Combined genetic approaches yield a 48% diagnostic rate in a large cohort of French hearing-impaired patients

Abstract

Hearing loss is the most common sensory disorder and because of its high genetic heterogeneity, implementation of Massively Parallel Sequencing (MPS) in diagnostic laboratories is greatly improving the possibilities of offering optimal care to patients. We present the results of a two-year period of molecular diagnosis that included 207 French families referred for non-syndromic hearing loss. Our multi-step strategy involved (i) DFNB1 locus analysis, (ii) MPS of 74 genes, and (iii) additional approaches including Copy Number Variations, in silico analyses, minigene studies coupled when appropriate with complete gene sequencing, and a specific assay for STRC. This comprehensive screening yielded an overall diagnostic rate of 48%, equally distributed between DFNB1 (24%) and the other genes (24%). Pathogenic genotypes were identified in 19 different genes, with a high prevalence of GJB2, STRC, MYO15A, OTOF, TMC1, MYO7A and USH2A. Involvement of an Usher gene was reported in 16% of the genotyped cohort. Four de novo variants were identified. This study highlights the need to develop several molecular approaches for efficient molecular diagnosis of hearing loss, as this is crucial for genetic counselling, audiological rehabilitation and the detection of syndromic forms.

Figure 1

Decision-making tree for molecular diagnosis of isolated hearing loss. *74 HL genes, see Supplementary Table 1. **See Supplementary Figure 1. ^$CNV: Copy Number Variation, deletions/duplications involving at least one exon and occurring in an isoform involved in hearing. ^$$aCGH: array Comparative Genomic Hybridization; QMPSF: Quantitative Multiplex PCR of Short Fluorescent fragments. All SNVs and small indels are confirmed by Sanger sequencing or long-range PCRs followed by Sanger sequencing for STRC variants.

Figure 2

Minigene analysis to assess the impact of variants on splicing. Full-length gels showing: (A) STRC variants c.3100-2 A > T and c.3100-18 G > A leading to complete skipping of exon 12. The lower band observed for the wild-type construction may be due to an artefact of the minigene system or alternative splicing of exon 12; (B) TJP2 c.2880G > A leading to complete skipping of exon 19; (C) USH2A variant c.14134-3169 A > G leading to the inclusion of a pseudoexon between native exons 63 and 64 (PE64); bold: termination codon.

Figure 3

3D analysis of POU4F3 p.(Phe322Ser) using homolog Oct-1 PDB structure 1E3O. The homeodomain directly binds DNA through helix 3 and the positively charged amino acids Arg or Lys (red). The equivalent of Phe 322 (A, white), also located in helix 3, is involved in the hydrophobic core with helices 1 and 2 stabilizing the region. Introduction of a small polar amino acid (Ser, B, white) is likely to modify the hydrophobic core and the domain and might modify DNA binding properties of the protein.

Figure 4

Diagnostic rates of the French NSHL cohort. After DFNB1 screening, 158 patients underwent MPS and pathogenic genotypes could be defined for 50 of them, involving 18 different genes.