Single-incision cochlear implantation and hearing evaluation in piglets and minipigs

Objectives: Various animal models have been established and applied in hearing research. In the explo- ration of novel cochlear implant developments, mainly rodents have been used. Despite their important contribution to the understanding of auditory function, translation of experimental observations from ro- dents to humans is limited due to the size differences and genetic variability. Large animal models with better representation of the human cochlea are sparse. For this reason, we evaluated domestic piglets and Aachen minipigs for the suitability as a cochlear implantation animal model with commercially available cochlear implants. Methods: Four domestic piglets (two male and two female) and six Aachen minipigs were implanted with either MED-EL Flex24 or Flex20 cochlear implants respectively, after a step-by-step surgical approach was trained with pig cadavers. Electrophysiological measurements were performed before, during and after implantation for as long as 56 days after surgery. Auditory brainstem responses, electrocochleography as well as electrically and acoustically evoked compound action potentials were recorded. Selected cochleae were further analyzed histologically or with micro-CT imaging. Results: A surgical approach was established using a retroauricular single incision. Baseline auditory thresholds were 27 ± 3 dB sound pressure level (SPL; auditory brainstem click responses, mean ± stan- dard error of the mean) and ranged between 30 and 80 dB SPL in frequency-speciﬁc responses (0.5 – 32 kHz). Follow-up measurements revealed deafness within the ﬁrst two weeks after surgery, but some animals partially recovered to a hearing threshold of 80 dB SPL in certain frequencies as well as in click responses. Electrically evoked compound action potential thresholds increased within the ﬁrst week after surgery, which led to lower stimulation responses or increase of necessary charge input. Immune re-actions and consecutive scalar ﬁbrosis following implantation were conﬁrmed with histological analysis of implanted cochleae and may result in increased impedances. A three-dimensional minipig micro-CT segmentation revealed cochlear volumetric data similar to human inner ear dimensions. Conclusions: This study underlines the feasibility of cochlear implantation with clinically used cochlear implants in a large animal model with representative inner ear dimensions comparable to humans. To bridge the gap between small animal models and humans in translational research and to account for the structural and size differences, we recommend the minipig as a valuable animal model for hearing research. First insights into the induced trauma in minipigs after cochlear implant surgery and a partial hearing recovery present important data of the cochlear health changes in large animal cochleae. © 2022 The Authors. Published by Elsevier B.V. This is an open )


Introduction
Cochlear implantation surgery today serves as a relatively lowrisk procedure with improvements in speech understanding and quality of life in patients with severe hearing loss ( McRackan et al., 2018 ). Many different animal models have been established to elucidate the mechanisms of cochlear implant (CI) insertion trauma or the potential to use the CI electrode array for drug delivery. Rodents (e.g., mouse, rat, gerbil, and guinea pig) represent the most commonly used animal group in this research field. These have led to the identification of hearing preservation after insertion trauma ( Attias et al., 2016 ;Braun et al., 2011 ;Kopelovich et al., 2015 ) and yield translational solutions for intracochlear drug delivery ( Frisina et al., 2018 ). Many of these animal models may seem ideal to test efficacy of device or pharmacological interventions to improve hearing preservation. However, due to distinct anatomical and genetic differences compared to humans, a direct translation from rodent findings to humans is difficult. Therefore, additional validation is required before starting clinical trials. Non-human primates are regarded as an ideal model to bridge the evolutionary gap between rodents and humans ( Marx et al., 2013 ). For ethical and financial reasons however, the use of these animals in research is complex and even prohibited in some countries of the world. This raises the question whether other large animals may be more suitable for CI research. The ideal animal model features anatomical inner ear dimensions and number of cochlear turns similar to humans.
Several species have been studied to find the best humanlike animal model for inner ear research. Among those, cats were evaluated in the early 1950s and have served as one of the largest non-rodent animal model for electrical stimulation and drug delivery in the past decades ( Duan et al., 2004 ;Klinke et al., 1999 ;Kretzmer et al., 2004 ;Leake et al., 2011 ;Schuknecht, 1953 ;Spoendlin and Gacek, 1963 ). Nonetheless, for the insertion of commercially available CIs used in humans, cats are not suitable due to a smaller and shorter cochlea with an average basilar membrane length of 22 mm ( Lovell and Harper, 2007 ). Cadaveric studies in lamb, pig and sheep ( Cordero et al., 2011 ;Elsayed et al., 2019 ;Gocer et al., 2007 ;Gurr et al., 2010aGurr et al., , 2010bHoffstetter et al., 2011 ;Mantokoudis et al., 2016 ;Schnabl et al., 2012 ;Trinh et al., 2021 ) show the advantages and disadvantages of each species as a representative animal model for otologic research. Kaufmann et al. were able to perform partial insertion of regular CIs in living sheep and demonstrated CI insertion depths ranging between 4.6 to 12 mm ( Kaufmann et al., 2020 ). Although the round window membrane thickness in the sheep model is similar to human ( Han et al., 2021 ), the shorter sensory epithelium and cochlear length seem to result in relatively small insertions depths. Gurr et al. first described the long and steep course of the external auditory canal in pig cadavers ( Gurr et al., 2010a ). In contrast to humans, the mastoid in pigs is located underneath the middle ear, partially covered by the atlanto-occipital joint. These porcine external ear anatomical characteristics complicate the access to the cochlea and make this animal model appear rather unsuitable for cochlear implantation with human-sized CIs. However, first cochlear implantations with commercially available electrode arrays were performed in another minipig breed and thus proved general feasibility of CI insertions in these animals ( Chen et al., 2017 ;Yi et al., 2016 ). In fact, the porcine basilar membrane length of approximately 32 mm (human ∼ 33.5 mm), and its 3.5 turns (human 2.75 turns) appear to be more similar to human cochleae as opposed to most other large animal models ( Lovell and Harper, 2007 ). Porcine temporal bones were previously compared to human temporal bones in high-resolution computed tomography scans ( Knoll et al., 2019 ) and the porcine cochlea has been analyzed in detail with latest technological advancements such as light sheet fluorescence microscopy ( Moatti et al., 2020 ). This highlights growing interest and potential suitability of the pig model for hearing research.
The objective of this study was to evaluate the suitability of commercially available CI devices in the domestic pig and in the Aachen minipig and to describe a detailed surgical protocol. Outcome measures included objective audiometry at two months post-implantation and correlation of electrophysiological data to histological findings in the extracted cochleae. Furthermore, a detailed analysis of minipig inner ear volumetric data were compared with the human inner ear for the proof of concept to implant both pig strains with conventional CIs.

Animals and anesthesia
Animal experiments were approved by the local animal welfare committee and the Austrian Federal Ministry of Education, Science and Research (BMBWF-2020-0.272.252). Four domestic piglets (two female and two male) with mixed strains ("Edelschwein", "Landrace" and "Pietrain" strains) and six male Aachen minipigs (Heinsberg, Germany) were acquired for this study. Mean animal body weight (BW) was 20.93 ± 1.43 kg (piglets) or 6.31 ± 0.30 kg (minipigs) at the beginning of the experiment at 9 weeks of age in both porcine animal models (corresponding to early years in humans). While the weight of the piglets remained relatively stable, it increased to 12.92 ± 0.63 kg in minipigs up to day 56 after surgery (26.18 ± 3.59 kg for piglets up to two weeks after insertion). For initial sedation, a mix of fentanyl (0.65 μg/kg BW), ketamine (10 mg/kg BW), medetomidine (100 μg/kg BW) and midazolam (37.5 μg/kg BW) was injected intramuscularly behind the right ear. After a waiting period of 5-10 min, the animal was brought into a calm room and a venous catheter was inserted into an auricular vein. If only hearing tests were to be performed, propofol was infused intravenously at a rate of 0.5-2 ml/kg BW per hour and the animal's heart rate and oxygen saturation were monitored continuously. For cochlear implantation surgery, the animal was transported to the operating room, where it was intubated after an initial propofol induction and anesthesia was maintained with 4% sevoflurane inhalation. A constant body temperature between 38.3 and 39.5 °C was obtained with a bair hugger TM (3 M, Minnesota, USA). Intraoperative pain medication consisted primarily of 0.5-2 μg/kg BW sufentanil and was adapted to 0.15 -0.7 μg/kg BW 30 min after the start of surgery. All animals received postoperative analgetic treatment with piritramide (0.15 mg/kg BW i.m., three consecutive injections with a time interval of 8 h), metamizole (50 mg/kg BW i.v.) and fentanyl patches (6 μg/h, attached onto the left front leg). Perioperative antibi-otic treatment was carried out with tulathromycin (2 mg/kg BW i.m.). Anti-emetic therapy consisted of intravenous application of droperidol (50 μg/kg BW).

Surgical approach
A step-by-step approach was implemented to receive a thorough understanding of the anatomical landmarks of the outer ear as well as the cartilaginous and bony parts of the external auditory canals of pigs. Twenty porcine cadavers originally used in other animal experiments at the Biomedical Research Center (Medical University of Vienna) served for training purposes to optimize the surgery. The in vivo surgical approach was performed in four piglets. After refinement of the surgical steps, the conclusive surgical access was performed in six Aachen minipigs. At the end of the experiments, all animals were euthanized with an intracardial pentobarbital injection (300 mg/kg BW). For optimal preservation of cochlear tissues, a previously described fixation method was used ( Musigazi et al., 2018 ). In short, after dissection of the external jugular veins and carotid arteries, a safe closed system for perfusion with solvents was prepared. Carotid arteries were used for fluid inflow and external jugular veins for fluid outflow. Internal jugular veins were ligated. At first, two liters of a heparinized 0.9% sodium chloride solution were perfused with a peristaltic pump at an infusion rate between 80 and 200 ml/min. Next, four liters of a 4% paraformaldehyde solution were perfused, and finally cochleae were extracted according to a previously described method ( Schuknecht, 1968 ).

Electrophysiological measurements
Audiometric tests were performed in a specially equipped room with a quiet environment preoperatively, intraoperatively, and at days 7, 14, and 56 after surgery. After sedation, the animals were placed on a mobile padded vehicle in prone position. Heart rate and oxygen saturation were continuously monitored with a pulse oximeter, which was attached to the tail of the animal. One subcutaneous needle electrode (-) was placed one centimeter behind each auricle. Two additional needle electrodes ( + ) were placed on the vertex and one needle (ground) electrode was placed on the snout (see Supplementary Fig. 1A). Bone conduction, acoustic clickand tone burst auditory brainstem responses (ABRs) were recorded with a mobile audiometric device (Intelligent Hearing Systems, Miami, USA) preoperatively after calibration with ten porcine auditory canals. Stimuli were presented in decreasing steps of 5 dB starting from 95 or 120 dB for the determination of click and tone burst ABR thresholds. For click ABRs, the stimulus was presented with a duration of 0.1 ms, a rate of 27.7/ sec , rarefaction polarity and the recordings were averaged following 1024 sweeps. Based on data of previous studies, which indicate a hearing range of 0.3 to 45 kHz in pigs ( Heffner and Heffner, 1990 ;Lovell and Harper, 2007 ), the tone bursts were acquired in frequencies of 0.5, 1,2,4,8,16,20, and 32 kHz with alternating polarity and 1024 sweeps. For 0.5 kHz the condensation polarity was chosen, and stimulus duration was determined with 8 ms, whereas all higher frequencies were recorded in alternating polarity with 4 ms stimulus duration. ABR thresholds equal or above 95 dB SPL were considered as "functionally deaf". Bone conduction ABRs (BC-ABRs) were recorded with 0.1 ms stimulus duration and a rate of 7.1/ sec and 512 sweeps. Masking of the contralateral ear was set to 40 dB above the stimulation intensity of the ipsilateral ear. All acoustic measurements were performed after removal of excessive cerumen. Acoustic compound action potentials (CAPs) after click stimuli and cochlear microphonics (CMs) after 1 kHz acoustic stimuli were recorded intraoperatively with a gold wire electrode, which was placed on the round window membrane after minor ear canal wall drilling ("PromStim", MED-EL, Innsbruck) as indicated in Supplementary Fig. 1B Electrically evoked compound action potentials (eCAPs) as well as impedances were recorded in sedated animals after CI insertion intraoperatively, and on days 7, and 14. To represent basal, middle, and apical parts of the cochlea, CI electrodes 1, 5, and 10 were evaluated. With the MED-EL Maestro 9.0.3 (MED-EL, Innsbruck, Austria) software stimulation parameters were applied with increasing charge between 0 and 50 charge units (qu, according to Maestro software) and a phase duration of 60 μs. For objective determination of eCAP thresholds, we applied the previously described "window method" ( Skidmore et al., 2021 ) that calculates the steepest slope of the amplitude growth function. As the stimulation intensity is continuously increased in very small steps in Maestro software, the exported data was analyzed for the steepest slope and its eCAP thresholds were analyzed with a custom-made code in MATLAB version R2021a (MathWorks, Inc., Massachusetts, USA).
Immediately after cochlear implantation surgery, and on followup days 7 and 14, electrically evoked ABR (eABR) thresholds were recorded. Stimulation parameters were provided via the Maestro software and eABR recording was performed with Intelligent Hearing Systems software. Thresholds were determined by decreasing the stimulation intensity from 600 current units (cu, according to Maestro software) to 100 cu for each stimulated cochlear implant electrode (electrodes 2, 6, and 11).

Histology and micro-CT scanning
All extracted cochleae were fixed in 4% paraformaldehyde for 24 h and decalcified with phosphate-buffered 12% ethylenediaminetetraacetic solution (pH 7.4) at 37 °C for at least four weeks. After washing and dehydration of decalcified cochleae, samples were embedded in paraffin with a parallel orientation of the modiolus axis to the paraffin cutting face. Samples were sectioned with a semiautomatic microtome (HM 355 S, Microm, Walldorf, Germany). After trimming each sample until two cochlear turns were visible, 4 μm sections were cut and sequentially collected. Every eighth slide, a trimming step of 60 μm thickness was performed. For each paraffin-embedded cochlea, about 60 slides were stained with Masson's trichrome stain kit (Polysciences Inc., Pennsylvania, USA) to detect the fibrotic reaction after cochlear implantation. All stained slides were imaged with the automated multimodal digital scanner Vectra® Polaris TM (Perkin Elmer Inc.) and each staining was quantitatively analyzed with the image analysis platform HALO (Indica Labs, Albuquerque, USA). Spiral ganglion neurons (SGNs) in Rosenthal's canal of the basal turn were counted in mid-modiolar sections of paraffin-embedded tissues by three researchers in a single-blinded way via Qupath freeware ( Bankhead et al., 2017 ), and all three counts were averaged for each sample (see Supplementary Fig. 2). For histology with the electrode array in situ, methylmethacrylate embedding was performed according to a previously described protocol ( Plenk, 1986 ). In brief, cochleae were fixed, dehydrated in ethanol and subsequently embedded in methylmethacrylate, before grinding and staining of para-modiolar sections with Giemsa stain (Merck, Darmstadt, Germany). For the micro-computed tomography (micro-CT) scans, one minipig ear and one piglet ear were placed in osmium tetroxide solution (1%) overnight.
The sample was dehydrated sequentially, fixed with gauze, and scanned using a Xradia MicroSCT-400 Carl Zeiss x-Ray Micropscopy (Pleasanton, California, USA) with an isotropic voxel resolution of 8.3 μm. Specimens were imaged over a 360 °sample rotation using a 45kVp / 177 μA x-ray spectrum. The scan was imported to the software package AMIRA® 6.0.1 (Thermo Fisher Scientific, Mérignac Cédex, France) and enhanced with bilateral, gaussian and median filter for optimal contrast of measured structures. Each cochlear compartment was segmented manually in several planes to obtain a three-dimensional overview of the minipig inner ear. Surface smoothing of the segmented compartments was performed and subsequently calculated volumetric data were acquired.

Data analysis
Data was analyzed with GraphPad Prism version 8.4.3 (Graph-Pad Software, San Diego, California, USA). Data points represent either mean or individual values, and error bars indicate SEM if not described otherwise in the figure legends. Descriptive statistical analysis was performed using paired t -test after the confirmation of normal distribution with the Kolmogorov-Smirnov test. P-values less than 0.05 were considered statistically significant.

Results
The anatomical differences of the external ear and the temporal bone between humans and pigs require an adaptation of the surgical approach for cochlear implantation in the porcine animal model. A comparably small bony part behind the entrance of the porcine external auditory canal impedes cochlear implant coil placement similar to how it is performed in humans. Additionally, the long and angled external auditory canal in pigs differs from the relatively straight canal in humans (see Fig. 1 ). However, we were able to successfully implant four piglets and six Aachen minipigs with standard MED-EL Flex 20 CIs or MED-EL Flex 24 CIs made for humans. The overall surgical anatomical landmarks were the same in all animals, yet slight variations of round window accessibility due to the varying degree of occlusion by the facial nerve was observed.

Single-incision cochlear implantation -surgical approach
After several approaches had been performed in cadavers and living animals, the transcanal approach proved to be the most di-rect, simple, and fast alternative. The detailed surgical approach is presented in the attached online Supplementary Material (Video 1). As we performed all surgeries on left ears, the pigs were positioned on their right side. The surgeon approaches the animal from the dorsal side to obtain optimal line of sight into the outer ear. With an otoscope, the ear canal is inspected and blocking cerumen is carefully removed. After disinfection and sterile draping of the body, the planned incision is marked approximately one to two centimeters postauricularly in length of about four centimeters (see Supplementary Fig. 3A). After the incision, the soft tissue is gently moved aside, and bleeding can be controlled by cautery. Orientation along the cartilaginous outer ear allows a quick detection of the posterior bony wall in the ear canal. The cartilaginous canal is cut in a cross section at the transition between the cartilaginous and bony part of the ear canal. At this point, sight-restricting cartilage is taken away. In the next step, the skin of the auditory canal is gently separated from the bone with a round knife and carefully shifted towards the middle ear to enable drilling of the posterior and superior bony wall. The posterior part of the tympanic membrane is gently removed to obtain sight of the round window. Depending on the size of the animal, a small part of the inferior wall may be drilled as the auditory canal is steep and forms a bony hill in some animals, which drastically hampers the surgical approach. As the dimensions of the access to the porcine middle ear are relatively limited, the surgical assistant plays a crucial role in optimizing the surgeon's sight. A step-by-step approach with repeated mobilization of the skin (see Supplementary Fig. 3 C-E) and careful drilling finally uncovers a signature landmark, which would commonly appear as a bony groove at the junction of the middle ear and the superior auditory canal wall. After the discovery of this "ramp"-shaped bony part, the skin of the auditory canal is gently mobilized with angled micro-scissors. The extracted skin is stored in sterile and moist gauze for later re-implantation after electrode array positioning. After drilling away excess bone, the facial nerve, the chorda tympani nerve, the short crus of the incus, the malleus and the eardrum are visualized. In some animals hyperplastic middle ear mucosa may be mistaken for a round win- There is no accessible bony part to place the coil of a cochlear implant behind the external auditory canal. (C) Coronal section of a right porcine temporal bone. As opposed to humans, the external auditory canal is noticeably longer and more curved. The mastoid is located beneath the cochlea. Legend: a mastoid, b styloid process, c entrance to the external auditory canal, d border of the linea temporalis, e atlantooccipital joint, f external auditory canal, g cochlea. dow membrane. However, this is found characteristically beneath the angle spanning from chorda tympani nerve to the facial nerve (chorda-facial angle; see Supplementary Fig. 3D). In contrast to humans, the facial nerve has no bony covering, and the identification of the nerve is crucial to access the middle ear. After the supporting bony piece in the chorda-facial angle had been removed, the vascularized round window membrane is unveiled. Furthermore, the chorda tympani nerve may be gently moved inferiorly without damage to facilitate the subsequent CI insertion.
Next, an implant bed is created in the retroauricular soft tissue and the coil of the cochlear implant is fixed before opening a small tunnel into the ear canal, where the electrode array is passed through (see Supplementary Fig. 3B). In this step, any bleeding to the middle ear should be prevented by plugging a sterile gelatin sponge (Spongostan; Ethicon, Raritan, USA) into the ear canal. After gently opening the round window membrane, the CI is inserted until first resistance is felt and up to twenty millimeters into the cochlea. The skin of the auditory canal is then repositioned over the part of the CI electrode closer to the receiver/stimulator and the ear canal is filled with iodine-dipped absorbable Spongostan gelatin sponges. The soft tissue is closed with subcutaneous nonabsorbable Ethibond 2-0 (Ethicon, Raritan, USA) and the skin is sewn with absorbable monofilament 4-0 Prolene (Ethicon, Raritan, USA). Net surgical time amounts to about 1.5 h, while intraoperative measurements took approximately another 30-45 min.
In our series, four minipigs were partially inserted with only the last electrode contact of the Flex 20 CI remaining above the round window entrance. The other two minipigs were fully inserted with the same CI. In contrast, the piglets were implanted with a MED-EL Flex 24 CI, with two full insertions and two partial insertions. In the latter two, the last electrode resided above the round window entrance as well. In all animals, the electrode array position within the inner ear was verified with a radiography immediately after surgery. All inserted cochlear turns were visible in the x-ray (see Supplementary Fig. 4A and 4B). Finally, a custom-made jacket was put on the animals to cover the wound and to carry the battery for a speech processor after an acclimatization period of two weeks. At the end of two weeks post-surgery, a 3D-printed grid was sewn above the speech processor, which magnetically attached to the implant coil. Grid and collar of the jacket both protected the implant from mechanical trauma (see Supplementary Fig. 4D). X-ray scans were repeated at days 14 and 56 after cochlear implantation. While at day 14 all electrode arrays remained in the cochlea, the x-ray scans at day 56 revealed that the electrode arrays had moved toward the middle ear in all minipigs.

Click and tone burst auditory brainstem responses
Baseline tone burst ABR thresholds varied between 30 and 80 dB sound pressure level (SPL) in all six minipigs within all measured frequencies. Click ABR thresholds ranged between 15 and 30 dB SPL pre-surgery (27 ± 3 dB SPL, mean ± SEM). Fig. 2 A shows mean ABR threshold values before and after surgery. We observed a lower hearing threshold level in the low frequencies (0.5 -8 kHz, range 30-35 dB SPL) compared to higher frequencies (16 -32 kHz, range 60 -80 dB SPL). One week after surgery, all animals were functionally deaf as confirmed with isolated thresholds found between 95 and even up to 120 dB SPL in some animals. Overall hearing improved until the last day of follow-up (day 56) with a hearing threshold of around 80 dB SPL in the best animals. Although we observed an increasing number of detectable thresholds over time, residual hearing at the end of experiments was very low. BC-ABR confirmed this finding, as baseline mean threshold was 35 ± 2.2 dB and follow-up measurements revealed no recordable thresholds with an exception at day 56 in one animal (60 dB SPL).

Cochlear health assessment
Prior to cochlear implantation, we assessed acoustic CAP hearing thresholds with a gold wire electrode (PromStim electrode, MED-EL) in five minipigs. The mean CAP threshold was 86 ± 3 dB SPL (individual values: 80, 80, 85, 90, and 95 dB SPL). Due to time restriction during surgery, the first operated minipig was spared from CAP evaluation. CMs were successfully assessed with the gold wire electrode after acoustic stimulation with 110 dB SPL. After the insertion of the first 3 to 5 electrode leads of MED-EL Flex20 cochlear implants, CMs were again measured, this time via the implant electrode array. Among six animals, four demonstrated recordable potentials. This measurement was repeated after full insertion but subsequent CM thresholds could not be detected in any of the animals.
At the conclusion of surgery, we were able to determine eABR thresholds in all animals by consecutively stimulating electrodes 2, 6, and 11. The initial eABR threshold amplitude ranged between 200 and 400 cu. While one minipig had an implant dysfunction (possibly caused by self-induced implant damage during the time at the housing facility), all other animals were included in follow-up eABR threshold evaluation. At day 7, eABR thresholds were still successfully recorded, but currents increased until day 14 (see Fig. 2 B). To represent a synchronized response of electricallyactivated auditory nerve fibers, eCAPs were analyzed. Supplementary Fig. 5 represents all measured eCAP thresholds. We observed a mean increase of 4.58 qu (56.51%), 5.71 qu (188.70%) and 4.46 qu (35.08%) between surgery and 7 days later in electrodes 1, 5, and 10 of the CI, respectively. Two weeks after surgery, only three animals showed representative eCAP thresholds and a representative mean threshold evaluation was not possible as the threshold range varied between 1.56 and 16.55 qu. Fig. 3 represents all impedance measurements at different time points. Although we measured higher impedances in the first three (apical) electrodes intraoperatively compared to other (middle and basal) electrodes of the cochlear implant, the apical impedances two weeks after surgery were lower than the ones measured in the middle or basal turn. Acquired mean impedance increased over time in all subjects as indicated in Fig. 3 A (baseline 7.30 ± 0.90 k , day 7 10.64 ± 1.88 k , and day 14 19.40 ± 1.28 k ).

Fibrosis development and volumetric data
All implanted piglets were observed until day 7 or 14 after CI insertion and implants remained within the porcine cochleae until euthanasia. Implants in minipigs had accidentally dislocated by day 56 due to the movement of the animals. After euthanasia and fixative perfusion as described above, all cochleae were extracted from piglets and minipigs for further histological analysis. During the removal of the minipig cochleae, we observed that the surgically reinserted skin of the ear canal healed very well, and no inflammation was visible. Among all extracted cochleae, four implanted and three contralateral non-implanted cochleae of minipigs were decalcified and embedded in paraffin. Staining with Masson's Trichrome is presented in Fig. 4 and shows the dimensions of the capsular fibrous sheath with lymphocyte collections just around the original position of the electrode array. The size of the capsule in comparison to the actual electrode array dimension appears to be reduced as indicated in Fig. 4A/B, which can either be a result of the continued remodeling after dislocation of the electrode or the postmortem tissue processing. Morphological damage was present in all implanted cochleae, especially in the basal turns. An example of an ipsi-and contralateral cochlea is also shown in Fig. 4 . Here, a hydrops of the scala media was detected, but as expected, despite the induced damage, we observed no significant reduction of SGN density in Rosenthal's canal as indicated for the basal turn (Supplementary Fig. 6).
One implanted (Flex 24, MED-EL) piglet cochlea was stained with Giemsa stain after methylmethacrylate embedding to acquire approximate data on electrode array location. Fig. 5 shows electrode in-situ histology with the implant covering the first two cochlear turns. This observation corresponds to the measured diameter of the scala tympani within the mid-modiolar sectioned micro-CT scan; the smallest diameter measured in the midmodiolar micro-CT section was 0.3 mm within the first two turns (see Fig. 6 ). Further volumetric data are presented after segmentation of the contralateral (right) minipig cochlea (see Fig. 7 ). The total inner ear volume comprised approximately 40 mm ³ with scala tympani (13.54 mm 3 ) and vestibuli (12.27 mm 3 ) covering most of it. The overall basilar membrane length measured 27.54 mm (Supplementary Fig. 7). Both segmented micro-CTs were compared with common non-human primate and human cochlear measurements of previous studies, which are presented in Table 1 .

Adverse events
While all animals tolerated the surgery well, post-operative wound infection with abscess development was observed in some animals despite prophylactic antibiotic medication. One piglet and three minipigs suffered from abscess formations in the area of the retroauricular CI electrode coil one week after surgery. Said piglet was euthanized at that point due to veterinary advice. Additional antibiotic treatment was given in cases of infections and all animals were monitored regularly by professional veterinarians for their well-being. All affected minipigs responded well to the antibiotic treatment and recovered quickly.   ( Dai et al., 2017 ), d ( Kirk and Gosselin-Ildari, 2009 ), e ( Marx et al., 2013 ), f ( Felix, 2002 ), g ( Felix, 2002 ;Marx et al., 2013 ;West, 1985 ).

Discussion
There is an indispensable need for large animal models in translational hearing research, specifically when focusing on CIs. Several hurdles have to be overcome when trying to establish such an animal model for cochlear implantation with humanlike dimensions. These are mainly of ethical, financial, and administrative nature. The pig seems to be an ideal candidate and a compromise compared to other species; it is available almost everywhere due to high supply and demand, is relatively easy to handle and has a large human-like cochlea. In long-term experiments, minipigs seem to be a better alternative due to their lower weight gain in comparison to conventional domestic pigs. Based on our results, pigs are suitable animals for CI research as we were able to insert commercially available CIs (Flex24 and Flex20 CIs, MED-EL) in both piglets and minipigs and perform functional assessments over a period of up to two months.

Macro-and micro-anatomy and surgical approach
There are clear differences in the outer ear anatomy between pigs and humans. Important anatomical discrepancies include the pinna, the external auditory canal with a contiguous anteriorly located temporomandibular joint and the ventrally located mastoid with close relation to the atlantooccipital joint (as indicated in Supplementary Fig. 1 B-C). However, as the middle and inner ear are approached, the anatomy resembles more the human setting and dimensions ( Gurr et al., 2010a ;Zhong et al., 2018 ). Herein, we present the (to our knowledge) first step-by-step surgical approach for cochlear implantation in piglets and Aachen minipigs with commercially available CIs used in humans. The detailed methodology lays the foundation for various experiments in hearing research, specifically as it appears to be safe and reproducible. However, it should be mentioned that preliminary practice with pig cadavers is recommended to optimize the steps, reduce operative time and increase the likelihood of a successful implantation.
In regard to cochlear dimensions, the micro-CT scan of the minipig inner ear revealed an overall inner ear volume of approximately 40 mm ³ and the measured minipig basilar membrane length of 27.54 mm is within human normative data (28 -40.1 mm, ( Ulehlova et al., 1987 )), considering we excluded the most apical part of the minipig cochlear measurement due to perfusion hole drilling. Additionally, the diameters of the first cochlear turn provide a representative model of the human inner ear. Our three-dimensional volumetric data of the minipig scala tympani (13.54 mm ³, Table 1 ) exceeds even the scala tympani volume of some non-human primate models such as the macaca fascicularis, which covers around 4.15 mm ³ ( Manrique-Huarte et al., 2021 ). Although, the inner ear volume of rhesus macaques is larger (59.4 μl) ( Dai et al., 2017 ) than the scanned minipig's in our series (39.96 μl), it is still smaller than the measured piglet inner ear (81.10 μl, Supplementary Fig. 8). Considering that the human inner ear is even larger (191 μl) ( Dai et al., 2017 ), the porcine animal models seem to mimic the human cochlear size at least in a similar way to commonly used non-human primate models.

Induced trauma -electrophysiology and histology
Intraoperative acoustic CAPs were detectable in all animals before CI insertion and the gold wire electrode established for human patients under clinical conditions (PromStim, MED-EL) can be placed in an optimal alignment to the round window niche to generate reproducible results. Electrocochleography (CAPs and CMs) responses via the inserted cochlear implants after partial insertion were present in half of the cases. To our knowledge, these electrophysiological measurements are the first in Aachen minip- igs. We could show the general feasibility of these measurements in the minipig model with commercially available CIs. eABR and eCAP thresholds were evaluated within the first two weeks after surgery. Increasing thresholds were observed and may correlate with an increasing immune reaction caused by the implant. In some cases, no eCAP thresholds were found at day 14 after implantation as it is shown in Supplementary Fig. 5. There are some potential explanations for this finding, one of them being an electrode array dislocation. However, we did not observe changes in the electrode array positions in x-rays performed two weeks after surgery. Therefore, a dislocation on the same day during eCAP measurement seems rather unlikely. Also, technical issues such as implant failure or device malfunction can be excluded as all devices were tested and their functionality was approved before implantation, and the recorded impedance measurements during the follow-up revealed normal device function in all animals (with the exception of the one implant failure mentioned above). Alternatively, the marked increase or even loss of eCAPs could be due to the significant fibrosis, which could develop excessively during the second week after implantation. Therefore, further adjustments to the experimental procedure may improve cochlear health results after successful surgery, leading to an overall better outcome in hearing performance.
In addition to establishing the surgical approach, we characterized the induced functional trauma with multiple electrophysiological measurements. Interestingly, we found lower baseline click ABR thresholds in Aachen minipigs when compared to other previously published pig models ( Bergler et al., 1996 ;Wilkes et al., 1989 ). In contrast, baseline tone burst ABRs are consistent with the previously measured minipig breeds from Asia ( Guo et al., 2015 ;Yi et al., 2016 ). Two functionally deafened animals had click ABR thresholds of 95 dB SPL seven days after CI insertion, and in the remaining four minipigs no recordable thresholds were found. This is in contrast to the recorded 77 ± 3 dB SPL presented by Yi et al., which might be due to a more shallow insertion in the animals included in this earlier publication ( Yi et al., 2016 ). Hearing thresholds did not change two weeks after cochlear implantation, but improved at day 56 in follow-up audiometry tests. In this context, it must be mentioned that the postoperative negative BC-ABRs were likely due to the limitation to the relatively low output capacity of bone oscillators, which had been observed in other studies ( Seo et al., 2018 ). Interestingly, tone burst ABRs revealed hearing improvement especially in the frequency range of 20 kHz on days 14 and 56 after implantation, even though higher frequencies are functionally correlated to the basal turn. At the latter time point, hearing recovery was visible in all frequencies, but at a low level. The hydrops development in the histologically assessed inner ear ( Fig. 4 ) might be one reason for the massive hearing loss, however, it cannot be defined with certainty when the hydrops occurred. Similarly, as mentioned above, the reduced size of the capsular fibrous sheath as depicted in Fig. 4 B may either be explained by the continued remodeling after dislocation of the electrode or the postmortem tissue processing. The fact that we did not observe a reduction in SGN numbers in the implanted ears as compared to the contralateral, non-implanted control ears ( Supplementary  Fig. 2) merits further investigation. A potential explanation for this finding is, that the degeneration of SGNs in pigs requires a longer period of time. Even though longitudinal data on SGN degeneration in pigs is lacking at the moment, a slower degeneration of SGNs in pigs as compared to rodents would not come as a surprise as this has already been described for other large animal models like the cat and also for humans ( Araki et al., 20 0 0 ;Zimmermann et al., 1995 ).
Regularly performed radiographies of the implanted temporal bones did not show any electrode array shift within the first two weeks after cochlear implantation. For the sake of animal well-being, additional sedation of minipigs after that time point until day 56 was not performed. Before euthanasia, final x-rays showed displacement of cochlear implants in all animals. Postoperative infection or mechanical trauma may be potential reasons for CI displacement. We could see fibrosis in histological sections, especially in the basal cochlear turn. Although the organ of Corti seemed to be damaged in the most basal part of the cochlea as indicated in our mid-modiolar sections (see Fig. 4 ), further apical sites revealed intact structures. This observation is consistent with the obtained micro-CT data in our study. In accordance with the measured minimum scala tympani diameter within the first two cochlear turns of the representative minipig cochlea, an electrode array insertion with a tip diameter up to 0.3 mm is feasible (see middle turn measurements in Fig. 6 ). As the Flex 20 CI electrode array covers the measured dimensions with a length of 20 mm and a tip diameter of 0.3 mm, we assume that the insertion did not induce excessive trauma.

Limitations of the study
One limitation for evaluation of residual hearing after cochlear implantation of the study may be the inserted gel sponges (Spongostan), which help padding the external ear canal but might stop the transmission of acoustic stimuli to the middle ear and the cochlea. Yet, while potentially relevant for some of the performed tests, the results were also confirmed with BC-ABRs, which should not have been affected by conductive hearing loss. While these bone conduction levels were not calibrated for the porcine species and the device was used with a pre-shipped calibration for humans, pre-and post-insertion BC-ABRs demonstrated threshold shifts verifying the acquired sensorineural trauma. Another important fact is that during our long-term experiment, the CI electrode array was dislocated from the inner ear at some point between the first and second month of follow-up. This conclusion was drawn as the x-ray on day 14 presented implanted electrode arrays within the cochlea and the same imaging technique on day 56 showed dislocated electrodes. In this sense, an adaption of the electrode fixation technique must be performed and is currently a limiting factor for long-term experiments. Despite electrode dislocation, we acquired important information on the induced trauma and improvement of hearing function over time. These insights lay the foundation in the scope of further experiments and the obtained results present the minipig as a suitable large animal model in CI research.

Conclusion
With the proposed step-by-step surgical approach, we show the general feasibility of cochlear implantation in piglets and minipigs. Partial or full insertions up to 20 mm with commercially available CIs used in humans are possible in both animal models. Furthermore, we present the induced trauma in both, electrophysiological measurements, and histological analysis over a time period of up to 56 days after CI surgery. With consistent anatomical landmarks between animals and structural similarities to humans, the surgical approach and overall cochlear implantation in this animal model is an advanced representation of both, the clinical surgery, and the induced trauma in large cochleae. Based on similar cochlear dimensions and a comparable hearing range to humans, we recommend this animal model as a possible alternative for the conduction of future experiments in translational hearing research.

Source of funding
Christoph Arnoldner and Clemens Honeder receive funding from MED-EL Corporation, Innsbruck, Austria. Erdem Yildiz, Matthias Gerlitz and Anselm Joseph Gadenstaetter were financed by this funding while conducting the studies. Dominic Schum and Martin Schmied received a research grant by MED-EL. Rudolf Glueckert received funding from MED-EL. Lukas D. Landegger receives funding from Decibel Therapeutics, Boston, USA and Amgen, Thousand Oaks, USA and has worked as an independent consultant for Gerson Lehrman Group and Conclave Capital.

Declaration of Competing Interest
The other authors declare no conflicts of interest.

Data availability
The original data presented in the study are included in the article.

Acknowledgments
We thank Johanne Charlott Krueger (University Hospital Aachen, Germany) who provided cadavers from previous research projects to prepare and investigate the surgical approach in Aachen minipigs. We also thank Jochen Tillein for the initial pictures of the pig cadaver model. Intraoperative microscopic photos were taken with the Haag-Streit surgical microscope HS 5-10 0 0, which was provided by Martin Romann (Neumed AG, Harmannsdorf, Austria). A special thanks to Stephan Handschuh for his advice on micro-CT scanning and segmentation methods of the cochlear structures. Additional thanks go to Josef Yu for his advice on HALO software and video illustration, and to the technical support from MED-EL Corporation. The graphical abstract of this manuscript was created with BioRender.com.

Meeting information
Preliminary results have been presented in a poster presentation at the 45th Annual ARO 2022 MidWinter Meeting virtual conference organized by the Association for Research in Otolaryngology and in a presentation at the 65th Annual Meeting of the Austrian Society of Otorhinolaryngology, 22nd -26th of September 2021, Austria.

Supplementary materials
Supplementary material associated with this article can be found, in the online version, at doi: 10.1016/j.heares.2022.108644 .