Optimising POU3F4 variant interpretation through gene-specific evidence in X-linked hearing loss

BACKGROUND: Standardised variant interpretation frameworks inadequately accommodate the distinct mutational and phenotypic architectures inherent to specific genes or diseases. POU3F4 exemplifies this limitation in X-linked hearing loss, where stringent genotype-phenotype correlations and localised mutational hotspots dominate the pathogenic landscape. We sought to establish a calibrated, gene-specific evidentiary model to resolve interpretive inconsistencies for POU3F4. METHODS: We sequenced 20,666 unrelated individuals with hearing loss and 7,258 controls using targeted panel and genome sequencing to capture single-nucleotide, indel, and structural variants in POU3F4. By integrating clinical phenotyping, paralog-based residue constraint modelling, and Bayesian likelihood-ratio analysis, we redefined evidence criteria for this gene. Representative stop-loss alleles underwent functional characterisation to elucidate their molecular consequences. FINDINGS: Genomic analysis identified 123 distinct POU3F4 variants, 87 of which were classified as pathogenic (P) or likely pathogenic (LP). Structural variants accounted for 29 of these P/LP variants, with breakpoints enriched in repeat-dense sequence across the locus. Incomplete partition type III cochlear malformation showed near-diagnostic coupling to POU3F4, with 96.4% of affected cases carrying P/LP variants, supporting escalation of PP4 to strong evidence. P/LP missense variants were significantly concentrated at paralog-conserved residues, and Bayesian modelling of this constraint produced a positive likelihood ratio of 33.0 (95% confidence interval 10.68-102.28), consistent with moderate-strength hotspot evidence of PM1. Applying the calibrated rules reclassified 17 single nucleotide variants or small indels and resolved 16 of 34 variants of uncertain significance. Functional assays demonstrated that stop-loss variants generate hydrophobic C-terminal extensions that destabilise POU3F4, driving nuclear depletion and loss of transcriptional activity. Enforced nuclear targeting restored localisation but not transcription, implicating protein instability rather than impaired nuclear import as the dominant mechanism. INTERPRETATION: POU3F4 pathogenicity is defined by a highly specific cochlear signature and profound topological constraint. Quantitative calibration of these features substantially refines diagnostic resolution in POU3F4-mediated hearing loss. The convergence of genomic modelling and functional validation establishes a calibrated evidentiary framework for consistent variant classification. FUNDING: This work was supported by National Natural Science Foundation of China (82530036, 82471889, 82171836), National Key Research and Development Program of China (2024YFC3405704), West China Hospital, Sichuan University 1.3.5 Project for Disciplines of Excellence grant ZYJC20002, Sichuan Provincial Natural Science Foundation (2024NSFSC0648).