Background As a critical prognostic factor in ovarian cancer which is the most lethal gynecologic malignancy, vasculogenic mimicry (VM) has not been systematically incorporated into prognostic evaluation frameworks in ovarian cancer (OC). This underscores the necessity to develop and validate a gene subtyping-based prognostic model through comprehensive analysis of VM-related biomarkers. Methods This study integrated multi-omics data from TCGA, GEO and GTEx, forming a primary set and an external validation cohort. Through literature mining, 28 VM-related genes were identified. Univariate Cox and LASSO regression distilled 9 genes as vasculogenic mimicry-related prognostic index (VMRPI), establishing a risk model validated by ROC and constructing a nomogram with clinical prognostic factors. Consensus clustering stratified patients into VM-high/-low subgroups. Multi-angle assessments connected risk scores with tumor mutational burden, immune infiltration, and chemotherapy sensitivity. Clinical validation encompassed IHC-PAS detection of VM structures in 36 HGSOC paraffin specimens and qRT-PCR confirmation of gene expression in matched frozen tissues. Results vasculogenic mimicry-related genes (VMGs) exhibited differential expressions in HGSOC versus normal tissues, with consensus clustering stratifying 474 patients into prognostically distinct VM-high/low subgroups. Prognosis-associated DEGs (n=758) enriched in ECM-receptor and PI3K-AKT pathways. A 9-gene prognostic model demonstrated robust predictive accuracy. Risk scores correlated with immune infiltration and drug sensitivity. Multivariate-validated nomogram integrating clinical factors and risk scores achieved precise survival prediction. IHC-PAS confirmed VM structures, with VM-positive cases showing upregulated VMGs and VMRPIs. Conclusions VMG-based stratification revealed distinct prognostic ovarian cancer subgroups and a 9-VMRPI demonstrated robust prognostic power with validated immune-microenvironment, drug-response associations, IHC-PAS staining, and qRT-PCR confirmation.
Zhao et al. (Thu,) studied this question.