Combined-Modality Therapy With Radiation and Chemotherapy for Elderly Patients With Glioblastoma in the Temozolomide Era: A National Cancer Database Analysis

Importance  The optimal management for elderly patients with glioblastoma (GBM) is controversial. Following maximal safe resection or biopsy, accepted treatment paradigms for elderly patients with GBM include combined-modality therapy (CMT) with both radiotherapy (RT) and chemotherapy (CT), RT alone, and CT alone.

Objective  To evaluate the overall survival (OS) outcomes associated with RT, CT, and CMT for elderly patients with GBM in the modern temozolomide era.

Design, Setting, and Participants  In this retrospective cohort study of a prospectively maintained, multi-institutional national cancer registry, the National Cancer Database was queried for elderly patients (≥65 years) with newly diagnosed GBM from January 1, 2005, through December 31, 2011, with complete data sets for RT, CT, tumor resection, Charlson-Deyo comorbidity scores, age, sex, and year of diagnosis. Data analysis was performed from October 2015 through December 2015.

Interventions  Combined-modality therapy, RT, CT.

Main Outcomes and Measures  Survival by treatment cohort was estimated using the Kaplan-Meier method and analyzed using the log rank test, univariate and multivariate Cox models, and propensity score–matched analyses.

Results  A total of 16 717 patients (median [range] age, 73 [65-≥90 y]; 8870 [53%] male) were identified. The median OS by treatment was 9.0 (95% CI, 8.8-9.3) months with CMT (8435 patients), 4.7 (95% CI, 4.5-5.0) months with RT alone (1693 patients), 4.3 (95% CI, 4.0-4.7) months with CT alone (1018 patients), and 2.8 (95% CI, 2.8-2.9) months with no therapy (5571 patients) (P < .001). On multivariate analysis, CMT was superior to both CT alone (hazard ratio, 1.50 [95% CI, 1.40-1.60]; P < .001) and RT alone (hazard ratio, 1.47 [95% CI, 1.39-1.55]; P < .001), whereas no differences were observed between CT alone vs RT alone (P = .60). Propensity score–matched analyses redemonstrated improved OS with CMT over CT alone (P = .002) and RT alone (P < .001); no differences were observed between CT alone vs RT alone (P = .44). On subgroup analyses, a consistent OS advantage was observed with CMT over both CT alone and RT alone across each age stratification (65-69, 70-74, 75-79, and ≥80 years) and among patients treated with or without tumor resection (all P < .001).

Conclusions and Relevance  In this analysis of multimodality therapy for elderly patients with GBM, OS was superior with CMT compared with CT alone and RT alone. Survival was similar between CT alone and RT alone, and both CT alone and RT alone were superior to no therapy. This analysis supports the use of CMT for suitable elderly candidates.

For elderly patients (≥65 years) with newly diagnosed glioblastoma (GBM), the optimal integration of radiotherapy (RT) and chemotherapy (CT) represents an ongoing controversy in the field of neuro-oncology. Although level I evidence exists to support an overall survival (OS) advantage of RT over best supportive care (BSC),¹ 2 randomized trials evaluating elderly populations have demonstrated no advantage of RT alone over CT alone with temozolomide, and subgroup analyses from each of these trials suggest that CT may be the more important single-modality therapy for patients with O⁶-methylguanine-DNA methyltransferase (MGMT) promoter-methylated tumors.^2,3 Nevertheless, elderly patients who present with a favorable performance status are often suitable candidates for combined-modality therapy (CMT) with both RT and CT, which is offered as the standard of care for younger patients.^4-6

At present, no randomized data are available for elderly patients with GBM to clarify the effect on OS of CMT vs RT alone or CT alone. In this analysis, using data from the National Cancer Database (NCDB) we evaluated associated OS outcomes for CMT, RT, and CT for elderly patients with GBM in the modern temozolomide era.

Box Section Ref ID

The NCDB is a hospital-based cancer registry that collects data from American College of Surgeons–Commission on Cancer accredited facilities. The database is sponsored by the American College of Surgeons and the American Cancer Society and includes approximately 70% of all malignant cancers diagnosed in the United States.⁷ Patient demographic characteristics, comorbidities, tumor characteristics, and overall survival (OS) are recorded, as well as therapies delivered during the first course of treatment and prior to any documented disease progression including surgery, RT, and CT. This analysis was performed with the approval of the University of Colorado School of Medicine institutional review board.

The NCDB was queried for patients 65 years of age or older³ with newly diagnosed GBM, with complete data sets for RT, CT, surgery, Charlson-Deyo comorbidity score, age, sex, and year of diagnosis. Molecular data including MGMT status were not available for analysis. The search was confined to the years 2005 or later in an effort to limit analyses to the temozolomide era for GBM management. The landmark EORTC-NCIC trial, published in 2005, demonstrated improved OS with the addition of temozolomide to RT and established temozolomide as the standard CT agent for adult GBM.⁶ It is important to note, however, that data regarding specific CT agents administered are not available in the NCDB. Patterns of care analyses suggest that temozolomide-based regimens may have been administered in approximately 97% of patients receiving systemic CT for GBM even during the early years of this study interval.^8,9 Cases in which death occurred within 1 month of diagnosis were omitted from analysis. Therapeutic RT was defined as 10 to 35 fractions (Fx) to coincide with the minimum Fx number to show equivalence to the standard regimen of 60 Gy delivered over 30 Fx in a randomized trial specific to elderly patients.² Radiotherapy to 9 Fx or less was considered palliative and was not used to define treatment cohort status. For analysis, patients were divided into treatment cohorts including CMT with both RT and CT, CT alone, RT alone, and no therapy.

Survival was estimated using the Kaplan-Meier method. Univariate comparisons were performed using the log rank test and unadjusted Cox proportional hazards models. Multivariate comparisons were made using Cox models adjusted for variables selected a priori including comorbidity score, age, sex, race, year of diagnosis, and tumor resection. The proportional hazards assumption was assessed for covariates in all final models and returned no significant results.^{10(pp161-200)} Median follow-up was calculated using the reverse Kaplan-Meier method.

To further account for confounding variables, propensity score–matched analyses were performed to compare the survival outcomes among the 3 treatment cohorts (CMT vs CT alone, CMT vs RT alone, and CT alone vs RT alone). Significance levels for pairwise propensity score–matched comparisons were corrected using the Bonferroni method to reduce the likelihood of a type I error.¹¹ One-to-one matching without replacement was completed using the nearest-neighbor match on the logit of the propensity score for treatment approach (derived from age, sex, race, year, comorbidity, and tumor resection status). The caliper width was 0.05 × the standard deviation of the logit of the propensity score,¹² which is estimated to eliminate more than 99% of the bias due to confounding variables.¹³

Our search returned 16 717 patients at least 65 years old with newly diagnosed GBM from 2005 to 2011. The median (range) age was 73 (65-≥90) years, and 8870 (53%) were male. A total of 8435 (50%) patients received CMT, 1693 (10%) received RT alone, 1018 (6%) received CT alone, and 5571 (33%) received no therapy. Treatment patterns over time are displayed in Figure 1 (corresponding percentages are displayed in eTable 1 in the Supplement). The median and mode number of RT Fx was 30 (range, 10-35 Fx). Patient characteristics are displayed in Table 1. Increases in CMT administration were observed in the setting of tumor resections, younger age, male sex, white race, lower comorbidity scores, and during the latter years of the study interval.

At last follow-up, 15 677 patients (94%) were deceased. The median (range) follow-up was 49.5 (1-105) months. Kaplan-Meier OS estimates are displayed in Figure 2. The median OS estimates were 9.0 (95% CI, 8.8-9.3) months for patients receiving CMT, 4.7 (95% CI, 4.5-5.0) months with RT alone, 4.3 (95% CI, 4.0-4.7) months with CT alone, and 2.8 (95% CI, 2.8-2.9) months with no therapy (P < .001). On univariate survival analysis, when compared with CMT, the hazard of mortality was increased for patients treated with RT alone (HR, 1.73 [95% CI, 1.64-1.82]; P < .001), CT alone (HR, 1.56 [95% CI, 1.46-1.67]; P < .001), and no therapy (HR, 2.59 [95% CI, 2.50-2.68]; P < .001).

On multivariate survival analyses adjusted for tumor resection, comorbidity scores, age, sex, race, and year, CMT remained superior to RT alone (HR, 1.47 [95% CI, 1.39-1.55]; P < .001), CT alone (HR, 1.50 [95% CI, 1.40-1.60]; P < .001), and no therapy (HR, 2.15 [95% CI, 2.07-2.23]; P < .001) (Table 2). No significant differences were observed between CT alone and RT alone (RT alone HR, 0.98 [95% CI, 0.90-1.06]; P = .60). In the same model, when compared with no therapy, a reduction in the hazard of mortality was observed with CT alone (HR, 0.70 [95% CI, 0.65-0.75]; P < .001), RT alone (HR, 0.68 [95% CI, 0.65-0.72]; P < .001), and CMT (HR, 0.47 [95% CI, 0.45-0.48]; P < .001) (eTable 2 in the Supplement).

Propensity score–matched analyses with one-to-one matching were performed to compare the survival outcomes within the 3 treatment cohorts. Patient characteristics were well balanced overall with the exception of small differences in race (eTables 3-5 in the Supplement). The corrected α value for significance was set at P = .017. On comparison of 1017 CT alone with 1017 matched CMT patients, the hazard of mortality remained increased for patients treated with CT alone (HR, 1.15 [95% CI, 1.05-1.26]; P = .002) vs those treated with CMT. Similarly, on comparison of 1643 RT alone with 1643 matched CMT patients, the hazard of mortality remained increased for patients treated with RT alone (HR, 1.21 [95% CI, 1.12-1.29]; P < .001) vs those treated with CMT. On comparison of 977 CT alone with 977 matched RT alone patients, no differences in OS were observed (RT alone HR, 0.97 [95% CI, 0.88-1.06]; P = .44).

Subgroup analyses evaluating the effect of treatment were performed for stratifications by age (65-69, 70-74, 75-79, and ≥80 years) and tumor resection (eTable 6 in the Supplement). When stratified by age, the magnitude of the observed OS advantage with CMT over single-modality therapy remained consistent across all age subgroups (HRs with RT alone or CT alone, range 1.44-1.73, all P < .001). Kaplan-Meier OS estimates by treatment for each age subgroup are displayed in Figure 3. When stratified by tumor resection, the observed OS advantage with CMT over single-modality therapy was also consistent for patients treated with and without tumor resection (HRs with RT alone or CT alone, range 1.46-1.59, all P < .001). No significant differences were observed between CT alone and RT alone within any age or tumor resection subgroup.

The optimal management for elderly patients with newly diagnosed GBM represents an ongoing controversy in the field of neuro-oncology.¹⁴ Although the survival advantages of RT^15,16 and later temozolomide CT^6,17 have been established by randomized trials reported over the past 4 decades, the benefit of CMT over single-modality therapy remains unclear in aging populations. Early studies from the 1970s and 1980s demonstrated a consistent survival advantage with RT among trial-eligible patients,^15,16 establishing adjuvant RT as the standard of care for nonelderly patients who presented with a favorable performance status. In 2005, the landmark EORTC-NCIC trial⁶ demonstrated improved OS with temozolomide CT in addition to RT and established CMT as the new standard of care for nonelderly patients; although, individuals older than 70 years were notably excluded from this trial. In 2007, a French cooperative (ANOCEF) randomized trial limited to patients 70 years or older demonstrated an unequivocal OS advantage with RT over BSC, providing the first level I evidence for RT specific to elderly patients.¹ Subsequently, 2 randomized trials comparing single-modality strategies for patients older than 65 years demonstrated no significant OS advantage of RT alone over CT alone with temozolomide,^2,3 introducing new controversies to the management paradigms for this population. At present, no randomized data are available to clarify the effect on OS of CMT over single-modality strategies with either RT alone or CT alone among elderly patients who are candidates for multimodality therapy.

In this analysis, we report the outcomes for more than 16 000 patients in the United States 65 years or older with newly diagnosed GBM who were treated with varying combinations of RT, CT, and surgery in the modern temozolomide era. In this population, a consistent OS advantage was observed for patients treated with CMT over single-modality therapy with RT alone or CT alone. The benefits of CMT remained significant in multivariate models and propensity score–matched analyses controlling for important clinical variables notably including age, comorbidity scores, and tumor resection. Subgroup analyses demonstrated a consistent OS benefit with CMT over single-modality therapy across sequential age stratifications (65-69, 70-74, 75-79, and ≥80 years) and tumor resection subgroups. Conversely, no significant differences in OS were observed in multivariate, propensity score–matched analyses, or subgroup analyses between the single-modality treatment strategies involving RT alone vs CT alone.

To our knowledge, this analysis represents the largest report of multimodality therapy outcomes specific to elderly patients with GBM in the temozolomide era. To guide the management of this population, which has historically been underrepresented in randomized trials,¹⁸ the National Comprehensive Cancer Network stratifies contemporary recommendations on the basis of performance status.⁴ Recommended therapies for patients with a favorable Karnofsky performance status (KPS, ≥60) include CMT (ie, RT and temozolomide), RT alone, and temozolomide alone, whereas patients with an unfavorable performance status (KPS, <60) are recommended to receive single-modality therapy with RT alone, temozolomide alone, or BSC. Similar recommendations are also reported in European Society for Medical Oncology guidelines.⁵ Acceptance in the neuro-oncology community of single-modality adjuvant strategies for elderly patients with GBM generally arises from the challenges that older patients may have in completing aggressive multimodality therapies,^14,19 concerns for increased RT-related central nervous system effects in older patients,²⁰ and the reports of randomized trials showing no OS advantage of RT alone over temozolomide alone.^2,3

In the NOA-08 trial, 373 patients older than 65 years were randomized to dose-dense temozolomide alone vs standard RT alone (60 Gy/30 Fx) and demonstrated no significant differences in survival.³ Similarly, the Nordic trial randomized 342 patients 60 years or older (later amended to ≥65 years) to standard dose temozolomide alone, hypofractionated RT alone (34 Gy/10 Fx), or standard fractionation RT alone (60 Gy/30 Fx), and demonstrated a prolongation in median OS with either temozolomide alone or hypofractionated RT alone over standard fractionation RT.² Subgroup analyses based on MGMT status from each of these trials suggest that, among patients who are candidates for only 1 modality of therapy, temozolomide may be the more important intervention for MGMT promoter-methylated tumors (approximately 35%-45% of patients) whereas RT may be more important for tumors that are nonmethylated (approximately 55%-65% of patients).²¹ However, neither study can inform the selection of elderly patients who would benefit most from the combination of both temozolomide and RT, and oncologists are left with the challenge of reconciling these results with level I evidence supporting CMT in younger patients.^6,17

In light of the aforementioned controversies, the most important findings of the present analysis are related to the significant OS advantage observed with CMT over single-modality approaches and the consistent OS benefits observed on subgroup analyses across sequential age stratifications (65-69, 70-74, 75-79, and ≥80 years). A number of smaller retrospective institutional series, prospective observational cohorts, and population-based analyses have reported similar advantages with CMT over single-modality therapy with RT alone or CT alone (eTable 7 in the Supplement). The NCIC-EORTC phase 3 trial (NCT00482677), which has now completed accrual, will compare the outcomes for patients at least 65 years of age treated with RT+temozolomide vs RT alone in a randomized setting. However, because all patients in the NCIC-EORTC trial received RT, this study will not be able to address the benefit of adding RT to CT vs CT alone, which was associated with a substantial survival advantage in the present analysis. While the results of prospective trials are awaited, this large analysis of a prospectively maintained national tumor registry offers meaningful support to CMT for elderly patients who are suitable candidates and wish to pursue multimodality therapy.

Comparisons of single-modality treatment strategies with RT alone vs CT alone demonstrated no significant differences in OS in this analysis. These observations appear generally consistent with the aforementioned randomized NOA-08 trial.³ The 3-arm Nordic trial provided more nuanced results, with the observation of improved OS with temozolomide alone over standard fractionation RT but similar OS between temozolomide alone and hypofractionated RT. Moreover, the Nordic trial also reported improved OS for patients older than 70 years treated with hypofractionated RT over standard fractionation RT² and separate randomized trials have demonstrated equivalence of hypofractionated RT with longer RT schedules,^22,23 supporting the role of hypofractionated RT in the elderly GBM population. However, similar comparisons of RT schedules were considered inappropriate for this registry analysis because national guideline recommendations contemporary to the study interval still reserved hypofractionated RT for patients with poor prognosis and limited KPS.²⁴ This analysis also demonstrated superior outcomes with single-modality strategies of both CT alone and RT alone over no therapy. These outcomes are largely consistent with the randomized ANOCEF trial of RT vs BSC¹ and data supporting temozolomide over BSC from phase 2 and retrospective analyses.^25,26 To our knowledge, this NCDB data set is the largest reported analysis to support an OS advantage of CT alone over BSC in the temozolomide era.

This analysis has several limitations. Although the demographic characteristics of patients in this analysis are similar to those of contemporary trials and population-based data,^2,3,27 it is important to acknowledge that no centralized pathologic review could be performed. The status of MGMT promoter methylation, which represents a favorable prognostic factor for survival and a predictive factor for CT response,^2,3,28 was unavailable and potential imbalances favoring CT delivery among patients with MGMT promoter-methylated tumors could have artificially improved the survival outcomes associated with CT administration. Thus, it is possible that analyses adjusted for MGMT status might have suggested smaller survival differences in comparisons involving the addition of CT (ie, CMT vs RT alone and CT alone vs no therapy) and potentially superior outcomes for RT alone over CT alone, rather than equivalence. The validated prognostic factor of patient performance status was also unavailable.²⁹ Although comorbidity scores and factors such as age, sex, race, and surgical status can provide some surrogates for patient performance and health care access, these factors cannot directly account for disparities in health and performance status. The range of unmeasured performance statuses in this analysis may have also contributed to the inferior OS outcomes observed by treatment strategy when compared with data from clinical trials and series from high-volume centers.^2,3,30-32 No data were available for analysis regarding treatment-related toxic effects, cognitive performance, or quality of life. Furthermore, there is no information on the use of salvage therapies for recurrent disease. Chemotherapy data were limited to CT administered during the temozolomide era without data regarding specific agents or treatment schedules. Available surgical coding was limited to tumor resection vs biopsy only, without prognostic information regarding the extent of tumor resection.³³ Disparities were also observed among baseline characteristics across treatment groups. Efforts to account for these disparities and the critical limitation of patient selection included the use of multivariate models, propensity score–matched analyses with well-balanced patient characteristics, and subgroup analyses evaluating the impact of therapy stratified by age and tumor resection status.

In this analysis, we report the survival outcomes associated with different treatment strategies in a large cohort of elderly patients with newly diagnosed GBM treated in the temozolomide era. When controlling for critical variables including age, comorbidity scores, and tumor resection, CMT was associated with a consistent OS advantage over single-modality therapy with either CT alone or RT alone. The benefits of CMT remained significant in propensity score–matched analyses and across sequential age stratifications (65-69, 70-74, 75-79, and ≥80 years). Analyses of single-modality treatment strategies demonstrated comparable survival with CT alone and RT alone, and both were superior to no therapy. Overall, this analysis supports the use of CMT for suitable elderly patient candidates.

Accepted for Publication: March 1, 2016.

Corresponding Author: Chad G. Rusthoven, MD, Department of Radiation Oncology, University of Colorado School of Medicine, 1665 N Aurora Ct, Ste 1032, Mail Stop F706, Aurora, CO 80045 (chad.rusthoven@ucdenver.edu).

Published Online: May 23, 2016. doi:10.1001/jamaneurol.2016.0839.

Author Contributions: Dr Rusthoven had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Rusthoven, Koshy, Amini, Kavanagh.

Acquisition, analysis, or interpretation of data: Rusthoven, Koshy, Sher, Ney, Gaspar, Jones, Karam, Ormond, Youssef, Kavanagh.

Drafting of the manuscript: Rusthoven, Koshy, Ney, Youssef.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Rusthoven, Koshy, Sher, Jones, Karam.

Administrative, technical, or material support: Rusthoven, Ormond.

Study supervision: Rusthoven, Youssef, Kavanagh.

Conflict of Interest Disclosures: None reported.