Abstract 16274: Identification of Novel CAD Genetic Loci by 1000 Genomes-Based Imputation and a Non-Additive Discovery Screen
Introduction: Known common coronary artery disease (CAD) risk variants explain only 10% of the predicted genetic heritability of the disease, suggesting that important genetic signals remain to be discovered.
Hypothesis: The 1000 Genomes imputation training set, and non-additive discovery screens, may allow detection of additional CAD-associated genetic variants that contribute to missing heritability.
Methods: As part of the CARDIoGRAMplusC4D Consortium, we assembled 48 GWAS of CAD that included 1000 Genomes imputed data. These consisted of 60,801 CAD cases and 123,504 controls, 23% being of non-European ancestry. In each study, GWAS analysis was carried out by assuming an additive, dominant or recessive model of inheritance, and followed by meta-analysis to combine the GWAS results for each model.
Results: After QC filtering, 9.4 million variants (91% SNPs, 9% INDELs) were available for meta-analysis. 29% of these were lower frequency variants (0.005 < MAF < 0.05). Novel associations (P < 5 x 10-8, and outside of known CAD genomic regions) under the additive model were detected for 38 variants (8% INDELs) with imputation info score of 0.94 [0.88-0.96] (median [IQR]). Of note, 34% of novel variants (N=13) were of low allele frequency (MAF < 0.05; median [IQR] = 0.03 [0.02-0.03]); these exhibited much larger effect sizes (P < 0.0001; Cohen’s d = 2.3) as compared to the common variants (MAF ≥ 0.05). The newly identified variants were mapped to 10 novel genomic loci for CAD. Together, these variants explained 2.5% of the heritability of CAD and majority (60%) were intronic. Three of these loci fit a dominant mode of inheritance. An additional two novel loci were identified by a recessive mode of inheritance. Among the newly identified loci, three had been previously reported at GWAS levels of significance for metabolic traits.
Conclusions: These findings demonstrate the value of using a global imputation training set to enhance coverage of low allele frequency and incompletely tagged variants. Consideration of non-additive models of inheritance enabled identification of additional genetic variants associated with CAD.
Author Disclosures: M. Nikpay: None. A. Goel: None. H. Won: None.
- © 2014 by American Heart Association, Inc.