11Cochlear synaptopathy
A patient with a normal audiogram who cannot follow speech in noise. The animal data are unambiguous: noise and aging silently destroy the synapses between inner hair cells and the auditory nerve, leaving thresholds intact but degrading the suprathreshold neural code. The human translation is contested. This module is about what is known, what is hoped, and what remains a research question.
This module differs in tone from the diagnostic-confidence-builders of Ménière's, SCD, and ANSD. Here ECochG sits on the research frontier: an SP/AP ratio measurement that might be the first non-invasive marker for a histologically real but clinically still-being-defined entity. The honest assessment from the field is that synaptopathy is well-established in animals, demonstrated in human post-mortem temporal bones, but remains without a validated non-invasive diagnostic test in living humans. The clinical lesson is to understand the concept and the evidence, not to apply a confident test result.
FThe concept
The traditional model of hearing loss centred on the hair cells: outer hair cell loss raises pure-tone thresholds (the audiogram), inner hair cell loss removes the sensory input entirely. Both produce audible signs that the audiogram detects. The synaptopathy hypothesis adds a third layer of damage — loss of the ribbon synapses that connect inner hair cells to type I auditory nerve fibres — that the audiogram does not detect.
The key empirical claim, established by Kujawa and Liberman in 2009 in mice, is that this synaptic loss is:[2009]
- Caused by noise that produces only temporary threshold shifts — exposures the audiogram-relevant hair cells fully recover from.
- Permanent in mammalian species without synaptic regeneration capacity (mice, rats, and importantly humans; guinea pigs may partially regenerate).
- Selective for low- and medium-spontaneous-rate auditory nerve fibres — the ones with high thresholds that encode suprathreshold and in-noise temporal information.[2013]
- Audiometrically silent — thresholds remain normal because high-spontaneous-rate fibres mediating threshold detection are relatively spared.[2014]
If correct, this explains a long-standing clinical puzzle: patients with normal audiograms who report difficulty understanding speech in noisy environments — the "hidden hearing loss" complaint that audiologists have heard for decades without an objective test to substantiate it.
TThe animal evidence
The animal literature is robust and consistent across species:
| Model | Finding | Reference |
|---|---|---|
| Mouse, noise-induced | Up to 50% of IHC ribbon synapses lost after 100 dB SPL × 2 h exposure that produced only temporary threshold shift; ABR wave I reduced 60% at 32 kHz, 30% at 12 kHz, despite recovered thresholds and intact hair cells. | Kujawa & Liberman 2009[2009] |
| Mouse, age-related | Steady decline in IHC synapse counts throughout life, reaching ~50% by older age, beginning before threshold elevation or OHC loss. | Sergeyenko 2013 |
| Guinea pig | Synapse loss after noise; some evidence of partial regeneration over time — a species difference that complicates direct extrapolation. | Reviewed by Encina-Llamas 2024 |
| Rhesus macaque | Synapse loss observed, but required higher noise levels than mouse models, and total synapse loss was lower. | Valero 2017[2017] |
| Single-unit recordings | Confirmed preferential damage to high-threshold, low- and medium-SR fibres. | Furman 2013[2013] |
The macaque data are particularly important for clinical extrapolation. Rhesus macaques required higher noise levels to produce equivalent synapse loss than mice did — suggesting that humans, with similarly long cochleae and even more robust efferent systems, may also be more resistant to noise-induced synaptopathy than rodent models predict.[2017] This does not refute the synaptopathy concept; it complicates the question of how much real-world noise exposure is required to produce clinically meaningful synapse loss in humans.
TSynapse-loss simulator
The schematic below illustrates the proposed mechanism. A normal inner hair cell carries roughly twelve ribbon synapses, divided between high-spontaneous-rate fibres (which mediate threshold detection) and low/medium-spontaneous-rate fibres (which encode suprathreshold and in-noise temporal information). Noise exposure and aging preferentially silence the low/medium-SR ribbons. The audiogram, supported by the surviving high-SR fibres, remains normal; the suprathreshold neural code degrades.
Slide from 0% to 100% to watch the low/medium-SR ribbons fade out across a four-cell IHC array — the “audiogram normal, suprathreshold neural code degraded” signature predicted by the animal model.[2016] The high-SR fibres (red) stay throughout, which is why pure-tone thresholds remain near-normal even at heavy synapse loss; the suprathreshold verdict that updates on the right is the predicted behavioural correlate, and the human evidence for whether it actually maps onto speech-in-noise scores is contested in the next section.
CThe human translation — supporting and contesting evidence
This is where the field becomes uncertain. The animal data are unambiguous; the human data are mixed. The honest summary, after roughly a decade of human work:
Evidence supporting human synaptopathy
- Liberman et al. 2016 — high-risk noise-exposed adults with normal audiograms showed larger SP/AP ratios than low-risk controls (driven by SP elevation), and poorer speech-in-noise scores at low presentation levels.[2016]
- Wu et al. 2019, 2021 — human temporal bone studies show age-related synapse loss with intact hair cells, and noise-exposed cochleae show primary neural degeneration correlating with reduced word-discrimination scores. This is the only direct human evidence.[2019, 2021]
- Stamper & Johnson 2015 — reduced ABR wave I amplitude as a function of recent noise exposure in normal-audiogram adults.[2015]
Evidence not supporting (or failing to replicate)
- Prendergast et al. 2017 — Manchester cohort: no significant ABR wave I, SP/AP, or EFR differences attributable to noise exposure in young adults with normal audiograms.[2017]
- Guest et al. 2017 — Tinnitus + noise exposure cohort: no electrophysiological evidence for cochlear synaptopathy despite the clinical picture suggesting it.[2017]
- Systematic review— Across 23 human studies covering 41 experiments, < 33% found a significant relationship between noise-exposure history and speech-in-noise performance, the behavioural correlate most often predicted by the hypothesis.
TProposed clinical metrics — and their problems
Several non-invasive measures have been proposed as functional biomarkers of synaptopathy. None has yet been validated for clinical diagnostic use:
| Measure | Rationale | Status |
|---|---|---|
| ECochG SP/AP ratio | Loss of AP fibres without SP change → ratio rises. | Liberman 2016 positive; mechanism for SP rise still unclear; replication mixed. |
| ABR wave I amplitude | Direct measure of synchronous distal nerve firing; lost synapses reduce it. | Stamper 2015 positive; Prendergast 2017 negative; high inter-individual variability limits sensitivity. |
| Wave V latency in noise | Predicted to slow with reduced synaptic drive in noisy conditions. | Mehraei 2016 positive in some cohorts; not widely replicated. |
| Envelope-following response (EFR) | Suprathreshold temporal coding selectively dependent on low/medium-SR fibres. | Promising mouse-to-human translation; carrier-frequency choice still being optimised for humans.[2014] |
| Middle-ear muscle reflex thresholds | Reflex arc depends on low-SR fibres; threshold elevation may indicate fibre loss. | Mixed results in human studies. |
| Speech-in-noise scores | Behavioural endpoint — the patient symptom synaptopathy is hypothesised to explain. | The behavioural correlate is intuitive but only weakly tied to specific electrophysiology. |
Even Liberman's 2016 paper, the most-cited human ECochG evidence for synaptopathy, leaves an awkward question unresolved: why does synapse loss elevate the SP? The conventional model has the SP generated by the same hair-cell and dendritic currents that contribute to the AP. If lost synapses simply remove fibres from the population that contributes to both potentials, both should fall together — the ratio should not rise. Liberman 2016 actually found that the SP itself was larger in the high-risk group, not just that the ratio rose.[2016] The current best explanation is that the Pappa 2019 / Hutson 2022 dissection of the SP into hair-cell and dendritic components leaves room for differential effects on each — the synapses lost are downstream of where the SP is mostly generated, but the dendritic component shift remains poorly characterised.[2019, 2022] This is not a fatal objection, but it is a reason for caution.
FClinical status — what to tell a patient
A reasonable practical stance for 2026:
- The condition is biologically real. Temporal bone studies in humans demonstrate cochlear synapse loss with age and noise exposure, with no other plausible mechanism.[2019, 2021]
- The clinical syndrome is harder to pin down. Speech-in-noise difficulty with a normal audiogram is common; how much of it is cochlear synaptopathy versus central auditory processing or attention is genuinely uncertain.
- No validated diagnostic test exists. ECochG SP/AP ratio and ABR wave I both show group-level signals in some studies, but neither has the sensitivity or specificity needed to apply confidently to an individual patient.
- Management is generally supportive. Hearing aids with appropriate frequency-specific gain and assistive listening technologies (FM systems, directional microphones, captioning) help substantially even without a definitive diagnosis.
- Hearing protection remains the single most defensible recommendation — synaptopathy from preventable noise exposure is, by definition, preventable. The therapeutic implications are stronger than the diagnostic ones.
FClinical case
Which is the most appropriate response and management plan?
TCSelf-assessment
In Kujawa & Liberman's 2009 mouse model of noise-induced cochlear synaptopathy, why did pure-tone thresholds return to normal despite permanent loss of 50% of IHC ribbon synapses?
A 2026 patient asks: “Can ECochG show whether I have hidden hearing loss?” The most honest answer is:
Which is the single most defensible clinical recommendation for a young adult with significant recreational noise exposure?