An 8-year-old Chinese girl presented with hearing loss after suspicion based on regression of speech. She is the second child born to healthy, non-consanguineous parents and has a healthy elder sister. No significant medical history was noted during the prenatal, delivery, and postpartum periods. The patient successfully passed the newborn hearing screening, with no abnormalities detected at that time. Her early language milestones were within the expected range. She began babbling at 8–9 months and spoke her first words by 12 months. By the age of 2 years, she was capable of speaking longer words. However, a gradual regression in her language skills was observed starting at the age of 3 years. Currently, her verbal communication is predominantly limited to the use of repetitive and simple words. She is using a hearing aid, which provides some improvement in her auditory capabilities.

The girl was thoroughly examined in the department of paediatrics and otolaryngology at Ganzhou Maternal and Child Health Hospital. Detailed medical histories and physical examinations, such as computerized tomography (CT) scans of the temporal bonehead, magnetic resonance (MR) labyrinthography and eye fundus were carried out to exclude any possible environmental causes or syndromic forms of hearing loss. Audiological evaluation included tympanometry, distortion product otoacoustic emission (DPOAE), click-evoked air-conduction auditory brain-stem response (AC-ABR), click-evoked bone-conduction auditory brain-stem response (BC-ABR), pure-tone audiometry (PTA). The hearing levels were determined by the thresholds of AC-ABR, NB CE-chirp ASSR, or PTA. Further, according to the World Report on Hearing in 2021, the grades of hearing loss were classified as mild (20 to < 35 dB HL), moderate (35 to < 50 dB HL), moderately severe (50 to < 65 dB HL), severe (65 to < 80 dB HL), or profound (80 to < 95 dB HL), complete or total (> 95 dB HL) by the average threshold of PTA air conduction in 500 Hz, 1000 Hz, 2000 Hz and 4000 Hz. In order to estimate the progress of her hearing levels, we evaluated her hearing in her 7 and 7.5 years old respectively by DPOAE, AC-ABR and BC-ABR. PTA was conducted when she was 8 years old.

This research was approved by the Ethics Committee of the Ganzhou Maternal and Child Health Hospital. Peripheral venous blood samples were taken from the girl, her parents and her elder sister following the acquisition of informed consent from all participants. RNA was extracted using the PAXgene Blood RNA Kit (PreAnalytiX, Hombrechtikon, Switzerland). In addition, DNA was extracted from the peripheral blood using the QIAamp DNA Mini Kit for blood (Qiagen, Hilden, Germany). Trio-based WES was performed by KingMed Diagnostics (Guangzhou, China), including exome library preparation, sequencing, and data analysis. Variant screening was based on clinical phenotypes of the affected subjects, population database (dbSNP, 1,000 Genome, ExAC), disease database (OMIM, HGMD, Clinvar), and biological information prediction tools (SIFT, Polyphen2, Mutation Taster and Splice AI). Two novel variants of CDH23 were identified in the girl, one is a splicing variant (CDH23: c.2398-6G > A) inherited her mother, and the another is a nonsense variant (CDH23: c.6068C > A), inherited from her father. The two CDH23 variants were confirmed by Sanger sequencing. To identify the potential impact of the c.2398-6G > A noncanonical splice variant of CDH23, the RDDCSC online in silico splice site prediction software (https://rddc.tsinghua-gd.org/search-middle?to=SplitToolModel, accessed on 23 April 2024) was used.

Following prior confirmation of CDH23 expression in blood, samples from the patient’s mother and a control were utilized to analyze the splicing pattern of exons 22, 23, and 24 via RT-PCR. Their blood was collected in PAXgene Blood RNA tubes, and RNA was extracted using PAXgene Blood RNA Kit (PreAnalytiX, Hombrechtikon, Switzerland). Two rounds of PCR were performed using nested primers: the first PCR was performed using Complementary DNA (a total of two sets of DNA) as a template, with CDH23-1F (5-GAAATCACCACCACGTCTCT-3) and CDH23-1R (5-TGCTGCTGTTGATGAGGAAG-3) as primers for 30 cycles. The second PCR was performed using products from the first round of PCR as a template, with CDH23-2F (5-GTGGGCCACAACCAGAAAAC-3) and CDH23-2R (5-TTCAAGCACCTCGGCCACAA-3) as primers for 30 cycles.

The protein sequence with 3354 amino acid residues of CDH23 was downloaded from uniprot web (https://www.uniprot.org/). The wild-type, mutant-type1 (CDH23:c.6068C > A) and mutant-type2 (CDH23:c.2398-6G > A) 3D structure of the CDH23 protein were predicted using AlphaFold web server (https://golgi.sandbox.google.com/about) (Abramson et al. 2024). The best model was selected based on pLDDTs prediction scores. The higher the score, the more confident is the structure. Prediction models editing was performed was visualized using PyMOL program (https://pymol.org/).

The protein–protein interaction (PPI) network associated with CDH23 was constructed using STRING (https://string-db.org/). The amino acid sequence of the CDH23 protein, comprising 3354 residues, was retrieved from the UniProt database (https://www.uniprot.org/). This sequence was used as the input query for STRING, where the parameters were set to ensure high-confidence interactions were captured. The confidence score threshold was adjusted to include only interactions supported by strong evidence such as experimental data, co-expression, and co-occurrence analyses.