Introduction

Many of the characteristics of the neural responses elicited by a cochlear implant can be demonstrated using far-field evoked potentials that reflect the activity of different levels of the auditory pathway. The peripheral neural auditory pathway can be investigated in this way through electrically evoked auditory brainstem responses (EABRs), which are early latency electrical potentials recorded from the scalp and form a sequence of peaks and troughs that occur within the first 10 ms after the onset of the electrical stimulus. These responses reflect activity from the eighth nerve and brainstem (Jewett & Williston, 1971; Møller & Jannetta, 1985). They have been used to estimate spiral ganglion cell survival (Simmons & Smith, 1983; Walsh & Leake-Jones, 1982) and evaluate the critical synchronous components of neural encoding at the periphery of the auditory neural pathway. Wave I cannot be seen in cochlear implant recipients since it has a latency of approximatively 0.35 ms (as seen in recordings of the electrically evoked action potential, Abbas et al., 1998; Frank & Norton, 2001) and is embedded in stimulus artifact. While waves IIe and IVe are not systematically present in electrically evoked responses, waves IIIe and Ve can easily be recorded. Many studies have been dedicated to these two waves in animals and humans and after comparison to acoustically evoked ABRs the EABR waves peak latencies have been described as being generally shorter than normal ABRs: 2.1 vs. 3.39 ms for wave IIIe and 4.0 vs. 5.5 ms for wave Ve (Gyo & Yanagihara, 1980; Starr & Brackmann, 1979). Moreover, Waring’s studies (1992, 1995) showed no significant difference for the interpeak interval IIIe-Ve (i.e., approximately 1.9 ms), while Gardi (1985), Firszt et al. (2002), and Gordon et al. (2006) found that interpeak intervals IIIe-Ve are shorter intervals than those that are typical for the acoustic ABR (i.e., 1 ms). Shorter latencies in implantees may be due to the fact that the delay in the traveling acoustic wave as well as the transduction delay are bypassed and that there may be better synchronization of neurons. Generators of waves IIIe and Ve are nevertheless probably respectively located in the ipsilateral cochlear nucleus and the inferior colliculus as for the corresponding data from acoustical stimulation (e.g., van den Honert & Stypulkowski, 1986).

The behaviour of EABR waves IIIe and Ve latencies with respect to stimulation site in the cochlea has been investigated both in the normal and the implanted ear. For normal hearing subjects, latencies of waves III and V were found to increase for low frequency stimuli (Gorga et al., 1988). In the cochlear implant recipient, wave IIIe has been much less studied than wave Ve and no significant effect of stimulation site could be found on its latency as Firszt et al. (2002) reported only a tendency to increase for the apical electrodes. However, several studies examined the relationship between site of stimulation and latency of wave Ve. According to van den Honert & Stypulkowski (1986) and Abbas & Brown (1988), there is no significant difference in wave Ve latencies when various parts of the cochlea are stimulated in cochlear implant users. However, other studies show that the electrically stimulated fibers of the eighth nerve do not respond with the same latency. Nagel (1974) stimulated each turn of the guinea pig cochlea with bipolar electrodes and reported longer response latencies with more apical stimulation. Miller et al. (1993) also reported an increasing baso-apical gradient in the guinea pig. In addition, a gradient of EABR latencies was reported in human ears stimulated by longitudinal electrode arrays (Hermann & Thornton, 1990): longer wave Ve latencies were recorded with more apical stimulation within the first turn. Conversely, other studies showed that apical electrodes had shorter wave Ve latencies than basal electrodes in humans (Shallop et al., 1990; Abbas & Brown, 1991; Miller et al., 1993) by approximately 0.4 ms (Allum et al., 1990; Firszt et al., 2002). These controversial findings may result from differences across the studies in the way the subjects were stimulated and the potentials recorded. Various types of cochlear implant devices were also used and this could infer some variability should the distance from the modiolus have an effect on EABR latencies. As the length of the electrode arrays varies across the cochlear implant types, different areas of the cochlea may also have been activated from one study to another. It should be stressed as well that only a few contacts on the electrode array were considered and that those electrodes differed across the studies. Moreover, all subjects were not included in the statistical analysis in all studies (see Miller et al., 1993). Hence since data is missing on this topic for both waves IIIe and Ve, in the present study EABRs were recorded from all stimulation sites in a group of subjects implanted with the same electrode array.

Since different EABR latencies may reflect the selective excitation of normal nerve fiber subpopulations with different response latencies, different synaptic delays or various lengths, characterizing the relationship between EABR latency and cochlea site will give an insight about the influence of the auditory pathway anatomy on EABR latency. EABR latencies may also be influenced by deafness. Elongated EABR latencies may indeed indicate pathologies of the ascending pathway as it has been shown that poorer speech recognition scores in implant users are correlated with longer latency of wave Ve (Hermann & Thornton, 1990, Gallego et al., 1998). Thai-Van et al. (2002) also showed differences in EABR latency for subjects implanted bilaterally and with asymmetric deafness: the ear with the longest duration of deafness showing the longest EABR latency. In order to study whether the quality of the auditory fibers at a specific site in the cochlea can have an impact on EABR latency, relationships with M levels (most comfortable levels) at first fitting of the cochlear implant were investigated. Levels and dynamic range are indeed closely dependent on subjects’ aetiology and spiral ganglion neuron population (Pfingst et al., 1980, 1984; Shannon, 1983; Lusted et al., 1984). The use of M levels was motivated by the fact that EABR recording is performed with levels perceived as being relatively high in our study as well as in clinical conditions. This may result in broad stimulation levels, not as specific to cochlear place, and using the same intensity range would allow approximately the same anatomical areas to be addressed in both subjective and objective parts of the study. Besides, as only M levels are measured for fitting the type of cochlear implant that is used in this study (HiRes90K® from Advanced Bionics Corporation), investigation of M levels rather than threshold levels or upper limit of comfortable levels seems also more interesting from a clinical point of view. In addition, since auditory performance is strongly related to duration of deafness (Blamey et al., 1996) and audiometric results (Brimacombe & Eisenberg, 1984; Wable et al., 2001), relationships of EABR latency with duration of deafness and audiometric tone threshold before implantation were also studied to investigate the possible influence of deafness on EABR latencies. Understanding the influence of anatomical versus pathological factors on them will allow a better interpretation of response latency in clinical use. Instances of use could be assessment of neural integrity, implant function, placement of intracochlear electrode arrays, and programming level. This is also important for understanding the coding effectiveness of cochlear implants.