Introduction
Despite the success of cochlear implants (CIs) in restoring hearing to more than half a million people worldwide, auditory perception by CI listeners suffers from fundamental limitations in spatial selectivity and in temporal processing compared to normal hearing (NH). Both of these can be revealed using psychophysical and physiological experiments in which simple stimuli are presented to one or more CI electrodes.
Limitations in spatial selectivity are reflected in the broad spread of neural excitation resulting from stimulation of a single electrode, as measured both using psychophysical techniques in humans and neural recordings in animals (Shannon
1983; Middlebrooks and Bierer
2002; Snyder et al.
2004; Carlyon et al.
2017). The processing of temporal fine structure (TFS) is also limited. When a single electrode is stimulated at moderate rates, pitch increases with increasing rate but the minimum detectable rate difference is usually substantially larger than for pure tones in NH (Moore and Carlyon
2005). As rate increases further, pitch no longer increases once the rate exceeds an upper limit, which varies between about 200–700 pps depending on the listener and on the electrode stimulated (Townshend et al.
1987; Kong and Carlyon
2009; Carlyon et al.
submitted). When bandpass filtered harmonic complexes, designed to minimise place of excitation cues to pitch, are presented to normal hearing listeners the highest upper limit observed is approximately 700 pps (Macherey and Carlyon
2014). This is consistent with the upper limit in the “best” CI listeners being approximately equal to that obtained in NH, but with many CI subjects showing a much lower limit.
A possible neural correlate of the upper limit has been observed in single-neuron recordings in the inferior colliculus (IC) of anaesthetized cats, which phase lock to electrical pulse trains up to a certain rate beyond which they exhibit only an onset response (Snyder et al.
1995; Vollmer et al.
2005; Middlebrooks
2008; Middlebrooks and Snyder
2010; Hancock et al.
2013; Vollmer et al.
2017). Although it is not known whether the limitation arises at or before the IC, there is evidence from humans that the limitations on TFS processing arise centrally to the auditory nerve (AN). We have measured the electrically evoked compound action potential (ECAP) and pulse-rate discrimination in the same subjects, and found good encoding of pulse rate in the ECAPs even at rates where behavioural discrimination was at chance (Carlyon and Deeks
2015).
There is direct evidence that the physiological upper limit of temporal processing is reduced by auditory deprivation. For example, juvenile deafened cats show a higher limit when they have grown up listening through a CI than when they have grown up deaf (Hancock et al.
2013; Vollmer et al.
2017). Indirect evidence from humans, consistent with an effect of auditory deprivation and chronic stimulation on the psychophysical upper limit, comes from the finding in one study that it correlates negatively with the duration of deafness (Cosentino et al.
2016). There is also some evidence, discussed in “
Efficacy” under the “
Discussion” section, that in both cats and humans, the upper limit following auditory deprivation can be increased by a period of chronic stimulation (Vollmer et al.
2005; Carlyon et al.
submitted).
Fine temporal processing in the auditory system relies on the ability of neurons, at and central to the AN, to fire in a sustained and temporally accurate fashion at high stimulus repetition rates (Song et al.
2005). This firing property is dependent on the expression of Kv3 high-voltage-activated potassium channels. Kv3 channels are activated by depolarization of the plasma membrane to potentials above − 20 mV; they open rapidly during the depolarising phase of the action potential in order to initiate repolarisation and prevent significant sodium channel inactivation. As the neuron begins to repolarise, the channels deactivate quickly and thus do not contribute significantly to the after-hyperpolarisation (Rudy et al.
1999; Rudy and McBain
2001). As a consequence, neurons expressing Kv3 channels are able to sustain action potential firing at high frequencies. Kv3.1 and Kv3.3 channel subtypes are expressed in fast spiking neurons throughout the auditory brainstem (Grigg et al.
2000; Li et al.
2001). Loss of Kv3.1 channel expression in the auditory brainstem is associated with ageing (Jung et al.
2005; Zettel et al.
2007) and with auditory deprivation arising from hearing impairment (von Hehn et al.
2004).
AUT00063 is a novel small-molecule drug that selectively enhances Kv3 channel function. In vitro electrophysiology studies with recombinant human Kv3.1 channels expressed in mammalian cells have shown that AUT00063 can increase the amplitude of hKv3.1-mediated potassium currents with a pEC50’s of 5.1 ± 0.17 (Anderson et al.
2018). In addition, two studies showed that AUT00063 reduces the elevation in spontaneous firing rate that results from noise exposure, both in the dorsal cochlear nucleus (“DCN”: Glait et al.
2018) and IC (Anderson et al.
2018). Those studies also showed that AUT00063 can increase neural thresholds (Glait et al.
2018) and reduce driven rates (Anderson et al.
2018) to acoustic stimulation. More importantly for the present study, it has been shown that AUT00063 can improve fine temporal coding in the auditory brainstem in mice. Chambers et al. (
2017) exploited their previous finding (Chambers et al.
2016) that Oubain administration, which killed 95 % of AN type 1 neurons, degraded the temporal representation of trains of acoustic chirps in the IC and auditory cortex, and showed that this degradation could be partially reversed by AUT00063, in vivo. Specifically, they demonstrated that AUT00063 increased the precision of phase locking in the IC, particularly at pulse rates faster than about 40 Hz, and improved the accuracy of a classifier that was trained to decode pulse rate from the responses of cells in the IC or auditory cortex. They also showed that, in vitro, AUT00063 reduced the width and increased the precision of action potentials recorded from fusiform neurons of the DCN, which provide a principal input to the IC. Further evidence for the effect of AUT00063 on temporal coding comes from a preliminary report showing that, whereas aged rats exhibit higher gap detection thresholds than younger rats, this elevation can be partially reversed by AUT00063 (Rybalko et al.
2014).
Two previous clinical trials of the effect of AUT00063 on acoustic hearing confirmed the safety of the drug at doses up to 800 mg/day, but revealed no significant effect on either tinnitus or on speech perception in people with hearing loss in older age (Autifony Therapeutics
2014,
2017b). However, acoustic studies of the effects of auditory deprivation, and its possible amelioration by a pharmaceutical agent, are limited by the fact that one can only test patients who have some useable residual hearing. This necessarily excludes those patients who will have experienced most deprivation, namely those who are profoundly deaf. CIs provide an almost unique opportunity to study such patients. Accordingly, Autifony Therapeutics Ltd., who are the inventors of AUT00063, decided to test its effects on hearing among profoundly deaf patients whose hearing was restored by a CI. The initial design of this “QuicK
+fire” trial tested speech and music perception using stimuli presented via the patients’ clinical processor, and the results of those investigations are described elsewhere (Sanchez et al.
2018). However, as the clinical processors typically remove TFS, and because AUT00063 has been shown to restore the processing of fine temporal information, we decided to evaluate it using direct-stimulation psychophysical experiments that were designed to measure temporal processing by CI users. To do so, we used tests that were well-established in our laboratory and that we had shown to be sufficiently sensitive to modest effects of chronic stimulation and/or stimulus level (Carlyon et al.
submitted). The rationale was to maximise the possibility of observing a significant effect by using methods that measure the processing that the drug was designed to improve. If—as turned out to be the case—no significant benefits were found, one could exclude the explanations that either the CI processor removed the appropriate (TFS) information, and/or that the tests were not sufficiently sensitive to reveal a significant effect.