A combined genome-wide linkage and association approach to find susceptibility loci for platelet function phenotypes in European American and African American families with coronary artery disease

Abstract

Background: The inability of aspirin (ASA) to adequately suppress platelet aggregation is associated with future risk of coronary artery disease (CAD). Heritability studies of agonist-induced platelet function phenotypes suggest that genetic variation may be responsible for ASA responsiveness. In this study, we leverage independent information from genome-wide linkage and association data to determine loci controlling platelet phenotypes before and after treatment with ASA.

Methods: Clinical data on 37 agonist-induced platelet function phenotypes were evaluated before and after a 2-week trial of ASA (81 mg/day) in 1231 European American and 846 African American healthy subjects with a family history of premature CAD. Principal component analysis was performed to minimize the number of independent factors underlying the covariance of these various phenotypes. Multi-point sib-pair based linkage analysis was performed using a microsatellite marker set, and single-SNP association tests were performed using markers from the Illumina 1 M genotyping chip from deCODE Genetics, Inc. All analyses were performed separately within each ethnic group.

Results: Several genomic regions appear to be linked to ASA response factors: a 10 cM region in African Americans on chromosome 5q11.2 had several STRs with suggestive (p-value < 7 x 10-4) and significant (p-value < 2 x 10-5) linkage to post aspirin platelet response to ADP, and ten additional factors had suggestive evidence for linkage (p-value < 7 x 10-4) to thirteen genomic regions. All but one of these factors were aspirin response variables. While the strength of genome-wide SNP association signals for factors showing evidence for linkage is limited, especially at the strict thresholds of genome-wide criteria (N = 9 SNPs for 11 factors), more signals were considered significant when the association signal was weighted by evidence for linkage (N = 30 SNPs).

Conclusions: Our study supports the hypothesis that platelet phenotypes in response to ASA likely have genetic control and the combined approach of linkage and association offers an alternative approach to prioritizing regions of interest for subsequent follow-up.

Figure 1

Summary of the factors derived from the Principal Components Analysis (PCA) in European Americans and African Americans on 37 phenotypes measured pre-aspirin (A), post aspirin (B) and post-adjusted-for-pre aspirin (C). Each column represents a single factor and the color block indicates a phenotype with PCA loading > 0.4 for the specific factor. (* = where applicable).

Figure 2

Tests for association under peak areas of linkage in African American [A] and European American [B] pedigrees selected for increased CAD risk. Significance plotted as -log10(p-value) against chromosomal distance for each factor with suggestive or significant linkage to the specific chromosome region from Tables 3 and 4. Region-specific Bonferroni thresholds are indicated by horizontal blue lines specific to each region (defined as 0.05/n; where n = number of SNPs tested in a region of +/- 5 Mb on either side of the linked STR).

Figure 3

Genome wide tests for association for Factor 4B(POST):ADP in African American pedigrees illustrating [A] P-values from linear mixed models along with the linkage traces in red, and [B] linkage weighted P-values.