Integrative statistical genetics prioritizes candidate ASCVD susceptibility genes across tissues

Atherosclerotic cardiovascular disease (ASCVD) remains a leading cause of mortality worldwide, yet its genetic architecture and underlying mechanistic hypotheses have not been fully elucidated.

We applied genomic structural equation modeling (SEM) to five ASCVD-related phenotypes with a total combined sample size of approximately 3.8 million individuals across the five component GWAS. Leveraging these genetic insights, we performed a cross-tissue transcriptome-wide association study (TWAS) using the UTMOST framework by integrating ASCVD genetic data with gene expression from 49 tissues (GTEx v8). Complementary analyses included single-tissue TWAS (FUSION) and gene-level analysis (MAGMA). By triangulating evidence from these multiple convergent statistical approaches, we identified robust candidate genes, which were further prioritized using Mendelian randomization, Bayesian colocalization, phenome-wide association studies (PheWAS), tissue/cell-type enrichment analyses, and AlphaFold-based structural predictions integrated with chemical interaction profiling.

We identified 14 genes robustly associated with ASCVD susceptibility, including four genes-ARVCF, GFPT1, NFU1, and USP39-some of which have been previously implicated in coronary artery disease genetics (ARVCF) or are newly prioritized in the composite ASCVD phenotype (USP39), exhibiting broad tissue expression patterns. Mendelian randomization and colocalization provided supportive evidence consistent with potential causal relationships, with pronounced tissue specificity (e.g., ARVCF in artery aorta, GFPT1 in heart atrial appendage, NFU1 in adipose subcutaneous). Cell-type enrichment highlighted erythroid progenitor cells and immune populations. PheWAS revealed potential horizontal pleiotropy, while network and pathway analyses implicated these genes in cell adhesion, amino sugar metabolism, iron-sulfur cluster assembly, and RNA splicing. Structural predictions confirmed protein model reliability, and chemical interaction profiling linked these genes to cardiovascular and metabolic disease pathways.

Our integrative statistical genetics approach advances the functional understanding of ASCVD genetics, implicates four genes in disease pathogenesis with tissue-specific mechanisms, and highlights promising candidate genes for future investigation.