Abstract
BACKGROUND: Visceral obesity-related genes (VORGs) have been implicated in cancer progression, but their biological and clinical relevance in ovarian cancer (OC) remains uncertain. METHODS: Prognostic modeling of VORGs was carried out using the The Cancer Genome Atlas (TCGA) and GSE53963 datasets. This model was further investigated in its association with tumor immunity, immune checkpoint inhibitor (ICI) response, and chemotherapy outcomes. Additionally, single-cell data were analyzed to explore gene expression within the tumor microenvironment of this model, specifically identifying high expression of the key gene MARVELD1 in tumor stem cells. The functionality of MARVELD1 was validated experimentally through Cell Counting Kit-8 (CCK-8), transwell, and colony formation assays. RESULTS: OC patients were stratified into four molecular subtypes based on overall survival (OS)-associated VORG expression profiles. Differentially expressed genes (DEGs) between subtypes with prognostic relevance were identified and subsequently analyzed using least absolute shrinkage and selection operator (LASSO) regression and stepwise Cox modeling, resulting in a robust four-gene prognostic model comprising CXCL9, IGF2, FCGBP, and MARVELD1. This model effectively classified patients into high- and low-risk groups, with the high-risk group demonstrating significantly poorer outcomes (P < 0.001). In addition, risk scores showed strong correlations with immune checkpoint gene expression and predicted chemotherapy sensitivity, underscoring their utility in forecasting therapeutic responses. Single-cell transcriptomic analysis revealed heterogeneous expression of model genes across subpopulations, with MARVELD1 prominently enriched in tumor stem cells. MARVELD1 was found to modulate early cellular processes through pathways such as reactive oxygen species. Finally, in vitro experiments confirmed that MARVELD1 knockdown significantly inhibited OC cell proliferation, invasion, and colony formation, further validating its oncogenic role. CONCLUSION: A VORG-based prognostic model was established that correlates with OC prognosis, tumor immune microenvironment, and treatment sensitivity. These findings suggest that MARVELD1 and related pathways may serve as potential targets to improve prognostic stratification and guide therapy in OC.