Auditory neuropathies and electrocochleography

Auditory neuropathy (AN) is a hearing disorder characterized by disruption of temporal coding of acoustic signals in audi- tory nerve ﬁ bres resulting in impairment of auditory perceptions relying on temporal cues (1,2). Mechanisms suggested include both pre-synaptic and post-synaptic disorders affecting inner hair cell (IHC) depolarization, neurotransmitter release from ribbon synapses, spike initiation in auditory nerve terminals, loss of nerve ﬁ bres and impaired conduction, all occurring in the presence of normal physiological measures of outer hair cell (OHC) activities (otoacoustic emissions (OAEs) and cochlear microphonic (CM)). Disordered synchrony of auditory nerve discharge has been suggested as the basis of both the profound alterations of auditory brainstem responses (ABRs) and impairment of speech perception (3). We review how electrocochleography (ECochG) can provide detailed information to help deﬁ ne objectively the sites of auditory neural dysfunction as affecting IHC receptor summating potential (SP) and compound action potential (CAP), the latter reﬂ ecting disorders of ribbon synapsis and auditory nerve ﬁ bres.


Differential diagnosis of AN: neural and inner hair cell (IHCs) disorders
The ECochG response results from the superimposition of three components (4,5), two originating from receptor elements, the cochlear microphonic (CM) and summating potential SP (SP), and the other, the compound action potential (CAP), arising from auditory nerve fi bres ( Figure 1). CM potentials are believed to originate mainly from the sum of the extracellular components of receptor potentials arising from OHCs located in the basal portion of the cochlea (4). SP responses recorded from the promontory originate from the DC component of receptor potentials arising in the basal IHCs (6).
In ECochG recordings evoked by click stimuli, CM appears as an oscillatory activity superimposed on the SP-CAP complex (4,5) (Figure 1). Since this activity is strictly related to the basilar membrane motion, the procedure of averaging the responses evoked separately by condensation and rarefaction stimuli is usually used to extract both SP and CAP (4,5,7). In normal hearing ears the ECochG waveform resulting from CM cancellation ( Figure 1) begins with an abrupt negative defl ection, the receptor summating potential (SP); this is followed by a negative peak, the neural compound action potential (CAP), which returns to baseline by 1.5 -2 ms at high stimulus intensities. Both SP and CAP potentials are reduced in amplitude and delayed in latency at decreasing stimulation inte n sities, while the duration of the whole SP-CAP complex is slightly prolonged (4,5).
Auditory neuropathy (AN) has a wide range of aetiologies (genetic, infectious, toxic-metabolic, immunological) (1,8). Currently, defi nite aetiological factors can be identifi ed in about half of patients (1). All forms of the disorder may be present in isolation or associated with multisystem involvement including peripheral and/or optic neuropathies (3,9). AN disorders associated with neurological diseases are invariably post-synaptic since alterations of auditory nerve fi bres discharge result directly from neuronal loss and demyelination of peripheral and/ or cranial nerves. ECochG recordings performed in these AN disorders show both CM and SP receptor potentials of normal amplitude consistent with preservation of cochlear function, while the neural CAP is decreased in amplitude and prolonged in duration compared to normal hearing ears (7). This ECochG pattern should be considered as distinctive of ' neural ' AN. These forms are rarely seen in small children because in most patients the neuropathies affecting cranial or peripheral nerves begin in late childhood or adolescence (9).
In several forms of AN underlain by different aetiologies (genetic, immunological, metabolic) the cochlear potentials obtained after CM cancellation consist of a prolonged negative defl ection that is attenuated in amplitude and markedly prolonged in duration compared to the CAP recorded from normally-hearing ears (7). One example is displayed in Figure 2 for one AN patient carrying the R445H mutation in the OPA1 gene (10). ECochG waveforms are superimposed on the ECochG recordings obtained from one normally-hearing child at decreasing stimulation intensities. The cochlear potentials recorded from the OPA1 patient consist of a prolonged negative defl ection showing no separation between SP and CAP components. Compared to the normal ear, the amplitude of this response is reduced, the peak latency is delayed, while the duration appears remarkably increased.
In order to clarify whether the prolonged potentials originate from neural or receptor activations, we used a neural adaptation paradigm consisting of an initial click, followed 15 ms later by 10 clicks with an inter-stimulus interval of 2.9 ms. This sequence was repeated every 191 ms (5,7).
The potentials recorded in response to this stimulus sequence are displayed for 110dB SPL intensity . ECochG potentials from one patient with AN. ECochG waveforms recorded from one patient carrying the R445H mutation in the OPA1 gene are superimposed on the corresponding recordings obtained from one normal-hearing child at decreasing stimulus intensities. In the normal ear, the response begins with negative SP followed by the CAP, which increases in peak latency and decreases in amplitude from 120 to 60dB SPL. The ECochG responses recorded from the AN patient consist of a low-amplitude negative potential showing increased peak latency and prolonged duration compared to the normal ear. in Figure 3 for one normally-hearing child (as in Figure 2) and the AN patient carrying the R445H mutation in the OPA1 gene. In the normally-hearing ear CAP amplitude shows a marked attenuation from the fi rst (#1) to the second (#2) click of the stimulus sequence. Subsequently, further decrease in amplitude is observed during the following 3 -4 clicks during train stimulation. As a result, CAP amplitude, as measured from baseline to the CAP peak, decreases on average to 53% at the end of the stimulus sequence (mean value from 34 normal ears 43%, range 25 -58%). Also SP amplitude attenuates at high stimulus rate, but the amount of amplitude reduction was much smaller (78%; mean value from 34 normal ears 71%, range 45 -100%).
In the OPA1 patient, amplitude changes during adaptation were only measured over the whole SP-CAP response (5,7) due to the lack of separation between SP and CAP components. SP-CAP potentials decrease in amplitude from the initial click of the sequence (click #1) to the fi rst click of the train (#2) and show further attenuation during the following 4 -5 clicks. As a result, the size of the SP-CAP potential after adaptation (50%) equals the amplitude of the adapted SP-CAP response in normal ears (mean value from 34 normal ears 43%, range 25 -58%). These fi ndings indicate that the prolonged negative defl ections recorded from this particular category of AN patient result from neural rather than receptor activation. Different mechanisms may underlie the generation of the prolonged negative potentials in the various forms of AN. In subjects with neural AN such as those affected by Charcot-Marie-Tooth disease, demyelination of auditory nerve fi bres is expected to induce slowed conduction velocity with reduced probability of synchronous fi ring, while the decrease in number of auditory nerve fi bres would result in reduced auditory input. This may lead to prolonged duration and decreased amplitude of the neural CAP potential (5). Other mechanisms may be involved in the forms of AN showing no separation between SP and CAP components. In these AN disorders the reduction or absence of SPs in ECochG recordings point to a pre-synaptic disorder of IHCs, which can occur in the presence of preserved function of OHCs as indicated by the detection of CMs with normal amplitude. Therefore, information provided by ECochG recordings are relevant in that they help to identify the lesion site by distinguishing the AN disorders involving IHCs from the neural disorders, in which SPs are normal. This information may be of value in predicting the outcome of cochlear implantation. A good outcome is invariably expected in implanted patients affected by AN disorders involving IHCs, whereas the benefi ts of cochlear implantation in neural forms are strictly dependent on the number of spared auditory neurons. In the normal-hearing child, CAP amplitude was markedly attenuated after adaptation whereas SP attenuation was much smaller.
In the three AN subjects, SP-CAP amplitude was markedly reduced after adaptation, and the amount of response attenuation was comparable with that calculated for the normal-hearing child.

Cochlear potentials from children with mutations in the otoferlin gene: a ribbon synaptic disorder
Cochlear potentials have been recorded at our department from children with biallelic mutations in the OTOF gene as part of the standard assessment protocol for cochlear implantation (11). Mutations in the OTOF gene (DFNB9) with a recessive pattern of inheritance are responsible for the most well-known form of isolated [12][13][14][15]. This is considered a pre-synaptic disorder with preservation of auditory nerve fi bres. The OTOF gene encodes otoferlin, a transmembrane protein belonging to the ferlin protein family, which contains several repeating C2 domains involved in calcium binding (16). Otoferlin plays a crucial role in vesicle release at the synapse between IHCs and auditory nerve fi bres by interacting with syntaxin1 and SNAP25 (17), and it has recently been implicated in vesicle replenishment at the pre-synaptic membrane (18). It has been estimated that OTOF mutations are responsible for at least 3% of cases of prelingual non-syndromic hearing impairment in the Spanish population (12,14). To date, more than 50 pathogenic mutations of the OTOF gene have been identifi ed (14). Most of them cause loss of otoferlin function, thus resulting in a very homogeneous phenotype of prelingual, profound hearing loss with absence, or marked threshold elevation, of ABRs. Over 50% of subjects carrying biallelic mutations in the OTOF gene also show preserved function of OHCs as indicated by OAEs recording (11,12,14). This is matter of concern when using OAEs for detecting hearing impairment in universal screening programmes since children with profound hearing loss underlain by mutations in the OTOF gene may be missed by the screening procedure.
Mutations that do not fully inactivate the otoferlin protein result in a milder phenotype characterized by the classical picture of AN, which is apparent only in febrile subjects (19 -21). This has been attributed to an instability of otoferlin, which develops when the body temperature rises (15).
In children with mutations in the OTOF gene, CMs are recorded with normal amplitude. The cochlear potentials obtained after CM cancellation show the prolonged negative defl ection found in other forms of AN (11). The ECochG waveforms recorded from fi ve children at 120dB SPL (corresponding to 90dB nHL, referred to the psychoacoustical threshold of normal hearing subjects) are reported in Figure 4 (left side) superimposed on the grand average of the ECochG recordings obtained from 26 normally-hearing children. In patients with OTOF mutations the ECochG response begins with a fast negative defl ection with peak amplitude and latency consistent with SP. This is followed by a lowamplitude negative potential with markedly prolonged duration compared to the neural responses recorded from the normal hearing ear. Similar ECochG waveforms have been obtained by McMahon et al. (22) from a group of young children with AN. In some patients a small CAP is superimposed on this prolonged activity. Interestingly, the sustained negative potentials are recorded as low as 50 -90dB below behavioural thresholds ( Figure 4, right side).  . ECochG potentials from children with biallelic mutation in the OTOF gene. Cochlear potentials recorded from fi ve children at 120dB SPL are reported on the right side superimposed on the grand average of the corresponding waveforms obtained from 26 normalhearing ears. The shadowed area refers to 95% confi dence limits. ECochG responses recorded from one representative child are displayed on the left side for decreasing stimulus levels. At high stimulus intensity ECochG responses begin with a fast negative defl ection with peak amplitude and latency consistent with SP. This is followed by a low-amplitude negative potential with prolonged duration compared to the neural responses recorded from normal ears. This response is identifi able as low as 60dB SPL, which is about 50dB lower than behavioural threshold.
Moreover, stimulation at a high rate (Figure 3) induces an attenuation in amplitude of SP-CAP potentials (53% in the example of Figure 3), which is similar to that obtained for SP-CAP in normal-hearing ears (mean value from 34 normal ears 43%, range 25 -58%) consistent with their neural generation.
The initial fast SP response recorded from children with mutations in the OTOF gene points to a normal activation of IHCs. This is followed by the prolonged neural potentials that seem analogous to the dendritic responses recorded by Sellick et al. (23) from the scala tympani of guinea pigs after blocking neural spiking in terminal dendrites of auditory nerve fi bres. Similarly, the prolonged responses recorded from children with OTOF mutations may result from the sum of small dendritic potentials arising in the distal portion of the afferent fi bres that are not followed by neural spiking. Abnormal dendritic activation, in turn, results from alteration of multivesicular release. A reduction in otoferlin activity may lead to reduced vesicle replenishment (18) and to abolition of the fast phase of exocytosis (17). The decrease of neurotransmitter release is likely to result in a decrease in amplitude of EPSPs with reduced synaptic reliability. Moreover, besides the reduction of neurotransmitter availability at the synaptic cleft, there is an impairment of multivesicular release leading to the generation of small EPSPs with abnormal morphology and dispersed in time, which only occasionally reach the threshold to trigger action potentials recorded as high-threshold CAP.
These mechanisms are relevant from the point of view of rehabilitation programmes for deaf children carrying mutations in the OTOF gene, since auditory nerve fi bres are not believed to be directly involved by the disorder. This might explain the excellent outcome of cochlear implantation in restoring auditory function in this category of patients (24,25).

Cochlear potentials from AN children discharged from the Neonatal Intensive Care Unit (NICU): a neural or inner hair cells (IHCs) disorder?
Infants admitted to the Neonatal Intensive Care Unit (NICU) are at increasing risk for sensorineural hearing loss due to exposure to many risk factors such as prematurity, low birth weight, respiratory distress, anaemia, noise and ototoxic drugs (26,27). The incidence of hearing loss ranges from 1.5% (28) to 3.2% (27). This variability is due to differences in the characteristics of the NICU population among hospitals, age at testing and variability in defi ning and assessing hearing loss (26). In these children, the combined effects of prematurity and other risk factors impinge on cochlear function, resulting in various types and degree of damage (26). Variable degrees of sensorineural hearing loss have been reported in infants discharged from the NICU resulting from lesion of OHCs, IHCs or both.
About 5.6% of children failing newborn hearing screening show abnormal ABRs and presence of OAEs consistent with the electrophysiological profi le of AN (26,28). Post mortem examination carried out on temporal bones of deceased neonates has shown extensive hair cell damage with a higher frequency of selective IHC loss in premature infants compared to full-term babies (29). Since no signs of auditory nerve damage were found, the dysfunction of IHCs has been proposed as the primary mechanism underlying AN in children discharged from the NICU.
In cooperative subjects clinical criteria for AN diagnosis include impairment of speech perception beyond that expected for the hearing loss, absence or marked abnormality of ABRs and preserved OHCs activities (OAEs and CMs). In newborn babies, the presence of AN relies only on the absence of ABRs associated with the detection of OAEs (26). However, no detailed information of cochlear nerve and hair cell activities could be provided by an ABR recording due to the low signal-to-noise ratio (7,24). In addition, in children discharged from the NICU, ABR abnormalities may result also from reduced synchrony occurring at any level from auditory nerve to brainstem generators. A reduced correlation between ABR and hearing thresholds has been documented in premature babies and in those suffering perinatal asphyxia (30 -32). In these cases, alterations of ABRs might ensue from abnormal fi ring of brainstem generators, which retain little or no connection with the dynamics of auditory periphery activation. Recently, the use of intratympanic electrocochleography (ECochG) has been proposed for this category of patients to defi ne the details of cochlear potentials (7,22) and, in general, to evaluate hearing threshold for rehabilitative purposes.
Twenty-three children from the NICU who showed the electrophysiological profi le of AN (mean age 23.6 months) have been submitted to ECochG recording at our department from 2001 to 2012 because of reduced ability to perform in behavioural audiometry (5,24). All children showed one or more disabilities including cerebral palsy, delayed motor development and cognitive impairment. The cochlear potentials recorded by ECochG at decreasing stimulus levels are displayed for one representative patient in Figure 5, superimposed on the corresponding waveforms recorded from one normal hearing child. The ECochG profi le obtained from children discharged from the NICU is indistinguishable from the most common pattern observed in patients with AN. Specifi cally, the response consists of an abrupt negative potential with the same latency as the SP in normal ears. This is followed by the prolonged negative response with reduced amplitude and delayed peak latency compared to the normal-hearing subjects. In this particular child, the amplitude of the SP potential is reduced compared to the normal ear. This result appears to be in accordance with the fi nding of a reduced number of IHCs in children discharged from the NICU who showed the electrophysiological profi le of AN (29). In this perspective, the auditory dysfunction found in these children should be related to an abnormal activation of IHCs. Nevertheless, more extensive analyses of ECochG recordings are needed to draw any defi nite conclusion.
Stimulation at high rate ( Figure 3) is consistent with the prolonged potentials being generated by neural rather than receptor elements since the size of attenuation after adaptation was 60% and, thus, it was comparable to that obtained from normal ears (mean value from 34 normal ears 43%, range 25 -58%) (5).
The ECochG fi ndings collected from children discharged from the NICU point to an abnormal activation of auditory periphery. Specifi cally, the fi nding of prolonged potentials similar to those obtained in other forms AN is consistent with alterations of auditory nerve discharge. Whether this abnormal pattern of activation is entirely related to a presynaptic disorder associated with IHCs loss or, rather, to a combination of mechanisms acting at both pre-and post-synaptic levels, remains to be determined.
Surprisingly, the low-amplitude prolonged negative potentials were recorded also from children (59 subjects) discharged from the NICU who showed abnormal ABRs and no OAEs (5). Specifi cally, 61% of these children showed prolonged responses with no separation between SP and CAP; 13% had distinct SP and CAP associated with an increase of response duration; while 25% showed no response at the maximum stimulation intensity (120dB SPL). Overall, the ECochG pattern observed in the majority of these children seems indistinguishable from that obtained from the group of infants with the electrophysiological profi le of AN. Figure 6 reports the distribution of SP-CAP thresholds in the two samples of children discharged from the NICU, one with OAEs (AN) and the other without OAEs (N-AN). It can be noted that patients showing the profi le of AN show normal or moderately elevated ECochG thresholds, whereas higher thresholds were found in children showing no OAEs. Threshold differences between the two groups of children may refl ect different mechanisms and extension of damage at the level of the auditory periphery. Taking into account the high variability of risk factors acting in the NICU, it is likely that ECochG threshold distribution result from a ' spectrum ' of lesions consisting of different amounts of IHCs and OHCs loss, synaptic damage and depression of auditory nerve fi bre activity. In this perspective, what really matters is not the hallmark of AN (absence of ABRs/presence of OAEs), but rather what ECochG recordings tells us about the receptor potentials and neural response. Abnormal SPs followed by prolonged potentials are likely to be associated with IHCs disorders, whereas normal SP responses followed by the sustained potentials may result from synaptic dysfunction. However, the latter ECochG pattern could be underlain by alternative mechanisms such as damage to the terminal dendrites as suggested for some forms of AN in humans (10) as well as for noise-induced trauma in guinea pigs (33).  Figure 5. ECochG potentials from one AN child discharged from the NICU. The waveforms recorded at decreasing stimulus levels are superimposed on the corresponding recordings obtained from one normal-hearing child. Cochlear potentials recorded from the child discharged from the NICU consist of an abrupt negative potential showing the same latency as the SP in normal ears but with reduced amplitude. This is followed by the prolonged negative response similar to that found in other forms of AN.

Conclusions
ECochG recordings collected from patients with AN can reveal abnormalities of both IHCs receptor potentials, and activation of auditory nerve fi bres. This fi nding may be crucial to defi ne the diagnosis in some forms of AN since OAEs are absent in over one-third of AN patients on retesting (1). Moreover, impairment of speech perception may be apparent only in the presence of noise in cooperative subjects (1), while it cannot be evaluated in newborn babies. In these cases the fi nding of abnormal cochlear potentials may help to refi ne the diagnosis of AN by identifying the site of lesion and by elucidating the underlying pathophysiological mechanisms. Moreover, ECochG recordings are mandatory in children discharged from the NICU who show abnormal ABRs and reduced ability to perform at behavioural audiometry since no reliable information on the function of auditory periphery can be obtained through ABR recordings.