Genome-Wide Variants Associated With Longitudinal Survival Outcomes Among Individuals With Coronary Artery Disease

Abstract

Objective: Coronary artery disease (CAD) is an age-associated condition that greatly increases the risk of mortality. The purpose of this study was to identify gene variants associated with all-cause mortality among individuals with clinically phenotyped CAD using a genome-wide screening approach.

Approach and results: We performed discovery (n = 1,099), replication (n = 404), and meta-analyses (N = 1,503) for association of genomic variants with survival outcome using secondary data from White participants with CAD from two GWAS sub-studies of the Duke Catheterization Genetics Biorepository. We modeled time from catheterization to death or last follow-up (median 7.1 years, max 12 years) using Cox multivariable regression analysis. Target statistical screening thresholds were p × 10^-8 for the discovery phase and Bonferroni-calculated p-values for the replication (p < 5.3 × 10^-4) and meta-analysis (p < 1.6 × 10^-3) phases. Genome-wide analysis of 785,945 autosomal SNPs revealed two SNPs (rs13007553 and rs587936) that had the same direction of effect across all three phases of the analysis, with suggestive p-value association in discovery and replication and significant meta-analysis association in models adjusted for clinical covariates. The rs13007553 SNP variant, LINC01250, which resides between MYTIL and EIPR1, conferred increased risk for all-cause mortality even after controlling for clinical covariates [HR 1.47, 95% CI 1.17-1.86, p(adj) = 1.07 × 10^-3 (discovery), p(adj) = 0.03 (replication), p(adj) = 9.53 × 10^-5 (meta-analysis)]. MYT1L is involved in neuronal differentiation. TSSC1 is involved in endosomal recycling and is implicated in breast cancer. The rs587936 variant annotated to DAB2IP was associated with increased survival time [HR 0.65, 95% CI 0.51-0.83, p(adj) = 4.79 × 10^-4 (discovery), p(adj) = 0.02 (replication), p(adj) = 2.25 × 10^-5 (meta-analysis)]. DAB2IP is a ras/GAP tumor suppressor gene which is highly expressed in vascular tissue. DAB2IP has multiple lines of evidence for protection against atherosclerosis.

Conclusion: Replicated findings identified two candidate genes for further study regarding association with survival in high-risk CAD patients: novel loci LINC01250 (rs13007553) and biologically relevant candidate DAB2IP (rs587936). These candidates did not overlap with validated longevity candidate genes. Future research could further define the role of common variants in survival outcomes for people with CAD and, ultimately, improve longitudinal outcomes for these patients.

Keywords: age-related disease; candidate gene analyses; coronary artery disease; genome-wide association study; survival analysis.

FIGURE 1

Study design and sample schema. This figure depicts the primary study (dark gray box) and extant sub-study data (light gray boxes) from which our retrospective datasets for the present analyses were derived (white boxes). The CATHGEN Biorepository containing data from 10,000 individuals recruited after cardiac catheterization (dark gray box) supplied the samples for two separate GWASs of CAD sub-studies (light gray boxes), providing the extant GWAS genotype data for our secondary analysis. The first GWAS sub-study contained data from a sequential sample of 2,203 CATHGEN CAD cases and controls (light gray box, left); the other CAD GWAS sub-study had data from 1,490 CATHGEN CAD cases and matched controls (light gray box, right). The white boxes display the secondary datasets we analyzed in the present retrospective study. Specifically, we derived our discovery cohort of 684 White CAD cases (white box, left) from the sequential case-control GWAS data; this discovery dataset was analyzed for Step 1. Our replication cohort of 404 White CAD cases (white box, right) was derived from the matched case-control GWAS dataset; this replication dataset was analyzed for Step 2. We then performed meta-analyses of our discovery and replication cohorts (black arrows converging on the bottom white box) for Step 3.

FIGURE 2

Manhattan plot for discovery phase. Manhattan plot of discovery phase SNPs in genomic order by chromosome and position on the chromosome (X-axis), plotted against the negative log (adj)p-value of each SNP’s association with survival events in CAD cases, adjusted for age, sex, and ancestry. Lower horizontal (blue line) represents p = 10^–5; upper horizontal (red) line represents p = 5×10^–8.

FIGURE 3

Kaplan-Meier curves for top two meta-analyses genome-wide SNP exemplars. X-axis, days from enrollment to death or last follow up; Y-axis, probability of survival; inset box, genotype frequencies. (A) Survival curve by rs13007553 genotype indicates 1.47 times increased hazard of all-cause mortality for each addition of the minor allele (T nucleotide). (B) Survival curve by rs587936 genotype indicates reduced hazard of all-cause mortality with each addition of the minor allele (C nucleotide) (HR 0.65).