Storage of Corneas With Predissected Descemet Membrane in an Active Storage Machine

Abstract

Purpose

We previously demonstrated the superiority of an active storage machine (ASM; which restores a physiologic intraocular pressure [IOP]) over conventional passive storage for whole corneas. To reduce surgical time, corneas are often predissected in eye banks, 2 to 4 days before surgery for endothelial keratoplasty. The aim is here to demonstrate the possibility of also storing predissected Descemet membrane endothelial keratoplasty (DMEK) in ASM so that their use is not limited to penetrating keratoplasty and to extend the storage life of these endothelial grafts for much more flexibility.

Methods

In this preclinical study, pairs of human corneas were predissected, leaving the Descemet membrane attached only by a small central area. After randomization, they were stored in the same commercial organ culture medium, either in the ASM (21 millimeters of mercury [mm Hg], 2.6 µL/min) or in a sealed flask, for 3 or 10 days. Final assessment consisted of measuring the viable endothelial cell density (ECD), corneal thickness, and assessing the expression of CD166 and NCAM, ZO-1 (morphology and junctions), Na⁺/K⁺ATPase (endothelial pump) and COX-IV (mitochondria) by immunostaining.

Results

Final endothelial cell (EC) viability was comparable between the two groups for both storage periods, as were the markers studied by immunofluorescence. However, corneal thickness was lower in the ASM group.

Conclusions

In conclusion, the ASM can be used for predissected DMEK for at least 10 days, thus enabling to organize runs of graft preparation and subsequent safe storage while simplifying logistics.

Translational Relevance

ASM can be used to preserve pre-cut DMEK grafts, providing greater flexibility in surgical planning.

Objectif

Nous avons précédemment démontré la supériorité d'un dispositif de stockage actif (ASM, qui rétablit une pression intraoculaire physiologique) par rapport au stockage passif conventionnel pour les cornées entières. Afin de réduire la durée de la chirurgie, les cornées sont souvent prédécoupées dans les banques de tissus, 2 à 4 jours avant la chirurgie, pour une greffe endothéliale. L'objectif ici est de démontrer la possibilité de stocker également des DMEK prédécoupées dans l'ASM afin que son utilisation ne se limite pas à la kératoplastie transfixiante et de prolonger la durée de conservation de ces greffons endothéliaux pour une plus grande flexibilité.

Méthodes

Dans cette étude préclinique, des paires de cornées humaines ont été prédisséquées, en laissant l'endothélium attaché seulement par une petite zone centrale. Après randomisation, elles ont été conservées dans le même milieu commercial d'organoculture, soit dans l'ASM (21 mmHg, 2,6 µL/min), soit dans un flacon scellé, pendant 3 ou 10 jours. L'évaluation finale a consisté à mesurer la densité cellulaire endothéliale viable, l'épaisseur de la cornée, et à évaluer l'expression de CD166 et NCAM, ZO-1 (morphologie et jonctions), Na+/K+ATPase (pompe endothéliale) et COX-IV (mitochondries) par immunomarquage.

Résultats

La viabilité cellulaire endothéliale finale était comparable entre les deux groupes pour les deux périodes de conservation, tout comme les marqueurs étudiés par immunofluorescence. Cependant, l'épaisseur cornéenne était plus faible dans le groupe ASM.

Conclusions

En conclusion, l'ASM peut être utilisée pour les DMEK prédécoupées pendant au moins 10 jours, ce qui permet d'organiser des séries de préparation de greffons et leur stockage ultérieur en toute sécurité, tout en simplifiant la logistique.

Pertinence translationnelle

L'ASM peut être utilisé pour préserver des greffons de DMEK prédécoupés et ainsi apporter plus de flexibilité dans l'organisation d'une chirurgie.

Introduction

The number of corneal transplantations (called keratoplasty) is steadily increasing worldwide, and the number of available donor corneas remains limited, even if in theory corneas can be procured from most deceased people within the first 48 to 72 hours. The shortage of donor tissue and the increase in indications for keratoplasty is an important concern in numerous countries. We estimated, 10 years ago, that only 1 cornea was available for the 70 needed worldwide.¹

Diseases of the corneal endothelium – the innermost layer of the cornea – cause permanent corneal edema only curable by keratoplasty. They are frequent and utilize approximately half of the donor corneas in the Western world.^2–4 For these diseases, selective replacement of the deficient endothelium with a healthy one, known as endothelial keratoplasty, has supplanted full-thickness keratoplasty because of its undoubted superiority in terms of restoration of visual acuity (speed and quality), maintenance of eye integrity, and major reduction in the frequency of rejection.^5,6

There are two surgical techniques: Descemet membrane endothelial keratoplasty (DMEK) and Descemet stripping automated endothelial keratoplasty (DSAEK). Each has its advantages and disadvantages, but the proportion of DMEKs, which is the only one to offer genuine anatomic restoration, is constantly increasing.³ Each requires specific graft preparation. In DMEK, the endothelium, composed of a monolayer of cells and their basal membrane (the Descemet membrane [DM]), is manually separated (typically peeled off) from the underlying stroma.⁷ In DSAEK, the endothelium and a thin layer of stroma are mechanically separated (cut) from the rest of the cornea by the blade of a microkeratome. Both preparations are carried out either by the surgeon directly in the operating theatre, or by eye bank technicians who supply predissected or precut grafts, with comparable graft survival in recipients.^8,9 Advantages of eye bank-prepared tissues have nevertheless been perfectly demonstrated: they eliminate the risk of having to cancel surgery at the last minute due to preparation failure, reduce operating room occupancy time, and concentrate the preparation of large numbers of grafts on a small number of skilled personnel.¹⁰

For DMEK, predissected grafts can be sent to the operating room in two forms. The endothelium can be completely separated from the rest of the cornea, trephined to its final size (usually 8 mm in diameter) and stored in a medium until injection into the patient’s eye. Alternatively, the endothelium can be partially peeled off, leaving a small area of adherence to the cornea.¹¹ The whole cornea is then placed back in storage medium. The surgeon completes the peeling process in a few seconds, with no risk of failure, and continues with the procedure. In practice, the shelf life of the predissected grafts seems, however, limited in time to 3 to 4 days.^12–16 Based on clinical results, Parekh et al. recommended to limit the storage of descemetic rolls stored in a cartridge in an organ culture (OC) medium to 4 days.¹⁶ Only one study experimentally assessed endothelial cell (EC) survival of free descemetic rolls stored for 7 days and reported significant damage.¹⁷

We have recently developed an active storage machine (ASM) that replaces the cornea in a more physiological environment by restoring intraocular pressure (IOP) and renewing the storage medium. IOP plays an important role in corneal physiology, contributing to control its hydration. The absence of IOP contributes significantly to corneal swelling during storage, creating endothelial folds that lead to significant EC death. Corneal thickness is conditioned not only by stromal swelling pressure and IOP, but also by endothelial pump function.¹⁸ We showed that the ASM improves EC survival and corneal thickness compared with conventional OC storage in the same nutrient medium, after 1 and 3 months.^19,20 The ASM was first validated using whole corneas. In the present work, we tested the ASM’s ability to also store corneas with predissected endothelium.

Materials and Methods

Study Design

A prospective, single-center, controlled, laboratory study was performed with two parallel groups of predissected DMEK grafts. For each pair of corneas from the same donor, after peeling, one was randomly stored in the ASM and the other in OC. We studied 2 storage times: 3 and 10 days. The main objective of the study was to compare the quality of predissected DMEK grafts stored in ASM versus in conventional passive OC. The primary endpoint was the vital endothelial cell density (vECD) after staining with Hoechst, Ethidium, and Calcein-AM²¹ after 3 and 10 days of storage. Secondary endpoints were corneal thickness measured by optical coherence tomography (OCT) at day 0 (D0; before dissection) and at the end of storage, and the expression of five proteins used as markers of endothelial quality,²² using immunostaining on flat mounted endothelium: ZO-1 for the integrity and morphology of the apical tight-junctions, CD166 and NCAM for the interdigitations of lateral membranes, Na⁺/K⁺ ATPase for the endothelial pump function, and COX IV for the mitochondrial mass.

Human Corneas

They were either corneas retrieved on deceased patients presenting contraindications for donation for therapeutic purpose, at the Saint-Etienne University Hospital, or organ cultured corneas deemed unsuitable for transplantation by the Besançon and Saint-Etienne tissue banks. All corneas were retrieved in accordance with French bioethics laws, after consultation of the National Refusal Register and confirming the absence of opposition to donate by interviewing the next of kin by telephone or in person. All experiments were in accordance with the principles of the Declaration of Helsinki. Corneas were retrieved by in situ corneoscleral excision, in the same conditions as those for therapeutic purposes, using a 17-mm diameter trephine as recommended for transplantation and storage in ASM.¹⁹ The corneas were each immersed in 100 mL of CorneaMax (Eurobio, Les Ulis, France) and transferred to the laboratory or the eye bank within 60 minutes of retrieval.

Graft Preparation: Peeling Technique

Peeling of the endothelium was performed under an operating microscope, in sterile conditions, by the same surgeon (author P. Goin) using the Submerged Corneas Using Backgrounds Aways technique (SCUBA, also known as no-touch technique with a Barron Vacuum Punch support (Katena Products, Denville, NJ). Care was taken to keep the largest possible endothelial area intact, with the aim of minimizing EC migration away from the graft during storage. The DM has been peeled away from the trabeculum and was left attached to the center of the cornea by approximately 1 mm² (Fig. 1). Trypan blue 0.06% (TB; VisionBlue, Dorc, Zuidland, The Netherlands) was used to stain the stromal surface of the peeled DM and verified the extent of peeled area, whereas sterile BSS (Alcon, Rueil Malmaison, France) was used to keep the endothelium submerged throughout the dissection.

Figure 1.

Endothelial dissection. (A) Trypan blue (TB) staining to screen for pre-existing endothelial lesions (area of dead cells) and better visualize the edges of the endothelium. (B) Tearing of the trabeculum with forceps to initiate the endothelial peripheral detachment. (C) Additional 360 degrees peripheral dissection. (D) Endothelium detachment with forceps. (E) Trypan blue staining of the stromal face of the endothelium to visualize the area of detachment. (F) Final state with small area of central attachment which appears unstained.

Graft Storage

Organ Culture

Corneas randomized to the OC group were stored in sealed bottles of CorneaMax placed in a dry incubator at 31°C, as per recommended by the manufacturer. These conditions were the same as those applied in tissue banks for transplantation. CorneaMax was a commercial OC medium containing 2% of fetal calf serum used for decades in Europe for eye banking.

Active Storage Machine

We used the same prototype version of the ASM for both studies performed with whole corneas and previously reported.^19,20,23 Corneas randomized to the ASM group were placed in the center of the bioreactor, held in position by a clamping ring, and the lid was closed (Fig. 2). The cornea thus separated an endothelial chamber and an epithelial chamber. The same OC medium as for the OC group was used. Pushed by a custom-made peristaltic pump, it circulated successively from an infusion bag through the endothelial chamber, a pressure sensor, a hose clamping valve, the epithelial chamber to end in the waste bag. The pump and valve were controlled by a microcontroller, which slaved to the measurement of the differential pressure between the two chambers. A custom-made software was used to control all ASM functions: priming, cruise control (with adjustable pressure and flow), temporary stop with pressure maintained, and draining. In this experiment, the medium flow rate was of 2.6 µL/minute and the differential pressure between both chambers was of 21 millimeters of mercury (mm Hg), close to the physiological values and similar to those used in the ASM validation studies.^19,20 The ASM chamber was placed horizontally to avoid curling of the partially detached endothelium during priming. The ASM was placed in an incubator at 31°C with a 5% CO₂ atmosphere to ensure correct pH control of the bicarbonate-based OC medium, because some ASM tubing was permeable to gas.

Figure 2.

Active storage machine (ASM). (A) Schematic diagram. (B) Front view. The organ culture medium stored in a bag (1) was pushed by a peristaltic pump (2) into the endothelial chamber (3) of the ASM. The pressure in the endothelial chamber was measured continuously by a pressure sensor (4) and regulated precisely by a solenoid valve (5) controlled by a microcontroller. The target pressure was 21 mm Hg, and the flow of medium renewal was 2.6 µL/minute. The medium then enters the epithelial chamber before ending up in a waste bag (6). The ASM was stored in an incubator at 31°C and 5% CO₂.

Initial Endothelial Cell Density Counting Method

The endothelial cell density (ECD) was measured at day 0 (D0), before dissection, using a conventional eye banking method. Briefly, the corneal endothelium was submerged for 4 minutes at room temperature in 0.9% sodium chloride (NaCl) to make intercellular spaces temporarily visible. Five separate images were taken in the central 8 mm with a light-transmission microscope (Olympus, Tokyo, Japan). The calculation of the ECD was performed following a semi-automated counting method with the FIJI software (a distribution of ImageJ available at https://fiji.sc) and the ECD3D plugin that we previously developed.²⁴

Thickness

Corneal thickness was measured with an anterior segment OCT (OCT; CASIA SS-1000, Tomey, Nagoya, Japan) on corneas placed in sterile Petri dishes. To take account of possible posterior folds, we calculated the mean of five measurements performed on the central cross section (center, 1, and 2 mm on each side).

Viable Endothelial Cell Density

The vECD was determined by triple staining with Hoechst 33342 (10 µM in Opti-MEM: minimal essential medium; Gibco), Ethidium homodimer (4 µM) and Calcein-AM (2 µM), as previously described.²¹ Predissected corneas were incubated with a mix of the 3 dyes for 45 minutes in the dark at room temperature (200 µL per graft). Images were taken with an MVX10 fluorescence macroscope (macrozoom microscope; Olympus, Tokyo, Japan). The measurements of the endothelial area covered by living cells (Calcein positive) and the number of cells per square millimeter were performed automatically with manual correction whenever necessary, with the CorneaJ plugin of FIJI. We measured the vECD on the whole area of the dissected endothelium and also on an 8-mm diameter disc corresponding to the most frequent graft size.

Immunofluorescence

Immunofluorescence (IF) was performed on the endothelium left attached to the center of the cornea. The whole cornea was fixated in pure methanol for 30 minutes. The endothelium was then submerged for 30 minutes with the saturation medium consisting of phosphate-buffered saline (PBS), 2% goat serum, and 2% bovine serum albumin. The primary antibodies (Table 1) were diluted 1:200 in the saturation medium and incubated with the endothelium for 1 hour at 37°C. Three 1-minute PBS washes were performed prior to incubation with the secondary antibodies. The secondary antibodies were diluted 1:500 and incubated with the endothelium for 45 minutes at 37°C in the dark. They were Alexa Fluor 488 goat anti-mouse (A32723; Invitrogen, Carlsbad, CA) and Alexa Fluor 555 goat anti-rabbit (A32732; Invitrogen). After 3 more washes with PBS, the endothelium was incubated for 10 minutes in the dark at room temperature with DAPI (D1306; Invitrogen) diluted 1:200 for counterstaining of nuclei. The endothelium was then completely detached from the corneal stroma and flat mounted on a glass slide under an operating microscope (Zeiss) with NB-23-00159-2 mounting medium (NeoBiotech, Nanterre, France). To study EC junctions, 2 markers were used: CD 166 and NCAM. Na+K+ ATPase was used as a membrane marker of human corneal ECs. This protein is involved in endothelial pump function. COX IV was used to study the mitochondrial content of ECs. Images were acquired with the IX81 inverted epifluorescence microscope (Olympus) and processed with CellSens imaging software (Soft Imaging System GmbH, Munster, Germany).

Table 1.

List of Primary Antibodies Used for Endothelial Characterization

Target Protein	Source	Clonality	Company	Reference	Applications/Fixative
CD166	Mouse	Monoclonal	BD Pharmingen, San Jose, CA	559260	IF/methanol
COX IV	Rabbit	Monoclonal	Cell Signaling Technology, Danvers, MA	#4850	IF/methanol
Na+/K+ ATPase	Mouse	Monoclonal	Millipore, Billerica, MA	#05‐369	IF/methanol
NCAM	Mouse	Monoclonal	R&D systems, Minneapolis, MN	MAB24081	IF/methanol
ZO-1	Rabbit	Polyclonal	Invitrogen, Carlsbad, CA	40-2200	IF/methanol

IF, immunofluorescence.

Statistics

All tests were paired and bilateral. The use of pairs of corneas ensured the best comparability as possible. The alpha risk was 5%. The normality of the data distribution was analyzed with the Shapiro-Wilk test and a 5% non-normality threshold. In case of normal distribution, we used a Student's t-test for comparison of means. For variables with a non-normal distribution, we used a nonparametric Wilcoxon test. GraphPad Prism software was used (GraphPad Software, San Diego, CA).

Results

Baseline Donors and Tissue Characteristics

Eight pairs of corneas were stored for 3 days (series 1) and 5 pairs for 10 days (series 2). The baseline characteristics of the donors are detailed in Table 2. The initial ECD before endothelial dissection did not differ between the ASM and the OC groups.

Table 2.

Baseline Characteristics of Donors and Corneas of the Two Series

	Series 1 (3-d Storage), N = 8 Pairs	Series 2 (10-d Storage), N = 5 Pairs
Age, y	69 ± 12 (50 to 88)	78 ± 4 (68 to 88)
Time between death and storage, h	20 ± 10 (6 to 34)	11 ± 5 (6 to 17)
Storage duration before endothelial dissection, d	10 ± 6 (1 to 18)	5 ± 4 days (1 to 12)
ECD immediately before dissection, cells/mm2	2185 ± 232 (1887 to 2600) (ASM)	2680 ± 416 (2114 to 3257) (ASM)
	vs.	vs.
	2276 ± 328 (1851 to 2957) (OC)	2644 ± 420 (2041 to 3121) (OC)
	P = 0.2242	P = 0.7571
Thickness, µm	1135 ± 131 (927 to 1363) (ASM)	1026 ± 194 (732 to 1229) (ASM)
	vs.	vs.
	1142 ± 132 (942 to 1414) (OC)	1043 ± 215 (721 to 1250) (OC)
	P = 0.6416	P = 0.8125

ASM, active storage machine; ECD, endothelial cell density; OC, organ culture.

Data were expressed as mean ± standard deviation (minimum-maximum).

Viable Endothelial Cell Density After Storage

There was no significant difference in vECD between the 2 groups, either on the total surface or on the 8 mm diameter central disc, after either 3 or 10 days of storage (Table 3; Fig. 3). Figure 4 illustrated the endothelial viability of the 5 pairs of corneas stored for 10 days. Notably, the predissected endothelium of all corneas showed very few folds, whereas these are usually observed on a whole cornea (with its undissected endothelium) preserved in OC. The ECDs obtained were compatible with the usual release criteria of many corneal banks for endothelial grafting (over 2200 cells/mm² in the center of the cornea).

Table 3.

Viable Endothelial Cell Density (vECD in Cells/mm²) at the End of Storage of Predissected Corneas

	Series 1 (3-d Storage), N = 8 Pairs	Series 2 (10-d Storage), N = 5 Pairs
vECD in the 8 central mm	2378 ± 501 (1371 to 3174) (ASM)	2482 ± 288 (2122 to 2739) (ASM)
	vs	vs
	2342 ± 503 (1470 to 3232)	2579 ± 315 (2200 to 2937)
	(OC)	(OC)
	P = 0.6238	P = 0.1756
vECD on the total surface	2069 ± 420 (1224 to 2719) (ASM)	2307 ± 259 (2022 to 2635) (ASM)
	vs	vs
	2056 ± 494 (1119 to 2750)	2369 ± 268 (2131 to 2684)
	(OC)	(OC)
	P = 0.8424	P = 0.3125

ASM, active storage machine; OC, organ culture.

Data were expressed as mean ± standard deviation (min-max).

Figure 3.

Comparison of the viable endothelial cell density (vECD) between corneas stored in the active storage machine (ASM) or in conventional passive organ culture (OC). (A) After 3 days (D). (B) After 10 days. The vECD was measured on the whole endothelial area and on the 8-mm diameter central disc, corresponding to the usual graft size used by surgeons worldwide. Data were expressed as mean ± standard deviation (minimum-maximum). Scatter plots represent vECD values. The horizontal line corresponds to the mean and the whiskers to the standard deviation.

Figure 4.

Endothelial viability of the predissected endothelium after 10 days of storage. (A) Staining with calcein-AM, which fluoresces green only in living cells that metabolize it. We measured and expressed as a percentage the viable endothelial area in the central 8 mm (V8) and in the whole endothelium (VT). (B) Image of the other 4 pairs of corneas stored 10 days after dissection. (C) For comparison, calcein-AM staining of another cornea stored for 10 days in organ culture in the same medium but with its endothelium intact, without dissection. Folds were systematic.

Central Corneal Thickness

Before predissection central thicknesses did not differ between the 2 groups for the 2 series at 3 and 10 days. On the other hand, after storage, the thickness was significantly smaller in the ASM group than in the OC group, for both storage times. Overall, for the 2 periods, in the OC group the mean thickness increased by 99 ± 118 µm and decreased by 161 ± 136 µm in the ASM group. For each group, in the OC group, thickness increased by 51 ± 49 µm between D0 and D3 and by 177 ± 158 µm between D0 and D10, whereas in the ASM group, thickness decreased by 219 ± 98 µm in 3 days and by 69 ± 147 µm in 10 days (Fig. 5). The results in terms of corneal thickness were similar for the central stromal attachment and the difference was statistically significant after 3 and 10 days (Fig. 6).

Figure 5.

Central 8 mm corneal thickness at the beginning and the end of storage. (A) After 3 days (D). (B) After 10 days. (C) Representative example of optical coherence tomography cross sections at D0 and D3 showing corneal thinning in the active storage machine (ASM). Data were expressed as mean ± standard deviation (minimum-maximum). Scatter plots represented thickness values. The horizontal line corresponded to the mean and the whiskers to the standard deviation. OC, organ culture.

Figure 6.

Central corneal thickness. (A) After 3 days (D). (B) After 10 days. The corneal thickness difference at the central point was statistically significant after 3 and 10 days. Data were expressed as mean ± standard deviation (minimum-maximum). Scatter plots represented thickness values. The horizontal line corresponded to the mean and the whiskers to the standard deviation. ASM, active storage machine; OC, organ culture.

Expression of Endothelial Markers

Regardless of storage time, we observed no difference in expression patterns for ZO-1, COX-IV, and Na⁺K⁺-ATPase, which were consistent with labeling in healthy endothelium.²² Lateral membrane labeling for CD166 and NCAM was expressed in both groups. Figure 7 illustrated the staining patterns observed in the 3-day group. They were similar in the 10-day group (data not shown).

Figure 7.

Expression of endothelial integrity markers. (A) Flat mount of the whole endothelium. Green labeling of CD166 (lateral junctions) obtained by multiple image alignment (MIA), × 10 and × 64 magnifications. (B) Red labeling of ZO-1 (apical junctions) and Cox IV (mitochondrial mass). Green staining of Na+/K+ ATPase (enzyme pump) and NCAM (lateral junctions) at × 64 magnification. Counterstaining of nuclei was performed with DAPI (D1306; Invitrogen, Carlsbad, CA). ASM, active storage machine; OC, organ culture.

Discussion

In this work, we show that the survival of partially predissected corneal endothelium is excellent both in ASM and on OC for at least 10 days in the CorneaMax medium. The quality of the endothelium remains entirely compatible with endothelial grafting.

Optimizing EC survival during the artificial ex vivo storage phase, between donor death and transplantation into the recipient, is at the heart of eye banking. However, it has been estimated that the decline in ECD during this phase is 0.7% per storage day in conventional (passive) OC,²⁵ whereas in vivo it is only 0.3²⁶ to 0.6%²⁷ per year. Among the many postmortem stresses that trigger this excess mortality, corneal edema plays a central role. It inevitably occurs when IOP collapses and the EC ceases to function. Even after immersion in a storage medium, the edema forms numerous folds in the endothelium, modifying biomechanical parameters and triggering the local death of a large number of ECs.²⁸ There are probably other associated phenomena (loss of survival signals linked to imperfect medium composition) but none has been identified to date.

Eye banking began with whole corneas and continued for almost 70 years, from the first New York eye bank in 1941 to the first stored endothelial grafts in 2007.²⁹ In just a few years, endothelial keratoplasty has become the most frequent corneal graft, and the demand for storage of corneas preprocessed by eye banks has also increased. The survival of ECs on isolated DMEK grafts, consisting only of ECs and their DM, has not been described beyond 7 days without damage.¹⁷ The causes of this additional mortality are not elucidated but might combine the effects of mechanical stress caused by dissection on an endothelium already weakened by storage, and the loss of survival signals transmitted by stromal keratocytes. Spontaneous rolling of the membrane on itself³⁰ is also likely to create a new form of mechanical stress.

We demonstrated that restoring IOP and slow renewal of the storage medium, corresponding to a more physiological environment, significantly increased the survival of ECs in whole grafts. To restore these parameters, we developed an ASM^19,20 for laboratory research and an industrial version of ASM is currently under development for eye banking. Corneas stored in ASM had 23% more viable ECs after 1 month of storage and 53% after 3 months compared with paired corneas stored conventionally (passively) in 100 mL of the same OC medium.

Given the ever-increasing number of DMEKs being performed worldwide, it was important to analyze whether the ASM could also be used to preserve corneas preprocessed by eye bank technicians. The significant changes undergone by the endothelium during dissection implied that the results obtained on whole corneas could not be extrapolated directly. Given the ASM principle (the cornea is used to separate 2 chambers), the only DMEK preparation method that allowed the use of ASM was to leave the endothelium partially attached to the stroma.

Here, we show that endothelial survival does not differ between ASM and OC. Endothelial cellular integrity is also similar. This lack of difference between the two methods is due to improved survival of the endothelium partially detached from the stroma in the OC group compared with survival of the endothelium on the whole cornea.³¹ Indeed, by separating the endothelium from the stroma, the mechanical constraints caused by the folds disappear and survival improves. This confirms the preponderant role of endothelial folds in EC mortality and the main mode of action of ASM and endothelial peeling: the removal of deleterious mechanical constraints.

The survival of ECs on DM partially attached to the stroma appears to be better than that previously described for endothelium completely detached from the cornea and spontaneously forming a roll in the preservation medium.^12–17 This suggests that the presence of nearby stroma positively influences endothelial survival, but only a direct comparison of the two storage methods (partial peeling versus free roll) would answer this question.

We paid particular attention to endothelial dissection, keeping it as peripheral as possible for 3 reasons: to preserve as many ECs as possible, to leave a safety margin so that the surgeon could complete peeling of the central 1 mm² in the operating room without damaging the central 8 mm zone (the forceps can grab the Descemet outside the transplanted central tissue), and to limit the consequences of centrifugal migration of ECs closest to the endothelial tear.³² The migration of ECs could theoretically have two consequences, especially when storage is prolonged as in this work: to reduce the central ECD (the only useful one) by mobilizing the entire endothelial mosaic; to induce the passage of ECs in mesenchymal transition on the other side of the DM and potentially interfere with graft adhesion to the recipient stroma. Note that we did not observe any cells on the stromal side of the DM, but we did not use mesenchymal cell-specific immunostaining. Nor did we observe any cells near the point of attachment of the DM to the stroma. This is another point of concern, because, in theory, stromal keratocytes could migrate onto the DM during prolonged storage after partial dissection, thereby disturbing donor-recipient adhesion.

We placed the ASM in a horizontal position to prevent the endothelium from folding in on itself due to gravity. We observed that the endothelium remained well spread out with no tendency to curl up on itself. Nor did the low flow of medium cause the endothelium to fold in on itself. We opted for air buffering for 72 hours, but our device also allows evaluating different gas buffering times and different dilutions.

The partial reduction of corneal edema in ASM despite the almost complete separation of the endothelium may seem surprising. It should be remembered that the CorneaMax medium we used does not contain Dextran and therefore does not exert deswelling. This is due in part to the restoration of IOP, which tensions the cornea and limits the posterior development of edema. In other experiments using endothelium-free corneas (data not shown), we observed that restoring IOP to endothelium-free corneas reduced thickness by a median value of 70 µm (from 25 to 350 µm, the greater the initial thickness). The reduction observed here appears much greater after 3 days, suggesting that the endothelium detached from the stroma is in contact with the latter and can therefore continue to exert a certain function. After 10 days, this effect seems to diminish, suggesting that the graft is no longer fully applied and that the endothelial cells are not fully exercising their pump function. In fact, when the cornea is placed in ASM, the pre-cut graft under AIR is perfectly applied against the stroma. However, it is possible that the graft may not adhere properly to the stroma under physiological conditions and may become partially detached.

In this work, we used paired corneas to ensure the best possible comparability between the two groups at the start of the experiment. The storage medium is a commercial medium that has been used routinely for many years. We also used an objective method for quantifying vECD^21,33 at the end of storage and a qualitative analysis of the main markers of endothelial structure and function. This study has its limitations, however, because we used a prototype version of ASM that cannot yet be disseminated to other banks. The number of corneas tested is also limited due to the limited availability of tissues suitable for scientific use. We also used corneas that had had a longer or shorter period of conventional OC before being predissected. This is both a strength (as we show that these corneas can withstand the additional stress of dissection) and a weakness (as we cannot exclude better survival with fresh corneas). We did not quantify the expression of the various markers observed by immunofluorescence (unlike what we had done for whole corneas), as there was no sufficiently marked difference between the two groups. Finally, the shelf life was not extended beyond 10 days. An extension of the duration might be necessary, but we initially assumed that the extension from 4 (the maximum duration in the usual practice¹⁶) to 10 days already constituted an important progress by allowing eye banks to organize endothelial graft preparation campaigns on a given day and then have a comfortable time to organize their allocations to recipients in different centers without urgency.

Conclusions

Corneas prepared for DMEK endothelial grafts with the endothelium attached to the center can be stored for 10 days in conventional OC medium without dextran. Separation of the physical bond between endothelium and stroma removes the stress usually induced by swelling of the OC-stored cornea and improves EC survival. Predissected corneas can be preserved in ASM or passive OC with comparable results. ASM can therefore be used to store both whole and predissected corneas. It will enable banks to have just one storage method for all corneas.