- Email: [email protected]
A population-based study of glioblastoma multiforme
Int. J. Radiation Oncology Biol. Phys., Vol. 51, No. 1, pp. 100 –107, 2001 Copyright © 2001 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/01/$–see front matter
PII S0360-3016(01)01572-3
CLINICAL INVESTIGATION
Brain
A POPULATION-BASED STUDY OF GLIOBLASTOMA MULTIFORME LAWRENCE PASZAT, M.D., M.S., F.R.C.P.C.,*†‡ NORMAND LAPERRIERE, M.D., F.R.C.P.C.,‡ PATTI GROOME, PH.D.,*§ KARLEEN SCHULZE, M.MATH.,* WILLIAM MACKILLOP, M.B., F.R.C.P.C.,*§ 㛳 AND ERIC HOLOWATY, M.D., M.SC., F.R.C.P.C. *Radiation Oncology Research Unit, Department of Oncology and §Department of Community Health and Epidemiology, Queen’s University, Kingston, Canada; †Institute for Clinical Evaluative Sciences, Toronto, Canada; ‡Department of Radiation Oncology, University of Toronto, Toronto, Canada; 㛳Cancer Surveillance Unit, Cancer Care Ontario and University of Toronto, Toronto, Canada Purpose: To describe (1) the use of surgery and radiotherapy (RT) in the treatment of patients with glioblastoma (GBM) in Ontario, (2) survival, and (3) proportion of survival time spent in the hospital after diagnosis. Methods and Materials: We performed a population-based cohort study of all Ontario Cancer Registry (OCR) cases of GBM diagnosed between 1982 and 1994. We linked OCR records, hospital files containing surgical procedure codes from the Canadian Institute for Health Information, and province-wide RT records. We studied the odds of treatment using multivariate logistic regression. We expressed the time spent in the hospital as the mean number of days per case, and as a proportion of the interval between diagnosis and death, or 24 months following diagnosis, whichever came first. We used the life-table method and Cox proportional hazards regression to describe survival. Results: The proportion of patients with GBM undergoing any surgery directed at the tumor varied with age (p < 0.0001) and region of residence (p < 0.0001). The proportion undergoing RT varied with age (p < 0.0001), region of residence (p < 0.0001), and year of diagnosis (p ⴝ 0.01). RT dose > 53.5 Gy varied with age (p < 0.0001), region of residence (p < 0.0001), and year of diagnosis (p ⴝ 0.0002). Median survival was 11 months among patients receiving RT and 3 months among those not receiving RT. The percentage of survival time spent in the hospital was similar among those who received from 49.5 to < 53.5 Gy, compared to > 53.5 Gy. Overall survival and the adjusted relative risk of death varied with age and region of residence. Conclusion: We observed practice variation in the treatment of patients with GBM according to age, region of residence, and year of diagnosis. Survival did not increase during the study period. The variation in RT dose between those receiving from 49.5 to < 53.5 Gy compared to > 53.5 Gy was not paralleled by variation in survival between regions where one or the other of the dose ranges predominated, nor was variation in dose ranges among the regions paralleled by variation in the proportion of survival time spent in the hospital. © 2001 Elsevier Science Inc. Glioblastoma, Population-based study, Radiation therapy, Outcomes research.
Glioblastoma multiforme (GBM) is a highly malignant form of brain tumor. Primary treatments include surgery and radiotherapy (RT). Contemporary practice varies with respect to (1) radiation dose, fractionation, planning and delivery, and (2) whether or not to use chemotherapy. GBM and its treatment may each contribute to declining performance status and quality of life. Health-related quality of life may improve transiently during therapy (1), and fatigue may increase and quality of life decrease after completion of therapy (2). Loss of verbal memory is independently associated with survival, adjusting for age and performance status (3).
Over several decades, investigators have attempted to improve survival by altered RT fractionation schemes (4), RT dose escalation (5), use of boosts by external RT or by brachytherapy (6), complex treatment-planning techniques (7), radiosensitizers (8), and chemotherapy (9). These studies have yielded rich information about prognostic factors, but randomized clinical trials have not demonstrated major improvements in patient outcomes. Several randomized clinical trials (RCTs) have shown that median survival is longer with RT than without RT (10 –14). Another RCT has shown that median survival is longer with 60 Gy than with 45 Gy (15); however, no additional improvement has been found in other studies of higher doses with or without hyperfractionation or acceler-
Correspondence to: Dr. Lawrence Paszat, Toronto Sunnybrook Regional Cancer Centre, 2075 Bayview Avenue, Toronto, Ontario, Canada M4N 3M5. E-mail: [email protected] This work was supported in part by an operating grant from Cancer Care Ontario to the Radiation Oncology Research Unit.
Acknowledgments—Dr. Paszat and Dr. Groome are career scientists of the Ministry of Health of Ontario. The radiation oncology departments of Ontario provided the radiation therapy data. Received Nov 9, 2000, and in revised form Feb 26, 2001. Accepted for publication Mar 20, 2001.
INTRODUCTION
100
Population-based study of glioblastoma multiforme
ated fractionation. Meta-analysis of RCTs, studying adjuvant chemotherapy combined with RT (9), demonstrates that any benefit would likely be restricted to the small subset of young patients with good performance status and minimal residual GBM postoperatively. Patients enrolled within clinical trials for their treatment often have better outcomes than patients treated outside of clinical trials. They are rigorously selected to be comparable, and their physicians, treating institutions, and treatment processes are subject to stringent guidelines (16). The treatment benefits seen in randomized clinical trials may be diminished when the treatment is implemented in the general population, because the spectrum of patients treated will be broader and there will be a broader range of treatment processes, physicians, and institutional expertise outside of the context of a randomized clinical trial (17, 18). In contrast to the efficacy of interventions demonstrated in controlled clinical experiments, the effectiveness of an intervention is the outcome of care provided under ordinary conditions by average practitioners for typical patients (19). Evidence of the effectiveness of an intervention is the extent to which the benefits are observed in the general population (20). We have performed a population-based cohort study of all patients with GBM residing in Ontario at the time of diagnosis between January 1, 1982 and December 31, 1994. The objectives of this study were to report (1) the use of surgery and radiotherapy (RT) for glioblastoma (GBM) in Ontario, (2) survival, and (3) time spent in the hospital after diagnosis. We will describe, in a complete and entirely unselected population, the utilization and effectiveness of interventions for which efficacy has been demonstrated in clinical trials. METHODS AND MATERIALS We performed a population-based cohort study of all patients with GBM residing in Ontario at the time of diagnosis between January 1, 1982 and December 31, 1994. The study is based on linked electronic databases from the Ontario Cancer Registry (OCR), the Canadian Institute for Health Information (CIHI), and the Radiation Oncology Research Unit (RORU) at Queen’s University, Kingston, Ontario, Canada. From the electronic files of the OCR, we identified all cases of ICD (International Classification of Diseases) 191 with the ICD-O morphology code for GBM (94403). The morphology codes are assigned to each pathology report received by the OCR by health records staff. Cases with morphology codes for anaplastic astrocytoma or any other histologic type of brain tumor were excluded. In the OCR electronic records, each case has a unique numeric identifier, as well as the following variables: diagnosis date, age at diagnosis, Ministry of Health residence code, postal code, vital status, numeric cancer center label, and cancer center chart number. The methods of case ascertainment by the OCR are
● L. PASZAT et al.
101
described by Clarke et al. (21), and the quality of case ascertainment by Holowaty et al. (22). The OCR ascertains cases on the basis of data from the following records originally created for other purposes: (1) registrations at cancer centers, (2) pathology reports received directly from acute care hospitals, (3) hospital discharge abstracts received from the Canadian Institute for Health Information (CIHI), and (4) death certificate diagnoses. Electronic hospital procedure and discharge abstracts from CIHI were previously labeled with the OCR unique numeric identifier for each case. The hospital records contain the admission date, discharge date, major treatment procedures, and ICD diagnoses for each admission. The accuracy of electronic records of major cancer-directed surgery in Ontario is described by Morovan et al. (23). Electronic RT records from all cancer centers in Ontario were compiled and processed by RORU and linked to the OCR and CIHI databases. The variables in the RT records used for linkage included the cancer center chart number and the numeric cancer center label. Each RT record includes the cancer center chart number, the date RT commenced, the anatomic region, the dose delivered, and the fraction number. The records contain no information concerning the size of the tumor or its location in the brain, pathologic features other than the histology code, the radiologic appearance of the tumor, the performance status of the cases, or of the use of chemotherapy. We categorized the surgical procedures directed at the brain tumor as (1) total or subtotal resection, (2) biopsy, and (3) lessor or no procedure. We selected RT records with appropriate anatomic region codes. We divided the province of Ontario into regions defined as the areas served by each cancer center. We assigned each census subdivision in Ontario to the cancer center most frequently visited by the residents of each census subdivision (24). In our analyses of treatment utilization, we will consider region of residence as a surrogate process indicator of practice and policy variations based on cancer centers and academic health sciences centers distributed across the province of Ontario. We described the population univariately by age at diagnosis, gender, region of residence, and year of diagnosis. We described the utilization of treatment procedures performed within 4 months of diagnosis by age, region of residence, and year of diagnosis. We tested the comparison of proportions using the 2 test. We modeled the odd ratios for treatment procedures using multivariate logistic regression. We described survival following diagnosis using the lifetable method and multivariable Cox proportional hazards regression analyses. We expressed the percentage of survival time spent in the hospital as the percentage of the time between the date of diagnosis and death interval between diagnosis and death spent in hospital, or, if a patient is still alive 24 months following diagnosis, the percentage of time between the
102
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 51, Number 1, 2001
Table 1. Population description and surgical treatment within 4 months of diagnosis Number of cases
Total or subtotal resection
Biopsy
No surgical procedure
81 219 414 733 1105 620 107
58 (71.6%) 153 (69.9%) 319 (74.9%) 541 (73.8%) 769 (69.6%) 391 (63.1%) 39 (36.5%)
11 (13.6%) 34 (15.5%) 71 (17.1%) 106 (14.5%) 209 (18.9%) 134 (21.6%) 35 (32.7%)
12 (14.8%) 32 (14.6%) 33 (8.0%) 86 (11.7%) 127 (11.5%) 95 (15.3%) 33 (30.8%)
356 1477 504 129 444 78 92 173 26
205 (57.6%) 1031 (69.8%) 353 (70.0%) 87 (67.4%) 321 (72.3%) 37 (47.4%) 67 (72.8%) 153 (88.4%) 7
67 (18.8%) 290 (19.6%) 115 (22.8%) 22 (17.1%) 74 (16.7%) 10 (12.8%) 11 (13.0%) 10 (5.8%) 0
84 (23.6%) 156 (10.6%) 36 (7.1%) 20 (15.5%) 49 (11.0%) 31 (39.7%) 13 (14.1%) 10 (5.8%) 19
852 1002 1425 2108
647 (75.9%) 714 (71.3%) 900 (63.2%) 1640
91 (10.8%) 155 (11.5%) 354 (24.2%) 295
114 (13.3%) 133 (13.3%) 171 (12.0%) 173
Age at diagnosis (years) 20–29 30–39 40–49 50–59 60–69 70–79 ⱖ 80 Region Ottawa Toronto Hamilton Kingston London NW Ontario Windsor NE Ontario Missing Year of diagnosis 1982–1985 1986–1989 1990–1994 Radiotherapy
date of diagnosis and the second anniversary of the diagnosis. RESULTS We identified 3279 cases of GBM diagnosed in Ontario between 1982 and 1994. Table 1 describes the population by age at diagnosis, region of residence, and year of diagnosis. Mean age at diagnosis is 59.3 years (standard deviation, 13.1), and median age is 61 years. Males comprise 59.0% of the population. Table 1 describes the utilization of surgical procedures directed at glioblastoma multiforme (GBM) within 4 months of diagnosis. The proportion of patients undergoing total or subtotal resection compared to biopsy alone varied with age (p ⬍ 0.0001), region of residence (p ⬍ 0.0001), and year of diagnosis (p ⬍ 0.0001). The proportion of patients undergoing any surgery directed at the tumor (including biopsy) compared to no surgery directed at the tumor varied with age (p ⬍ 0.0001) and region of residence (p ⬍ 0.0001), but not with year of diagnosis (p ⫽ 0.48). Table 1 describes the combinations of surgery and RT observed. Fifty percent of cases underwent total or subtotal resection plus RT, 37.2% of cases underwent surgery without RT, 5.3% of cases received RT without any surgery, and 7.5% of cases received neither surgery nor RT. Table 2 describes the utilization of RT within 4 months of diagnosis. The proportion of patients receiving RT within 4 months of diagnosis varied with age (p ⬍ 0.0001), region of residence (p ⬍ 0.0001), and year of diagnosis (p ⫽ 0.01). The mean dose was 48.8 Gy (median, 50 Gy). Mean fraction
number was 24 (median, 25). The most common dosefractionation combinations were 50 Gy in 25 fractions (37.6%), 60 Gy in 30 fractions (11.1%), and 54 Gy in 30 fractions (5.0%). The proportion of patients receiving RT dose ⬎ 53.5 Gy, compared with ⱕ 53.5 Gy varied with age (p ⬍ 0.0001), region of residence (p ⬍ 0.0001), and year of diagnosis (p ⫽ 0.0002). The majority of patients in the Toronto area received from 49.5 Gy to ⬍ 53.5 Gy, whereas the majority of patients elsewhere in Ontario received ⱖ 53.5 Gy. The RT records do not distinguish complete vs. incomplete treatment doses: these are all delivered doses. Linear accelerators were used to administer RT to 55.4% of patients, whereas 44.6% were treated with cobalt units. Table 3 presents the odds ratios and 95% confidence intervals from multiple logistic regression analyses on (1) receiving total or subtotal resection; (2) receiving RT; and (3) among those receiving RT, receiving dose ⱖ 53.5 Gy. For each of these dependent variables, there was variation according to age, region of residence, and year of diagnosis. Median survival was seven months from the date of diagnosis. Overall survival from the date of diagnosis by the life-table method was 29.4% (95% confidence interval [CI], 27.8 –30.9%) at 1 year, 11.1% (95% CI, 10.0 –12.1%) at 2 years, 7.4% (95% CI, 6.5– 8.3%) at 3 years, 5.5% (95% CI, 4.7– 6.3%) at 4 years, and 4.8% (95% CI, 4.0 –5.6%) at 5 years. Overall survival varied by age at diagnosis (Wilcoxon and log–rank tests each yielded p ⬍ 0.0001), by region of residence (Wilcoxon test, p ⬍ 0.0001; log–rank test, p ⫽ 0.0012) but not by year of diagnosis (Wilcoxon test, p ⫽ 0.13; log–rank test, p ⫽ 0.49). Median survival was eleven months among patients re-
Population-based study of glioblastoma multiforme
● L. PASZAT et al.
103
Table 2. Description of radiotherapy by dose
Age at diagnosis (years) 20–29 30–39 40–49 50–59 60–69 70–79 ⱖ 80 Region Ottawa Toronto Hamilton Kingston London NW Ontario Windsor NE Ontario Year of diagnosis 1982–1985 1986–1989 1990–1994
Radiotherapy
Dose ⬍39.5 Gy
Dose 39.5–⬍49.5 Gy
Dose 49.5–⬍53.5 Gy
Dose ⱖ 53.5 Gy
60 (74.1%) 168 (76.7%) 324 (78.3%) 555 (75.7%) 715 (64.7%) 265 (42.7%) 21 (19.6%)
3 (3.7%) 7 (8.6%) 29 (9.0%) 42 (7.6%) 115 (16.1%) 123 (46.4%) 11 (52.4%)
7 (8.6%) 7 (4.2%) 24 (7.4%) 39 (7.0%) 72 (10.1%) 32 (12.1%) 1 (4.8%)
29 (35.8%) 76 (45.2%) 147 (45.4%) 263 (47.4%) 302 (42.2%) 49 (18.5%) 7 (33.3%)
21 (25.9%) 78 (45.6%) 123 (38.0%) 210 (37.8%) 226 (31.6%) 60 (22.6%) 2 (9.5%)
174 (48.9%) 979 (66.3%) 305 (60.5%) 87 (68.2%) 304 (68.5%) 52 (66.7%) 75 (81.5%) 129 (74.6%)
35 (20.1%) 148 (15.1%) 39 (12.8%) 19 (21.6%) 60 (19.7%) 7 (13.5%) 8 (10.7%) 14 (10.9%)
15 (8.6%) 92 (9.4%) 17 (5.6%) 5 (5.8%) 28 (9.2%) 7 (13.5%) 11 (14.7%) 7 (5.4%)
23 (13.2%) 698 (71.3%) 53 (17.4%) 19 (21.8%) 13 (4.3%) 7 (13.5%) 3 (4.0%) 55 (42.6%)
101 (58.1%) 41 (4.2%) 194 (63.6%) 44 (50.6%) 203 (66.8%) 31 (59.5%) 53 (70.6%) 53 (41.1%)
549 (64.4%) 676 (67.5%) 883 (62.0%)
58 (10.6%) 77 (11.4%) 195 (22.1%)
68 (12.4%) 317 (3.7%) 89 (10.1%)
273 (49.7%) 320 (47.3%) 280 (31.7%)
149 (27.1%) 254 (37.6%) 317 (35.9%)
ceiving RT and 3 months among patients not receiving RT. Overall survival was 4.1% (95% CI, 3.1–5.0%) at 5 years among RT patients compared to 2.6% (95% CI, 1.6 –3.7%) among patients not receiving RT (p ⬍ 0.0001). The mean length of stay in hospital between the date of diagnosis and the date of death, or the second anniversary of diagnosis, declined from 90.7 days among patients diagnosed in 1982 to 53.7 days among patients diagnosed in 1994. Table 4 presents the percentage of survival time spent in hospital, according to age at diagnosis, surgical proce-
dure, use of RT, and RT dose. The percentage of survival time in hospital is defined as the time between date of diagnosis and date of death, or if alive at 2 years, between the date of diagnosis and the second anniversary of diagnosis. The percentage of survival time spent in hospital varied by age at diagnosis (age ⱕ 59 years compared to age ⱖ 60 years, p ⬍ 0.0001), by resection compared to biopsy (p ⬍ 0.0001), by the use of RT compared to no RT (p ⬍ 0.0001), and by RT dose ⬍ 49.5 Gy compared to ⱖ 49.5 Gy (p ⬍ 0.0001) among those who received RT. There was no dif-
Table 3. Odds ratios for total or subtotal resection, for radiotherapy (RT) and for RT dose ⬎53.5 Gy
Age at diagnosis (years) 20–29 30–39 40–49 50–59 60–69 70–79 ⱖ 80 Region Ottawa Toronto Hamilton Kingston London NW Ontario Windsor NE Ontario Year of diagnosis 1982–1985 1986–1989 1990–1994
Odds ratios for total or subtotal resection
Odds ratios for radiotherapy
1.00 (0.60, 1.66) 1.00 (0.72, 1.38) 1.26 (0.97, 1.64) 1.14 (0.92, 1.41) 1.00 0.77 (0.62, 0.95) 0.25 (0.17, 0.38)
1.59 (0.95, 2.67) 1.80 (1.28, 2.54) 2.0 (1.53, 2.61) 1.69 (1.36, 2.09) 1.00 0.40 (0.32, 0.49) 0.13 (0.08, 0.21)
0.58 (0.45, 0.73) 1.00 1.00 (0.80, 1.26) 0.86 (0.59, 1.28) 1.20 (0.94, 1.52) 0.38 (0.27, 0.60) 1.22 (0.76, 1.98) 3.10 (1.91, 5.02)
0.50 (0.39, 0.65) 1.00 0.77 (0.62, 0.96) 1.05 (0.70, 1.57) 1.29 (1.01, 1.64) 1.10 (0.67, 1.83) 2.84 (1.62, 5.00) 1.30 (0.90, 1.88)
1.00 0.79 (0.64, 0.98) 0.54 (0.45, 0.66)
1.00 1.25 (1.02, 1.54) 1.00 (0.83, 1.21)
Odds ratios for dose ⬎53.5 Gy 1.86 (0.90, 3.85) 3.03 (1.87, 4.90) 2.78 (1.24, 2.57) 1.87 (1.36, 2.53) 1.00 0.43 (0.30, 0.64) 0.11 (0.02, 0.50) 38.52 (24.53, 60.50) 1.00 51.40 (34.17, 77.31) 27.74 (16.10, 77.31) 58.53 (38.76, 88.38) 47.26 (24.06, 92.82) 91.94 (48.81, 173.18) 15.98 (9.87, 25.88) 1.00 2.13 (1.56, 2.91) 2.11 (1.56, 2.84)
104
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 51, Number 1, 2001
Table 4. Percentage of survival time spent in hospital
Age at diagnosis (years) 20–29 30–39 40–49 50–59 60–69 70–79 ⱖ 80 Surgical procedure Resection Biopsy None Radiotherapy Yes No Radiotherapy dose ⬍39.5 Gy 39.5–⬍49.5 Gy 49.5–⬍53.5 Gy ⱖ 53.5 Gy Year of diagnosis 1982–1985 1986–1989 1990–1994 Overall
⬍10%
10–⬍25%
25–⬍50%
50–⬍100%
100%
61.7% 59.5% 36.0% 24.6% 14.0% 9.4% 3.8%
25.9% 19.3% 27.1% 26.3% 20.8% 14.9% 10.5%
6.2% 9.6% 14.7% 22.3% 20.0% 16.3% 9.5%
3.7% 6.4% 12.1% 16.2% 23.4% 21.4% 26.7%
2.5% 5.5% 10.1% 10.6% 21.9% 38.0% 49.5%
22.1% 16.7% 31.1%
23.6% 16.2% 17.3%
18.5% 16.7% 15.8%
17.9% 21.3% 17.8%
18.0% 29.2% 18.1%
27.1% 13.3%
27.3% 10.6%
21.0% 11.9%
17.0% 21.3%
7.6% 42.9%
12.1% 20.7% 30.9% 31.0%
15.8% 25.0% 29.3% 31.0%
23.9% 23.9% 20.2% 19.8%
30.6% 17.9% 14.2% 13.8%
17.6% 12.5% 5.4% 4.5%
17.7% 20.6% 27.3% 22.2%
21.7% 19.8% 22.0% 21.4%
19.0% 17.7% 16.9% 17.8%
17.7% 20.6% 17.2% 18.5%
23.8% 21.6% 16.6% 20.0%
ference in the percentage of survival time spent in hospital between those who received from 49.5 Gy to ⬍ 53.5 Gy and those who received ⱖ 53.5 Gy. Table 5 presents the relative risk (RR) of death following diagnosis of GBM. The RR of death varied with age at diagnosis, and in one region of Ontario relative to the reference region, and did not vary by year of diagnosis.
Table 5. adjusted relative risk of death Relative risk of death Age at diagnosis (years) 20–29 30–39 40–49 50–59 60–69 70–79 ⱖ 80 Region Ottawa Toronto Hamilton Kingston London NW Ontario Windsor NE Ontario Year of diagnosis 1982–1985 1986–1989 1990–1994
0.22 (0.17, 0.29) 0.30 (0.25, 0.35) 0.52 (0.46, 0.59) 0.71 (0.65, 0.79) 1.00 1.70 (1.54, 1.88) 2.65 (2.16, 3.25) 1.26 (1.12, 1.42) 1.00 1.08 (0.97, 1.20) 1.19 (0.99, 1.43) 1.01 (0.91, 1.13) 0.93 (0.73, 1.17) 1.12 (0.91, 1.40) 0.91 (0.77, 1.07) 1.00 1.00 (0.91, 1.09) 1.00 (0.92, 1.10)
DISCUSSION Diagnosis of GBM The median survival rates described in this populationbaed cohort are typical of those described in the literature. 12.7% of patients in this cohort identified by the ICD-O morphology code for GBM (99403) did not have a record of open biopsy or resection. Electronic records of surgery might be missing for some of the patients, others might have had a more limited biopsy procedure, and the remainder might have had a radiologic diagnosis of GBM. The morphology codes are based on paper histopathology reports from dozens of hospitals and pathologists, in academic as well as community settings, and are subject to interobserver variation as described in the clinical literature. Despite some reports of very high reproducibility of some classification systems of malignant glial brain tumors (25, 26), unreliability is likely present. For example, Aldape et al. reported 89% concordance between original histopathologic diagnoses and diagnoses given after expert review of cases of GBM in the San Francisco area. The diagnosis changed from GBM to another histology in 13 of 253 patients, and that, among initial diagnoses of anaplastic astrocytoma, 20 of 57 patients initially labeled as anaplastic astrocytoma were labeled GBM by expert review (27). Central histopathology review by the Radiation Therapy Oncology Group resulted in 96% agreement on the diagnosis of GBM between initial and central review, but 34% of anaplastic astrocytomas on initial review were subsequently diagnosed as glioblastoma on central review (28).
Population-based study of glioblastoma multiforme
Treatment variations–surgery The percentage of patients for whom the maximal surgical procedure was biopsy increased from 10.6% in 1982–1985 to 22.1% in 1990 –1994, and the percentage undergoing total or subtotal resection decreased from 75.9% to 63.2%. During the same interval, the percentage of patients without a record of surgery remained unchanged. Patients older than 70 years of age were less likely to have resection and more likely to have biopsy or no record of surgery. Treatment variations—RT A total of 63.4% of all patients received RT, and 75.6% of those received ⱖ 49.5 Gy. Only a minority of patients aged 70 years or older received RT, and, of those who did, the majority received ⬍ 39.5 Gy. Lower dose palliative RT has been recommended in recent clinical literature as appropriate therapy for those with poor prognosis GBM, which includes patients aged 70 or older (29 –31). Comparing RT dose in the Toronto area to RT dose in the rest of Ontario: the RT dose was between 49.5 Gy and ⬍ 53.5 Gy for 698/979 (71.3%) of patients in the Toronto receiving RT, whereas the dose was ⱖ 53.5 Gy for 679/ 1178 (57.6%) for patients receiving RT in the rest of Ontario. This is a remarkable practice variation. Interpreting this variation is difficult. We do not know if the clinical indications for RT used in the Toronto area were different from those in the rest of Ontario; however, the adjusted odds ratios for the use of RT within 4 months of diagnosis (Table 3) show lower odds of RT among patients living in the Ottawa and Hamilton regions, and higher odds among patients living in the London and Windsor areas, relative to Toronto. The systematic difference in dose may provide the context of a natural experiment and suggest hypotheses about the outcomes associated with populous regions, where divergent doses prevailed. Survival The overall survival experience we report for this cohort is higher than expected, although Polednak and Flannery reported 5-year relative survival of 4% for patients diagnosed 1983–1987 in Surveillance, Epidemiology, and End Results (SEER) data (32). Davis et al. reported 2- and 5-year overall survival as 1% in a cohort study of the SEER data from 1973 to 1991 (33). In the province of Alberta, Canada, the mean survival following diagnosis of GBM has been reported to be 11.8 ⫾ 1.3 months, with 2.2% survival at 3 years following diagnosis (34). However, in this Alberta study, only 15 of 29 patients surviving 3 years were found to have true cases of GBM after expert histopathologic review. Similarly, a Swedish study of 48 cases of 4-year survivors following the diagnosis of GBM and anaplastic astrocytoma, histopathologic review found that none of the survivors had correct diagnoses of GBM, and only 8 of the 48 patients had correct diagnoses of anaplastic astrocytoma (35). The presence of a few long-term survivors who did not in
● L. PASZAT et al.
105
truth have GBM, out of a population of 3279 patients, would not influence the computation of median survival or the direction of adjusted RRs, but would overestimate the precision of the RRs. We did not find any improvement in overall survival during the years of the study. This is consistent with the lack of noteworthy survival improvements in the literature of clinical investigations in GBM during the study period. Survival did not vary between the Toronto area, where the majority of patients received 50 Gy, and the rest of Ontario, where the majority of patients received ⱖ 53.5 Gy. We have refrained from making any direct comparisons based on treatment and have not included treatment variables in Cox proportional hazards regression analysis because of the impossibility of controlling for treatment-selection variables. Given the universal health insurance available without charge to all residents of Ontario and the single governmental source of payment for all health services and facilities, it would seem unlikely that patients of GBM resident in Toronto would be diagnosed at an earlier point in the course of the disease compared to those in other large populous regions of the province, most of which are served by large academic health sciences centers. It is therefore unlikely that earlier diagnosis in Toronto GBM patients is compensating for inferior survival due to lower RT dose. The linked databases for this study do not contain information about the use of chemotherapy; however, during this study period, it was the practice in the Toronto region to refrain from the use of adjuvant chemotherapy in GBM. An interesting hypothesis is that, in reality, RT doses ⱖ 53.5 Gy do not prolong survival, compared to doses between 49.5 Gy and ⬍ 53.5 Gy. The study outcomes contain no information about fatal brain necrosis or neurologic morbidity related to RT for GBM. However, another interesting hypothesis is that the therapeutic ratio between any benefit of increasing RT dose and associated harms from increasing RT dose favors RT dose in the range between 49.5 Gy and ⬍ 53.5 Gy, in the real-world situation of typical patients treated by average clinicians in average institutions, as compared to patients who are treated in RCTs. Percent of survival time spent in the hospital The percentage of suvival time spent in the hospital is a nonspecific surrogate outcome measure of treatments, and, in particular, surrogate descriptors of performance status or health-related quality of life. Interregional practice variation in RT dose is not paralled by variation in the proportion of survival time spent in the hospital. Approximately 20% of all patients who receive high-dose RT spend the majority or all of the remainder of survival time between diagnosis and death in the hospital. General Studies of this design are valuable for describing broad issues on a population basis and for hypothesis generation,
106
I. J. Radiation Oncology
●
Biology
●
Physics
but cannot establish a causal link between treatment and outcome and are constrained by the limited number of covariates available in the data sources. By analogy with the clinical investigative reports supporting lower RT doses for poor-prognosis GBM, and given the lack of evidence for major survival improvements in recent RCTs, it would be appropriate to pursue further investigation of doses in the range between 49.5 Gy and ⬍ 53.5 Gy relative to higher doses, along with other procedures, such as sophisticated treatment-planning methods to minimize mordibity and mortality associated with RT.
Volume 51, Number 1, 2001
CONCLUSION We observed practice variation in the treatment of GBM according to age, region of residence, and year of diagnosis. Survival did not increase during the study period. Survival was superior in the cohort receiving RT (11 months), compared to the cohort that did not (3 months). The variation in RT dose between those receiving from 49.5 to ⱕ 53.5 Gy compared to ⬎ 53.5 Gy was not paralleled by variation in survival between regions, where one or the other of the dose ranges predominated, nor was variation in dose ranges among the regions paralleled by variation in the proportion of survival time spent in the hospital.
REFERENCES 1. Osoba D, Brada M, Yung WK, et al. Health-related quality of life in patients treated with temozolomide versus procarbazine for recurrent glioblastoma multiforme. J Clin Oncol 2000; 18(7):1481–1491. 2. Lovely MP, Miaskowski C, Dodd M. Relationship between fatigue and quality of life in patients with glioblastoma multiformae. Oncol Nurs Forum 1999;26(5):921–925. 3. Meyers CA, Hess KR, Yung WK, et al. Cognitive function as a predictor of survival in patients with recurrent malignant glioma. J Clin Oncol 2000;18(3):646 – 650. 4. Papsdorf K, Wolf U, Hildebrandt G, et al. Outcome and side effects in radiotherapy of glioblastoma—Conventional fractionation versus accelerated hyperfractionation. Front Radiat Ther Oncol 1999;33:158 –165. 5. Nelson DF, Curran WJ Jr, Scott C, et al. Hyperfractionated radiation therapy and bis-chlorethyl nitrosourea in the treatment of malignant glioma—possible advantage observed at 72.0 Gy in 1.2 Gy b.i.d. fractions: Report of the Radiation Therapy Oncology Group Protocol 8302. Int J Radiat Oncol Biol Phys 1993;25(2):193–207. 6. Laperriere NJ, Leung PM, McKenzie S, et al. Randomized study of brachytherapy in the initial management of patients with malignant astrocytoma. Int J Radiat Oncol Biol Phys 1998;41(5):1005–1011. 7. Mock U, Dieckmann K, Wolff U, et al. Comparison of coplanar and non-coplanar irradiation techniques for malignant glioma: An analysis of dose–volume histograms. Front Radiat Ther Oncol 1999;33:202–213. 8. Miralbell R, Mornex F, Greiner R, et al. Accelerated radiotherapy, carbogen, and nicotinamide in glioblastoma multiforme: Report of European Organization for Research and Treatment of Cancer trial 22933. J Clin Oncol 1999;17(10): 3143–3149. 9. Fine HA, Dear KB, Loeffler JS, et al. Meta-analysis of radiation therapy with and without adjuvant chemotherapy for malignant gliomas in adults. Cancer 1993;71(8):2585–2597. 10. Walker MD, Alexander E Jr, Hunt WE, et al. Evaluation of BCNU and/or radiotherapy in the treatment of anaplastic gliomas. A cooperative clinical trial. J Neurosurg 1978;49(3): 333–343. 11. Walker MD, Green SB, Byar DP, et al. Randomized comparisons of radiotherapy and nitrosoureas for the treatment of malignant glioma after surgery. N Engl J Med 1980;303(23): 1323–1329. 12. Sandberg-Wollheim M, Malmstrom P, Stromblad LG, et al. A randomized study of chemotherapy with procarbazine, vincristine, and lomustine with and without radiation therapy for astrocytoma grades 3 and/or 4. Cancer 1991;68(1):22–29. 13. Andersen AP. Postoperative irradiation of glioblastomas. Re-
14.
15.
16. 17. 18.
19. 20. 21.
22. 23.
24. 25. 26. 27.
sults in a randomized series. Acta Radiol Oncol Radiat Phys Biol 1978;17(6):475– 484. Kristiansen K, Hagen S, Kollevold T, et al. Combined modality therapy of operated astrocytomas grade III and IV. Confirmation of the value of postoperative irradiation and lack of potentiation of bleomycin on survival time: A prospective multicenter trial of the Scandinavian Glioblastoma Study Group. Cancer 1981;47(4):649 – 652. Bleehen NM, Stenning SP. A Medical Research Council trial of two radiotherapy doses in the treatment of grades 3 and 4 astrocytoma. The Medical Research Council Brain Tumour Working Party. Br J Cancer 1991;64(4):769 –774. Begg CB, Engstrom PF. Eligibility and extrapolation in cancer clinical trials. J Clin Oncol 1987;5(6):962–968. Yusuf S, Collins R, Peto R. Why do we need some large, simple randomized trials? Stat Med 3 1984;(4):409 – 422. Layde PM, Broste SK, Desbiens N, et al. Generalizability of clinical studies conducted at tertiary care medical centers: A population-based analysis. J Clin Epidemiol 1996;49(8):835– 841. Brook RH, Lohr KN. Efficacy, effectiveness, variations, and quality. Boundary-crossing research. Med Care 1985;23(5): 710 –722. Epstein RS, Sherwood LM. From outcomes research to disease management: A guide for the perplexed. Ann Intern Med 1996;124(9):832– 837. Clarke EA, Marrett LD, Krieger N. Cancer registration in Ontario: A computer approach. In: Jensen OM, Park DM, Maclennan R, et al., editors. Cancer registration: Principles and methods. Lyon, France: IARC Scientific Publications, 1991. p. 246 –257. Holowaty EJ, Dale D. The hospital-only project. Health Rep 1993;5:91–95. Morovan V, Lee G, Holowaty EJ, et al. A reabstraction study to estimate the completeness and accuracy of data elements in the Ontario Cancer Registry. A report to Health Canada. HC contract H4078-3-C098/01-SS, 1996. Mackillop WJ, Groome PA, Zhang-Solomons J, et al. Does a centralized radiotherapy system provide adequate access to care? J Clin Oncol 1997;15(3):1261–1271. Daumas-Duport C, Scheithauer B, O’Fallon J, et al. Grading of astrocytomas. A simple and reproducible method. Cancer 1988;62(10):2152–2165. Daumas-Duport C. Histological grading of gliomas. Curr Opin Neurol Neurosurg 1992;5(6):924 –931. Aldape K, Simmons ML, Davis RL, et al. Discrepancies in diagnoses of neuroepithelial neoplasms: The San Francisco Bay Area Adult Glioma Study. Cancer 2000;88(10):2342– 2349.
Population-based study of glioblastoma multiforme
28. Scott CB, Nelson JS, Farnan NC, et al. Central pathology review in clinical trials for patients with malignant glioma. A Report of Radiation Therapy Oncology Group 83-02. Cancer 1995;76(2):307–313. 29. Bauman GS, Gaspar LE, Fisher BJ, et al. A prospective study of short-course radiotherapy in poor prognosis glioblastoma multiforme. Int J Radiat Oncol Biol Phys 1994;29(4):835– 839. 30. Thomas R, James N, Guerrero D, et al. Hypofractionated radiotherapy as palliative treatment in poor prognosis patients with high grade glioma. Radiother Oncol 1994;33(2):113– 116. 31. Jeremic B, Shibamoto Y, Grujicic D, et al. Short-course radiotherapy in elderly and frail patients with glioblastoma multiforme. A phase II study. J Neurooncol 1999;44(1):85–90.
● L. PASZAT et al.
107
32. Polednak AP, Flannery JT. Brain, other central nervous system, and eye cancer. Cancer 1995;75(Suppl. 1):330 –337. 33. Davis FG, Freels S, Grutsch J, et al. Survival rates in patients with primary malignant brain tumors stratified by patient age and tumor histological type: An analysis based on Surveillance, Epidemiology, and End Results (SEER) data, 1973– 1991. J Neurosurg 1998;88 (1):1–10. 34. Scott JN, Rewcastle NB, Brasher PM, et al. Which glioblastoma multiforme patient will become a long-term survivor? A population-based study. Ann Neurol 1999;46(2): 183–188. 35. Ullen H, Mattsson B, Collins VP. Long-term survival after malignant glioma. A clinical and histopathological study on the accuracy of the diagnosis in a population-based cancer register. Acta Oncol 1990;29(7):875– 878.
PII S0360-3016(01)01572-3
CLINICAL INVESTIGATION
Brain
A POPULATION-BASED STUDY OF GLIOBLASTOMA MULTIFORME LAWRENCE PASZAT, M.D., M.S., F.R.C.P.C.,*†‡ NORMAND LAPERRIERE, M.D., F.R.C.P.C.,‡ PATTI GROOME, PH.D.,*§ KARLEEN SCHULZE, M.MATH.,* WILLIAM MACKILLOP, M.B., F.R.C.P.C.,*§ 㛳 AND ERIC HOLOWATY, M.D., M.SC., F.R.C.P.C. *Radiation Oncology Research Unit, Department of Oncology and §Department of Community Health and Epidemiology, Queen’s University, Kingston, Canada; †Institute for Clinical Evaluative Sciences, Toronto, Canada; ‡Department of Radiation Oncology, University of Toronto, Toronto, Canada; 㛳Cancer Surveillance Unit, Cancer Care Ontario and University of Toronto, Toronto, Canada Purpose: To describe (1) the use of surgery and radiotherapy (RT) in the treatment of patients with glioblastoma (GBM) in Ontario, (2) survival, and (3) proportion of survival time spent in the hospital after diagnosis. Methods and Materials: We performed a population-based cohort study of all Ontario Cancer Registry (OCR) cases of GBM diagnosed between 1982 and 1994. We linked OCR records, hospital files containing surgical procedure codes from the Canadian Institute for Health Information, and province-wide RT records. We studied the odds of treatment using multivariate logistic regression. We expressed the time spent in the hospital as the mean number of days per case, and as a proportion of the interval between diagnosis and death, or 24 months following diagnosis, whichever came first. We used the life-table method and Cox proportional hazards regression to describe survival. Results: The proportion of patients with GBM undergoing any surgery directed at the tumor varied with age (p < 0.0001) and region of residence (p < 0.0001). The proportion undergoing RT varied with age (p < 0.0001), region of residence (p < 0.0001), and year of diagnosis (p ⴝ 0.01). RT dose > 53.5 Gy varied with age (p < 0.0001), region of residence (p < 0.0001), and year of diagnosis (p ⴝ 0.0002). Median survival was 11 months among patients receiving RT and 3 months among those not receiving RT. The percentage of survival time spent in the hospital was similar among those who received from 49.5 to < 53.5 Gy, compared to > 53.5 Gy. Overall survival and the adjusted relative risk of death varied with age and region of residence. Conclusion: We observed practice variation in the treatment of patients with GBM according to age, region of residence, and year of diagnosis. Survival did not increase during the study period. The variation in RT dose between those receiving from 49.5 to < 53.5 Gy compared to > 53.5 Gy was not paralleled by variation in survival between regions where one or the other of the dose ranges predominated, nor was variation in dose ranges among the regions paralleled by variation in the proportion of survival time spent in the hospital. © 2001 Elsevier Science Inc. Glioblastoma, Population-based study, Radiation therapy, Outcomes research.
Glioblastoma multiforme (GBM) is a highly malignant form of brain tumor. Primary treatments include surgery and radiotherapy (RT). Contemporary practice varies with respect to (1) radiation dose, fractionation, planning and delivery, and (2) whether or not to use chemotherapy. GBM and its treatment may each contribute to declining performance status and quality of life. Health-related quality of life may improve transiently during therapy (1), and fatigue may increase and quality of life decrease after completion of therapy (2). Loss of verbal memory is independently associated with survival, adjusting for age and performance status (3).
Over several decades, investigators have attempted to improve survival by altered RT fractionation schemes (4), RT dose escalation (5), use of boosts by external RT or by brachytherapy (6), complex treatment-planning techniques (7), radiosensitizers (8), and chemotherapy (9). These studies have yielded rich information about prognostic factors, but randomized clinical trials have not demonstrated major improvements in patient outcomes. Several randomized clinical trials (RCTs) have shown that median survival is longer with RT than without RT (10 –14). Another RCT has shown that median survival is longer with 60 Gy than with 45 Gy (15); however, no additional improvement has been found in other studies of higher doses with or without hyperfractionation or acceler-
Correspondence to: Dr. Lawrence Paszat, Toronto Sunnybrook Regional Cancer Centre, 2075 Bayview Avenue, Toronto, Ontario, Canada M4N 3M5. E-mail: [email protected] This work was supported in part by an operating grant from Cancer Care Ontario to the Radiation Oncology Research Unit.
Acknowledgments—Dr. Paszat and Dr. Groome are career scientists of the Ministry of Health of Ontario. The radiation oncology departments of Ontario provided the radiation therapy data. Received Nov 9, 2000, and in revised form Feb 26, 2001. Accepted for publication Mar 20, 2001.
INTRODUCTION
100
Population-based study of glioblastoma multiforme
ated fractionation. Meta-analysis of RCTs, studying adjuvant chemotherapy combined with RT (9), demonstrates that any benefit would likely be restricted to the small subset of young patients with good performance status and minimal residual GBM postoperatively. Patients enrolled within clinical trials for their treatment often have better outcomes than patients treated outside of clinical trials. They are rigorously selected to be comparable, and their physicians, treating institutions, and treatment processes are subject to stringent guidelines (16). The treatment benefits seen in randomized clinical trials may be diminished when the treatment is implemented in the general population, because the spectrum of patients treated will be broader and there will be a broader range of treatment processes, physicians, and institutional expertise outside of the context of a randomized clinical trial (17, 18). In contrast to the efficacy of interventions demonstrated in controlled clinical experiments, the effectiveness of an intervention is the outcome of care provided under ordinary conditions by average practitioners for typical patients (19). Evidence of the effectiveness of an intervention is the extent to which the benefits are observed in the general population (20). We have performed a population-based cohort study of all patients with GBM residing in Ontario at the time of diagnosis between January 1, 1982 and December 31, 1994. The objectives of this study were to report (1) the use of surgery and radiotherapy (RT) for glioblastoma (GBM) in Ontario, (2) survival, and (3) time spent in the hospital after diagnosis. We will describe, in a complete and entirely unselected population, the utilization and effectiveness of interventions for which efficacy has been demonstrated in clinical trials. METHODS AND MATERIALS We performed a population-based cohort study of all patients with GBM residing in Ontario at the time of diagnosis between January 1, 1982 and December 31, 1994. The study is based on linked electronic databases from the Ontario Cancer Registry (OCR), the Canadian Institute for Health Information (CIHI), and the Radiation Oncology Research Unit (RORU) at Queen’s University, Kingston, Ontario, Canada. From the electronic files of the OCR, we identified all cases of ICD (International Classification of Diseases) 191 with the ICD-O morphology code for GBM (94403). The morphology codes are assigned to each pathology report received by the OCR by health records staff. Cases with morphology codes for anaplastic astrocytoma or any other histologic type of brain tumor were excluded. In the OCR electronic records, each case has a unique numeric identifier, as well as the following variables: diagnosis date, age at diagnosis, Ministry of Health residence code, postal code, vital status, numeric cancer center label, and cancer center chart number. The methods of case ascertainment by the OCR are
● L. PASZAT et al.
101
described by Clarke et al. (21), and the quality of case ascertainment by Holowaty et al. (22). The OCR ascertains cases on the basis of data from the following records originally created for other purposes: (1) registrations at cancer centers, (2) pathology reports received directly from acute care hospitals, (3) hospital discharge abstracts received from the Canadian Institute for Health Information (CIHI), and (4) death certificate diagnoses. Electronic hospital procedure and discharge abstracts from CIHI were previously labeled with the OCR unique numeric identifier for each case. The hospital records contain the admission date, discharge date, major treatment procedures, and ICD diagnoses for each admission. The accuracy of electronic records of major cancer-directed surgery in Ontario is described by Morovan et al. (23). Electronic RT records from all cancer centers in Ontario were compiled and processed by RORU and linked to the OCR and CIHI databases. The variables in the RT records used for linkage included the cancer center chart number and the numeric cancer center label. Each RT record includes the cancer center chart number, the date RT commenced, the anatomic region, the dose delivered, and the fraction number. The records contain no information concerning the size of the tumor or its location in the brain, pathologic features other than the histology code, the radiologic appearance of the tumor, the performance status of the cases, or of the use of chemotherapy. We categorized the surgical procedures directed at the brain tumor as (1) total or subtotal resection, (2) biopsy, and (3) lessor or no procedure. We selected RT records with appropriate anatomic region codes. We divided the province of Ontario into regions defined as the areas served by each cancer center. We assigned each census subdivision in Ontario to the cancer center most frequently visited by the residents of each census subdivision (24). In our analyses of treatment utilization, we will consider region of residence as a surrogate process indicator of practice and policy variations based on cancer centers and academic health sciences centers distributed across the province of Ontario. We described the population univariately by age at diagnosis, gender, region of residence, and year of diagnosis. We described the utilization of treatment procedures performed within 4 months of diagnosis by age, region of residence, and year of diagnosis. We tested the comparison of proportions using the 2 test. We modeled the odd ratios for treatment procedures using multivariate logistic regression. We described survival following diagnosis using the lifetable method and multivariable Cox proportional hazards regression analyses. We expressed the percentage of survival time spent in the hospital as the percentage of the time between the date of diagnosis and death interval between diagnosis and death spent in hospital, or, if a patient is still alive 24 months following diagnosis, the percentage of time between the
102
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 51, Number 1, 2001
Table 1. Population description and surgical treatment within 4 months of diagnosis Number of cases
Total or subtotal resection
Biopsy
No surgical procedure
81 219 414 733 1105 620 107
58 (71.6%) 153 (69.9%) 319 (74.9%) 541 (73.8%) 769 (69.6%) 391 (63.1%) 39 (36.5%)
11 (13.6%) 34 (15.5%) 71 (17.1%) 106 (14.5%) 209 (18.9%) 134 (21.6%) 35 (32.7%)
12 (14.8%) 32 (14.6%) 33 (8.0%) 86 (11.7%) 127 (11.5%) 95 (15.3%) 33 (30.8%)
356 1477 504 129 444 78 92 173 26
205 (57.6%) 1031 (69.8%) 353 (70.0%) 87 (67.4%) 321 (72.3%) 37 (47.4%) 67 (72.8%) 153 (88.4%) 7
67 (18.8%) 290 (19.6%) 115 (22.8%) 22 (17.1%) 74 (16.7%) 10 (12.8%) 11 (13.0%) 10 (5.8%) 0
84 (23.6%) 156 (10.6%) 36 (7.1%) 20 (15.5%) 49 (11.0%) 31 (39.7%) 13 (14.1%) 10 (5.8%) 19
852 1002 1425 2108
647 (75.9%) 714 (71.3%) 900 (63.2%) 1640
91 (10.8%) 155 (11.5%) 354 (24.2%) 295
114 (13.3%) 133 (13.3%) 171 (12.0%) 173
Age at diagnosis (years) 20–29 30–39 40–49 50–59 60–69 70–79 ⱖ 80 Region Ottawa Toronto Hamilton Kingston London NW Ontario Windsor NE Ontario Missing Year of diagnosis 1982–1985 1986–1989 1990–1994 Radiotherapy
date of diagnosis and the second anniversary of the diagnosis. RESULTS We identified 3279 cases of GBM diagnosed in Ontario between 1982 and 1994. Table 1 describes the population by age at diagnosis, region of residence, and year of diagnosis. Mean age at diagnosis is 59.3 years (standard deviation, 13.1), and median age is 61 years. Males comprise 59.0% of the population. Table 1 describes the utilization of surgical procedures directed at glioblastoma multiforme (GBM) within 4 months of diagnosis. The proportion of patients undergoing total or subtotal resection compared to biopsy alone varied with age (p ⬍ 0.0001), region of residence (p ⬍ 0.0001), and year of diagnosis (p ⬍ 0.0001). The proportion of patients undergoing any surgery directed at the tumor (including biopsy) compared to no surgery directed at the tumor varied with age (p ⬍ 0.0001) and region of residence (p ⬍ 0.0001), but not with year of diagnosis (p ⫽ 0.48). Table 1 describes the combinations of surgery and RT observed. Fifty percent of cases underwent total or subtotal resection plus RT, 37.2% of cases underwent surgery without RT, 5.3% of cases received RT without any surgery, and 7.5% of cases received neither surgery nor RT. Table 2 describes the utilization of RT within 4 months of diagnosis. The proportion of patients receiving RT within 4 months of diagnosis varied with age (p ⬍ 0.0001), region of residence (p ⬍ 0.0001), and year of diagnosis (p ⫽ 0.01). The mean dose was 48.8 Gy (median, 50 Gy). Mean fraction
number was 24 (median, 25). The most common dosefractionation combinations were 50 Gy in 25 fractions (37.6%), 60 Gy in 30 fractions (11.1%), and 54 Gy in 30 fractions (5.0%). The proportion of patients receiving RT dose ⬎ 53.5 Gy, compared with ⱕ 53.5 Gy varied with age (p ⬍ 0.0001), region of residence (p ⬍ 0.0001), and year of diagnosis (p ⫽ 0.0002). The majority of patients in the Toronto area received from 49.5 Gy to ⬍ 53.5 Gy, whereas the majority of patients elsewhere in Ontario received ⱖ 53.5 Gy. The RT records do not distinguish complete vs. incomplete treatment doses: these are all delivered doses. Linear accelerators were used to administer RT to 55.4% of patients, whereas 44.6% were treated with cobalt units. Table 3 presents the odds ratios and 95% confidence intervals from multiple logistic regression analyses on (1) receiving total or subtotal resection; (2) receiving RT; and (3) among those receiving RT, receiving dose ⱖ 53.5 Gy. For each of these dependent variables, there was variation according to age, region of residence, and year of diagnosis. Median survival was seven months from the date of diagnosis. Overall survival from the date of diagnosis by the life-table method was 29.4% (95% confidence interval [CI], 27.8 –30.9%) at 1 year, 11.1% (95% CI, 10.0 –12.1%) at 2 years, 7.4% (95% CI, 6.5– 8.3%) at 3 years, 5.5% (95% CI, 4.7– 6.3%) at 4 years, and 4.8% (95% CI, 4.0 –5.6%) at 5 years. Overall survival varied by age at diagnosis (Wilcoxon and log–rank tests each yielded p ⬍ 0.0001), by region of residence (Wilcoxon test, p ⬍ 0.0001; log–rank test, p ⫽ 0.0012) but not by year of diagnosis (Wilcoxon test, p ⫽ 0.13; log–rank test, p ⫽ 0.49). Median survival was eleven months among patients re-
Population-based study of glioblastoma multiforme
● L. PASZAT et al.
103
Table 2. Description of radiotherapy by dose
Age at diagnosis (years) 20–29 30–39 40–49 50–59 60–69 70–79 ⱖ 80 Region Ottawa Toronto Hamilton Kingston London NW Ontario Windsor NE Ontario Year of diagnosis 1982–1985 1986–1989 1990–1994
Radiotherapy
Dose ⬍39.5 Gy
Dose 39.5–⬍49.5 Gy
Dose 49.5–⬍53.5 Gy
Dose ⱖ 53.5 Gy
60 (74.1%) 168 (76.7%) 324 (78.3%) 555 (75.7%) 715 (64.7%) 265 (42.7%) 21 (19.6%)
3 (3.7%) 7 (8.6%) 29 (9.0%) 42 (7.6%) 115 (16.1%) 123 (46.4%) 11 (52.4%)
7 (8.6%) 7 (4.2%) 24 (7.4%) 39 (7.0%) 72 (10.1%) 32 (12.1%) 1 (4.8%)
29 (35.8%) 76 (45.2%) 147 (45.4%) 263 (47.4%) 302 (42.2%) 49 (18.5%) 7 (33.3%)
21 (25.9%) 78 (45.6%) 123 (38.0%) 210 (37.8%) 226 (31.6%) 60 (22.6%) 2 (9.5%)
174 (48.9%) 979 (66.3%) 305 (60.5%) 87 (68.2%) 304 (68.5%) 52 (66.7%) 75 (81.5%) 129 (74.6%)
35 (20.1%) 148 (15.1%) 39 (12.8%) 19 (21.6%) 60 (19.7%) 7 (13.5%) 8 (10.7%) 14 (10.9%)
15 (8.6%) 92 (9.4%) 17 (5.6%) 5 (5.8%) 28 (9.2%) 7 (13.5%) 11 (14.7%) 7 (5.4%)
23 (13.2%) 698 (71.3%) 53 (17.4%) 19 (21.8%) 13 (4.3%) 7 (13.5%) 3 (4.0%) 55 (42.6%)
101 (58.1%) 41 (4.2%) 194 (63.6%) 44 (50.6%) 203 (66.8%) 31 (59.5%) 53 (70.6%) 53 (41.1%)
549 (64.4%) 676 (67.5%) 883 (62.0%)
58 (10.6%) 77 (11.4%) 195 (22.1%)
68 (12.4%) 317 (3.7%) 89 (10.1%)
273 (49.7%) 320 (47.3%) 280 (31.7%)
149 (27.1%) 254 (37.6%) 317 (35.9%)
ceiving RT and 3 months among patients not receiving RT. Overall survival was 4.1% (95% CI, 3.1–5.0%) at 5 years among RT patients compared to 2.6% (95% CI, 1.6 –3.7%) among patients not receiving RT (p ⬍ 0.0001). The mean length of stay in hospital between the date of diagnosis and the date of death, or the second anniversary of diagnosis, declined from 90.7 days among patients diagnosed in 1982 to 53.7 days among patients diagnosed in 1994. Table 4 presents the percentage of survival time spent in hospital, according to age at diagnosis, surgical proce-
dure, use of RT, and RT dose. The percentage of survival time in hospital is defined as the time between date of diagnosis and date of death, or if alive at 2 years, between the date of diagnosis and the second anniversary of diagnosis. The percentage of survival time spent in hospital varied by age at diagnosis (age ⱕ 59 years compared to age ⱖ 60 years, p ⬍ 0.0001), by resection compared to biopsy (p ⬍ 0.0001), by the use of RT compared to no RT (p ⬍ 0.0001), and by RT dose ⬍ 49.5 Gy compared to ⱖ 49.5 Gy (p ⬍ 0.0001) among those who received RT. There was no dif-
Table 3. Odds ratios for total or subtotal resection, for radiotherapy (RT) and for RT dose ⬎53.5 Gy
Age at diagnosis (years) 20–29 30–39 40–49 50–59 60–69 70–79 ⱖ 80 Region Ottawa Toronto Hamilton Kingston London NW Ontario Windsor NE Ontario Year of diagnosis 1982–1985 1986–1989 1990–1994
Odds ratios for total or subtotal resection
Odds ratios for radiotherapy
1.00 (0.60, 1.66) 1.00 (0.72, 1.38) 1.26 (0.97, 1.64) 1.14 (0.92, 1.41) 1.00 0.77 (0.62, 0.95) 0.25 (0.17, 0.38)
1.59 (0.95, 2.67) 1.80 (1.28, 2.54) 2.0 (1.53, 2.61) 1.69 (1.36, 2.09) 1.00 0.40 (0.32, 0.49) 0.13 (0.08, 0.21)
0.58 (0.45, 0.73) 1.00 1.00 (0.80, 1.26) 0.86 (0.59, 1.28) 1.20 (0.94, 1.52) 0.38 (0.27, 0.60) 1.22 (0.76, 1.98) 3.10 (1.91, 5.02)
0.50 (0.39, 0.65) 1.00 0.77 (0.62, 0.96) 1.05 (0.70, 1.57) 1.29 (1.01, 1.64) 1.10 (0.67, 1.83) 2.84 (1.62, 5.00) 1.30 (0.90, 1.88)
1.00 0.79 (0.64, 0.98) 0.54 (0.45, 0.66)
1.00 1.25 (1.02, 1.54) 1.00 (0.83, 1.21)
Odds ratios for dose ⬎53.5 Gy 1.86 (0.90, 3.85) 3.03 (1.87, 4.90) 2.78 (1.24, 2.57) 1.87 (1.36, 2.53) 1.00 0.43 (0.30, 0.64) 0.11 (0.02, 0.50) 38.52 (24.53, 60.50) 1.00 51.40 (34.17, 77.31) 27.74 (16.10, 77.31) 58.53 (38.76, 88.38) 47.26 (24.06, 92.82) 91.94 (48.81, 173.18) 15.98 (9.87, 25.88) 1.00 2.13 (1.56, 2.91) 2.11 (1.56, 2.84)
104
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 51, Number 1, 2001
Table 4. Percentage of survival time spent in hospital
Age at diagnosis (years) 20–29 30–39 40–49 50–59 60–69 70–79 ⱖ 80 Surgical procedure Resection Biopsy None Radiotherapy Yes No Radiotherapy dose ⬍39.5 Gy 39.5–⬍49.5 Gy 49.5–⬍53.5 Gy ⱖ 53.5 Gy Year of diagnosis 1982–1985 1986–1989 1990–1994 Overall
⬍10%
10–⬍25%
25–⬍50%
50–⬍100%
100%
61.7% 59.5% 36.0% 24.6% 14.0% 9.4% 3.8%
25.9% 19.3% 27.1% 26.3% 20.8% 14.9% 10.5%
6.2% 9.6% 14.7% 22.3% 20.0% 16.3% 9.5%
3.7% 6.4% 12.1% 16.2% 23.4% 21.4% 26.7%
2.5% 5.5% 10.1% 10.6% 21.9% 38.0% 49.5%
22.1% 16.7% 31.1%
23.6% 16.2% 17.3%
18.5% 16.7% 15.8%
17.9% 21.3% 17.8%
18.0% 29.2% 18.1%
27.1% 13.3%
27.3% 10.6%
21.0% 11.9%
17.0% 21.3%
7.6% 42.9%
12.1% 20.7% 30.9% 31.0%
15.8% 25.0% 29.3% 31.0%
23.9% 23.9% 20.2% 19.8%
30.6% 17.9% 14.2% 13.8%
17.6% 12.5% 5.4% 4.5%
17.7% 20.6% 27.3% 22.2%
21.7% 19.8% 22.0% 21.4%
19.0% 17.7% 16.9% 17.8%
17.7% 20.6% 17.2% 18.5%
23.8% 21.6% 16.6% 20.0%
ference in the percentage of survival time spent in hospital between those who received from 49.5 Gy to ⬍ 53.5 Gy and those who received ⱖ 53.5 Gy. Table 5 presents the relative risk (RR) of death following diagnosis of GBM. The RR of death varied with age at diagnosis, and in one region of Ontario relative to the reference region, and did not vary by year of diagnosis.
Table 5. adjusted relative risk of death Relative risk of death Age at diagnosis (years) 20–29 30–39 40–49 50–59 60–69 70–79 ⱖ 80 Region Ottawa Toronto Hamilton Kingston London NW Ontario Windsor NE Ontario Year of diagnosis 1982–1985 1986–1989 1990–1994
0.22 (0.17, 0.29) 0.30 (0.25, 0.35) 0.52 (0.46, 0.59) 0.71 (0.65, 0.79) 1.00 1.70 (1.54, 1.88) 2.65 (2.16, 3.25) 1.26 (1.12, 1.42) 1.00 1.08 (0.97, 1.20) 1.19 (0.99, 1.43) 1.01 (0.91, 1.13) 0.93 (0.73, 1.17) 1.12 (0.91, 1.40) 0.91 (0.77, 1.07) 1.00 1.00 (0.91, 1.09) 1.00 (0.92, 1.10)
DISCUSSION Diagnosis of GBM The median survival rates described in this populationbaed cohort are typical of those described in the literature. 12.7% of patients in this cohort identified by the ICD-O morphology code for GBM (99403) did not have a record of open biopsy or resection. Electronic records of surgery might be missing for some of the patients, others might have had a more limited biopsy procedure, and the remainder might have had a radiologic diagnosis of GBM. The morphology codes are based on paper histopathology reports from dozens of hospitals and pathologists, in academic as well as community settings, and are subject to interobserver variation as described in the clinical literature. Despite some reports of very high reproducibility of some classification systems of malignant glial brain tumors (25, 26), unreliability is likely present. For example, Aldape et al. reported 89% concordance between original histopathologic diagnoses and diagnoses given after expert review of cases of GBM in the San Francisco area. The diagnosis changed from GBM to another histology in 13 of 253 patients, and that, among initial diagnoses of anaplastic astrocytoma, 20 of 57 patients initially labeled as anaplastic astrocytoma were labeled GBM by expert review (27). Central histopathology review by the Radiation Therapy Oncology Group resulted in 96% agreement on the diagnosis of GBM between initial and central review, but 34% of anaplastic astrocytomas on initial review were subsequently diagnosed as glioblastoma on central review (28).
Population-based study of glioblastoma multiforme
Treatment variations–surgery The percentage of patients for whom the maximal surgical procedure was biopsy increased from 10.6% in 1982–1985 to 22.1% in 1990 –1994, and the percentage undergoing total or subtotal resection decreased from 75.9% to 63.2%. During the same interval, the percentage of patients without a record of surgery remained unchanged. Patients older than 70 years of age were less likely to have resection and more likely to have biopsy or no record of surgery. Treatment variations—RT A total of 63.4% of all patients received RT, and 75.6% of those received ⱖ 49.5 Gy. Only a minority of patients aged 70 years or older received RT, and, of those who did, the majority received ⬍ 39.5 Gy. Lower dose palliative RT has been recommended in recent clinical literature as appropriate therapy for those with poor prognosis GBM, which includes patients aged 70 or older (29 –31). Comparing RT dose in the Toronto area to RT dose in the rest of Ontario: the RT dose was between 49.5 Gy and ⬍ 53.5 Gy for 698/979 (71.3%) of patients in the Toronto receiving RT, whereas the dose was ⱖ 53.5 Gy for 679/ 1178 (57.6%) for patients receiving RT in the rest of Ontario. This is a remarkable practice variation. Interpreting this variation is difficult. We do not know if the clinical indications for RT used in the Toronto area were different from those in the rest of Ontario; however, the adjusted odds ratios for the use of RT within 4 months of diagnosis (Table 3) show lower odds of RT among patients living in the Ottawa and Hamilton regions, and higher odds among patients living in the London and Windsor areas, relative to Toronto. The systematic difference in dose may provide the context of a natural experiment and suggest hypotheses about the outcomes associated with populous regions, where divergent doses prevailed. Survival The overall survival experience we report for this cohort is higher than expected, although Polednak and Flannery reported 5-year relative survival of 4% for patients diagnosed 1983–1987 in Surveillance, Epidemiology, and End Results (SEER) data (32). Davis et al. reported 2- and 5-year overall survival as 1% in a cohort study of the SEER data from 1973 to 1991 (33). In the province of Alberta, Canada, the mean survival following diagnosis of GBM has been reported to be 11.8 ⫾ 1.3 months, with 2.2% survival at 3 years following diagnosis (34). However, in this Alberta study, only 15 of 29 patients surviving 3 years were found to have true cases of GBM after expert histopathologic review. Similarly, a Swedish study of 48 cases of 4-year survivors following the diagnosis of GBM and anaplastic astrocytoma, histopathologic review found that none of the survivors had correct diagnoses of GBM, and only 8 of the 48 patients had correct diagnoses of anaplastic astrocytoma (35). The presence of a few long-term survivors who did not in
● L. PASZAT et al.
105
truth have GBM, out of a population of 3279 patients, would not influence the computation of median survival or the direction of adjusted RRs, but would overestimate the precision of the RRs. We did not find any improvement in overall survival during the years of the study. This is consistent with the lack of noteworthy survival improvements in the literature of clinical investigations in GBM during the study period. Survival did not vary between the Toronto area, where the majority of patients received 50 Gy, and the rest of Ontario, where the majority of patients received ⱖ 53.5 Gy. We have refrained from making any direct comparisons based on treatment and have not included treatment variables in Cox proportional hazards regression analysis because of the impossibility of controlling for treatment-selection variables. Given the universal health insurance available without charge to all residents of Ontario and the single governmental source of payment for all health services and facilities, it would seem unlikely that patients of GBM resident in Toronto would be diagnosed at an earlier point in the course of the disease compared to those in other large populous regions of the province, most of which are served by large academic health sciences centers. It is therefore unlikely that earlier diagnosis in Toronto GBM patients is compensating for inferior survival due to lower RT dose. The linked databases for this study do not contain information about the use of chemotherapy; however, during this study period, it was the practice in the Toronto region to refrain from the use of adjuvant chemotherapy in GBM. An interesting hypothesis is that, in reality, RT doses ⱖ 53.5 Gy do not prolong survival, compared to doses between 49.5 Gy and ⬍ 53.5 Gy. The study outcomes contain no information about fatal brain necrosis or neurologic morbidity related to RT for GBM. However, another interesting hypothesis is that the therapeutic ratio between any benefit of increasing RT dose and associated harms from increasing RT dose favors RT dose in the range between 49.5 Gy and ⬍ 53.5 Gy, in the real-world situation of typical patients treated by average clinicians in average institutions, as compared to patients who are treated in RCTs. Percent of survival time spent in the hospital The percentage of suvival time spent in the hospital is a nonspecific surrogate outcome measure of treatments, and, in particular, surrogate descriptors of performance status or health-related quality of life. Interregional practice variation in RT dose is not paralled by variation in the proportion of survival time spent in the hospital. Approximately 20% of all patients who receive high-dose RT spend the majority or all of the remainder of survival time between diagnosis and death in the hospital. General Studies of this design are valuable for describing broad issues on a population basis and for hypothesis generation,
106
I. J. Radiation Oncology
●
Biology
●
Physics
but cannot establish a causal link between treatment and outcome and are constrained by the limited number of covariates available in the data sources. By analogy with the clinical investigative reports supporting lower RT doses for poor-prognosis GBM, and given the lack of evidence for major survival improvements in recent RCTs, it would be appropriate to pursue further investigation of doses in the range between 49.5 Gy and ⬍ 53.5 Gy relative to higher doses, along with other procedures, such as sophisticated treatment-planning methods to minimize mordibity and mortality associated with RT.
Volume 51, Number 1, 2001
CONCLUSION We observed practice variation in the treatment of GBM according to age, region of residence, and year of diagnosis. Survival did not increase during the study period. Survival was superior in the cohort receiving RT (11 months), compared to the cohort that did not (3 months). The variation in RT dose between those receiving from 49.5 to ⱕ 53.5 Gy compared to ⬎ 53.5 Gy was not paralleled by variation in survival between regions, where one or the other of the dose ranges predominated, nor was variation in dose ranges among the regions paralleled by variation in the proportion of survival time spent in the hospital.
REFERENCES 1. Osoba D, Brada M, Yung WK, et al. Health-related quality of life in patients treated with temozolomide versus procarbazine for recurrent glioblastoma multiforme. J Clin Oncol 2000; 18(7):1481–1491. 2. Lovely MP, Miaskowski C, Dodd M. Relationship between fatigue and quality of life in patients with glioblastoma multiformae. Oncol Nurs Forum 1999;26(5):921–925. 3. Meyers CA, Hess KR, Yung WK, et al. Cognitive function as a predictor of survival in patients with recurrent malignant glioma. J Clin Oncol 2000;18(3):646 – 650. 4. Papsdorf K, Wolf U, Hildebrandt G, et al. Outcome and side effects in radiotherapy of glioblastoma—Conventional fractionation versus accelerated hyperfractionation. Front Radiat Ther Oncol 1999;33:158 –165. 5. Nelson DF, Curran WJ Jr, Scott C, et al. Hyperfractionated radiation therapy and bis-chlorethyl nitrosourea in the treatment of malignant glioma—possible advantage observed at 72.0 Gy in 1.2 Gy b.i.d. fractions: Report of the Radiation Therapy Oncology Group Protocol 8302. Int J Radiat Oncol Biol Phys 1993;25(2):193–207. 6. Laperriere NJ, Leung PM, McKenzie S, et al. Randomized study of brachytherapy in the initial management of patients with malignant astrocytoma. Int J Radiat Oncol Biol Phys 1998;41(5):1005–1011. 7. Mock U, Dieckmann K, Wolff U, et al. Comparison of coplanar and non-coplanar irradiation techniques for malignant glioma: An analysis of dose–volume histograms. Front Radiat Ther Oncol 1999;33:202–213. 8. Miralbell R, Mornex F, Greiner R, et al. Accelerated radiotherapy, carbogen, and nicotinamide in glioblastoma multiforme: Report of European Organization for Research and Treatment of Cancer trial 22933. J Clin Oncol 1999;17(10): 3143–3149. 9. Fine HA, Dear KB, Loeffler JS, et al. Meta-analysis of radiation therapy with and without adjuvant chemotherapy for malignant gliomas in adults. Cancer 1993;71(8):2585–2597. 10. Walker MD, Alexander E Jr, Hunt WE, et al. Evaluation of BCNU and/or radiotherapy in the treatment of anaplastic gliomas. A cooperative clinical trial. J Neurosurg 1978;49(3): 333–343. 11. Walker MD, Green SB, Byar DP, et al. Randomized comparisons of radiotherapy and nitrosoureas for the treatment of malignant glioma after surgery. N Engl J Med 1980;303(23): 1323–1329. 12. Sandberg-Wollheim M, Malmstrom P, Stromblad LG, et al. A randomized study of chemotherapy with procarbazine, vincristine, and lomustine with and without radiation therapy for astrocytoma grades 3 and/or 4. Cancer 1991;68(1):22–29. 13. Andersen AP. Postoperative irradiation of glioblastomas. Re-
14.
15.
16. 17. 18.
19. 20. 21.
22. 23.
24. 25. 26. 27.
sults in a randomized series. Acta Radiol Oncol Radiat Phys Biol 1978;17(6):475– 484. Kristiansen K, Hagen S, Kollevold T, et al. Combined modality therapy of operated astrocytomas grade III and IV. Confirmation of the value of postoperative irradiation and lack of potentiation of bleomycin on survival time: A prospective multicenter trial of the Scandinavian Glioblastoma Study Group. Cancer 1981;47(4):649 – 652. Bleehen NM, Stenning SP. A Medical Research Council trial of two radiotherapy doses in the treatment of grades 3 and 4 astrocytoma. The Medical Research Council Brain Tumour Working Party. Br J Cancer 1991;64(4):769 –774. Begg CB, Engstrom PF. Eligibility and extrapolation in cancer clinical trials. J Clin Oncol 1987;5(6):962–968. Yusuf S, Collins R, Peto R. Why do we need some large, simple randomized trials? Stat Med 3 1984;(4):409 – 422. Layde PM, Broste SK, Desbiens N, et al. Generalizability of clinical studies conducted at tertiary care medical centers: A population-based analysis. J Clin Epidemiol 1996;49(8):835– 841. Brook RH, Lohr KN. Efficacy, effectiveness, variations, and quality. Boundary-crossing research. Med Care 1985;23(5): 710 –722. Epstein RS, Sherwood LM. From outcomes research to disease management: A guide for the perplexed. Ann Intern Med 1996;124(9):832– 837. Clarke EA, Marrett LD, Krieger N. Cancer registration in Ontario: A computer approach. In: Jensen OM, Park DM, Maclennan R, et al., editors. Cancer registration: Principles and methods. Lyon, France: IARC Scientific Publications, 1991. p. 246 –257. Holowaty EJ, Dale D. The hospital-only project. Health Rep 1993;5:91–95. Morovan V, Lee G, Holowaty EJ, et al. A reabstraction study to estimate the completeness and accuracy of data elements in the Ontario Cancer Registry. A report to Health Canada. HC contract H4078-3-C098/01-SS, 1996. Mackillop WJ, Groome PA, Zhang-Solomons J, et al. Does a centralized radiotherapy system provide adequate access to care? J Clin Oncol 1997;15(3):1261–1271. Daumas-Duport C, Scheithauer B, O’Fallon J, et al. Grading of astrocytomas. A simple and reproducible method. Cancer 1988;62(10):2152–2165. Daumas-Duport C. Histological grading of gliomas. Curr Opin Neurol Neurosurg 1992;5(6):924 –931. Aldape K, Simmons ML, Davis RL, et al. Discrepancies in diagnoses of neuroepithelial neoplasms: The San Francisco Bay Area Adult Glioma Study. Cancer 2000;88(10):2342– 2349.
Population-based study of glioblastoma multiforme
28. Scott CB, Nelson JS, Farnan NC, et al. Central pathology review in clinical trials for patients with malignant glioma. A Report of Radiation Therapy Oncology Group 83-02. Cancer 1995;76(2):307–313. 29. Bauman GS, Gaspar LE, Fisher BJ, et al. A prospective study of short-course radiotherapy in poor prognosis glioblastoma multiforme. Int J Radiat Oncol Biol Phys 1994;29(4):835– 839. 30. Thomas R, James N, Guerrero D, et al. Hypofractionated radiotherapy as palliative treatment in poor prognosis patients with high grade glioma. Radiother Oncol 1994;33(2):113– 116. 31. Jeremic B, Shibamoto Y, Grujicic D, et al. Short-course radiotherapy in elderly and frail patients with glioblastoma multiforme. A phase II study. J Neurooncol 1999;44(1):85–90.
● L. PASZAT et al.
107
32. Polednak AP, Flannery JT. Brain, other central nervous system, and eye cancer. Cancer 1995;75(Suppl. 1):330 –337. 33. Davis FG, Freels S, Grutsch J, et al. Survival rates in patients with primary malignant brain tumors stratified by patient age and tumor histological type: An analysis based on Surveillance, Epidemiology, and End Results (SEER) data, 1973– 1991. J Neurosurg 1998;88 (1):1–10. 34. Scott JN, Rewcastle NB, Brasher PM, et al. Which glioblastoma multiforme patient will become a long-term survivor? A population-based study. Ann Neurol 1999;46(2): 183–188. 35. Ullen H, Mattsson B, Collins VP. Long-term survival after malignant glioma. A clinical and histopathological study on the accuracy of the diagnosis in a population-based cancer register. Acta Oncol 1990;29(7):875– 878.