Children born deaf heard their parents for the first time

Researchers at Fudan University and Mass Eye and Ear delivered a working copy of the OTOF gene into the inner ear of six children born deaf; five turned toward voices within weeks.

One gene, one protein, one missing step

Autosomal recessive deafness 9, usually written DFNB9, is caused by mutations in a single gene called OTOF. The gene was identified in 1999 by Shin'ichiro Yasunaga, Christine Petit and colleagues, who found the same nonsense mutation in four unrelated Lebanese families with profound prelingual deafness. OTOF encodes otoferlin, a protein of 1,230 amino acids with three C2 domains and a single transmembrane anchor. Its structure pointed to a role in vesicle membrane fusion, and in the mouse inner ear the gene was expressed mainly in the sensory hair cells.

That placement explains the disease. The cochlea's inner hair cells convert sound into a chemical signal by releasing neurotransmitter onto the auditory nerve. Otoferlin is the calcium sensor that drives that release. A child born without working otoferlin has hair cells that detect sound normally and a nerve that is intact, but the synapse between them does not fire. The hardware is present. One protein is absent.

Until this trial the only treatment for these children was a cochlear implant, which bypasses the hair cells entirely and stimulates the nerve with electrodes. The implant works, but it discards the cochlea's own machinery. A gene therapy that put otoferlin back would be the first treatment to use it.

A gene too large for the delivery vehicle

The obstacle was size. Adeno-associated virus, the standard vector for delivering genes to the inner ear, cannot carry a transgene as large as otoferlin's. The gene does not fit.

The solution, worked out in mice, was to split the gene in two. In 2019 Hanan Al-Moyed, Ellen Reisinger and colleagues reported in EMBO Molecular Medicine that two AAV vectors, each carrying half of the otoferlin complementary DNA, could be co-injected into the cochleae of otoferlin knockout mice six to seven days old. Inside the hair cells the two halves recombined into a full-length gene. Full-length otoferlin was expressed in up to 50 percent of inner hair cells, fast exocytosis at the synapse was fully restored, and clicks evoked auditory brainstem responses with thresholds of 40 to 60 decibels in animals that had previously heard nothing.

The rescue was partial. Vesicle replenishment reached only 35 to 50 percent of wild-type levels, and the mice had already lost about 40 percent of their synaptic ribbons, a loss the therapy did not prevent. But the dual-vector approach had worked in a mammal, and that was the proof needed to move toward children.

Six children in Shanghai

The human trial was run at the Eye and ENT Hospital of Fudan University in Shanghai, led by Yilai Shu, with Zheng-Yi Chen of Mass Eye and Ear and Harvard Medical School as co-leader. It was a single-arm, single-centre study registered with the Chinese Clinical Trial Registry as ChiCTR2200063181. Between October 19, 2022 and June 9, 2023 the team screened 425 candidates and enrolled six children aged one to seven years, each with severe-to-complete hearing loss, confirmed mutations in both copies of OTOF, and no bilateral cochlear implants.

Each child received a single injection into one cochlea through the round window membrane. The vector was AAV serotype 1 carrying the human OTOF gene in two halves. One child received 9 × 10¹¹ vector genomes; five received 1.5 × 10¹². The primary endpoint was dose-limiting toxicity at six weeks. All six completed follow-up to week 26.

The results were published in The Lancet on January 24, 2024, by Jun Lv and colleagues. No dose-limiting toxicity and no serious adverse event occurred. Forty-eight adverse events were recorded; 46 were grade 1 or 2, and two were grade 3, both a reduced neutrophil count in one participant.

Five of six

Hearing was measured by the auditory brainstem response, or ABR, an electrical recording from the scalp that does not depend on the child cooperating. Before treatment every child had an average threshold above 95 decibels at 0.5 to 4 kilohertz, meaning that sound at that intensity produced no detectable neural response.

Five of the six children recovered hearing, with average ABR thresholds falling by 40 to 57 decibels. In the child who received the lower dose, the threshold improved from above 95 decibels at baseline to 68 at four weeks, 53 at 13 weeks and 45 at 26 weeks. In the four responders at the higher dose, thresholds at 26 weeks were 48, 38, 40 and 55 decibels. Speech perception improved in every child whose hearing recovered. According to Harvard's account of the trial, the first signs of hearing appeared four to six weeks after injection, and by 26 weeks three of the older children, with their cochlear implants switched off, could understand speech and respond to it; two could follow speech in a noisy room and hold a telephone conversation.

One child did not respond. The paper does not establish why.

Both ears

The protocol was then amended to allow treatment of both cochleae. In June 2024 the same group reported in Nature Medicine an interim analysis of five additional children given bilateral injections. All five had bilateral ABR thresholds above 95 decibels at baseline. At 26 weeks the right and left thresholds were 58 and 58 decibels in the first patient, 75 and 85 in the second, and 55 and 50 in the third; at 13 weeks they were 75 and 78 in the fourth and 63 and 63 in the fifth. Thirty-six adverse events were recorded, the most common being raised lymphocyte counts and raised cholesterol, and none was serious. Speech perception and the ability to locate the source of a sound, which requires two working ears, were restored in all five.

Against the implant

The question that follows is how the therapy compares with the standard treatment. In 2025 the Fudan group published in JAMA Neurology a non-blinded cohort study of 72 children with congenital deafness, 11 treated with gene therapy and 61 with cochlear implants, matched on duration of deafness, baseline thresholds and baseline speech ability. In nine gene therapy patients with 12-month data, mean ABR thresholds were restored from above 95 to 54.8 decibels. On the parent-reported Meaningful Auditory Integration Scale, gene therapy patients scored higher than implant patients at six months (median 31.0 versus 23.5) and at 12 months (32.0 versus 28.0). Children who had gene therapy in one ear and an implant in the other performed better on speech in noise with the implant switched off than children with a single implant.

These are small numbers in a non-randomised comparison at a single hospital, and the gene therapy children were older at baseline (mean 3.7 years versus 1.9). The finding is suggestive, not conclusive.

What the story does and does not show

The 2024 Lancet trial showed that a single injection of a dual-AAV OTOF vector restored measurable hearing in five of six children with DFNB9 over 26 weeks, with no serious adverse events. The follow-on study showed the same in both ears in five more children. The 2025 comparison suggested outcomes at least as good as cochlear implantation in a small matched cohort.

The story does not show that gene therapy works for deafness in general. OTOF mutations account for a small fraction of congenital hearing loss, and the therapy works because the hair cells in these children are intact and need only one missing protein. Most genetic deafness involves hair cells that are damaged or absent, which this approach does not address. The trials were single-arm, single-centre and small, with the longest published follow-up measured in months. Thresholds of 38 to 68 decibels are hearing, and they are enough for speech, but they are not normal hearing. How long the effect lasts, whether it holds as the cochlea grows, and whether a second injection would be possible after immunity to the vector develops, are open questions.

What the sequence does show is that a gene found in four Lebanese families in 1999 was split in two and reassembled inside the hair cells of deaf mice in 2019, and that five years later the same method produced auditory brainstem responses in children who had never had them.

One injection, and hearing switched on.