Most experimental tinnitus drugs try to quiet the brain. AC102 tries to repair the ear. In noise-damaged animals, a single dose restored lost connections in the inner ear, and the signs of tinnitus nearly disappeared. Here is what that study showed, where the human trials stand, and what it might mean for those of us who have lived with the sound for decades.
Read in: English · Español · Português · Français · Deutsch · हिन्दी · 中文 · العربية
Promising in animals, safe in a first human trial, unproven for tinnitus in people.
Nearly every drug tried for tinnitus works on the brain. Anticonvulsants, benzodiazepines, antidepressants, and the primidone I wrote about in an earlier issue all aim to calm overactive nerve signaling or the distress that comes with it. At best, they turn the volume down while you take them.
AC102 starts at the other end. It is designed to protect and restore the inner ear itself, specifically the tiny connections, called ribbon synapses, that link the sensory hair cells of the cochlea to the fibers of the auditory nerve.
Those connections are fragile. Loud noise can break them even when the hair cells survive and a hearing test comes back normal. Researchers call this hidden hearing loss. When the brain stops receiving its normal input from the ear, the auditory system appears to turn up its own internal gain to compensate, and one leading theory holds that this compensation is what we hear as tinnitus.
If the missing input is a cause of the sound, then restoring the input might take away the brain’s reason to produce it. That is the idea behind AC102.
AC102 is a small molecule from a drug family called pyridoindoles. In earlier animal work it reduced inflammation and cell death in the inner ear, improved how cells there produce energy, and promoted the regrowth of nerve fibers. It is given as a single injection through the eardrum into the middle ear, suspended in a gel that is liquid at room temperature and firms up at body temperature, so it rests against the cochlea and seeps in slowly.
The study, published in May 2025 in the International Journal of Molecular Sciences, came from the University Hospital Erlangen in Germany working with AudioCure. Researchers exposed Mongolian gerbils to loud noise in one ear: enough to break ribbon synapses and trigger tinnitus behavior, not enough to kill hair cells. Then they treated the damaged ear with either AC102 gel or the same gel without the drug.
Animals cannot tell us what they hear, so tinnitus was measured indirectly. A sudden loud sound makes a gerbil startle, and a brief silent gap just before it normally softens the startle. An animal whose silence is filled with ringing cannot detect the gap as well. It is a widely used test, but it is a stand-in for tinnitus, not a report of it.
Single-dose experiment. Both groups started at the same level; only the treated group recovered.
Behavioral measure in gerbils, not human tinnitus reports. Source: Tziridis et al., Int J Mol Sci, 2025.
The pattern is what makes this study stand out. A few days after the noise, roughly six in ten animals in both groups showed tinnitus behavior, so the treatment did not prevent it from starting. Over the next five weeks, the untreated animals stayed stuck. By week five, only one of the fifteen treated animals still showed it.
A second experiment, using three weekly doses in a separate group of animals, pointed the same way: at five weeks, 1 of 11 treated gerbils still showed tinnitus behavior, compared with 6 of 12 untreated ones.
When the researchers examined the cochleas afterward, the untreated ears had lost a significant number of ribbon synapses. The treated ears had lost far fewer; in the three-dose group they ended up with slightly more synapses than the animals’ unexposed ear. Hearing in the treated ears also returned to, or within a few decibels of, where it started, while the untreated ears were still measurably impaired at five weeks.
These are striking results, and they come with limits worth stating plainly.
Safety. The first human trial, run at Radboud University Medical Center in the Netherlands and the Medical University of Vienna, gave a single injection to 42 healthy volunteers: 21 received AC102 and 21 received the gel alone. There were no serious side effects. Nearly everyone had something mild and brief, mostly ear discomfort, along with some ear pain, dizziness, a muffled feeling, and in some cases temporary tinnitus. These happened about as often with the plain gel as with the drug, and the researchers attributed them to the injection and the gel filling the middle ear. Nearly all cleared within a week, and hearing returned to normal in everyone.
Effectiveness. The Phase 2 trial is testing AC102 for sudden hearing loss, not tinnitus. Patients receive one AC102 injection within five days of losing hearing, and the results are compared with oral steroid tablets, today’s standard treatment. It is double-blind and planned for roughly 200 patients at more than 45 hospitals in Austria, Germany, the Czech Republic, the Netherlands, Poland, Serbia, and Ukraine.
EU Clinical Trials Information System · 2024-513658-31-00 · reviewed October 2026
Trial status: Ended. Overall end date: 16 January 2026. No results summary posted.
The American registry, ClinicalTrials.gov (NCT05776459), has not caught up and still lists the trial as active. Either way, nine months after the European end date, no results have been published. That is not unusual. Companies often take a year or more to analyze and release trial data. But it means the central question, whether AC102 works in people at all, is still unanswered.
Many people with sudden hearing loss also develop tinnitus, so the Phase 2 results may offer the first human hint about tinnitus. A dedicated tinnitus trial has not been registered.
This is the question I care about most, and the honest answer is that nobody knows.
Long-standing tinnitus is probably not only an ear problem. After years, the brain’s own networks for attention, emotion, and sound may have reorganized around the ringing. Even if AC102 could restore some lost connections in an old injury, it is unknown whether that new input would undo patterns the brain has held for twenty or thirty years.
There is also an open question about what AC102 actually does. In the gerbil study, it is not fully clear whether the drug preserved synapses that were about to die or regrew ones already lost. For someone injured long ago, that distinction is everything. Protection does nothing for connections that disappeared decades ago; regrowth might.
Still, I would not dismiss the possibility. If damaged synapses keep starving the brain of normal input, restoring them might help even late, perhaps not by silencing the sound, but by giving the brain a reason to stop amplifying it. That is a reasonable hypothesis. It is not yet a demonstrated effect, and only a trial that enrolls people with chronic tinnitus can test it.
AC102 belongs on a short list of experimental treatments worth following. It targets a real, measurable injury rather than masking a symptom, it has passed its first safety test in people, and its animal results are among the strongest I have seen for any tinnitus drug.
But it helps to separate two different promises. Preventing tinnitus right after a noise injury or sudden hearing loss is where the evidence points, and it is the nearer-term possibility. Reversing chronic tinnitus is much less certain and has not been studied at all.
When the Phase 2 results are published, Tinnitus Monthly will report what they show, good or bad.
Tinnitus research and device news from around the world, translated into plain English. One issue a month. No hype, no selling.