A novel bilayered expanded polytetrafluoroethylene glaucoma implant creates a permeable thin capsule independent of aqueous humor exposure

Abstract The purpose of these studies was to evaluate clinical, functional, and histopathological features of glaucoma drainage implants (GDIs) fabricated from novel, custom‐tailored expanded polytetrafluoroethylene (ePTFE). Implants of matching footprints were fabricated from silicone (Control) and novel, bilayered ePTFE. ePTFE implants included: (a) one that inflated with aqueous humor (AH) (High), (b) one that inflated with a lower profile (Low), (c) an uninflated implant not connected to the anterior chamber (Flat), and (d) one filled with material that did not allow AH flow (Filled). All implants were placed in adult New Zealand White rabbits and followed over 1–3 months with clinical exams and intraocular pressure. The permeability of tissue capsules surrounding GDIs was assessed using constant‐flow perfusion with fluoresceinated saline at physiologic flow rates. After sacrifice, quantitative histopathological measures of capsule thickness were compared among devices, along with qualitative assessment of cellular infiltration and inflammation. Capsular thickness was significantly reduced in blebs over ePTFE (61.4 ± 53 μm) versus silicone implants (193.6 ± 53 μm, p = .0086). AH exposure did not significantly alter capsular thickness, as there was no significant difference between High and Filled (50.9 ± 29, p = .34) implants. Capsules around ePTFE implants demonstrated permeability with steady‐state pressure: flow relationships at physiologic flow rates and rapid pressure decay with flow cessation, while pressure in control blebs increased even at low flow rates and showed little decay. Perfused fluorescein dye appeared beyond the plate border only in ePTFE implants. ePTFE implants are associated with thinner, more permeable capsules compared to silicone implants simulating presently used devices.


| INTRODUCTION
There has been renewed interest in improving the success and complication rate of incisional glaucoma surgery. Commonly used approaches create an outlet for aqueous humor (AH) release from the anterior chamber to the subconjunctival space through transscleral techniques (trabeculectomy, deep sclerectomy) or by inserting an implant (glaucoma drainage implant [GDI], XEN Gel Stent [Allergan], PreserFlo Microshunt [Santen]) that facilitates AH release to a subconjunctival tissue pocket (a bleb). Failure rates due to inadequately controlled intraocular pressure (IOP) or need for reoperation have been reported as high as 46.9% over 5 years for trabeculectomy, 1 and the equivalent rates for Ahmed and Baerveldt GDIs are 32 and 18%, respectively. 2 Complications of incisional surgeries include excessive IOP reduction (hypotony), diplopia, bleb leaks, device extrusion, and infection. 3 The 5-year rate of erosion for GDIs has been reported between 1 and 5%. 3,4 Risk of wound leak following trabeculectomy is significant as well, reported as 6% over 5 years. 3 Inadequate IOP reduction following incisional glaucoma surgeries most commonly results from loss of bleb capsule permeability to AH flow (also referred to as capsular porosity or hydraulic conductivity) due to excessive fibrosis. 5,6 Bleb capsule thickness and permeability are potentially affected by several factors, including individual patient healing rate, the actions of cytokines and anti-proliferative factors present in AH, and characteristics of the implant itself. [7][8][9][10][11][12][13] Wilcox et al. showed that larger GDI plate radius led to thicker capsules. 14,15 They hypothesized that the greater height of a larger bleb increases capsule tension as modeled by Laplace's law, which leads to collagen deposition and capsular thickening. 14 The influences of bleb geometry and AH exposure were explored further by Coote and coworkers, who showed that outflow resistance around GDIs implanted in rabbit eyes increased with exposure to either AH 16 or balanced salt solution, suggesting that capsule tension alone, and not AH cytokines, is sufficient to reduce permeability. 17 Currently, glaucoma surgeons utilize a variety of techniques to prevent excessive fibrosis. The primary technique used in conjunction with trabeculectomy and XEN Gel Stent placement is antifibrotic application at the time of surgery. 18,19 Mitomycin C application has been associated with improved clinical outcomes, but comes with an increased risk of postoperative infection and hypotony. 20,21 The benefits of antifibrotic use with GDI placement have not been established. 22 Surgeons use several ad hoc modifications in order to reduce postoperative fibrosis following GDI placement, such as twostage implantation to delay AH flow, or placement of the GDI plate more posteriorly or above the Tenon's capsule. 23,24 Large-scale, prospective randomized trials have not been performed to establish the potential benefit of these modifications. A promising and incompletely explored approach to fibrosis control utilizes biomaterials that modulate healing. An ideal biomaterial for use in bleb-forming glaucoma surgery would (a) incite neither toxicity nor immune response (biocompatibility), (b) permit sufficient healing to minimize risk of exposure and infection, (c) provide optimal permeability of the capsule formed around the device, and (d) abrogate any pro-inflammatory effects of AH.
Expanded polytetrafluoroethylene (ePTFE) is a highly stable polymer of tetrafluoroethylene that was patented by Gore. 25 Due to its biocompatibility, biostability, and high compliance, ePTFE incorporates well into many tissues and is approved for use in numerous biomedical implants, including: vascular grafts, bypass grafts, hernia membranes, and sutures. 26

| Implant device materials and construction
Control glaucoma drainage devices were made of 40-durometer silicone sheeting of 0.25 mm nominal thickness with a smooth finish. A silicone tube with an outer diameter of 0.5 mm and inner diameter of 0.25 mm, chosen to facilitate surgical placement through a 25G needle tract, was attached using a silicone adhesive. Experimental implants devices were fashioned from novel, bilayered ePTFE membranes; two such membranes were continuously bonded to one another around the device outer perimeter to form a central reservoir chamber with a total nominal material thickness of 0.2 mm prior to inflation. ePTFE has a microporous structure comprised of dense "nodes" joined by thin "fibrils"; its pore size refers to the distance between nodes. 28 A unique aspect of the ePTFE used here was that the membranes were comprised of an exterior "open" layer with a larger pore size into which surrounding cells could integrate, and an inner "tight" layer with pores too small for cellular ingress into the reservoir, but through which AH could pass readily. A silicone tube identical to that used for control devices was sandwiched between the layers and attached in watertight fashion with a silicone adhesive. The preimplantation resistance to flow through experimental devices was negligible at physiologic flow rates. All devices were sterilized by ethylene oxide after manufacturing and packaged in sterilization pouches prior to implantation.

| Characterization of implants
Five implant designs were evaluated, all with identical twodimensional footprints prior to implantation. Implant dimensions were chosen to fit between the extraocular muscles of NZW rabbits and anterior to their posterior orbital venous sinuses. Schematics are shown in Figure 1, and a brief description of experimental protocols, including the number of each implant type placed and duration of follow-up, is found in Table  2. An ePTFE implant with an inflatable reservoir attached to a silicone tube. In the uninflated state, the footprint of the reservoir was the same as the silicone control; however, AH entry through the tube fully inflated the reservoir (High).
3. An ePTFE implant identical to (2), but with the tube occluded by a silicone plug and the reservoir filled with additional ePTFE material to mimic the shape produced by inflation, without AH exposure (Filled).
4. An ePTFE implant similar to (2), but with an inflatable reservoir whose height was limited by a central rivet (Low).
5. An ePTFE implant with a reservoir identical to (2)   Health) solution was applied to the eye at the conclusion of surgery, twice daily for 7 days and, once daily for 7 more days.

| Bleb grading
The anterior segment (AS) and zone overlying the GDI were inspected twice daily for the first seven postoperative days, then daily until the end of each study. Then, 14-and 28-days after surgery, IOP was mea-

| Bleb permeability measurement and fluorescein egress
Measurements were performed using constant-flow perfusion of the GDI tube by anterior chamber cannulation, measuring the pressure within the device system at various flow rates. This differed from previously published methods, 16,17 in which constant pressure

| Histopathology
Rabbits were humanely sacrificed using a lethal injection of sodium  Figure S1). Slides were evaluated by a masked, trained pathologist (M. T.) and the zone of fibrous tissue between the implant and normal-appearing conjunctiva was measured as the capsule. For capsular thickness measurements, at least 10 measurements were included per eye, representing a minimum of 5 different sections. In addition, the device area was qualitatively assessed for tissue inflammation adjacent to implants, cellular integration into device material, and undesirable folding of ePTFE devices, which are inherently more flexible than silicone.

| Statistical analysis
All results are displayed as mean ± SD. For the IOP, bleb height, vascularity, and bleb extent analyses, two-way analysis of variance (ANOVA) with Dunnett correction for multiple comparisons was used.
Nested t tests were used for comparisons of capsular thickness of control and High implants. Two-way ANOVA with Tukey correction for multiple comparisons was used for comparison of capsular thickness over High, Low, Filled, and Flat devices.

| Bleb morphology
Implant blebs, shown in Figure 2a,b, were graded for height, extent, and vascularity using the IBAGS tool for all devices at 2 and 4 weeks after placement, and at 6, 8, and 12 weeks for control and High devices. 29 Mean grades for bleb extent and vascularity did not differ significantly among devices, but there were significant differences in bleb height between control devices and some ePTFE devices (Figure 3b). At both 2 and 4 weeks after implantation, control device blebs (IBAGS score 1.50 ± 1.05 and 1.68 ± 0.77, respectively) were significantly higher than Low (0.56 ± 0.38, p = .05 at 2 weeks and 0.75 ± 0.29, p = .04 at 4 weeks) and Flat (0 ± 0 at 2 and 4 weeks, p = .0005) blebs. There was no significant difference between control and High device bleb heights at 2 and 4 weeks (p = .65 and p = .69, respectively) or between control and Filled device bleb heights at 2 and 4 weeks (p = .51 and p = .19, respectively). High blebs were significantly higher at 2 and 4 weeks than Low (p = .05 and p = .03, respectively) and Flat (p = .0005 and p < .0001, respectively). At 6 and 12 weeks, control and High devices were not significantly different in height (Figure 3b).

| Intraocular pressure
IOPs were measured for all implant models for 1 month after implantation and for 12 weeks for devices with tubes open to the anterior chamber ( Figure 3a). IOPs did not decrease significantly from baseline in any group and did not differ significantly among implant models.

| Tissue integration and capsular thickness
Histopathological examination of implants demonstrated cellular integration into ePTFE material and reduced capsular thickness around the ePTFE versus the control devices. Cellular integration into the outer layer, but not into or through the inner layer, of the ePTFE devices was apparent microscopically, as was the formation of a fibrous capsule outside the integrated cellular layer (Figure 4a Table S1).
While there was a trend toward increasing High capsular thickness between 1 and 3 months, this change was not significant (p = .18).

AS-OCT images of control and High implants qualitatively suggest
greater capsule density over control devices (Supplemental Figure S2).
To assess the effect of bleb height on capsular thickness in ePTFE devices, we compared ePTFE devices with identical footprints but differing heights when inflated: High, Low, and Flat  Table S1). There was no significant difference in capsular thickness between High and Low implants (p = .18), both of which were able to inflate with AH exposure, though to different degrees.

| Permeability assessment under constant-flow perfusion
The reduced capsular thickness of blebs associated with ePTFE implants versus controls suggested that ePTFE-associated blebs could be more permeable to AH. To test this hypothesis, we performed fluo-  implants. For both control and High devices, after the highest perfusion rate was reached, fluorescein infusion was stopped and pressure was monitored continuously while the seal between the irrigation cannula and the tube was maintained. Pressure remained high in blebs overlying control devices, with minimal detectable decay, or decrease in "p ratio" (Figure 5d). In ePTFE-associated blebs, a rapid pressure decay (p ratio decrease) was observed over the course of several minutes (Figure 5d). The rate of decay shown in Figure 5d was notably higher in ePTFE-associated blebs (red lines) compared to control blebs (black lines). Due to challenges associated with prolonged perfusion (involuntary animal movement, cannula dislocation, fluorescein backflow from control blebs) complete p ratio curves obtained were too few for statistical analysis (five total; two control, three ePTFE), but observed decay behavior was consistent with that described for all devices.

| Complications
Three implants, all control type, were excluded from the study data due to complications. One had an area of conjunctival erosion noted on postoperative Day 40 and it was explanted. A second control was explanted due to exposed implant on Day 41. The third implant had separation of tube and plate noted immediately prior to sacrifice. Our studies showed no difference in bleb capsule thickness whether devices of the same geometry were exposed or not exposed   Several bleb features may contribute to permeability, including capsular thickness and surface area. Prior permeability studies using a different method from ours to examine non-ePTFE implants in rabbits found very low permeability in healed implant blebs that decreased both with increased thickness and with fluid challenge; this relationship is not fully understood. 16,17,28 In addition to capsule thickness, bleb surface area also determines permeability. 13 In our studies, all implants had identical footprints and their overlying blebs were generally restricted to their footprint. Since bleb heights were the same in ePTFE (inflatable) devices and controls, the internal surface area for AH diffusion was similar. Yet, ePTFE devices had better permeability, indicating that their thinner capsules were more permeable either due to thinness alone, or due to favorable characteristics of the interaction of ePTFE membranes with the capsule wall.
ePTFE has been evaluated previously in glaucoma surgery. [33][34][35][36][37][38][39] The Ahmed implant was modified with an outer ePTFE membrane to create the PRIME-Ahmed, which showed increased resistance to fluid outflow in vitro, but created a thinner capsule in rabbit eyes when compared to unmodified Ahmed implants. 34,36 Of note, blebs associated with the PRIME-Ahmed showed a greater degree of vascularization, collagen disorganization, and chronic inflammation compare to control implants; these features were not noted in our histopathological exams. More recently, an ePTFE-based membrane-tube-type glaucoma shunt device was described as having good biocompatibility and IOP reduction in rabbit eyes, as well as IOP reduction in patients with refractory glaucoma. 37,40,41 Our implants differed from both previous devices in that the microstructure of the unique, bilayered ePTFE used here was chosen and manufactured specifically for this application. We further add to previous studies by (a) including silicone controls with matching footprints to ePTFE devices, (b) varying device height to clarify the effect of device geometry on capsule thickness, (c) evaluating the independent contribution of AH flow to capsule thickness, and (d) evaluating bleb permeability.
These studies have several limitations. First, while NZW rabbits are an established model in glaucoma surgery, there are aspects of their anatomy and fibrotic response that do not mirror human eyes. In NZW rabbits, implants needed to be placed significantly closer to the limbus than would be traditionally done in humans to avoid venous sinuses within the orbit. Additionally, rabbit filtering surgeries typically fail within 4 weeks, 42 which is significantly more rapid than in the average human eye. Our surgeries allowed for immediate, unrestricted flow from the anterior chamber into the subconjunctival space; while this approach is tolerated in rabbit eyes, 43 it does not recapitulate surgery in humans, in which early postoperative flow restriction is common with some implants. Our approach, however, could potentially expose the subconjunctival space to more inflammatory cytokines in the AH in the immediate postoperative period. Our sample size was not powered to detect a change in IOP in response to surgery. The exclusion of IOP as a primary outcome was based on previous studies that did not observe IOP reduction after GDI placement in normotensive eyes; 30 to detect IOP change would have required not only a larger sample but also a protocol to induce ocular hypertension in rabbit eyes. Finally, we followed implants through 3 months after placement. While this time period was described previously as sufficient for bleb stabilization, 30,44 future studies should follow implants for longer durations.

| CONCLUSIONS
These studies demonstrated thinner capsule formation and increased permeability of ePTFE implants compared to silicone controls in rabbit eyes, which was sustained at 3 months. Capsule geometry-and not AH flow-determined capsule thickness in this model as well. These findings could aid in the future design of implants for glaucoma surgery.