 18%, and the remaining 6% either did not have imaging of spine or were unable to undergo the procedure. CSF cytology was performed in 95% of patients. Staging was performed according to the Chang classification (4). The T stage for each patient was determined and the M stage was evaluable for 95% of the patients. Because of the
Table 2. Treatment characteristics for 100 medulloblastoma patients treated from 1979 to 1994 Treatment
characteristics
Surgical Extent of resection GTR STR Biopsy Shunt Radiation therapy Sequencing Initial Postchemotherapy Radiation therapy dosages Primary total dose (mean) Whole-brain dose (mean) Posterior-fossa boost (mean) Spine dose (mean) Special radiation techniques Cribriform plate Spinal electron treatment Chemotherapy Sequencing Adjuvant At recurrence Not given Treatment in failure Additional surgery Reirradiation
n
-
%
43 54 3 40
43% 54% 3% 40%
97 3
97% 3%
53.2 35.9 17.3 33.9
Gy Gy Gy Gy
48-58 Gy 30.6-50.4 Gy 4-25.0 Gy 30.6-45 Gy
17 13
17% 13%
49 35 16
48% 36% 16%
20 18
20% 18%
GTR = gross total resection; STR = sub total resection.
Medulloblastoma:
Long-term results
ambiguity of the staging system, T staging was not included as a variable in the analyses of prognostic faCtors (7). Radiation therapy Patients were treated with neuraxis radiation therapy followed by a boost to the primary site to achieve a total dose of at least 50 Gy. The craniospinal radiation therapy technique was similar for all patients, and the posterior fossa boost was delivered via parallel-opposed photon beams in most cases.Recently, the posterior fossa boost has been carried out using a CT-based, multifield, conformal approach. Patients were treated using 6oCoor a 6MV linear accelerator. In 13% of patients electrons were used to treat the spine. and in 17% an additional boost to the cribriform plate region was performed. Chemotherapy Only 16% of the patients received no chemotherapy. A total of 49% received adjuvant chemotherapy, whereas 35% received chemotherapy at recurrence. The regimens and dosagesused varied considerably over the time period of the study. Dejkitions and analysis of disease-free and overall survival Overall survival and disease-free survival were measured from the date of pathologic diagnosis.Failures were determined by site, including the primary site (posteriorfossa). the nonposterior fossa brain, the spine, meninges (widespread leptomeningeal disease), and extraneural sites (primarily bone and thorax). Failure included recurrence or progression of diseasein sites identified to have active diseaseat presentation or relapse in a site free of diseaseat presentation. The time of failure was determined as the date at which radiologic or biopsy confirmation of relapse at one of these sites occurred. Failure at these sites was determined either as the only site of first failure or as a component of first failure. Extraneural relapse occurred in 9% of patients (8 in bone and 1 in the thorax) or in 14% (9 of 64) of patients who failed treatment. Deaths due to other causes (n = l), treatment-related deaths (brain necrosis [ir = 21, sepsis[n = 11, pneumonia [n = I]), and the development of second malignancies (n = 3) were included in the survival analysis. These seven patients were not analyzed as treatment failures. No patients were lost to follow-up. Statistical methods Using Cox and logistic regression,the clinical and treatment variables were entered in a forward, step-wise manner using p < 0.05 as criterion for inclusion in the model. For evaluation of end points of disease-free and overall survival, Kaplan-Meier methodology was usedwith Wilcoxon-Breslow test statistics (5, 8). The median followup was 100 months.
0
T. E.
MERCHANT
31
ef al.
RESULTS Clinical characteristics The median age of the 100 patients in the study was 14 years (range l-35). They were evenly divided between males (53%) and females (47%). Gross total resection was performed in 43%, subtotal resection in 54%, and biopsy only in 3% of patients. According to the staging system of Chang et al. (4), the distribution of T stagesincluded 3% Tl, 39% T2, 47% T3, and 11% T4. Sixty-eight patients (68%) were determined to be MO, 12% M 1,2% M2, 13% M3, and 5% Mx. The total percentage of patients with nonlocalized disease(i.e., M 1, M2, or M3) was 27%. These data are detailed in Table 1. Treatment characteristics In addition to their extent of resection asoutlined above, the patients were further characterized basedon their treatment profiles. Forty percent required ventriculoperitoneal shunting as a part of their initial therapy. Radiation therapy was given in the immediate postoperative period in 97% of casesand after someform of preradiation adjuvant chemotherapy in only 3%. The mean dosesto the primary site (53.2 Gy). whole-brain (35.9 Gy), and spine (33.9 Gy) were determined. Whereas 16% of the patients received no chemotherapy, 49% received adjuvant chemotherapy and 35% were treated with chemotherapy at the time of recurrence. In failure, 20% of patients underwent additional resection and 18% received palliative radiation therapy to the site of recurrence. These data are described in Table 2. Survivalfor patients with localized and nonlocalized diseaseand other prognostic factors With a median follow-up of 100 months, the median, 5-year, and lo-year actuarial overall survivals for the entire group are 58 months, 50%, and 25%, respectively (Fig. 1). The median (85 vs. 44 months), 5-year (59% vs. 30%), and lo-year (30% vs. 20%) overall survivals are significantly longer @ < 0.02) for patients with localized
Overall Entire
__
Survival Group
~
- --~
_~_~~
100
150
550
Months
Fig. 1. Overall survival (actuarial)for 100medulloblastoma patients treated with definitive radiation therapy from 1979 to 1994.
32
I. J. Radiation Oncology l Biology l Physics % Overall
Survival
MO (n=66)
vs Ml
(n=27)
Fig. 2. Overall survival (actuarial) for 100 medulloblastoma patients treated with definitive radiation therapy from 1979 to 1994. Localized disease (MO) patients have a significantly longer overall survival (p < 0.02, log rank) than those with nonlocalized (M+) disease.
disease(MO) as compared to those with nonlocalized disease (Fig. 2). By subgroup, the median survival of Ml, M2, and M3 patients is, 71, 26, and 19 months, respectively (Table 3). Prognostic factors influencing overall survival by univariate analysis (improved survival vs. decreasedsurvival) included spinal radiation modality (electrons vs. photons, p < 0.04) and the use of chemotherapy (none vs. any, p < 0.004; adjuvant vs. recurrent, p < 0.005; none vs. adjuvant, p < 0.07). By multivariate analysis, only cytology (negative vs. positive, p < 0.002) and initial ventriculoperitoneal shunting (no shunt vs. shunt, p < 0.01) influenced (improved vs. decreased)overall survival. Actuarial survival
curves were compared
(improved
survival
Volume 36, Number 1, 1996
original MO patients and as a component of failure in 32% (22 of 68). Prognostic factors influencing relapse-free survival by univariate analysis included cytology (negative vs. positive, p < O.Ol>,extent of resection (grosstotal vs. subtotal, p < 0.02) spinal radiation modality (electrons vs. photons, p < 0.04) initial ventriculoperitoneal shunting (no shunting vs. shunting, p < 0.04) and useof chemotherapy (none vs. any, p < 0.001; adjuvant vs. recurrent, p < 0.007; none vs. adjuvant, p < 0.07; and no adjuvant vs. adjuvant, p < 0.07). By multivariate analysis, factors influencing disease-free survival included cytology (negative vs. positive,p < O.OOl), shunting (no shunt vs. shunt), and use of chemotherapy (none vs. any, p < 0.002; adjuvant vs. recurrent, p < 0.03; and none vs. adjuvant, p < 0.03). Logistic regression was performed to determine which categoric variables influenced failure at the primary site, in the spine, in the meninges, or at extraneural sites. Positive CSF cytology at presentation predicted for failure in the spine 0, < 0.01) and meninges 0, < 0.04) as a component of failure. The patient receiving chemotherapy. whether
adjuvant
or for recurrence,
Table 3. Outcomes for 100 medulloblastoma from 1979 to 1994
vs. de-
creased survival) for patients based on M stage (p < O.OOl), cytology @ < 0.007), and chemotherapy (none vs. any chemotherapy, p < 0.04; and none vs. adjuvant chemotherapy, p < 0.04).
Patterns of failure and disease-free survival The median time to failure was 37 months for the entire group. Patients with MO diseasehad a significantly longer median time to failure than those with M+ disease: 58 compared with 35 months @ < 0.02). First failure is listed in Table 3 as a sole site of failure or as a component of failure. First failure may represent progression of disease or recurrence in the primary site or in a site of dissemination determined at diagnosis. As the only site of first failure among the 64 patients who failed therapy, failure occurred in the primary site in 22% (14 of 64), nonposterior fossa brain 3% (2 of 64), spine 13% (8 of 64), meninges 9% (6 of 64), and extraneural 9% (6 of 64). As any component of failure, failure in the primary site occurred in 55% (35 of 64) brain 26% (17 of 64), spine 38% (24 of 64), meninges 36% (23 of 64), and extraneural 14% (9 of 64). Local failure occurred in only 18% (12) of the 68
was most
likely
to
have failure at the primary site 0, < 0.02). Patients receiving chemotherapy for recurrence were more likely to have had failures in the spine (p < 0.05) or at extraneural sites(p < 0.01). Patients receiving adjuvant chemotherapy compared with no chemotherapy were more likely to fail at the primary site 0, < 0.04). Finally, patients receiving
Site of first failure Posterior fossa Non-osterior fossa Spine Meninges Extraneural
Only site of failure 14 (16%) 2 8 6 6
(2%) (8%) (6%) (6%)
35 (35%) 17 24 23 9
(15%) (16%) (17%) (3%)
%
(actuarial)
1 7 5-year
f/u = follow-up.
21 (21%) 15 16 17 3
34% 3%
Survival
M+
Any failure
n
55
p < 0.02 (log rank)
Component of failure
34 3
Disease status (70 mo f/u) No evidence of disease Alive with disease Dead from other causes Treatment related death Dead of disease
Disease-free Entire group MO M+ p < 0.02 (log rank) Overall (actuarial) Entire group MO
patients treated
41% 49%
(17%) (24%) (23%) (9%)
1% 7% 55%
lo-year (actuarial)
Median
29%
27% 31% 29%
37 58 35
50% 59% 30%
25% 30% 20%
85 44
mo mo mo
58 mo mo mo
Medulloblastoma:
Long-term
adjuvant chemotherapy were less likely to fail at extraneural sites than were those who did not receive adjuvant chemotherapy @ < 0.04). Actuarial curves were compared for patients divided among the clinical and treatment variables; significant differences were found for cytology 0, < 0.008) and M stage (p < 0.001). DISCUSSION This study provides a useful look at the long-term results of patients treated in the modem era with definitive radiation therapy. Each one of several issues-length of follow-up, treatment era, and radiation therapy techniques and doses-must be carefully considered in the interpretation of historic and recent series. Many of the largest reported series to date describe results associated with short-term follow-up, data which span several decades, multi-institution studies, and patients treated with a variety of radiation therapy and chemotherapy regimens. Table 4 provides an overview of the largest series of medulloblastoma patients for interstudy comparison. The poor survival results at 10 years There are few series with sufficient length of follow-up to determine the survival at 10 years (2, 6-8, 14-16). Although many series, including our own, report good survival rates at 5 years, there is a continuous rate of relapse and mortality beyond 5 years. Thus, our poor survival results are not singular. It appears that the early failures can be attributed primarily to the poor-prognosis patients (M+), and that late failures are attributable to good-prognosis patients (MO). Our data showing an actuarial survival of 25% at 10 years for the entire group and 30% for the MO patients are much worse than the 63% lo-year
Table 4. Results from large, contemporary Series (date) details Evans et nl. (1990)
Reference 6
results
0 T. E. MERCHANT
survival reported by Jenkin et al. (8) and the 45% IO-year survival reported by Tait et al. (15) both contemporary series. The comparison of these series is made difficult because of a lack of staging information in the former study (only 15 of 72 patients underwent any form of spinal imaging), the lack of CSF cytology data, and the lack of follow-up information in the latter study (15% of patients were lost to follow-up at 5 years and 66% at 10 years). Similar problems are encountered when comparing our patients to an earlier series from the Princess Margaret Hospital (2), where a 43% lo-year survival was reported for 122 patients treated from 1958 through 1978. In these patients, neither CSF cytology nor spinal imaging was routinely analyzed or performed. Comparing our series to that of the Royal Marsden (3) (patients treated from 1950 to 1964), their reported survival of 26% at 10 years demonstrates that there has been no improvement in more than 30 years of craniospinal radiation therapy. Local failure as a component of failure Of 100 patients, 64 failed therapy. These failures are represented as either progression of disease or recurrences at or beyond sites of disease that were once in remission. One would expect that the greatest risk for recurrence would be for sites of preexisting disease and that failure would also occur as a result of the inability of treatment to eradicate residual or subclinical disease. Our data and those of others (16) show that the most important site of first failure is the primary site or posterior fossa. Local failure as any component of failure occurred in 35% of all patients or 55% (35 of 64) of all treatment failures. Local failure as the only component of failure occured in 14% of patients, or 22% (14 of 64) of all failures. As the only site of first failure, failure in the posterior fossa occurred in 18% (12 of 68) of patients with localized disease and in only 7% (2
series or series with IO-year survival data employing Study (5 yr)
1975-1981
Hershatter et al. (1986) Jenkin et al. (1990) Stiller and Lennox (1983) Tait et al. (1990)
7 8 14 15
1940- 1983 1977-1987 1971-1977 1975-1979
Tarbell er al (1991) Merchant et al. (1995)
16
1970- 1989 1979- 1994
(I&r) 88 91 127 72 (v) 304 141 145 89 100
33
et nl.
RT follow-up * IB # **
full-dose radiation
Chemotherapy Adjuvant (n = 88) Adjuvant (n = 0) Adjuvant (4/127) Adjuvant (3%) Adjuvant (94/304) Adjuvant (n = 141) Adjuvant (n = 0) Pre-RT (n = 39) Adjuvant (49%)
Entire group
therapy
Median
65%
Dates 5 yr
33% 71% 35% 53%
21% 63% 30% 45%
26 yr 7 yr 9 yr 12yr
65% 50%
48% 25%
9 yr 8 yr
Full-dose radiation therapy is defined as 30-35 Gy to the craniospinal axis and 50-56 Gy to the posterior fossa. * Violations in radiation therapy design (15% < 35 Gy to brain, 26% < 35 Gy to spine). ’ Only 17 received > 52 Gy (PF) and > 30 Gy to spine; 13 received > 40 Gy PF and > 20 Gy to spine. ? Of 72 patients, 37 did not have spinal staging. B Population-based study for which treatment details are available in 94 of 304 patients (posterior fossa > 45 Gy, II = 22. spine > 35 Gy, II = 56). # RT details known for 92% of patients. None excluded for RT protocol violations; 15% of patients lost to follow-up at 5 years, 66% at 9 years. *;rMedian PF dose 54 Gy, and spine dose 29.5 Gy.
34
I. J. Radiation
Oncology
0 Biology
0 Physics
of 27) of patients with nonlocalized disease. This difference may simply be because patients with nonlocalized disease have a shorter median time to recurrence and dissemination of already nonlocalized disease. The overall survival is poor for the nonlocalized group; thus, these patients lack sufficient time to have disease recur at the primary site. Of the 12 patients with originally localized disease who failed in the posterior fossa, the group was evenly divided between those who had undergone a gross total resection (n = 6) and those who underwent a subtotal resection (n = 6). Collin’s law has previously been regarded as applicable to medulloblastoma patients (3, 7) in which cure is assured if patients survive their age and 9 months beyond diagnosis. In our series, of the 64 patients failing therapy, only 10.9% violated this law and failed beyond the time interval defined as their age plus 9 months. Treatment-related morbidity The toxicity of radiation therapy and chemotherapy alone or in combination cannot be overestimated. In this series, two patients suffered from fatal necrosis of neural tissue both in and out of the high-dose regions of treatment. These women, ages 30 and 35 years, received radiation therapy. They deteriorated neurologically after radiation therapy and died. Autopsy showed diffuse white-matter changes and parenchymal necrosis. Two patients suffered from the acute effects of chemotherapyinduced neutropenia and died of pneumonia (n = 1) and sepsis (n = 1). Three additional patients developed second primary brain tumors, anaplastic astrocytoma (n = l), and glioblastoma multiforme (n = 2); thus, 7% of the patients suffered a treatment-related death if one can attribute the second tumors to radiation therapy. One of the worst scenarios suffered by patients, their families, and the treating physicians is a radiation-induced tumor with malignant propensity exceeding that of medulloblastoma. In our series, three patients developed primary brain tumors other than medulloblastoma after radiation therapy. It is conceivable that two of the second primary tumors (glioblastoma multiforme) were radiation therapy induced with intervals from treatment to presentation of 88 and 107 months. The patient with the shortest interval (12 months) developed an anaplastic astrocytoma. It is probable that other patients will develop second primary tumors and that others would have had they outlived their disease. The rate of second tumor formation is probably underestimated owing to the poor treatment results. Treatment volume, dose, and fractionation Craniospinal treatment fields have been used as the standard treatment approach since the early 1950s (3, 13). Fewer controversies surround the radiation therapy treatment volumes than dose and fractionation schemes. In our series, all patients were treated with a consistent treatment policy that included craniospinal radiation and posterior fossa boost portals similar to those previously described.
Volume
36, Number
1, 1996
Three additional measures were used to treat these patients which are not commonly discussed in the methods of the major series. Additional treatment (boost) to the cribriform plate was performed in 17% of patients, and electrons were used to treat the spinal fields in small children. More recently we have used conformal technology and CT-based treatment planning to minimize the dose to the nonposterior fossa brain. The use of cribriform boost had no impact on the patients treated in this series. The use of electrons for treatment of the spine field caused a positive change. Patients treated to the spine with electrons were less likely to fail and had an improved survival, although the numbers are small. Our data as well as those of others show that the control rates at the primary site are inadequate and that the posterior fossa should probably be treated to a higher dose than the 54-56 Gy which is used at present (17). Conformal technology may permit an escalation of dose without an increase in the acute and late effects of treatment. In this study, all patients were treated with conventional fractionation. At the present time there are a number of cooperative group studies employing hyperfractionated radiation therapy (HFRT) in the treatment of medulloblastoma (1). HFRT, 1.O- 1.2 Gy delivered twice a day, is used with the intention of reducing treatment related late effects while at the same time increasing the total dose to improve local control. To date, no study has demonstrated a benefit for HFRT in the treatment of primary brain tumors, including medulloblastoma. Furthermore, the relative radiosensitivity of medulloblastoma and the methods used to determine radiosensitivity are controversial. Inclusion of patients treated with HFRT in some series confounds the results. However, in a recent report (17), no difference in outcome or toxicity was noted when comparing patients treated with HFRT and CERT. Any benefit derived from escalating the total dose with HFRT may be offset by the lower dose per fraction given with each treatment. The hypothesis that HFRT reduces radiation-induced neurologic sequelae may be difficult to test. Considering that the portion of the brain that is treated to the highest dose (posterior fossa) is that which can tolerate the most treatment with the least amount of sequelae, current methods to determine differences in late effects may lack the sensitivity to demonstrate a difference. Conformal technology may serve the patient in a manner equivalent to hyperfractionation. A retrospective analysis of the effect of chemotherapy on outcome Given the retrospective nature of this study, the use of adjuvant chemotherapy in only 49% of the patients, and the multitude of regimens used in both the adjuvant and recurrent settings, it is difficult to take a stand regarding chemotherapy from our data. Chemotherapy was associated with a poor outcome in our series, probably because the patients perceived to be at greated risk were those who
Medulloblastoma:
Long-term
received chemotherapy in the adjuvant setting and the remainder received chemotherapy for recurrence. There are three published randomized trials comparing radiation therapy to radiation therapy and adjuvant or preirradiation chemotherapy (6, 10, 15). None of these trials or other studies (11, 12) demonstrated a survival benefit except for subgroups of patients with advanced disease. Treatment at failure Sixty-four patients failed treatment or progressed while being treated in this study. The median survival following relapse for these patients is 12 months. The actuarial 5year survival following relapse is a dismal 4%. At the time of analysis of this report, 8 of 64 patients who failed are alive at 12, 30,44, 71, 77, 134, 136, and 152 months. Six patients were MO at diagnosis and two had M3 disease. Five of the six MO failures were salvaged with chemotherapy. The two patients originally diagnosed with M3 disease are alive with disease at 12 and 44 months. The group from UCSF reported an actuarial 25% 5-year sur-
results
0 T. E. MERCHANT
et al.
35
viva1 after relapse, although no specific details were given regarding their salvage therapy (17). In summary, these data show that survival at 5 years is not a reliable indicator of treatment efficacy for patients with medulloblastoma receiving with full-dose radiation therapy. Poor long-term survival, treatmentrelated deaths, deaths from second primary tumors, and extraneural dissemination should be carefully considered when determining treatment options for patients with medulloblastoma. Finally, local relapse as a component of first failure in patients with localized disease at presentation remains a significant problem for patients treated with standard-dose radiation therapy with or without the addition of adjuvant chemotherapy. Selecting the appropriate treatment for patients based on perceived risk factors determined from retrospective studies may not be justified. Strategies to increase the dose to the primary site and more intensive chemotherapy regimens should be considered for all patients.
REFERENCES J. C.; Nirenberg, A.; Donohue, B. Hyperfractionated radiotherapy and adjuvant chemotherapy for high-risk PNET. J. Neurooncol. 12:262; 1992. Berry, M. P.; Jenkin, R. D. T.; Keen, C. W.; Nair, B. D.; Simpson, W. J. Radiation treatment for medulloblastoma. J. Neurosurg. 55:43-51; 1981. Bloom, H. J. G.; Wallace, E. N. K.; Henk, J. M. The treatment and prognosis of medulloblastoma in children. Am. J. Roentgenol. 105:43-62; 1969. Chang, C. H.; Housepian, E. M.; Herbert, C., Jr. An operative staging system and a megavoltage radiotherapeutic technique for cerebellar medulloblastoma. Radiology 93: 135 l1359; 1969. Cox, D. Regression models and life tables. J. R. Stat. Sot. 34: 187-220; 1972. Evans, A. E.; Derek, R.; Jenkin, 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. Results of a prospective randomized trial of radiation therapy with and without CCNU, vincristine, and prednisone. J. Neurosurg. 72:572-582; 1990. Hershatter, B. W.; Halperin, E. C.; Cox, E. B. Medulloblastoma: The Duke experience. Int. J. Radiat. Oncol. Biol. Phys. 12:1829-1837; 1986. Jenkin. D.; Goddard, K.; Armstrong, D.; Becker, L.; Berry, M.: Chan, H.; Doherty, M.; Greenberg, M.; Hendrick, B.; Hoffman, H.; Humphreys. R.; Sonley, M.; Weitzman, 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; ! 990. Kaplan, E. S.; Meier, P. Nonparametric estimations from incomplete observations. Am. Stat. Assoc. J. 53:457-482; 1958. Krischer, J. P.; Ragab, A. H.; Kun, L. Nitrogen mustard, vincristine, procarbazine, and prednisone as adjuvant che-
1. Allen,
2. 3. 4.
5. 6.
7. 8.
9. 10.
11.
12.
13. 14. 15.
motherapy in the treatment of medulloblastoma. A Pediatric Oncology Group study. J. Neurosurg. 74:905-909; 1991. Packer, R. J.; Siegel, K. R.; Sutton, L. N.; Evans, A. E.; D’ Angio, G.; Rorke, L. B.; Buning, G. R.; Schut, L. Efficacy of adjuvant chemotherapy for patients with poor-risk medulloblastoma: A preliminary report. Ann. Neurol. 24:503508; 1988. Packer, R. J.; Sutton, L. N.; Goldwein, J. W.; Perilongo, G.; Bunin, G.; Ryan, J.; Cohen, B. H.; D’Angio, G.; Kramer, E.; Zimmerman, R. A.; Rorke, L. B.; Evans, A. E.; Schut, L. Improved survival with the use of adjuvant chemotherapy in the treatment of medulloblastoma. J. Neurosurg. 74:433440; 1991. Patterson, E.; Farr, R. F. Cerebellar medulloblastoma: Treatment by irradiation of the whole central nervous system. Acta Radiol. 39:323-336; 1953. Stiller, C. A.; Lennox, E. L. Childhood medulloblastoma in Britain 1971-77: Analysis of treatment and survival. Br. J. Cancer 48:835-841; 1983. Tait, D. M.: Thornton-Jones, H.; Bloom, H. J. G.; Lemerle, J.; Morris-Jones, P. Adjuvant chemotherapy for medulloblastoma: The first multi-centre control trial of the intema-
tional societyof pediatriconcology (SIOP 1). Eur. J. Cancer 26:464469; 1990. 16. Tarbell, N. J.; Loeffler, J. S.; Silver, B.; Lynch, E.; Lavally, B. L.; Kupsky, W. J.: Scott, R. M.; Sallan, S. E. The change in patterns of relapse in medulloblastoma. Cancer 68: 16001604; 1991. 17. Wara, W. M.; Le, Q.-T. X.; Sneed, P. K.; Larson, D. A.; Prados, M. D.; Levin, V. A.; Edwards, M. S. B.; Weil, M. D. Pattern of recurrence of medulloblastoma after lowdose craniospinalradiotherapy.Int. J. Radiat. Oncol. Biol.
Phys. 30:55l-556; 1994.
Oncology
Biol.
Phys.. Vol. 36, No. I. pp. 29-3.5, 1996 Copyright 0 1996 Elsevier Science Inc. Printed 10 the USA. ALI rights reserved 0360.3016/96 $15.00 + .OO
PII: SO360-3016(96)00274-X
0 Clinical
Original
Contribution
MEDULLOBLASTOMA: WITH DEFINITIVE
LONG-TERM RESULTS FOR PATIENTS TREATED RADIATION THERAPY DURING THE COMPUTED TOMOGRAPHY ERA
THOMAS E. MERCHANT, D.O., PH.D.,* MING-HSIEN WANG, B .S .,* TONI HAIDA, B.A.,* KAREN L. LINDSLEY, M.D.,* JONATHAN FINLAY, M.B., CH.B.,’ IRA J. DUNKEL, M.D.,+ MARC K. ROSENBLUM, M.D.’ AND STEVEN A. LEIBEL, M.D.* *Departments
of Radiation
Oncology,
‘Pediatrics, and zPathology, Memorial New York, NY
Sloan-Kettering
Cancer Center,
Purpose: We performed a retrospective evaluation of the patterns of failure and outcome for medulloblastoma patients treated with craniospinal irradiation therapy during the conputed tomography (CT) era. Materials and Methods: The records of 100 patients treated at Memorial Sloan-Kettering Cancer Center between 1979 and 1994 were reviewed. CT scans or magnetic resonance imaging were used to guide surgical intervention and evaluate the extent of resection postoperatively. All patients were treated with conventional fractionation (1.8 Gy/day) and the majority received full-dose neuraxis radiation therapy and > 50 Gy to the primary site. Results: With a median follow-up of 100 months, the median, S-year, and IO-year actuarial overall survival for the entire group were 58 months, 50%) and 25 % , respectively. The median, 5- and lo-year actuarial disease-free survivals were 37 months, 41%, and 27%, respectively. Patients with localized disease (no evidence of disease beyond the primary site) had significantly improved overall @ < 0.02) and disease-free (p < 0.02) survivals compared to those with nonlocalized disease. For patients with localized disease, the 5- and lo-year overall survival rates were 59% and 31%) whereas the disease-free survivals were 49% and 31%) respectively. Diseasefree and overall survivals at similar intervals for patients with nonlocalized disease were 29% and 30% (5 years), and 29% and 20% (10 years), respectively. Sixty-four of 100 patients failed treatment. Local failure as any component of first failure occurred in 35% of patients or 55% (35 of 64) of all failures and as the only site of first failure in 14% or 22% (14 of 64) of all failures. For patients presenting with localized disease (n = 68), local failure as any component of first failure occurred in 32% (22 of 68) and in 18% (12 of 68) as the only site. A multivariate analysis showed that M stage was the only prognostic factor to influence overall survival. For diseasefree survival, M stage and the extent of resection were prognostic factors. Ventriculoperitoneal shunting and the use of chemotherapy were associated with a poor outcome; however, these results were confounded by the positive impact of chemotherapy in decreasing the risk of extraneural metastases and the use of these therapies in the more advanced patients. Conclusion: These long-term follow-up data represent one of the largest series of patients with complete followup who were treated with a consistent radiation therapy treatment policy during the CT era. Local failure in patients with localized disease, the persistent risk of late failures, treatment-related toxicity, and the ever-present risk of secondary malignancies demonstrate the limitations of standard therapies. Strategies used to increase the total dose to the primary site should be pursued along with other adjuvant therapies such as intensive chemotherapy.
INTRODUCTION Medulloblastoma remains one of the most intriguing tumors studied and treated by the multidisciplinary oncology team because of its unique clinical and pathologic characteristics and apparent radiocurability. Standard therapy consists of maximal resection compatible with good neurologic outcome and postoperative craniospinal radiation therapy with a boost to the primary site of disease in the posterior fossa using con-
ventional fractionation schemes. Although the treatment volumes and doses have changed little during the past 40 years, the treatment of medulloblastoma continues to evolve. Radiation doses are altered based on the age of the patient and the perceived risk for recurrence. Likewise, chemotherapy is used based on the age of the patient and the presence of factors prognostic for a high risk of failure, whether distant or local. The specific role of chemotherapy and the optimal regimen remain unclear.
Reprint requests to: Thomas E. Merchant, D.O., Ph.D., Department of Radiation Oncology. St. Jude Children’s Research
Hospital, 332 North Lauderdale, Memphis, Accepted for publication 18 May 1996.
29
TN 38105.2794.
30
I. J. Radiation Oncology
0 Biology
0 Physics
Table 1. Clinical characteristics for 100 medulloblastoma patients treated from 1979 to 1994 Clinical characteristics Age Sex (M:F) T stage Tl T2 T3 T4 CSF fluid cytology Positive Negative Unknown M stage (Chang) MO Ml M2 M3 Mx Initial imaging study CT MRI Spine imaging Myelogram MRI Unknown
n
%
14 yr (mean) (53:47)
(l-35 yr, range) (53%:47%)
3 39 47 11
3% 39% 47% 11%
25 70 5
25% 70% 5%
68 12 2 13 5
68% 12% 2% 13% 5%
80 20
80% 20%
76 18 6
76% 18% 6%
CSF = cerebrospinal fluid; CT = computed tomography; MRI = magnetic resonance imaging.
Alterations have been made in the treatment of medulloblastoma and new regimens have been recommended without the benefit of long-term follow-up data for patients treated with standard radiation therapy technique in the era of modern medical imaging and neurosurgical technique. To this end we have compiled complete, long-term follow-up data on 100 patients treated with conventional radiation therapy at our institution during the CT era. The clinical-pathologic characteristics and outcomes of these patients, all of whom were treated with radiation therapy in a similar manner, form
the basis of this report.
MATERIALS
AND METHODS
Patient data The records of 100 evaluable patients treated at Memorial Sloan-Kettering Cancer Center between 1 January 1 1979 and 31 December 31 1994 were reviewed. The information obtained from these records included the following patient and treatment characteristics: date of birth; sex; date and age at presentation; initial imaging study; cerebrospinal fluid (CSF) analysis; imaging study of spine; extent of resection based on operative reports and postoperative imaging; initial use of ventriculoperitoneal shunt; use and sequencingof radiation and chemotherapy; radiation dosesto the neuraxis, whole-brain, and posterior
Volume 36, Number 1, 1996
fossa; and additional use of surgery, radiation therapy, or chemotherapy in failure. The frequencies of the categoric variables and the mean values of the continuous variables are presented in Table 1 and Table 2. Surgery A total of 97 patients underwent an attempted resection and 3 patients were only biopsied. Resections were performed to the extent compatible with good neurologic outcome. A gross-total resection was achieved in 43% and a subtotal resection in 54% basedon the surgeon’soperative impression and postoperative neuroimaging. Ventriculoperitoneal shunting was performed as an initial part of the management in 40% of the cases.
Pretreatment
evaluation
The initial evaluation included X-ray CT scanning or magnetic resonance imaging (MRI). For patients treated in our study, 80% had a CT performed following the onset of symptoms and the remainder had MRI. Contrast myelography was performed in 76% of patients, spinal MRI in 18%, and the remaining 6% either did not have imaging of spine or were unable to undergo the procedure. CSF cytology was performed in 95% of patients. Staging was performed according to the Chang classification (4). The T stage for each patient was determined and the M stage was evaluable for 95% of the patients. Because of the
Table 2. Treatment characteristics for 100 medulloblastoma patients treated from 1979 to 1994 Treatment
characteristics
Surgical Extent of resection GTR STR Biopsy Shunt Radiation therapy Sequencing Initial Postchemotherapy Radiation therapy dosages Primary total dose (mean) Whole-brain dose (mean) Posterior-fossa boost (mean) Spine dose (mean) Special radiation techniques Cribriform plate Spinal electron treatment Chemotherapy Sequencing Adjuvant At recurrence Not given Treatment in failure Additional surgery Reirradiation
n
-
%
43 54 3 40
43% 54% 3% 40%
97 3
97% 3%
53.2 35.9 17.3 33.9
Gy Gy Gy Gy
48-58 Gy 30.6-50.4 Gy 4-25.0 Gy 30.6-45 Gy
17 13
17% 13%
49 35 16
48% 36% 16%
20 18
20% 18%
GTR = gross total resection; STR = sub total resection.
Medulloblastoma:
Long-term results
ambiguity of the staging system, T staging was not included as a variable in the analyses of prognostic faCtors (7). Radiation therapy Patients were treated with neuraxis radiation therapy followed by a boost to the primary site to achieve a total dose of at least 50 Gy. The craniospinal radiation therapy technique was similar for all patients, and the posterior fossa boost was delivered via parallel-opposed photon beams in most cases.Recently, the posterior fossa boost has been carried out using a CT-based, multifield, conformal approach. Patients were treated using 6oCoor a 6MV linear accelerator. In 13% of patients electrons were used to treat the spine. and in 17% an additional boost to the cribriform plate region was performed. Chemotherapy Only 16% of the patients received no chemotherapy. A total of 49% received adjuvant chemotherapy, whereas 35% received chemotherapy at recurrence. The regimens and dosagesused varied considerably over the time period of the study. Dejkitions and analysis of disease-free and overall survival Overall survival and disease-free survival were measured from the date of pathologic diagnosis.Failures were determined by site, including the primary site (posteriorfossa). the nonposterior fossa brain, the spine, meninges (widespread leptomeningeal disease), and extraneural sites (primarily bone and thorax). Failure included recurrence or progression of diseasein sites identified to have active diseaseat presentation or relapse in a site free of diseaseat presentation. The time of failure was determined as the date at which radiologic or biopsy confirmation of relapse at one of these sites occurred. Failure at these sites was determined either as the only site of first failure or as a component of first failure. Extraneural relapse occurred in 9% of patients (8 in bone and 1 in the thorax) or in 14% (9 of 64) of patients who failed treatment. Deaths due to other causes (n = l), treatment-related deaths (brain necrosis [ir = 21, sepsis[n = 11, pneumonia [n = I]), and the development of second malignancies (n = 3) were included in the survival analysis. These seven patients were not analyzed as treatment failures. No patients were lost to follow-up. Statistical methods Using Cox and logistic regression,the clinical and treatment variables were entered in a forward, step-wise manner using p < 0.05 as criterion for inclusion in the model. For evaluation of end points of disease-free and overall survival, Kaplan-Meier methodology was usedwith Wilcoxon-Breslow test statistics (5, 8). The median followup was 100 months.
0
T. E.
MERCHANT
31
ef al.
RESULTS Clinical characteristics The median age of the 100 patients in the study was 14 years (range l-35). They were evenly divided between males (53%) and females (47%). Gross total resection was performed in 43%, subtotal resection in 54%, and biopsy only in 3% of patients. According to the staging system of Chang et al. (4), the distribution of T stagesincluded 3% Tl, 39% T2, 47% T3, and 11% T4. Sixty-eight patients (68%) were determined to be MO, 12% M 1,2% M2, 13% M3, and 5% Mx. The total percentage of patients with nonlocalized disease(i.e., M 1, M2, or M3) was 27%. These data are detailed in Table 1. Treatment characteristics In addition to their extent of resection asoutlined above, the patients were further characterized basedon their treatment profiles. Forty percent required ventriculoperitoneal shunting as a part of their initial therapy. Radiation therapy was given in the immediate postoperative period in 97% of casesand after someform of preradiation adjuvant chemotherapy in only 3%. The mean dosesto the primary site (53.2 Gy). whole-brain (35.9 Gy), and spine (33.9 Gy) were determined. Whereas 16% of the patients received no chemotherapy, 49% received adjuvant chemotherapy and 35% were treated with chemotherapy at the time of recurrence. In failure, 20% of patients underwent additional resection and 18% received palliative radiation therapy to the site of recurrence. These data are described in Table 2. Survivalfor patients with localized and nonlocalized diseaseand other prognostic factors With a median follow-up of 100 months, the median, 5-year, and lo-year actuarial overall survivals for the entire group are 58 months, 50%, and 25%, respectively (Fig. 1). The median (85 vs. 44 months), 5-year (59% vs. 30%), and lo-year (30% vs. 20%) overall survivals are significantly longer @ < 0.02) for patients with localized
Overall Entire
__
Survival Group
~
- --~
_~_~~
100
150
550
Months
Fig. 1. Overall survival (actuarial)for 100medulloblastoma patients treated with definitive radiation therapy from 1979 to 1994.
32
I. J. Radiation Oncology l Biology l Physics % Overall
Survival
MO (n=66)
vs Ml
(n=27)
Fig. 2. Overall survival (actuarial) for 100 medulloblastoma patients treated with definitive radiation therapy from 1979 to 1994. Localized disease (MO) patients have a significantly longer overall survival (p < 0.02, log rank) than those with nonlocalized (M+) disease.
disease(MO) as compared to those with nonlocalized disease (Fig. 2). By subgroup, the median survival of Ml, M2, and M3 patients is, 71, 26, and 19 months, respectively (Table 3). Prognostic factors influencing overall survival by univariate analysis (improved survival vs. decreasedsurvival) included spinal radiation modality (electrons vs. photons, p < 0.04) and the use of chemotherapy (none vs. any, p < 0.004; adjuvant vs. recurrent, p < 0.005; none vs. adjuvant, p < 0.07). By multivariate analysis, only cytology (negative vs. positive, p < 0.002) and initial ventriculoperitoneal shunting (no shunt vs. shunt, p < 0.01) influenced (improved vs. decreased)overall survival. Actuarial survival
curves were compared
(improved
survival
Volume 36, Number 1, 1996
original MO patients and as a component of failure in 32% (22 of 68). Prognostic factors influencing relapse-free survival by univariate analysis included cytology (negative vs. positive, p < O.Ol>,extent of resection (grosstotal vs. subtotal, p < 0.02) spinal radiation modality (electrons vs. photons, p < 0.04) initial ventriculoperitoneal shunting (no shunting vs. shunting, p < 0.04) and useof chemotherapy (none vs. any, p < 0.001; adjuvant vs. recurrent, p < 0.007; none vs. adjuvant, p < 0.07; and no adjuvant vs. adjuvant, p < 0.07). By multivariate analysis, factors influencing disease-free survival included cytology (negative vs. positive,p < O.OOl), shunting (no shunt vs. shunt), and use of chemotherapy (none vs. any, p < 0.002; adjuvant vs. recurrent, p < 0.03; and none vs. adjuvant, p < 0.03). Logistic regression was performed to determine which categoric variables influenced failure at the primary site, in the spine, in the meninges, or at extraneural sites. Positive CSF cytology at presentation predicted for failure in the spine 0, < 0.01) and meninges 0, < 0.04) as a component of failure. The patient receiving chemotherapy. whether
adjuvant
or for recurrence,
Table 3. Outcomes for 100 medulloblastoma from 1979 to 1994
vs. de-
creased survival) for patients based on M stage (p < O.OOl), cytology @ < 0.007), and chemotherapy (none vs. any chemotherapy, p < 0.04; and none vs. adjuvant chemotherapy, p < 0.04).
Patterns of failure and disease-free survival The median time to failure was 37 months for the entire group. Patients with MO diseasehad a significantly longer median time to failure than those with M+ disease: 58 compared with 35 months @ < 0.02). First failure is listed in Table 3 as a sole site of failure or as a component of failure. First failure may represent progression of disease or recurrence in the primary site or in a site of dissemination determined at diagnosis. As the only site of first failure among the 64 patients who failed therapy, failure occurred in the primary site in 22% (14 of 64), nonposterior fossa brain 3% (2 of 64), spine 13% (8 of 64), meninges 9% (6 of 64), and extraneural 9% (6 of 64). As any component of failure, failure in the primary site occurred in 55% (35 of 64) brain 26% (17 of 64), spine 38% (24 of 64), meninges 36% (23 of 64), and extraneural 14% (9 of 64). Local failure occurred in only 18% (12) of the 68
was most
likely
to
have failure at the primary site 0, < 0.02). Patients receiving chemotherapy for recurrence were more likely to have had failures in the spine (p < 0.05) or at extraneural sites(p < 0.01). Patients receiving adjuvant chemotherapy compared with no chemotherapy were more likely to fail at the primary site 0, < 0.04). Finally, patients receiving
Site of first failure Posterior fossa Non-osterior fossa Spine Meninges Extraneural
Only site of failure 14 (16%) 2 8 6 6
(2%) (8%) (6%) (6%)
35 (35%) 17 24 23 9
(15%) (16%) (17%) (3%)
%
(actuarial)
1 7 5-year
f/u = follow-up.
21 (21%) 15 16 17 3
34% 3%
Survival
M+
Any failure
n
55
p < 0.02 (log rank)
Component of failure
34 3
Disease status (70 mo f/u) No evidence of disease Alive with disease Dead from other causes Treatment related death Dead of disease
Disease-free Entire group MO M+ p < 0.02 (log rank) Overall (actuarial) Entire group MO
patients treated
41% 49%
(17%) (24%) (23%) (9%)
1% 7% 55%
lo-year (actuarial)
Median
29%
27% 31% 29%
37 58 35
50% 59% 30%
25% 30% 20%
85 44
mo mo mo
58 mo mo mo
Medulloblastoma:
Long-term
adjuvant chemotherapy were less likely to fail at extraneural sites than were those who did not receive adjuvant chemotherapy @ < 0.04). Actuarial curves were compared for patients divided among the clinical and treatment variables; significant differences were found for cytology 0, < 0.008) and M stage (p < 0.001). DISCUSSION This study provides a useful look at the long-term results of patients treated in the modem era with definitive radiation therapy. Each one of several issues-length of follow-up, treatment era, and radiation therapy techniques and doses-must be carefully considered in the interpretation of historic and recent series. Many of the largest reported series to date describe results associated with short-term follow-up, data which span several decades, multi-institution studies, and patients treated with a variety of radiation therapy and chemotherapy regimens. Table 4 provides an overview of the largest series of medulloblastoma patients for interstudy comparison. The poor survival results at 10 years There are few series with sufficient length of follow-up to determine the survival at 10 years (2, 6-8, 14-16). Although many series, including our own, report good survival rates at 5 years, there is a continuous rate of relapse and mortality beyond 5 years. Thus, our poor survival results are not singular. It appears that the early failures can be attributed primarily to the poor-prognosis patients (M+), and that late failures are attributable to good-prognosis patients (MO). Our data showing an actuarial survival of 25% at 10 years for the entire group and 30% for the MO patients are much worse than the 63% lo-year
Table 4. Results from large, contemporary Series (date) details Evans et nl. (1990)
Reference 6
results
0 T. E. MERCHANT
survival reported by Jenkin et al. (8) and the 45% IO-year survival reported by Tait et al. (15) both contemporary series. The comparison of these series is made difficult because of a lack of staging information in the former study (only 15 of 72 patients underwent any form of spinal imaging), the lack of CSF cytology data, and the lack of follow-up information in the latter study (15% of patients were lost to follow-up at 5 years and 66% at 10 years). Similar problems are encountered when comparing our patients to an earlier series from the Princess Margaret Hospital (2), where a 43% lo-year survival was reported for 122 patients treated from 1958 through 1978. In these patients, neither CSF cytology nor spinal imaging was routinely analyzed or performed. Comparing our series to that of the Royal Marsden (3) (patients treated from 1950 to 1964), their reported survival of 26% at 10 years demonstrates that there has been no improvement in more than 30 years of craniospinal radiation therapy. Local failure as a component of failure Of 100 patients, 64 failed therapy. These failures are represented as either progression of disease or recurrences at or beyond sites of disease that were once in remission. One would expect that the greatest risk for recurrence would be for sites of preexisting disease and that failure would also occur as a result of the inability of treatment to eradicate residual or subclinical disease. Our data and those of others (16) show that the most important site of first failure is the primary site or posterior fossa. Local failure as any component of failure occurred in 35% of all patients or 55% (35 of 64) of all treatment failures. Local failure as the only component of failure occured in 14% of patients, or 22% (14 of 64) of all failures. As the only site of first failure, failure in the posterior fossa occurred in 18% (12 of 68) of patients with localized disease and in only 7% (2
series or series with IO-year survival data employing Study (5 yr)
1975-1981
Hershatter et al. (1986) Jenkin et al. (1990) Stiller and Lennox (1983) Tait et al. (1990)
7 8 14 15
1940- 1983 1977-1987 1971-1977 1975-1979
Tarbell er al (1991) Merchant et al. (1995)
16
1970- 1989 1979- 1994
(I&r) 88 91 127 72 (v) 304 141 145 89 100
33
et nl.
RT follow-up * IB # **
full-dose radiation
Chemotherapy Adjuvant (n = 88) Adjuvant (n = 0) Adjuvant (4/127) Adjuvant (3%) Adjuvant (94/304) Adjuvant (n = 141) Adjuvant (n = 0) Pre-RT (n = 39) Adjuvant (49%)
Entire group
therapy
Median
65%
Dates 5 yr
33% 71% 35% 53%
21% 63% 30% 45%
26 yr 7 yr 9 yr 12yr
65% 50%
48% 25%
9 yr 8 yr
Full-dose radiation therapy is defined as 30-35 Gy to the craniospinal axis and 50-56 Gy to the posterior fossa. * Violations in radiation therapy design (15% < 35 Gy to brain, 26% < 35 Gy to spine). ’ Only 17 received > 52 Gy (PF) and > 30 Gy to spine; 13 received > 40 Gy PF and > 20 Gy to spine. ? Of 72 patients, 37 did not have spinal staging. B Population-based study for which treatment details are available in 94 of 304 patients (posterior fossa > 45 Gy, II = 22. spine > 35 Gy, II = 56). # RT details known for 92% of patients. None excluded for RT protocol violations; 15% of patients lost to follow-up at 5 years, 66% at 9 years. *;rMedian PF dose 54 Gy, and spine dose 29.5 Gy.
34
I. J. Radiation
Oncology
0 Biology
0 Physics
of 27) of patients with nonlocalized disease. This difference may simply be because patients with nonlocalized disease have a shorter median time to recurrence and dissemination of already nonlocalized disease. The overall survival is poor for the nonlocalized group; thus, these patients lack sufficient time to have disease recur at the primary site. Of the 12 patients with originally localized disease who failed in the posterior fossa, the group was evenly divided between those who had undergone a gross total resection (n = 6) and those who underwent a subtotal resection (n = 6). Collin’s law has previously been regarded as applicable to medulloblastoma patients (3, 7) in which cure is assured if patients survive their age and 9 months beyond diagnosis. In our series, of the 64 patients failing therapy, only 10.9% violated this law and failed beyond the time interval defined as their age plus 9 months. Treatment-related morbidity The toxicity of radiation therapy and chemotherapy alone or in combination cannot be overestimated. In this series, two patients suffered from fatal necrosis of neural tissue both in and out of the high-dose regions of treatment. These women, ages 30 and 35 years, received radiation therapy. They deteriorated neurologically after radiation therapy and died. Autopsy showed diffuse white-matter changes and parenchymal necrosis. Two patients suffered from the acute effects of chemotherapyinduced neutropenia and died of pneumonia (n = 1) and sepsis (n = 1). Three additional patients developed second primary brain tumors, anaplastic astrocytoma (n = l), and glioblastoma multiforme (n = 2); thus, 7% of the patients suffered a treatment-related death if one can attribute the second tumors to radiation therapy. One of the worst scenarios suffered by patients, their families, and the treating physicians is a radiation-induced tumor with malignant propensity exceeding that of medulloblastoma. In our series, three patients developed primary brain tumors other than medulloblastoma after radiation therapy. It is conceivable that two of the second primary tumors (glioblastoma multiforme) were radiation therapy induced with intervals from treatment to presentation of 88 and 107 months. The patient with the shortest interval (12 months) developed an anaplastic astrocytoma. It is probable that other patients will develop second primary tumors and that others would have had they outlived their disease. The rate of second tumor formation is probably underestimated owing to the poor treatment results. Treatment volume, dose, and fractionation Craniospinal treatment fields have been used as the standard treatment approach since the early 1950s (3, 13). Fewer controversies surround the radiation therapy treatment volumes than dose and fractionation schemes. In our series, all patients were treated with a consistent treatment policy that included craniospinal radiation and posterior fossa boost portals similar to those previously described.
Volume
36, Number
1, 1996
Three additional measures were used to treat these patients which are not commonly discussed in the methods of the major series. Additional treatment (boost) to the cribriform plate was performed in 17% of patients, and electrons were used to treat the spinal fields in small children. More recently we have used conformal technology and CT-based treatment planning to minimize the dose to the nonposterior fossa brain. The use of cribriform boost had no impact on the patients treated in this series. The use of electrons for treatment of the spine field caused a positive change. Patients treated to the spine with electrons were less likely to fail and had an improved survival, although the numbers are small. Our data as well as those of others show that the control rates at the primary site are inadequate and that the posterior fossa should probably be treated to a higher dose than the 54-56 Gy which is used at present (17). Conformal technology may permit an escalation of dose without an increase in the acute and late effects of treatment. In this study, all patients were treated with conventional fractionation. At the present time there are a number of cooperative group studies employing hyperfractionated radiation therapy (HFRT) in the treatment of medulloblastoma (1). HFRT, 1.O- 1.2 Gy delivered twice a day, is used with the intention of reducing treatment related late effects while at the same time increasing the total dose to improve local control. To date, no study has demonstrated a benefit for HFRT in the treatment of primary brain tumors, including medulloblastoma. Furthermore, the relative radiosensitivity of medulloblastoma and the methods used to determine radiosensitivity are controversial. Inclusion of patients treated with HFRT in some series confounds the results. However, in a recent report (17), no difference in outcome or toxicity was noted when comparing patients treated with HFRT and CERT. Any benefit derived from escalating the total dose with HFRT may be offset by the lower dose per fraction given with each treatment. The hypothesis that HFRT reduces radiation-induced neurologic sequelae may be difficult to test. Considering that the portion of the brain that is treated to the highest dose (posterior fossa) is that which can tolerate the most treatment with the least amount of sequelae, current methods to determine differences in late effects may lack the sensitivity to demonstrate a difference. Conformal technology may serve the patient in a manner equivalent to hyperfractionation. A retrospective analysis of the effect of chemotherapy on outcome Given the retrospective nature of this study, the use of adjuvant chemotherapy in only 49% of the patients, and the multitude of regimens used in both the adjuvant and recurrent settings, it is difficult to take a stand regarding chemotherapy from our data. Chemotherapy was associated with a poor outcome in our series, probably because the patients perceived to be at greated risk were those who
Medulloblastoma:
Long-term
received chemotherapy in the adjuvant setting and the remainder received chemotherapy for recurrence. There are three published randomized trials comparing radiation therapy to radiation therapy and adjuvant or preirradiation chemotherapy (6, 10, 15). None of these trials or other studies (11, 12) demonstrated a survival benefit except for subgroups of patients with advanced disease. Treatment at failure Sixty-four patients failed treatment or progressed while being treated in this study. The median survival following relapse for these patients is 12 months. The actuarial 5year survival following relapse is a dismal 4%. At the time of analysis of this report, 8 of 64 patients who failed are alive at 12, 30,44, 71, 77, 134, 136, and 152 months. Six patients were MO at diagnosis and two had M3 disease. Five of the six MO failures were salvaged with chemotherapy. The two patients originally diagnosed with M3 disease are alive with disease at 12 and 44 months. The group from UCSF reported an actuarial 25% 5-year sur-
results
0 T. E. MERCHANT
et al.
35
viva1 after relapse, although no specific details were given regarding their salvage therapy (17). In summary, these data show that survival at 5 years is not a reliable indicator of treatment efficacy for patients with medulloblastoma receiving with full-dose radiation therapy. Poor long-term survival, treatmentrelated deaths, deaths from second primary tumors, and extraneural dissemination should be carefully considered when determining treatment options for patients with medulloblastoma. Finally, local relapse as a component of first failure in patients with localized disease at presentation remains a significant problem for patients treated with standard-dose radiation therapy with or without the addition of adjuvant chemotherapy. Selecting the appropriate treatment for patients based on perceived risk factors determined from retrospective studies may not be justified. Strategies to increase the dose to the primary site and more intensive chemotherapy regimens should be considered for all patients.
REFERENCES J. C.; Nirenberg, A.; Donohue, B. Hyperfractionated radiotherapy and adjuvant chemotherapy for high-risk PNET. J. Neurooncol. 12:262; 1992. Berry, M. P.; Jenkin, R. D. T.; Keen, C. W.; Nair, B. D.; Simpson, W. J. Radiation treatment for medulloblastoma. J. Neurosurg. 55:43-51; 1981. Bloom, H. J. G.; Wallace, E. N. K.; Henk, J. M. The treatment and prognosis of medulloblastoma in children. Am. J. Roentgenol. 105:43-62; 1969. Chang, C. H.; Housepian, E. M.; Herbert, C., Jr. An operative staging system and a megavoltage radiotherapeutic technique for cerebellar medulloblastoma. Radiology 93: 135 l1359; 1969. Cox, D. Regression models and life tables. J. R. Stat. Sot. 34: 187-220; 1972. Evans, A. E.; Derek, R.; Jenkin, 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. Results of a prospective randomized trial of radiation therapy with and without CCNU, vincristine, and prednisone. J. Neurosurg. 72:572-582; 1990. Hershatter, B. W.; Halperin, E. C.; Cox, E. B. Medulloblastoma: The Duke experience. Int. J. Radiat. Oncol. Biol. Phys. 12:1829-1837; 1986. Jenkin. D.; Goddard, K.; Armstrong, D.; Becker, L.; Berry, M.: Chan, H.; Doherty, M.; Greenberg, M.; Hendrick, B.; Hoffman, H.; Humphreys. R.; Sonley, M.; Weitzman, 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; ! 990. Kaplan, E. S.; Meier, P. Nonparametric estimations from incomplete observations. Am. Stat. Assoc. J. 53:457-482; 1958. Krischer, J. P.; Ragab, A. H.; Kun, L. Nitrogen mustard, vincristine, procarbazine, and prednisone as adjuvant che-
1. Allen,
2. 3. 4.
5. 6.
7. 8.
9. 10.
11.
12.
13. 14. 15.
motherapy in the treatment of medulloblastoma. A Pediatric Oncology Group study. J. Neurosurg. 74:905-909; 1991. Packer, R. J.; Siegel, K. R.; Sutton, L. N.; Evans, A. E.; D’ Angio, G.; Rorke, L. B.; Buning, G. R.; Schut, L. Efficacy of adjuvant chemotherapy for patients with poor-risk medulloblastoma: A preliminary report. Ann. Neurol. 24:503508; 1988. Packer, R. J.; Sutton, L. N.; Goldwein, J. W.; Perilongo, G.; Bunin, G.; Ryan, J.; Cohen, B. H.; D’Angio, G.; Kramer, E.; Zimmerman, R. A.; Rorke, L. B.; Evans, A. E.; Schut, L. Improved survival with the use of adjuvant chemotherapy in the treatment of medulloblastoma. J. Neurosurg. 74:433440; 1991. Patterson, E.; Farr, R. F. Cerebellar medulloblastoma: Treatment by irradiation of the whole central nervous system. Acta Radiol. 39:323-336; 1953. Stiller, C. A.; Lennox, E. L. Childhood medulloblastoma in Britain 1971-77: Analysis of treatment and survival. Br. J. Cancer 48:835-841; 1983. Tait, D. M.: Thornton-Jones, H.; Bloom, H. J. G.; Lemerle, J.; Morris-Jones, P. Adjuvant chemotherapy for medulloblastoma: The first multi-centre control trial of the intema-
tional societyof pediatriconcology (SIOP 1). Eur. J. Cancer 26:464469; 1990. 16. Tarbell, N. J.; Loeffler, J. S.; Silver, B.; Lynch, E.; Lavally, B. L.; Kupsky, W. J.: Scott, R. M.; Sallan, S. E. The change in patterns of relapse in medulloblastoma. Cancer 68: 16001604; 1991. 17. Wara, W. M.; Le, Q.-T. X.; Sneed, P. K.; Larson, D. A.; Prados, M. D.; Levin, V. A.; Edwards, M. S. B.; Weil, M. D. Pattern of recurrence of medulloblastoma after lowdose craniospinalradiotherapy.Int. J. Radiat. Oncol. Biol.
Phys. 30:55l-556; 1994.