Integrative network analysis identifies an immune-based prognostic signature as the determinant for the mesenchymal subtype in epithelial ovarian cancer

BACKGROUND: Epithelial ovarian cancer (EOC) has been classified into four molecular subtypes, of which the mesenchymal subtype has the poorest survival. Our goal is to develop an immune-based prognostic signature by incorporating molecular subtypes for EOC patients. METHODS: The gene expression profiles of EOC samples were collected from seven public datasets as well as an internal retrospective validation cohort, containing 1192 EOC patients. Network analysis was applied to integrate the mesenchymal modalities and immune signature to establish an immune-based prognostic signature for EOC (IPSEOC). The signature was trained and validated in eight independent datasets. RESULTS: Seven immune genes were identified as key regulators of the mesenchymal subtype and were used to construct the IPSEOC. The IPSEOC significantly divided patients into high- and low-risk groups in discovery (OS: P < .0001), 6 independent public validation sets (OS: P = .04 to P = .002), and an internal retrospective validation cohort (OS: P = .025). Furthermore, pathway analysis revealed that differences between risk groups were mainly activation of mesenchymal-related signalling. Moreover, a significant correlation existed between the IPSEOC values versus clinical phenotypes including late tumor stages, drug resistance. CONCLUSION: We propose an immune-based signature, which is a promising prognostic biomarker in ovarian cancer. Prospective studies are needed to further validate its analytical accuracy and test the clinical utility.