IMMU-39. EFFICACY OF IMMUNE CHECKPOINT INHIBITORS IN GLIOBLASTOMA FROM REAL-WORLD DATA ANALYSIS

Abstract

Abstract Immune checkpoint inhibitors (ICIs) failed to show efficacy in a randomized trial in glioblastoma (GBM). We utilized the patient-led XCELSIOR platform (NCT03793088) to evaluate ICIs in real-world GBM treatment. XCELSIOR centralizes EMR data from all care sites, creating longitudinal, source-verified data. Identity verification permits accurate overall survival (OS) calculations. As of 6/8/2023, 726 GBM patients were enrolled in XCELSIOR. Median diagnosis date was 1/13/2021 and median age was 56. Seventy-five patients (10%) received ICIs: 53 received pembrolizumab (71%) and 26 received nivolumab (35%) including 12 with ipilimumab (16%). Median days from diagnosis to ICI initiation was 263 days with 22 patients treated at 1st line, 39 at 2nd line, and 14 at 3rd line or later. Data was annotated for 71 patients treated with ICIs and 394 patients not treated with ICIs (no-ICI). Median OS (mOS) from diagnosis for all ICI patients vs. no-ICI was 4.95 years (95% CI 2.18-NR) vs. 1.75 years (1.56-2.03, p < 0.0001). Key features in the ICI cohort were 46% methylated MGMT, 47% gross total resection, 61% exposed to non-SOC agents besides ICI. Creating a propensity score matched cohort for these features, ICI vs. no-ICI had a numerically increased but non-significant mOS of 2.58 years (1.78-NR, n=62) vs. 2.05 years (1.76-2.61, n=113) but a 5-year survival rate of 44% for ICI vs. 16% for no-ICI. Optune use was similar, 39% ICI and 29% no-ICI. Common non-SOC agents used in the ICI cohort were everolimus (11%), osimertinib (8%), lenvatinib (7%), abemaciclib (7%), and neoantigen peptide vaccines (7%). Among unmethylated MGMT patients, mOS was 1.78 years for ICI vs. 1.8 years for no-ICI. These data suggest activity of ICIs for a subset of GBM patients, but do not support use in patients with unmethylated MGMT. Next steps will focus on discrete combination regimens in this evolving data.