Non-Contact Irreversible Electroporation in the Esophagus With a Wet Electrode Approach

Patients with esophageal cancer can experience local disease recurrence following initial treatment with surgery [1–3]. For residual or recurrent esophageal cancers refractory to other forms of therapy, additional surgical options are often limited [2]. Tumor infiltration into the esophageal wall and obstruction of the lumen causes dysphagia, pain, and substantial reduction in quality of life. Endoscopic ablation or esophageal stent placement can be performed for palliation, and clinical evidence suggests that ablative tumor debulking prior to stenting can further improve quality of life when compared to stenting alone [4]. However, the use of ablation in esophageal cancer patients has been limited to debulking of stent ingrowth, or for the treatment of superficial intraluminal disease [5] as penetrating probe/applicator based treatment with current devices is associated with risk of severe adverse effects such as esophageal perforation or stricture formation [6]. An ablation technique that can provide clinically beneficial tumor destruction with a more favorable safety profile would be valuable for management of postsurgical recurrence of esophageal cancer.

Irreversible electroporation (IRE) is an ablation technique where high-voltage, ultrashort electric pulses are applied using needle electrodes to kill cells by catastrophic membrane permeabilization [7,8]. The mechanism of cell death with IRE does not rely on sustained elevation of tissue temperature and hence has been termed a nonthermal ablation technique [9]. Therefore, IRE has gained clinical application as a safe option for the percutaneous ablation of tumors abutting heat sensitive structures such as the renal pelvis and bile ducts [10–14] and has been shown to spare the esophageal wall from serious injury during ablation of cardiac tissue [15]. Despite purported safety benefits relative to conventional percutaneously delivered thermal ablation techniques, clinical and preclinical studies [16–18] have demonstrated the potential for substantial tissue temperature increases in the vicinity of IRE electrodes used for pulse delivery. In proximity to the esophagus, these temperatures may induce sufficient damage to impact the integrity of the esophageal wall and lumen patency. Tissue temperature increase during IRE is associated with Joule heating from passage of electric currents through tissues [7]. The region immediately adjacent to the electrodes is the site of highest current density and thus the typical location of maximum heating, temperature increase and thermal damage [16].

Theoretically, avoiding direct tissue-electrode interfacing may enable tumor ablative effects while minimizing thermal effects. This approach is tenable due to the working principle of IRE that requires the induction of a strong electric field but not heat transport for tissue ablation. The use of electrically conductive gels between electrode and tissue is common during IRE of cutaneous tissue or subcutaneous tumors in rodents [19]. The presence of a conductive gel at the tissue-electrode interface produces favorable electrical properties that improves IRE efficacy, and may also reduce incidental arcing or tissue heating [20]. Recent experience in using IRE for cardiac ablation also suggests that direct tissue-electrode contact may not be essential to derive treatment effect [15]. In our study, we applied finite element modeling techniques to develop a “noncontact” wet-electrode IRE approach for esophageal tumors of clinically relevant sizes while avoiding thermal damage to the healthy esophageal wall. Further, we evaluate the in vivo feasibility and safety of noncontact, wet-electrode IRE in a swine model using a custom-designed catheter electrode. The ability to perform IRE using an endoscopic or image-guided approach without thermal damage to healthy tissues is desirable and would support translation of the technique to treat patients with esophageal cancer.

A catheter mounted electrode was designed to test the feasibility and safety of the wIRE in healthy swine esophagus (Fig. 1(a)). The custom-designed catheter device consisted of a medical grade stainless steel electrode wire (2.5 mm diameter, 10 mm length) that was coiled onto the shaft of a 5F, 16 × 40 mm inflatable balloon catheter (Atlas, Bard Peripheral Vascular, Tempe, AZ). The electrode was located approximately 3 cm from the proximal edge of the balloon. This device sizing allowed placement into the esophagus using the working channel of a flexible gastroscope or using a 0.035' guide wire under fluoroscopy guidance. Figure 1(b) shows a fluoroscopy projection of the device within the swine, where a balloon is introduced to mechanically block the gastro-esophageal junction, allowing creation of a column of diluted saline above that fully covered an electrode mounted on the catheter shaft distal to the location of the balloon, mirroring the wIRE-Swine simulation model (Fig. 2(c)).

A geometric model of the human esophagus (45 mm diameter, 7 mm thickness) with a tumor (13 mm wide, 25 mm long, Fig. 2(d)) obstructing the lumen, and the peri-esophageal visceral tissue (200 mm diameter) was constructed using a computer aided modeling software (Inventor, Autodesk Inc., San Rafael, California) with dimensions derived from published data [21]. The peri-esophageal visceral tissue was assumed to be comprised of connective tissue. The geometry was imported into Comsol Multiphysics (Comsol Inc., Burlington, MA) for mesh generation and finite element analysis (FEA). Two configurations were tested; conventional IRE (conIRE) where the treatment electrode (20 mm length, 1 mm diameter) was placed directly into the tumor (Fig. 2(a)) and wet-electrode IRE (wIRE) (Fig. 2(b)) where the electrode was placed adjacent to the tumor in the esophageal lumen that was filled with chilled diluted saline (infused at 4 degrees Centigrade) to provide electrical conduction. For the conventional treatment, simulations were run with air or chilled diluted saline (conIRE-Sal) in the esophageal lumen. For the wet electrode approach, an additional simulation was performed to estimate treatment effect in a healthy swine esophagus without tumor, with the electrode centered within the lumen (wIRE-Swine) (Fig. 2(c)). A diagram describing the configurations with their associated simulations is shown in Supplemental Table 1 available in the Supplemental Materials on the ASME Digital Collection.

The finite element mesh for wIRE consisted of 25,787 elements, 24,344 elements for con-IRE, and 27,049 elements for wIRE-Swine, respectively. The mesh generation was refined via a physics-controlled algorithm. The simulations were run for 120 s, a time period that included the duration of pulse delivery (90 s) and a short post-IRE period (30 s) to observe heat dissipation. Simulations were started 0.5 s prior to the delivery of the first square wave pulse. The pulse parameters used for all the simulations were an applied voltage of 3000 V for 90 pulses with a pulse width of 100 μs and an interpulse delay of 1 s. The pulses were assumed to be delivered between the active electrode placed within the tumor or the esophageal lumen and a distal ground electrode, represented by grounding the outer boundary of the visceral tissue surrounding the esophagus. All equations, variables, and constants used to solve computational simulations are depicted in Tables 1 and 2. The three-dimensional electric current module in Comsol Multiphysics was used to solve the Laplace Equation to determine the electric field distribution in the mesh during pulse application. Initial simulations were calculated using static time conditions with DC current for assessment of IRE extent at various voltages, then simulations were solved using time dependent methods with application of a pulse train to observe thermal damage. The associated temperature changes were accounted for by coupling Joule heating from passage of electric currents, heat transport, and metabolic heat generation. The Pennes Bioheat equation was used to represent metabolic heat flux, arterial blood temperature, specific heat of blood, blood perfusion rate, and blood density. An initial temperature of 37 °C was used for all biological domains while a temperature of 4 °C was used for the chilled saline domain. Thermal injury to tissue was estimated using an implementation of the Arrhenius Equation and associated thermal damage equation. The implementation of our simulation models closely follows work reported by Edd and Davalos [20] and Corovic et al. [22]. The electrical, thermal, biological, and other relevant material properties using in the finite element models are defined in Tables 3 and 4. Additional simulations were conducted where the tumor conductivity was modeled as a function of electric field strength using the symmetric sigmoid function model used in Sel et al. [23] for conIRE, conIRE-Sal, and wIRE conditions. These results are shown in Supplemental Figure 1 available in the Supplemental Materials on the ASME Digital Collection and Table 5. Due to the lack of published experimental data specifically detailing the esophageal tumor gradient as a result of EP, conductivities and sigmoid constants were assumed to be comparable to pancreas tissue found [24]. There was not a significant different in IRE extent when comparing the spatially dynamic conductivity versus a constant conductivity; therefore, the tumor conductivities for time dependent simulations were assumed to be constant.

Following a protocol approved by the Institutional Animal Care and Use Committee, a single fluoroscopy guided endoluminal wIRE ablation was performed in the esophagus of seven Yorkshire pigs (weight range 35–45 kg, male) acquired from one supplier (Archer Farms, Darlington, MD). The wIRE device was inserted into the esophagus over a 0.035' guide wire under fluoroscopy guidance (Innova, GE, Milwaukee, WI). The device balloon was positioned within the distal esophagus/gastro-esophageal junction and inflated to anchor the device and mechanically occlude the lumen, while also preventing dislocation during treatment (Fig. 1(b)). The esophagus proximal to the balloon was filled (∼50 cc) with chilled diluted saline (approximately 1:3 with DI water) to provide an electrically conductive pathway from the electrode to the mucosa while also reducing the thermal damage to the healthy esophageal wall. The head of the animal was elevated to prevent aspiration of the diluted saline into the airways, and the animal was restrained to ensure maintenance of the saline column within the esophagus. Imaging was utilized to ensure that the electrode remained centered in the lumen during pulse delivery. IRE was performed by delivering ninety 100 microsecond pulses at 3000 V between the electrode in the esophagus and a grounding pad (4 × 8 in., ESRE grounding pad, Bovie Medical) placed on the flank of the animal using the Nanoknife generator (Angiodynamics, Latham, NY). The grounding pad was approximately at 2-3 feet distance from the estimated location of the electrode within the esophagus. Pulse delivery was performed using ECG gating, after administration of a neuromuscular paralytic (Rocuronium). Following IRE, fluoroscopy and flat panel cone beam computed tomography using intraluminally injected contrast was performed to examine the immediate post-treatment patency of the esophagus and to exclude perforation. Animals were euthanized with barbiturate overdose approximately 4 h post-IRE. A simplified CAD diagram of the device during pulse application can be seen in Fig. 3.

The esophagus of each animal was resected enbloc from the larynx to the stomach and sectioned longitudinally to allow identification of the ablation zone based on superficially visible discoloration of the mucosa or change in shape or texture. Photographs were taken for measuring the length of the lesion along the esophageal wall, which was then sectioned transversally across the ablation zone for fixation in 10% neutral buffered formalin. Surrounding adventitial connective tissue and adjacent nerves were included in the sections whenever possible. These samples were then processed for histology by paraffin embedding, sectioned at 4 μm thickness, and stained with hematoxylin and eosin (H&E). All specimens were assessed for histopathologic findings of nonthermal or thermal cell injury, cell degeneration and necrosis, inflammatory infiltrate, and fibrotic changes by a board-certified veterinary pathologist (SM).

All results were summarized by mean and standard deviation.

Placement of the electrode into the tumor during conIRE induced an electric field gradient that increased in intensity proportionally to the applied voltage. When considering a critical electric field threshold, there is a range of published values at which esophageal tissue is expected to undergo IRE [25–27]. In these simulations, a value of 900 V/cm was the assumed to demarcate the IRE threshold in both tumor and healthy esophagus as it was the median value from published datasets. A summary of the results for all simulations is shown in Table 6. A maximum of 70% of the tumor was predicted to undergo IRE during treatment at 3000 V (Figs. 4(a) and 4(b)). At this setting, 11% of adjacent healthy esophageal wall was also predicted to undergo IRE. While reducing the treatment voltage reduced the potential for damage to the healthy esophagus, there was concomitant, drastic reduction in the tumor volume expected to undergo IRE (reducing from 70% at 3000 V to 17% at 1000 V). These results are shown in Table 7. The estimated percentage of thermal damage in the healthy adjacent esophagus was 65% (Figs. 1(a), 1(b), and 5).

In the wIRE condition, electric field penetration into the tumor reduced with increasing distance between the electrode and the tumor surface, while simultaneously decreasing the area of healthy esophagus expected to undergo IRE (Fig. 6). When applying 3000 V at the electrode, 28% of the tumor volume was predicted to undergo IRE while damage to healthy esophagus was limited to 0% if the electrode was within 1 mm of the tumor surface (Figs. 4(g)–4(h)). This volume of tumor ablation during wIRE was clinically relevant to what could be achieved with conIRE using similar pulse parameters, with substantially lower risk of injury to healthy esophagus.

When using electric pulse parameters for maximum tumor coverage, wIRE induced a peak temperature of 102.7 °C and 37.2 °C in the tumor and healthy esophagus respectively (Figs. 5(a)–5(d), 7, and 8(d)). Temperatures rapidly returned to physiologic levels immediately after cessation of pulse application, and the estimated thermal damage in the adjacent healthy esophagus less than 1% (Figs. 4(a), 4(b), and 5). This was substantially lower when compared to the conIRE condition where temperatures reached 128.4 °C within the tumor and 75.1 °C in the healthy adjacent esophagus (Figs. 1(a)–1(d), 7, and 8(a)), leading to greater thermal damage within both tissue types 65% of tissue within the adjacent esophagus).

From conIRE-Sal simulations, it was noted that the addition of chilled diluted saline to the lumen somewhat reduced electric pulse induced temperature changes with mitigation of thermal damage to the adjacent esophageal lining (136.9 °C, in the tumor and 69 °C with 17% thermal damage in the adjacent esophagus) (Figs. 1(a), 1(b), 3(a)–3(d), 5, 7, and 8(b)) while maintaining a comparable IRE extent (Figs. 4(e) and 4(f)), but levels remained considerably greater when compared to wIRE simulations. Reducing the applied voltage during conIRE reduced temperature increases in both the tumor and healthy esophagus but with corresponding reduction in overall ablation volumes as well.

wIRE-Swine simulations were used to plan the in vivo study, and determine whether nonthermal, transmural, circumferential ablations in the esophageal wall can be achieved using this approach. For the selected pulse parameters, simulation results indicated that sufficient electric field strength can be concentrated in the esophageal wall to induce transmural ablation without substantial temperature increase and thermal damage to the tissue. Simulations indicated that IRE could penetrate up to 29% of the esophageal lumen with limited temperature elevation in the esophageal wall, and circumferential ablations can be achieved by a single electrode centered within the lumen while surrounded by chilled diluted saline (Figs. 2(a)–2(d), 4(c), 4(d), 6, and 8(c)). These simulation findings were validated through our in vivo swine model studies.

A figure symbol legend for Figs. 4–7 is detailed in Table 8.

Seven endoluminal esophagus ablation using the wIRE-Swine approach were successfully completed, one in each animal that underwent the procedure. In three animals, transient ventricular tachycardia (VT) occurred during electric pulse delivery that could be immediately controlled by an intravenous injection of 1 mg/kg lidocaine. In the remaining animals lidocaine was administered prior to delivery of the electric field, successfully preventing the occurrence of VT. No other clinical or procedure related complications were observed in the animals prior to euthanasia at the designated timepoint.

Following IRE ablations, fluoroscopy and computed tomography imaging was used to confirm that the electrode-bearing catheter had not been dislodged during treatment. There was no evidence that the patency of the esophagus was compromised in any of the animals. Immediate organ perforation was ruled out by the absence of extravasation of intraluminal contrast into the extra-esophageal space (Fig. 1(c)). The ablation zone was not directly visualized and could not be distinguished from surrounding tissue on CT imaging performed immediately after the procedure due to the technique's spatial resolution, the low soft-tissue contrast and beam hardening artifacts from the presence of the metallic electrode coil.

Post-treatment gross visual examination of the electrode did not reveal residual tissue or debris indicative of thermal damage or mechanical injury. All seven planned ablations were visually distinguishable and the ablated areas harvested from the animals. Gross pathology examination of esophagus specimens demonstrated macroscopically visible circumferential lesions characterized by transmural discoloration and edematous thickening of the submucosa at the site of ablation, consistent with appearance of luminal organs following IRE (Figs. 9(a) and 9(b)) [28]. The boundary between the ablation zone and untreated areas appeared sharply demarcated. The depth of penetration of the ablation zone was consistently found to comprise the full thickness of the esophagus wall on cross sections. The length of the lesion was found to be 18.66 ± 7.53 mm long as measured using photographs taken during gross analysis. In comparison our simulations predicted an ablation length of 22.67 mm and a penetration of 0.7 mm into the esophagus.

Histopathology sections stained with H&E exhibited full thickness coagulative necrosis of the esophagus wall from the mucosa to the adventitia (Fig. 10(a)). The depth of full thickness ablation was measured to be 3.52 ± 0.89 mm. The area of necrosis was characterized by nuclear pyknosis and cytoplasmic hypereosinophilia with retention of architecture, and separation of the mucosal epithelium from the lamina propria (Figs. 10(b)–10(d)). The area of necrosis was accompanied by full thickness hyperemia and mild neutrophilic inflammation, and submucosal edema. In four cases where nerves attached to the esophagus were included into the histologic sections no histologic damage was evident within the nerves (Fig. 10(e)). Complete necrosis of submucosal glands were observed in all sections examined.

Our simulation models suggested the feasibility of using a wet electrode approach for noncontact tumor ablation in the esophagus with IRE. Simulations also indicated that such an approach could minimize off-target heating or damage to the healthy esophageal wall. The simulation findings were validated in a swine model experiment where we demonstrated predictable ablation of the esophageal wall using wIRE. Our animal model experiment demonstrated preservation of lumen patency even after performing transmural circumferential ablation of the esophagus. We anticipate that similar noncontact IRE should be feasible in other luminal organs such as the bronchus, bile duct or urinary tract with the use of appropriately designed devices for the wet electrode technique.

While most previously studied ablation techniques such as electrocautery, argon plasma coagulation, radio frequency ablation, and cryotherapy are based on a thermal mechanism [28], irreversible electroporation (IRE) is a predominantly nonthermal technique that is capable of inducing cell-death while conserving the extracellular scaffold of the organ [8,29]. Previous work has shown that structures and organs with a lumen such as the small bowel [30], the rectum [31], the bile ducts [32], and the urinary collecting system [33] remained intact when IRE was applied close to these structures. More recently, IRE pulse parameters used for cardiac ablation have been tested for safety from view of injury to the esophagus. Our experiments adds to this body of literature documenting safety of IRE in the esophagus as our electric field strength and energy delivered far exceeds what has been tested in prior studies [34,35]. Our experimental findings are consistent with recent reports in literature where IRE ablation in the esophagus has been tested for cardiac [15,36] and oncologic applications [37], with a promising long-term safety profile. In all these studies, it has been reported that the acute patency of the lumen was maintained and in the follow up period, intact extra cellular matrix was seen to promote the recovery of mucosal layers without evidence of perforation. Taking these results in context, it is possible that esophagus may react in a similar fashion following wIRE ablation, but this remains to be proven conclusively.

The mechanism of ablation in IRE is considered nonthermal, however, temperature increase can usually be observed in close proximity to the electrodes due to high current densities. For efficient thermal ablation or coagulative tissue damage, temperatures higher than 50 °C must be achieved and maintained for prolonged periods of time (>5 min) to cause heat induced cell death by permanent damage or denaturation of proteins, and membrane fusion [38]. Our simulations suggest that significant thermal injury cannot be expected within the esophagus wall when using wIRE. However, when the electrode is placed directly inside the tumor during conIRE, temperatures exceeding 70 °C may be reached in the esophageal wall with the potential for damage. In our simulations, we performed 1:1 comparison of ablation with a single electrode/treatment with conIRE and wIRE. While individual wIRE ablation volumes were smaller when compared to matched conIRE condition, repeating the treatment with wIRE can readily allow complete tumor coverage while maintaining nonthermal benefit (Supplemental Figure 2 available in the Supplemental Materials on the ASME Digital Collection.). IRE is well established to evoke a robust immune response where we anticipate that even subtotal nonthermal IRE of the tumor can eventually lead to complete clearance from this mechanism. Finally, we used a single value of 900 V/cm as the threshold for IRE in both healthy esophagus and tumors. The exact IRE threshold for these tissue types is not available in literature and incorporating this into our simulations would increase the accuracy of our findings. Validating our findings, however, is impeded by the lack of tumor-bearing patient-sized animal models.

The lumen of the esophagus varies in diameter and shape, where achieving predictable, circumferential ablations in the lumen can be challenging. Currently, balloon mounted electrodes are used in applications such as eradication of Barrett's Esophagus using Radio frequency Ablation and in preliminary studies with IRE [26]. wIRE provides a simplified approach to electrode construction, allowing uniform electrical contact, reducing temperature increase in the tissue, and at the same time allows for rapid repositioning for achieving larger treatment volumes. The actual length of ablation could be potentially varied by altering the length of the electrode within the esophagus, and the penetration into the esophageal wall could be titrated by adjusting the electric pulse parameters (voltage and number of pulses), but these were not tested in our preliminary experiments. We expect that minor alterations to the electrode design would allow using wIRE the same to achieve focal (noncircumferential) ablation as well. We expect the ability to perform noncontact IRE will be relevant to the growing use of the technique for ablation in luminal organs such as the heart, where complex organ geometry and physiological conditions (e.g., beating heart) may preclude uniform or predictable contact with lumen surface.

Regarding the safety of the procedure, we did not observe acute perforation of the esophageal lumen as confirmed by immediate postinterventional imaging and histologic analysis. However, it must be noted that treatment response to IRE make take up to 24 h to completely manifest, and our subacute timepoint precluded confirmation of this outcome [39]. The preservation of the organ's integrity is crucial since the esophagus is located within the mediastinum and surrounded by vital structures such as blood vessels and nerves. Perforation of the esophagus could lead to potentially life threatening adverse events such as mediastinitis or fistula formation to surrounding structures. In contrast to results reported by Schoellnast et al., who has demonstrated effects of IRE on nerves with potential axonal recovery [40,41], our histology results did not indicate an immediate effect on adjacent nerves. This may be due to the fact that our specimens were harvested early after the procedure whereas such changes may manifest at later timepoints. Furthermore, in Schoellnast's study [40], the sciatic nerve was in the center of the ablation zone, whereas the nerves in our experiments were located at the periphery of the treatment zone and thus exposed to significantly lesser electrical field strengths. Moreover, recent studies with longer observation period looking at impact of IRE on nerves also indicate the lack of permanent damage. While the mechanisms are unclear, the cumulative results of our and prior studies are supportive of advancing the use of IRE in the esophagus.

There are several limitations to our study. Currently, no tumor model exists in swine, and hence ablations were performed on normal healthy tissue. wIRE is yet to be validated in a suitable tumor model for focal ablation while sparing the peripheral healthy esophagus. Further, the concept of wIRE is predicated on placing a single electrode into the tumor through the working channel of an endoscope. Placing multiple electrodes may pave the way for optimization of pulse parameters and sculpting the electric field with conIRE that may replicate some of the features of wIRE. Our simulation models did not fully incorporate electroporation and heating induced electrical property changes in tissue due to lack of prospective data collection during experiments and information in literature. We expect this may underestimate ablation volumes and also electric field distribution in healthy esophagus. Its impact on accuracy of our model findings is somewhat offset as the gap is present in all our test conditions. In some simulations a peak temperature of 100 °C or higher was encountered which would cause phase change in the saline. However, this phenomenon was not fully represented within our models. Our preliminary data were collected at an early postprocedural time point. This was necessary as this was a preliminary safety study and adverse events in the esophagus such as perforation could have severe impact on animal well-being. This may also have impacted ablation measurements in the lumen. Given our acceptable acute safety profile, we expect future studies to delineate the ablation size measurements more accurately. Adverse events such as perforation or strictures may evolve over time and a longer follow-up period will be necessary to exclude potential adverse events. Thermal effects were determined using a simulation model rather than in vivo measurements due to the technical difficulties of introducing a temperature probe without interfering with treatment delivery. While preliminary results from our swine study indicate that wIRE may constitute a promising novel technique for endoluminal catheter directed ablations in the esophagus and other luminal organs, further research will be necessary to establish the safety and efficacy of this method and to examine mid- and long term outcomes.

In conclusion, we have used simulation models to develop the novel concept of wIRE, where a wet electrode is used to perform ablation of tissue within a luminal organ in a noncontact fashion. Ablation with wIRE presents the benefit of largely mitigating the thermal effects of IRE, especially in conditions ablations are to be performed using a single electrode. A custom designed electrode catheter was developed to test and validate the feasibility and early of safety of wIRE in swine esophagus, with favorable results.

The authors report no relevant disclosures related to the work presented here. G.S. holds stock options in Aperture Medical. JCD holds equity in Adient Medical, Serpex Medical and Cordis. SBS is a consultant to BTG, Johnson & Johnson, XACT. SBS has funding support from GE Healthcare and Angiodynamics and holds stock in Aperture Medical.

• Congressionally Directed Medical Research Programs (Award Nos. CA170630 and CA190888; Funder ID: 10.13039/100000090).

• National Cancer Institute (Award No. R01CA236615, Funder ID: 10.13039/100000054).

• National Institute of Diabetes and Digestive and Kidney Diseases (Award No. R01DK129990, Funder ID: 10.13039/100000062).

• The Institute for Applied Life Sciences in the University of Massachusetts at Amherst (Funder ID: 10.13039/100008975).

Computer simulations: M.C.S, S.C., G.S. Animal experiments: T.W, N.B.G, S.B.S, J.C.D. Histology: S.M, G.S. Statistical analysis and results: M.C.S, G.S., and Manuscript preparation: M.C.S, S.C., T.W., S.B.S, J.C.D, and G.S.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.