The auditory nerve implant—concept and device description of a novel electrical auditory prosthesis

Abstract

The cochlear implant (CI) is considered one of the most successful neural prostheses, enabling deaf individuals to achieve intelligible speech perception. However, CI performance remains limited in noise and with complex acoustic scenes, including music and multi-talker speech. One major issue for CIs is the poor electrode-neural interface where electrodes are positioned within the bony cochlea and distant from the auditory nerve fibers. Due to recent advances with microelectrode technologies designed for peripheral nerves, there has been rekindled interest in the auditory nerve implant (ANI), in which a novel prosthesis with a microelectrode array has been developed for direct stimulation of the auditory nerve. Animal studies demonstrate that the ANI achieves substantially lower thresholds and more selective neural activation compared to CI stimulation, which could lead to greater hearing performance. To successfully translate the ANI to patients, the ANI device components need to be further designed for safe and reliable implantation in humans through development of alternative surgical techniques, and validated in chronic animal studies. New stimulation strategies also need to be developed, especially with the potential to insert tens to hundreds of microelectrodes across the spiraling tonotopy of the auditory nerve to activate more spatially and temporally distinct nerve fiber patterns than is possible with the CI. Once in humans, extensive perceptual experiments can be performed with the ANI to characterize thresholds, loudness growth functions, pitch patterns, temporal coding properties, and spectral selectivity, as well as evaluating novel stimulation strategies that will guide the development of the next generation ANI system.

Introduction

Hearing loss is the most prevalent sensory disorder, affecting over 466 million individuals, which is more than 5% of the world population according to the World Health Organization. The cochlear implant (CI) has become the standard treatment for severe or profound hearing loss since the U.S. Food and Drug Administration (FDA) approval in the mid-1980s, with approximately one million recipients worldwide and about 65,000 new implantations each year (Zeng 2022). A CI bypasses the non-functioning hair cells in the cochlea and directly stimulates the spiral ganglion neurons (SGNs) with electrical pulses that correspond to the acoustic signal recorded via a microphone. Over the past two-plus decades, advancements in technology, particularly in microelectronics and signal processing informed by significant scientific research in hearing and speech perception, have driven the development of highly effective CIs, particularly for speech in quiet environments (Carlyon and Goehring 2021; Lenarz et al. 2022). Modern CIs integrate multi-channel electrode arrays with advanced audio processing strategies (Clark et al 1987; Wilson et al. 1991; Boyle et al. 2009), enabling the average user to achieve 70–80% accuracy in sentence recognition in quiet environments without lip-reading. Many users can even converse over the phone in quiet settings (Wilson 2000; Skinner et al. 1994). Additionally, CIs offer opportunities for children to acquire language skills (Lenarz et al. 2022; Niparko et al. 2010; Sharma et al. 2002; Holt and Kirk 2005; Harrison et al. 2005), prompting FDA approval for implantation in children as young as 9 months. The standard of care in many countries recommends implanting deaf children from the age of 6 months, some even recommend bilateral implantation (e.g., German Guideline on Cochlear Implantation) (Cochlear implant care 2020). Access to binaural cues has been shown to provide significant benefits, as demonstrated by improved outcomes in individuals with either bilateral implants or bimodal hearing, where residual acoustic hearing is combined with a CI (Tyler et al. 2002; Müller et al. 2002; Litovsky et al. 2006; Blamey et al. 2015; Olson and Shinn 2008; Graaff et al. 2021; Yang and Zeng 2017; Gifford 2020; Hoesel and Tyler 2003).

Limitations of the CI

Despite the remarkable successes of CIs, users still face major limitations such as a restricted dynamic range (the perceived range of loudness with electrical stimulation) (Zeng et al. 2002; Moore 2003), poor perception of pitch and timbre (Oxenham 2008; Moore et al 2005; Gfeller et al. 2002), and dissatisfaction with the unnatural quality of auditory percepts. They also experience significant difficulties recognizing speech in background noise (Müller-Deile et al. 1995; Nelson et al. 2003; Stickney et al. 2004), perceiving tonal languages (Fu et al. 2004; Wei et al. 2004), and enjoying music (Limb and Roy 2014; Dorman et al. 1991; McDermott 2004).

One challenge for CIs is the variability in electrode insertion depth caused by differences in the cochlear duct length (25—45 mm) (Alexiades et al. 2015) among individuals. Such variability prevents some patients from accessing frequency percepts below roughly 600 Hz despite the proven benefits of low-frequency information for word understanding, speech in noise, and melody recognition (Luo and Fu 2006; Gantz et al. 2005). Although attempts have been made to make longer arrays, there remains challenges of safely inserting the electrode array into the decreasing space towards the apex of the cochlea.

A second key bottleneck in CI performance results from the electrode-neuron interface (Zeng 2017). The CI array is immersed in conductive perilymph (cochlear fluids), leading to broad current spread and shunting. Additionally, these electrodes are separated from the targeted SGNs by the poorly conductive bony cochlear wall, the modiolus. As a result, higher current levels are required to activate the SGNs, and overlapping electric fields generated by different electrodes (Kral et al. 1998) reduce spectral resolution, hindering speech perception in noise and music appreciation (Nelson et al. 2003; Fu et al. 1998; Qin and Oxenham 2003; Nelson and Jin 2004). Although innovations like bipolar, partial tripolar, or dynamic current focusing configurations aim to reduce the spread of neural activation (Berenstein et al. 2008; Bierer and Litvak 2016; Zhu et al. 2012; Jong et al. 2019; Donaldson et al. 2011), current CI designs with 12–24 channels still provide only 4–8 effective channels of information (Wilson and Dorman 2008). Research involving normal-hearing listeners and vocoder simulations of a CI suggests that at least ten effective channels are necessary for optimal performance in challenging listening conditions (Fu and Nogaki 2005; Dorman et al. 1998; Shannon et al. 2004), while over 32 channels are required to convey sufficient information for salient pitch perception (Mehta et al. 2020; Mehta and Oxenham 2017). Limited by the functional spectral resolution, audio-processing strategies of the contemporary CI primarily utilize the slow temporal envelope of acoustic signals extracted in different frequency bands to modulate the amplitude or the pulse duration of the electrical pulses. This approach inherently limits the transmission of temporal fine structure in acoustic signals (Loizou 2006; Zeng et al. 2005). Although efforts have been made to convey this information by varying the pulse rates within each channel, evidence remains insufficient to demonstrate that CI users can perceive or effectively utilize such cues (Riss et al. 2016; Nogueira et al. 2009).

CI stimulation also presents unique challenges from a neural coding perspective. Electrical currents directly depolarize auditory nerve fibers introducing abnormal synchronization across nerve populations (Raggio et al 1999; Schreiner and Raggio 1996). The lack of stochastic variability in response timing combined with broad activation (Snyder et al. 2000; Snyder et al. 2004; Bierer and Middlebrooks 2002) can interfere with information transmission to higher auditory centers (Moore and Shannon 2009; Middlebrooks et al. 2005). CI stimulation has been shown to be less effective at activating neurons in the primary auditory cortex that are highly selective to fine spectral structure (Johnson et al. 2016). These effects likely contribute to altered neural computations underlying perception and cognition. Notably, even CI users with good performance in speech tasks may rely on alternative acoustic cues, such as spectral tilts, when spectral peaks (formants) are compromised (Winn and Litovsky 2015; Winn et al. 2012). Studies have also shown increased cognitive demands for CI users (Winn and Teece 2021; O’Neill et al. 2021; Hughes et al. 2018) and abnormal context effects (Stilp 2017; Feng et al 2018; Aravamudhan et al 2005), a process in which the auditory system adapts to the acoustic variability from different environments or speaker characteristics to maintain perceptual constancy (Barlow et al 1961; Dean et al. 2005; Ladefoged and Broadbent 1957; Watkins 1991).

Advancements in alternative hearing prostheses

Neurotrophic agents have been used to promote closer proximity between electrodes and auditory neurons. Previous studies show that such delivery results in rescue or enhanced survival of SGNs over extended cochlear regions, though additional study is needed to achieve targeted, sustained, and safe delivery (Pinyon et al 2014; Leake et al. 2020; Budenz et al. 2012). Alternative prostheses, such as the auditory midbrain implant (AMI) or auditory brainstem implant (ABI), are being developed for patients without functional auditory nerves (Lim and Lenarz 2015; Wong et al. 2019). However, stimulating the central auditory system is challenging. Unlike auditory nerve fibers that map frequencies tonotopically, phase lock to temporal modulations up to 4—6 kHz, and show monotonic responses to increasing loudness (Sumner and Palmer 2012; Evans 1972; Johnson 1980; Taberner and Liberman 2005; Møller 1983; Sachs and Abbas 1974), central neurons encode acoustic features intrinsically. They exhibit sophisticated and diverse frequency receptive field and rate level functions (Syka et al. 2000; Palmer et al. 2013) and transition from phase locking to rate coding above 1000 Hz (Liu et al. 2006; Bartlett and Wang 2007; Lu et al. 2001; Wallace et al. 2007) complicating stimulation strategies.

Researchers are also investigating light-based approaches to achieve more precise spatial activation of auditory nerves. Techniques include infrared lasers (IR), which are hypothesized to alter membrane capacitance (Littlefield and Richter 2021), although the exact mechanism is still not well understood, and optogenetic methods that use light-sensitive proteins to control ion channels (Jeschke and Moser 2015; Keppeler et al 2020). Optogenetic stimulation, in particular, outperforms electric stimulation at medium and high intensities and achieves spectral selectivity comparable to acoustic stimulation at moderate intensities (Dieter et al. 2019). However, these approaches are still in the early stages of development. Optogenetic stimulation requires gene transfer into target cells, raising concerns about unknown long-term effects and sustainability, and the light source is still distant to the target SGNs. IR stimulation may be diffracted by the bony modiolar wall resulting in less focused auditory nerve activation. Substantial technological advancements, as well as rigorous safety and efficacy evaluations, are essential before these methods can transition to clinical applications.

This concept paper focuses on the development of an auditory nerve implant (ANI) hearing prosthesis, which involves the placement of a penetrating microelectrode array directly into the auditory nerve to achieve spatially focused stimulation, thereby allowing the transmission of spectrally and temporally specific information in acoustic sounds. For an initial clinical trial for demonstrating proof-of-concept, the ANI will be intended as an alternative to the ABI candidates who still have a functional auditory nerve but cannot sufficiently benefit from a CI. For example, the cochlea may be ossified, scarred, or otherwise occluded (e.g., post-meningitic, fracture, inflammatory) and a CI array cannot be properly inserted. Another example is that CI use is limited by facial nerve or vestibular side effects and/or insufficient hearing performance. Through success in this initial patient cohort and advancements in surgical approaches to the auditory nerve, future ANI studies will broaden the inclusion criteria as appropriate. In this concept paper, we evaluate the feasibility of an ANI from historical context, electrode fabrication, and surgical perspectives, and we comment on key challenges in safety and stimulation efficacy. A primary goal of our translational effort is to implant an ANI device in an initial cohort of deaf patients with a planned 12-month follow-up to assess device safety and stability and to characterize the percepts generated from direct auditory nerve stimulation. Continuation, removal, or re-implantation after 12 months will be determined case-by-case based on risk–benefit considerations, clinical judgment, participant preference, and ongoing monitoring. These first-in-human findings can then guide the development of an optimal ANI design with proper surgical implantation techniques and novel stimulation strategies for evaluation in a multi-site efficacy clinical trial to open the doors for a new type of hearing prosthesis beyond the CI.

ANI concept and design

Prior ANI studies

Direct auditory nerve stimulation was initially explored alongside the development of CIs (House 1976; House and Urban 1973). In 1963, Zoellner and Keidel inserted a 350 µm diameter electrode wire into the cochlea eliciting auditory sensations, although the claimed placement within the modiolus was unconfirmed (Zoellner et al 1963). In 1965, Simmons et. al. implanted a bundle of five 75 µm diameter electrodes with exposed tips linearly spaced 1 mm apart into the auditory nerve through the modiolar wall (Simmons et al. 1965), achieving stable function for over five years with histological evidence of healthy neurons (Simmons et al. 1979). Pitch ranking was possible but highly dependent on the type of stimuli (Simmons et al. 1981).

Despite initial promise, ANIs received far less attention than CIs due to surgical complexity and technological limitations, particularly in the lack of proper microelectrode array technologies suitable for long-term implantation and sufficient spatial coverage of the auditory nerve at that time. The cochlea’s bony structure, though suboptimal for electrode-neuron interfacing, provides a stable and protective cavity for electrodes that reduces movement and minimizes the risk of trauma to the eighth nerve. Its tonotopic organization (Greenwood 1961) with high frequencies mapping to the base and low frequencies mapping to the apex of the cochlea apparently simplifies electrode design. In contrast, the auditory nerve’s location in the narrow internal auditory canal (IAC) near the vestibular and facial nerves may complicate surgical access and long-term electrode placement. Furthermore, the bundled auditory nerve fibers that possibly twist and spiral en route to the brainstem require a three-dimensional array of closely spaced electrodes to sufficiently span the cross-sectional array of the nerve to cover the full range of frequencies encoded by those fibers. It is important to note that even for CIs, tonotopic stimulation is challenging due to the need for current to transmit through the bony modiolar wall to access distant and remaining components of SGNs that are also bundled up and traverse towards the IAC.

Interest in ANIs was renewed with advancements in higher-count electrode array designs and thin-film multichannel electrode technologies. In 1986, Naumann et al. implanted a 12-channel wire-bundle into the auditory nerve via the middle ear and oval window, targeting the nerve as it exits the modiolus into IAC (Naumann et al. 1986). Each electrode wire was arranged to diverge within the auditory nerve for targeted distribution. Although reproducible auditory percepts were achieved, the study was discontinued after 66 days due to possible moisture damage to the receiver. Later, Zappia et al. (1990)demonstrated low thresholds averaging 10 µA for charge-balanced biphasic pulses (200 µs per phase, 400 µs total) using silicon-based microelectrodes in guinea pigs, which were significantly lower than average CI thresholds (Chatterjee 1999). However, they also observed localized SGN loss and vascular injury with chronic implantation.

In 2003, a Utah Electrode Array (UEA; three-dimensional 12-channel penetrating silicon-substrate electrode array) was implanted into the auditory nerve of cats for up to 52 h under anesthesia (Badi et al. 2002; Badi et al. 2003; Hillman et al. 2003). Electrically-evoked auditory brainstem responses (eABRs) confirmed successful activation of the auditory nerve with a median threshold of 15 µA and no obvious activation of vestibular or facial nerves (Badi et al. 2007; Kim et al. 2007). Postoperative imaging showed that the UEAs remained intact and in position with an uncompromised vascular supply. However, long-term stability and biocompatibility were not assessed. In 2007, Middlebrooks and colleagues demonstrated neural responses in the central auditory system using a multisite silicon-substrate array (NeuroNexus, Ann Arbor, Michigan) implanted into the modiolar trunk of the auditory nerve in cats (Middlebrooks and Snyder 2007; Middlebrooks and Snyder 2008; Middlebrooks and Snyder 2010). Their findings highlighted substantially lower thresholds, greater dynamic ranges, access to lower frequencies not typically possible with a CI, and significantly reduced spread of excitation compared to a CI. However, the array’s fragility and challenges in encapsulation and explanation limited its potential for long-term use and clinical translation.

Re-evaluating the feasibility of the ANI

Advances in microelectrode array design, manufacturing, and longevity have addressed many obstacles that previously hindered ANIs. In particular, the novel Utah Slanted Electrode Array (USEA) features a three-dimensional architecture designed to interface with the cross-section of peripheral and cranial nerves (Davis et al. 2016; Wendelken et al. 2017; George et al. 2020a; George et al 2019; Page et al. 2018), making it well-suited for an ANI. The introduction of low impedance arrays through iridium oxide tip metallization processes was a critical breakthrough, allowing effective stimulation within the voltage compliance limits of CIs. Additionally, the development of a helical lead designed for peripheral nerve applications (George et al. 2020a) has successfully passed the stringent lead fatigue tests required by the CI86 standard (Instrumentation A for the A of M, others 2017).

The NeuroPort™ Electrode array (Blackrock Neurotech, Salt Lake City, Utah) is a penetrating microelectrode array that offers up to 100 silicon microneedles arranged with a 400 µm spacing. Each electrode is electrically isolated by glass and is equipped with patterned metal bond pads that connect to a helical wire bundle. The needle tips are metalized with a layer of sputtered iridium oxide film (SIROF) (Negi et al. 2010), providing a highly conductive and biocompatible interface for neural stimulation. The microelectrode array is encapsulated with a Parylene-C layer for insulation and protection (Caldwell et al. 2018; Caldwell et al. 2020; Hsu et al. 2008). Following encapsulation, the Parylene-C layer is selectively etched from the needle tips, exposing the SIROF coating at the active sites (Campbell et al. 1991; Leber et al 2019; Bhandari et al. 2010). This process reduces the impedance of the electrodes to typical values of 5 to 20 kΩ in saline, enabling neural recording and selective low-current neural stimulation (e.g., < 10 µA) (Campbell et al. 1991; Leber et al 2019; Aoyagi et al. 2003; Branner and Normann 2000; Branner et al. 2004; Normann et al 2005; Wark et al. 2013).

The NeuroPort Electrode array has been extensively studied and is cleared for human use in specific applications. In the United States, it is FDA cleared for acute, temporary (< 30 days) recording and monitoring of brain electrical activity. Additionally, the NeuroPort Electrode and its variants have been used in several successful investigational device exemption (IDE) studies in the U.S. and European Union (EU), some lasting over numerous years (Sponheim et al. 2021; Hughes et al. 2021). The array has demonstrated versatility in various applications including stimulation in the somatosensory cortex (Hughes et al. 2021; Flesher et al. 2016) with a more recent major breakthrough of stimulation in peripheral arm nerves (George et al. 2020a; Paskett et al. 2021; George et al. 2020b). For instance, Flesher et al. showed that somatosensory cortex microstimulation evoked tactile sensations (Hughes et al. 2021; Flesher et al. 2016), and Davis et al. and Wendelken et al. reported successful recording and microstimulation in human peripheral nerves to control prostheses (Davis et al. 2016; Wendelken et al. 2017; George et al 2019; Paskett et al. 2021; Page et al. 2021).

Complementing these technical advancements, established intracranial skull base surgical techniques and innovative tools have significantly improved the feasibility of the ANI (Jackler et al 2009; Schwam et al. 2023; Lucas et al. 2023; Kashani et al. 2023). Decades of experience with CIs (Naples and Ruckenstein 2020), along with extensive research on cochlear anatomy, otopathology, physiology, and surgical interventions, provide a robust foundation for optimizing ANI array placement while minimizing damage to surrounding tissues. This surgical precision is crucial for maintaining long-term functionality and patient safety. Collectively, these developments have laid the groundwork for future clinical applications with an ANI.

Proposed ANI system design

The proposed ANI system (Fig. 1) combines advanced CI technology components (i.e., the Sonnet 2 audio processer and the Synchrony 2 implantable stimulator by MED-EL, Innsbruck, Austria), with a novel ANI microelectrode array and cabling (Blackrock Neurotech, Salt Lake City, Utah) to interface directly with the auditory nerve. The ANI array is a modified USEA tailored to the anatomical and functional characteristics of the human auditory nerve. It is comprised of 15 electrodes arranged in a 3 × 5 grid with an electrode-to-electrode spacing of 400 µm (Fig. 2, feature B.1). Electrode length increases linearly from 0.5 mm to 0.9 mm across rows to maximize cross-sectional area coverage of the auditory nerve. A platinum foil handling fin embedded in medical-grade silicone facilitates safe handling and precise placement during surgery (Fig. 2, feature B.2).

Fig. 1.

Concept of the auditory nerve implant (ANI) system consisting of an audio processor and cochlear implant simulator (MED-EL, Innsbruck, Austria) integrated with the ANI array (Blackrock Neurotech, Salt Lake City, Utah), which is placed directly into the auditory nerve medial to its exit from the cochlea. Image courtesy of Blackrock Neurotech and MED-EL. This schematic portrays the concept of the ANI system in which individual components are not drawn to scale or do not represent the final design that will be implanted in humans, and the placement along the nerve is expected to be closer to the cochlea within the IAC

Fig. 2.

The components of the ANI system. (A) Backside of the integrated device, showing the silicone wing (A.1) for surgical handling, a MED-EL Synchrony 2 cochlear stimulator (A.2 and A.3), and the helical lead (A.4) for enhanced mechanical stability. (B) Close-up of the ANI array, highlighting its 3 × 5 microelectrode configuration (B.1), platinum handling fin for surgical manipulation (B.2), bendable plastic section of the wire bundle (B.3), and the beginning of the helical lead (B.4)

The wire bundle connecting the electrode array to the stimulator is constructed from the same gold alloy wire used in the NeuroPort Electrode and is overmolded in medical-grade silicone. Most of the wire bundle is coiled into a helix (Fig. 2, feature A.4 and B.4). This helical design improves the flexibility and minimizes the risk of wire breakage from bending fatigue or surgical handling. These improvements have been demonstrated previously through innovative development and testing in median and ulnar nerve stimulation in arm amputees for control of a robotic arm (George et al 2019; George et al. 2020b; Duncan et al. 2019). A short, uncoiled segment of the lead extends from the edge of the electrode array before the start of the helical coil. This unique portion creates a bendable, plastic section that retains its orientation and shape, allowing for precise positioning of the array on the auditory nerve during surgery (Fig. 2, feature B.3). To further assist surgeons during placement, a silicone handling wing is molded into the helical coil just after the beginning of the helix to enhance torsional control (Fig. 2, feature A.1). Both the silicone handling wing and the platinum handling fin are innovations that were created based on numerous cadaver studies performed by the co-authors of this paper that greatly facilitate the surgical positioning and implantation of the electrode array into the auditory nerve. Finally, the helical lead is precision molded allowing for press fitting of the lead into the mastoid bone groove.

The initial ANI device (Fig. 2) has been developed with available technologies and materials that have already been extensively used safely in humans (i.e., Blackrock NeuroPort Electrode and MED-EL Synchrony 2 components). The rationale for using available technologies is to enable safe and expedient translation to patients so that critical aspects of device design and function can be examined. The MED-EL Synchrony 2 stimulator has 12 stimulation channels and electrodes, therefore, a 3 × 5 electrode array was used in this first device design (with three inactive electrodes). This design will allow for examination and rigorous psychophysical testing of chronic microelectrode stimulation of the auditory nerve in awake and responding patients. Additionally, this design will hopefully allow for open-set speech recognition for these patients. Future ANI systems will have increased channel counts as well as more effective stimulation strategies specific to microstimulation of the auditory nerve based on results from this first-in-human study.

Surgical considerations for ANI implantation

Our team has carefully evaluated multiple surgical approaches for placing the ANI array in the auditory nerve. We evaluated standard skull base approaches to the IAC and vestibulocochlear nerve bundle, including translabyrinthine, retrosigmoid, retrolabyrinthine, and middle fossa approaches, as well as endoscopic-assisted approaches. This research progressed to fabrication of prototype electrode arrays and surgical placement of electrodes in more than 40 auditory nerves in human cadaver studies, which will be reported upon separately. Subsequently, intraoperative acute experiments were performed with human subjects under Ethics Committee approval (Clinical Trial NCT06306534) in Hannover, Germany. Patients requiring vestibular schwannoma resection with auditory nerve preservation gave informed consent to have an ANI array briefly implanted in the auditory nerve. Brief pulse trains of electrical stimulation were presented on a subset of the implanted electrodes, and the resulting eABRs were recorded. These results will also be reported in a future publication.

As the auditory nerve exits the cochlea, it converges with the vestibular nerve in the IAC to form a nerve bundle (cranial nerve VIII) that is immediately next to the facial nerve (cranial nerve VII). Thus, insertion in the lateralmost part of the auditory nerve, proximal to the cochlea, is more likely to result in auditory-specific stimulation compared to a medial IAC or cerebellopontine angle insertion. The translabyrinthine approach provides access to the auditory nerve bundle within the lateral IAC while also affording a natural bony support for the nerve during electrode insertion. The precision-molded helical lead can be press fit and securely anchored in precisely sized and oriented grooves in the mastoid bone, while packing materials can be placed around the inserted array to minimize the effects of cerebrospinal fluid (CSF) pulsations in the area for increased device stability. This approach allows for accommodation of anatomical variations across patients and facilitates optimal placement of the electrode array for the initial human pilot study. Based on our observations, the translabyrinthine approach will be used for the first-in-human scientific study with the ANI in deaf patients. Once the auditory nerve is exposed, the implantable stimulator is placed in a standard skull bed (same as with a CI), and the lead is fixed in the mastoid grooves. Then, the electrode array is positioned above and inserted into auditory nerve. After insertion of the electrode array, packing materials are added to the cavity, and the final aspects of the surgery are completed.

Future directions

The successful clinical translation of an ANI will require a multidisciplinary collaboration among engineers, clinicians, physiologists, and experts in speech and hearing to evaluate the device’s safety and efficacy. Future research and development should prioritize the following key areas:

Ensuring long-term safety and durability of the electrode array

A comprehensive pre-clinical testing plan will evaluate electrode impedance, structural integrity, and material degradation over an extended equivalent lifetime. Long-term stimulation testing under conditions exceeding typical clinical requirements, such as various pulse waveforms, higher currents, and higher pulse rates, is essential to ensure device durability. The Parylene-C encapsulation layer, a vital biocompatible component, must demonstrate minimal degradation with over 90% of electrode sites maintaining stable impedances. Additionally, monitoring for particulates or signs of corrosion is necessary to confirm that materials, including iridium and tip metallization materials, are not released during stimulation.

The lead is designed to achieve an optimal balance between flexibility and durability, facilitating surgical handling, reducing tethering forces, and ensuring secure anchoring. Mechanical and electrical reliability will be validated through rigorous benchtop testing. This includes bend and elongation assessments for mechanical resilience, continuity testing for electrical integrity, and insulation evaluations to confirm dielectric robustness based on the ANSI/AAMI CI86 regulatory guidelines (Instrumentation A for the A of M, others 2017). Additionally, biocompatibility testing, chemical characterization, sterilization validation, and electromagnetic compatibility testing according to ISO 14708–7 will be a part of the final verification and validation of the ANI.

Another critical consideration for long-term safety is the biological response to chronic implantation, particularly the risk of tissue encapsulation and neural loss due to micromotion of rigid electrodes (Woeppel et al. 2021; Polikov et al. 2005; Salatino et al. 2017; Christensen et al. 2014; Biran et al. 2007; Biran et al. 2005). This tissue response has been well-documented in long-term cortical (Szymanski et al 2021) and peripheral nerve (Christensen et al. 2014) clinical investigations with the UEA, where micro-movements between the array and neural tissue can provoke glial scarring, fibrotic encapsulation, cause electrode migration, and potentially isolate the device from nearby neurons (Biran et al. 2007; Kim et al. 2004). However, unlike peripheral nerve implants, where substantial displacement may occur in the surrounding muscle activity that might generate mechanical loadings on the device, the ANI is implanted within the IAC, a constrained bony structure that offers mechanical isolation. Additionally, the simulator and lead are anchored to the skull, substantially limiting large movement of the array relative to the nerve, while still having enough flexibility in the lead to allow the array to accommodate slight movements of the nerve, such as during vessel pulsations. This anatomical context, more akin to that of cortical applications, is expected to minimize micromotion-induced injury. While some degree of encapsulation is anticipated, long-term studies of cortical UEA implants have shown that stimulation and recording can be maintained for several years despite the presence of a tissue response or damage to the electrode arrays (Sponheim et al. 2021; Hughes et al. 2021; Woeppel et al. 2021; Bjånes et al. 2025; Patrick-Krueger et al. 2024; Greenspon et al. 2024). These observations provide encouraging evidence for the chronic viability of rigid electrode arrays in anatomically protected environments such as the IAC.

Refining surgical techniques and pre-clinical testing

Continuing to refine minimally invasive and precise surgical methods is critical for reducing complications and ensuring the long-term stability of the electrode array. Cadaver studies have already provided valuable insights into the anatomical dimensions and placement requirements for the ANI array and the lead. Further cadaver experiments are being pursued by the University of Minnesota and Hannover Medical School in Germany to identify and develop less invasive surgical approaches to the auditory nerve that can be more widely implementable by the otology field. Additionally, co-authors at the University of Utah have developed a feline model and have successfully implanted ANI arrays into auditory nerves via a transbullar approach (Thomas et al. 2024). These feline surgical methods and techniques closely simulate the translabyrinthine approach used in humans. Pre-clinical validation in several animal models at the University of Utah and University of Minnesota is an ongoing effort to demonstrate the safety and efficacy of ANI implantation and stimulation for human use.

Auditory nerve degeneration is a critical consideration for candidate selection and outcome prediction with ANI. Long-term sensorineural hearing loss (SNHL) is frequently accompanied by an associated loss of auditory nerve fibers (Nadol et al. 1989), resulting in reduced nerve diameter and neural density, and the extent of degeneration can vary considerably across individuals (Herman and Angeli 2011; Naguib et al. 2017; Reimann et al. 2024; Nadol and Xu 1992; Zimmermann et al. 1995). This variability can influence ANI efficacy and patient outcomes. Nevertheless, clinical experience with CIs suggest that many individuals retain sufficient residual fibers to support functional stimulation (Kamakura and Nadol 2016; Cheng and Svirsky 2021; Seyyedi et al. 2014). To mitigate the risk associated with degeneration, preoperative imaging, such as high-resolution T2-weighted MRI with multiplanar reconstruction, can aid in assessing nerve integrity and guiding candidacy decisions. When a sufficient quantity of auditory-nerve fibers remains, the USEA could, in principle, offer design flexibility (e.g., electrode length and density) to better match smaller nerves. For individuals with severe degeneration, an ABI may be more appropriate. Additionally, neuroprotective strategies such as neurotrophin delivery or chronic low-level electrical stimulation are under active investigation and may help preserve or partially restore neural populations and promote long-term survival of SGNs (Leake et al. 2020; Leake et al. 2013; Shepherd et al. 2005; Shepherd et al. 2008). These strategies could potentially be used alongside the ANI.

Optimizing stimulation strategies

Direct stimulation of the auditory nerve has been shown to provide lower thresholds and more selective neural activation, but maximizing these potential benefits requires experimental testing and validation. Studies in animals and humans measuring evoked compound action potentials (eCAPs) and eABRs can help characterize nerve activation, establish safe stimulation ranges, and lead to improved stimulation strategies that avoid activation of nearby structures like the facial nerve. Current spread and channel interaction may also be assessed. Recording neural responses in the inferior colliculus or auditory cortex to ANI stimulation in animal models can be a useful approach for quantifying thresholds, dynamic ranges, frequency activation patterns, and phase-locking properties, as well as comparing neural patterns with acoustic responses to guide parameter optimization such as pulse rate, pulse width, current levels, and multi-polar stimulation patterns. These various approaches can lead to more effective stimulation strategies for activating the central neurons and for providing higher fidelity transmission of hearing information to the brain.

ANI introduces potential challenges that can be distinct from CIs. Although the three-dimensional array designs of the ANI can access more spatially distinct fibers spanning a broader frequency range compared to the CI, the frequency-to-place mapping in the spiraling auditory nerves may not fully follow the monotonic gradient observed along the cochlea. Consequently, spatially adjacent sites can, especially at higher currents, produce frequency crossover, activating neurons representing disparate frequencies, which can distort intended spectral features and alter loudness growth. In addition, frequency maps can vary across patients depending on the array placement. Identifying the individual frequency map and employing appropriate, low-current stimulation to minimize unwanted frequency crossover are critical for stimulation strategies. Importantly, despite these complexities, direct nerve access with ANI retains a more peripheral, tonotopically grounded interface than auditory brainstem or midbrain implants, and thus may offer better hearing outcomes when it is appropriately mapped and fitted.

In addition, action potentials primarily initiate on the central axon of the SGNs with ANI stimulation, which results in shorter latency but increased neural synchrony (Kiang and Moxon 1972; Hartmann et al. 1984). Since spiking timing across neural population carries stimulus information (e.g. mean and standard deviation of first spike timing are correlated with sound loudness) (Huet et al. 2018; Heil and Irvine 1997; Rubinstein and Hong 2003), this timing change could alter neural coding and contribute to steep loudness growth and limited dynamic range under electrical stimulation (Hong and Rubinstein 2003; Zeng et al. 1998). The loss of spontaneous activity in deafferented nerves (Johnson and Kiang 1976) further limits the independence of spike trains across fibers, thereby reducing stochastic resonance (Rubinstein and Hong 2003), a nonlinear mechanism to improve signal detection. Strategies like adding Gaussian noise (Matsuoka et al. 2000) or using high pulse rates (Hong and Rubinstein 2003) show limited potential in CI research, but they may show promise with ANI stimulation. Moreover, delivering spectro-temporal fine structure information demands coordinated pulse trains across electrode sites that may be possible with ANI stimulation. Both physiological and modeling studies may provide insights that help overcome these challenges and guide sound coding and stimulation strategy development for the ANI.

Looking forward, while increasing intracochlear contact density has not yielded significant improvements in speech outcomes (Fishman et al. 1997), largely due to current spread, direct nerve access via ANI may allow more effective use of higher-density stimulation. In principle, this could better enable transmission of spectrotemporal fine structure in complex sounds (speech, music), potentially improving sound quality, aesthetic perception, and speech understanding in noise. The USEA platform has already been demonstrated to scale toward high-density versions (HD-USEA) with up to four times more channels, and further advances in microfabrication, electronics miniaturization, and packaging may enable even greater scaling (Wark et al. 2013). These innovations could expand the number and diversity of evoked percepts, but will also pose challenges for psychophysical fitting, coding strategies, and power management. Similar to CIs, the relationship between ANI channel count and perceptual benefit is unlikely to be linear, with potential plateaus (Fishman et al. 1997). Accordingly, future clinical and psychophysical studies will be needed to determine the optimal balance among electrode density, coding approaches, and user outcomes for ANI.

Finally, a scientific study involving chronic implantation of an ANI in human patients would provide a unique opportunity to characterize important perceptual effects and capabilities of different stimulation patterns, as well as hearing performance for varying stimulation strategies. These and future studies will help guide the development of a new generation of auditory prostheses that have the potential to push hearing performance beyond the traditional CI.

Abbreviations

CI

Cochlear implant

ANI

Auditory nerve implant

eCAP

Evoked compound action potential

eABR

Electrically-evoked auditory brainstem response

SGN

Spiral ganglion neuron

IAC

Internal auditory canal

CSF

Cerebrospinal fluid

UEA

Utah Electrode Array

USEA

Utah Slanted Electrode Array

IDE

Investigational device exemption

SIROF

Sputtered iridium oxide film

Authors’ contributions

LF, the corresponding author, coordinated the manuscript structure and led drafting and revisions. HHL, TL, and FS provided the overall vision and conceptual framework for the project, and contributed to manuscript organization and revisions. KD, IS, LR, and MA contributed to sections of the initial draft. MA, TL, RS, and AS contributed to content related to surgical procedures. LR, ML, FS, JC, and RF contributed to sections and figures addressing electrodes, including manufacturing considerations. KH, SS, IH, and CB contributed to device design and provided engineering review of the processor. LAJ, GMG, MA, DJW, HH, and WMT contributed content related to large-animal models for evaluating device efficacy and safety. AJO and WN contributed to sections on cochlear implant research. All authors reviewed and approved the final manuscript.

Funding

NIH UG3 NS107688, NSF UtB DGE 1734815, German Hearing4All, DFG Cluster of Excellence EXC 2177/1, Minnesota Lions Hearing Foundation, The Hamilton and Mildred Haley Kellogg Charitable Trust, MED-EL, Blackrock Neurotech, Feinstein Institutes for Medical Research (Internal Funding), WVU (Startup Funding).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.