Genomic study of nonsyndromic hearing loss in unaffected individuals: Frequency of pathogenic and likely pathogenic variants in a Brazilian cohort of 2,097 genomes

Abstract

Hearing loss (HL) is a common sensory deficit in humans and represents an important clinical and social burden. We studied whole-genome sequencing data of a cohort of 2,097 individuals from the Brazilian Rare Genomes Project who were unaffected by hearing loss to investigate pathogenic and likely pathogenic variants associated with nonsyndromic hearing loss (NSHL). We found relevant frequencies of individuals harboring these alterations: 222 heterozygotes (10.59%) for sequence variants, 54 heterozygotes (2.58%) for copy-number variants (CNV), and four homozygotes (0.19%) for sequence variants. The top five most frequent genes and their corresponding combined allelic frequencies (AF) were GJB2 (AF = 1.57%), STRC (AF = 1%), OTOA (AF = 0.69%), TMPRSS3 (AF = 0.41%), and OTOF (AF = 0.29%). The most frequent sequence variant was GJB2:c.35del (AF = 0.72%), followed by OTOA:p. (Glu787Ter) (AF = 0.61%), while the most recurrent CNV was a microdeletion of 57.9 kb involving the STRC gene (AF = 0.91%). An important fraction of these individuals (n = 104; 4.96%) presented variants associated with autosomal dominant forms of NSHL, which may imply the development of some hearing impairment in the future. Using data from the heterozygous individuals for recessive forms and the Hardy-Weinberg equation, we estimated the population frequency of affected individuals with autosomal recessive NSHL to be 1:2,222. Considering that the overall prevalence of HL in adults ranges from 4-15% worldwide, our data indicate that an important fraction of this condition may be associated with a monogenic origin and dominant inheritance.

Keywords: GJB2 (C×26) gene mutations; STRC gene; deafness; genomics; hearing loss; nonsyndromic hearing loss; whole genome sequencing.

FIGURE 1

Flowcharts showing the selection workflow for patients, genes, and variants. The first column shows the patient selection workflow: 2,199 patients who had been referred for molecular investigation using whole genome sequencing from 2020–2021 were preselected; 13 failed sample quality controls, and 89 patients presented with hearing loss (88 syndromic hearing loss and one with nonsyndromic hearing loss [NSHL]): all were excluded from our cohort and population calculations. The second column shows gene selection: 129 genes associated with NSHL were preselected from the OMIM database, reviews (Shearer et al., 2017; Van Camp and Smith, 2021), and other available literature; three of them were initially excluded because one of them (DFNX3) refers to a locus without a known gene, another (KCNJ10) did not present a valid transcript in MANE or RefSeq databases, and the other (ATP2B2) was not associated with a Mendelian form of hearing loss; four genes (FOXI1, GJB3, KCNJ10, and TSPEAR) were excluded for presenting disputed gene-disease association and another gene (MYO1A) refuted association. The third column shows the variant selection workflow: 1) for sequence variants, 996 variants were preselected in 5,330 variant coordinates, but only 118 were reported in ClinVar as pathogenic or likely pathogenic (P/LP); the remaining 878 were excluded; after our internal curation and reclassification, 29 variants were excluded because they were reclassified as variants of unknown significance (VUS, n = 28) or likely benign (LB, n = 1); 2) for copy-number variants (CNV), the filtering process of 5,285 preselected CNV events eliminated 5,053 CNVs; our internal curation, visual inspection, and classification eliminated another 220.