LAUSR

PD-1 inhibitors have shown limited efficacy in glioblastoma (GBM) due to microenvironment immunosuppression and low tumor mutational burden. In GBM, PD-L1 expression is not a predictive marker for response to PD-1 or PD-L1 inhibitors. Multiplex immunostaining panel technology allows for detailed analyses of tumor microenvironment cells and their interaction. Pre-treatment tumor tissue was collected retrospectively from 27 patients in our neurooncology database at Columbia University Irving Medical Center with primary glioblastoma who were diagnosed within the past three years, had surgery here, and were either treated with SOC therapy (n= 8) or PD-1 inhibitors at recurrence (n= 19). Multiplex immunofluorescence was done for 1) CD11b/IDO1/HLADR/GFAP, 2) PD1/PD-L1/GFAP, and 3) CD4/CD8/CD25/FoxP3/Ki67/GFAP. Multiplex immunofluorescence panels did not show any correlation with outcomes in patients treated with SOC therapy (non-immunotherapy). Among the 19 patients treated with PD-1 inhibitors, those with more HLA-DR positive cells had worse outcomes (p= 0.02). PD-L1 expression on tumor cells was not predictive of outcomes. There was a correlation trend between PD-1/PD-L1 interaction score (p= 0.08), which measures density of PD-1-positive cells in proximity to PD-L1-positive cells, and outcomes. PTEN loss was correlated with higher Ki67 expression in both tumor cells (p= 0.05) and non-tumors cells (p= 0.03); however, this relationship was not found when looking at Ki67 in CD4+ cells, CD8+ cells, or CD4+CD8+ cells combined. This assay allowed us to evaluate tumor- associated macrophages, myeloid-derived suppressor cells, CD8+ lymphocytes, and CD4+ T regulatory cells; however, none of these were significant predictive markers for outcome. Quantitative spatial profiling by multiplex immunofluorescence is feasible in FFPE glioblastoma tissue. More refined and extensive quantitative and spatial microenvironment analyses may allow for the development of biomarkers for immunotherapy in GBM.