Lessons from ten years of genome-wide association studies of asthma

Abstract

Twenty-five genome-wide association studies (GWAS) of asthma were published between 2007 and 2016, the largest with a sample size of 157242 individuals. Across these studies, 39 genetic variants in low linkage disequilibrium (LD) with each other were reported to associate with disease risk at a significance threshold of P<5 × 10^-8, including 31 in populations of European ancestry. Results from analyses of the UK Biobank data (n=380 503) indicate that at least 28 of the 31 associations reported in Europeans represent true-positive findings, collectively explaining 2.5% of the variation in disease liability (median of 0.06% per variant). We identified 49 transcripts as likely target genes of the published asthma risk variants, mostly based on LD with expression quantitative trait loci (eQTL). Of these genes, 16 were previously implicated in disease pathophysiology by functional studies, including TSLP, TNFSF4, ADORA1, CHIT1 and USF1. In contrast, at present, there is limited or no functional evidence directly implicating the remaining 33 likely target genes in asthma pathophysiology. Some of these genes have a known function that is relevant to allergic disease, including F11R, CD247, PGAP3, AAGAB, CAMK4 and PEX14, and so could be prioritized for functional follow-up. We conclude by highlighting three areas of research that are essential to help translate GWAS findings into clinical research or practice, namely validation of target gene predictions, understanding target gene function and their role in disease pathophysiology and genomics-guided prioritization of targets for drug development.

Figure 1

Effect size (odds ratio) for 31 previously reported asthma risk SNPs, comparing results reported in the original GWAS describing each association with those obtained in the analysis of the UK Biobank study. Each SNP is represented by a circle, with its size and color indicating total GWAS sample size (as per Supplementary Table 2) and PubMed identifier (PMID), respectively. The red diagonal line represents equality of odds ratio between published GWAS and UK Biobank (that is, x=y). The genomic location of the four SNPs that deviate markedly from the diagonal are shown next to the respective circle. For three of these (SNPs in PDE4D, CRB1 and CDHR3), the association in the UK Biobank study had a P-value>0.05 (see Table 2).