Identification of ovarian cancer subtype-specific network modules and candidate drivers through an integrative genomics approach

Identification of cancer subtypes and associated molecular drivers is critically important for understanding tumor heterogeneity and seeking effective clinical treatment. In this study, we introduced a simple but efficient multistep procedure to define ovarian cancer types and identify core networks/pathways and driver genes for each subtype by integrating multiple data sources, including mRNA expression, microRNA expression, copy number variation, and protein-protein interaction data. Applying similarity network fusion approach to a patient cohort with 379 ovarian cancer samples, we found two distinct integrated cancer subtypes with different survival profiles. For each ovarian cancer subtype, we explored the candidate oncogenic processes and driver genes by using a network-based approach. Our analysis revealed that alterations in DLST module involved in metabolism pathway and NDRG1 module were common between the two subtypes. However, alterations in the RB signaling pathway drove distinct molecular and clinical phenotypes in different ovarian cancer subtypes. This study provides a computational framework to harness the full potential of large-scale genomic data for discovering ovarian cancer subtype-specific network modules and candidate drivers. The framework may also be used to identify new therapeutic targets in a subset of ovarian cancers, for which limited therapeutic opportunities currently exist.