Genome-Wide Association Study of Down Syndrome-Associated Atrioventricular Septal Defects

Abstract

The goal of this study was to identify the contribution of common genetic variants to Down syndrome-associated atrioventricular septal defect, a severe heart abnormality. Compared with the euploid population, infants with Down syndrome, or trisomy 21, have a 2000-fold increased risk of presenting with atrioventricular septal defects. The cause of this increased risk remains elusive. Here we present data from the largest heart study conducted to date on a trisomic background by using a carefully characterized collection of individuals from extreme ends of the phenotypic spectrum. We performed a genome-wide association study using logistic regression analysis on 452 individuals with Down syndrome, consisting of 210 cases with complete atrioventricular septal defects and 242 controls with structurally normal hearts. No individual variant achieved genome-wide significance. We identified four disomic regions (1p36.3, 5p15.31, 8q22.3, and 17q22) and two trisomic regions on chromosome 21 (around PDXK and KCNJ6 genes) that merit further investigation in large replication studies. Our data show that a few common genetic variants of large effect size (odds ratio >2.0) do not account for the elevated risk of Down syndrome-associated atrioventricular septal defects. Instead, multiple variants of low-to-moderate effect sizes may contribute to this elevated risk, highlighting the complex genetic architecture of atrioventricular septal defects even in the highly susceptible Down syndrome population.

Keywords: aneuploidy; birth defect; complex trait; congenital heart defect; trisomy.

Figure 1

Manhattan plot of the genome-wide association analysis based on the case-control dataset for the non-chromosome 21 autosomal single-nucleotide polymorphisms. The horizontal red line denotes the P-value threshold for genome-wide significance.

Figure 2

Genomic region at 1p36.3 with suggestive evidence of association with Down syndrome−associated atrioventricular septal defect. (A) and (B) show the locus zoom plot and University of California, Santa Cruz (UCSC) image at the corresponding genomic region (human genome build 19, hg19). (A) Locus zoom plot of the region of association at single-nucleotide polymorphism (SNP) with strongest signal (rs1698973). Linkage disequilibrium (LD) between this SNP (purple diamond) and nearby markers is color-coded based on the strength of the LD. The left Y-axis shows the -log₁₀ of the association P-value and the right Y-axis indicates the recombination rate across each region. The position on the chromosome (hg19) and the nearby genes are shown below the X-axis. (B) UCSC browser image shows evidence of regulatory activity. Included are a custom track showing the location of most significant SNPs and the annotation tracks from ENCODE showing regulatory activity.

Figure 3

Genomic region at 5p15.3 with suggestive evidence of association with Down syndrome−associated atrioventricular septal defect. (A) Locus zoom plot of the region of association at single-nucleotide polymorphism (SNP) with strongest signal (rs1428986). Linkage disequilibrium (LD) between this SNP (purple diamond) and nearby markers is color-coded based on the strength of LD. The left Y-axis shows the -log₁₀ of the association P-value and the right Y-axis indicates the recombination rate across each region. The position on the chromosome (hg19) and the nearby genes are shown below the X-axis. (B) University of California, Santa Cruz browser image shows evidence of regulatory activity. Included are a custom track showing the location of most significant SNPs and the annotation tracks from ENCODE showing regulatory activity.

Figure 4

Genomic region on chromosome 21q22.1 at the KCNJ6 gene showing suggestive evidence of association. (A) Locus zoom plot of the region of association at single-nucleotide polymorphism (SNP) with strongest signal on ch21 (rs860795). Linkage disequilibrium (LD) between this SNP (purple diamond) and nearby markers is color-coded based on the strength of LD. The left Y-axis shows the -log₁₀ of the association P-value and the right Y-axis indicates the recombination rate across each region. The position on the chromosome (hg19) and the nearby genes are shown below the X-axis. (B) University of California, Santa Cruz browser image shows evidence of regulatory activity. Included are a custom track showing the location of most significant SNPs and the annotation tracks from ENCODE showing regulatory activity.

Figure 5

Genomic region on chromosome 21q22.3 at the PDXK gene showing suggestive evidence of association. (A) Locus zoom plot of the region of association at single-nucleotide polymorphism (SNP) with strongest signal on ch21 (rs2838355). Linkage disequilibrium (LD) between this SNP (purple diamond) and nearby markers is color-coded based on the strength of LD. The left Y-axis shows the -log₁₀ of the association P-value and the right Y-axis indicates the recombination rate across each region. The position on the chromosome (hg19) and the nearby genes are shown below the X-axis. (B) University of California, Santa Cruz browser image shows evidence of regulatory activity. Included are a custom track showing the location of most significant SNPs and the annotation tracks from ENCODE showing regulatory activity.