This was a retrospective study which included adult subjects with valid pure-tone thresholds before CI implantation. Preoperative audiometric and speech measurements were acquired within the CI candidacy screening procedure. Objective T-ECAP measurements were conducted intraoperatively during CI implantation.

All subjects agreed to the use of their data in this study as part of a general declaration of consent to clinical research (Ethics Committee approval no. 162_17 Bc Erlangen).

Participants were native German speakers and at least 18 years of age at the time of implantation. Onset of severe to profound hearing loss had to be postlingual. All subjects received a Slim Modiolar electrode CI632 (Cochlear Ltd., Sydney, NSW, Australia). Additional inclusion criteria were: fully inserted electrodes with complete scala tympani (ST) positioning. Exclusion criteria were: tip foldover of the electrode; reimplantation or reinsertion of the electrode; abnormalities of the cochlea or of the auditory nerve; cognitive disorders which prevent undergoing audiometric assessment.

The subject group consisted of 42 females and 42 males; their age at time of implantation ranged from 25 to 85 years, with a mean (±standard deviation, SD) of 65 ± 13 years. Of the 84 subjects, 38 received their CI on the left and 46 on the right side. Mean ± SD duration of deafness was 12 ± 13 (range: 1–59) years. Demographic data are shown in Table 1.

Subjects underwent an audiological test battery during the CI assessment. The unaided pure-tone audiometric threshold for octave-spaced frequencies from 125 Hz to 8 kHz were obtained by using air-conduction headphones (DT48; beyerdynamic, Heilbronn, Germany) and the AT 1000 audiometer (Auritec Medizindiagnostische Systeme GmbH, Hamburg, Germany) which was calibrated according to the DIN EN 60318 standard. The contralateral ear was masked appropriately. The pure-tone average threshold (4FPTA) was calculated from frequency thresholds: 0.5, 1, 2, and 4 kHz. If a pure tone was “not heard” at the maximum output level of the audiometer, the subject’s hearing loss was estimated as the maximum output level plus 10 dB.

Word recognition score (WRS) for phonemically balanced monosyllabic words (DIN 45621-1:1995-08, 1995) under unaided conditions was recorded. The headphone presentation level was increased until the maximum score of 100% was reached. Otherwise the maximum score achievable (WRS_max) that was still below the patient’s loudness level of discomfort was noted (Hoppe et al., 2019).

All CI surgical procedures were performed by the same experienced surgeon according to the manufacturer’s guidelines (Cochlear Ltd, 2020) by the soft-surgery technique. The implantation was carried out by round-window insertion (N = 77) or round-window enlargement (N = 7). ECAP data were recorded intraoperatively after insertion and closure of the round window opening.

Plain X-ray imaging was used intraoperatively to ensure that no electrodes were inserted incorrectly (e.g., with tip foldover), and digital volume tomography (DVT) was performed within the 2 days after surgery in order to verify in detail that the intracochlear electrode positioning was correct.

Postoperative DVT scans were analyzed in “cochlear view” (Cohen et al., 1996; Xu et al., 2000) with regard to scalar positioning of electrodes, cochlear size (diameter A), insertion depth angle for the most apical electrode E22 (θ_apical) and most basal electrode E1 (θ_basal). Diameter A was measured as the distance between the round window and the outer wall passing through the modiolus, while the insertion depth angles were calculated relative to the reference line drawn between the round window and the modiolus (Figure 2). As described by Aschendorff et al. (2017) the medial-lateral position (MP) can be calculated as the ratio of the active electrode length (L_AE) according to the manufacturer’s specifications (CI632: 13.4 mm, Cochlear Ltd., Macquarie University, Australia) and the length of the outer wall (L_LW) for the matching angular insertion depth according to Escudé et al. (2006).

The MP describes the average proximity of the electrode contacts to the modiolus. A large MP (up to 1) corresponds to an electrode array that is placed laterally (comparable to straight electrodes), while a small MP describes an electrode array that is very close to the modiolus.

In order to analyze the ECAP responses in relation to the actual cochlear place pitch, the frequency allocation for each electrode’s insertion angle was calculated.

The remaining 20 insertion depth angles between θ_apical and θ_basal were estimated by linear interpolation for each subject. For each electrode contact, the spiral ganglion frequency was determined according to the findings of Stakhovskaya et al. (2007). Thus, the angular location of the electrode contact corresponds to the tonotopic map of the spiral ganglion cells and to the tonotopic coding of the cochlea. The spiral ganglion map was used, because we investigated only perimodiolar electrode arrays, i.e., those with close proximity to the modiolus and thus to the spiral ganglion cells (Rebscher et al., 2008; Li et al., 2021).

Intraoperatively, the Custom Sound EP software with its AutoNRT algorithm (Van Dijk et al., 2007) was used to measure T-ECAPs at all 22 electrode contacts. Stimulation and recording parameter settings were kept at default settings (stimulation rate, 250 Hz; stimulus and masker pulse width, 25 μs; stimulus-masker gap, 400 μs; sweeps, 35; gain, 50 dB). The starting level was set to 170 current units (CU). The algorithm uses the forward masking paradigm (Brown et al., 1990) in order to reduce the stimulus artifact. In general, AutoNRT increases the biphasic stimulation pulses in 6 CU steps until two reliable responses (N₁ and P₁) in a row were found; subsequent the stimulation levels decrease in 3 CU steps until a nonresponse is found. Then, the T-ECAP is estimated as the mean from the lowest recorded response and the highest recorded nonresponse (Van Dijk et al., 2007). The T-ECAP measurement was performed after impedance measurement and conditioning of all electrode contacts.

In postoperative T-ECAP measures, the conscious CI subject might show discomfort at certain stimulation levels and/or electrodes. If the “loudest acceptable presentation level” is reached before a reliable response was recorded, results will show an incomplete T-ECAP profile. By measuring T-ECAPs intraoperatively, we were able to collect data for all 22 intracochlear electrode contacts in all subjects.

Statistical analyses were performed using IBM SPSS Statistics for Windows, version 24 (IBM Corp., Armonk, N.Y., USA) and MATLAB™ software (The MathWorks, Inc., Natick, Massachusetts, USA).

Normal distribution (Shapiro–Wilk test) did not apply to all variables. Therefore, the nonparametric Friedman test and the Kruskal–Wallis test were used for comparing the equality of paired and unpaired samples, respectively. Post hoc analysis was performed and p values were adjusted by using the Bonferroni correction for multiple comparisons. Correlation analysis was performed by Spearman rank correlation. Statistical significance was defined as p < 0.05.

The mean preoperative hearing loss ranged from 55 dB HL (125 Hz) to 100 dB HL (8 kHz) and is illustrated in Figure 3 and Table 1. Differences among pure-tone frequencies were found to be significant [Friedman test: χ²(6) = 269.89, p < 0.001]. Median 4FPTA was 82 dB HL (SD 18 dB HL; range 47–129 dB HL).

Mean WRS_max was 35% (SD 26%; range 0–100%). WRS_max and 4FPTA were significantly correlated (r = −0.51, p < 0.001).

The cochlear parameters of the basal turn diameter A value (mean 9.0 mm; SD 0.5 mm) ranged from a minimum of 8 mm to a maximum of 10 mm. An average insertion depth angle θ_apical of 380° (SD 24.2°) was achieved, ranging from 315° to 425°. Mean insertion depth at the most basal electrode (θ_basal) was 9.4° (SD 5.2°; range 1–25°). A negative correlation was found between the insertion depth angle θ_apical and the cochlear size with regard to diameter A (r = −0.33; p = 0.003).

The medial-lateral position MP ranged between 0.53 and 0.70 with a mean of 0.62 (SD 0.03). The most apical electrode E22 reached a mean SGC frequency of 749 Hz (SD 100 Hz) with a range between 615 and 1,162 Hz, while the most basal electrode E1 covered the SGC region around 16 kHz (SD 0.63 kHz; range 14.1–17.1 kHz).

Intraoperative T-ECAP measurements were performed in all 84 subjects at all 22 electrode contacts. In 57 out of the 1,848 electrodes tested, no valid T-ECAP could be measured. The overall success rate was 96.9%. In 61 subjects (73%) all 22 T-ECAPs could be measured. In eleven and two subjects respectively, 21 and 20 T-ECAPs were still detectable. In four subjects each, 19 and 18 T-ECAPs were measured. There was one subject with only 16 and one with only 14 valid T-ECAPs.

The mean T-ECAP was 177 CU (SD 22 CU; range 91–252 CU). The mean T-ECAP profile shows a pattern of increasing T-ECAPs from apical to medial electrodes, reaching a plateau at the medial region before thresholds increased further toward the basal region of the electrode array (Figure 4). T-ECAPs differ significantly across the electrode array [Kruskal–Wallis test: χ² (21) = 749.33, p < 0.001]. The pairwise comparison from post hoc analysis showed main differences among apical (e.g., EL20), medial (e.g., EL12) and basal (e.g., EL2) electrodes.

Figure 5 illustrates ECAP thresholds in relation to their place of stimulation. The Kruskal–Wallis test showed differences between T-ECAP and place of stimulation [χ²(3) = 367.06, p < 0.001]. Post hoc comparison revealed significantly higher ECAP thresholds for electrodes within the SG octave frequency band 8 kHz compared with 1, 2 and 4 kHz. Conversely, T-ECAPs around 1 kHz were significantly lower compared with those at 2, 4 and 8 kHz.

No correlation was found between the medial-lateral position MP and mean T-ECAP (p = 0.724). Also, the insertion depth angles θ_apical and θ_basal were not correlated with T-ECAPs at the corresponding EL22 (p = 0.539) and EL1 (p = 0.131).

There was no correlation found between cochlear diameter A and mean T-ECAP (p = 0.345).

Hearing thresholds as represented by the 4FPTA and by T-ECAP were significantly correlated (r = 0.2, p < 0.05). Additionally, Figure 6 shows the significant correlation of hearing loss at single SG-frequency 1 kHz (r = 0.18, p < 0.001), 2 kHz (r = 0.13, p < 0.05), 4 kHz (r = 0.25, p < 0.001), and 8 kHz (r = 0.42, p < 0.001) with T-ECAPs at the corresponding placed electrodes. Linear regression analysis was performed for each subset.

No correlation was found between mean T-ECAP and WRS_max (p = 0.575). Also, there was no correlation found between cochlear diameter A and 4FPTA (p = 0.989).

Before cochlear implantation, audiometric measures are assessed in order to identify CI candidates. As Hoppe et al. (2019) pointed out, CI candidates with even substantial residual hearing on the side of CI implantation benefit from the CI. Since we conducted our investigation only in subjects with preoperatively measurable residual hearing, the mean 4FPTA of 82 dB HL in our cohort was lower than in earlier studies of subjects with Slim Modiolar electrodes (86–98 dB HL; Ramos-Macías et al., 2018; Hey et al., 2019; Liebscher et al., 2021). Hearing levels ranged in our study from 58 dB HL at the lower (125 Hz) to 110 dB HL at the higher (8 kHz) frequencies. This decline toward higher frequencies is typical for CI candidates with moderate-to-severe hearing loss (Hughes et al., 2014; Ramos-Macías et al., 2018; Hoppe et al., 2021). Sensorineural hearing loss is caused by changes in peripheral processing due to damage to the inner hair cells and/or the auditory nerve. This damage is generally more severe in the basal half of the cochlea than in the medial and more apical regions.

Pure-tone thresholds in CI candidates for frequencies above 6 kHz are usually beyond the technical limits of the audiometer (e.g., in this study 130 dB SPL at 6 kHz), indicating complete loss of function of the inner hair cells at and above that frequency those frequencies. Therefore, we could not measure and include frequencies beyond 8 kHz in our analysis; even though it can be expected that these basal regions are affected the most by auditory sensory loss and auditory deprivation.

We investigated only fully inserted electrodes with complete scala tympani insertion. The results of the average insertion depth angle θ_apical of 380 ± 24.2° is comparable to results of other studies using the Slim Modiolar electrode (Aschendorff et al., 2017) or comparable perimodiolar electrode arrays with similar length (Landsberger et al., 2015), in which θ_apical of 403° and 381° respectively were reported.

Aschendorff et al. (2017) additionally reported a slightly deeper insertion of the most basal electrode (θ_basal) of 18° compared with the 9.4° found in this study. These variations can result from various factors, e.g., different surgical approaches (round window, extended round window or cochleostomy). The round window approach was more frequently used in this study (88% in our study vs. 44% in that of Aschendorff et al., 2017); an extension of the round window or cochleostomy can lead to an increased insertion depth angle (Briggs et al., 2006). Additionally, the cochlear anatomy is known to influence the achievable insertion depth angle, since a larger diameter A and cochlear duct length will reduce the overall possible insertion depth (Escudé et al., 2006; Franke-Trieger and Mürbe, 2015; Ketterer et al., 2018), even though the cochlear diameters A in both studies were in line (average of 8.9 vs. 9.0 mm). Lastly, Aschendorff et al. (2017) illustrated the influence of positioning of the “depth markers,” which provide assistance during surgery indicating the electrode array’s insertion depth in relation to the round window opening, to the actual achieved insertion depth and to the medial-lateral position. The best perimodiolar placement was achieved when the first marker was placed at the cochlea opening. The findings of Aschendorff et al. investigating the Slim Modiolar electrode were highly relevant for surgeons and their insertion approach with this electrode array later on.

The average medial-lateral position from this study of 0.62 matches the results of Aschendorff et al. (2017). The lower variation of 0.03 found in this study (compared with 0.05) and smaller maximum outliers of 0.7 (compared with 0.77) may be the result of the overall increase in experience with this electrode.

Intraoperative T-ECAPs could be recorded reliably for all subjects. We had a high overall success rate in finding ECAP thresholds (96.9%). This success rate is even higher than the previously rate of 95.6% reported by Liebscher et al. (2021), who investigated different perimodiolar electrode arrays, and it is also higher than the 90% reported by Hey et al. (2019), who studied the Slim Modiolar electrode. We interpret our high success rate as resulting from the fact that we only investigated in correctly placed ST insertions and had no pathological cases. However, there are still some limitations due to the use of the AutoNRT algorithm in the intraoperative test setting. Intraoperatively, the main reasons for missing T-ECAPs are electrical compliance complications, in which no ECAP response are detected as the maximum current level could not be reached owing to increased impedances (Spivak et al., 2011; Hey et al., 2019). Additionally, even with pre-conditioning applied, recording artefacts can occur (Van Dijk et al., 2007) that lead to a response being missing. However, all electrodes and electrode contacts worked within the specifications; throughout the test there were no open- or short-circuits.

The average T-ECAP profile of perimodiolar electrodes has been demonstrated in numerous studies. The profile typically shows increasing ECAP thresholds from the apical to the medial electrodes, where a plateau or in some cases a maximum is reached, with a slight decrease before increasing further (Figure 4) toward the basal part of the electrode array (Spivak et al., 2011; Müller et al., 2015; Hey et al., 2019; Mewes et al., 2020; Liebscher et al., 2021). Interestingly, the Slim Modiolar electrode shows lower T-ECAPs at the apical electrodes than its “perimodiolar counterpart” the Contour Advance electrode (Liebscher et al., 2021). One reason for this is the much less frequent incidence of an electrode being translocated from ST into the SV (Aschendorff et al., 2017; Shaul et al., 2020; Liebscher et al., 2021), a displacement that is accompanied by significantly increased T-ECAPs (Mittmann et al., 2015; Venail et al., 2015; Liebscher et al., 2021).

Electrode impedance can also influence T-ECAPs. Venail et al. (2015) demonstrated that higher impedance leads to lower ECAP thresholds in postoperative measures, and they concluded that fibrotic tissue growth surrounding the electrode contacts changes the electrical current path and therefore the impedance value. However, since we only investigated intraoperative T-ECAPs measured right after the insertion, tissue growth should not be an issue. Nonetheless, if the contact surfaces of single electrodes are surrounded by different fluids or tissues, then the impedance values will differ. Both electrode types (Slim Modiolar and Contour Advance) show different impedances patterns (Mewes et al., 2020), since the Slim Modiolar electrode is significantly smaller in volume than the Contour Advance. Therefore, less space within the cochlea is used and the intracochlear electrodes’ positioning differs, which can lead to different current paths.

Regarding the electrodes’ intracochlear positioning, in our results the average medial-lateral position of 0.62 shows that there is an overall small distance between electrode contacts and the modiolus. Regardless of this, the physical electrode-to-modiolus distance is not correlated with ECAP thresholds, which confirms the findings of Venail et al. (2015).

Lastly, the Slim Modiolar electrode is designed to be more effective with regard to the preservation of residual hearing. Therefore, patient recruitment might have changed over time, since more subjects with substantial amount of residual hearing nowadays receive this perimodiolar electrode array instead a straight (e.g., CI622) or shorter (e.g., CI624) electrode array.

The effects of reduced ECAP responses caused by a smaller amount of surviving SGCs have been shown in animal studies (Prado-Guitierrez et al., 2006; Ramekers et al., 2014). Ramekers et al. (2014) demonstrated in their histological and electrophysiological analysis that a reduced packing density of SGCs in the deafened group of guinea pigs led to a decreased ECAP amplitude growth function. The neural survival of the peripheral auditory system can be revealed in an objective manner by using ECAPs with CI Systems.

Our cohort of CI users showed a similar relationship. Subjects with better preoperative hearing also had lower ECAP thresholds, meaning that a smaller amount of electrical stimulation was needed to determine the objective auditory response. We therefore take this as further evidence that T-ECAPs can reflect intracochlear neuronal health status. This correlation is evident for frequencies 1, 2, 4 and 8 kHz (Figure 6). Since the hearing loss is generally less distinct in the apical than in the basal cochlear region, T-ECAPs are lower at the apical than at the basal electrodes. Our results support the findings of Nassiri et al. (2019), who also found lower T-ECAPs at apical electrodes in subjects with residual hearing. The relationship of decreasing ECAP thresholds between apical and basal regions can also be seen for the characteristic frequencies below 1 and beyond 8 kHz (Figure 5). However, since the average insertion depth (θ_apical) of the most apical EL22 is 380°, which corresponds to 749 Hz (SD 100 Hz) SG frequency, there are only few subjects with the most apical electrode inserted within the octave band with the 500 Hz center frequency; therefore, this frequency group was not included in the final analysis. Additionally, sufficient hearing thresholds beyond 8 kHz were not encountered in any subject; hence, we could not include higher-frequency groups in our correlation analysis with ECAP measures.

Nonetheless, it is important to point out that around an SG frequency of 16 kHz the overall highest ECAP thresholds were found. With regard to our findings, we assume that in these basal regions the neural survival of the peripheral auditory system is particularly low. Studies that used histopathological analysis support this assumption. For example, Wu et al. (2021) recently examined the survival rate of inner and outer hair cells, and of peripheral axons of the auditory nerve fibers, from temporal bones in normal-hearing and noise-exposed human subjects. They found an age-related loss in hair-cell and peripheral-axon counts in both groups in the apical and basal regions. However, the noise-exposed group showed smaller counts for high frequencies and lower hearing thresholds. Additionally, studies investigating the primary neuronal loss by counting SGC in temporal bones of hearing-impaired human subjects show a negative correlation with word-recognition scores and audiometric thresholds (Otte et al., 1978; Sagers et al., 2017). Makary et al. (2011) also displayed the age-related decline in SGC counts in healthy subjects. Their findings showed also a trend of increased degeneration in the basal turn compared to the upper turns.

Overall, histopathological studies show that auditory neuronal loss occurs at various stages (hair cells, SGCs) and is driven by age, etiology and severity of the hearing impairment.

Lastly, we found no correlation between intraoperative mean T-ECAP and preoperative WRS_max. The arithmetic T-ECAP mean of all electrodes might be too vague, since speech recognition requires central auditory processing and T-ECAPs do not provide information about temporal processing. Other study groups who investigated in the relationship between ECAP measures and postoperative speech perception outcome with CI showed a more complex picture. Instead of just using T-ECAPs, Skidmore et al. (2021) generated an ECAP index value that consists of various different ECAP measures, including T-ECAP, slope of the amplitude growth function, N₁-latency and refractory recovery function. In adults, they showed a significant correlation of Consonant-Nucleus-Consonant-word (CNC) lists and AzBio sentence measures in quiet with the ECAP index. He et al. (2022) showed that subjects with a prolonged speed of recovery from neural adaptation tend to have poorer speech perception outcome. Similar findings were shown by another publication of Skidmore et al. (2023), who additionally found significant correlation of the ECAP index with CNC words and AzBio sentences in noise. These studies show, that in order to find a correlation of auditory performance with ECAP measures, a more extensive analyses from the characteristics of neural response is needed.

In CI candidates with preoperative speech perception, WRS_max can be a predictor of the minimum speech perception obtained with CI later on (Hoppe et al., 2019). In subjects without speech perception, pure-tone thresholds are still useful in order to estimate neuronal health status.

It was shown recently (Fontenot et al., 2019; Kim, 2020), that electrocochleography correlates with auditory perception. When using acoustic stimulation ECoG measurements are influenced by both the hair cell functioning and the neural processing. For electrically evoked responses as used in our study no acoustic-electric conversion is necessary. Hence, ECAP deliver information on the neural components without being influenced by hair cell functioning.

There are still factors which are not recorded but might affect the correlation in our findings. E.g., in cases of retrocochlear hearing loss at certain frequencies; there are no valid hearing levels, but one might find peripheral responses (T-ECAPs). Additionally, we have no knowledge of the individual existence of potential cochlear dead regions (Moore and Malicka, 2013). If there are non-functioning inner hair cells at a certain region for a specific frequency, adjacent hair cells may evoke neural excitation. Due to this “off-place listening” the measured pure-tone audiometric thresholds might not reflect the actual hearing loss properly.

A similar effect might occur at the electrophysiological level in ECAP measures. Due to the monopolar stimulation mode, the spatial spread generated by the electrical stimulus of one single electrode can cause excitation of neighboring SGNs (Cohen et al., 2003). It has been shown, that current focused stimulation modes can provide less current spread (Bonham and Litvak, 2008; Padilla and Landsberger, 2016), providing a more place-specific electrical stimulation. However, this technique is not yet available in standard (T-)ECAP measures.