Targeted Resequencing of Otosclerosis Patients from Different Populations Replicates Results from a Previous Genome-Wide Association Study

Abstract

Otosclerosis is one of the most common causes of hearing loss in young adults. It has a prevalence of 0.3-0.4% in the European population. Clinical symptoms usually occur between the second and fifth decade of life. Different studies have been performed to unravel the genetic architecture of the disease. Recently, a genome-wide association study (GWAS) identified 15 novel risk loci and replicated the regions of three previously reported candidate genes. In this study, seven candidate genes from the GWAS were resequenced using single molecule molecular inversion probes (smMIPs). smMIPs were used to capture the exonic regions and the 3' and 5' untranslated regions (UTR). Discovered variants were tested for association with the disease using single variant and gene-based association analysis. The single variant results showed that 13 significant variants were associated with otosclerosis. Associated variants were found in five of the seven genes studied here, including AHSG, LINC01482, MARK3, SUPT3H and RELN. Conversely, burden testing did not show a major role of rare variants in the disease. In conclusion, this study was able to replicate five out of seven candidate genes reported in the previous GWAS. This association is likely mainly driven by common variants.

Keywords: gene analysis; hearing loss; otosclerosis; replication; targeted resequencing.

Figure 1

QQ plot of results of single SNP variant analysis and gene-based analysis. The QQ plot shows the observed and expected distribution of p-values of all variants used for single variant analysis. The diagonal line shows the expected distribution in absence of any association and the dots and squares represent the observed p-values. The QQ plot of the single SNP variant analysis (dots) shows a clear diversion of the dots compared to the red line, which indicates an enrichment of significant p-values. The observed p-values after gene-based analysis (squares) show only a slight diversion, indicating that there is probably no cumulative effect of (very) rare variants on the phenotype. This observation is in line with the lack of significance upon correcting the p-values for the number of tests.

Figure 2

Forest plot of odds ratios of all 13 significant variants across all subpopulations and the total population. For each variant the odds ratio (OR) is represented by a square and the 95% confidence interval by a horizontal line. Association is considered to be consistent across all populations when the OR is lower (or higher) than one across all subpopulations. The plot shows a (near) consistent association in 9 out of 13 variants.