COL14A1 drives ischemic cardiac injury in PCOS: an artificial intelligence-identified biomarker

Polycystic ovary syndrome (PCOS) and ischemic cardiomyopathy (ICM) share metabolic and cardiovascular risk factors, including insulin resistance and chronic inflammation, but the molecular links remain unclear.

To identify shared diagnostic biomarkers between PCOS and ICM and clarify their biological basis.

Transcriptomic datasets from multiple PCOS and ICM GEO cohorts were integrated. Differential expression analysis, WGCNA, and 105 machine-learning model combinations were applied to identify robust cross-disease biomarkers. The top-performing Elastic Net (Enet) model was used to screen candidate genes. Immune infiltration, single-cell RNA sequencing (scRNA-seq), and spatial transcriptomics defined cell-type specificity and tissue localization. Drug-target interactions were explored by molecular docking.

Eleven candidate genes were identified, and COL14A1 was the only gene consistently validated across all independent datasets. scRNA-seq identified COL14A1 enrichment in TPM2+ activated fibroblasts, and cell-cell communication analysis highlighted a macrophage-fibroblast crosstalk axis. Spatial transcriptomics confirmed that COL14A1 is tightly linked to fibroblast activation, supporting its role as a mediator of immune-fibrotic responses during postischemic cardiac remodelling. Molecular docking revealed that rofecoxib is a high-affinity potential therapeutic agent targeting COL14A1.

COL14A1 may represent a shared molecular link between PCOS and increased susceptibility to ischemic cardiac injury, potentially acting through macrophage-associated immune activation and fibroblast-mediated ECM remodelling. Rofecoxib may serve as a potential repurposed therapeutic candidate pending further validation. These findings provide mechanistic insights into the ovario-cardiac axis and highlight COL14A1 as a promising target for future risk stratification and intervention studies in women with PCOS.