Integrative Multi-Omics Analysis Reveals Molecular Subtypes of Ovarian Cancer and Constructs Prognostic Models.

Abstract

Abstract: Ovarian cancer (OV) remains the most lethal gynecological malignancy. The aim of this study was to identify molecular subtypes of OV through integrative multi-omics analysis and construct machine learning-based prognostic models for predicting the efficacy of immunotherapy. In here, the mutation, copy number variation, RNA sequencing expression profiles, and clinical information were obtained from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases. Multi-omics data were stratified using the MOVICS package, identifying different molecular subtypes. Our analysis identified 2 molecular subtypes (CS1 and CS2) with significant survival differences. Transcriptional regulatory network analysis revealed differential activation of transcription factors such as FOXA1 and GATA3 in CS1, whereas AR and ESR2 were enriched in CS2. A robust prognostic signature comprising 5 key genes was developed through the integration of 10 machine learning algorithms, demonstrating high predictive power across data sets. Immune cell infiltration analysis revealed that anti-tumor immune cells were more abundant in low-risk groups, whereas pro-tumor immune cells predominated in high-risk groups. Furthermore, low-risk patients exhibited better immunotherapy responses and higher tumor mutational burden (TMB). In conclusion, our findings underscore the potential of multi-omics integration in unveiling novel OV subtypes and constructing predictive models that inform personalized treatment strategies. Future research should focus on validating these findings in larger cohorts to enhance OV management through targeted therapeutic approaches.