Deciphering the molecular basis of tumor immunogenicity in high grade serous ovarian cancer (200)

Abstract

<b>Objectives:</b> Response rates to immunotherapy in high-grade serous ovarian cancer (HGSOC) have thus far been disappointing, consistent with the understanding that HGSOC is not immunogenic cancer. We hypothesize that a subset of HGSOC may experience an immunemediated survival benefit and that these patients can be reproducibly identified by a gene signature developed through a novel cancer informatics approach. <b>Methods:</b> Gene expression signatures reflecting T cell-mediated antitumor activity show poor prognostic power in HGSOC. We hypothesized that this might be true for some tumors but not others. We developed a novel, Cox-regression-based, gene interaction algorithm to identify genes from tumor gene expression profiles whose expression levels are capable of stratifying patients into subgroups whereby a previously validated marker of T cell cytolytic activity (CYT gene signature) either does or does not achieve significant prognostic power. Six publicly available datasets of HGSOC gene expression profiles (<i>n</i>=1,525 tumors) were used to identify and validate a metagene (composed of many genes) in training and validation cohorts that could stratify HGSOC patients into population tertiles that differed according to the prognostic performance of the CYT signature. Kaplan-Meier analyses and multivariable Cox regression models were used to evaluate the significance of the association between the CYT signature and overall survival in the metagene patient tertiles. Gene ontology analyses using the Database for Annotation, Visualization and Integrated Discovery (DAVID) and Ingenuity Pathway Analysis (IPA) tools were performed to assess the pathway-level differences among metagene tertiles related to tumor immunogenic potential. <b>Results:</b> A metagene consisting of 43 genes was identified by our algorithm, and a percentile ranking procedure was applied to two independent training cohorts. In four independent validation cohorts, the metagene reproducibly identified a group of patients representing approximately one-third of each cohort where high expression of the CYT gene signature was associated with improved overall survival consistent with an immune-mediated survival benefit (Figure 1). This finding appeared independent of other prognostic variables in cox models, including stage, age, and tumor debulking status. In this metagene group, gene ontology analysis suggested a tumor molecular configuration in HGSOC reflective of increased immunogenicity via reduced immunosuppression by TGF-beta and Wnt signaling pathways. <b>Conclusions:</b> Approximately one-third of HGSOC tumors exhibit an immunogenic molecular background that may confer an immunemediated survival advantage to patients. The metagene described here may be a tool to better identify HGSOC patients likely to be responsive to immunotherapies. Our findings suggest that immune suppression by TGF-beta and Wnt pathways may be viable targets for further study and future drug development.