Neural Hearing Loss: Mechanisms, Diagnosis and Treatment Horizons

Abstract

Neural hearing loss, characterized by dysfunction of the auditory nerve, including the spiral ganglion neurons (SGNs) and/or their synaptic connections, is increasingly recognized as a critical contributor to auditory deficits across diverse conditions, including Auditory Neuropathy Spectrum Disorder (ANSD), presbycusis, and noise-induced hearing loss (NIHL). It is possible that neural hearing loss is underdiagnosed, due to the lack of clinical tools with sufficient sensitivity and specificity to detect poor neural health. Current interventions, such as hearing aids and cochlear implants (CIs), primarily target sensory deficits and offer limited benefit in cases of significant neural compromise. Therapeutically, there is a growing shift towards biologically driven strategies aimed at restoring neural function. Recent developments in novel therapies, including pharmacological, gene-based, neurotrophic, and cell-based approaches, have opened new possibilities demonstrating the potential to protect, repair, and/or replace damaged SGNs, and re-establish auditory pathways. This perspectives article explores the evolving understanding of neural hearing loss, emphasizing its complex pathophysiology and the limitations of current diagnostic and therapeutic approaches, while highlighting how a diverse range of emerging solutions are moving closer to clinical application.

Introduction

Hearing loss is the most widespread sensory deficit, affecting over 1.5 billion people globally, with its prevalence rising sharply in ageing populations [1]. More than two-thirds of adults over the age of 60 are affected, and it is estimated that by 2050, one in four people worldwide will experience hearing loss [1]. This trend represents a significant burden on global public health systems and economies, with untreated hearing loss linked to social isolation [2], cognitive decline [3], and reduced workforce participation [4], contributing to an estimated $980 billion in annual global costs. Developing effective interventions is therefore both a clinical and socioeconomic priority.

Sensorineural hearing loss (SNHL) accounts for the vast majority of hearing loss cases and encompasses a broad range of auditory dysfunctions, ranging from damage to cochlear hair cells to impairments affecting the cochlear nerve. However, clinical assessment and treatments have historically focused almost exclusively on sensory deficits, particularly those caused by hair cell loss, while largely overlooking the contribution of neural pathology..

Neural hearing loss involves dysfunction of the auditory nerve, including the spiral ganglion neurons (SGNs) and/or their synaptic connections, and is increasingly recognized as a key contributor to hearing difficulties in conditions such as Auditory Neuropathy Spectrum Disorder (ANSD), age-related hearing loss (presbycusis), and noise-induced hearing loss (NIHL). Unlike sensory loss, neural hearing loss primarily affects processing and transmission of auditory information, rather than the initial detection of sound. Individuals often struggle with speech perception, particularly in background noise, even when pure-tone thresholds appear normal. The reliance on standard diagnostic tests such as pure-tone audiometry (PTA) and speech perception in quiet can therefore miss these deficits, leading to a disconnect between clinical findings and patient-reported difficulties [5, 6].

The historical emphasis on diagnosing and treating the sensory component of SNHL has been shaped by the long-held assumption that the degeneration of auditory neurons is a secondary consequence of hair cell loss [7–9]. However, human temporal bone studies have shown that primary neural degeneration can occur independently of hair cell damage [10, 11]. Despite growing recognition of its significance in the scientific literature, neural hearing loss may often go undetected and poorly addressed in clinical practice.

Current interventions for SNHL, including hearing aids and cochlear implants (CIs), are designed to compensate for the loss of hair cells. While effective for many individuals with sensory deficits, they offer limited benefit for those with significant neural degeneration. Indeed, compared with other recipients, outcomes can be worse and adoption rates lower among older adults [12, 13] and individuals with auditory neuropathy [14], suggesting that existing technologies often fail to meet the needs of patients. Currently, no clinically approved therapies directly address neural hearing loss, resulting in an unmet need in the management of hearing impairment attributable to neural pathology.

Recognizing neural pathology as a core component of SNHL is therefore critical for meeting the needs of this underserved patient population. A shift in perspective is required, not only for improving diagnostic accuracy and clinical decision-making, but also for driving innovation in the development of targeted interventions that more effectively treat both sensory and neural elements of SNHL.

This paper presents an overview of the evolving understanding of neural hearing loss, including its mechanisms, clinical presentation, and current limitations in assessment and treatment, and explores a broad range of emerging therapeutic strategies that are advancing towards clinical application.

Mechanisms of Neural Hearing Loss

Overview of Anatomy and Physiology

The cochlea contains a highly organized and specialized network of auditory neurons that transmit auditory signals from the peripheral auditory system to the brain (Fig. 1). SGNs are bipolar cells that extend a peripheral process (dendrite) to the organ of Corti and a central axon, through the auditory nerve, into the brainstem cochlear nucleus. The human auditory nerve contains approximately 40,000 nerve fibres, the majority (~95 %) of which are myelinated and originate from type I SGNs. These are afferent fibres, responsible for carrying sensory information from the cochlea to the brain. They form monosynaptic connections with inner hair cells (IHCs), the primary sensory receptors of the cochlea. Although each type I SGN innervates only a single IHC, each IHC forms synaptic connections with multiple SGNs, enabling robust and redundant encoding of auditory information across a range of sound intensities and frequencies [15].

Fig. 1.

Illustration of the innervation of the cochlea. Yellow: afferent fibres; Green: efferent fibres; Blue: IHCs; Orange: OHCs. SGNs Spiral ganglion neurons, IHCs Inner hair cells, OHCs outer hair cells

Auditory signal transmission at the synapses between the IHCs and peripheral terminals of the SGNs is mediated by the release of the neurotransmitter glutamate. This release is triggered by sound-induced vibrations of the basilar membrane, which deflect the stereocilia on the apical surface of IHCs. This mechanical deflection opens mechanoelectrical transduction channels, leading to hair cell depolarisation and subsequent neurotransmitter release. The released glutamate activates postsynaptic receptors on the SGNs, initiating action potentials that propagate along the auditory nerve to the brainstem, where further processing occurs in the cochlear nuclei. Type I SGNs are primarily responsible for encoding temporal fine structure (TFS) and spectral features of acoustic signals, forming the basis for complex auditory functions such as speech perception and sound localisation [15].

Type I SGNs exhibit heterogeneity in gene expression and function. Traditionally, they are classified into three functional subtypes, low, medium, and high spontaneous rate neurons, each linked to distinct response properties, anatomical characteristics, and molecular markers [15–17]. Recent single-cell RNA sequencing studies have identified three molecular subtypes (type IA, IB, and IC) in mice; however, the exact correspondence between these molecular profiles and physiological subtypes remains under investigation [18–21]. Moreover, strict stratification is not always straightforward, and interspecies differences must be considered [18, 20, 22, 23]. Despite these uncertainties, some differential characteristics of type I subtypes observed in non-human species could be relevant to humans, including differences in apical to basal distribution along the cochlea and vulnerability to degenerative factors such as noise, ototoxic drugs, and ageing (with low spontaneous rate neurons shown to be more vulnerable) [e.g. 24–27].

The remaining ~5 % of auditory nerve fibres are unmyelinated type II neurons, which also serve an afferent role but form divergent, polysynaptic connections with multiple outer hair cells (OHCs). The precise role of type II neurons is not yet fully understood, but they are thought to contribute to cochlear protection mechanisms and the detection of high-intensity sound stimuli [15]. In addition to this afferent innervation, the cochlea receives efferent input from the brainstem via the olivocochlear system. These efferent fibres modulate cochlear sensitivity and protect against acoustic overstimulation, primarily by acting on OHCs and, to a lesser extent, on the afferent synapses of IHCs [15].

Pathophysiology

The pathophysiology of neural hearing loss is complex and heterogeneous but at its most fundamental level, it reflects the degeneration or dysfunction of the auditory nerve. Auditory neural dysfunction can involve damage to any compartment of the SGN, including the cell body, central axon, and peripheral processes (dendrites) including the synaptic terminal (Fig. 2), each of which plays a critical role in the encoding and transmission of auditory information [28–32]. Based on location of damage, these pathologies can be categorized as postsynaptic (referring to the SGN-IHC synapse), whereas those involving the SGN synaptic terminal may also be grouped with other non-neural synaptic pathologies and classified as synaptic (see further discussion on synaptopathies below). Examples of abnormalities include loss of synaptic connections, disruption of ion channel function, shrinkage of cell bodies, demyelination of central axons, or complete neuronal loss [e.g. 10, 11, 33, 34, 35]. The consequences of neural degeneration include reduced conduction velocity, impaired spike timing, and desynchronised neural firing [36].

Fig. 2.

Illustration of pathologies categorized based on location of damage. Postsynaptic pathologies of the SGNs can involve deficits in any compartment of the neuron (blue background). Synaptopathies can involve the postsynaptic SGNs terminal and/or presynaptic IHC ribbon. SGN Spiral ganglion neuron, IHC Inner hair cell

The mechanisms underlying neural degeneration can also be categorized based on the origin and progression of damage. Neural degeneration can result from pathologies directly affecting the neurons (primary degeneration), or from pathologies affecting sensory input or other elements of the neurons’ environment, such as supporting cells and the vascular system (secondary degeneration) [37]. Retrograde degeneration is a commonly described pattern of spiral ganglion degeneration whereby deficits in the peripheral processes precede those of the cell body and central axon [38]. However, simultaneous degeneration of different neuron compartments is also possible [39], and some neurons may survive even after losing their peripheral processes [40].

Damage to the synaptic terminal between IHCs and auditory nerve fibres can lead to cochlear synaptopathy. Synaptopathies can originate within the IHC and/or at the peripheral axon or synaptic terminal of the neuron (Fig. 2), highlighting the continuum between sensory and neural pathologies, particularly in age and noise-related hearing impairment [30, 41–43]. Distinguishing the precise origin of synaptopathy, whether primarily sensory, neural, or a combination, would have important implications for the selection and development of targeted interventions. Hidden hearing loss (HHL) is a conceptually overlapping term with both cochlear synaptopathy and neural hearing loss [44]. It is often used to refer to conditions whereby individuals experience hearing difficulties despite normal audiometric thresholds [30, 44]. HHL may be caused by deficits in IHCs, SGNs, or synapses between these. Given the overlap in terminology and underlying concepts, findings from studies investigating both synaptopathies and HHL are often relevant to understanding neural hearing loss. Although an in-depth review of these conditions is out of scope for this paper, many publications focusing on various aspects including pathophysiology, assessment methods, functional implications, and treatment options are available [14, 30, 45–48]. These provide comprehensive syntheses of current knowledge and highlight ongoing debates and gaps in the fields of synaptopathies and HHL for interested readers.

At the molecular and cellular level, several pathophysiological mechanisms contribute to SGN damage, including oxidative stress, mitochondrial dysfunction, excitotoxicity, DNA damage, abnormal protein accumulation, chronic inflammation, and dysregulation of neurotrophins [23, 49–52]. Oxidative stress is a significant contributor in several conditions and involves the accumulation of reactive oxygen species (ROS) that damage DNA, proteins, and lipids. This process can be triggered by ageing, noise trauma, and exposure to ototoxic agents [49]. Mitochondria are especially vulnerable to ROS-induced damage, and mitochondrial dysfunction accelerates cochlear ageing and neural degeneration [50]. Persistent oxidative stress also activates inflammatory pathways, leading to cytokine release and neuronal apoptosis [51].

These cellular-level changes, impacting individual neurons, have a cumulative impact along the length of the cochlea, forming what is increasingly referred to as a neural health profile, a spatially and functionally variable representation of auditory nerve integrity [53]. This profile reflects both the distribution and severity of degeneration across the SGN population. Although measuring neural health remains challenging, and often indirect (see section on 'Diagnosing neural hearing loss'), growing evidence from various study designs demonstrates that it can vary significantly along the cochlear spiral, between the two ears, and across different individuals [38, 54]. Some regions may retain relatively intact SGN populations with preserved function, while others may exhibit localized degeneration or complete neuronal loss, resulting in so-called “dead regions” where no neural activity can be elicited despite preserved hair cell function [53, 55]. The term poor neural health may be used to describe different patterns of deficits, including generalized reduction in auditory nerve function across the cochlea, localized deficits confined to specific cochlear regions, such as the commonly observed increased loss of SGNs towards the basal end of the cochlea [11, 38, 56, 57], or combinations of pathologies [38, 58]. In all cases, it reflects the cumulative impact of SGN degeneration but is most relevant for profiles where the extent and distribution of dysfunction result in significant perceptual or clinical consequences. Neural hearing loss and poor neural health are closely intertwined, shaped by a complex interplay of genetic, environmental, and age-related factors.

Our understanding of the pathophysiological processes underlying neural hearing loss continues to evolve. For example, auditory nerve degeneration was, until relatively recently, considered mainly a secondary consequence of cochlear hair cell loss, particularly in conditions such as presbycusis and NIHL. However, accumulating evidence from human temporal bone studies and animal models has challenged this view, demonstrating that SGN degeneration can occur independently of inner hair cell pathology and may both precede and exceed it [10, 11]. Makary et al. (2011) showed that SGN counts decline by around 1000 cells per decade, resulting in up to 30 % loss by age 90 in individuals with normal populations of hair cells [59]. Further evidence from temporal bone studies in older adults suggests that there is an even greater percentage of SGNs with a loss of peripheral axons, likely exceeding 60 % over the age of sixty [11, 60]. Similarly, progressive SGN degeneration can occur in the absence of IHC loss following significant noise exposure due to the loss of synapses at the IHC-SGN junction [48, 61].

In addition to age-related degeneration and noise exposure, primary neural degeneration can occur as the result of a diverse range of aetiologies including genetic factors, congenital anomalies, ototoxicity, infections, neoplasms, nutritional deficiencies, trauma such as temporal bone fractures or following cochlear implantation, immune-mediated processes, metabolic disorders, and idiopathic conditions [31, 49, 57, 62, 63]. These diverse causes reflect the vulnerability of SGNs to both systemic and localized insults.

Genetic factors are increasingly recognized as playing a significant role in auditory neural dysfunction. Studies of ANSD, a heterogeneous group of hearing disorders characterized by impaired IHC and/or SGN function with preserved OHC activity, have identified several genetic mutations affecting synaptic transmission and neuronal integrity [64]. For example, mutations in OTOF1, which encodes otoferlin, a calcium sensor essential for neurotransmitter release, are associated with pre-synaptic synaptopathy. Similarly, postsynaptic auditory neuropathy, the most relevant to this review, has been linked to mutations in genes such as DIAPH3, OPA1, GJB2, MPZ, PMP22, AIFM1, and PJVK, which affect dendritic structure, axonal integrity, and oxidative stress responses. Central transmission deficits, potentially associated with NARS2 mutations, may further impair auditory processing. These genetic insights highlight the diverse molecular pathways that can disrupt auditory nerve function, independent of hair cell pathology [14, 65].

Challenges in Understanding Mechanisms in Humans

While sensory deficits have traditionally dominated the clinical and research focus in hearing disorders, it is increasingly evident that neural hearing loss represents a distinct and critical pathology. Despite growing recognition of its importance, neural hearing loss, and the pathophysiological processes that underpin it, remains challenging to study in humans due to the limited accessibility of cochlear neural structures and the limitations of current diagnostic tools. As a result, much of our understanding is derived from animal models, human temporal bone studies, and computational simulations [e.g. 10, 28, 35, 59, 66, 67]. Animal studies have contributed significantly to the evidence, with a diverse range of methodological approaches used; however, they are insufficient for full translation to human applications [see discussion in 68]. Beyond ethical considerations limiting their use, there are also significant interspecies differences which complicate extrapolation of findings. These differences include variations in SGN anatomy, physiology, and pathophysiology, such as cell body and process size, myelination, electrical response properties, and vulnerability to degenerative factors [see discussions and evidence from 16, 20, 30, 34, 40, 66, 68, 69, 70].

Human temporal bone studies provide direct insights into cochlear pathology, and procedural advances, such as machine-learning-based quantification and three-dimensional (3D) mapping as demonstrated by Wu et al. (2023), can enhance their precision and usability. Despite these improvements, such studies remain constrained by small sample sizes, reliance on post-mortem tissue, and the inability to capture dynamic processes in living systems.

Computational models complement these efforts by allowing scientists to simulate and analyse complex biological processes, such as auditory nerve function and degeneration, using mathematical frameworks and computer algorithms [e.g. 28, 67]. These models can rapidly test a wide range of variables, predict outcomes, and generate hypotheses that would be difficult, time-consuming, and costly to investigate in living systems. For example, they can help researchers understand how specific neural pathologies might affect an individual neuron’s function, or overall hearing performance. However, it is important to recognize that the accuracy of computational models depends on the quality of the data and assumptions used to build them. While these are powerful tools for hypothesis generation and exploration, they cannot replace direct experimental evidence and must be validated against real-world observations. Ultimately, bridging findings from different methodologies remains essential for advancing our understanding of neural hearing loss pathophysiology and predicting treatment response mechanisms.

Clinical Presentation and Diagnostic Challenges

Functional Implications of Neural Hearing Loss

The functional consequences of neural hearing loss are variable and depend on the underlying neural health profile. Neural hearing loss can coexist with other auditory pathologies, including cochlear and central auditory dysfunctions, making it difficult to attribute specific perceptual deficits to peripheral neural damage alone. Moreover, associating functional outcomes with specific pathophysiological mechanisms (such as synaptopathy, demyelination, or axonal degeneration) is challenging due to the redundancy and plasticity of the auditory system. Despite these diagnostic challenges, several functional deficits that may not be detected by standard clinical tests have been associated with neural hearing loss.

Performance in simple auditory tasks might not be affected in many cases of neural hearing loss. For example, while pure tone threshold elevation may occur with extensive SGN loss, many individuals exhibit normal thresholds despite significant neural damage [30, 71, 72]. This is partly due to the dense innervation of the IHCs by multiple SGNs, providing a level of redundancy in the auditory system. In early stages of neural degeneration, surviving neurons can compensate for the loss, preserving threshold sensitivity and masking the underlying neural deficit. More complex auditory tasks, such as word recognition in quiet, place greater demands on the integrity of neural coding than pure tone detection yet still tolerate substantial neural loss. For instance, it has been suggested that over 60 % auditory nerve fibre loss is required before word recognition scores in quiet fall below 90 % [73].

Neural hearing loss is often best understood as a disruption in complex auditory processing, including deficits in temporal coding at the neural level, such as impaired phase-locking to TFS, envelope modulation sensitivity, and disrupted spectro-temporal integration. These encoding deficits can degrade performance in complex listening situations, including speech-in-noise perception (particularly with competing talkers), sound localisation (from degraded binaural timing cues), and music perception (from diminished pitch discrimination) [73].

One proposed mechanism underlying suprathreshold auditory deficits involves the differential vulnerability of different auditory nerve fibre types, with low and medium spontaneous rate fibres thought to be more vulnerable [24–27, 70]. Low and medium spontaneous rate fibres, which have higher response thresholds and broader dynamic ranges, are believed to play a crucial role in encoding sounds in complex acoustic environments [74]. These fibres remain responsive at higher sound levels, where high spontaneous rate fibres may already be saturated [48], making them particularly important for speech in noise processing. While these peripheral mechanisms are crucial, they constitute only a portion of the broader processing deficit associated with neural hearing loss.

Carney (2018) proposes that the representation of suprathreshold sounds relies on the contrast between neural responses across different frequency channels, which serves as an important coding strategy, especially in noisy environments [75]. This contrast is enhanced by central auditory structures such as the cochlear nucleus and inferior colliculus, and by efferent feedback pathways, which may themselves depend on the integrity of low and medium spontaneous rate fibres [17, 75]. In addition, the auditory system may engage in reactive plasticity and compensatory responses to peripheral damage. These include synaptic regeneration and central gain enhancement, both of which can alter functional consequences of neural hearing loss [45, 76]. Taken together, these findings suggest that temporal coding deficits in neural hearing loss are likely to be shaped by complex peripheral and central mechanisms.

Although evidence remains limited and sometimes contradictory, neural hearing loss may also be associated with other perceptual disorders, such as tinnitus and hyperacusis. Both conditions are thought to associate, at least in part, with central hyperexcitability or maladaptive gain control mechanisms in response to reduced afferent input. This reflects the brain’s attempt to compensate for diminished neural signals by amplifying spontaneous activity or increasing sensitivity in central auditory pathways. Such mechanisms may be shared with neural hearing loss, where degeneration of SGNs or synaptic connections leads to reduced peripheral input, potentially triggering similar compensatory changes [77, 78].

Lastly, as our understanding of neural hearing loss deepens, evidence suggests that the degeneration of SGNs may underlie increased listening effort, even in individuals with normal hearing thresholds. Using a cross-species design, Zink et al. (2025) found parallel age-related decreases in neural activity generated by the peripheral auditory system in response to sound in humans and in gerbils with confirmed SGN loss [79]. Middle-aged adults also showed increased pupil-indexed listening effort during speech-in-noise tasks, even when behavioural performance was maintained. These findings suggest that neural degeneration contributes to hidden hearing difficulties and elevated cognitive load in midlife [79].

Diagnosing Neural Hearing Loss

Despite its contribution to a range of perceptual difficulties, neural hearing loss could potentially be undiagnosed in a larger group of patients than currently believed. The identification of neural hearing loss is an important challenge as its effects can be underestimated, largely due to the widespread reliance on PTA, the gold standard clinical assessment for hearing loss. While PTA is effective for detecting changes in hearing sensitivity, it provides limited insight into the integrity of auditory neural pathways and lacks both the specificity for distinguishing between sensory, neural and central lesions and the sensitivity for detecting deficits in more complex listening tasks [80]. Clinicians frequently encounter patients whose subjective hearing difficulties, particularly in complex auditory environments, are not aligned with their audiometric profiles [5, 6, 30]. This mismatch underscores the inadequacy of relying solely on threshold-based measures to assess auditory function. Although patients with this clinical presentation are frequently described as having HHL, the term encompasses a wide range of distinct pathophysiological processes, limiting its usefulness for precise clinical classification or treatment selection.

Other standard clinical behavioural assessments, such as the Threshold Equalizing Noise (TEN) test, can aid in identifying cochlear dead regions [81]. By masking specific frequencies with calibrated noise, the TEN test helps determine whether a tone is being detected via off-frequency listening, which is indicative of a dead region. However, it is less suitable for individuals with severe-to-profound hearing loss, as the high levels of masking noise required can be uncomfortable and may lead to unreliable results due to reduced dynamic range and poor residual hearing. Moreover, the TEN test does not provide direct information about neural synchrony or auditory nerve integrity and cannot determine the relevant contribution of loss of IHCs or SGNs to the dead region [82].

In contrast, standard clinical objective assessments such as the Auditory Brainstem Response (ABR) offer direct evaluation of neural function. Whilst ABR thresholds seem to be generally insensitive to neural hearing loss, suprathreshold measures such as the Wave I amplitude have emerged as more promising indicators, especially when audiometric thresholds appear normal. Animal studies demonstrate persistent reductions in Wave I amplitude following noise-induced neuropathy, despite recovery of ABR thresholds, suggesting sensitivity to synaptic or neural damage [24]. In humans, reduced Wave I amplitude has also been linked to NIHL [83]. However, its clinical utility is limited by technical challenges, including the need for high stimulus levels, difficulty identifying Wave I in adults, high inter-subject variability, and lack of standardized protocols [42]. Moreover, ABR measures may underestimate neural deficits due to non-linear relationships between neural survival and response amplitude. Chambers et al. (2016) demonstrated that ABR Wave I amplitude and behavioural tone detection in mice were logarithmically related to the number of surviving auditory nerve fibres, with central compensatory mechanisms potentially masking early neural loss [76].

There is growing recognition that behavioural psychophysical tests targeting temporal processing, which are not yet widely used in clinical assessments, are more sensitive to neural hearing deficits. These deficits stem from disruptions in the temporal fidelity of auditory nerve signalling and are not reflected in standard audiometric thresholds, underscoring the need for more sensitive diagnostic tools. Temporal deficits can be assessed using a variety of methods. For example, challenging speech perception tasks, such as speech-in-noise, depend on auditory mechanisms that require precise neural timing and synchrony, both of which are impaired in neural hearing loss [84]. These temporal processing impairments have real-world consequences, particularly in complex auditory environments. Difficulty understanding speech in noise is one of the most common complaints among individuals with hearing loss, and temporal deficits are a key contributor to this challenge. Despite the diagnostic value of speech-in-noise testing, it remains underutilized in clinical practice. Factors contributing to its limited use include lack of time, training, standardized protocols [85], and a lack of perceived clinical utility due to current hearing technologies being less effective in noisy environments [86]. It is, however, important to recognize that speech testing outcomes can be influenced by various factors, including cognitive load, linguistic complexity, and attentional demands. Therefore, a multimodal approach that combines speech-in-noise tests with other temporal processing measures could offer a more comprehensive evaluation of neural function.

Temporal processing measures less affected by central compensatory mechanisms can be particularly useful in identifying peripheral neural dysfunction. Such measures include temporal modulation detection and binaural processing tests, which assess the auditory system’s ability to encode timing and spatial cues, functions that are particularly vulnerable to neural degradation [76, 87]. Complementary electrophysiological measures, such as the envelope-following response (EFR), frequency-following response (FFR), acoustic reflex, and compound action potential (CAP), can provide objective insights into auditory nerve integrity and neural synchrony [e.g. 45, 87, 88, 89, 90].

Furthermore, advances in neuroimaging, especially high-resolution magnetic resonance imaging (MRI) using specialized protocols, offer the potential to directly visualize structural abnormalities in the auditory nerve and cochlear pathways [29, 91]. For example, a recent study proposed assessing neural health with diffusion-weighted MRI by calculating a metric of the density of the audiovestibular nerve fibres [91]. This study showed that participants with auditory neuropathy (having absent ABRs but either normal otoacoustic emissions or cochlear microphonics) had significantly lower fibre density than participants with normal hearing or hearing loss not classified as auditory neuropathy. Similarly, with high-resolution T2-weighted structural MRI, Harris et al. (2021) showed that the density of the audiovestibular nerve was lower for older adults and associated with a metric of neural synchrony. Such studies demonstrate that although challenging, advances in acquisition and analysis of neural images have the potential to detect changes in neural health.

Despite advances in diagnosing neural hearing loss, and its increasing recognition as a distinct and critical pathology, efforts to understand its true burden remain limited. The lack of established, standardized methods for diagnosis and phenotyping limits our ability to accurately determine the exact prevalence of neural hearing loss. Dedicated research towards developing new methods and standardizing existing methods is therefore paramount.

Implications for Intervention

Limitations of Existing Devices

The lack of understanding and potential underdiagnosis of neural hearing loss has had significant implications on the development of effective treatment options. Traditional hearing devices, such as hearing aids and CIs, are designed to address sensory impairments arising from hair cell loss that are typically defined and measured by clinical assessments focused on sound detection.

Hearing aids, for example, can enhance audibility but do not restore the temporal precision or neural synchrony required for accurate speech perception, especially in noisy or complex listening environments. For some, amplification may even exacerbate listening difficulties by increasing the intensity of background noise without improving the clarity of speech clarity [92]. These limitations in challenging acoustic settings are a common source of dissatisfaction among hearing aid users, who often report that their devices fall short of expectations in real-world environments despite technological advancements [92, 93].

CIs, on the other hand, bypass damaged hair cells and directly stimulate the auditory nerve. However, their effectiveness depends on the survival and functional integrity of the SGNs [94]. In instances of severe auditory nerve fibre loss or dysfunction, such as for postsynaptic ANSD and severe-to-profound presbycusis, the benefits of CIs can be limited, particularly in challenging listening situations [14, 55, 65, 95]. Moreover, CIs do not address the temporal coding deficits central to many forms of neural hearing loss. Deactivating specific CI electrodes has been proposed as a strategy to mitigate the effects of neural hearing loss by reducing stimulation through electrodes that are poorly matched to surviving neurons. However, beyond failing to address the underlying neural degeneration, inappropriate channel deactivation can further impair auditory processing [55, 96].

Increasingly, the impact of neural health on CI outcomes is being recognized, with mounting evidence suggesting that refined and standardized assessments of neural integrity could enhance clinical decision making and device performance [e.g. 14, 28, 55, 88, 97, 98, 99]. While some findings remain inconclusive or contradictory, often due to methodological variability [65, 94, 100], there is growing optimism that advancements in neural health measurement will improve clinical outcomes by enabling clinicians to recommend the most effective interventions and deliver more personalized rehabilitation.

In cases where the cochlear nerve is entirely absent, such as cochlear nerve aplasia, or in patients with neurofibromatosis type 2 (NF2) or certain other cochlear anomalies, an Auditory Brainstem Implant (ABI) may be offered [101]. These devices bypass the cochlear and auditory nerve entirely by placing the stimulatory electrode array adjacent to the cochlear nucleus in the brainstem. Although ABIs provide a critical option for patients with no viable cochlear nerve, the procedure is invasive, technically challenging and outcomes are variable [102]. Indeed, ABIs can cause undesired non-auditory side effects due to stimulation of adjacent neural structures [103].

These limitations across current technologies highlight the urgent need for a paradigm shift in both diagnosis and treatment of neural hearing loss. Until such advancements become clinically available, many individuals with neural hearing loss will continue to face significant communication challenges, particularly in noisy environments, despite having access to existing hearing technologies.

Using CIs to Explore Neural Health

Despite their limitations, CIs provide a unique opportunity to advance our understanding of neural health. Electrophysiological assessments leveraging intracochlear telemetric monitoring, are emerging as valuable tools for directly measuring auditory nerve function. Standard electrically evoked compound action potentials (eCAPs), which are routinely measured in clinical settings, can provide valuable insights into the underlying neural status of CI users [104]. Key eCAP-derived measures include response latency and threshold, as well as amplitude-related variables such as the maximum amplitude at the highest current level, the slope of the amplitude growth function (AGF), and the failure index [58, 66, 105]. In addition, more advanced paradigms such as panoramic eCAPs (PECAPs) [106] and protocols that manipulate stimulus parameters, including interphase gap, phase duration, polarity, and inter-pulse intervals, offer the potential to bypass pre-neural factors and yield more robust estimates of neural health [e.g. 28, 56, 66, 107].

In addition to objective responses, assessing behavioural responses to electrical stimuli can also provide valuable information on the underlying status across the auditory pathway. Changes in standard behavioural assessments such as threshold and comfort levels and corresponding dynamic ranges may indicate changes in the condition of the auditory nerve. More sophisticated assessments, such as the polarity effects test can provide an even more reliable indicator of neural health [108, 109].

These capabilities make CI candidates, especially those with presbycusis and postsynaptic ANSD, an ideal initial target population for advanced therapies aimed at restoring auditory nerve function, as CIs would allow for real-time monitoring of therapeutic effects and neural integration. Moreover, the existing surgical infrastructure and clinical follow-up pathways associated with CI care provide a practical framework for evaluating the safety and efficacy of regenerative interventions in a controlled and well-characterized setting.

Emerging Therapies for Neural Hearing Loss

To effectively address the needs of individuals with neural hearing loss, there needs to be a shift towards approaches that aim to restore auditory neuron function, rather than solely compensating for sensory deficits. This section provides an overview of the emerging therapeutic strategies most relevant to neural hearing loss that target neural protection, synaptic repair and regeneration, highlighting how these are advancing towards clinical application. Neural protection aims to preserve the survival and function of existing neurons, synaptic repair focuses on restoring or enhancing the connections between hair cells and auditory neurons, and SGN regeneration seeks to replace lost or damaged neurons (Fig. 3). Table 1 provides a comparative overview of the main emerging therapeutic strategies for neural hearing loss discussed in this review, highlighting their mechanisms, clinical stage, advantages, and limitations.

Fig. 3.

Upper part (grey background): Examples of two main types of pathologies leading to neural hearing loss i.e. synaptopathies and complete neural loss. Lower part (purple background): Evolving therapeutic approaches including neuroprotection, synapse repair, and SGN regeneration. Yellow: healthy neurons; White: damaged or degenerated neurons/synaptic terminals; Purple: restored or regenerated neurons/synaptic terminals. SGN Spiral ganglion neuron

Table 1.

Overview of emerging therapeutic strategies for neural hearing loss

Therapy/Approach	Mechanism/Target	Clinical Stage	Advantages	Limitations/Challenges
Neurotrophins	Neuroprotection, neurite outgrowth, synaptic repair	Preclinical, early clinical (e.g., OTO-413)	Well-studied, multiple delivery methods, neuroprotective and synaptogenic effects	No SGN regeneration, delivery challenges, limited clinical efficacy so far
Small molecules	Modulation of signalling pathways (e.g., ROCK inhibitors, TrkB/TrkC agonists)	Preclinical, some clinical (e.g., sodium thiosulfate for otoprotection)	Easy delivery, potential for local administration	Limited direct evidence for SGN regeneration, off-target effects
Gene-based approaches	Sustained neurotrophin expression, synaptic repair, correction of genetic defects, direct reprogramming	Preclinical, early clinical (e.g., AAV, electrotransfer)	Long-term effect, cell-type specificity possible	Immune response, delivery, regulatory hurdles
Cell-based therapies	Replacement of lost SGNs using pluripotent stem cell-derived neural progenitors	Early clinical (e.g., Rincell-1 Phase I/IIa)	Potential for true regeneration, functional restoration	Safety, integration, functional heterogeneity, immune response
Direct reprogramming	Conversion of glia/fibroblasts to SGN-like neurons in situ	Preclinical (in vitro, animal models)	Avoids transplantation, uses endogenous cells, potential for in situ repair	Early stage, specificity, functional integration, scalability
Drug delivery systems	Sustained, targeted delivery (e.g., hydrogels, nanoparticles, drug-eluting electrodes)	Preclinical, some clinical (e.g., OTO-413)	Improved local delivery, reduced systemic exposure	Technical complexity, translation to clinic

Notably, several methodologies can be applied across therapeutic strategies. For example, neurotrophins have demonstrated potential both for neuroprotection and for promoting synaptic repair, while gene therapy and small molecule drugs may contribute to both neural preservation and regeneration depending on their target and delivery. This overlap highlights the versatility of emerging approaches and the importance of tailoring interventions to the specific underlying pathology in neural hearing loss.

Neuroprotection and Synaptic Repair

Neurotrophins

Neurotrophins have been extensively studied for over two decades as a means to enhance SGN survival. Neurotrophic compounds are promising for protecting neurons from degeneration and stimulating neurite outgrowth and synaptic restoration, but they cannot trigger the regeneration of new SGNs once they have been lost [110].

Attention has focused particularly on neurotrophins belonging to the nerve growth factor (NGF) family, primarily brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT-3). These are known to play major roles during the normal development of the cochlea, supporting differentiation, and in the adult cochlea, supporting the maintenance and survival of SGNs [111–114]. Neurotrophic support to SGNs is provided mainly by the hair cells and supporting cells of the organ of Corti [112, 115, 116].

Loss of this neurotrophic support after deafness leads to the gradual degeneration of SGNs through apoptotic cell death [112, 117]. NT-3 appears to regulate ribbon synapse density and can induce synaptic regeneration after excitotoxic damage in vitro [118] and after acoustic trauma in an animal model [119]. Moreover, cochlear overexpression of NT-3 by supporting cells, when induced at middle age, increases ribbon synapse density in mice and ABR Wave I amplitude, demonstrating its neuroprotective effect against ageing [120]. Administration of neurotrophins to the cell culture medium is necessary for the survival of SGNs in vitro, with type II SGNs being especially dependent on supplementation by BDNF [121].

Beyond these, a novel family of neurotrophins, structurally unrelated to the NGF superfamily, may offer a promising avenue for promoting SGN regeneration. The protein cometin is expressed in the developing otic vesicle in the mouse embryo but is absent in the adult cochlea [122], suggesting a role during cochlear development. Remarkably, an infusion of cometin protein via an osmotic pump into the inner ear of neomycin-treated adult guinea pigs significantly protected them against neural degeneration [122]. The surviving neurons appeared morphologically normal and functionally active, as evidenced by stable eABR thresholds in the cometin-treated animals, compared to the gradual threshold increase observed in controls over the two weeks of infusion. The neuroprotective effects of the infusion lasted for a further two weeks after withdrawal of the pump, suggesting that cometin may be useful for long-term SGN protection. Similar neuroprotective and neurite-outgrowth promoting activities have been shown in vitro with neonatal rat SGN cultures, raising the possibility that cometin could be worthy of further investigation as a tool for protecting SGN health [123, 124].

The clinical translation of neurotrophin-based therapies is limited by challenges in sustained and targeted delivery. Numerous studies have demonstrated that exogenous neurotrophins delivered directly to the cochlea by an osmotic pump over several weeks can protect SGNs and promote improved neuronal survival after deafness caused by various insults [123, 124], and also in neonatally deafened cats [125]. Neurotrophic effects have also been reported with other growth factors such as glial-cell-line-derived neurotrophic factor (GDNF) [126–129] and fibroblast growth factor (FGF) [130]. However, sustained delivery of neurotrophins appears to be key for maintaining their protective effect. A study in guinea pigs reported accelerated degeneration of SGNs after cessation of BDNF administration [131], while another report showed maintenance of improved neural survival for only two weeks after treatment with neurotrophins was terminated [132].

In animal models, long-term delivery of neurotrophins has been implemented using osmotic pumps with a reservoir, but this is an unfeasible solution for clinical application. Cochlear implants and slow-release gel formulations offer promising opportunities to overcome the challenge of sustained neurotrophin delivery, with CIs providing not only an effective route for controlled administration but also the potential added therapeutic benefit of electrical stimulation, which has been shown to help maintain treatment efficacy. Following the cessation of BDNF delivery in guinea pigs, electrical stimulation improved SGN survival for up to six weeks, however neurotrophic effects were observed only in the basal cochlea near the stimulating electrode, and accelerated neural degeneration was seen in the upper cochlear turns [133]. Combined use with a CI appears to be indeed beneficial; when BDNF infusion was administered together with a CI, a highly significant improvement in SGN survival (more than a 50 % increase compared to the contralateral ear) was maintained when electrical stimulation from the CI continued for 3–4 months after the termination of BDNF delivery [134].

Modifying the CI to incorporate a drug-eluting electrode array offers the dual benefits of sustained therapeutic delivery and the neuroprotective effects of electrical stimulation. Richardson et al. (2009) explored CI electrodes coated with polypyrrole/para-toluene sulfonate containing NT-3 (Ppy/pTS/NT3) [135]. These electrode arrays stored 2 ng of NT-3 and released 0.1 ng/day with electrical stimulation. Guinea pigs were implanted with Ppy/pTS or Ppy/pTS/NT3 electrode arrays two weeks after deafening, with some receiving electrical stimulation. Guinea pigs implanted with electrically stimulated Ppy/pTS/NT3-coated electrodes had lower electrically evoked ABR thresholds and greater SGN densities in implanted cochleae compared to non-implanted cochleae and to animals implanted with Ppy/pTS-coated electrodes. CI companies are actively developing drug-eluting electrodes, although the current emphasis is on controlling inflammation, using steroids [136]. In addition, the duration of drug release is likely to be finite and limited to the period over which the reservoir remains active.

Beyond direct infusion, alternative strategies have emerged involving the use of slow-release gels administered into the middle ear. While this method initially appeared promising, it encountered significant challenges in clinical translation. Otonomy developed OTO-413, a formulation of human recombinant BDNF combined with a thermo-reversible polymer called poloxamer P407, which allowed drug delivery for several weeks from a single intra-tympanic injection. Preclinical data from noise-induced cochlear synaptopathy were very promising [137]. Nevertheless, clinical trials failed to show any meaningful improvement over placebo controls, leading to the company winding down in late 2022 [138]. Overall, despite extensive research, neurotrophins have yet to demonstrate clinical efficacy and there are no approved therapies.

Small Molecules

Small molecules represent a diverse class of therapeutic agents that could be used to modulate cellular pathways, promote neuroprotection, and facilitate regeneration within the cochlea. Their relatively low molecular weight allows for efficient delivery and diffusion. Use of small molecule analogues of BDNF and NT-3, which act as agonists of the TrkB and TrkC receptors respectively, have shown some promise in vitro and in vivo. Amitryptaline and 7, 8 dihydroxyflavone (DHF), TrkB activators, were both shown to have some neuroprotective activity in response to excitotoxic or noise-induced damage, with the effects of the former still being evident a year after exposure. Synaptic loss was reduced and treated mice were posited to have been protected from the noise-induced deafferentation observed in control animals [139]. Similar results have been seen in vitro with the TrkC agonist and NT-3 analogue 1aA. 1aA was shown to promote SGN neurite outgrowth in cultured explants and could restore synaptic density in organ of Corti explants treated with kainic acid to induce excitotoxic damage [140]. Both DHF and 1aA can be covalently conjugated to risedronate, which can bind hydroxyapatite, allowing for the future potential of a ‘bone-binding’, slow-release compound in the cochlea [140].

Beyond neurotrophic support, small molecule drugs are being investigated to further enhance neural protection and regeneration. For example, the otoprotectant sodium thiosulfate, marketed as Pedmark/Pedmarqsi in the USA/EU, can shield against cisplatin-induced hair cell damage in those undergoing chemotherapy [141]. While this intervention may indirectly protect SGNs by preserving hair cell integrity, its primary mechanism targets sensory cells rather than directly safeguarding SGNs. A potential candidate for neural regeneration is the ROCK inhibitor. ROCK is a kinase protein which acts as a relay in the intracellular cascade involved in the remodelling of the actin cytoskeleton, triggered by the binding of the Rho small GTPase to ROCK. When this pathway was inhibited by compounds such as Y-27632, this was found to have positive effects on peripheral axon regeneration in an in vitro organotypic mouse cochlea culture, where kainic acid was used to induce excitotoxic damage [142]. Moreover, the same compound showed positive results in vivo in a mouse model of laser-induced shockwave synaptopathy [143]. Although evidence was provided of synaptic restoration, no SGN counts were given so it remains tantalizing to speculate whether neural preservation was also improved.

Gene-Based Approaches

Vector-mediated gene therapy could also potentially offer sustained supplementation of neurotrophins within a target cell population in the cochlea. Numerous studies (reviewed in [144]) have explored the use of viral vectors in animal models to deliver neurotrophins, mainly BDNF, NT-3, and GDNF, showing a reduction in SGN degeneration and an improvement in survival. Initially, studies employed Ad vectors, but their use has now been superseded by the application of adeno-associated viruses (AAVs). Although these are substantially better than Ad vectors, allowing for a higher level of transduction, they still elicit an inflammatory reaction and immune response that remains poorly characterized [145].

To avoid the immune reaction triggered by viral vectors, gene augmentation by electrotransfer has been used (reviewed in [146]). In this approach, the CI electrode array is used for a novel “close field” electroporation to transduce mesenchymal cells lining the cochlear perilymphatic canals with a “naked” DNA construct (plasmid), that does not elicit an immune response, driving the expression of BDNF and a green fluorescent protein (GFP) reporter. The focusing of electric fields by particular CI electrode configurations led to surprisingly efficient gene delivery to adjacent mesenchymal cells. The resulting BDNF expression stimulated the regeneration of atrophied spiral ganglion neurites. In this model, delivery of a control GFP-only vector failed to restore neurites, with atrophied neurons indistinguishable from those in non-implanted cochleae. However, in cochleae transfected with BDNF DNA, regenerated spiral ganglion neurites extended close to the CI electrodes, exhibiting localized ectopic branching. This neural remodelling enabled bipolar stimulation via the CI array, with low stimulus thresholds and an expanded dynamic range of the cochlear nerve, as determined via eABRs [147]. For clinical use, a gene delivery electrode is needed (the Hearing-BaDGE® device), which is inserted for the electroporation and then withdrawn and replaced with a conventional implant. An initial human trial (CINGT) has been registered, although its outcome has not yet been reported (ANZCTR ACTRN12618001556235).

Gene-based strategies are also relevant to direct reprogramming discussed in the following section. Gene-based therapies aiming at repairing synaptic defects are not included here as they do not directly address the restoration of SGN function; readers are referred to existing comprehensive reviews on this topic [148, 149].

Cell Replacement and Regeneration

Cell-Based Therapies

Cell-based therapies aim to replace lost spiral ganglion neurons by transplanting exogenous neural progenitors derived from pluripotent stem cells. As there is not an endogenous source for SGN regeneration in the adult mammalian cochlea, transplantation of exogenous otic neural progenitors (ONPs) is a logical therapeutic strategy. Also, from the cell manufacturing point of view, the cochlea is a very good candidate organ to target with a stem cell-based therapy. With around 42,000 SGNs per cochlea [150], numbers are easier to achieve when compared to those needed for regenerative strategies in other organs, such as the heart, liver, and others, requiring cell numbers on the order of 10e⁷.

To restore a lost cell population successfully, the first requirement is to generate them efficiently from a renewable and well-characterized source. Neural otic phenotypes have been obtained from several human cell sources, including bone marrow [151, 152] and dental pulp-derived mesenchymal stem cells [153, 154]. These starting tissues, although conceptually attractive as they could provide a patient-specific (i.e. autologous) source for cell replacement, present with limitations as they produce immature cells with limited or no functional capacity and their use for direct neuronal cell replacement is debated, still lacking clear evidence [155].

An alternative is the use of pluripotent stem cells. Pluripotent stem cells are undifferentiated stem cells that have the capacity to produce derivatives from the three germ layers, and based on their origin, can be segregated into two kinds: human embryonic (hESCs) and induced pluripotent stem cells (iPSCs). Embryonic stem cells are derived from early-stage blastocysts, commonly surplus to in vitro fertilization. These represent the original state of pluripotency present in the inner cell mass of the embryo and can be driven to differentiate into specific cell lineages using methods that mimic development in vivo. Induced pluripotent stem cells are generated by reprograming a somatic cell using a cocktail of factors (such as transcription factors or small molecules) to reset the cell into a pluripotent state comparable to that of an embryonic stem cell. These cells, once established as cell lines, can be induced to differentiate just like their embryonic counterparts. The use of pluripotent stem cells for cell-based therapy has been demonstrated, with clinical trials taking place in neurological conditions including Parkinson’s disease and drug-resistant focal epilepsy (reviewed in [156]).

It could be argued that a pluripotent stem cell, due to its very nature, could present a safety risk if used as a therapeutic agent. However, several strategies have been developed to mitigate potential risks associated with PSC-derived therapies. For example, only differentiated cells are typically used for transplantation. During manufacture, rigorous purging of pluripotent cells takes place, and the final cell populations are screened for chromosomal aberrations before administration. Long term safety studies in a variety of animal models have been conducted, in which the potential for migration of cells to other, off-target organs is tracked. Such processes have ensured that to date, in over 100 clinical trials for PSC-derived therapies, with a cumulative dose of 10¹¹ cells in >1200 patients, no safety concerns have been identified at the time of writing [156]. For example, there were no reported adverse outcomes in patients who received hESC-derived retinal pigmented epithelial cells [157]. Moreover, several studies have been completed with PSC-derived dopaminergic neurons for Parkinson’s Disease and inhibitory GABAergic interneurons for mesial temporal lobe epilepsy (reviewed in [156]), again with no long-term issues related to the transplanted cells.

Another significant perceived limitation of cell-based therapies is the requirement for long-term immunosuppression to prevent host rejection. However, recent studies have demonstrated that short-term immunosuppressant regimens can be sufficient to allow the successful engraftment and long-term survival of transplanted cells, even after cessation of external immunosuppression. This has been reported in immune-privileged sites such as the retina and the brain for conditions like macular degeneration and Parkinson’s disease [158, 159], setting an encouraging precedent for the cochlea. Furthermore, the development of hypoimmune cells, which are engineered to disrupt the expression of surface antigens involved in immune recognition, may eventually eliminate the need for immunosuppression altogether [160] as would the use of autologous, patient-derived iPSCs.

The Use of Human Pluripotent Stem Cells for Auditory Nerve Repair

An increasing body of evidence supports the rationale of using human pluripotent stem cells (hPSCs) to generate neural progenitors that can restore the loss of SGNs. Chen et al. (2012) described one of the first developmentally informed protocols to produce ONPs, generating an otic placodal-intermediate stage cell expressing defined otic markers, which could be matured into cochlear neurons [161]. In this study, a gerbil model of auditory neuropathy was used, in which ouabain was applied to the round window. This treatment selectively depleted type I SGNs while preserving the hair cells. When these human ONPs were delivered into the cochleae of gerbils that had been treated with ouabain, the transplanted cells engrafted, made connections to the hair cells and partially restored auditory brainstem thresholds [161]. Re-innervation of the ouabain-treated gerbil cochlea was also demonstrated when using neural crest-like sensory neurons derived from hESCs, although there was no evidence of hearing restoration [162]. Neurons derived from hPSCs have shown to be capable to establish synapses with hair cells in vitro [163, 164] and in vivo [161], and also connect with the brainstem, in vitro [164] and in vivo [161]. However, the ability of transplanted cells to recapitulate the full functional heterogeneity of SGNs remains an open question. Future research should focus on advancing our understanding and engineering of SGN subtype specialization and connectivity.

The use of biological scaffolds could potentially enhance the differentiation of the transplanted cells. By using a 3D collagen matrix with a conventional neural-inducing protocol, Ishikawa et al. (2017) showed that the proportion of human iPSC-derived neurons expressing VGLUT1 when transplanted into the guinea pig scala tympani was >50 % of the total transplanted cells [165]. However, this did not prolong the durability of the graft [165]. Self-assembling peptide amphiphile molecules have been used to create a niche for hESC-derived ONPs, enhancing differentiation and extension of neurites in vitro and promoting cell survival in a rat in vivo model [166].

Differentiation of pluripotent stem cells into 3D organoid systems has acquired great momentum lately, as they create a more physiological environment that favours maturation and terminal differentiation recreating some of the functional properties of complex structures [167]. This approach presents clear advantages when modelling an organ, either for the study of a disease-causing mutation or for testing new drugs. Although the use of 3D cell aggregates has been explored for the delivery of ONPs [168], 3D cultures and aggregates are not necessarily advantageous for cell-based therapy. The methods for generating organoids are still intrinsically very variable and difficult to adapt to an industrial manufacturing process. In addition, organoids are very complex structures harbouring multiple cell types, severely compromising the purity required for a defined cell-based therapy product [169].

A formulation of hESC-derived ONPs, Rincell-1, has recently received clinical trial authorisation to commence its testing in a Phase I/IIa trial in adults with presbycusis and postsynaptic auditory neuropathy (ClinicalTrials.gov: NCT07032038). It is under development by Rinri Therapeutics, a UK based biotechnology company. Although this will be an early phase first-in-human trial focusing on safety and the treatment efficacy is yet to be established, this advancement marks an important milestone in the progress towards a regenerative treatment for neural hearing loss.

Direct Reprogramming

In addition to transplantation of exogenous cells, recent advances have highlighted the potential of direct reprogramming strategies to regenerate SGNs from endogenous cell populations. Direct reprogramming offers an alternative regenerative strategy by converting endogenous non-neuronal cells into SGN-like neurons in situ. It refers to the conversion of a cell’s fate without transition through an intermediate pluripotent or multipotent cell state and is a promising emerging therapeutic strategy [170]. Also known as transdifferentiation, this approach seeks to harness the potential of resident glial cells or fibroblasts, potentially overcoming challenges related to cell sourcing and immune compatibility. Various molecular mechanisms and strategies for manipulating cell fate are being investigated involving forced expression of transcriptional factors, epigenetic modifications, metabolic factors, non-coding RNAs and small molecules [170]. Following loss or damage of auditory neurons, there often remains a population of non-neuronal, mainly glial cells. Leveraging direct reprogramming could enable regeneration of SGNs by guiding glial cells to adopt a neuronal identity, using developmental cues from the cochlea as a blueprint.

Using this approach, the transcription factors NeuroD1 and Ascl1 have been shown to convert non-neural spiral ganglion cells into neurons in vitro when introduced through gene transfection [171, 172]. Recent advances have further demonstrated that direct reprogramming can be achieved not only from glial cells but also from fibroblasts, as shown by Huang et al. (2025), who successfully converted fibroblasts into SGN-like cells using defined transcription factors in vitro [173]. Although this approach could be therapeutically viable, particularly with the advances made recently in gene therapy approaches to the cochlea, translation in vivo will require achieving cell-type specificity to avoid off-target effects, ensuring that reprogrammed neurons integrate functionally into existing auditory circuits, and developing efficient, clinically feasible delivery methods [170, 172]. Further research is needed to demonstrate efficacy and safety in vivo, particularly in large animal models and, ultimately, in humans. Combination approaches, such as providing neurotrophic support alongside reprogramming, may further enhance the survival and integration of newly generated neurons [172].

Treatment Delivery

Beyond the challenges with sustained treatment delivery discussed in the 'Neurotrophins' section, developing and refining surgical procedures for accessing treatment target areas and drug delivery systems are important across all therapeutic strategies.

In many of the studies mentioned above, substances have been directly applied to the round window membrane or infused into the cochlear fluid compartments. A new delivery device involving a 3D-printed series of microneedles to make tiny punctures and preserve the structure of the round window membrane has been trialled and can successfully deliver siRNA into the cochlea in guinea pigs, with transfection of SGNs evident [174]. However, these methods are relatively invasive as they still require surgical access to the cochlea. Recent innovations, such as microneedle-based delivery devices and slow-release materials administered via a trans-tympanic approach, offer less invasive alternatives, but further refinement and validation are needed before widespread adoption [175].

Another key challenge that was been partially addressed is the development of minimally invasive surgical techniques for cell therapies [176–178]. For example, an access approach previously used in animal models has been adapted with the aid of high-resolution synchrotron imaging and a minimally invasive route for delivery into the internal auditory meatus (IAM) has been identified [177, 178]. This procedure was developed to integrate access to the IAM into routine CI surgery, enabling a combined therapeutic approach. However, for pluripotent stem cell-derived ONPs to be delivered as a monotherapy, alternative minimally invasive access routes will be required to reach the neural compartments without need for cochlear implantation.

Conclusions and Future Directions

Neural hearing loss remains a significant unmet clinical need, with profound implications for communication, cognitive health, and quality of life. It is not merely an under-recognized condition but a key determinant of hearing deficits. The pathophysiology is complex, spanning from the early loss of synaptic connections to the retraction of neural fibres and the degeneration of SGNs, the consequences of which are not adequately addressed by current interventions. While hearing aids and CIs are the gold standard of auditory rehabilitation, they fall short in restoring the intricate neural architecture necessary for listening in more challenging acoustic environments.

The limitations of these conventional approaches underscore the rationale for novel therapies aimed at restoring neural integrity. Elevating neural health to equal clinical importance as sensory impairments is essential for developing effective interventions that address the full complexity of hearing loss. Therapeutic approaches aiming to protect or repair SGNs, such as neurotrophin delivery, have shown promise in preclinical models and would be extremely valuable if validated in clinical studies. Beyond these, treatments aiming to fully regenerate damaged SGNs, including cell-based therapies, are being developed and entering clinical trial stage. Collectively, these emerging therapies represent a paradigm shift in the management of neural hearing loss, moving from compensatory devices to biologically restorative interventions.

Beyond challenges with treatment delivery discussed earlier, the refinement of neural health biomarkers is essential for accurately assessing the status and progression of neural degeneration, as well as monitoring therapeutic efficacy. Current diagnostic tools are limited in their ability to capture the functional benefits of interventions, particularly those that extend beyond improvements in pure-tone thresholds. There is a pressing need to develop, validate, and embed outcome measures that reflect real-world listening abilities and cognitive load, ensuring that clinical trials and care pathways capture the full impact of novel therapies.

Safety remains an important consideration, particularly for cell-based and gene therapies. While preclinical and early clinical studies have not identified major safety concerns to date, it is essential to maintain careful, long-term monitoring as these therapies progress towards broader clinical use. Drawing on experience from other fields, robust protocols for patient follow-up and risk assessment will be crucial to ensure that emerging treatments remain both effective and safe over time.

Ultimately, these advancements are poised to reshape treatment paradigms, moving from compensatory devices to biologically restorative interventions. Bridging the gap between pre-clinical promise and clinical application will demand robust translational models and interdisciplinary collaboration. As we refine our tools and expand our understanding, the prospect of restoring natural hearing through neural regeneration is becoming an increasingly tangible goal.

Author Contribution

Conceptualization: EG, EP, CMCB, LA, RH, DEHH, MNR; Investigation: EG, EP, CMCB, LA, DEHH, MNR; Writing - Original Draft: EG, EP, LA, MNR; Writing - Review & Editing: EG, EP, CMCB, LA, RH, DEHH, MNR; Visualization: EG, EP, CMCB, LA, DEHH, MNR; Project Administration: EG, EP, CMCB; Supervision: DEHH, MNR; Funding Acquisition: RH, DEHH, MNR.

Funding

This work was funded by Rinri Therapeutics Ltd.

Data Availability

This article is a review of previously published studies. No new data were created or analysed.

Declarations

Use of AI Technology

Artificial Intelligence (AI) assisted tools (Microsoft Copilot; Generative Pre-trained Transformer-based text-refinement functions; Microsoft Corporation; used intermittently between August 2025 and January 2026) were employed to improve readability and ensure a cohesive writing style throughout the manuscript. These tools were used solely for grammar, spelling, punctuation, clarity, and style refinement, and for generating brief summaries of author-provided text. The AI tools did not generate scientific content, perform data analysis, or create figures or artwork. All authors reviewed and approved the AI-refined text and accept full responsibility for the integrity and accuracy of the final manuscript.

Competing Interests

All authors are employees of and/or financially supported by Rinri Therapeutics Ltd. Marcelo Rivolta is the Founder, Director, and Chief Scientific Officer and Douglas Hartley is the Chief Medical Officer of Rinri Therapeutics Ltd.