Journal Article

Integrating genetic and gene expression data in network-based stratification analysis of cancers

Wang, Ji-Ping, Liou, Kenny

2025

Overview

Background Cancers are complex diseases that have heterogeneous genetic drivers and varying clinical outcomes. A critical area of cancer research is organizing patient cohorts into subtypes and associating subtypes with clinical and biological outcomes for more effective prognosis and treatment. Large-scale studies have collected a plethora of omics data across multiple tumor types, providing an extensive dataset for stratifying patient cohorts. Network-based stratification (NBS) approaches have been presented to classify cancer tumors using somatic mutation data. A challenge in cancer stratification is integrating omics data to yield clinically meaningful subtypes. In this study, we investigate a novel approach to the NBS framework by integrating somatic mutation data with RNA sequencing data and investigating the effectiveness of integrated NBS on three cancers: ovarian, bladder, and uterine cancer. Results We show that integrated NBS subtypes are more significantly associated with overall survival or histology. Specifically, we observe that integrated NBS subtypes for ovarian and bladder cancer were more significantly associated with patient survival than single-data type NBS subtypes, even when accounting for covariates. In addition, we show that integrated NBS subtypes for bladder and uterine are more significantly associated with tumor histology than single-data type NBS subtypes. Integrated NBS networks also reveal highly influential genes that span across multiple integrated NBS subtypes and subtype-specific genes. Pathway enrichment analysis of integrated NBS subtypes reveal overarching biological differences between subtypes. These genes and pathways are involved in a heterogeneous set of cell functions, including ubiquitin homeostasis, p53 regulation, cytokine and chemokine signaling, and cell proliferation, emphasizing the importance of identifying not only cancer-specific gene drivers but also subtype-specific tumor drivers. Conclusions Our study highlights the significance of integrating multi-omics data within the NBS framework to enhance cancer subtyping, specifically its utility in offering profound implications for personalized prognosis and treatment strategies. These insights contribute to the ongoing advancement of computational subtyping methods to uncover more targeted and effective therapeutic treatments while facilitating the discovery of cancer driver genes.

• Publisher: BioMed Central,BioMed Central Ltd,Springer Nature B.V,BMC
• Subject: Algorithms / Analysis / Anopheles / Bioinformatics / Biomarkers / Biomedical and Life Sciences / Bladder / Breast cancer / Cancer / Cancer subtyping / Cell proliferation / Chemokines / Cluster analysis / Computational Biology - methods / Computational Biology/Bioinformatics / Computer Appl. in Life Sciences / Data integration / Data mining / DNA methylation / Effectiveness / Female / Gene expression / Gene Expression Profiling / Gene Expression Regulation, Neoplastic / Gene mutations / Gene Regulatory Networks / Gene sequencing / Genes / Genetic aspects / Genomes / Health services / Histology / Homeostasis / Humans / Informatics / Integrated approach / Life Sciences / Medical prognosis / Medical research / Medicine, Experimental / Methods / Microarrays / Mutation / Neoplasms - genetics / Network-based stratification / Omics / Oncology, Experimental / Ovarian cancer / Ovarian Neoplasms - genetics / p53 Protein / Patients / Precision medicine / Prognosis / Propagation / Survival / Survival analysis / Tumors / Ubiquitin / Urinary Bladder Neoplasms - genetics / Uterine cancer / Uterine Neoplasms - genetics