Stanford University Sunnyvale, California, United States
Disclosure(s): No financial relationships with ineligible companies to disclose
Background/Purpose: Patients with rheumatic conditions are at increased risk for cardiovascular (CV) problems, striking on average a decade before peers and conferring substantial morbidity and mortality. Understanding the mechanisms underlying this elevated CV risk could lead to improved care for rheumatology patients and novel treatments for CV conditions. While genetic studies have illuminated several mutual risk loci between certain pairs of conditions, these studies have been limited in size, scope, and replication. Our goal was to conduct a comprehensive joint genetic analysis across 8 rheumatic conditions and 8 CV conditions based on two nation-wide data repositories, characterizing the shared genetic risk, loci, and putative causal SNPs. Methods: We used the Million Veteran Program (MVP; n = 453,878) and UK Biobank (UKB; n = 406,492) EUR cohorts and genome-wide association study (GWAS) statistics based on PheCode definitions of 8 rheumatic (rheumatoid arthritis [RA], lupus [SLE], ankylosing spondylitis [AS], psoriatic arthropathy [PA], regional enteritis, sicca syndrome, ulcerative colitis [UC], and chronic UC) and 8 CV traits (coronary artery disease [CAD], arterial embolism, aortic aneurysm [AA], venous embolism, myocardial infarction [MI], congestive heart failure [CHF], peripheral vascular disease [PVD], and cerebrovascular disease [CVD]) (MVP: Verma et al. 2024, UKB: Zhou et al. 2018) (Figure 1). Genetic heritability (LDSC), global genetic correlation (LDSC), local genetic correlation (HESS), systematic colocalization (coloc, with adaptive locus identification), fine-mapping (SuSiE), and functional enrichment (STRING) were performed for each rheumatic-CV trait pair in a bidirectional replication framework. Results: Heritabilities were generally higher in MVP than UKB, particularly for CV traits. A total of 31 significant genetic correlations (false discovery rate [FDR] < 0.05) were identified, all positive. Four rheumatic-CV trait pair genetic correlations replicated in the reverse cohort setup (all involving RA; range: 0.08 – 0.30). A total of 76 regions showed nominal local genetic correlation (P < 0.05), with 23 of those regions at the 6p21.32/6p21.33 locus. All genome-wide-significant SNPs (P < 5e-8) were tested for colocalization (posterior probability for H3 or H4 > 0.8) and fine-mapped (4,773 loci), yielding 25 loci estimated to share the same causal variant (H3) and 26 loci estimated to have different causal variants (H4). These 51 loci were functionally enriched for genes related to antigen processing and peptide presentation (Gene Ontology BP; FDR < 0.05). Implicated loci for RA included ATXN2 (with CAD, MI, CVD, venous embolism), SH2B3 (with MI, PVD), RSBN1 (with CHF), and the 11p34.1 locus near ST3GAL4 and KIRREL3 (with CAD). Conclusion: Our findings support the presence of shared genetic risk between numerous rheumatic and CV conditions, both on the level of mutually implicated genes and of potentially causal variants. Genes related to antigen processing and peptide presentation were enriched among the colocalized loci of rheumatic and CV traits. The strongest shared genetic signal was seen for RA, implicating ATXN2, SH2B3, RSBN1, and the 11p34.1 locus.
