Integrin β8: A novel prognostic biomarker for ovarian cancer identified through multi-omics-based prognostic models and experimental validation
This study aimed to identify specific prognostic biomarkers for ovarian cancer (OC), develop a prognostic model, and validate a key candidate target. We established a co-expression network using single-cell transcriptomic data and high-dimensional weighted gene co-expression network analysis (hdWGCNA). A 13-gene prognostic signature was constructed via Cox/LASSO regression, effectively stratifying patients into high- and low-risk groups with distinct prognoses (external validation 1-year AUC = 0.83). Comprehensive analyses including CIBERSORT, pRRophetic, GSVA/GSEA, and TIDE assessment revealed that high-risk patients exhibited poorer outcomes, reduced CD8+ T cell infiltration, diminished sensitivity to multiple targeted agents, activation of Hedgehog, NF-κB, and PI3K-Akt pathways, and lower predicted immunotherapy response. A nomogram incorporating the risk score demonstrated good prognostic utility. The risk score constituted an independent prognostic factor (multivariate HR = 1.98). From the signature, ITGB8 was selected for experimental validation. ITGB8 demonstrated significant overexpression in OC tissues and cell lines. Its knockdown suppressed OC cell proliferation, invasion, and migration in vitro, and tumorigenicity in vivo. Furthermore, we identified that ITGB8 promotes invasion by activating the PI3K-AKT-NF-κB signaling pathway, as evidenced by reduced phosphorylation upon knockdown and rescued invasion with a PI3K agonist. This work successfully establishes a validated OC prognostic model and identifies ITGB8 as a novel prognostic biomarker and functional oncogene, supporting its future investigation as a candidate therapeutic target.