Integrative transcriptome-wide analysis of shared splicing and expression loci affecting coronary heart disease and lipid traits: a mendelian randomization mediation approach

Abstract Background Coronary heart disease (CHD) is a leading cause of morbidity and mortality, with lipid metabolism playing a key role in its pathogenesis. While lipid-associated genetic variants have been linked to CHD risk, the molecular mechanisms remain unclear. Splicing quantitative trait loci (sQTLs) are increasingly recognized for their potential impact on gene regulation and lipid homeostasis. However, the extent to which splicing variation influences CHD risk through lipid-mediated pathways remains underexplored. Purpose This study investigates whether shared splicing and expression loci contribute to CHD risk through lipid traits. Using a transcriptome-wide Mendelian randomization mediation framework, we quantify the mediation effects of lipid traits and explore their implications for CHD risk assessment and prevention. Methods We integrated sQTLs from GTEx and eQTLs from eQTLGen to identify causal loci shared between lipid traits (LDL, HDL, TG, APOB, APOA1) and CHD/MI. Genetic associations were derived from FinnGen (discovery) and replicated in CARDIoGRAMplusC4D and UK Biobank. Mendelian randomization (MR) was used to assess causal relationships between splicing/expression loci and CHD and their mediation by lipid traits. Summary data-based MR (SMR) evaluated exposure-outcome and exposure-mediator associations, while two-sample MR assessed mediator-outcome relationships. Mediation effects were estimated using the product and Delta methods. Sensitivity analyses included colocalization, pleiotropy assessment, multi-tissue replication, and single-cell RNA-seq validation. Results In FinnGen, we identified HLA-DRB5, FES, EHBP1L1, and HP as sQTLs associated with CHD and LDL. Mediation analysis showed that LDL partially mediated their effects, with mediation proportions ranging from 7.59% (HLA-DRB5) to 16.64% (HP). Expression QTL (eQTL) analysis confirmed EHBP1L1 and FES as consistent CHD-associated genes across multiple datasets. Sensitivity analyses and single-cell transcriptomic data further validated their relevance in coronary artery tissue. Conclusion This study identifies HLA-DRB5, FES, EHBP1L1, and HP as key splicing loci associated with CHD risk, with LDL acting as a partial mediator. FES and EHBP1L1 were the most promising candidates, showing consistent validation across datasets. These findings provide insights into lipid-related genetic regulation in CHD and suggest that splicing variation may serve as a novel target for risk assessment and prevention, though further verification is needed.