Proteomic Mendelian randomization and machine learning reveal causal plasma biomarkers in cardiorenal comorbidity

Chronic kidney disease (CKD) significantly increases the risk of coronary artery disease (CAD), but the causal plasma proteins linking these conditions are poorly understood. This study aimed to identify genetically validated plasma biomarkers and therapeutic targets for this cardiorenal link.

We integrated multi-omics data using a Mendelian randomization (MR) framework. The analysis incorporated protein quantitative trait loci (pQTL) data and gene expression datasets for CAD and CKD. A mediation MR approach identified plasma proteins that genetically mediate the causal pathway from CKD to CAD. Machine learning, including Boruta feature selection and support vector machine (SVM) modeling, was used to refine a diagnostic panel from the candidate proteins.

Analysis identified six key causal mediator proteins: LIMA1, PLCG1, and PZP (which promoted CAD risk in CKD), along with HGF, SERPINE2, and TXNDC15 (which exerted protective effects). A diagnostic panel based on these proteins showed strong predictive performance. Functional analysis indicated these proteins converge on lipid-inflammation pathways and modulate immune responses. Molecular docking suggested imatinib as a high-affinity binder to HGF, indicating drug repurposing potential.

This study defines a genetically anchored plasma protein panel that mechanistically links CKD to CAD via a lipid-inflammation axis. The findings support the development of precision blood-based tools for early risk stratification and highlight potential targets for therapeutic intervention in cardiorenal disease.