IMMU-02. VALIDATION OF PD-L1 EXPRESSION IN HIGH GRADE GLIOMAS AS AN INDEPENDENT NEGATIVE PROGNOSTIC MARKER

Checkpoint inhibition is believed to be a promising method of immune modulation for cancer patients. PD-L1 expression has correlated with improved response to anti-PD-1 therapy in other tumors. However, evaluation of PD-L1 expression in high grade glioma (HGG) patients remains fragmented and controversial. We aimed to compare existing methods of PD-L1 assessment in HGG, validate the status PD-L1 expression as an independent prognostic marker and define the expression profile of PD-L1 in the HGG microenvironment. We tested widely used PD-L1 antibodies to determine the expression of PD-L1 both in tumor cells and within the HGG microenvironment on large core (4mm) tissue microarrays. Antibodies were validated using Western blots of HEK293 cells transfected with a PD-L1 expressing plasmid. RNAScope in situ hybridization was used to validate PD-L1 expression results, and multiplex immunohistochemistry (MP-IHC) was used to immunophenotype selected cases. SP263 and SP142 emerged as our most reliable PD-L1 antibodies. Among HGG samples, 23.5% tumors expressed PD-L1 on greater than 5% of their cells. 36% of IDH wild type glioblastomas expressed PD-L1 at this level, consistent with prior reports. HGG patients expressing PD-L1 on >5% of their cells had a median survival of 19.5 months versus 26 months for patients with <5% (p=0.0072, log-rank test) in our patient cohort, and this was an independent prognostic variable. These expression data are being validated using RNAScope in-situ hybridization. Finally, MP-IHC demonstrated cell-surface PD-L1 expression on T cells, microglia, macrophages and glial cells in the immune microenvironment of HGG patients, demonstrating the heterogeneity of the cellular expression of this marker. High PD-L1 expression by IHC is an independent negative prognostic marker in HGG, thus justifying PD-L1 as a stratification factor in clinical trials, particularly in immunotherapeutic protocols. Our results also underscore the heterogeneity of the tumor microenvironment.