Unifying candidate gene and GWAS Approaches in Asthma

Abstract

The first genome wide association study (GWAS) for childhood asthma identified a novel major susceptibility locus on chromosome 17q21 harboring the ORMDL3 gene, but the role of previous asthma candidate genes was not specifically analyzed in this GWAS. We systematically identified 89 SNPs in 14 candidate genes previously associated with asthma in >3 independent study populations. We re-genotyped 39 SNPs in these genes not covered by GWAS performed in 703 asthmatics and 658 reference children. Genotyping data were compared to imputation data derived from Illumina HumanHap300 chip genotyping. Results were combined to analyze 566 SNPs covering all 14 candidate gene loci. Genotyped polymorphisms in ADAM33, GSTP1 and VDR showed effects with p-values <0.0035 (corrected for multiple testing). Combining genotyping and imputation, polymorphisms in DPP10, EDN1, IL12B, IL13, IL4, IL4R and TNF showed associations at a significance level between p = 0.05 and p = 0.0035. These data indicate that (a) GWAS coverage is insufficient for many asthma candidate genes, (b) imputation based on these data is reliable but incomplete, and (c) SNPs in three previously identified asthma candidate genes replicate in our GWAS population with significance after correction for multiple testing in 14 genes.

Figure 1

a) Schematic depiction of literature query process. * genes not further investigated were (a) ORMDL3, originally discovered in this GWA; (b) discovered by other GWAS: CHI3L1, IL1RL1, MYB, WDR36 (c) NOS1 where CA-repeats and not SNPs were reported; (d) gene deletions for GSTM1, FLG. b) Number of populations that show an association between SNPs in the 14 selected candidate genes and asthma diagnosis. (based on literature query from October 22^nd 2009).

Figure 2. Correlation between effect sizes observed by concomitant genotyping and imputation of candidate gene SNPs for which both measures were available.

(red indicates change of effect direction between genotyping and imputation). For the comparison based on HapMapII (HM2) data 241 SNPs and for HapMapIII (HM3) 216 SNPs were available.

Figure 3. Comparison of effect sizes and p-values derived from regenotyped SNPs with data based on imputation.

Log transferred p-values are shown on the y-axis. Each individual SNP effect is depicted as a circle where the diameter of the circle reflects the respective odds ratios. The dark grey circles represent the regenotyping the light grey circles stand for the imputation based on HapMap2 and the transparent show the imputation data based on HapMap3. Significance level are marked as solid (p = 0.05) and dashed grey lines (p = 0.0035).

Figure 4. Association of 566 tagging SNPs in 14 candidate genes with asthma.

Genes are positioned along the x-axis in alphabetical order. Each dot represents SNPs in their relative position within the gene. Log transferred p-values are shown on the y-axis. Each individual SNP effect is depicted as a circle where the diameter of the circle reflects the respective odds ratios. Significance level are marked as solid (p = 0.05) and dashed green lines (p = 0.0035).

Figure 5. Study Design and population selection strategy.

(ISAAC  =  International Study for Asthma and Allergies in Childhood; MAGICS  =  Multicentre Asthma Genetics in Childhood Study; GWAS  =  Genome wide Association Study; QC  =  quality control).