CTIM-25. NEOADJUVANT ANTI-PD1 IMMUNOTHERAPY FOR SURGICALLY ACCESSIBLE RECURRENT GLIOBLASTOMA: CLINICAL AND MOLECULAR OUTCOMES OF A STAGE 2 SINGLE-ARM EXPANSION COHORT

Glioblastoma is immunologically “cold”, with little clinical response to single-agent immune-checkpoint inhibitors (ICI). Our previous randomized study of neoadjuvant and adjuvant pembrolizumab versus adjuvant pembrolizumab in recurrent glioblastoma identified a molecular signature for ICI and suggested that neoadjuvant pembrolizumab improved survival (PMID30742122). We extended these original findings with a single-arm expansion cohort of glioblastoma patients at first or second relapse undergoing surgical resection for recurrent disease. Eligibility was identical to the original study. Patients receiving > 4mg dexamethasone, or with prior immunotherapy or anti-angiogenic therapy were excluded. The primary objective was to determine the effect of neoadjuvant pembrolizumab on cell cycle/cancer proliferation genes. Secondary objectives were PFS-6 and OS. Twenty-five patients received neoadjuvant pembrolizumab followed by tumor resection and adjuvant pembrolizumab until progression or unacceptable toxicity. The median age of participants was 60 years and median KPS was 90. Ninety-six percent of tumors were IDH wild-type and 48% were MGMT promoter methylated. Neoadjuvant pembrolizumab continued to induce alterations in the tumor microenvironment (T-cell/IFN upregulation, cell cycle downregulation; molecular responders (MR)), but the proportion of MR in this trial was 36% compared with 70% in the original treatment cohort. Median OS was 6.8 months. There were no CTCAE grade 4-5 AEs attributable to pembrolizumab. Further analysis of prognostic differences between stage 1 and 2 subjects is underway, as are additional immune and genotypic correlates to further define molecular changes induced by pembrolizumab. Anti-PD1 monotherapy triggers an intra-tumoral genetic signature associated with interferon-γ response and suppression of cell cycle/cancer proliferation in recurrent glioblastoma, though does not extend OS. This surgical window-of-opportunity paradigm allows for attribution of molecular effects to single agents. Future work will evaluate the differences in patient population, tumor characteristics, and address additional immune checkpoints and/or signaling pathways that can be targeted to augment the effectiveness of anti-PD1 in glioblastoma.