The Institutional Animal Care and Use Committee at North Carolina State University (NCSU IACUC #22-118) approved this study. All methods were conducted according to the national guidelines under which the institution operates, and NIH Guidelines for the Care and Use of Laboratory Animals (8th edition). Ten, 2–4 weeks-old female Yorkshire piglets were included in this study. The mean animal body weight (BW) was 6.1 ± 1.52 kg (mean ± SD) at the time of surgery. Piglets had food withheld for 1–2 h prior to sedation and induction of anesthesia. For initial sedation, (TKX, 1.6–3.9 mg/kg body weight [BW] intramuscularly [i.m.]) a mix of tiletamine (50 mg/mL), zolazepam (50 mg/mL), ketamine (50 mg/mL), and xylazine (50 mg/mL) was injected behind the right ear. After a waiting period of 5–10 min, the animal was brought into an anesthesia preparation room and preoxygenated (10 L/min) for at least 5 min via a tight-fitting facemask with 100% oxygen at a flow rate of 10 L/min. Isoflurane (1%–5%, vaporized in 100% O₂) was initiated if needed to maintain sedation. A venous catheter was inserted into an auricular vein and the piglet was intubated. The piglet was then transported to the operating room, where anesthesia was maintained on isoflurane (1%–5% vaporized in 100% O₂). A constant body temperature between 38.3°C and 39.5°C was obtained using T/Pump water-circulated heat pads (Stryker, Minnesota, United States) and a bair hugger TM (3 M, Minnesota, United States). Intravenous balanced polyionic fluids (2.2 mL/kg/hr BW) were administered during anesthesia. Intraoperative pain medication consisted of 0.02 mg/kg BW buprenorphine. Locoregional anesthesia was provided using a mix of bupivacaine (1.25–2.5 mg/kg BW), epinephrine (0.005 mg/kg BW), and lidocaine (up to 4 mg/kg BW) in a subcutaneous line block at the intended surgery site.

Following injection, general anesthesia was maintained for an additional 0.5 h to allow the perfusate to diffuse through the RWM prior to humane euthanasia with pentobarbital sodium (100 mg/kg BW intravenous [i.v.] or direct cardiac injection in anesthetized animals) and sample collection.

The stimulation and recording were controlled by an Intelligent Hearing Systems Duet with Smart EP (Miami, FL). Sound delivery was through an Etymotic speaker (ER-3b) connected by a sound tube to an in-ear insert. Bone conduction was through the B-81 Bone Vibrator. Needle electrodes were placed on the mastoid bone ventral to the stimulated ear (inverting) the vertex (non-inverting) and the forehead (ground). For bone conduction, the contralateral ear was masked with 80 dB nHl broad-band noise. Stimuli were primarily clicks although tone pips were also tested in some cases. Clicks were a 100 µs monophasic pulse and tones were 2 ms long with 1s rise/fall times. All stimuli were presented in alternating condensation and rarefaction phases, with up to 1,000 repetitions to each phase (fewer repetitions were used if the response was large). Levels for the transducers were calibrated by the manufacturers and are expressed in dB nHl relative to human hearing thresholds. The gain was 100,000x and recording epochs were 1,024 points collected in 25 ms.

The dexamethasone sodium phosphate (DSP) J64083.06 was purchased from ThermoFisher and dexamethasone (Dex) D4902 was purchased from Sigma. The Dex was dissolved in ethanol for a total of 10 mg/mL master batch and then diluted in sterilized deionized (DI) water 1:100 for a final concentration of 100 μg/mL. DSP solution was made at 4 mg/mL in sterilized DI water and filtered with a 40 µm syringe filter.

Perilymph was collected from the RWM immediately postmortem. Piglets were positioned in right lateral recumbency to facilitate collection from the left, operated side. The incision edges were retracted and bone rongeurs were used to gently disrupt the mastoid process of the temporal bone and expose the round window. The connection of the inner ear to the brain was kept intact to prevent leakage of cerebrospinal fluid. Blood or tissue fluid was cleared from the round window with gauze. A 5 μL microhematocrit capillary tube (Fisherbrand™, Norwood, MA) with a distal end core diameter ranging from 0.11 to 0.12 cm was introduced into the RWM. Perilymph filled the tube by capillary action with a time interval of approximately 60 s. This process was repeated until a total of five microhematocrit tubes were collected for a total of ∼20 µL (∼4 µL in each vial was collected) and stored on dry ice. The piglet was repositioned in the left lateral recumbency to access the right, unoperated, control side. A 5 cm retroauricular skin incision was made centered at the level of the lateral canthus of the right eye. The incision was extended through the fibromuscular layers to access the mastoid process of the temporal bone. Perilymph collection was repeated as described previously. The perilymph samples were sent on dry ice in the hematocrit tubes (frozen) overnight to an outside vendor (Carbon Dynamic Institute) for mass spectrometry analysis. The samples from five tubes were pooled together for analysis after Dex injections. After DSP injections, for sequential sampling the first sample was not pooled with the other four samples, and the data is represented separately.

The BoneClear procedure (Wang et al., 2019) was tailored for the porcine cochlea due to the size and intricacy of porcine tissues as previously described (Moatti et al., 2020; Li et al., 2021; Moatti et al., 2023b). The custom-built light-sheet microscopy was used for 3D imaging.

As other groups have noted, there are anatomical differences between porcine and human external ear (Yildiz et al., 2022). The ear canal in pigs is tilted and narrow, which makes it almost impossible to access the TM directly down the external canal. Thus, a transcanal surgical approach was adopted and described to access the TM. The detailed surgical approach is demonstrated in Figure 1. Figure 1A shows the schematic of the porcine external, middle, and inner ear. With the transcancal approach and the removal of part of the external canal, the TM becomes accessible for the injection of the therapeutics into the middle ear cavity. The therapeutics can then penetrate the inner ear via the RWM and the oval window (OW). Supplementary Figure S1 shows the 3D scanned skull of a domestic pig depicting the inner ear in relation to the external ear canal detailing TM access. A step-by-step guide is provided below and in Supplementary Movies S1–S3:

1. The piglets were positioned in the right lateral recumbency, and a small surgical towel was used to tilt the head 45° sternal. All surgeries were performed on the left ear.

We used ABR to evaluate the hearing of the animal and to make sure the surgical approach did not affect the animal’s ability to hear. Immediately after surgery, due to the temporary damage introduced to the TM and middle ear, an air-conduction ABR cannot be performed. Thus, a bone-conduction ABR was used as our outcome measure to assess damage to the inner ear postoperatively.

First, we established the ABR measurements in the healthy pigs and compared air- and bone-conduction ABR. In Figure 2A, we show comparisons between air- and bone-conduction responses (blue and orange, respectively). The figure shows three different cases and in the top panel the three main peaks typically seen in the porcine ABR are shown (Hansen et al., 1992; Guo et al., 2010; Anderson et al., 2019). These three peaks are labeled as waves I, III/IV, and V with reference to human ABR waveforms. An example of a bone-conduction threshold series is shown in Figure 2B. The peaks show the expected decrease in latency with threshold increases (dotted line for wave III/IV). We found the air- and bone-conduction click-evoked ABR thresholds to be similar with mean values of 37 ± 4 dB (n = 3) and 38 ± 6.6 dB (n = 6) nHl, respectively. After surgery, there was no difference in the bone conduction thresholds (39 ± 10.0 dB nHL, n = 4). We observed click-evoked ABR bair-conducted latencies at 80 dB nHl to be slightly shorter than bone-conducted latencies. The latencies of wave I were 1.7 ± 0.5 (n = 8) compared to 1.9 ± 0.5 ms (n = 8), respectively, and latencies of wave III/IV were 3.3 ± 0.4 compared to 3.7 ± 0.4 ms, and latencies of wave V were 4.4 ± 0.4 compared to 4.8 ± 0.4 ms.

In Figure 2C, overlapped tracings are shown for clicks and tone stimuli with bone-conduction ABR before and after intratympanic injections. After surgery, the wave morphology and threshold to clicks were similar in both cases, and responses were present across the frequencies tested.

Next, we evaluated the efficiency of our surgical approach with the delivery of the corticosteroids, dexamethasone (Dex) and dexamethasone sodium phosphate (DSP). The drugs were delivered in separate experiments to the middle ear through an intratympanic injection, i.e., either Dex or DSP was injected. The drug then passed from the middle ear through the RWM to the base of the inner ear. Although Dex passes through the RWM readily, the low solubility of Dex in solution is a significant issue, as its maximum solubility is low (0.1 mg/mL). To overcome the Dex solubility issue, DSP is often used; however, DSP has a lower permeability through the RWM. It is important to note that intratympanic injection of DSP is a common clinical procedure for the treatment of idiopathic sudden hearing loss (Chandrasekhar et al., 2019). In DSP the phosphate group is covalently bound to Dex, but through hydrolyses, DSP becomes enzymatically dephosphorylated to Dex (Derendorf and Hochhaus, 1995; Samtani et al., 2004; Samtani and Jusko, 2005). DSP is inactive until the phosphate group is cleaved by phosphatases. Therefore, when injecting DSP, both concentrations of Dex and DSP in the perilymph should be characterized.

In non-survival procedures, piglets were euthanized 30 minutes after the injection, and perilymph (roughly 20 µL) was sampled from the RWM using microcapillary tubes (5 µL), as shown in Figure 3A (see also methods section). For clarity, a schematic of the perilymph collection from the RWM is also provided in Figure 3A. The concentrations of Dex and DSP (Dex injected at 100 μg/mL and DSP injected at 4 mg/mL) in the perilymph were measured using mass spectrometry. Figures 3B, C show the drug concentrations in the perilymph following treatment with Dex and DSP, respectively.

Figure 3B shows that Dex was present in the inner ear after intratympanic injection, but not in the perilymph of the contralateral ears (p-value: 0.0175, n = 5). Note, that in survival surgeries at one- and two-weeks post-surgery, no Dex was detected in the perilymph, which indicated a complete removal of Dex (n = 2). This was expected as Dex is cleared rapidly from the middle and inner ear of small rodents. The data presented in Figure 3B was obtained from two different surgery dates (IT-1 and IT-2). We found the mean ± SD for two surgeries performed at the same date to be 0.07 ± 0.04 (n = 2) while the mean for the other three surgeries in another date was 0.58 ± 0.05 μg/mL (n = 3).

After solely DSP injection, both DSP and dephosphorylated DSP (as Dex) content in the perilymph were measured. There was significantly more Dex content than DSP in the inner ear (p-value: 0.0256). To capture the gradient differences between the base and apex of the cochlea, perilymph from RWM was collected in two consecutive orders (sequential sampling from the base) that are shown by two separate data points per piglet in the graph. Most probably due to the sampling from the base, there was no difference in concentration of the drug after DSP injection between the 1st (the first 5 µL sample of perilymph from the base of the cochlea) and 2nd (a 15 µL sample) collection using an unpaired t-test (mean ± SD DSP content: 1.43 ± 1.10 versus 0.99 ± 1.04 μL/mL and mean ± SD Dex content: 4.47 ± 5.76 versus 5.64 ± 6.61 μL/mL, respectively).

FluidSim software is used to simulate the diffusion of drugs within the perilymph of multiple animal models and various delivery methods including systemic and intratympanic application (Salt, 2002; Plontke et al., 2007; Lehner et al., 2023). FluidSim estimates the middle ear kinetics, round and oval window drug entry, drug elimination to the vasculature, and more. To incorporate the porcine model into FluidSim, cross-sectional data of various structures in the inner ear was required. The cross-sectional data needed to be taken along the longitudinal diffusion axis of the drug. To obtain this data, an intact porcine inner ear was tissue-cleared and 3D imaged using a custom-built light-sheet microscope as previously described (Moatti et al., 2023b).

IMARIS was used to segment in 3D different organs in the inner ear as shown in Figure 4A. From the 3D segmented structures, the cross-section data was derived using IMARIS and MATLAB (see Methods section) and incorporated into FluidSim code. Figure 4B shows a 2D representation of the scala tympani across species, the x-axis represents the length of the organ, and the y-axis represents the equivalent tube diameter, where the diameter (d) is calculated as d = 2 C r o s s − S e c t i o n   A r e a π . The comparison of the other segmented volumes between pigs, humans, and other species is shown in Figure 4C. This data confirms the similarity of porcine cochlear volumes to humans. The vestibular volume; however, is much smaller in pigs than in humans.

We used the segmentation data and simulated the diffusion of Dex and DSP in the porcine inner ear. The RWM permeability measurements were obtained using an ex vivo porcine model (123 × 10⁻⁹ m/s for Dex and 12 × 10⁻⁹ m/s for DSP) and other parameters were set to the default FluidSim parameters (Moatti et al., 2023a). The Dex and DSP concentration as a function of time from injection and location along the scala tympani (ST) is shown in Figure 5A for a total of 300 min after intratympanic injection.

FluidSim predicted 30 min after intratympanic injection of Dex, the total amount of Dex in the whole scala tympani to be 0.39 μg/mL. At 1 mm away from the base, 30 min after delivery, FluidSim predicted that the Dex concentration would be 1.72 μg/mL.

FluidSim also predicted 30 min after intratympanic injection of DSP, the total amount of dephosphorylated DSP (if all hydrolyses) and DSP (if none hydrolyses) in the whole ST to be 1.65 and 1.27 μg/mL, respectively. At 1 mm away from the base, 30 min after delivery, FluidSim predicted that the Dex and DSP concentration would be 7.32 and 5.84 μg/mL, respectively.

Figure 5C shows the total diffusion time for Dex to reach the apex from the base of the cochlea in different animal models. The simulation of diffusion of Dex from the base to the apex of the cochlea shows a total diffusion time of ∼8 days and ∼6 days in pigs and humans, respectively.

Additionally, we measured the angle of the basal turn of the cochlea against the horizontal plane in a porcine fetal head. The fetal head was tissue cleared and imaged in 3D using a light sheet microscope (Supplementary Figure S2) we found the angle to be ∼34°. This angle in humans is measured at about 34° ± 3.6° as well (Tang et al., 2018). Given the similarity in the angle between humans and pigs, if there is an influence on fluid diffusion, the effect should be similar.

The cochlear aqueduct, situated within the petrous temporal bone’s bony labyrinth, serves as a conduit for the perilymphatic duct. This duct facilitates the drainage of perilymph into the subarachnoid space of the posterior cranial fossa, merging with the cerebrospinal fluid (CSF).

We have conservatively measured the aqueduct dimensions from our 3D images. Tracing the aqueduct bone, i.e., without any soft tissue, yielded a diameter at the narrowest part of the aqueduct to be about 200–400 µm. We observed that the soft tissue occupies space within the aqueduct, as shown in Supplementary Figure S3, and therefore effectively decreasing the flow rate. As such 200–400 µm diameter is a conservative and overestimated measurement. The extent to which soft tissue is affected by the tissue-clearing process is unknown. This value of the bone diameter without consideration of the soft tissue is in line with the results reported in the literature (Yildiz et al., 2022). These values might not be substantially wider than those reported for humans 138 (±58) μm (Gopen et al., 1997).

The development of a translatable large animal model for assessing intratympanic delivery could present an opportunity to improve the drug delivery of new and current therapeutics. The dimensions of the middle and inner ears, along with their associated barriers, significantly impact intratympanic delivery. While non-human primates have been used as the large animal model, from the ethical and financial perspective, their use in preclinical studies is limited and even banned in some places. The lower number of animals tested also results in lower statistical validity. On the other hand, the resemblance between the human and porcine middle and inner ears renders the pig an optimal large animal model for investigating intratympanic delivery. Nevertheless, the angled, narrow external auditory canal in pigs poses a challenge by obstructing direct access to the TM, thereby hindering the adaptation of simple intratympanic delivery in porcine models. To overcome this hurdle, we have developed and experimentally validated a surgical procedure for the safe intratympanic delivery of substances to the porcine inner ear (Figure 1).

This innovative approach offers the opportunity to evaluate a diverse range of promising therapeutics in a preclinical setting. For instance, in cancer patients, cisplatin therapy often results in hearing loss. To mitigate this debilitating side effect, systemic delivery of sodium thiosulphate was proposed. However, clinicians have commented on the potential for sodium thiosulphate to affect the survival rate when delivered systemically among patients with disseminated disease (Freyer et al., 2018; Viglietta et al., 2020; Witty and Cox, 2022). Local delivery of sodium thiosulphate could confine the drug predominantly to the inner ear, minimize any attenuation of cisplatin’s effects as needed for clinical applications in patients, and reduce the cost of treatment. The optimization of drug dosage, administration timing, and formulation could be facilitated in a predictable porcine model to reduce the cost of clinical trials. From an ethical standpoint, the potential expansion of data gathered from preclinical studies in pigs can enhance the robustness of results, aiding in the assessment of the feasibility and value of initiating human clinical trials.

The notion that local delivery achieves higher drug concentration in the perilymph and lower concentration in the plasma is supported by a 2007 study in humans (Bird et al., 2007). Not only did intratympanic delivery of Dex in humans produce an 88× higher concentration in the inner ear than systemic delivery, but also decreased the Dex concentration in plasma by 40×.

We’ve created and experimentally confirmed a surgical method for safely delivering therapeutics to the inner ear of pigs through intratympanic administration. This addresses the challenge presented by their angled and narrow external auditory canal.

ABR measurements (Figure 2) were used as a tool to identify the safety of our procedure. Pure tone ABR of domestic pigs showed similar thresholds of 38 and 37 dB for air- (n = 3) and bone-conduction (n = 6) ABR, respectively. The bone-conduction ABR has been previously performed on minipigs (tone-burst thresholds were reported) but there is no report in standard pigs to the best of our knowledge (Yildiz et al., 2022). The observed differences between the air- and bone-conduction ABR can be attributed to the inherent differences between these two techniques. The bone-conduction ABR has been reported to be longer than the air-conduction ABR in humans (Sohmer and Freeman, 2001). Overall, the waves evoked by bone-conduction ABR in standard pigs were clear, stable, reproducible, and well differentiated. In other studies, air-conduction ABR thresholds were between 40- and 50-dB sound pressure levels (SPL) for Rongchang pigs and unknown breed pigs, not reported in the paper (Hansen et al., 1992; Guo et al., 2010; Anderson et al., 2019). Here, we observed a slightly lower threshed of 38 dB SPL (n = 3). The observed differences could be due to breed differences between piglets. Also, when comparing our ABR results to those reported for humans, wave V of the ABR in young piglets is about 8 ms, compared to 8–9 ms in children and 5 ms in older adults (Sharma et al.,2016).

Interestingly, in three cases, after the surgery, no ABR shift or change in the ABR morphology was observed in piglets confirming their intact hearing. In only one case we did record a 20 dB shift in the ABR threshold of hearing. We can attribute this case either to our surgical procedure or a change in electrode position; nevertheless, it emphasizes the importance of performing ABR before and after surgical procedures. For the next experiments, we plan to keep the electrode positions intact before and after surgery to minimize that source of variability. If the ABR shift is observed consistently in patients where the drill is used to remove the bone, the use of rongeur is suggested instead.

The mass spectrometry data (Figure 3) from the collected perilymph confirms the successful delivery of the drugs (Dex and DSP) to the inner ear. The low concentration of the Dex in the inner ear after Dex injection (0.38 ± 0.28 μg/mL, mean ± SD) is limited by the removal processes in the inner and middle ear including through the aqueduct and vasculature. The initial Dex concentration of 100 μg/mL is also limited due to the low solubility of Dex, which prevents the injection of higher drug concentrations. Significantly different concentrations of Dex inside the inner ear were recorded during the two surgeries, n = 5 (Figure 3B, 0.07 ± 0.04 versus 0.58 ± 0.05 μg/mL, mean ± SD). The observed variation could be due to the natural variability in the animals, contamination of Dex from the middle ear during perilymph collection, or surgical procedure variation such as head position. Any delay in sample collection after perilymph removal could result in lower drug concentration. This variation could be also related to the presence of a false round window membrane which could hinder the substance passage and therefore affect the total concentration of the observed drug in the perilymph. There is a report on the presence of fibrous plug and false round window membrane of 11% and 21%, respectively, in human patients causing the same problem clinically (Juan and Linthicum, 2010). A large range of Dex concentrations (0.03–95.2 μg/mL) is also reported after intratympanic injection of DSP in humans (Bird et al., 2007).

Thirty minutes after the DSP injection, we observed higher Dex (dephosphorylated) concentration than DSP inside the inner ear. It was previously reported that after intratympanic DSP (0.4–1.8 mL of a 4 mg/mL solution, n = 13) injection in humans, the concentration of Dex in the perilymph was 1.4 μg/mL (range, 0.1–16.3) compared to 7.5 μg/mL for DSP (range, 0.03–95.2) (Bird et al., 2007). In porcine inner ear after intratympanic injection of DSP (1 mL of 4 mg/mL, n = 3), we found the Dex concentration in perilymph to be 5.06 μg/mL (range, 0.52–13.10) and the DSP concentration to be 1.21 μg/mL (range, 0.25–2.60). We noticed a higher ratio of Dex to DSP in domestic pigs than in humans, which could be due to the higher dephosphorylation rate in pigs, or lack of stabilization of DSP after perilymph collection in our technique (not stored in the presence of phosphatase enzyme inhibitor such as ethylenediaminetetraacetic acid--EDTA). To clarify this issue, in the future, we plan to study the DSP to Dex conversion rate in the perilymph collected from the pigs using HPLC or mass spectrometry measurements. Overall, the mean total concentration of the Dex + DSP in the perilymph is very similar in both pigs and humans (6.3 versus 8.9 μg/mL) after equivalent intratympanic injections of DSP.

The porcine volumetric data of different cochlear structures (Figure 4) shows the likeness to humans. Specifically, the similarity between the size of scala tympani in humans and pigs would seem to make the diffusion predictions from base to apex in pigs more relevant than other animal models, including non-human primates. The volumes of the porcine inner ear that we calculated using 3D imaging were very similar to the data collected from micro-CT scans of domestic pigs (Yildiz et al., 2022). The predicted time of diffusion from the cochlear base to the apex was calculated to be 6 days for humans and 8 days for pigs. The vestibular volumes, however, are smaller in pigs confirming similarity between pigs and humans.

Using the volumetric data and derived cross-sections, FluidSim was adapted to the porcine model (Figure 5). Our porcine volumetric data is available at FluidSim4. Please note that other than the cross sections of cochlear organs, many experimentally determined parameters, such as RMW permeability and rate of elimination from the perilymph, can be adjusted in FluidSim to positively impact the simulation results. Therefore, as more experimental data becomes available for pigs, the FluidSim model can be increasingly refined for greater fidelity. For example, here we used the original setup in the FluidSim and changed the permeability of the RWM which seems to have the highest impact on the drug content in the perilymph. However, for fine-tuning, other factors such as drug elimination from the middle ear, elimination from scalas, CSF entry flow rate, oscillation, blood elimination, and more can be adjusted as well. Given our previous ex vivo permeability measurements for Dex and DSP and using default elimination rates from the middle and inner ear, FluidSim predicted the total amount of Dex, dephosphorylated DSP (if all hydrolyses), and DSP (if none hydrolyses) in the whole scala tympani to be 0.39, 1.65, and 1.27 μg/mL, respectively. Additionally, Dex, dephosphorylated DSP (if all hydrolyses), and DSP (if none hydrolyses) concentrations at 1 mm from the base are 1.7, 7.32, and 5.84 μg/mL, respectively. The values we experimentally found in the 20 µL of perilymph for Dex and DSP (total of DSP and dephosphorylated DSP) were 0.38 ± 0.25 and 6.27 ± 5.62 μg/mL (mean ± SD), respectively. Overall, the simulated values are close to the experimental values, and future work will optimize FluidSim for specific drugs. In general, accurate simulation will extremely be helpful in studying inner ear pharmacokinetics.

In an attempt to capture the gradient differences between base and apex, we collected perilymph samples sequentially starting from the base of the RWM. Although we expected to see drug concentration differences between the samples, in all animals injected with DSP, the drug concentration in samples from the base and further distally in the cochlea were equivalent (mean ± SD DSP content: 1.43 ± 1.10 versus 0.99 ± 1.04 μL/mL; Dex content: 4.47 ± 5.76 versus 5.64 ± 6.61 μL/mL, respectively). This indicates the sequential sampling from the RWM is not appropriate for observing the gradient differences in the biodistribution and apical sampling may be necessary to study gradient differences (Salt et al.,2006).

Our experimental results for the concentration of drugs in the perilymph suggest that there is no significant exchange of fluid between the CSF and perilymph while collecting perilymph. This conclusion is supported by the fact that we did not observe a significant difference between the sequential perilymph sampling (Figure 3C, first sample—5 μL, second sample—15 µL). If CSF would enter and dilute the perilymph after the first measurement, we would expect to see a drop in the drug concentration in the second sample. Since we did not observe this difference, we inferred the rate of the flow of the CSF to the perilymph in the event of the RWM puncture is not substantial. On the other hand, when collecting the perilymph from pigs, 5 µL can be extracted quite rapidly, ∼1 min, which is different from non-human primates. This relatively rapid perilymph extraction can be indicative of a higher CSF flow to the perilymph through the aqueduct in pigs than non-human primates. This point can be further studied in the future.