hyperfractionated radiation therapy for gliomas of the brainstem in children and in adults

$: Hyperfractionated radiation therapy for gliomas of the brainstem in children and in adults$
Int J Radrolron Oncology RIOI Ph.m Vol. 24, pp. 599-610 0360-3016/92 $5.00 + .OO

Printed in the U.S.A. All rights reserved. Copyright 0 1992 Pergamon Prw Ltd.

0 Clinical Original Contribution

HYPERFRACTIONATED RADIATION THERAPY FOR GLIOMAS OF THE BRAINSTEM IN CHILDREN AND IN ADULTS

DENNIS C. SHRIEVE, M.D., PH .D., * WILLIAM M. WARA, M.D., * MICHAEL S. B. EDWARDS, M.D., + PENNY K. SNEED, M.D., * MICHAEL D. PRADOS, M.D., +

PHILLIP H. COGEN, M.D., + DAVID A. LARSON, M.D., PH.D.* AND VICTOR A. LEVIN, M.D.+

*Dept. of Radiation Oncology, Qept. of Neurological Surgery, University of California, School of Medicine, San Francisco, CA

Between February- 1984 and September 1990,60 patients with brainstem gliomas were treated with hyperfractionated radiotherapy in the Department of Radiation Oncology at the University of California, San Francisco. Forty-one children (g 18 years) and 19 adults were treated with 100 cGy twice daily with 4-8 hr between doses. Thirty-one patients (21 children and 10 adults) received total doses of 66-72 Gy and 29 patients (20 children and nine adults) received 74-78 Gy. Median follow-up was 208 weeks for all patients (214 weeks for children, 157 weeks for adults). Twenty-three patients (14 children and nine adults) were alive at the time of analysis, surviving 59-359 weeks following treatment. Median actuarial survival was 73.6 weeks overall (72 weeks for children, 190 weeks for adults; p = 0.43). Survival at 12 and 24 months was 65% and 38%, respectively (63% and 32%, for children; 68% and 53% for adults). All patients had pretreatment magnetic resonance imsging by which tumors were classified as either focal or diffuse. No significant pretreatment prognostic factors for adults were identified. In children, significant favorable prognostic factors on univariate analysis were older age (p = 0.001 ), tumor location in thalamus or midbrain (p = 0.002), focal appearance on MRI scan (p < 0.001) and duration of symptoms > 2 months prior to treatment (p < 0.001). Thirty-five patients had tumor biopsies, leading to a diagnosis in 33 (22 children and 11 adults). Children with moderately anaplastic astrocytomas survived significantly longer than those with glioblastoma multiforme or unbiopsied tumors (p < 0.001). Only duration of symptoms > 2 months remained significant as a favorable prognostic indicator for children on multivariate analysis (p < 0.001). Survival was not significantly different for patients receiving < 72 Gy and those receiving > 72 Gy (p = 0.18). No subgroup of patients showed significantly better survival with the higher dose. These findings indicate that hyperfractionated radiotherapy is effective treatment for adults and a subgroup of better prognosis children with brainstem gliomas. There is a subgroup of pediatric patients with extremely poor prognosis for whom even this aggressive treatment does little to extend survival. We conclude that there is no benefit to increasing total dose above 72 Gy for any of the groups analyzed.

Brainstem glioma, Radiotherapy, Hyperfractionation, Pediatric brain tumors.

INTRODUCTION

Primary gliomas of the brainstem comprise lo- 15% of brain tumors in children and are less common in adults. Most patients present with neurologic dysfunction, that is, cranial nerve deficits, long tract signs, or ataxia. His- tologies range from differentiated astrocytoma to glioblastoma multiforme and tumors may arise focally in the thalamus, midbrain, pons or medulla or, more commonly, as diffuse brainstem lesions ( 1, 12, 13 ) . Conventional radiotherapy has resulted in reported median survivals of 7-15 monthsforchildren (1,5,6, 11, 14, 19-21,24,26).

Many of these results include patients treated as early as 1950 for whom diagnosis is uncertain. Levin et al. have previously reported a median survival of 44 weeks and a 1 year survival of 47% for children with brainstem gliomas treated with conventional radiation therapy and chemotherapy (24). Survival data for adults treated with radiation therapy (XRT) for brainstem gliomas are scarce (19, 23).

Due to the poor prognosis of children with brainstem gliomas, especially high-risk patients, there have been several trials initiated that utilize hyperfractionated radiotherapy (HFRT) (10, 15, 16, 25, 28-31). HFRT is

Reprint requests to: Dennis C. Shrieve, M.D., Ph.D., De- partment of Radiation Oncology, University of California, Box 0226, San Francisco, CA 94143-0226. Acknowledgements-The authors wish to thank Mary Malec

for assistance in compiling patient data and David Ahn, Ph.D., for performing multivariate analyses.

Accepted for publication 30 March 1992.

599

600 I. J. Radiation Oncology 0 Biology 0 Physics Volume 24, Number 4, 1992

based on theoretical, experimental, and clinical data in- dicating that late responding tissues, such as normal brain, are preferentially spared relative to tumor and early responding tissues as dose per fraction is decreased (35, 38). This sparing of the dose-limiting tissue allows treatment to higher total doses which remain within normal tissue tolerance, but should result in an increased effective tumor dose. Results available to date have not suggested any dose-limiting acute toxicities with HFRT of pediatric brainstem tumors to total doses of 64.8-72 Gy delivered in l-1.2Gyfractionsover6-7weeks(lO, 15, 16,28,29). Reported data have suggested a benefit of HFRT for children with brainstem gliomas and some patients are now being treated to total doses greater than 72 Gy in hope of further improving outcome, especially for poor risk patients.

with vascular proliferation. Highly anaplastic astocytomas (HAA) does not qualify as GBM but does have at least focally high cellularity and two of the following: high nuclear to cytoplasmic ratio, coarse nuclear chromatin, high mitiotic activity, nuclear or cytoplasmic pleomorphism. Moderately anaplastic astrocytomas (MOAA) are not HAA and have mildly to moderate increased cellularity, enlarged nuclei and relatively uniform cytoplasm. HAA are most similar to anaplastic astrocytoma by the WHO classification. MOAA are most similar to low grade astrocytomas.

Between February 1984 and September 1990,41 children and 19 adults with brainstem or thalamic gliomas have been treated with HFRT at the University of Cali- fornia, San Francisco (UCSF).. Some of these patients were included in previously reported results ( lo,25 ) . The present paper updates these results and compares the outcome to those obtained in patients treated more recently to higher total doses.

The majority of patients were taking glucocorticoids, which were tapered as appropriate after the first weeks of radiation therapy (XRT). None of the patients received chemotherapy prior to or during radiation therapy. Two patients had previous diagnoses of neurofibromatosis and one of Osler-Weber-Rendu disease and a family history of brain tumors.

Radiation

METHODS AND MATERIALS

Patients characteristics Between February 1984 and September 1990, 60 pa-

tients with brainstem tumors were treated with hyperfractionated radiotherapy at the Department of Radiation Oncology at UCSF. There were 4 1 children, less than 18 years old, and 19 adults. Median follow-up was 208 weeks for all patients ( 2 14 weeks for adults and 157 weeks for children). Minimum follow-up was 52 weeks. The median age for all patients was 9.3 years (range 2-69 years). All patients had pretreatment magnetic resonance imaging and 35 (24 children, 11 adults) had stereotactic biopsies leading to histologic diagnoses in 33 patients. Eighty-four percent of unbiopsied patients had diffuse lesions on MRI. All patients were treated as part of Brain Tumor Research Center (BTRC) protocol 832 1 or 8725 approved by the Institutional Review Board. Informed consent was obtained from the patient or the patient’s parents prior to treatment. Tumors were classified according to histology, when available, site, and MRI appearance (focal vs. diffuse) ( 3). Age, gender, cranial nerve involvement, and duration of symptoms prior to treatment were also considered as potential prognostic factors.

All but one patient * were treated using megavoltage radiation (4- 18 MV ), 100 cGy twice daily ( 5 days/ week) to the midplane of opposed lateral fields encompassing the tumor and a l-2 cm margin, based on the tumor volume transferred from the sagittal MRI scan to the sim- ulation film. The field normally covered the third ventricle superiorly and included the second cervical vertebral body inferiorly. The anterior border was normally the anterior clinoid and the tumor was covered posteriorly with ade- quate margin. Average field size was 96.8 cm2 before blocking of critical structures and normal brain to the extent possible. There was a minimum of 4 hr between daily treatments.

Initially, doses were escalated to 72 Gy, calculated to be approximately isoeffective with 54 Gy of conventional XRT in terms of effective dose to normal brain. Subse- quently, the dose was escalated to 78 Gy, which represents a 10% increase in effective dose to normal brain compared to 72 Gy (Appendix).

Survival statistics Most patients were followed through the Neuro-On-

cology Service and/or the Department of Radiation On- cology at UCSF. Several patients were referred from or have moved to places too distant to make direct follow- up feasible. Such patients were followed through referring physicians.

Tumor histologies were classified according to criteria established at UCSF ( 8 ). Glioblastoma multiforme (GBM) is a glial neoplasm with at least focally high cellularity and both nuclear and cytoplasmic pleomorphism

A computerized database was compiled using all available patient data. Survival from the first day of radiotherapy was calculated using the Kaplan-Meier method (22). Survivals of patients according to different subgroups were compared using the Mantel-Haenszel log-rank test (27) and multivariate analysis was performed using the

* One adult patient who was 26 wk pregnant at diagnosis was treated with an arc field for a pontine GBM.

HFRT for brainstem gliomas 0 D. C. SHRIEVE et ai. 601

Cox proportional hazards model ( 7 ). Statistical analyses and 3A). These differences based on age were not statis- were performed in September 199 1. tically significant (p = 0.43, Table 1).

RESULTS

Compliance with treatment Thirty-one patients were assigned to the group receiving

I 72 Gy. Twenty-five of 29 patients prescribed 72 Gy received that dose; one received 7 1 Gy and two received 70 Gy. One pediatric patient prescribed 72 Gy received only 68 Gy at the discretion of the attending physician. Two adult patients received 66 Gy as prescribed, one as the first patient in the series and one with a history of diabetes and stroke (both are alive at 359 and 84 weeks, respectively).

Twenty-nine patients were assigned to the group receiving > 72 Gy. Nine patients initially were prescribed and received 76 Gy. Subsequently, 17 of 19 patients prescribed 78 Gy received that dose; two received 77 Gy. One pediatric patient with a juvenile pilocytic astrocytoma received 74 Gy.

There were few delays in treatment with median overall treatment time of 50 days (range, 50-71 days) and 55 days (range, 5 l-88 days), for the groups receiving I 72 Gy and > 72 Gy, respectively.

Overall survival Median actuarial survival for the entire population

studied was 73.6 weeks. Survivals at 1 year and 2 years were 65% and 38%, respectively (Fig. IA). Median survival was longer for adults than for children ( 190 vs. 7 1.7 weeks). In all, 14 children (34%) were alive 59-353 weeks following treatment and nine adults (47%) were living 77-359 weeks post-treatment. Ninety-seven percent of all deaths occurred within 2 years of treatment with only one patient dying at a time longer than 2 years after treatment. One and 2 year survivals were 68.4% and 52.6% for adults and 63.4% and 32% for children, respectively (Fig. 2A

1.0

0.0

3 0.6 3 L ; 0.4

A

, All Patients

0.2 F t

0.0 L ’ * * ’ ’ ’ a ’ * - * ’ - * * ’ 0 100 200 300 400

Time (weeks)

Survival by dose There was no significant difference in survival between

patients treated with doses I 72 Gy and those treated with higher doses (Fig. 1B). Median survival for all patients in the lower dose group was 74.1 weeks and 52.9 weeks for the higher dose group (p = 0.18). Survivals at 1 and 2 years were 77% and 45%, respectively, for the low dose group and 52% and 31% for the high dose group. There was no significant difference between the dose groups when adults or children were considered separately (Fig. 2B and 3B).

E,trect of pretreatment characteristics on prognosis Seven patient characteristics evaluable before treatment

were assessed for potential prognostic value: age, gender, tumor location, tumor appearance on MRI (diffuse vs. focal), the presence of cranial neuropathy, duration of symptoms prior to treatment, and tumor histology, when available.

For adult patients, none of the studied features was a significant predictor of outcome (Table 2). In children, age, tumor location, MRI appearance, duration of symptoms, and histology were statistically significant prognostic factors by univariate analysis (Table 2 ) . Neither gender nor the presence of cranial neuropathies were predictive of outcome.

Age at treatment. The difference in survival seen between adults and children was not significantly different (see above). However, among children, age was a significant prognostic factor. Children less than or equal to the median age of children in this study (6.6 years) had a significantly poorer outcome than older children (p = 0.00 1; Table 1). In the younger population, 17 /2 1 tumors ( 8 1% ) were diffuse compared to 7 / 20 ( 35% ) in the older group (p = 0.008 ) .

1.0

0.8

5 0.6 ._ L ; 0.4

>72 Gy

~~0.18 0.0 I , . , I , , , I , . , ,

0 100 200 300 400 Time (weeks)

Fig. 1. (A) Actuarial survival of all patients (n = 60); (B) Survival of all patients by total dose (2 72 Gy, n = 3 1; > 72 Gy, n = 29); p = 0.18.

602 Volume 24, Number 4, 1992 1. J. Radiation Oncology 0 Biology 0 Physics

1.0 n

0.8 iii

t 0.0 L ’ * ’ ’ ’ ’ - ’ * - ’ ’ ’ * ’ ’

0 100 200 300 400 Time (weeks)

n = 9); p = 0.07.

Tumor location. Survival of children with tumors of the thalamus or midbrain was significantly better than in children with pontine or medullary tumors (p = 0.002). Median survival was not reached for children with thalamic or midbrain tumors. One and 2 year survivals were 85.7% and 62.3%, respectively. Median survival was 52.7 weeks for children with tumors of the pons or medulla with 1 and 2-year survivals of 5 1.9% and 16.3%, respectively (Fig. 4A and Table 2). In neither group was there improved survival in patients receiving the higher dose of radiation (Fig. 4B and C).

MRI appearance. Survival was significantly longer for children with tumors appearing as focal brainstem lesions on MRI scan compared to those with diffuse tumors (p < 0.00 1). Median survival for children with diffuse tumors was 52.7 weeks. One and 2 years survivals were 50% and 4.7%, respectively. Children with focal tumors

All Children

0.2 -

0.0 I * .a 1. a. I *. . I 0 100 200 300 400

Time (weeks)

1.0

0.8

0.6

n-l

PI- B

I 7 >72 Gy I I

1 s72Gy

lIdLQm 0.0 I a * . I . . . I . . . ,

0 100 200 300 400 Time (weeks)

Fig. 2. (A) Survival of all adult patients (n = 19); (B) Survival of adults by dose (5 72 Gy, n = 10; > 72 Gy,

had 1 year and 2 year survivals of 82.4% and 70%, respectively, with median survival not reached (Fig. 5A). There was no significant survival advantage for patients with focal or diffuse brainstem tumors treated to higher doses (Fig. 5B and C).

Duration of symptoms. The median duration of symptoms for the entire 60 patients and for the children as a group was 2 months. Children with symptoms for I 2 months prior to treatment had a poorer outcome (median survival 48.4 weeks) compared to those who had mani- fested symptoms for longer times (median survival not reached; p < 0.00 1, Fig. 6A). One and 2 year survivals were 43.5% and 4.3%, respectively, for patients with a history of symptoms for I 2 months and 89% and 7 l%, respectively, for the more favorable group. The prognostic significance of the duration of symptoms remained when 6 months was used as the analysis criterion (p = 0.003,

1.0

0.8

Z 0.6 2 L 1 0.4

0.2 t ~72 Gy

fi?EQz 0.0 I . . . I . , I I , . , ,

0 100 200 300 400 Time (weeks)

Fig. 3. (A) Survival of all children (n = 41); (B) Survival of children by dose (I 72 Gy; n = 21; > 72 Gy, n = 20); p = 0.77.

HFRT for brainstem gliomas 0 D. C. SHRIEVE et al. 603

Table I. Survival by age

Survival

Age

> 18 < 18 6.7-18 < 6.7

Median

190 71.7 NR 50.7

1 year

68% 63% 80% 48%

2 year

53% 32% 57% 10%

p-value

0.43

0.001

data not shown). Increasing total dose above 72 Gy did not significantly improve survival in either group of children (Fig. 6B and C).

Twenty of 23 children presenting with a history of symptoms I 2 months had diffuse tumors on MRI. Only one child with a short history of symptoms remains alive ( 145 weeks without progressive disease). Conversely, only 4 of 24 children with diffuse tumors presented with symptoms of > 2 months duration. Two of these children are alive, one with a thalamic MOAA without evidence of disease at 157 weeks and the other being the single surviving unbiopsied child with progressive disease at 63 weeks. Of 27 children with either diffuse tumors on MRI or duration of symptoms I 2 months, 20 had both poor prognostic factors. However, duration of symptoms in children was the only prognostic factor studied that had significance on multivariate analysis (p < 0.00 1) .

Tumor histology. Survival was significantly longer for children with MOAAs compared with GBMs (p < 0.00 1) or no biopsy (p < 0.001) (Table 2). One and 2 year survivals for children with MOAAs were 91.7% and 75%, respectively. Children with HAAs had 75% survival at 1 year and 50% at 2 years. One year survivals for unbiopsied

children and those with GBMs were 4 1.2% and 50%, respectively, and there were no 2-year survivors in these two groups.

Seventeen children were not biopsied. Sixteen of these patients demonstrated diffuse tumors on MRI. Only 1 of the 17 unbiopsied children is alive, 63 weeks following treatment with progressive disease (see above). 75% of children with biopsied MOAA or HAA had focal tumors on MRI.

Two children with JPAs (juvenile pilocytic astrocytoma) were treated with hyperfractionated radiotherapy. One child was treated following a second subtotal resection and second recurrence. Another child was treated early in the protocol. Both patients are alive and doing well 1 and 5 years following treatment. None of the above results are changed significantly by the exclusion of these two patients with low-grade tumors (data not shown).

Treatment-related toxicities Treatment-related toxicities were more frequent with

hyperfractionated radiotherapy than would be expected with conventional radiotherapy (Table 3 ). A total of 24 patients had treatment-related toxicities (40% ). Besides the effects on skin, patients had difficulties with otitis media and externa, nonspecific ear pain. tinnitus, and sub- jective hearing loss. There was not an increased frequency of acute toxicities in the > 72 Gy group compared to the group receiving 5 72 Gy. All acute toxicities occurred initially at doses < 72 Gy.

One pediatric patient had documented 75% hearing loss 3 years following treatment with 76 Gy. (Hearing was not formally tested in most patients.) One child expired with bilateral carotid thromboses 9 months after treatment

Table 2. Survival by pretreatment characteristics

Gender Male Female

Crania1 neuropathy Absent Present

Location Thai/midbrain Pans/medulla

MRI appearance Focal Diffuse

Duration of symptoms 5 2 months > 2 months

Histology MOAA HAA GBM No biopsy

?? Based on univariate analysis. + p-values compared to MOAA.

Children (4 1) Adults ( 19)

Median Median in weeks 2-year in weeks

(n) (%) p-value’ 2-year

(n) (%) p-value’

74.1 (22) 40.9 190 (6) 66.7 70.9 (19) 19.3 0.29 73.6 (13) 46.2 0.86 73.7 (14) 25.3 NR (6) 64.70 71.7 (27) 36.4 0.5 1 73.6 (13) 44.4 0.33 NR (14) 62.3 71.9 (5) 40 52.7 (27) 16.3 0.002 190 57.1 0.67 (14) NR (17) 70 NR (6) 50 52.7 (24) 4.7 < 0.001 190 (13) 54 0.71 48.4 (23) 4.3 190 (6) 50 NR (18) 71 < 0.001 NR 53.8 0.64 (13) NR (12) 75 - NR (5) 80 - NR (4) 50 0.18+ 190 (5) 60 0.18+ 53 (4) 0 < 0.001’ 71.8 (1) 0 0.24+ 48 (17) 0 < 0.001+ 73.6 (8) 37.5 0.19+


A

Thalamus / Midbrain

Pons / Medulla

p=O.OO? 0.0 I . ..I . . . I . . . (

0 100 200 300 400 Time (weeks)

572 Gy 172 Gy

0.2 - QAlz >72 Gy f?SQASM

0.0 1 I . . 1 B . . 1 I * . I , 0 100 200 300 400 0 100 200 300 400

Time (weeks) Time (weeks)

Fig. 4. (A) Survival of children by tumor location (thalamus/midbrain, n = 14; pans/medulla, n = 27); p = 0.002. (B) Survival of children with thalamic or midbrain tumors by dose (I 72 Gy, n = 7; > 72 Gy, n = 7); p = 0.17. (C) Survival of children with tumors of the pons or medulla by dose (I 72 Gy, n = 14; > 72 Gy, n = 13); p = 0.094.

of an unbiopsied, diffuse pontine tumor with 78 Gy. Post- mortem examination was not obtained. MRI scan showed recurrent tumor. One adult patient has reported chondronecrosis of the ear 5 years after treatment with 72 Gy.

Three pediatric patients developed opportunistic infec- tions associated with depressed lymphocyte counts near the end of HFRT to 72 Gy. Two proven cases of pneu- mocystis carinii pneumonia (PCP) occurred. The third child developed herpes esophagitis and presumed PCP.

Comparison to conventional XRT Adults. Between 8 173 and 5 / 86, 11 adults were treated

for brainstem gliomas with conventional XRT at UCSF (all adult patients available in UCSF database). Doses ranged from 16 Gy to 59 Gy. One patient was intolerant of treatment and another died during treatment. Each received a total dose of 16 Gy. 9/ 11 patients have died of disease and only one remains alive. Median survival was 37 weeks with 1 year and 2 year survivals of 36% and

18%, respectively. Comparing these results with those obtained using HFRT, there was a trend toward improved outcome with HFRT (p = 0.08, Fig. 7A).

Children. Between 4178 and l/82, 10 children were treated for brainstem gliomas with conventional XRT at UCSF (all pediatric patients available in UCSF database). Median dose was 54 Gy (range 32-60 Gy ). Median survival was 47 weeks. One and 2-year survivals were 30% and 20%, respectively. Comparison to the current HFRT data indicates a trend toward improved survival with HFRT (p = 0.07, Fig. 7B).

DISCUSSION

The diagnosis of brainstem gliomas carries a grim prognosis. Resection is rarely a therapeutic option ( 12, 13, 34) and even biopsy may not be advisable in a large number of cases (4, 12, 13). Radiotherapy has been the only treatment for the majority of patients. Cohen and

HFRT for brainstem gliomas 0 D. C. SHRIEVE et a/ 605

0.6

0.2 - 0.0 *.‘c Diffuse .,..., ““,

0 100 200 300 400 lime (weeks)

F ’ ~72 Gy 0.6

i; 1,. , , , , , , , , , ,p”.“9 (

0 loo 200 300 400 Time (weeks)

0.8

0.6

0.4

0.2

572 Gy

>72 I

GY

c

0.0 ’ ’ ’ ’ d ’ m ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ 0 100 200 300 400

Time (weeks)

Fig. 5. (A) Survival of children by appearance of tumor on MRI (focal, n = 17; diffuse, n = 24); p < 0.00 1. (B) Survival of children with focal tumors by dose (I 72 Gy, n = 11; > 72 Gy, n = 6); p = 0.39. (C) Survival of children with diffuse tumors by dose (I 72 Gy, n = 10; > 72 Gy, n = 14); p = 0.76.

Duffner have reported 2 year and 5 year survivals of 20% and 16%, respectively, for 117 children using data from five United States Surveillence, Epidemiology, End Re- sults registries (6). This is in close agreement with the 17% long-term survival reported by Bloom et al. from their single institution study of 73 children with pontine and medullary tumors (5). Stewart et al. reported a 6% 5-year survival for 144 children from the United Kingdom National Series (32)) although not all of these patients were treated with XRT. Five-year survivals of up to 45% are reported by individual institutions or groups ( 1, 5, 6, 11, 14, 19-2 1, 24, 26). These reports include many patients diagnosed and treated in the pre-CT era. Results may also vary depending on distribution of patient age, sites within the brainstem (with or without inclusion of thalamic tumors), type and duration of presenting symp toms, tumor size, and radiographic appearance. Based on such pretreatment characteristics, pediatric patients may be identified as being either poor-risk or standard-risk (3, 4, 10-14, 17-2 1, 34). Improved imaging with CT and,

more recently, with MRI has led to better characterization of brainstem tumors, has likely eliminated benign lesions which may have been treated presumptively in the past, and has identified a group of patients with diffuse brainstem lesions in whom biopsy is not warranted (4, 12, 13). Most studies include children exclusively and survival data for adults with brainstem tumors are scarce ( 19, 23).

Approximately 90% of failures following XRT for brainstem gliomas are caused by failure to achieve local control ( 14, 19,29 ), suggesting that an increase in tumor dose may improve results. There is evidence for a radiation dose response for brainstem gliomas in some studies ( 19, 23)) but not in others ( 14,26). Walker et al. clearly demonstrated improved survival with increasing dose in patients with supratentorial malignant gliomas (37). Most have used doses of 50-55 Gy over 5-6 weeks as conventional treatment for brainstem tumors. A total dose of 55-60 Gy to normal brain in conventional fractions is generally considered a safe dose ( 37). In an attempt to improve survival in patients with brainstem gliomas, sev-


0.8

0.6

0.4 ~72 Gy

0.2 - 7 4mo JBQau

0.0 L . . . ’ . - * ’ . . . ’ . . . ’ 0 100 200 300 400

Time (weeks)

B 1.0 ~72 Gy

s

0.8 -

3 0.6 ~72 Gy

- > $ 0.4 -

0.2 - Qall

0.0 I . _ . 1 . . . I . . _ , 0 100 200 300 400

Time (weeks) Time (weeks)

Fig. 6. (A) Survival of children by duration of symptoms (5 2 months, n = 23; > 2 months, n = 18); p < 0.001. (B) Survival of children with short history of symptoms by dose (I 72 Gy, n = 9; > 72 Gy, n = 14); p = 0.84. (C) Survival of children with longer history of symptoms by dose (5 72 Gy, n = 12; > 72 Gy, n = 6); p = 0.11.

era1 groups have used hyperfractionated radiotherapy (HFRT) as a strategy to increase the effective dose to the tumor without increasing the effective dose to normal brain above tolerance ( 10, 15, 16, 25, 28, 29, 3 1).

Hyperfractionated radiotherapy Radiobiological models of cell killing by radiation pre-

dict that as dose per fraction is decreased the total dose required for a given effect increases, that is, there is sparing of irradiated tissues. Modem cell survival and dose-effect models employ the linear-quadratic equation:

-lnE=cuD+PD’

where cr and @ are coefficients which describe a tissue’s relative sensitivity to damage caused by single-hit (a-type) or double-hit (p) events (35, 38). The ratio a/P indicates, for a given tissue, the relative contribution of these two components. The value cu/B is the dose at which the effect

due to the (Y component is equivalent to the effect due to P-type damage, that is,

aD = /!3D=

or

D = al/3

Using these equations and experimental data, (Y/P has been determined for a number of early and late effects in normal tissues as well as in experimental tumor systems (3538). Early responding tissues and tumors demonstrate LY//? greater than for late responding tissues. The same relationship holds for early versus late effects in the same tissue (e.g., skin). This indicates that the increment in total dose necessary to balance a decrement in dose per fraction is larger for late responding normal tissues (e.g., brain) than for early responding tissues (e.g., tumor or mucosal surfaces). Clinical data has also indicated a dis-

HFRT for brainstem gliomas 0 D. C. SHRIEVE el al. 607

Table 3. Treatment-related toxicities

Dose

Complication 172Gy >72Gy Total (n = 31) (n = 29) (n = 60)

Early (during XRT) Moist desquamation Otitis externa Otitis media Tinnitus/ear pain Hearing loss Opportunistic infection

Subacute ( l-9 months after XRT)

Carotid thrombosis Late (> 9 months)

Hearing loss Chondronecrosis

4 1 5 3 1 4 1 2 3

: 2 3

3 3 3

1 1

1 1 1 1

sociation of early and late effects (35, 38). a//3 may be used to compare the expected “effective dose” when dose per fraction and total doses are altered (Appendix).

In the above form, the model does not take into account the overall treatment time. Strictly speaking HFRT is the delivery of a larger number of smaller doses of radiation over the same overall time compared to the reference regimen. As demonstrated by low a/& late responding tissues are extremely sensitive to changes in dose per fraction. In a dose escalation study, one must choose between increasing dose per fraction (and losing normal tissue sparing) or increasing overall time (and risking decreased ef- ficacy caused by tumor repopulation ). Some studies have chosen the former approach ( 15, 16 ). We and others (28, 29) have chosen the latter approach based on the rationale that brain tumors are relatively slow growing (compared to head and neck cancers, for example) and that a major

1.0

0.8 R

IL HFRT

A

oo: . 1, j,, cry.y-lTal,, , , , , ,

0 100 200 300 400 500 600 Time(weeks)

goal of HFRT, especially in CNS tumors, is to keep late damage to a minimum.

Results of treatment in adults There are few reports on survival of adults treated for

brainstem or thalamic gliomas. Grigsby et al. ( 19) reported on 19 adults with pontine tumors and 3 1 with tumors of the thalamus treated with conventional radiation between 1950 and 1983. Five-year survivals were 46% and 2 1% for tumors of the pons and thalamus, respectively. Eighteen percent of all patients studied had histologic diagnoses. Kim et al. (23 ) reported a 40% 2-year survival for adults with tumors of the brainstem. Few of the patients in the latter study ultimately survived, however.

Linstadt et al. recently reported on a group of adult patients treated with HFRT at UCSF (25 ). The current analysis represents an update of those results including an additional five patients and confirms that HFRT may offer an advantage over conventional XRT with a 2-year survival of 53%. Comparison to patients treated with conventional XRT at UCSF indicates a trend toward improved survival with HFRT (p = 0.08 ), with nearly triple the 2-year survival rate. The number of patients is small and further investigation is required to be certain of a benefit. There was no apparent benefit of increasing the total dose above 72 Gy and no significant prognostic factors were identified for adults, although small numbers of patients makes comparisons difficult.

Results of treatment in children There are approximately 150-200 new cases of pediatric

brainstem glioma in the United States each year. Reported 2-year survivals are 18-46% for children treated with conventional XRT between 1950 and 1986 (1, 5, 6, 11, 14, 18-2 1, 24, 26). Freeman et al. compiled a series from

zf.1, . 0 100 200 300 400

Time(weeks)

Fig. 7. (A) Survival of adult patients with brainstem gliomas treated with conventional XRT (n = 11) or HFRT (n = 19); p = 0.08. (B) Survival of children with brainstem gliomas treated with conventional XRT (n = 10) or HFRT (n = 41); p = 0.07.

608 1. J. Radiation Oncology 0 Biology 0 Physics Volume 24, Number 4, 1992

among 13 institutions intended to indicate the outcome using modern diagnostic and therapeutic radiology ( 14 ). Sixty-two patients were treated between 1972 and 1981, 52 to focal fields, eight to whole brain and two to the entire craniospinal axis with a median tumor dose of 5 1 Gy. The 2-year survival was 29.1%. Levin et al. have reported a series of 28 children with brainstem gliomas treated with conventional XRT, median dose 55 Gy, and chemotherapy beginning in 1980 (24). Median survival was 44 weeks, l-year survival was 47% and survival at 70 weeks was 30%.

Currently there are at least three major programs of HFRT for brainstem gliomas in children. The Pediatric Oncology Group (POG) began with a regimen of 1.1 Gy BID to a total dose of 66 Gy in 6 weeks ( 15, 30). This is isoeffective in terms of normal brain and represents a 15% increase in effective dose to tumor compared to 1.8 Gy daily to 54 Gy (Appendix). Median survival for 33 high- risk patients was 11 months. One-year and 2-year survivals were 48% and 6.3%, respectively. A second phase of the POG study escalated total dose to 70.2 Gy in 6 weeks by increasing the dose per fraction to 1.17 Gy ( f6). Fifty- seven poor-risk children had a median survival of 10 months, l-year and 2-year survivals were 39.6% and 23%, respectively. There was no statistically significant im- provement in survival (p = 0.15) between the two dose regimens. The authors noted a trend toward improved survival at 2 years and felt that the patients treated to 70.2 Gy were overall a poorer-risk population. The Pediatric Oncology Group has continued the dose escalation study to 75.6 Gy in 6 weeks.

The Children’s Cancer Study Group (CCSG) began a program of HFRT for pediatric brainstem gliomas in 1984. The initial regimen was 1.2 Gy BID to a total dose of 64.8 Gy (28). This regimen is equivalent to the initial POG regimen in terms of effective doses to normal brain (isoeffective with 54 Gy conventional therapy) and tumor ( 14% increase in effective dose, Appendix). Median survival was 11 months. One-year and 2-year survivals were 48% and 7%, respectively. The CCSG study also included predominantly poor-risk patients. Total dose was escalated to 72 Gy with 1 Gy fractions twice daily (29), representing increases in effective dose of 5% and 24% in terms of normal brain and tumor, respectively, relative to conventional XRT (Appendix). Median progression-free survival for 31 poor-risk patients was about 10 months and 28% survived at 2 years. This was significantly better than the group treated to 64.8 Gy or a similar group of 3 1 patients treated with conventional XRT.

In 1984, a program of HFRT for brainstem gliomas in children was begun by the Brain Tumor Research Center and Department of Radiation Oncology at UCSF. Ini- tially, doses were increased to 72 Gy delivered as 1 Gy fractions twice daily, identical to the recent CCSG regimen and equivalent to the POG regimen of 70.2 Gy total dose (Appendix ) . Median survival for all children treated under the protocol was 64 weeks with a 35% 2-year survival

( 10). For children treated only at UCSF, median survival was 73.7 weeks and 2 year survival 33% (present study). There was no benefit to subsequently increasing the total dose to 78 Gy. Children with poor prognostic characteristics did poorly, irrespective of total dose.

It is difficult to compare results among studies. Patient characteristics profoundly influence outcome. Based on histology alone, our patients were remarkably similar to those of the two POG studies combined ( 15, 16). Of 24 biopsies among our pediatric patients, there were 8% pilocytic astrocytoma, 50% MOAA, 17% HAA and 17% GBM. The POG study found 9% pilocytic astrocytoma, 57% astrocytoma, 14% anaplastic astrocytoma, and 14% GBM ( 15, 16 ). Biopsies were obtained in 59% of all children in the present study, compared with 43% in the POG study. The CCSG study reports biopsy results on 22% of all patients with 9% low grade glioma, 73% anaplastic astrocytomas, and 18% GBM (28,29). We found duration of symptoms at presentation to be the only significant prognostic factor on multivariate analysis. The median duration of symptoms for children in our study was 2 months, identical to that reported by POG in their initial study ( 15) ( 1 months in the second POG report ( 16)). Eighty percent of children in the current series reported symptoms of 6 months or less, compared with 9 1% in the POG series. CCSG report duration of symptoms 5 1 months for 69% of patients.

It seems likely that the populations studied in the three series are not distinctly different in terms of pretreatment characteristics. Taken together, they support the use of HFRT for the treatment of brainstem gliomas in children. Two-year survivals of 28% and 33% were obtained in the CCSG and UCSF trials using an identical regimen of 1 Gy twice daily to a total dose of 72 Gy. POG reported a similar 2-year survival of 23% using a regimen calculated to deliver an effective tumor dose 1% greater than that used by POG and UCSF (Appendix). Comparison to results reported by CCSG and to those found in our own patients treated with conventional XRT indicates a sub- stantial benefit from HFRT. Both POG and CCSG found no benefit of HFRT when using lower total effective tumor doses ( 15, 28). The present study has found no benefit of increasing total dose above 72 Gy at 1 Gy twice daily.

Toxicity did not cause any treatment delays in our series. Acute local reactions occurred at a rate felt to be greater than those encountered with conventional therapy. The occurrence of moist desquamation and otitis media or externa were similar in the second POG ( 16) study and in the current series. Intralesional necrosis was reported in three patients (5%) in the POG study ( 16). There were three cases ( 5%) of intralesional necrosis in the current series. CCSG reported no evidence of radio- necrosis in seven patients undergoing postmortem examination 7-18 months after HFRT to 72 Gy (3 1). In the current series, one pediatric patient died with bilateral carotid artery thromboses and recurrent tumor 9 months following HFRT to 78 Gy. Postmortem examination was

not obtained and it is not known if this occurrence was stem and rarely had low grade tumors (8.7%). Approxi- treatment-related or secondary to tumor. Overall, three mately 40% of children presented as better prognosis pa- postmortem examinations revealed no clear evidence of tients (demonstrating none or a single poor prognostic brain necrosis. All three studies showed recurrent tumor. feature) with an overall two-year survival of 80%.

In conclusion, HFRT is effective treatment for brainstem gliomas in adults with an overall 2-year survival of 53%. There was no advantage of increasing total dose above 72 Gy. There were no useful pretreatment prognostic factors identified for adult patients.

Our results with HFRT in pediatric patients are con- sistent with those reported by POG and CCSG using equivalent regimens ( 16, 29). Two-year survivals were 23-32% in the three studies. Survival for children with poor prognostic features was extremely poor in the UCSF series. Duration of symptoms was found to be the only significant prognostic factor on multivariate analysis. Pa- tients with a short history of symptoms tended to be younger, have diffuse tumors located in the lower brain-

Based on these findings and on the findings of POG and CCSG, we are currently treating adult and pediatric brainstem gliomas with HFRT to a total dose of 72 Gy to focal fields. High-risk patients continue to do poorly. A new treatment protocol has recently been opened for high-risk children (diffuse or high-grade tumors) which includes HFRT to 72 Gy and P-interferon for 12 weeks. This and other combined modality approaches, including high-dose chemotherapy with autologous bone marrow transplant may be options for future trials in high-risk patients. A prospective randomized trial comparing HFRT to conventional XRT is needed to determine ef- ficacy in the different prognostic groups of pediatric patients with brainstem gliomas.

HFRT for brainstem gliomas 0 D. C. SHRIEVE et al. 609

REFERENCES I. Albright, A. L.; Guthkelch, A. N.; Packer, R. J.; Price.

R. A.; Rourke, L. B. Prognostic factors in pediatric brainstem gliomas. J. Neurosurg. 6575 l-755;1986.

2. Ang, K. K.; van der Kogel, A. J.; van der Schuren, E. Lack of evidence for increased tolerance of rat spinal cord with decreasing fraction doses below 2 Gy. Int. J. Radiat. Oncol. Biol. Phys. 11:105-l 10;1985.

3. Barkovich, A. J.; Krischer, J.; Kun, L. E.; Packer, R.; Zim- merman, R. A.; Freeman, C. R.; Wara, W. M.; Albright. L.; Allen, J. C.; Hoffman, H. J. Brain stem gliomas: A classification system based on magnetic resonance imaging. Pe- diatric. Neurosurg. 16:73-78;199 1.

4. Berger, M. S.; Edwards, M. S. B.; LaMasters, D.; Davis. R. L.; Wilson, C. B. Pediatric brain stem tumors: Radio- graphic, pathological, and clinical correlations. Neurosurg. 12:298-302;1983.

5. Bloom, H. J. G.; Glees, J.; Bell, J. The treatment and long- term prognosis of children with intracranial tumors: A study of 6 10 cases, 1950-198 1. Int. J. Radiat. Oncol. Biol. Phys. 18:723-745;1990.

6. Cohen, M. E.; Duffner, P. K. Brain tumors in children: Principles of diagnosis and treatment. New York: Raven Press; 1984.

7. Cox, D. R. Regression models and life tables. J. Royal Stat. Sot. 34: 187-220; 1966.

8. Davis, R. L.; Liu, H. C.; Vestnys, P.; Resser, K. J.; Levin, V. A. Correlation of survival and diagnosis in supratentorial gliomas. J. Neuro-Oncol. 2:267; 1984.

9. Edwards, M. S. B.; Prados, M. Current management of brain stem gliomas. Pediat. Neurosci. 13:309-3 15; 1987.

10. Edwards, M. S. B.; Wara, W. M.; Urtasun, R. C.; Prados, M.; Levin, V. A.; Fulton, D.; Wilson, C. B.; Hannigan, J.; Silver, P. Hyperfractionated radiation therapy for brain-stem glioma: A phase II trial. J. Neurosurg. 70:691-700;1989.

11. Eifel, P. J.; Cassady, J. R.; Belli, J. A. Radiation therapy of tumors of the brainstem and midbrain in children: Expe- rience of the Joint Center for Radiation Therapy and Chil- dren’s Hospital Medical Center ( 197 1 - 198 1) . Int. J. Radiat. Oncol. Biol. Phys. 13:847-852; 1987.

12. Epstein, F. A staging system for brain stem gliomas. Cancer 56:1804-1806;1985.

13. Epstein, F.; Wisoff, J. HR. Intrinsic brainstem tumors in childhood: Surgical indications. J. Neuro-Oncol. 6:309- 317;1988.

14. Freeman, C. R.; Suissa, S. Brain stem tumors in children: Results of a survey of 62 patients treated with radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 12: 1823-l 828; 1986.

15. Freeman, C. R.; Krischer, J.; Sanford, R. A.; Burger, P. C.; Cohen, M.; Norris, D. Hyperfractionated radiotherapy in brain stem tumors: Results of a Pediatric Oncology Group Study. Int. J. Radiat. Oncol. Biol. Phys. 15:31 l-318;1988.

16. Freeman, C. R.; Krischer, J.; Sanford, R. A.; Cohen, M. E.; Burger, P. C.; Kun, L.; Halperin, E. C.: Cracker, I.; Wharam, M. Hyperfractionated radiation therapy in brain stem tumors: Results of treatment at the 7020 dose level of Pediatric Oncology Group Study #8495. Cancer 68:474-48 1; 199 I.

17. Greenberger, J. S.; Cassady, J. R.; Levene, M. B. Radiation therapy of thalamic, midbrain and brain stem gliomas. Ra- diology 122:463-468;1977.

18. Grigsby, P. W.; Thomas, P. R. M.; Schwartz, H. G. G.; Fineberg, B. Irradiation of primary thalamic and brainstem tumors in a pediatric population. Cancer 60:290 l- 2906; 1987.

19. Grigsby, P. W.; Thomas, P. R. M.; Schwartz, H. G. G.: Fineberg, B. B. Multivariate analysis of prognostic factors in pediatric and adult thalamic and brainstem tumors. lnt. J. Radiat. Oncol. Biol. Phys. 16:649-655;1989.

20. Halperin, E. C. Pediatric brain stem tumors: Patterns of treatment failure and their implications for radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 11:1293-1298;1985.

2 1. Halperin, E. C.; Wehn, S. M.; Scott, J. W.; Djang, W.; Oakes, W. J.: Friedman, HR. S. Selection of management strategy for pediatric brainstem tumors. Med. Ped. Oncol. 17: 1 16- 125;1989.

22. Kaplan, E. L.; Meier, P. Nonparametric estimation from incomplete observations. J. Am. Stat. Assoc. 53:457- 481;1958.

23. Kim, T. K.; Chin, HR. W.; Pollan, S.; Hazel, J. HR.; Webs- ter, J. HR. Radiotherapy of primary brain stem tumors. Int. J. Radiat. Oncol. Biol. Phys. 6:51-57;1980.

24. Levin, V. A.; Edwards, M. S. B.; Wara, W. M.; Allen, J.; Ortega, J.; Vestnys, P. 5Fluorouracil and l-( 2-chloroethyl)- 3-cyclohexyl- I-nitrosourea (CCNU ) followed by hydroxy- urea, misonidazole, and irradiation for brain stem gliomas: A pilot study of the Brain Tumor Research Center and Children’s Cancer Group. Neurosurg. 14:679-681;1984.

25. Linstadt, D. E.; Edwards, M. S. B.; Prados, M.; Larson, D. A.; Wara, W. M. Hyperfractionated irradiation for adults

610 I. J. Radiation Oncology 0 Biology 0 Physics Volume 24. Number 4, 1992

26.

27.

28.

29.

30.

31.

with brainstem gliomas. lnt. J. Radiat. Oncol. Biol. Phys. 20:757-760;1991. Littman, P.; Jarrett, P.; Bilaniuk, L. T.; Rorke, L. B.; Zim- merman, R. A.; Bruce, D. A.; Carabeh, S. C.; &hut, L. Pediatric brain stem gliomas. Cancer 45:2787-2792;1980. Mantel, N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Che- mother. Rep. 50: 163- 170; 1966. Packer, R. J.; Littman, P. A.; Sposto, R. M.; D’Angio, G.; Priest, J. R.; Heideman, R. L.; Bruce, D. A.; Nelson, D. F. Results of a pilot study of hyperfractionated radiation therapy for children with brain stem ghomas. lnt. J. Radiat. Oncol. Biol. Phys. 13:1647-1651;1987. Packer, R. J.; Allen, J. C.; Goldwein, J. L.; Newall, J.; Zim- merman, R. A.; Priest, J.; Tomita, T.; Mandelbaum, D. E.; Cohen, B. HR.; Finlay, J. L.; Sutton, L. N.; D’Angio, G. Hyperfractionated radiotherapy for children with brainstem gliomas: A pilot study using 7200 cGy. Ann. Neurol. 27: 167-173;1990. Sanford, R. A.; Freeman, C. R.; Burger, P.; Cohen, M. E. Prognostic criteria for experimental protocols in pediatric brainstem ghomas. Surg. Neurol. 30:276-280; 1988. Smith, R. R.; Zimmerman, R. A.; Packer, R. J.; Hackney, D. B.; Bilaniuk, L. T.; Goldberg, HR. I.; Grossman, R. I.;

32.

33.

34.

35.

36.

37.

38.

APPENDIX

Calculation of efective doses for tumor and normal brain in hyperfractionation protocols

The linear-quadratic equation,

-In E = aD -t PD2,

is applicable to single doses. When multiple fractions of radiation are used the equation becomes:

Table A. Dose regimens used in HFRT of brainstem ghomas

Relative to 54 Gy conventional XRT’

Regimen

Dose/Fx Total dose (GY) (GY)

Normal tissue

(Y/p = 2 %

Tumor al@ = 10

% Ref.

1.8 (Con) 54 - 1.8 (Con) 59.4 110 1.2 (BID) 64.8 101 1.1 (BID) 66 100 1.17 (BID) 70.2 108 1 (BID) 72 105 1.26 (BID) 75.6 120 1 (BID) 78 114

- - 110 - 114 28 115 15 123 16 124 29* 134 t 135 *

?? Conventional XRT with treatment once daily compared to BID treatment with two treatments daily.

+ Current POG protocol. * Current study.

Schut, L. Pediatric brainstem glioma: Post-irradiation clinical and MR follow-up. Neuro-radiol. 32:265-27 1; 1990. Stewart, A. M.; Lennox, E. L.; Sanders, B. M. Group characteristics of children with cerebral and spinal cord turnouts. Br. J. Cancer 28:568-574;1973. Strange, P.; Wohlert, L. Primary brain stem tumours. Acta Neurochir. 62:2 19-232; 1982. Stronik, A. R.; Hoffman, HR. J.; Hendrick, E. B.; Hum- phreys, R. P. Diagnosis and management of pediatric brainstem gliomas. J. Neurosurg. 65:745-750;1986. Thames, HR. D.; Withers, HR. R.; Peters, L. J.; Fletcher, G. HR. Changes in early and late radiation responses with altered dose fractionation: Implications for dose-survival relationships. lnt. J. Radiat. Oncol. Biol. Phys. 8:219- 226; 1982. van der Schuren, E.; Landyuyt, W.; Ang, K. K.; van der Kogel, A. J. From 2 Gy to 1 Gy per fraction: Sparing effect in rat spinal cord? lnt. J. Radiat. Oncol. Biol. Phys. 14:297- 300; 1988.33. Walker, M. D.; Strike, T. A.; Sheline, G. E. An analysis of dose-effect relationship in the radiotherapy of malignant gliomas. lnt. J. Radiat. Oncol. Biol. Phys. 5: 1725- 1731;1979. Withers, H. R. Biological basis for altered fractionation schemes. Cancer 55:2086-2095;1985.

-In E = n(cud + pd2),

where d is the dose per fraction and n is the number of fractions. Equivalent equations are:

and

-In E = nd(a + @d)

-In E -z a

where D is the total dose and -In E/LX is the effective dose. Different dose fractionation regimens may be compared on the basis of effective dose, using appropriate a/p for the tissues of interest.

D,(l +-$,)= D2(l +-$)

or

DI alP + 4 -zz D2 QIP + d, *

cz//3 for normal brain is not known but may be esti- mated to be 2 Gy based on experimental work with rat spinal cord (2,36). (Y/P for tumors are generally near 10 Gy. Based on these values and the above equation we have calculated effective doses for tumor and normal CNS tissue based on dosing regimens relevant to the current study (Table A).

hyperfractionated radiation therapy for gliomas of the brainstem in children and in adults

Documents