Auditory Neuropathy

Timothy C. Hain, MD Last updated: July 20, 2021

Other links: Hearing LossHearing Aids Hearing TestingCentral hearing loss

Rance and Starr
Schematic of the ear. The lesion in auditory neuropathy is the nerve (6) Schematic from Rance and Starr (2015) defining putative sites for testing pattern defined as "auditory neuropathy"

Introduction: An auditory neuropathy (AN) is roughly defined anatomically as a hearing loss caused by damage or aplasia of the auditory portion of the eighth nerve (6 on the picture to the left), which is located between the inner ear (cochlea-D) and the brainstem. (see item 6 on figure above-right). It is an overlap syndrome between otology and neurology, as it causes symptoms of interest to otologists (hearing loss), but is caused by a disease of interest to neurologists (nerve damage). A related acronym, ANSD -- auditory neuropathy spectrum disorder -- is a less well defined version of AN.

AN was first reported in the late 1970's as paradoxical findings because of a discrepancy between absent ABR and present hearing thresholds and OAE (although with some reduction), this disorder has been referred to as central auditory dysfunction (not a good name), auditory neural synchrony disorder, auditory neuropathy, and most recently as auditory neuropathy spectrum disorder. We will stick to the term auditory neuropathy herein (AN).

It has been suggested that there are two forms of AN-- a "presynaptic" type involving damage to the auditory nerve, and a "postsynaptic" version reflecting damage at the synapse of the cochlear nerve with the cochlear nucleus (Rance and Starr, 2015). See the figure above for Rance/Starr's opinion regarding of the site of lesion. Some reports find cochlear aplasia -- a small cochlear nerve or bony aperture for the cochlear nerve (Jeong et al, 2013). This group, lacking the wiring for hearing, would be expected to do very poorly concerning acquisition of hearing.

Auditory neuropathy, when it is truly due to nerve damage, may also affect vestibular function (Sheykholeslami et al. 2000; Sinha et al, 2013). Sujeet et al (2014) reported that absent VEMP and reduced caloric responses were common in AN patients. Nash (2014) reported vestibulopathy in 42% of children getting cochlear implants for AN. Of course, these were likely very severe situations.

Similarly, there probably is an analogous "vestibular neuropathy" syndrome, characterized by absent vestibular evoked potentials (VEMP tests), and peculiarities in the ENG or rotatory chair test. The recently reported "CANVAS" syndrome (cerebellar ataxia, neuropathy, vestibular areflexia) would be a reasonable candidate for this as well. In other words, the vestibular findings of AN may have been largely overlooked. Otherwise, it would be very puzzling how a disorder of the 8th nerve could selectively damage the auditory portion or vestibular portion as the two parts of the nerve are extremely close and one would think that the same process that damages one would also damage the other.


Prevalence and Incidence

It is possible (but not very likely) that the medical profession underdiagnoses auditory neuropathy. First, the syndrome itself is little known. Second, one of the tools to diagnose auditory neuropathy (otoacoustic emissions or OAE) is generally little known to any specialty physicians other than otolaryngologists. On the other hand, it may be that the AN syndrome is just so imprecise that there is little clinical attention devoted to it and it deserves to remain obscure.

There is an interesting interaction between newborn hearing screening programs (which typically use OAE), and AN. One would expect that AN would be always missed in these programs, because AN is defined by present OAEs. (Korver et al, 2012). Thus we might be missing AN early on.

Supporting the second conjecture, that the AN syndrome is just so imprecisely defined, that it is often ignored, the estimates of the prevalence of AN varies from 1/200 patients with sensorineural hearing loss (SNHL) -- i.e. 0.5 % to 26.7% or more of children with SNHL. Starting with the high end, Liange and others (2001) felt that the prevalence in China was "high". Caldas et al (2012) found 26.7% of 15 children aged 10-12 had "results that are compatible with auditory neuropathy spectrum disorder". Rance (2014) found AN in 5/10 patients with type 1 diabetes (i.e. 50%). Our comment: These are very high estimates and likely wrong.

Moving now to to the lower estimates, Madden et al (2002) found 22/428 children with hearing loss had AN -- e.g. about 5% (Madden et al, 2002). Penido and Isaac (2013) reported only 1.2% in 2292 "individuals with SNHL". Nikolopoulos (2014) in a review suggested that the prevalence is between 0.23 and 2% of infants "at risk" for hearing loss, and that it is responsible for 8% of newly diagnosed cases of hearing loss in children per year. Mittal et al (2012) reported that 5.3% of children in an Indian hospital had ANSD defined by present cochlear microphonics but absent BAER. Our comment: these figures that do not control for age but are likely correct in suggesting a relatively small proportion of hearing loss is due to AN.

Several generalizations -- children seem to have much more AN. Overall, including adults, probably only 1-2% of hearing loss has any evidence for AN. There are some major outliers -- it may be that the criteria for AN vary enough across researchers, that there can be dramatic differences in "prevalence" of AN reported, which are meaningless.

Problems with the definition of AN

Where we practice in Chicago, AN is not a common diagnosis, even though some authors report a prevalence as high as 27% of AN in hearing loss in 10-12 year olds. Why does the prevalence of diagnosis of AN vary so widely ?

The reason may be that the definition of AN (or ANSD) is too loose.

As noted above, there are supposedly two types of AN - -pre and post synaptic. The first type of "AN", "presynaptic", should logically be called something else than "Auditory neuropathy", as the nerve is not the site of lesion. We do not call motor neuron disease such as ALS a neuropathy. Nevertheless, because this syndrome has been defined primarily by test findings rather than anatomic sources of data, the nomenclature has persisted of "AN" for a disorder that does not have an "N". (McMahon et al, 2008)

Another problem with AN case finding is the vulnerability of the definition to diagnostic error. Most aspects of the definition of AN are vulnerable to lack of cooperation or malingering. Ascertainment of hearing thresholds is intrinsically subjective -- if you don't admit to hearing the tones -- you fail the test. Yes there are methods of detecting malingering of course. ABR and ECOG testing requires excellent technique and lack of movement of subjects. Imaging doing this in an active child. It would be very easy to combine a normal OAE (which doesn't require any cooperation),with several tests that are greatly impaired by lack of cooperation (such as ABR), and diagnose AN. This may be the reason that AN is largely diagnosed in small children. If AN is truly a neuropathy, one would think that AN should increase with age as do neuropathies in general. But it doesn't -- in fact children seem to have a lot more AN than adults. How does the AN vanish with age ? Of course, the logical defense to the assertion that AN is an artifact of uncooperative patients is that everybody else is doing bad testing. Perhaps.

A third problem with AN is that it is just implausible that there should be selective nerve damage to the hearing part of the 8th nerve, without damage to the vestibular part as well. Of course, the logical defense to this is that diagnostic methods for vestibular function are missing vestibular nerve damage. Perhaps true.

A fourth problem is the instability in the criteria for AN. What does it mean to have an "abnormal" ABR ? What does it mean to have "detectable" OAE's ? With different investigators setting their criteria for AN differently -- perhaps this is why we have such wide variation.

To summarize this section, there are significant questions about the definition, existence and prevalence of AN. We suspect that it is uncommon -- perhaps 2% of infants with hearing loss. In adults, it is likely exceedingly rare to have a substantial loss of hearing from AN.


The diagnosis of AN is presently made by combining an absent or extremely abnormal ABR test (brainstem auditory evoked response, also called ABR or auditory brainstem response) with normal otoacoustic emissions (OAE), or reasonably intact hearing tests (e.g. absent ABR, present though impaired hearing). Absent acoustic reflexes to sound, but possibly preserved AR to cutaneous stimulation of the face, is also expected in auditory neuropathy (Rance and Starr, 2015). However, nobody does AR to cutaneous stimulation, and this is not a practical clinical criterion. Wave I and the cochlear microphonic can be present in the ABR/ECOG and the patient can still have "AN". What about the other waves in the ABR ? Should they all be absent? What if they are just delayed ? Or some are missing ? How does one define a "normal" OAE. A screener that tests 3 frequencies, or a sweep device that goes from 750 to 16K? It is enough to have OAE's on the screener (that just checks low pitches ?)

Logically, the diagnostic process might not necessarily prove that the site of lesion is the auditory nerve. Gibson and Sanli recently pointed out that a selective loss of inner hair cells might create a similar testing configuration (2007). The outer hair cells (OAE) are preserved, the inner gone. The acoustic neuropathy literature calls this a "presynaptic" auditory neuropathy (Rance and Starr, 2015). In our opinion, this is word-play that violates the definition of "neuropathy" -- basically they are saying that AN might not require a neuropathy. It is good to be inclusive, but not when one is diagnosing diseases. This use of words is similar to saying that ALS (motor neuron disease) was a subtype of peripheral neuropathy. Although they share some symptoms (i.e. weakness), one is due to cell damage, and the other to nerve damage. We don't think that the "presynaptic" AN term should be used.

McMahon and others pointed out that a brainstem disorder could also cause similar testing pattern as that said to be diagnostic of an auditory neuropathy (2008). This seems possible to us, but unlikely in the absence of imaging or clinical findings supporting a brainstem disorder. Of course, in neonates with brain damage associated with a premature or rough delivery, this might be more likely. So in other words, a combination of normal OAE, very abnormal ABR (only a wave 1 or a present cochlear potential on ECOG, and very impaired conventional hearing testing given that it can be performed at all, given that one can believe the results, would seem somewhat (not definitely) likely to identify an auditory nerve disorder. However, this is not a "sure thing". The weakest link is in the subjective nature of the conventional audiogram and the somewhat erratic nature of ECOG/ABR testing in many settings. In other words, an uncooperative patient pretending to have hearing loss (such as a child or a malingering person), combined with normal OAE, and a poor ABR due to movement artifact, might produce an imaginary "auditory neuropathy". Settings that have more uncooperative patients might reasonably have much more AN. This may explain the drastic difference in prevalence of AN across settings.

El-Badry et al (2018) recently reported that cVEMP testing is normal in almost all children with ANSD (auditory neuropathy spectrum disorder) of prelingual onset, and the majority with post-lingual onset had absent cVEMPs. It is difficult toknow what this means. We are impressed that these authors obtained cVEMPs in very young deaf children.

Neuroimaging studies are generally normal in AN -- almost by definition (Wang et al. 2003), but a few have small cochlear nerves or apertures for the nerve in the skull. In the present day where ABR's are no longer routinely obtained in the evaluation of hearing loss (MRI is generally preferred), this diagnosis might be easily overlooked. Diagnostic problems might also occur in persons with mild auditory neuropathy -- one might see a mildly abnormal ABR combined with normal otoacoustic emissions This to us seems to be an ambiguous situation.

In the authors opinion, in adult patients diagnosed with AN, a more generalized peripheral neuropathy should be considered including testing for the more common treatable types of neuropathy (i.e. Diabetes). The examiner should be watching for high arches and pes cavus. These are neurology activities.

Associated disorders and differential diagnosis

Auditory neuropathy has been reported in several hereditary neuropathies.

Auditory neuropathy has also been reported in acquired neuropathies.

Autoimmune inner ear disease is in the differential diagnosis. The extent to which individuals with sudden hearing loss actually have AN, is uncertain. Of course, this could be diagnosed by OAE testing. It is our general experience that patients with SHL have absent OAE's as well.

Auditory neuropathy may also affect vestibular function (Sheykholeslami et al. 2000; Fujikawa and Starr, 2001). When auditory neuropathy is combined with vestibular neuropathy, the diagnosis is made by combining the usual criteria for auditory neuropathy with abnormalities in vestibular function tests. It seems likely that some persons with bilateral vestibular impairment might have this due to neuropathy, and greater use of ABR testing and VEMP testing might be useful in this population. Somewhat contrary to this idea that nerve disease can be detected with a VEMP is the observation that VEMPs are generally normal in vestibular neuritis.

Buetti and Luxon (2014) reviewed vestibular involvement in peripheral neuropathy.

This review is problematic as it is based on a small number of reports, and suggests some rather amazing conjectures. For example, they state that "vestibular dysfunction is a common finding in 50-80% of patient in PN" (PN is peripheral neuropathy). As any general neurologist understands from experience, this is nonsense.

They also review a small number of publications concerning HMSN (hereditory motor and sensory neuropathy). There are a few reports here of absent vestibular function in these genetic neuropathies. With respect to CMT (Charcot Marie Tooth), another genetic neuropathy, there are again a few reports. There are also rare reports in patients with peroneal muscle atrophy (an inherited disorder with diverse mutations).

Concerning Fredreich's ataxia, which is related to the CANVAS syndrome mentioned above, There are reports of reduced vestibular function in a substantial number of patients. This is somewhat reasonable given that FA is thought to cause cerebellar damage through deafferentation.

Concerning CIDP, or chronic inflammatory demyelinating polyneuropathy, again vestibular function has been reported, presumably relating to the propensity for CIDP to affect cranial nerves. In the closely related condition of Guillain-Barre, almost nothing is known about vestibular function.

From this review published in 2014, it is clear that far more work is needed concerning the risk of vestibular impairment in neuropathy.


Treatment in adults should logically be aimed at the underlying peripheral neuropathy, if one is present. As treatment is often not available for peripheral neuropathy, this often translates into no treatment. Nevertheless, there are treatments for Guillain Barre, porphyria, and many other generalized neuropathies (such as from Diabetes). Screening tests should be performed for treatable causes and treatment selected accordingly. One would expect that this will be very rare.

One would expect that treatments aimed at stabilizing nerve function and reducing pain or paresthesias, would make vestibular impairment worse. Thus rationally, treatment that involves suppressing inflammation (e.g. steroids) when present, or blocking autoimmune damage (such as IVIG) would be most practical, when they are available for the disorder in question. For diabetic neuropathy, stricter control of blood sugar would be the rational treatment.

Zhang et al (2014) suggested that steroids were helpful for AN. This is not a common observation however, and of course, steroids are not a practical long term treatment.

One would expect that cochlear implants (CI)would be useless for AN caused by cochlear nerve damage (or absence), but potentially useful for AN caused by cochlear damage. This shows the disservice of the definition of AN that includes both !

A second issue is the "use it or lose it" problem (Cardon et al, 2013). If a child grows up with no hearing, they may not develop the central circuitry to hear (or it may die off). As AN patients are almost all children, perhaps there should be an aggressive attempt to implant very young children. This is a thorny issue, made worse by the fuzziness if the definition of AN or ANSD as you prefer.

Ji et al, reported that only a single one of their 8 implanted subjects attained speech capability (2015). On the other hand, Fernandez et al (2015), from Brazil, reported that An patients had similar response to CI as children with CI. Similarly, Kontorinis et al (2014) reported outcome of CI in ANSD "is favorable" in most of the cases". Quite a large difference ! Our thought is that Ji et al was using tighter criteria for AN. Jeong et al (2013) reported patients with radiologic abnormalities of their cochlear nerve did much worse concerning CI response.

Pelosi et al (2013) from Vanderbilt, reported that patients with CI and with hearing aids did equally well, but a year before (2012), reported that many with ANSD had "limited auditory skills", possibly due to comorbid conditions including cognitive delay. Our thought is that HA cost only $5000 and CI cost about $50,000 -- need we say more ?

Dean et al (2013) reported that ANSD patients did well when implanted at a young age. Our comment -- ANSD is vague.

Budenz et al (2013) reported that "Children with a diagnosis of AN without associated cognitive or developmental disorders have speech and language outcomes comparable to other children who received a CI. "

Jeon et al stated that patients with electrically evoked ABR did better with CI than those without, but that "CI provides at least partial measurable auditory benefit". Our comment -- is the huge cost of CI justified by "at least partial measurable auditory benefit" ? Humphriss et al (2013) reported that current "evidence is weak", and "Stronger evidence is needed to support cost-effective clinical policy and practice in this area."

As a summary of the literature regarding AN and CI, while "hope springs eternal", with tight criteria for AN that include strong evidence for a damaged cochlear nerve (i.e. radiology), CI implantation is unreasonable. This is easy to understand -- if there is no nerve to transmit hearing signals to the brain, a CI is wasted money and effort. If one has a vaguely defined ANSD, that could include inner hair cell damage, one could argue that there might be a reason to give it a try.

Literature review:

Collections of cases (someone needs to do a meta-analysis)

Can et al (2015) reported that 2% of late preterms with hyperbilirubinemia had AN (same as controls).

Doyle, Sininger and Starr (1998) reported 8 pediatric patients having hearing deficits which they attributed to auditory neuropathy. They comment that word discrimination was impaired out of proportion to pure tone performance. Their subject 8 also had Fredreich's ataxia, which is an inherited ataxia which is associated with neuropathy.

Gibson and Sanli (2007) studied 39 children with auditory neuropathy and concluded that the more likely mechanism was selective loss of inner hair cells. In other words, they didn't have a neuropathy at all -- and AN is a misnomer.

Kaga et al (2015) reported 17 adult patients.

Madden and others (2001) described 22 cases of auditory neuropathy from a pediatric otology clinic. 50% had a history of hyperbilirubinemia. 45% had a history of prematurity, 45% ototoxic drug exposure, 36% a family history of hearing loss, and 36% had a history of neonatal ventilator dependence. This would seem to indicate that AN is multifactorial. Cochlear implantation was successful in 4 children.

Maeda reported a single case with CMT (Charcot marie tooth) who was steroid dependent. As CMT is not steroid responsive, this is an unusual observation. One wonders whether this child didn't have two disorders.

Mohammadi et al (2015) reported that many neonates with "auditory neuropathy spectrum disorder" actually had "evidence for cochlear nerve aplasia. " One wonders how this is distinguished from auditory neuropathy.

Prabhu reported a single case after a viral illness (chikungunya).

Song et al (2021) reported on 44 patients (average age of 4.35 years). They reported "Some patients with UAN could show ipsilateral CND and MRI is essential to evaluate if the nerve is deficient. No pathogenic variants that directly related to the pathogenesis of UAN have been found in this study currently." Of the 44 patients, 7 were found to have CND (cochlear nerve deficiency). They diagnosed UAN (unilateral auditory neuropathy) using "Auditory brainstem responses were absent or abnormal in all 44 affected ears, while otoacoustic emissions and/or cochlear microphonics were present."

Starr and others (1996) reported 10 patients, all children or young adults. Cochlear microphonics and otoacoustic emissions were preserved in all, but auditory brainstem response were normal. Auditory brainstem reflexes were also negative. The shape of the pure tone loss varied, being mainly low frequency in 5, flat in 3, and high frequency in 3. Speech was affected out of proportion to that expected from a pure tone loss, which might lead to some confusion with a central hearing loss pattern. subsequently eight of these patients developed evidence of a peripheral neuropathy, which was hereditary in 3. Starr and others (1998) also reported a variant in three children in which transient deafness occurred when the children were febrile. This pattern is reminiscent of the typical exacerbation of neurological symptoms in individuals with demyelinating disease (e.g. MS).

Stein and others (1997) identified  infants who failed hearing screening on ABR, but passed their otoacoustic emission test. Four of five had hyperbilirubinemia.

Genetic studies: (not a big source of AN)

Zong et al (2015) reported a mutation in a family containing variation in a gene termed AIFM1. Tang et al (2015) reported no coherent genetic anomaly in 71 patients including some with AN. Silva et al (2015) reported "There are differences at the molecular level in patients with and without auditory neuropathy."

Lepcha et al (2015) reported that ANS occurred in 1.12% of patients in an Indian ENT clinic. Of these, 60% were "found to have neurological involvement" including cerebral palsy, peripheral neuropathy, spinocerebellar ataxia, hereditary motor-sensory neuropathy, spastic paresis, and ponto-bulbar palsy. Sixty-six percent were born of consanguineous marriages. Comment: this report suggests a very strong genetic component.

Morelet et al (2014) reported a group of 9 Amish persons, who were interrelated, with a mutation in SLITRK6, who had a syndrome with some aspects of AN. They had progressive hearing gloss, were nearsighted, had no DPOAE's (of course AN should have present OAE's), and had poor ABR's.

Runge (2013) reported an otoferlin mutation in two siblings with ANSD.

Bae found two mutations in 19 Korean patients with ANSD.

Matsunaga et al (2012) reported the effect of mutations in OTOF. Interestingly, some of these patients had a "temperature-sensitive auditory neuropathy". This is reminiscent of the situation with the hot bath test in MS.

As an overview, genetic causes of AN or ANSD seem rare.