Integrating Proteomics and Mendelian Randomization to Evaluate HGF as a Potential Mediator of Stroke Risk Factors

PURPOSE: Stroke prediction remains suboptimal. We aimed to identify plasma proteins associated with stroke risk and evaluate their potential as clinical biomarkers and causal mediators. EXPERIMENTAL DESIGN: We used a multitiered proteomic approach in the UK Biobank (UKB). In a matched nested case-control study (672 cases, 2,688 controls), 2,920 proteins were analyzed. Findings were internally validated in the UKB-PPP cohort and cross-validated using genetic instruments from seven large-scale protein GWAS meta-analyses. Mendelian randomization (MR) and mediation analyses were employed to assess causality. RESULTS: We identified 122 proteins significantly associated with stroke risk. A 122-protein panel improved the C-index from 0.610 to 0.693. MR analysis provided genetic evidence supporting an association between higher genetically predicted Hepatocyte Growth Factor (HGF) levels and stroke risk (OR 1.18, 95% CI 1.09-1.28). Mediation analysis indicated that HGF mediated 39.15% of the effect of weight and 37.28% of waist circumference on stroke risk. CONCLUSIONS AND CLINICAL RELEVANCE: Our findings support HGF as a potential mediator linking metabolic risk factors to stroke. The identified protein panel offers a robust tool for stroke risk stratification and highlights HGF as a biologically plausible candidate for therapeutic intervention.