Investigating the Tolerance of Corneal Endothelial Cells to Surgical Fluid Pressure Using Intact Porcine Eyes

Abstract

Purpose: It remains unclear if fluid pressures used during cataract surgeries contribute to iatrogenic corneal endothelial cell (CEC) loss.

Methods: A custom experimental platform was used to pressurize the anterior chamber of explanted porcine eyes to surgical fluid pressures of 60 mm Hg or 400 mm Hg for 5 minutes or 60 mm Hg for 45 minutes (n = 8 or 9 per group). The corneal endothelia were stained with a unique combination of nucleic acid viability dyes and were imaged using fluorescence microscopy without removing the cornea from the globe. The images were analyzed using custom code to quantify acute CEC loss as a percentage of CEC injury/death (PCI). The PCI values from the surgical pressure groups were compared to sham controls perfused to physiological pressures for each surgical duration. As a positive control, a group of eyes (n = 8) was perfused with deionized water to intentionally induce CEC injury/death.

Results: No significant differences were observed in mean PCI values between any of the surgical fluid pressure groups and sham control groups, at either duration. The mean PCI of the positive control group was significantly different from all other groups, indicating that the method is able to detect CEC injury/death.

Conclusions: Our results suggest that the magnitudes of fluid pressure used over the duration of a cataract surgery do not significantly contribute to acute iatrogenic CEC loss.

Translational Relevance: Not only do the findings of this study answer a longstanding clinical question related to cataract surgery, but the platform introduced will facilitate testing how new cataract surgery devices and techniques affect CEC viability.

Introduction

The non-proliferative corneal endothelial cells (CECs) located on the posterior side of the cornea are vital for maintaining corneal transparency by regulating the hydration and thickness of the corneal stroma through a “pump-leak” mechanism.¹ When the endothelial cell density (ECD) decreases ∼80% from healthy adulthood levels, a cornea is at risk of developing corneal decompensation, or failure of the CECs to maintain stromal hydration.² The resulting overhydration of the cornea is known as corneal edema or as bullous keratopathy in advanced stages. An individual with corneal edema may suffer from blurred vision, discomfort, severe pain, and even blindness if left untreated.^3,4 

A leading etiology for corneal decompensation, and subsequent corneal edema, is CEC damage during intraocular surgeries.^2–4 This includes after phacoemulsification cataract surgery with intraocular lens implantation (hereinafter referred to as cataract surgery), which has been associated with CEC losses of 2% to 42% within the first year after surgery.^5–15 The CEC damage during cataract surgery results in a specific form of irreversible corneal edema known as pseudophakic corneal edema (PCE) or pseudophakic bullous keratopathy (PBK). PCE or PBK occurs after approximately 1% to 2% of the more than 20 million cataract surgeries performed every year^16–18 and is a leading indication for corneal transplantation.^19–24 

A compelling argument can be made that mechanical loading may play a significant role in acute CEC damage during surgery. For example, during cataract surgery, CECs are subjected to contact forces from collisions with crystalline lens fragments,¹⁰ factors imposed by ultrasound-induced cavitation (e.g., oxidative stress and mechanical shock waves),^25,26 and fluid flow from phacoemulsification devices up to 40 cc/min (40,000 µL/min).²⁷ CECs are also subjected to intraocular pressures (IOP) greater than 60 mm Hg for 50% to 85% of the procedure (with peak pressures exceeding 400 mm Hg),²⁸ which previous studies suggest can put a patient at risk for other iatrogenic injuries including retinal damage.²⁹ These surgically imparted mechanical loads are in stark contrast to those normally experienced by CECs. Under normal physiological conditions, CECs live in a relatively stable mechanical environment, primarily experiencing IOP levels that are maintained within a range of 9 to 21 mm Hg and a steady fluid flow due to aqueous humor exchange (2.5 µL/min).^30–33 

Our lab has previously shown that porcine CECs are highly susceptible to mechanical trauma via contact indentation. Specifically, a custom corneal endothelial indentation platform was used in combination with fluorescence microscopy to determine that CEC injury/death occurs when contact pressures exceed a critical threshold of ∼43 mm Hg.³⁴ Although this critical contact pressure is consistent with findings that ECD decreases in acute angle-closure glaucoma patients (where IOP typically exceeds 40 mm Hg),^35–38 indentation-based contact pressure and IOP are distinct mechanical loading scenarios. Indentation exerts a spatially varying contact pressure on the endothelium, which, in conjunction with direct material interactions (i.e., friction and adhesion), induces complex distributions of shear and axial stresses in the endothelium. In contrast, IOP distributes the same (normal) fluid pressure evenly onto all parts of the corneal endothelium. Additionally, although sudden, large increases in IOP have been shown to disrupt the structure and function of CECs,³⁹ these observations were made in the context of acute glaucoma, which subjects the endothelium to lower magnitudes and longer durations of IOP compared with cataract surgery. Therefore, it remains unclear if surgical fluid pressure (magnitude and duration) contributes to CEC damage during cataract surgery. 

To address this knowledge gap, we used a custom experimental setup to apply controlled, measurable fluid pressures with surgical magnitudes and durations to the corneal endothelium of intact porcine eyes. A novel in situ staining and imaging (ISSI) method, which includes a unique viability staining assay and image processing/analysis pipeline, was introduced that facilitates quantification of CEC injury/death immediately after application of these fluid pressures without removing corneas from their globes. We hypothesized that more CEC injury/death would be observed with increasing fluid pressures and durations, with notable differences between eyes pressurized to extreme surgical magnitudes or durations and those maintained at physiological pressures. Surprisingly, our findings revealed that CECs can tolerate extremely large magnitudes of fluid pressure and suggest that surgical fluid pressure alone does not result in clinically significant acute CEC loss. 

Materials and Methods

Specimen Preparation

Fresh enucleated porcine eyes, from a random mix of male and female pigs with an average age of 4 to 6 months, were obtained from a slaughterhouse (Sioux-Preme Packing Co., Sioux City, IA, USA). The eyes were transported to the laboratory on ice (arriving ∼24 hours after euthanasia) and then placed in a refrigerator at 4°C for storage prior to experimentation. At the time of experimentation (less than 72 hours after euthanasia), an eye was removed from the refrigerator and the globe was carefully cleared of most of the extraocular muscles and orbital fat with gross dissection scissors, leaving a small amount of tissue on the posterior side of the globe. The intact eye was then placed in a Pad 4 Pigs eye holder (PN0154; eyecre.at GmbH, Tirol, Austria) such that the cornea was facing up and centered, and the remaining ocular tissue on the back of the globe was used to secure the eye to the pad with fixation pins (PN0307; eyecre.at GmbH). The cornea was hydrated with phosphate-buffered saline (PBS) throughout the preparation process. The eye remained fixed to the Pad 4 Pigs eye holder at room temperature (RT) throughout the fluid pressure experiment until it was prepared for image acquisition. 

Experimental Fluid Pressure Setup

A custom experimental platform (Fig. 1) was developed to perfuse the intact eyes to precise, measurable anterior chamber pressures (ACPs). A microfluidic pressure controller (OB1 MK3+; Elveflow, Paris, France), connected to a compressed air supply (the pressure source), was used to pressurize a reservoir (KPT-M-2; Elveflow) containing balanced salt solution (BSS; 0065079550 or 0065179504; Alcon, Geneva, Switzerland). This generated a pressure differential between the pressure inlet of the BSS reservoir and the liquid outlet, causing pulseless fluid flow through a series of standard three-way stopcock valves and out of a 30-gauge × ½ inch long PrecisionGlide perfusion needle (305106; Becton, Dickinson and Company, Franklin Lakes, NJ, USA). The desired input pressures were set using the Elveflow software. The series of three-way valves between the BSS reservoir and perfusion needle, along with the tubing and waste tube, were designed to allow the staining solution to be primed without interrupting the BSS perfusion and then enter the anterior chamber of the eye through the same perfusion needle at a controlled pressure. At the opposite side of the eye, a 30-gauge × 1 inch long PrecisionGlide vacuum needle (305128; Becton, Dickinson and Company) was connected to another three-way valve that routed fluid to the inlet of an Elveflow microfluidic pressure sensor (MPS-V2-S-3). The pressure sensor contains a deformable membrane that uses a method similar to piezoelectric strain gauges to measure the gauge pressure of the fluid inside the sensor. This allowed for precise measurement of ACPs within ±4 mbar (±3 mm Hg) downstream from the pressure controller, which was continuously monitored using the Elveflow software. Fluid was pulled out of the anterior chamber of the eye and into the pressure sensor at controllable flow rates (such as a flow rate of 2.5 µL/min to represent physiological aqueous humor exchange, although the rates depended on the experimental protocol step) using a custom-built syringe pump attached to the outlet of the pressure sensor. An additional three-way valve was placed between the pressure sensor outlet and the syringe pump to allow the syringe to be emptied into a waste tube while still maintaining a closed system. A column of PBS was hung directly above the eye to drip PBS onto the center of the corneal epithelium at a constant rate using a modified intravenous flow regulator. Note that all of the Elveflow instruments use units of millibar (mbar). Thus, even though the protocol and results discuss all pressures in terms of millimeters of mercury (mm Hg), the input pressures were actually set to mbar values that were equivalent to the desired pressures in mm Hg, and all mbar pressure measurements were converted to mm Hg (1 mbar = 0.7501 mm Hg) during the experiments. 

Figure 1.

 

Schematic of experimental fluid pressure setup highlighting important features/equipment (created with BioRender.com).

Study Design and Fluid Pressure Experiment Protocol

Fluid pressures were applied to the corneal endothelium of the intact porcine eyes by perfusing the anterior chambers with BSS to different ACP magnitudes (as measured by the pressure sensor) for specific durations using the aforementioned fluid pressure setup (Fig. 1). Five primary experimental groups were assessed: (1) a sham control group that was perfused to a physiological ACP of 15 mm Hg for 5 minutes (n = 8); (2) a surgical pressure group that was perfused to an ACP of 60 mm Hg for 5 minutes (n = 8); (3) a surgical pressure group that was perfused to an ACP of 400 mm Hg for 5 minutes (n = 9); (4) another sham control group that was perfused to a physiological ACP of 15 mm Hg for 45 minutes (n = 8); and (5) a final surgical pressure group that was perfused to an ACP of 60 mm Hg for 45 minutes (n = 8). 

A detailed experimental protocol is available in the Supplementary Materials (Supplementary Figs. S1–S15). In summary, the perfusion and vacuum needles were carefully placed into the anterior chamber of intact porcine eyes ∼180° from each other by piercing their tips through the cornea just anterior/superficial to the limbus. The eyes, regardless of experimental group, were then pressurized to a baseline ACP of 15 mm Hg for 10 minutes. After maintaining a baseline ACP, the eyes were pressurized to the group-specific physiological or surgical ACP for either 5 minutes or 45 minutes (depending on the experimental group). Up to this point, the syringe pump was set to pull fluid out of the anterior chamber at a flow rate of 2.5 µL/min. At the end of the experimental duration, the flow of BSS into the eye was stopped, and the syringe pump was fully advanced to pull all fluid out of the anterior chamber (i.e., to flush the anterior chamber). A staining solution was then perfused into the anterior chamber at a pressure of 15 mm Hg for 5 minutes. The staining solution was made up of 10 µL of hexidium iodide (HI, H7593; Thermo Fisher Scientific, Waltham, MA, USA) and 10 µL of SYTOX Green Nucleic Acid Stain (S7020; Thermo Fisher Scientific) in ∼0.5 mL of PBS. HI is a red fluorescent nucleic acid dye that is permeant to the cell membrane and therefore stains all CEC nuclei (alive or dead), whereas SYTOX Green is a green fluorescent nucleic acid dye that can only penetrate compromised membranes and therefore only labels CECs that are injured (i.e., cells with temporarily ruptured membranes) or dead. The anterior chamber was flushed once again and then perfused with the remaining staining solution for another 5 minutes. When the ACP reached 15 mm Hg, the needles were removed from the anterior chamber and the corresponding incisions were sealed with small drops of cyanoacrylate glue (Super Glue Ultra Gel Control; Loctite, Düsseldorf, Germany). PBS was dripped onto the corneal epithelium throughout the entire experiment to maintain anterior corneal hydration. 

Injured Cell Stain Validation

An additional positive control group (n = 8) was included to validate that injured/dead CECs were stained with SYTOX Green during the fluid pressure experiments. To intentionally induce CEC injury/death, additional steps were added to the experimental protocol to perfuse the anterior chamber of the eyes with deionized water (DI-H₂O), which is known to damage CECs,⁴⁰ likely due to osmotic lysis. The positive control eyes went through a similar protocol as the 5-minute sham control samples, with additional steps to flush the anterior chamber and introduce the DI-H₂O in a similar manner as the staining solution, after perfusing the eye to the physiological ACP of 15 mm Hg for 5 minutes and prior to staining (Supplementary Figs. S4–S10). 

Fluorescence Image Acquisition, Processing, and Analysis

Immediately after completing the fluid pressure experiments, the eyes were removed from the Pad 4 Pigs eye holder and placed cornea-down into a custom eye holder filled with PBS (Fig. 2A). The eyes mounted in the holder were covered, allowed to sit for 5 minutes at RT for stain incubation, and then placed on top of a thermomixer set to 300 rpm for an additional 5 minutes at RT to ensure that the staining solution was evenly distributed throughout the anterior chamber. The eye holder was then mounted onto the stage of an Olympus IX81 inverted microscope with fluorescence imaging capabilities (Olympus, Tokyo, Japan). For each eye, a single multichannel fluorescence tiled (stitched) z-stack image of the endothelium, large enough to cover the pupil of the eye and centered at the approximate center of the pupil, was acquired through the corneal epithelium and stroma at 4× magnification (30 tiles per image, ∼25 z-steps at a step size of 40 µm, and an exposure time of 300 ms for both channels). 

Figure 2.

 

Imaging workflow. (A) Fluorescent z-stack corneal endothelial images were acquired in intact porcine eyes using an epifluorescence microscope at 4× magnification. (B) The acquired images were then processed by marking locations of in-focus CECs for each plane (B-1), fitting a 3D surface to those points (B-2), and using the 3D surface fit to create height-focused circular ROIs with a diameter of 6.5 mm (B-3). (C) Zooming into a representative endothelial image demonstrates the ability of the staining assay to clearly label individual CEC nuclei. (D) The processed ROI images were then analyzed to obtain injured/dead and total CEC nuclei counts (images demonstrate how yellow markers are plotted on centroids of CEC nuclei that were identified by a custom MATLAB image analysis code).

The multichannel z-stack images were imported into Fiji (ImageJ), split into two different channel z-stacks, and saved as .tiff images. The HI channel z-stack dimensions were converted to pixels, and the ImageJ multipoint tool was used to manually plot points at locations of in-focus CEC nuclei in each plane. The point coordinates were saved as .csv files that were imported into a custom MATLAB (MathWorks, Natick, MA) image processing code that first fit a three-dimensional (3D) surface (2 × 2 polynomial) to the point coordinates. It then utilized the surface fit equation to create a height map that was used to generate height-focused, flattened images of both channels. The code then ran the height-focused images through tile-correcting and intensity normalization algorithms, before cropping out circular regions of interest (ROIs) with a 6.5-mm diameter. The ROIs were centered at the apex of the corneas, which the code identified by locating the minimum z-position of the 3D surface for each sample (Fig. 2B). 

The processed 6.5-mm-diameter HI channel ROI images (Fig. 2C) were analyzed using another MATLAB code adapted from particle tracking functions created by Blair and Dufresne⁴¹ that automatically identified the centroids of the CEC nuclei. We tested this code's ability to accurately count the number of CEC nuclei within a small area in all of the processed positive control images, which demonstrated that the code can automatically count CEC nuclei with an accuracy of ∼96% when compared to manual CEC counts within the same area (see the Supplementary Materials for details). The coordinates of the auto-counted CEC nuclei were imported into Fiji and overlaid onto the HI channel ROI image, and a manual adjustment was performed to remove or add any incorrectly marked or missed nuclei using the multipoint tool. The manually adjusted HI nuclei coordinates were used to obtain the total CEC count (TC) and saved as a new spreadsheet file for each sample. Due to the proclivity of the SYTOX Green to also stain keratocytes within the corneal stroma (which could interfere with CEC counts), an additional colocalization filtering step was added to the MATLAB particle tracking code when analyzing the processed SYTOX Green channel images. This colocalization filter eliminated incorrectly auto-counted nuclei for which the centroid coordinates were not within a specific Euclidian distance from the HI nuclei coordinates. Similar to the HI channel images, a manual adjustment of the auto-counted nuclei from the SYTOX Green channel was then applied in Fiji. For the manual adjustment of the SYTOX Green channel, the coordinates of the auto-counted injured/dead CEC nuclei were overlaid onto a merged image that contained both the SYTOX Green and HI channel nuclei. This allowed for visual comparison between the auto-counted SYTOX Green channel nuclear coordinates and the shape and position of the CEC nuclei in the HI channel, which were the primary criteria used to determine if a keratocyte was unintentionally counted. Note that the spindle-shaped keratocyte nuclei have a distinct morphology from CEC nuclei, which appear bean shaped. The manually adjusted SYTOX Green nuclei coordinates were used to obtain the final injured/dead CEC count (IC). One sample from the 60-mm Hg, 45-minute group was excluded from the study because the keratocytes were especially widespread in the acquired images and overlapped with CECs, making it difficult to distinguish CECs and obtain an accurate count of injured/dead CECs (even after manual adjustment). 

Quantification of CEC Injury and Statistical Analysis

The total and injured/dead CEC counts (TC and IC, respectively) obtained from image analysis were used to calculate the CEC loss for each sample as a percent CEC injury (PCI): 

 

\begin{eqnarray} {\ PCI (\rm \%) = (IC/TC) \times 100} \end{eqnarray}

(1)

 

 A one-way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test was used to compare the mean PCI values across all experimental groups (including the positive control group). The threshold for significance was set to α = 0.05. The ANOVA was performed in Prism 10.2.1 for Windows (GraphPad Software, Boston, MA). 

Results

There were no significant differences in mean PCI values between eyes that were perfused to a physiological ACP of 15 mm Hg for 5 minutes (mean = 0.44%, SD = 0.62%) and those that were perfused to surgical ACPs of 60 mm Hg (mean = 0.41%, SD = 0.52%) or 400 mm Hg (mean = 0.31%, SD = 0.33%) for 5 minutes (P > 0.9999 for both comparisons). Similarly, there was no significant difference in PCI values between eyes that were perfused to a physiological ACP of 15 mm Hg for 45 minutes (mean = 0.43%, SD = 0.30%) and those that were perfused to a surgical ACP of 60 mm Hg for 45 minutes (mean = 1.63%, SD = 2.79%; P = 0.9999). There were also no significant differences in PCI values between any of the aforementioned groups when comparing across durations. Conversely, there were significant differences in PCI values between the positive control group that was perfused with DI-H₂O (mean = 49.5%, SD = 25.2%) and all other groups (P < 0.0001 between the positive control group and all other groups) (Fig. 3). 

Figure 3.

 

Quantification of injured CECs for each experimental group. There was a significant difference (P < 0.0001) in PCI between the positive control group and all other groups, but no significant differences in PCI between any of the other groups. The yellow and purple dotted lines indicate the low and high ends of clinically observed CEC loss within 1 year following cataract surgery (2% and 42%), respectively.

Qualitatively, the acquired endothelial images were consistent with the quantitative results (Fig. 4). In particular, very few injured CEC nuclei were visible across the analyzed ROIs of the physiological and surgical pressure groups at both durations. In contrast, widespread injured/dead CEC nuclei were observed in the positive control group. 

Figure 4.

 

Representative corneal endothelial images from each experimental group, showing the full 6.5-mm processed/analyzed ROI (top) and zoomed in views corresponding to the yellow boxes to show individual CEC nuclei (bottom).

Discussion

In this study, we developed a custom experimental platform that enables precise application of measurable fluid pressures in the anterior chamber of intact eye explants, as well as a novel ISSI method that together allow for assessment of CEC damage triggered by fluid pressure alone. This approach was used to determine if surgically relevant magnitudes of fluid pressures applied for surgically relevant durations resulted in acute CEC injury/death in porcine corneas. Contrary to our hypothesis, we found no significant differences in mean PCI values between eyes that were perfused to a physiological pressure (15 mm Hg) versus those that were perfused to surgical magnitudes of pressure, for both short (5 minutes) and long (45 minutes) surgical durations. Notably, the mean PCI values measured in all of the surgical pressure groups, even including eyes perfused to 400 mm Hg (>25 times higher than physiological IOP), were less than the low end of CEC loss observed clinically within 1 year after cataract surgery (2% to 42%).^5–15 The positive control eyes that were perfused with DI-H₂O did have significantly higher PCI values than all other groups, demonstrating that our staining assay was capable of detecting CEC injury/death in our experimental setup. Interestingly, the mean PCI value for the positive control group (49.5%) closely resembled the high end of CEC loss observed clinically within 1 year after cataract surgery (42%),^5–15 allowing the DI-H₂O treatment to serve as a benchmark for extreme cases of iatrogenic CEC loss (Fig. 3). These results strongly suggest that fluid pressure alone does not result in clinically significant acute iatrogenic CEC damage. 

To our knowledge, this was the first study to isolate the effects of surgical fluid pressures, with magnitudes and durations specific to cataract surgery, on acute CEC loss. Numerous studies measuring CEC loss after cataract surgery report differences in ECD as a result of different surgical techniques,^6–9,11,12 but they are not able to distinguish how each of the individual factors that may affect CEC health during cataract surgery (e.g., ultrasound energy, free radical production, contact with lens fragments and/or instruments and fluid flow) contribute to the total iatrogenic CEC damage. Wenzel et al.⁴² used a research model that eliminated some of these factors by investigating the influence of bottle height during irrigation and aspiration (I/A) on CEC loss in porcine eyes. They found that short durations (10 minutes) of I/A without ultrasound energy led to significant CEC loss when compared to controls, with more CEC loss occurring at higher bottle heights. Although bottle height is closely related to IOP (i.e., a higher bottle height results in larger hydrostatic pressure), Wenzel et al.⁴² acknowledged that bottle height also directly influences turbulent fluid flow within the anterior chamber. Thus, their experimental approach did not isolate the effects of fluid pressure on CEC viability. We believe that the results of our study expand on the work by Wenzel et al. by delineating the specific contribution of IOP to CEC loss by (1) eliminating the confounding variable of pressure-dependent turbulent fluid flow, and (2) actively measuring the static fluid pressure applied to the endothelium (the ACP). 

Compared with previous studies, we investigated a broader range of pressure magnitudes (15–400 mm Hg) and durations (5–45 minutes). This approach should account for any level of pressure conceivably encountered during cataract surgery, even in extreme cases. For example, Khng et al.²⁸ measured mean IOPs as high as 220 mm Hg and transient spikes exceeding 400 mm Hg during simulated cataract surgeries, although these values are much higher than other values reported in the literature.^29,43–45 Similarly, it is difficult to clearly identify the duration for which CECs experience high pressures during a normal cataract surgery due to differences in how surgical durations are measured and reported.^{28,29,44–50} Consequently, we chose to pressurize eyes to 60 mm Hg for 45 minutes, which surpasses the International Council of Ophthalmology (ICO)–Ophthalmology Surgical Competency Assessment Rubrics (OSCARs) resident/surgeon training competent criteria of 30 minutes (for a total surgery).^51,52 Therefore, 45 minutes at an elevated pressure may represent what could occur during a complicated surgery or a surgery completed by a less experienced surgical resident.^46,49,50 

A key benefit of our platform is that it allows for in situ viability staining that reduces the risk of measuring CEC damage that occurs as a result of sample preparation/manipulation. Specifically, in contrast with other studies,^34,42 our ISSI approach allows for assessment of CEC loss without dissecting/removing the cornea from the globe. Additionally, the stain enters the anterior chamber of the intact eye using the same needle that is used to pressurize the eye earlier in the experiment. Hence, there is no need to remove and reinsert this needle, make any new incisions, or introduce any other instruments into the anterior chamber. We believe all of these considerations strengthen our conclusion that the fluid pressure used during cataract surgery alone is not a primary contributor to CEC loss during cataract surgery. 

Our findings are consistent with previous clinical studies. For example, it has been reported that LASIK surgery has no long-term effect on CEC loss.^53,54 These findings are notable because IOP can be as high as 82 mm Hg to >140 mm Hg during LASIK surgery,^55,56 but CECs in eyes undergoing this procedure do not experience other possible contributors to CEC loss including turbulent fluid flow, lens fragmentation, and ultrasound cavitation. 

Although our data appear to suggest that high surgical IOP levels are harmless, our findings should be interpreted with caution, as high IOP imparted during eye surgery could lead to other severe complications outside of CEC viability. For example, studies have found that acute IOP elevation common to cataract surgery can lead to impairment of ocular blood flow, inhibit transport of neurotrophins from the brain to the retina, and cause cellular and molecular retinal injuries.^29,44,57 We also acknowledge that, although our study provides a direct indication of immediate CEC death that is valuable for understanding acute CEC loss that occurs during an intraocular surgery, it may not capture the full extent of CEC damage. Li et al.³⁹ demonstrated that in rats sudden increases in IOP (∼83 mm Hg for 2 hours) disturbed the apical junctional complex (AJC) integrity of the corneal endothelium and the expression of Na, K-ATPase, thereby compromising the endothelial barrier function. Thus, it is possible that the CEC plasma membrane can tolerate larger mechanical loads than the endothelial AJC, and CECs might be functionally damaged prior to membrane rupture. It is also possible that acutely exposing CECs to high IOP might result in delayed/long-term CEC damage that was not captured in this study. For example, subjecting CECs to high fluid pressure may activate apoptotic pathways that trigger cell death over many hours or days, but our cell viability assessments were made <3 hours after pressurization. Nevertheless, we believe the platform we have developed is amenable to use in conjunction with other assays, such as the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) apoptosis assay or immunostaining of flatmounted whole corneas.⁵⁸ Our approach can also be adapted to facilitate assessment of CEC changes after several hours or days (e.g., by integrating perfusion/exchange of media into the eye with optimized eye storage conditions) to enable longer term evaluation of CEC health after anterior chamber pressurization. However, another animal model (e.g., rabbit or rat) would likely be required to confirm that acute application of surgical fluid pressure does not contribute to an increased rate of CEC loss over a longer duration (1 months or more). 

Although results obtained from animal models should not be overgeneralized to make definitive conclusions about human surgeries, we believe that the porcine eye was an appropriate model for this study due to its physiological and anatomical similarities to the human eye.⁵⁹ In particular, pigs have a similar CEC density to humans (∼2800–3400 cells/mm² in pigs 5 to 10 months old⁶⁰ versus ∼2500–3000 cells/mm² in adult humans),⁶¹ both porcine CECs and human CECs lack regenerative capacity in vivo,⁶⁰ and porcine eyes have the same general pattern of age-related CEC decline as human eyes.⁶⁰ Furthermore, as an accessible and cost effective model with similar macroscopic size and anatomy compared to human eyes, porcine eyes are the most frequently used model in surgical wet lab training.^59,62 This feature was an important consideration because developing our method in the porcine eye model demonstrates its utility for use in the wet lab setting for clinical training purposes and/or preclinical testing of new phacoemulsification devices. However, there are limitations to using porcine eyes for studies related to corneal biomechanics. For example, at the tissue level, porcine corneas are thicker^59,63,64 and less stiff⁶⁵ and exhibit different viscoelastic behavior when compared to human corneas.⁶⁶ It is also plausible that porcine CECs experience different cell-level stresses when subjected to external mechanical loads (such as fluid pressure) than human CECs because they have more intrinsic morphological heterogeneity than human CECs.⁶⁰ Although we do not expect the conclusions of this study to be model specific, repeating these experiments in human tissue is warranted. To this end, we believe that our experimental platform can be readily applied to human cadaver eyes in future studies. 

Even though contact pressure is a distinct loading environment from fluid pressure, it is striking that porcine CECs can withstand 400 mm Hg of fluid pressure but immediately die with 43 mm Hg of contact pressure during indentation.³⁴ These contrasting findings demonstrate that the out-of-plane compressive and in-plane tensile stresses triggered by contact pressure are likely not directly responsible for acute CEC injury/death during contact indentation, and they suggest that revisiting how and why CECs are harmed during contact indentation is warranted. 

To our knowledge, this study was the first to use the combination of HI and SYTOX Green to assess CEC viability. Although HI is typically used to identify Gram-positive bacteria,^67,68 we demonstrated that HI stain is highly effective at labeling CEC nuclei in porcine corneas (see Supplementary Fig. S16). Although the seminal studies of CEC damage after simulated surgeries by Terry and others have established robust CEC viability measurements using dyes such as Calcein AM and Trypan Blue,^69–78 we believe that our combination of stains can overcome some drawbacks of these traditional viability dyes. In particular, in our experience, Calcein AM and Trypan Blue live stains often do not define discernable borders between cells and can exhibit poor contrast between live and dead cells when not thoroughly washed. However, HI (in combination with SYTOX Green) allows for clear identification of CECs and easy quantification of CEC death at single-cell resolution, even when imaging is performed through the full thickness of the cornea. Interestingly, we have found that HI can also be used to stain and image corneal and lens epithelial nuclei in intact porcine eyes (see Supplementary Fig. S17). The primary challenge we encountered using this combination of stains was that the SYTOX Green stain had a tendency to diffuse into the stroma and stain keratocytes, sometimes with a fluorescence intensity similar to that of the CECs. This made it difficult, even when applying filtering algorithms (see Methods), to avoid counting keratocytes when assessing CEC death. Fortunately, in most cases it was easy to visually distinguish keratocytes (due to their unique nuclear morphology) and remove them from analysis during the manual adjustment step. 

Collectively, the results of this study strongly suggest that surgical fluid pressure alone is not responsible for acute CEC loss during cataract surgery. Our findings, along with the novel experimental platform and ISSI method introduced in this study, open new avenues for future work pinpointing the key factors that drive iatrogenic CEC injury in cataract and other surgeries. For example, our experimental approach can be readily deployed to assess how different surgical techniques or devices (e.g., new phacoemulsification handpieces) impact CEC injury/death in simulated surgeries with porcine or human eyes, ultimately paving the way toward our long-term goal of developing novel surgical techniques and medical devices that avoid iatrogenic CEC injury and improve intraocular surgical outcomes. 

Acknowledgments

The authors thank Albert J. Bae, PhD (Lewis & Clark College, Department of Physics), for developing MATLAB code that was an integral part of our image processing workflow. The authors also thank Lindsay Rathbun Wysocki, PhD, of the HCIC Microscopy Core at the University of Rochester for her technical support during the use of confocal microscopy to validate our staining assay, and Aziz Benamara, PhD (Elveflow), for his guidance while integrating the microfluidics equipment into our experimental platform. Finally, we thank Michael Giacomelli, PhD (University of Rochester, Department of Biomedical Engineering), for his recommendation to try hexidium iodide for cell viability staining. 

Supported by the National Eye Institute, National Institutes of Health (T32EY007125) and by Carl Zeiss Meditec Cataract Technology, Inc. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. 

Disclosure: A.J. McMullen, Carl Zeiss Meditec Cataract Technology, Inc. (F); Z.A. Dantsoho, Carl Zeiss Meditec Cataract Technology, Inc. (F); S. Chamness, Carl Zeiss Meditec Cataract Technology, Inc. (E); J. Brunelle, Carl Zeiss Meditec Cataract Technology, Inc. (C); J. Martiz, Carl Zeiss Meditec Cataract Technology, Inc. (E); Y.M. Khalifa, Carl Zeiss Meditec Cataract Technology, Inc. (F); M.R. Buckley, Carl Zeiss Meditec Cataract Technology, Inc. (F)