Genome-wide association analysis of 350 000 Caucasians from the UK Biobank identifies novel loci for asthma, hay fever and eczema

Abstract

Even though heritability estimates suggest that the risk of asthma, hay fever and eczema is largely due to genetic factors, previous studies have not explained a large part of the genetics behind these diseases. In this genome-wide association study, we include 346 545 Caucasians from the UK Biobank to identify novel loci for asthma, hay fever and eczema and replicate novel loci in three independent cohorts. We further investigate if associated lead single nucleotide polymorphisms (SNPs) have a significantly larger effect for one disease compared to the other diseases, to highlight possible disease-specific effects. We identified 141 loci, of which 41 are novel, to be associated (P ≤ 3 × 10-8) with asthma, hay fever or eczema, analyzed separately or as disease phenotypes that includes the presence of different combinations of these diseases. The largest number of loci was associated with the combined phenotype (asthma/hay fever/eczema). However, as many as 20 loci had a significantly larger effect on hay fever/eczema only compared to their effects on asthma, while 26 loci exhibited larger effects on asthma compared with their effects on hay fever/eczema. At four of the novel loci, TNFRSF8, MYRF, TSPAN8, and BHMG1, the lead SNPs were in Linkage Disequilibrium (LD) (>0.8) with potentially casual missense variants. Our study shows that a large amount of the genetic contribution is shared between the diseases. Nonetheless, a number of SNPs have a significantly larger effect on one of the phenotypes, suggesting that part of the genetic contribution is more phenotype specific.

Figure 1

Manhattan plots for asthma, for hay fever and/or eczema, and for asthma and/or hay fever and/or eczema (combined) for autosomal chromosomes. The black horizontal line indicates the genome-wide threshold (3 × 10⁻⁸). The black regions represent novel loci found in this study.

Figure 2

Comorbidity between asthma and hay fever/eczema.

Figure 3

Venn diagram showing the phenotype specificity of the GWA loci, based on the results from the polytomous logistic regression analyses. The Venn diagram show loci (SNPs) that are specific (significantly larger effect) to or shared between (no significant difference in effects) between two non-overlapping groups of cases. The name of each locus is denoted by the most likely gene(s). At some of the loci (e.g. IL2RA, LPP, and IL4R), more than one independent (R² < 0.8) lead SNP has been analyzed in the polytomous logistic regression. If those showed different specificity pattern, they have been included twice in the figure with the name of respective lead SNP(s) also included in the locus name. P-values and estimates for the genes can be found in Table S16 where the area number 1 (green in the figure) indicates specificity for the asthma only; area number 2 (blue in the figure), specificity for hay fever/eczema only; and area number 3 (white in the figure), specificity for asthma with hay fever/eczema (significantly larger estimate in the asthma with hay fever/eczema group of cases).