Background: Otitis media (OM) is a common childhood disease characterised by middle ear inflammation and effusion. Susceptibility to recurrent acute OM (rAOM; ≥ 3 episodes of AOM in 6 months) and chronic OM with effusion (COME; MEE ≥ 3 months) is 40-70% heritable. Few underlying genes have been identified to date, and no genome-wide association study (GWAS) of OM has been reported.
Methods and findings: Data for 2,524,817 single nucleotide polymorphisms (SNPs; 535,544 quality-controlled SNPs genotyped by Illumina 660W-Quad; 1,989,273 by imputation) were analysed for association with OM in 416 cases and 1,075 controls from the Western Australian Pregnancy Cohort (Raine) Study. Logistic regression analyses under an additive model undertaken in GenABEL/ProbABEL adjusting for population substructure using principal components identified SNPs at CAPN14 (rs6755194: OR = 1.90; 95%CI 1.47-2.45; P(adj-PCA) = 8.3 × 10(-7)) on chromosome 2p23.1 as the top hit, with independent effects (rs1862981: OR = 1.60; 95%CI 1.29-1.99; P(adj-PCA) = 2.2 × 10(-5)) observed at the adjacent GALNT14 gene. In a gene-based analysis in VEGAS, BPIFA3 (P(Gene) = 2 × 10(-5)) and BPIFA1 (P(Gene) = 1.07 × 10(-4)) in the BPIFA gene cluster on chromosome 20q11.21 were the top hits. In all, 32 genomic regions show evidence of association (P(adj-PCA)<10(-5)) in this GWAS, with pathway analysis showing a connection between top candidates and the TGFβ pathway. However, top and tag-SNP analysis for seven selected candidate genes in this pathway did not replicate in 645 families (793 affected individuals) from the Western Australian Family Study of Otitis Media (WAFSOM). Lack of replication may be explained by sample size, difference in OM disease severity between primary and replication cohorts or due to type I error in the primary GWAS.
Conclusions: This first discovery GWAS for an OM phenotype has identified CAPN14 and GALNT14 on chromosome 2p23.1 and the BPIFA gene cluster on chromosome 20q11.21 as novel candidate genes which warrant further analysis in cohorts matched more precisely for clinical phenotypes.