Intra-Cochlear Electrode Position Impacts the Preservation of Residual Hearing in an Animal Model of Cochlear Implant Surgery

In recent years, the use of cochlear implants (CIs) has expanded to include patients with residual low-frequency acoustic hearing based on preserved neurosensory structures located in the apical turn of the cochlea. It is assumed that the preservation of residual hearing (RH) is of particular importance to improve speech perception in difficult listening situations with background noise [1]. Furthermore, preventing hair cell (HC) loss may also delay the degeneration of the auditory pathway to some extent, potentially resulting in better electrical hearing outcomes for CI recipients [2].

Since the insertion of the CI electrode array usually does not reach the apical turn, the possibility of mechanical destruction of those neurosensory structures by the electrode array itself can be excluded. Although this would in principle allow RH to be preserved, over 60% of CI recipients currently experience only minimal hearing preservation or a complete loss of RH [3, 4]. Possible pathological mechanisms behind this phenomenon may include, e.g., an insertion-related increase in intra-cochlear pressure or an inflammatory immune response to the electrode array by fibrosis and neo-ossification. In this respect, various efforts have been made to preserve RH, including the modification of the surgical approach, pharmacological interventions, and electrode array design [4‒6]. Although those studies have shown positive short-term effects in the preservation of RH (e.g., atraumatic insertion of the electrode or local application of corticoids), medium or long-term data are not available which would support any of those approaches [7‒9].

One feature has not yet been experimentally investigated in detail, i.e., the electrode array’s position within the scala tympani (ST). In this respect, recent studies suggest that the use of different electrode arrays, designed to adopt distinct locations within the ST, has an impact on the preservation of RH. Mady et al. [10] investigated the preservation of RH (≤85 dB at 250 and 500 Hz) in 45 patients who obtained either a straight array that tends to result in an anti-modiolar position along the lateral wall or a curved array that adopts a peri-modiolar position. In the short-term follow-up (1-month post-surgery), lateral wall arrays showed significantly better preservation of RH than peri-modiolar arrays. This effect could no longer be observed in the long-term follow-up (1-year post-surgery). A retrospective study by Wanna et al. [11] concluded a significantly higher likelihood of hearing preservation in patients with lateral wall and mid-scala arrays compared to patients with peri-modiolar arrays in the short-term follow-up (3 weeks post-OP). After ≥1 year of post-surgery, this effect was only evident for lateral wall arrays. Further studies confirmed the superiority of a lateral wall electrode for preserving RH [4, 7, 12‒14]. Although this indicates an influence of the electrode’s position on the preservation of RH, it should be noted that present studies rely on the assumed position of the electrode, based on the electrode array’s design, rather than precisely evaluated electrode array position data. To date, studies on the impact of electrode position in CI recipients with RH only consider the scalar position and insertion depth but do not fully evaluate the precise location of the electrode array within the ST. Thus, the exact contribution of the array’s position to the preservation of RH is still a matter of debate.

New imaging techniques, such as high-resolution micro-computed tomography (µCT), allow for the visualization of internal cochlear structures and precise determination of electrode positioning. However, investigations on electrodes with metallic components still show artifacts which blur the region-of-interest and, thus, do not allow a precise evaluation of the electrode array’s position within the ST [15].

In the present study, the use of an array without metallic components enables the investigation of a possible correlation between the precisely determined electrode position within the ST and low-frequency hearing loss in the range of RH. In this context, a possible impact of the inter-individual ST diameter on RH loss was examined.

For the present study, 4 normal hearing adult (17–24 weeks) guinea pigs (Dunkin Hartley, Cavia porcellus) were used. All experiments were approved by the Governmental Commission for Animal Studies (LaGeSo Berlin, Germany, Approval No. G14614). The experiments were conducted in accordance with the EU Directive 2010/63/EU on the protection of animals used for scientific purposes. Every effort was made to minimize pain or discomfort.

All surgeries and measurements were performed under general anesthesia (fentanyl [0.025 mg/kg]-midazolam [1.0 mg/kg]-medetomidine [0.2 mg/kg], i.m.). Body temperature was maintained at 37°C by a heating mat placed under the animal.

One ear per animal was randomly assigned for insertion of customized guinea pig ST pure silicone arrays (without electrode contacts or wires) (Advanced Bionics GmbH, Hanover, Germany) into the basal part of the cochlea via a cochleostomy next to the round window. The electrode dimensions are shown in Figure 1. Subsequently, the cochleostomy was sealed with muscle tissue and the silicone array was fixed on the bulla tympanica using dental cement. For human CI surgery, it is quite common to insert the electrode array via the round window. However, the anatomy of the guinea pig generally makes this impractical, meaning that a cochleostomy was used for insertion of the electrode array.

The auditory brainstem responses (ABR) were measured on the assigned side at frequencies of 4, 8, 12, 16, 20, 24, 28, and 32 kHz. Measurements were made 1 week before and 4 weeks after the insertion of the pure silicone array (schedule in Fig. 2). ABRs were recorded using a differential amplifier (Opti-Amp 8002, Intelligent Hearing Systems, Miami, USA) and Smart EP 5.33 software (Intelligent Hearing Systems, Miami, USA). Tone pulses (31.10 pulses/s, 3 ms duration, 1.5 ms rise, 1.5 ms fall) were delivered into the external ear canal via an in-the-ear high-frequency transducer. Three needle electrodes (vertex [reference], mastoid [active], and leg [ground]) were attached to record 1,024 sweeps. The signal was filtered (high pass 100 Hz, low pass 3,000 Hz) and amplified (100k). The acoustic stimuli’s level was decreased in 5-dB steps following each recording until the auditory threshold was reached. The contralateral ear was masked with broadband noise (2–32 kHz, 30 dB below stimulus loudness) during all ABR measurements. The auditory threshold was defined as the stimulus level at which a visual wave III was no longer deducible. After postoperative ABR measurement (4 weeks after insertion), animals were sacrificed and perfused with 4% paraformaldehyde which was administered via the left ventricle. The cochleae were carefully removed from the petrous bone and stored in 4% paraformaldehyde for subsequent procedures.

The µCT scans were performed with a MILabs MicroCT scanner (MILABS B.V. Duwboot 7a 3991 CD Houten, Netherlands). The isotropic voxel size was 10 μm. To reduce artifacts, a high voltage of 50 kVp and currents of 21 mA and an integration time of 75 ms were used. All scans had a scale of 0.1 pixel/µm.

The µCT images of the cochlea were automatically aligned in the superior-inferior axis (apex-base of the cochlea) using the TransformJ script for ImageJ (National Institutes of Health, USA). The position of the silicone array was analyzed in the transversal z-plane projection (shown in Fig. 3).

The ICPI measurement represents the array’s position within the ST, where a value toward 0 represents a close position to the modiolus (peri-modiolar position) and a value toward 1 represents a close position to the lateral wall (anti-modiolar position) [16].

The insertion depth of the silicone array was determined by manually measuring the length of the inserted part of the array from the array tip to the cochleostomy on µCT images in the sagittal plane using ImageJ (National Institutes of Health, USA). The insertion depth of the array was assigned to the corresponding frequency range according to Greenwood [17]. The corresponding frequency range was defined as the range in the region of the cochlea occupied by the silicone array, whereas the remaining (lower) frequency range was defined as the frequency range apical to the silicone array’s area.

The petrous bone samples were incubated in 0.4 m ethylenediaminetetraacetic acid (EDTA) (pH 7.5; Roth, Germany) for 14 days (the solution was changed every 2 days). After embedding in paraffin, 6 μm tissue sections were prepared and subjected to AZAN staining according to Heidenhain correspondent to the manufacturer’s instructions (Morphisto, Germany). In brief, the tissue was deparaffinized in ROTI Histol (Roth, Germany) and dehydrated in ethanol. After pretreatment with aniline alcohol, nuclear staining with azokamine and differentiation with aniline alcohol were performed. The reaction was stopped with acetic acid alcohol. After a pickling step in phosphotungstic acid, bone and connective tissue were stained with aniline blue/orange G. Finally, the tissue was dehydrated in ethanol, 2-propanol, and ROTI Histol, and mounted with ROTI Mount (Roth, Germany). The documentation was performed with a Zeiss Axio Lab.A1 microscope equipped with an AxioCam ICc1 camera (Zeiss, Germany).

SPSS software (IBM SPSS Statistics version 25, IBM Corp., NY, USA) was used for all statistical analyses. The data distribution was tested with the Shapiro-Wilk test. Pre- and post-surgery ABR data were compared using either a one-way paired t test (for normally distributed data) or a one-way Wilcoxon test (for not normally distributed data).

Possible correlations between the ICPIs and the hearing loss (post-insertion ABR threshold normalized against pre-insertion ABR threshold) were analyzed in the previously determined frequency ranges of the array’s area, as well as in the defined frequency range apical to the array’s, area using two-way Pearson’s correlation analysis. Here, the ICPIs at measuring points 1 (most apical) to 8 (most basal) were correlated with hearing loss in the most apical frequency to the most basal frequency. The correlation analysis in the previously determined frequency ranges of the array area as well as in the defined frequency range apical to the array area was examined by using every second ICPI from measuring point 1 (most apical) to 8 (most basal). These were correlated with the hearing loss in the most apical frequency to the most basal frequency of the array’s area, or apical to the array’s area. Correlation analysis between the ST diameter and the hearing loss was performed for the total frequency range, the array area and the area apical to the array according to the procedure described above by using the ST diameter values at the defined measuring points.

All correlation analyses were performed using either a two-way Pearson’s correlation analysis or a two-way Spearman’s ρ depending on data distribution. The significance level for all statistical tests was set at p < 0.05.

In order to exclude an influence on the postoperative hearing threshold by incorrect insertion of the silicone array (e.g., translocation) or structural damage to the cochlea by the inserted array (e.g., fracture of the osseous spiral lamina or rupture of the basilar membrane), all 4 samples were subjected to histological examination. This analysis confirmed a consistent insertion of the array into the ST in all 4 animals (shown in Fig. 4). The silicone array was encapsulated by a fibrotic tissue layer. Furthermore, fibrosis and neo-ossification were observed in the ST as part of the foreign body reaction.

Auditory brainstem responses (ABRs) verified the absence of hearing problems in all animals 1 week before surgery. In comparison, insertion of the pure silicone array induced a significant hearing loss 4 weeks after implantation (shown in Fig. 5), with hearing loss occurring over the entire frequency range (4–32 kHz) (shown in Fig. 5). The averaged hearing loss ranged from 41.2 dB ± 17.4 dB at 4 kHz to 57.5 ± 5.6 dB at 12 kHz.

The evaluation of the µCT images revealed different array positions within the ST ranging from a more peri-modiolar position to an anti-modiolar position. These observations were reflected by measurements of 8 ICPIs per electrode array (shown in Fig. 6). Averaged ICPIs ranged from 0.33 to 0.59 with a standard deviation between 0.06 and 0.14.

Measurements of insertion depth resulted in insertions of the array into the basal turn of the cochlea of 2,498.35 µm to 3,187.34 µm ± 246.91 µm. The assignment of the insertion depth to the corresponding frequency range showed a penetration of the array from 32 kHz to approximately 19 kHz in three animals and from 32 kHz to approximately 17 kHz in one animal. Since corresponding ABR measurements were available at 16 kHz and 20 kHz, an array’s area was defined at the frequency range 32–20 kHz in the three animals and 32–16 kHz in the one animal. The more apical (lower) frequency range was accordingly defined as the frequency range apical to the array area (4–16 kHz in the one animal and 4–20 kHz in the three animals).

Correlation analysis of the ICPIs and the hearing loss in the total frequency range revealed a significant negative correlation (r_xy = −0.40; p = 0.02). In the previously determined frequency range associated with the silicone array’s area, no significant correlation (r_xy = −0.11; p = 0.66) was found. However, a correlation analysis for the frequency range apical of the area occupied by the silicone array (shown in Fig. 7) revealed a highly significant negative correlation between ICPIs and hearing loss (r_xy = −0.65; p = 0.008).

The analysis of a possible correlation between the ST diameter at the defined measuring point at the electrode and the observed hearing loss did not reveal a significant correlation in the total frequency range (p = 0.74), the array area (p = 0.8), or the area apical to the array (p = 0.6) (Fig. 8).

The insertion of a pure silicone array without metallic components allowed us to precisely determine the array’s position within the ST using artifact-free, high-resolution µCT images. While only a single type of silicone array was used, the dimensions were such that it occupied a similar amount of the guinea pig cochlea that a typical human-sized array would occupy in a human cochlea. The measured array locations ranged from a peri-modiolar to an anti-modiolar position and revealed a correlation of lower postoperative hearing loss with the proximity of the array to the lateral wall. This association was found exclusively in the low-frequency range. This is consistent with previous work from our laboratory, which showed a higher loss of HCs in the apical portion of the cochlea compared to the basal region 4 weeks post-insertion of an array into the guinea pig cochlea [18].

All of these findings raise the question of the reason for position-dependent hearing loss in the apical turn of the cochlea. The effect of electrode array position being exclusive to apical regions is possibly related to pressure changes, which occur during the electrode array’s insertion: when the electrode array acts like a piston in the tapering ST. Such high pressure changes were detected during electrode insertions into plastic cochleae [19‒21] and human cadavers [22, 23]. Consequently, mechanical overstimulation of HCs leads to programmed cell death, and ultimately to a significant HC loss which is assumed to be the main contributor to low-frequency hearing loss following CI electrode insertion [24, 25]. Naturally, the anatomy of the human cochlea is quite different from that of the guinea pig. With the guinea pig cochlea-sized array being used in this study, the main effects of how a clinical electrode array will behave in a human cochlea should be captured, while bearing in mind that there is a great deal of variability in the shapes and sizes of both guinea pig and human cochleae. In agreement herewith, it is assumed that an insertion-induced increase of intra-cochlear pressure affects apically located HCs more strongly than mechanical stress by the electrode array, which would affect basally located HCs [20, 26]. Insertion trauma marked by an intra-cochlear pressure increase is influenced by various factors, e.g., the specific CI electrode array insertion, electrode array type [22, 27], insertion speed and insertion technique [20, 23]. Although insertion of the silicone array was carried out in a controlled manner, our results show that it is possible to end up with the divergent array positions described, even with the same array type. Different array positions should be related to the various diameters of the ST at different parts of the cochlea, leading to the potential occlusion of the ST during the insertion. This ultimately could lead to different pressure levels being applied to apical cochlear structures as the array is advanced into the cochlea. Additionally, should a design of electrode array that is not on the lateral wall have its tip stick and free again repeatedly during insertion, this can also lead to pressure spikes that are applied mainly to the apical portion of the cochlea. These effects should be lower for a mid-scalar or straight lateral wall electrode position compared to, say buckled peri-modiolar ones. Recent clinical studies which found a higher preservation of RH for very slim or anti-modiolar electrodes support this hypothesis [4, 7, 10‒14, 28]. The width of the ST is reported to correlate with overall cochlear size [29] and the observation that more hearing loss tends to occur in smaller cochleae [30, 31] tends to further support this hypothesis. Furthermore, Mittmann et al. [27] found a significantly higher pressure peak amplitude during the insertion of a peri-modiolar array into an artificial cochlea model compared to the insertion of a straight electrode array. However, a study on human cadavers was not able to show a conclusive correlation between the design of the inserted electrode array (peri-modiolar array vs. straight array) and the resulting intra-cochlear pressure patterns. The insertion of the same (straight) array resulted in both the highest and lowest pressure ranges encountered during the experiment [22]. The observation of differential pressure patterns might be explained by the occurrence of deviating electrode array positions while using the same electrode design, as has been shown in the present study. This would further indicate that an electrode array’s position-dependent pressure patterns influence the preservation of RH.

Despite the absence of a correlation between the inter-individual ST diameters and hearing loss, our results suggest that, beyond surgical technique and electrode array design, inter-individual anatomical differences of the cochlea (e.g., angle of the ST in the basal turn) could influence the exact electrode array position within the ST. As the electrode array’s position was found in the guinea pig to correlate with preserved RH beyond the extent of the array within the cochlea, the investigation of inter-cochlear anatomical structures as a predictor for the exact electrode array position within the ST could improve individual CI surgery planning, resulting in a higher probability of preserving RH. Although the significance of the present study is limited by the small number of animals examined, as well as by the use of a single type of electrode array model in this study, we provide here evidence for the preference of an anti-modiolar array position to reduce the loss of residual low-frequency hearing induced by a CI electrode array insertion. Future work should clarify the cause of this finding by investigating the influence of electrode array position-dependent pressure patterns. Furthermore, investigations on the influence of electrode array position on RH preservation should be combined with CI-related electrical stimulation. The results shown above appear to be important for patients with substantial RH. However, clinical studies have also shown that closer proximity between electrode contacts and the modiolus correlates with a higher word recognition score [32, 33] and increased electrode contact (channel) discrimination 6 months after surgery [34]. Thus, the selection of the electrode array type should be made very much on an individual basis, depending on the status of RH and the cochlear anatomy of the patient. In this study, only one type of array has been studied here and there are a number of different electrode array designs available for clinical practice.

In making these observations, several limitations of the present study should be kept in mind. First is that a relatively small number of animals that have been studied. Despite this, the considerable variability in array position reported still comes out. The differences between guinea pig and human cochleae should also be considered. Its considerable anatomical differences between species, along with the quite different frequency ranges being addressed, mean that morphological and physiological differences do exist and need to be considered. While the higher level observations relating to position of the array and its impact on hearing levels apical to the array’s position appear reasonable, we should not dig too deeply into details that will be specific to a human hearing preservation surgery. Related to this is the need to use a cochleostomy for the current study of guinea pigs, while today’s state-of-the-art hearing preservation surgery tends to use a round window insertion. In clinical practice, there are different types of electrode array design routinely used, aiming for lateral, mid-scala, or peri-modiolar locations within the ST. As stated earlier, only a single type of array has been studied here and this is a further limitation of the current study when directly considering clinical application.

All experiments were approved by the Governmental Commission for Animal Studies (LaGeSo Berlin, Germany, Approval No. G14614).

M.M., S.S., M.T., D.B., and A.E. declare no conflict of interest in this study. P.B. is employed by Advanced Bionics GmbH in a scientific role.

Advanced Bionics GmbH funded this study. The funder had no role in the design, data collection, data analysis, and reporting of this study.

M.M. and D.B. designed the study. M.M. and S.S. performed the histology. M.M. and M.T. performed the µCT measures. M.M. prepared the manuscript draft. D.B., P.B., and A.E. revised the manuscript.

The data that support the findings of this study are not publicly available due to the protection of legal interests of the sponsor but are available from the corresponding author upon reasonable request.