Real-world evaluation of the impact of radiotherapy and chemotherapy in elderly patients with glioblastoma based on age and performance status

Abstract

Background

This retrospective study investigated the impact of, in addition to age, the management and outcomes of elderly patients with glioblastoma (GBM).

Methods

The National Cancer Database was queried between 2004 and 2015 for GBM patients age 60 years and older. Three age groups were created: 60 to 69, 70 to 79, and 80 years and older, and 4 age/KPS groups: “age ≥ 60/ KPS < 70” (group 1), “age 60 to 69/KPS ≥ 70” (group 2), “age 70 to 79/KPS ≥ 70” (group 3), and “age ≥ 80/KPS ≥ 70” (group 4). Multivariable (MVA) modeling with Cox regression determined predictors of survival (OS), and estimated average treatment effects analysis was performed.

Results

A total of 48 540 patients with a median age of 70 years (range, 60-90 years) at diagnosis, and a median follow-up of 6.8 months (range, 0-151 months) were included. Median survival was 5.0, 15.2, 9.6, and 6.8 months in groups 1, 2, 3, and 4, respectively (P < .001). On treatment effects analysis, all groups survived longer with combined chemotherapy (ChT) and radiation therapy (RT), except group 1, which survived longer with ChT alone (P < .001). RT alone was associated with the worst OS in all groups (P < .01). Across all groups, predictors of worse OS on MVA were older age, lower KPS, White, higher comorbidity score, worse socioeconomic status, community treatment, tumor multifocality, subtotal resection, and no adjuvant treatment (all P < .01).

Conclusions

In elderly patients with newly diagnosed GBM, those with good KPS fared best with combined ChT and RT across all age groups. Performance status is a key prognostic factor that should be considered for management decisions in these patients.

The incidence of glioblastoma (GBM) is highest in individuals between ages 75 and 84 years (median age at diagnosis, 64 years) and the incidence of GBM in this elderly population is rising, according to the latest report from the Central Brain Tumor Registry of the United States.¹ However, until recently the landmark trial by the European Organisation for Research and Treatment of Cancer (EORTC) Brain Tumor and Radiotherapy Groups, and the National Cancer Institute of Canada (NCIC) Clinical Trials Group establishing maximal safe resection followed by radiotherapy (RT) (60 Gy in 30 fractions) with concurrent and adjuvant temozolomide (TMZ) chemotherapy (ChT) as the standard of care upfront treatment for GBM only included patients age 18 to 70 years, and patients had a median age of 56 years at diagnosis in this trial.^2,3 Resultantly, the optimal treatment of a large proportion of patients diagnosed with GBM is still unclear.²

Population studies have shown that age is an important prognostic factor in patients with GBM, with older patients faring worse than their younger counterparts.^4,5 Older patients were defined as 60 years or younger in the first population study⁴ and 85 years or older in the second one.⁵ Chronological age was also shown to affect practice patterns, with practitioners’ tendency to be less aggressive in managing elderly patients.⁴ Health care professionals face challenges in treating elderly patients, including inadequate geriatric knowledge and training, underrepresentation of this population in clinical trials, and the presence of medical comorbidities that can interfere with the way these patients can withstand medical treatment.^6–9

Clinical management of GBM in elderly patients has recently been investigated in 5 randomized trials,^10–14 but still constitutes a controversial topic. It is unclear whether multimodal treatment is appropriate for elderly patients. In addition, studies have shown that a patient’s ability to tolerate toxic therapy did not necessarily correlate with chronological age in various malignancies.^15,16 Another controversial topic in the management of GBM is the interplay of chronological age and performance status, as most trials, including the EORTC/NCIC trial² and trials focused on elderly populations,^10,12–14 excluded patients with poor performance status, or studies included younger patients with poor performance status along with elderly patients.¹²

The primary aim of this population-based study is to evaluate the impact of treatment modality on overall survival (OS) based on age and KPS in elderly patients with GBM. Secondary aims consist of identifying independent prognostic factors for OS in elderly patients with GBM, as well as evaluating patterns of practice in the management of elderly patients both before and after publication of the most recent GBM trials focused on elderly populations. This study contributes real-world data on the impact of performance status and age in elderly patients with newly diagnosed GBM.

Methods

This study analyzed the National Cancer Database (NCDB), which is a joint project of the Commission on Cancer (CoC) of the American College of Surgeons and the American Cancer Society, and consists of deidentified information regarding tumor characteristics, patient demographics, and patient survival for approximately 70% of the US population. The NCDB contains information not included in the Surveillance, Epidemiology, and End Results database, including details regarding use of systemic therapy and radiation therapy (RT). Data used in this study were derived from a deidentified NCDB file. The American College of Surgeons and the CoC have not verified and are neither responsible for the analytic or statistical methodology used nor the conclusions drawn from these data by the investigators. Because all patient information in the NCDB database is deidentified, this study was exempt from an institutional review board evaluation.

The NCDB was queried for eligible patients defined by age 60 and older with histologically confirmed GBM diagnosed between 2004 and 2015. Three age groups were created: age 60 to 69, age 70 to 79, and age 80 years and older. The reported KPS was gathered to form 4 age/KPS groups: “age ≥ 60/KPS < 70” (group 1), “age 60 to 69/KPS ≥ 70” (group 2), “age 70 to 79/KPS ≥ 70” (group 3), and “age ≥ 80/KPS ≥ 70” (group 4). Patients were also evaluated based on year of diagnosis: 2004 to 2012 (prior to publication of prospective elderly GBM trials), and 2013 to 2015.

Statistical Analysis

Data analysis was performed using Stata/MP statistical software (version 15.1; Stata). The Pearson chi-square test was used to assess measures of association in frequency tables. The equality of group medians was assessed with nonparametric tests for equality. The survival function was carried out using Kaplan-Meier estimates. OS was calculated from the diagnosis date to the date of last contact or death. The log-rank test was used to assess the equality of the survivor function across groups, and Cox proportional hazards model was used for univariate (UVA) and multivariate analysis to assess the effect of patient, tumor, and other predictors of significance on the end point under assessment. All factors found to have a P value of .25 or less on UVA were included in the multivariable analysis (MVA), with each factor eliminated in a step-wise manner until the most significant variables were identified. The Wald test was used to assess the role of covariates in the model. The estimated hazard ratio (HR) is reported. A P value of .05 or less was considered statistically significant. Statistical tests were based on a 2-sided significance level.

Estimated average treatment effects analysis was carried out using survival-time inverse-probability-weighted regression adjustment.^17,18 Patients’ average treatment effects and potential-outcome means were assessed to compute averages of predicted potential survival outcomes means. Patients were assessed for OS by adjuvant treatment status in a model that adjusted for factors including age/KPS group, ethnicity, Charlson-Deyo comorbidity score, education level, median income quartile, treatment facility location (metropolitan, urban, or rural), facility type (community cancer program or academic program), unifocal/multifocal disease, O⁶-methylguanine-DNA methyltransferase (MGMT) promoter methylation status, extent of resection, adjuvant treatment modality, and RT modality.

Results

Patient Population

The methodology for our cohort derivation is depicted in Figure 1. Briefly, of 204 705 patients included in the NCDB glioma database, 48 919 patients older than 60 years were included in the general elderly analysis, and 3231 patients were included in the analysis by KPS/age groups. Tables 1 and 2 illustrate the patient and treatment characteristics for all elderly patients (age ≥ 60 years) and for the cohort of patients with reported KPS, respectively.

Within our study cohort, median age at diagnosis was 70 years (range, 60-90 years), and median follow-up was 6.8 months (range, 0-151 months). In the overall cohort, 56.2% of patients were men, 92.1% of White ethnicity, and 65.7% diagnosed between 2004 and 2012 (the rest were diagnosed between 2013 and 2015). In the cohort of patients with reported KPS, 56.7% of patients were men, 91.9% of White ethnicity, and 43.6% diagnosed between 2004 and 2012. After surgery/biopsy, 7353 (15.2%) were treated with RT alone, 1411 (2.9%) with ChT alone, 25 687 (52.9%) with both RT and ChT (ChT within 90 days of RT), and 14 090 (29.0%) had no/unknown adjuvant treatment.

The proportions of patients receiving ChT and RT were 97.5% in group 2, 74.3% in group 3, but only 47.8% in group 1, and 55.6% in group 4 (P < .001). The proportions of patients receiving ChT and RT were 62% in patients age 60 to 69 years, 49.6% in patients age 70 to 79 years, and 29.6% in patients 80 years or older (P < .001).

Survival and Patterns of Care Analysis

The 2-year OS rates were 27.9%, 15.7%, 12.9%, and 6.6% for groups 2, 3, 4, and 1, respectively (P < .01). The 2-year OS rates were 19.0%, 9.9%, and 4.6% for age groups 60 to 69, 70 to 79, and 80 and older, respectively (P < .01). Median survival was 15.2, 9.6, 6.8, and 5.0 months in groups 2, 3, 4, and 1, respectively (P < .01), and 10.5, 5.8, and 3.5 months in age groups 60 to 69, 70 to 79, and 80 and older, respectively (P < .01) (Figure 2).

The addition of ChT to RT was associated with improved OS in all age groups (P < .01), whereas RT alone was associated with the worst OS in all groups (P < .01). When considering the age/KPS groups, all patients survived longer with combined ChT and RT on both UVA and estimated average treatment effects OS analysis, except patients in group 1, who survived longer when treated with ChT alone (1-year OS of 31.5% with ChT and RT, vs 44.4% with ChT alone, P < .001) (Figures 3 and 4). For example, for patients in group 2, patients receiving ChT + RT survived longer than patients treated with RT alone (HR = 1.26; 95% CI, 1.03-1.53 P = 0.02) or those treated with ChT alone (HR = 2.66; 95% CI, 1.69-4.45, P < .001). Patients in group 3 also survived significantly longer when treated with ChT + RT vs RT alone (HR = 1.49; 95% CI, 1.16-1.90, P = .002).

On MVA, variables associated with worse OS in the entire cohort and in all age and age/KPS groups were older age, lower performance status, White ethnicity, higher Charlson-Deyo comorbidity score, lower education level, lower income, treatment in a community cancer program (vs an academic program), tumor multifocality, subtotal resection, non–intensity-modulated radiotherapy (IMRT) use, and no adjuvant treatment or adjuvant RT or ChT monotherapy (vs combined RT and ChT) (all P < .01).

Tumor MGMT status of 5884 patients was also reported in this database. Among 2468 patients with methylated MGMT status, the median OS and 2-year OS rates were respectively 15.5 months and 31.9% with RT plus ChT, as compared to 10.2 months and 21.8% respectively with RT only (P < .001). Among 3416 patients with nonmethylated MGMT status, the median OS and 2-year OS rates were respectively 12.6 months and 17.0% with RT plus ChT, as compared to 8.6 months and 11.8% with RT alone (P < .001) (Supplementary Figure 1).

Evaluation based on the year of treatment found that for all patients age 60 years or older, those diagnosed in the 2004 to 2012 period had shorter OS than those diagnosed in the 2013 to 2015 period (2-year OS: 12.6% vs 16.9%, P < .001). However, when looking at particular age/KPS subgroups, only patients in the best prognostic subgroup (group 2) had a significantly shorter survival in 2004 to 2012 as compared to the 2013 to 2015 period. The use of hypofractionated RT increased between the period 2004 to 2012 and 2012 to 2015 (15% vs 19%, respectively, P < .001), most markedly in patients age 80 years or older (28% vs 48%, respectively, P < .001), and in elderly patients with KPS less than 70 (26% vs 40%, respectively, P < .001). Although there was an increase in the use of combined modality treatment between 2004 and 2012 and 2013 and 2015 in the entire elderly cohort, there was a statistically nonsignificant increase in the use of ChT alone in patients in group 4, and a decrease in the use of combined modality treatment (Supplementary Figure 2).

Discussion

Despite growing research efforts and clinical trials to evaluate better treatments for elderly GBM patients,^10–14 the optimal clinical management of GBM in elderly patients remains a challenging and controversial topic in neuro-oncology. Practice patterns in the care of elderly patients with newly diagnosed GBM vary widely, and to date there has been no established standard of care.¹⁹ Chronological age continues to be a major factor that dictates management decisions in patients with GBM. In a Canadian population-based study, elderly patients were shown to be less likely to undergo gross total resection, which is a known positive prognostic factor for GBM. Elderly patients were also less likely to receive RT than their younger counterparts, and if they did receive RT, they were less likely to receive a dose greater than 53.5 Gy. In this study, patients age 60 years or older were also shown to have worse survival outcomes than younger patients, raising the question whether this inferior survival outcome may reflect less aggressive management.⁴

Clinical trials in the elderly population have largely compared different single modality treatments,^11,12,14,20 or a single modality treatment to best supportive care,¹⁰ reflecting an implicit bias that elderly patients would not tolerate multimodal treatment. Many of these trials also had a minimum KPS as an eligibility factor and therefore provided limited guidance for the management of patients with lower functional status. For instance, a French trial included patients age 70 years or older with a KPS of 70 or higher and showed that the addition of RT to supportive care significantly prolonged survival (median survival of 16.9 weeks with best supportive care only vs 29.1 weeks with the addition of RT). No significant differences were found in quality of life or neurocognitive function between patients in both arms.^10,12,14 The Neuro-oncology Working Group of the German Cancer Society led the NOA-08 trial that randomly assigned patients older than 65 years with a KPS greater than 60 and a new diagnosis of anaplastic astrocytoma or GBM to TMZ alone or RT alone (60 Gy in 30 fractions). TMZ was found to be noninferior to RT both in the intention-to-treat and per-protocol analyses.¹² In the Nordic trial, 291 patients older than 60 years with Eastern Cooperative Oncology Group performance status 0 to 2 were randomly assigned to standard-fractionation RT (60 Gy in 30 fractions), hypofractionated RT (34 Gy in 10 fractions), or TMZ. In the overall trial population, although survival was significantly longer in the TMZ arm compared with standard RT, it was not significantly different between patients treated with TMZ or with hypofractionated RT. For the subset of patients older than 70 years, OS was significantly better in patients treated with hypofractionation (HR = 0.59 [95% CI, 0.37-0.93], P = .02), or TMZ alone (HR = 0.35 [95% CI, 0.21-0.56], P < .0001) compared with standard fractionation RT.¹⁴

A recent study randomly assigned “vulnerable” patients diagnosed with GBM to 2 different hypofractionated RT regimens, 25 Gy in 5 fractions (short) and 40 Gy in 15 fractions (longer), and defined “vulnerable” as elderly (age ≥ 65 years) and/or KPS of 50% to 70% but any age. Outcomes between the 2 arms were similar in terms of OS, progression-free survival, and quality of life, and the authors concluded that the “short” regimen can offer a reasonable and convenient approach in the treatment of elderly and/or frail patients However, this study included a substantial number of patients younger than 65 years with poor performance status and therefore does not fully reflect a trial that addresses the elderly population specifically.¹¹ It is worth mentioning the low proportion of patients treated with hypofractionation in the present study. The use of hypofractionation is especially relevant in the current COVID-19 pandemic era. By decreasing time spent in an RT center, hypofractionation has the potential to reduce exposure of elderly patients during pandemics.

Another NCDB analysis of elderly patients with GBM diagnosed between 2005 and 2011 also showed that multimodal treatment was associated with improved OS as compared with ChT alone and RT alone,²¹ but did not analyze patients by KPS. Our study aimed primarily at examining the impact of a combination of age and performance status on outcomes and patterns of practice in the treatment of elderly patients with newly diagnosed GBM in the United States. Based on the data of more than 30 000 patients, elderly (age > 60 years) and very elderly (age > 80 years) patients with good performance status (KPS of 70 or higher) had better OS with combined ChT and RT as compared to single-modality adjuvant treatment. Our findings corroborate those of a recent Canadian prospective trial,¹³ in which the addition of TMZ to short-course RT in elderly patients with Eastern Cooperative Oncology Group performance status less than or equal to 2 (corresponding to KPS ≥ 70) was associated with longer survival than RT alone. Because this study accounted for performance status in the eligibility criteria, it analyzed the best prognostic group and did not look at differences in outcomes between elderly patients with a KPS less than 70 and those with a KPS of 70 or greater, the way our study did. Our results suggest that patients with poor performance status (KPS < 70) had better OS with ChT alone and patients treated with RT alone had worse OS when compared to patients treated with ChT alone or combined modality treatment, but this may reflect a selection bias of offering more palliative RT for patients with limited life expectancy.

Despite the demonstrated benefit of using both ChT and RT, our real-world analysis from NCDB data shows a trend toward decreased use of combined modality therapy in 2013 to 2015 compared with 2004 to 2012. This may have resulted from the publication of the elderly GBM trials, advocating “deescalation” of treatment in the elderly population. In terms of aggressiveness of management, surgeons tend to be less aggressive in the elderly population, probably because of perceived frailty, but also because of patients’ comorbidities. However, consistent with the findings in NOA-08 that reported the extent of surgery is an independent prognostic factor for OS among GBM patients age 65 years and older with a KPS of 60 or greater, our analysis confirms that the extent of resection is an independent prognostic factor in elderly patients with newly diagnosed GBM.^5,12

It is not clear whether MGMT promoter methylation status is a useful predictive biomarker in the elderly GBM population. The results described here suggest that in elderly patients with GBM, the addition of TMZ to RT resulted in longer survival than RT alone, regardless of MGMT status: 15.5 months with ChT vs 10.2 months without ChT, if MGMT promoter was methylated, and 12.2 vs 8.6 months if MGMT promoter was not methylated. These results corroborate the findings of a recent Canadian trial,¹³ in which the addition of TMZ to short-course RT was associated with longer survival than RT alone, regardless of MGMT methylation status. In the NOA-08 trial, median event-free survival was significantly longer for patients with MGMT promoter methylation treated with TMZ than in those treated with RT (8.4 months vs 4.6 months, respectively), and the opposite was true in patients with no MGMT promoter methylation (3.3 months vs 4.6 months).¹² Combined modality approaches were not analyzed in the NOA-08 trial; therefore it is difficult to compare it directly with the findings described here.

We acknowledge that this study has inherent limitations related to its retrospective nature and the use of a large population-based database. NCDB can have missing data (most notably type of ChT and MGMT methylation status), a risk of potential miscoding of variables, and possible selection bias because only centers accredited by the CoC can contribute to the NCDB, therefore limiting the generalizability of the study findings. Another limitation is that NCDB provides only all-cause survival and does not capture progression-free survival or cancer-specific survival, which are arguably less important in patients with GBM in whom most deaths occur secondary to GBM. Despite these limitations, it still provides an invaluable tool to study large real-world cohorts, and findings from this study can be used to complement those derived from prospective randomized controlled trials.

Conclusion

The results described in this study support that performance status is an independent prognostic factor that should be considered for clinical management decisions in elderly GBM patients. Integration of KPS in combination with age factors can improve clinical decisions on the optimal use of multimodal treatments that may consist of maximal safe resection, ChT, and/or RT for elderly GBM patients. Ideally, a comprehensive geriatric assessment evaluating physical function, neurocognition, psychological status, nutritional status, comorbidities, and social support should be performed at diagnosis,⁶ and guide shared treatment decision making.

Acknowledgments

This work was presented as an oral presentation at the Society for Neuro-Oncology meeting, New Orleans, Louisiana, November 2018.

Funding

This work was supported by The University of Texas MD Anderson Cancer Center (Support Grant P30CA01667) and The University of Texas MD Anderson Cancer Center, Division of Radiation Oncology (Radiation Oncology Strategic Research Initiative grant).

Conflict of interest statement. None declared.

References