Integrated digital pathology and transcriptome analysis identifies molecular mediators of T-cell exclusion in ovarian cancer

Abstract

Close proximity between cytotoxic T lymphocytes and tumour cells is required for effective immunotherapy. However, what controls the spatial distribution of T cells in the tumour microenvironment is not well understood. Here we couple digital pathology and transcriptome analysis on a large ovarian tumour cohort and develop a machine learning approach to molecularly classify and characterize tumour-immune phenotypes. Our study identifies two important hallmarks characterizing T cell excluded tumours: 1) loss of antigen presentation on tumour cells and 2) upregulation of TGFβ and activated stroma. Furthermore, we identify TGFβ as an important mediator of T cell exclusion. TGFβ reduces MHC-I expression in ovarian cancer cells in vitro. TGFβ also activates fibroblasts and induces extracellular matrix production as a potential physical barrier to hinder T cell infiltration. Our findings indicate that targeting TGFβ might be a promising strategy to overcome T cell exclusion and improve clinical benefits of cancer immunotherapy.