OS04.3 Extent of resection and overall survival in risk adjusted and exact matched analyses of 22,928 glioblastoma (all molecular subtypes) patients

Abstract

BACKGROUND: Maximal tumor resection, chemo-and radiotherapy are the standard of care for glioblastoma (GBM), of which robust assessment by clinical trial methodology has been limited by GBM's severely abridged survival and relatively rare prevalence. METHODS: The United States National Cancer Database was queried for GBM patients newly-diagnosed from 2004-2014. Overall survival (OS) was assessed by Cox proportional hazards, stratified by extent of resection, and risk adjusted for age, sex, race, Charlson's comorbidity index, insurance, facility location, facility annual case volume, chemotherapy, radiotherapy, tumor location, tumor laterality, and tumor size; and estimated by Kaplan-Meier methods. Subgroup analyses included MGMT methylation or Karnofsky Performance Score, in the risk adjustment. Coarsened exact matching was used to simulate allocation in an interventional trial. RESULTS: 82,960 adult GBM (all molecular subtypes) patients remained after exclusion, 22,928 with complete data for extent of resection and risk adjustment variables, 83% (n=19,007) of which reached endpoint for OS. Median OS for gross total resection (GTR) was 15.5 mos. (95%CI: 15.2-15.9), compared to 11.7 mos. for subtotal resection (STR; 95%CI: 11.3-12.0; HR 1.29, p<0.001) and 5.9 mos. for those who did not undergo resection (NR; 95%CI: 5.7-6.2; HR 1.56, p<0.001). Estimated 2yr OS rates were GTR 30.7% (95%CI: 29.7-31.8), STR 22.1% (95%CI: 21.1-23.0), and NR 14.8% (95%CI: 14.1-15.5). Notably, in the risk-adjustment adjuvant chemotherapy (HR 0.63, 95%CI: 0.60-0.66), radiotherapy (HR 0.71, 95%CI: 0.68-0.75), unilateral tumor (HR 0.66 vs. bilateral, 95%CI: 0.58-0.76), and supratentorial tumors (HR 0.77 vs. infratentorial, 95%CI: 0.71-0.86) were associated with significantly (p<0.001) improved OS. Additionally, coarsened exact matched (5,867 GTR: 5,867 STR) cohorts demonstrated similar outcomes, (GTR HR vs. STR: 0.74, 95%CI: 0.71-0.78). Median OS for patients with these favorable characteristics was 24.2 mos. vs. 1.5 mos. for those with unfavorable characteristics. In subgroup analyses, GTR significance remained after including MGMT (n=2,776) status in the risk-adjustment (HR vs. STR: 0.74) and MGMT methylation was associated with improved OS (HR 0.66, 95%CI: 0.60-0.72). Compared to STR, GTR was associated with reduced rates of 30-and 90-day mortality (OR 0.71 and 0.64, respectively). CONCLUSION: In the largest prospectively collected dataset of its kind, GTR for GBM was associated with significantly improved OS compared to STR or NR, in risk adjusted and exact matched analyses. Adjuvant chemotherapy, radiotherapy, unilaterality, supratentorial location, and MGMT methylation in the risk adjusted analysis were also associated with significantly improved OS.