Intra‐cochlear Flushing Reduces Tissue Response to Cochlear Implantation

Intraoperative trauma leading to bleeding during cochlear implantation negatively impacts residual hearing of cochlear implant recipients. There are no clinical protocols for the removal of blood during implantation, to reduce the consequential effects such as inflammation and fibrosis which adversely affect cochlear health and residual hearing. This preclinical study investigated the implementation of an intra‐cochlear flushing protocol for the removal of blood.


INTRODUCTION
4][5][6][7][8] For cochlear implant recipients, this could result in not only transient immediate hearing loss, but also further permanent loss of residual hearing.
There are currently no protocols clinically approved for the removal of blood or other unwanted substances, such as bone dust, entering the cochlea during implantation.The introduction of hyaluronic acid polymers, such as Healon ® , are widely practiced by surgeons to remove bone dust; however, this approach has not been validated, nor implemented with a view to preserving residual hearing.Therefore, a more proactive approach in removing the blood intraoperatively may help undertake this.
Here we developed and validated a flushing mechanism to remove blood during cochlear implantation.The study aimed to model clinical cochlear implantation with insertional trauma occurring, resulting in the infiltration of blood into the cochlea.The novel aspect of this study involved implementing a flushing mechanism to remove the blood while preserving residual hearing.Hearing thresholds and the volume of the tissue response constituents were compared between three subsets of animals: cochlear implant only; cochlear implant with blood injection; and cochlear implant, blood injection followed by flushing.This study design aimed to evaluate the efficacy of the flushing method in removing the blood and whether the intervention impacted the hearing and tissue response.

Experimental Design
This study was carried out in accordance with the St Vincent's Hospital (Melbourne) Animals Ethics Committee guidelines (project AEC 010/20) and adhered to the guidelines of the National Health and Medical Research Council of Australia.
The aim of this study was to examine the efficiency of flushing intra-cochlear bleeding to minimize the detrimental effects on auditory function and cochlear mechanics.To do this, the study had to emulate clinical cochlear implantation whereby bleeding may occur in the cochlea, and if it does, a flushing protocol is initiated.Thus, 30 adult tri-colour Dunkin-Hartley guinea pigs underwent cochlear implantation and observation across a 28-day period following surgery.The animals were divided into three experimental groups: control, cochlear implantation only; blood, cochlear implantation with an injection of 7 μL of blood into the cochlea during surgery; flush, cochlear implantation, an injection of 7 μL of blood into the cochlea during surgery followed by an injection of artificial perilymph into the cochlea to act as a flushing mechanism to remove the blood.The rationale for the control group undergoing cochlear implantation was to mimic an ideal implantation where no trauma is detected and compare the auditory function as well as tissue responses to the flush group to determine the success of the flushing.All the guinea pigs chosen were under 6 months of age and weighed <1 kg.All cochlear implantations were performed on the left ear of the animal, whereas the right ear was treated as an in-animal control.Animal allocations were contingent with implant failures to obtain a comparable amount of data in all three groups.
Animals were monitored for 28 days post-surgery.Auditory brainstem responses were measured prior to surgery, immediately post-surgery, on day 7 and day 28 (final day) for both ears.The cochleae of all animals in the study were harvested and optically cleared for light sheet microscopy (LSM) to analyze the tissue response within the cochlea.The animals were not electrically stimulated throughout the experiment.

Cochlear Implant Arrays
The cochlear implants were designed and produced by an in-house technician.The array consisted of four full-banded intra-cochlear platinum ring electrodes.The maximum ring diameter was 0.43 mm, narrowing to 0.33 mm toward the tip, with an interelectrode separation of approximately 0.4 mm.An extra-cochlear platinum ball electrode was also attached to the array.Each platinum electrode was connected to an internal 25 μm diameter Teflon-insulated platinum-iridium (90/10) wire, connected to Teflon-insulated stainless-steel lead wires, embedded in a silicone carrier.Gold pins were attached to the end of the lead wires and inserted into an 8-pin rectangular receptacle connector (Harwin Inc., UK) for connection to a laptop.

Auditory Brainstem Responses
Auditory brainstem responses (ABRs) were chosen to test the auditory function of the guinea pigs in this study.The technique used for obtaining ABRs has previously been described. 9he acoustic stimuli were computer-generated tone pips for a duration of 5 ms with a 1 ms rise/fall times, and presented at frequencies: 2, 8, 16, 24, and 32 kHz.The intensity of stimulation began at 90 dB and decreased by 5 dB increments until no waveforms were present in the signal via visual inspection.The resulting waveforms were evaluated using a custom-written program in Igor Pro (version 6, WaveMetrics Inc., USA) and analyzed with the threshold defined as the lowest stimulus intensity to evoke a wave III response >0.4 μV.

Cochlear Implant Surgery
All guinea pigs underwent anaesthetic induction using a 3:1 mixture of ketamine (60 mg/kg, Troy Laboratories Pty Ltd, Sydney, Australia) and xylazine (4 mg/kg, Troy Laboratories Pty Ltd, Sydney, Australia) injected intramuscularly (quadriceps), and glycopyrrolate (0.02 mg/kg) was administered subcutaneously to reduce mucosal secretions.A post-auricular incision was made, the muscle overlying the bulla retracted, and a 2-2.5-mm diameter superior bullostomy was made to expose the basal turn of the cochlea.An $0.8-1-mm cochleostomy was made approximately 1 mm from the round window niche.The array was then carefully inserted until resistance was felt.
For the blood and flush groups.Blood was collected from the ear veins (approximately 7 μL) and immediately injected into the same cochleostomy used for electrode insertion, using a 24G catheter placed 2 mm inside the cochleostomy, prior to implantation.The blood was injected manually by the surgeon and was performed in under 1 min.
For the flush group.A 24G catheter was placed 2 mm inside the cochleostomy site used for insertion, where 37 C artificial perilymph was injected by a syringe pump to flush the blood with a flow rate of 30 μL/min for 1 min.The artificial perilymph used for flushing is described in Salt et al. 10 It contained NaCl 125, KCl 3.5, NaHCO 3 25, MgCl 2 1.2, CaCl 2 1.3, NaH 2 PO 4 0.75, and C6H12O6 5.0 and was diluted in 500 mL distilled water.In vitro experiments were performed prior to the animal study to determine a sufficient flow rate to remove blood from a guinea pig shaped phantom cochlea.
The cochleostomy for all guinea pigs was sealed around the array with a fascia plug to prevent perilymph leakage.The implant was secured at the bullostomy site using carboxylate cement (Durelon™, 3 M ESPE AG, Seefeld, Germany) and the muscle and skin over the bulla was then sutured with 3-0 coated Vicryl (polyglactin suture; Ethicon, Johnston & Johnston Medical Pty Ltd, Australia) and lightly sprayed with water-resistant dressing spray Opsite (Smith Nephew, Medica Limited, U.K.).

Perfusion and Tissue Processing
The technique for cochleae harvesting and tissue processing has been previously described. 11On Day 28, animals were inducted with a 2:1 mixture of ketamine (40 mg/kg) and xylazine (4 mg/kg) via intramuscular injection, final ABRs were conducted, followed by euthanasia using an overdose of sodium pentobarbital (>300 mg/kg).The animals were perfused intracardially using 500 mL of 0.9% normal saline at 37 C followed by 500 mL of 10% neutral buffered formalin at 4 C, and their cochleae were collected.
Once removed, the cochleae were placed into 10% neutral buffered formalin overnight at room temperature.The cochleae were rinsed with phosphate buffered saline (PBS) and then decalcified for 2.5 weeks in regular (2 days) of 10% (w/v) EDTA at room temperature.Permeabilization with dimethyl sulfoxide (DMSO) was carried out for 10 minutes three times, followed by three 20-minutes rinses in PBS with agitation.

Immunolabeling and Whole Cochlea Imaging
The cochleae were incubated in the primary antibodies for 3 days, agitating at 37 C.The labels used in this study are detailed in Table I.The cochleae were washed three times in 20 mL of PBS-T for 20 minutes at 37 C and then incubated in the secondary antibodies overnight and washed again (Table II).Finally, the cochleae were dehydrated in ethanol (50%, 70%, 100%, and 100%) for 12 hours at each at 4 C and placed in ethyl cinnamate overnight at room temperature.
Myosin VIIa is a protein found in hair cells.Ionized calcium-binding adaptor molecule 1 (IBA1) is a microglial and macrophage-specific calcium-binding protein involved in phagocytosis.Smooth muscle actin (SMA) is a marker for the myofibroblasts, responsible for the production and organization of granulation tissue in contracting wounds and in angiogenesis. 12Fibronectin is a glycoprotein present in the extracellular matrix of the tissue response, regulating cell adhesion and motility.Lastly, Huc/HUD is a label for spiral ganglion neurons; however, this data were not analyzed for this study.
To image the cochlea, Ultramicroscope II (LaVision, BioTec, Bielefeld, Germany) with an Olympus MVPLAPO 2Â lens (NA: 0.5) attached to a stereomicroscope (MVX10 with Zoom body 0.63Â-6.3Â)was used to achieve a final magnification range from 1.26Â to 12.6Â.Diode lasers as specified in Table III were captured by a sCMOS camera (Andor Neo) at 250 ms exposure.
A surface for the implant was manually created by shadowing in the slice view and segmented as a 3D volume.Using the same technique, a surface of the total tissue response within the scala tympani was created, made visible by being labeled with IBA1, fibronectin, and SMA.Within the total tissue response area, 3D surface reconstruction of each immunolabel was performed using an automatic creation wizard within Imaris, using an algorithm based on Ridler 13 and background subtraction, resulting in rendered 3D surfaces of extra-cellar tissue response (fibronectin), myofibrotic tissue (SMA), and mononuclear phagocytes (IBA1).Specifically, for IBA1, previous studies have shown that predominant IBA1 positive cell population in the cochlea are resident and active macrophages. 14

Implant Position
A frequency map was generated to determine the position of the implant with respect to frequency.To do this, Myosin VII labeling of the hair cells was used to trace the Organ of Corti and calculate the frequency map.Using the 3D images, whole cochleae were imaged at magnification 1.25Â, where the formation of hair cells was clearly visible. 11Hair cells were traced as a 3D 'filament' object using the 'Filament tracer' module of Imaris.Each voxel point of the filament was given a spot object using 'Filament Analysis' XTension, a MATLAB plugin available for Imaris™ XT module.The x, y, z positions of each point were exported, and their accumulative distance from base was calculated in three dimensions.The frequency map was calculated using Equation 1 15 : where %d corresponds to percent distance from base of the cochlea and kHz represents a characteristic frequency.
The points along the filament closest to the implant tip and the implant closest to the cochleostomy site were found via an Imaris extension named 'Find the spot closest to a surface', a MATLAB plugin available for Imaris™ XT module.The frequencies corresponding to these spots were recorded.

Statistical Analysis
All statistical analyses were performed using R Statistical Software (version 4.04, 16 packages: emmeans). 17A series of linear models, specifically multiple one-way ANOVA models, were fitted to the ABR thresholds and were compared between experimental groups across both time and frequencies.The experimental group was treated as a categorical variable.The thresholds obtained prior to surgery were treated as a covariate for this analysis, instead of an outcome, to assess the difference in the post-surgery time points while accounting for the pre-surgery   thresholds.For the cochlear imaging data, the rendered volumes for all immunolabels in addition to the total tissue response were quantified and compared between experimental groups.A linear model was fitted to each immunolabel, whereas the experimental group was treated as a categorical variable.

Electrophysiology
Thirty animals were used for this study (control: 8, blood: 13, flush: 9).The unequal group numbers were due to animal death or implant failure (n = 13, control: 3, blood: 7, flush: 3), resulting in 17 animals totally (control: 5, blood: 6, flush: 6) that survived the full experimental period.Due to the pre-surgery thresholds being treated as a covariate, Figure 1 shows the average thresholds for all experimental groups after surgery, day 7 and day 28 measurements for frequencies of 2, 8, 16, 24, and 32 kHz.Immediately after surgery, thresholds were similar for all groups for frequencies 2, 8, and 32 kHz.For 16 kHz, the blood group had an average threshold of 75 dB, higher than both the control (mean = 12 dB, 95% CI = 25 to À3 dB, p = 0.12) and the flush group (mean = 17 dB, 95% CI = 33 to 1 dB, p = 0.04).Similarly for 24 kHz, blood had the highest threshold, averaging at 80 dB, and was statistically higher than the control group (mean = 10 dB, 95% CI = 20 to 1 dB, p = 0.04) and flush group (mean = 15 dB, 95% CI = 26 to 5 dB, p < 0.01).
At day 7, the blood and control groups had similar thresholds for all frequencies, with a marginally higher threshold for the blood group at 8 kHz (mean = 6.65, 95% CI = 26 to À13 dB, p = 0.48).The flush group had the lowest thresholds for all frequencies, specifically for 16 kHz as the threshold was lower than the blood group (mean = À18 dB, 95% CI = À37 to 2 dB, p = 0.07), and the control group (mean = À20 dB, 95% CI = À39 to À1 dB, p = 0.04).Similar trends were observed at day 28 as the flush group maintained significantly lower thresholds than the control group for 16 kHz (mean = À18 dB, 95% CI =-33 to -2 dB, p = 0.03) and 24 kHz (mean = À21 dB, 95% CI = À35 to À7 dB, p < 0.01), as well as the blood group at 24 kHz (mean = À20 dB, 95% CI = À34 to À7 dB, p < 0.01).For the contralateral ear, there were no statistical differences between the experimental groups at any frequency nor time point.

Tissue Response
Whole cochlea immunofluorescence was performed on 17 experimental animals (specifically, control: 5, blood: 6, flush: 6), whereas the remaining 13 animals either did not complete the experimental period or the cochleae were damaged during the processing.
Illustration of the whole cochlea imaging is shown in Figure 2. The total tissue response volume in the cochlea was calculated (refer Fig. 2B), in addition to the total volume of the tissue immunolabelled for IBA1, fibronectin, and SMA (refer Fig. 2C).The averages of these volumes for all experimental groups are shown in Figure 3, along with their respective 95% confidence intervals.The blood group had the highest volume for total tissue response and the immunolabels IBA1, fibronectin, and SMA.The flush group had the lowest volumes across labels, barring SMA, and its total tissue response was significantly lower than the blood group (mean = 1.1e08 μm 3 , 95% CI = 1.9e07 to 2.1e07 μm 3 , p = 0.02) and IBA1 (mean = 5.3e07 μm 3 , 95% CI = 1.7e06 to 1.1e08 μm 3 , p = 0.04).Additionally, the control group was significantly lower than the blood group for total tissue response, IBA1 and SMA (All p < 0.05).There was no statistical difference between the flush and control groups.
Further analysis was performed on the total tissue response to determine if significant differences were observed at specific regions along the array between experimental groups.To do this, all surfaces created from the light sheet images for each animal were exported as slices.The images were then compiled into a z-stack for each animal and divided into 10 segments corresponding to equal divisions of the implanted array.For these 10 segments, the average area of the tissue response was Fig. 1.Auditory brainstem response thresholds.Thresholds are divided based on their frequency, experimental group, and the day which they were measured, with frequency plotted on the x-axis.The thresholds from the auditory brainstem responses prior to surgery were treated as a covariate rather than an outcome for this analysis.The average and 95% confidence intervals are shown.calculated and compared between experimental groups.The results are shown in Figure 4A, with the average cross-sectional area displayed for each segment along the array, with 1 being the most basal and 0 being the most apical.
The blood group showed the largest tissue response in majority of the segments, particularly in the basal half of the array.From the most basal segment, the average tissue response area for the blood group gradually increased, reaching a maximum average of 4.9e05 μm 2 at segment 0.6, to then steeply decrease until the most apical segment.The control group followed a similar trend as the blood group, although reaching a maximum at segment 0.9 and decreases thereafter.After segment 0.8, the areas for the control group were significantly lower than the blood group until the most apical segment (all p < 0.01).
Interestingly the flush group showed a different trend to the former groups, by lacking a local maximum and exhibiting a plateau from segments 0.9-0.4,averaging an area of 2.5e05 μm 2 .Between segments 0.7 and 0.9, the flush group was significantly lower than both the blood and control groups (all p < 0.01), and significantly lower than the blood group for segments 0.4 and 0.5 (both p < 0.01).From segments 0.4 to 0.0 (most apical), the flush group closely mimicked the blood group in magnitude.
The varied distribution patterns drew our attention to the basal half of the array.The average tissue response area for the basal half of the array was calculated for each experimental group and is depicted in Figure 4B.
Implant position was analyzed to determine if insertion depth affected the tissue response.A frequency map was generated for every animal, with the frequency closest to the implant tip and most basal implant region  (at the cochleostomy site) recorded and compared between experimental groups (Fig. 5).The results showed no statistical difference at either location across experimental groups (tip: p = 0.58, cochleostomy: p = 0.48).The frequency corresponding to the location of the implant tip and cochleostomy site was on average 10.2 and 26.7 kHz, respectively.Referring these frequencies to Figure 4A, 24 and 16 kHz would reside within the basal half of the array, given the logarithmic nature of cochlea topography.Interestingly, the lower ABR thresholds for the flush group at these frequencies coincide with less tissue response in these regions when compared with blood and control groups, suggesting the tissue response affected auditory function.

Clinical Relevance
The emergence of potential biomarkers of the cochlear environment during surgery, and in particular, the possibility that intra-cochlear bleeding might be detected during surgery through impedance measurements, 18 has heightened the need for reliable methods of clearing scala tympani during implant surgery.It is envisaged that clearance of scala tympani might be triggered by a biomarker, with the hope that downstream consequences of intra-cochlear bleeding, such as more florid inflammation, and more exuberant fibrosis, may be prevented.A further consideration is the need for methods of cochlear flushing to not adversely affect residual cochlear function or structure.The main rationale of clearing blood and/or bone dust is to reduce the impact of consequential effects on cochlear implant function, which includes residual hearing.It is therefore important that the clearance/flushing method proposed has minimal impact upon residual acoustic function.
The cochlear flushing method adopted here succeeded in reducing downstream consequences of the presence of intra-cochlear blood in association with cochlear implantation, as evidenced by a reduction in the volume of tissue response.In fact, there was less fibrosis than in control animals around the basal half of the implant electrode.Flushing had little impact on ABR thresholds intraoperatively but was associated with better thresholds 1-4 weeks after surgery in the mid frequencies.Therefore, it appeared that cochlear flushing was also beneficial in the control group.

CONCLUSIONS
Here we suggest that flushing during cochlear implantation may be a useful surgical tool to reduce the tissue response post-surgery, both in routine implantation and to clear an intra-cochlear bleed.The ABR results suggest that the intra-cochlear pressure presented via the flushing in this study did not affect the structure of the cochlea and improved auditory function.

Fig. 2 .Fig. 3 .
Fig. 2. Immunofluorescence labeling of an implanted cochlea.(A) Whole cochlea imaging with the surface of the implant visualized in grey.The three immunolabels: ionised calcium-binding adaptor molecule 1 (IBA1, Green); fibronectin (Red); smooth muscle actin (SMA, Blue).(B) The implanted electrode surrounded by the total tissue response volumes (pink).(C) Surfaces of the individual immunolabels in the tissue response.

Fig. 4 .
Fig. 4. (A) The average cross-sectional area of the tissue response along the electrode array.The array was divided into 10 equal segments with the average area of the tissue response calculated for each experimental group.The 95% confidence intervals are also shown.(B) The average area of the tissue response for the basal half of the array.The averages for each experimental group are shown along with the 95% confidence intervals.

Fig. 5 .
Fig. 5. (A) An example of a frequency map generated for an implanted cochlea.The Myosin VIIa labeling is used in Imaris to trace the hair cells to construct a frequency map.The map is constructed of points which can be used to determine which frequency is closest to the regions of interest on the implant.(B) The average corresponding frequency at the tip of the implant.The averages for each experimental group are shown along with the 95% confidence intervals.(C) The average corresponding frequency at the most basal region of the implant at the cochleostomy site.The averages for each experimental group are shown along with the 95% confidence intervals.

TABLE I .
Primary Antibodies.

TABLE II .
Secondary Antibodies.