Distinct immune microenvironments in ovarian cancer subtypes indicate potential for immunotherapies

To enable immunotherapy for ovarian cancers, precision targets for immune priming as well as patient stratification approaches are required. The tumor microenvironment in particularly the rare subtypes of low-grade serous, mucinous, clear cell and endometrioid ovarian cancer, remains poorly characterized, and these tumors have been largely ignored in the immuno-oncology setting. We performed spatially resolved molecular profiling of 78 tumor and immune protein markers in defined tissue regions from 254 ovarian cancer patients of mixed histologies, using GeoMx. Network graph analysis was applied to compute spatial statistics from multiplex immunofluorescence images. GeoMx-compatible softwares were developed for data processing and analysis, based on linear mixed effect modelling, survival analysis and machine learning. Immune-regulatory targets associated with specific subtypes included STING in low-grade serous; CTLA-4 and PD-L1 in mucinous; CD40, IDO1 and VISTA in clear cell; and B7-H3 in endometrioid ovarian carcinomas. In high-grade serous ovarian carcinomas, intra-tumoral expression of SMA and PD-L1 emerged as strong prognostic indicators. Proximity of CD8 + T-cells to tumor cells as measured by group degree centrality was a marker of improved prognosis and of infiltration of all T cell subtypes, dendritic cells, and tumor-associated macrophages, along with elevated expression of PD-L1, IDO1, Tim-3, and CD40. In contrast, tumors with low CD8-tumor proximity were enriched in CD20 and CD25. Our findings highlight the potential for differential targeted treatment related to histotyping, tumor-immune spatial scoring and intra-tumoral expression of key prognostic markers.