EPID-21. INCORPORATING RECEIPT OF STANDARD-OF-CARE AS A KEY VARIABLE IN POPULATION-BASED STUDY OF GLIOBLASTOMA SURVIVAL

Abstract

Abstract Object Populational database research with real-world settings offers epidemiologic insight at scales that cannot be easily recapitulated, especially for rare cancers such as glioblastoma. However, most survival studies were conducted without consideration for whether the patients received the standard of care (SOC). METHODS Survival analysis was performed using the National Cancer Database (NCDB) and stratified for Methyl-guanine-methyl transferase (MGMT) promotor methylation. Glioblastoma patients age < 40 were excluded to minimize the prognostic impact of isocitrate dehydrogenase (IDH) mutations. SOC was defined as the receipt of chemotherapy and radiation within 2-6 weeks of surgical biopsy or resection. Sensitivity analyses were performed allowing chemotherapy and radiation start date to differ by 0, 3, 5, and 10 days. RESULTS Within the NCDB (N = 11231), 54.7% of newly diagnosed glioblastoma patients did not receive SOC. The median survival time (MST) estimates were significantly shorter for all glioblastoma patients (10.9, IQR: 4.3-21.5, months) relative to SOC treated patients (15.3, IQR: 8.6-26.2, months; p < 0.001). The MST estimates remain stable in sensitivity analysis allowing for 0, 3, 5, and 10-day discrepancy between chemotherapy and radiation start date. These results were recapitulated using the SEER-Medicare database. The MST estimates for MGMT methylated and unmethylated glioblastoma patients after SOC in the NCDB were 21.0 and 14.5-mo, respectively. The overall and MGMT-specific survival estimates mirror those reported by the SOC arms of five recent randomized controlled studies. Analysis of all glioblastoma patients indicates survival benefit of immunotherapy (HR: 0.90, p = 0.020) while such benefit was absent in SOC treated patients (HR: 0.97, p = 0.644) - with the latter observation more reflective with trial results and clinical experiences. CONCLUSION Survival analysis in the populational database study can be misleading without information on the receipt of SOC.