Mediators of the causal associations between protein ratios and ischemic stroke: a two-step Mendelian randomization study

Background
Proteomics has revealed links between plasma proteins and ischemic stroke (IS), but the relationship between protein ratios, IS, and the effects of blood cells and serum uric acid (SUA) is underexplored.

Methods
Using two-sample Mendelian randomization (MR), we assessed causal relationships between 2,821 protein ratios, 91 blood phenotypes, SUA, and IS subtypes. FDR correction was applied specifically to protein ratio analyses to account for multiple comparisons in the primary MR step. Significant associations were further validated through co-localization analysis, which assessed shared genetic architecture between exposure and outcome loci. This analysis used GWAS data from MEGASTROKE, GISCOME, minimizing confounding bias and reverse causation. Additionally, the total effects of protein ratio levels on IS were decomposed into direct and indirect effects mediated through multiple pathways. Sensitivity analyses ensured robustness.

Results
The CD34/ITGAV ratio exhibited distinct effects on stroke risk, showing 34.9% increased odds of LAS (OR=1.349, 95% CI=1.097-1.658) while demonstrating protective effects against IS outcome progression (OR=0.564, 95% CI=0.380-0.838). Bayesian co-localization analysis revealed complete genetic overlap (PPH4 = 1) for key protein ratio-stroke subtype pairs: AIS with TGFBR2/THBD ratio, LAS with LGALS8/VWF ratio, CES with BST2/CEACAM1 and CD209/CLEC4G ratios. In mediation pathways, neutrophil parameters accounted for 54.4% of the prognosis effect in the ABHD14B/STAMBP-IS association, whereas SUA mediated only 1.3% of the PODXL2/SDC1 ratio-IS relationship.

Conclusions
Our MR study combined with co-localization analysis identifies causal links between protein interactions and IS, highlighting potential targets to disrupt pathways connecting protein ratio changes to IS incidence and outcomes, offering promising intervention avenues.