European and multi-ancestry genome-wide association meta-analysis of atopic dermatitis highlights importance of systemic immune regulation

Abstract

Atopic dermatitis (AD) is a common inflammatory skin condition and prior genome-wide association studies (GWAS) have identified 71 associated loci. In the current study we conducted the largest AD GWAS to date (discovery N = 1,086,394, replication N = 3,604,027), combining previously reported cohorts with additional available data. We identified 81 loci (29 novel) in the European-only analysis (which all replicated in a separate European analysis) and 10 additional loci in the multi-ancestry analysis (3 novel). Eight variants from the multi-ancestry analysis replicated in at least one of the populations tested (European, Latino or African), while two may be specific to individuals of Japanese ancestry. AD loci showed enrichment for DNAse I hypersensitivity and eQTL associations in blood. At each locus we prioritised candidate genes by integrating multi-omic data. The implicated genes are predominantly in immune pathways of relevance to atopic inflammation and some offer drug repurposing opportunities.

Fig. 1. Manhattan plots of atopic dermatitis GWAS.

(a) the European-only fixed effects meta-analysis (n = 864,982 individuals) and (b) the multi-ancestry MR-MEGA meta-analysis (n = 1,086,394 individuals). −log₁₀(P-values) are displayed for all variants in the meta-analysis. Variants that meet the genome-wide significance threshold (5 × 10⁻⁸, red line) are shown in green.

Fig. 2. Cell type tissue enrichment analysis.

a GARFIELD enrichment analysis of open chromatin data. Plot shows enrichment for AD associated variants in DNase I Hypersensitive sites (broad peaks) from ENCODE and Roadmap Epigenomics datasets across cell types. Cell types are sorted and labelled by tissue type. ORs for enrichment are shown for variants at GWAS thresholds of P < 1 × 10⁻⁸ (black) and P < 1 × 10⁻⁵ (blue) after multiple-testing correction for the number of effective annotations. Outer dots represent enrichment thresholds of P < 1 × 10⁻⁵ (one dot) and P < 1 × 10⁻⁶ (two dots). Font size of tissue labels corresponds to the number of cell types from that tissue tested. b MAGMA enrichment analysis of gene expression data. Plot shows P-value for MAGMA enrichment for AD associated variants with gene expression from 54 GTEx ver.8 tissue types. The enrichment –log₁₀(P-value) for each tissue type is plotted on the y-axis. The Bonferroni corrected threshold P = 0.0009 is shown as a dotted line and the 7 tissue types that meet this threshold are highlighted as red bars.

Fig. 3. Prioritised genes at GWAS loci.

Prioritised genes at known (a) and novel (b) loci. For each independent GWAS locus the top prioritised gene (or genes if they were tied) from our bioinformatic analysis is presented along with a bar representing the total evidence score for that gene. A more detailed breakdown of the constituent parts of this evidence score is presented in Supplementary Fig. 5 and the total evidence scores for the top 3 genes at each locus are presented in Supplementary Data 10. NB. There are some cases of two independent GWAS signals implicating the same gene.

Fig. 4. Predicted interaction network of proteins encoded by the top prioritised genes from known and novel European GWAS loci.

Protein-protein interaction analysis carried out in STRING v11.5; nodes coloured red represent the GO term ‘Regulation of immune system process’ (GO:0002682) for which 28/1514 proteins are included (FDR P = 1 × 10⁻⁹). Full results for all identified pathways are available in Supplementary Data 12.