Genome-Wide Association Study of Non-syndromic Orofacial Clefts in a Multiethnic Sample of Families and Controls Identifies Novel Regions

Abstract

Orofacial clefts (OFCs) are among the most prevalent craniofacial birth defects worldwide and create a significant public health burden. The majority of OFCs are non-syndromic and vary in prevalence by ethnicity. Africans have the lowest prevalence of OFCs (~ 1/2,500), Asians have the highest prevalence (~1/500), Europeans and Latin Americans lie somewhere in the middle (~1/800 and 1/900, respectively). Thus, ethnicity appears to be a major determinant of the risk of developing OFC. The Pittsburgh Orofacial Clefts Multiethnic study was designed to explore this ethnic variance, comprising a large number of families and individuals (~12,000 individuals) from multiple populations worldwide: US and Europe, Asians, mixed Native American/Caucasians, and Africans. In this current study, we analyzed 2,915 OFC cases, 6,044 unaffected individuals related to the OFC cases, and 2,685 controls with no personal or family history of OFC. Participants were grouped by their ancestry into African, Asian, European, and Central and South American subsets, and genome-wide association run on the combined sample as well as the four ancestry-based groups. We observed 22 associations to cleft lip with or without cleft palate at 18 distinct loci with p-values < 1e-06, including 10 with genome-wide significance (<5e-08), in the combined sample and within ancestry groups. Three loci - 2p12 (rs62164740, p = 6.27e-07), 10q22.2 (rs150952246, p = 3.14e-07), and 10q24.32 (rs118107597, p = 8.21e-07) are novel. Nine were in or near known OFC loci - PAX7, IRF6, FAM49A, DCAF4L2, 8q24.21, NTN1, WNT3-WNT9B, TANC2, and RHPN2. The majority of the associations were observed only in the combined sample, European, and Central and South American groups. We investigated whether the observed differences in association strength were (a) purely due to sample sizes, (b) due to systematic allele frequency difference at the population level, or (c) due to the fact certain OFC-causing variants confer different amounts of risk depending on ancestral origin, by comparing effect sizes to observed allele frequencies of the effect allele in our ancestry-based groups. While some of the associations differ due to systematic differences in allele frequencies between groups, others show variation in effect size despite similar frequencies across ancestry groups.

Keywords: GWAS; family study; genetic factors in OFC; genetic heterogeneity; multiethnic.

Figure 1

Circular Manhattan plots of ALL, CSA, EUR, AFR and ASIA. Red lines are drawn at the 1e-06 significance level. Loci significant at the 1e-06 significance level are indicated by the gray dotted lines and numbered 1–18. Chromosomes are depicted by the black bands in the outermost circle.

Figure 2

Association peak in 17p13.1 (NTN1) consistent with category 1. SNPs are ordered by base-pair position in lower panels; Q-statistic p-value noted under each SNP in forest plot of effect size; Wright's F_ST value noted under each SNP name in heatmap of effect allele frequencies; lead variant and group with most significant p-value shown in red in heatmap.

Figure 3

Association peak in 1q32.2 (IRF6) consistent with category 2. SNPs are ordered by base-pair position; Q-statistic p-value noted under each SNP in forest plot; lead variant and sample with highest p-value shown in red in heatmap and F_ST values noted under each SNP. The top 10 associated variants fall below our MAF filter in the EUR and ASIA groups.

Figure 4

Association peak in 17q25.3 in the AFR group consistent with category 3. SNPs are ordered by base-pair position; Q-statistic p-value noted under each SNP in forest plot; lead variant and sample with highest p-value shown in red in heatmap, and F_ST values noted under each SNP.