We recently showed that use of an active storage machine (ASM) that restores 2 important parameters in corneal physiology (intraocular pressure [IOP] and constant renewal of the storage medium) improved the quality of corneal grafts compared to conventional passive organ culture (OC), which dates from the 1980s.1,2 In an ex vivo experiment on 50 pairs of human corneas, we compared ASM and OC for 1 month using the same commercial storage medium. We demonstrated that endothelial cells (ECs) survived significantly better in ASM, with +23% viable cells. In addition, while OC entails a major corneal edema that requires an artificial deswelling step of 1–3 days before surgery, the ASM prevents this edema and keeps corneas always ready for transplantation.2
Given its effectiveness, we wanted to extend storage in ASM beyond 1 month for 2 reasons: to further improve eye banks’ logistics, and to develop novel ex vivo experimental models. Although the systematic storage of all corneas in the very long term is of no interest given the global shortage of corneal grafts3 and banks’ usual throughput, being able to extend storage of some corneas beyond 1 month without tissue damage would give banks high flexibility in at least 3 situations: (1) grafts could be constantly available for emergencies (essentially corneal perforation complicating very severe systemic diseases, and infectious keratitis); (2) conversely, corneas would not perish during periods of lower demand: unlike organ transplants, corneal transplant activity is not constant, being significantly lower in patients’ and surgeons’ vacation periods.4 The 35-day period of OC, combined with the constantly deteriorating cellular quality of grafts over time, results in many grafts being destroyed during storage; (3) exports to countries with severe shortages would be possible without loss of quality or emergency logistics.
Although American eye banks currently use a short-term storage method at 4°C (originally for 5 to 8–10 d), which differs from OC, the recent Cornea Preservation Time Study that analyzed the quality of grafts stored up to 14 days5,6 clearly shows the will to extend the shelf life of corneal transplants without sacrificing quality, to facilitate transplant logistics, and to increase the overall efficiency of the system, each cornea having to find its recipient within a reasonable time, so as to optimize a limited resource.
In this project, we thus extended storage in ASM to 3 months to test the system’s limits.
MATERIALS AND METHODS
We designed a prospective controlled 3-month study, similar to that used for 1-month validation,2 with 2 parallel randomized groups, ASM versus OC, the reference method in Europe, on paired corneas to increase comparability. Although OC is valid for only 5 weeks at most, we kept this control group as there was no other possible comparator and, in addition, 3 transplant series with corneas stored in OC for >1 month had been reported.7-9 The primary endpoint was EC loss, assuming a time-stable decay constant λ per the equation ECDt2 = ECDt1 × eλ(t2–t1). Secondary endpoints were: endothelial cell density (ECD) measured every 3 weeks (to more precisely analyze the 3-mo decline curve); viable ECD (vECD in cells/mm2) determined with a laboratory live/dead assay optimized for corneas,10,11 which is the most reliable evaluation of the pool of living ECs at the end of storage; corneal thickness and transparency12; expression of Na+/K+ ATPase (enzymatic pumps) and Cox IV (mitochondria) in ECs; number of epithelial layers; and rate of microbiological contamination.
All procedures conformed to the tenets of the Declaration of Helsinki for biomedical research involving human subjects. The French Biomedical Agency specifically authorized us to retrieve corneas for this preclinical study (PFS16-010). Corneas were procured by in situ excision, with a 17-mm-diameter trephination, necessary for correct insertion in the ASM. Both corneas were immediately immersed in 100 mL of CorneaMax, a CE-marked OC Dulbecco’s Minimal Essential Medium-based medium containing 2% fetal calf serum, penicillin, and streptomycin (Eurobio, Les Ulis, France) and transferred to the laboratory within 20 minutes. The same batch was used for all corneas throughout the study.
We used the same ASM as in our previous 1-month validation. The cornea was secured tightly to the ASM base, with the scleral rim as a watertight seal to separate the epithelial and endothelial chambers. A sterile disposable storage tank was filled with CorneaMax. A peristaltic pump controlled by a pressure sensor and a microcontroller renewed the medium at a preset rate while creating higher pressure in the endothelial chamber. The medium circulated in the endothelial chamber, then in the epithelial chamber, and was finally directed into a waste tank. The epithelial chamber was at atmospheric pressure. Two transparent optical quality windows, on either side of the cornea, enabled optical controls during storage. The complete system, control panels excepted, was placed in a 31°C dry 5% CO2 incubator.
Per the randomization, 1 cornea of each pair was placed in ASM. We set a limit of 1 hour between procurement and transfer into the ASM to ensure rapid IOP restoration. The medium flow rate was set at 2.6 µL/min (normal aqueous humor flow rate13) and the transcorneal pressure gradient at 21.5 mm Hg (upper limit of normal IOP in humans, chosen for stability in this ASM version) with atmospheric pressure as a reference. The storage tank was refilled 3 times between day 2 and day 86. No deswelling step was necessary at day 86.
After procurement, the cornea randomized in the OC group remained in its sealed bottle and was incubated at 31°C in a dry incubator. The OC group followed the same standard storage process as corneas intended for graft, except that it was prolonged until D86. At D2, after the first quality and safety assessments (EC count, transparency, microbiological testing), the medium was renewed (100 mL) and the cornea suspended using a sterile floating device called a cornea claw (Eurobio). The medium was then renewed every 3 weeks, and the used medium was subjected to microbiological analysis. At D86, quality and safety assessments were repeated, then the cornea was immersed in 60 mL of CorneaJet, a dextran-containing deswelling medium (Eurobio), and incubated for 48 hours. In routine eye banking, this 2-day deswelling time is used for graft shipping to the operating theater.
Endothelial Cell Count Methods
For both groups, ECD was compared twice (day 2 and month 3) with methods based on standard eye-banking techniques. However, each group required a noninterchangeable method for EC visualization: for OC, transmitted light microscopy, and for ASM, specular microscopy (SM). We did not use a transmitted light microscope for ASM as this would have required turning off the system, removing the cornea for specific preparation and observing it in a Petri dish. This extra handling would have been unrepresentative of real-life use of the ASM, which was designed as a closed system, quality assessment being done without deconditioning the cornea. Conversely, SM was not possible for OC, because storage-induced stromal swelling prevented visualizing any ECs. At D2, to ensure the groups were initially comparable, only pairs of corneas with >2000 ECs/mm2 and <10% difference between them continued the experiment. All images were acquired and analyzed by 1 skilled observer (T.G.).
For OC, in order not to induce any additional stress, no additional tissue testing was performed during the 3 months. Indeed, in OC, the counting of ECs requires extracting the cornea from its medium for about 15 minutes to incubate it in the presence of sodium chloride and trypan blue. For the ASM group, on the other hand, SM was possible without extracting the cornea from the system and without specific incubation. As a result, the cell count was repeated every 3 weeks to better analyze cell survival over time. Corneas were defined as suitable for transplantation if ECD was >2000 cells/mm2, and for emergencies if ECD was >1600 cells/mm2. They were named premium if ECD was 2401–2600 cells/mm2 and superpremium if ECD was >2600 cells/mm2.
In addition, a final measurement was performed at the end of storage (D88) in both groups to determine vECD, using triple Hoechst, Ethidium, Calcein-AM staining (HEC) as previously reported.10,11
For transparency, the OC cornea in its sterile Petri dish and the cornea inside the ASM were directly placed horizontally between a backlit chart and a camera as previously described.12
Central corneal thickness was assessed with an anterior segment optical coherence tomography (OCT; CASIA SS-1000, Tomey, Nagoya, Japan). Acquisition was done in epithelial high definition mode (the most accurate of this OCT). The OC cornea was placed in its sterile Petri dish. The ASM’s epithelial chamber was temporarily emptied, as imaging with this OCT required exposure of the ocular surface to air, like in vivo. This measurement was repeated only at M3, to avoid medium consumption by this additional manipulation. In addition, the principle of the ASM and our experience of 1-month storage suggested that thickness varied little during pressurized storage.
Histology, Immunostaining, and Western Blot
Straight after triple HEC staining, corneas were processed for histology (n = 12 pairs), immunostaining (5 pairs), and Western blot (WB) (5 pairs), using standard methods described previously.2 Primary antibodies are listed in Table 1.
For both groups, media sterility was tested by 14-day inoculation in aerobic, anaerobic, and fungal blood culture bottles, which we pioneered and validated in eye banking.14-16 Tests were done at D2 and repeated every 3 weeks.
We calculated the number of donors to include using a standard method for a clinical trial. As no data are available on ECD after 3-month OC, we used the data obtained during our previous study comparing ASM and OC for 1 month, which used identical cell-counting methods. In said study, mean ECD was 2592 cells/mm2 at D2 with a standard deviation (SD) of 349 cells/mm2. Given that cell loss during OC was 0.93% per storage day, we estimated that the ASM would deliver real change if it reduced cell loss by 40% (with mean final ECD thus rising from 1187 to 1622 cells/mm2). As the groups (ASM and OC) were paired, the paired t-test was used. With these hypotheses, a sample size of 22 donor corneas (11 pairs) randomized to 1 of 2 groups provided adequate power (>80%) to demonstrate a difference of 435 cells/mm2 between the groups, using a 2-tailed type I error of 5%.17 As the microbiological contamination rate was almost negligible, a 12-donor sample was planned to achieve an evaluable sample size of 22 paired corneas.
Normality of continuous data distribution was analyzed using the Shapiro–Wilk test, with a nonnormality threshold of 5%. Normally distributed data were described by their mean and SD. Continuous nonnormally distributed variables were summarized as median (10–90 percentiles). The nonparametric Wilcoxon signed-rank test was used when the variable followed a nonnormal distribution, and a t-test when the variable followed a normal distribution. The proportions between the groups were compared by a χ2 test or Fisher exact test. All tests were 2-tailed. OC and ASM results were compared using paired tests, as corneas from the same donor had a major biological link. The null hypothesis was rejected by a type I error <0.05. Statistical analysis was done using SPSS 23.0 (IBM Corp, Armonk, NY).
Baseline Donor Characteristics
Of the 18 pairs of corneas that were enrolled and randomized for storage in ASM versus OC, 6 did not meet the inclusion criteria at D2 and 12 continued the protocol. Donors were 5 females and 7 males, with a mean age of 77 ± 13 years (range 56–93). Time from death to procurement was 15 ± 4 hours (5.30–23.50). Fourteen percent of eyes had undergone cataract surgery (same proportion in both groups).
Endothelial Cell Density Is Higher After Storage in the ASM
No contamination occurred, leaving 12 pairs in statistical analysis. Corneas in both groups had similar ECD at retrieval (day 2), with 2718 ± 295 versus 2779 ± 396 cells/mm2 for ASM and OC, respectively (P = 0.249). In ASM, they had 2331 ± 259 at month 1 and 1897 ± 208 cells/mm2 at month 2 (no parallel counting in OC as explained above). At month 3, ECD was significantly higher in ASM with 1840 ± 216 versus 1479 ± 237 cells/mm2 in OC (P < 0.001). The mean difference was 361 cells/mm2 (95% confidence interval [CI] 241-480), or 26% 95%CI(16-36) more ECs in ASM (mean of the individual differences between paired corneas). We calculated EC loss daily, assuming that the rate of decay (λ) was constant for the 3 months (with ECDt2 = ECDt1 × eλ(t2–t1)). It was 0.46% per storage-day 95%CI(0.44-0.47) for ASM and 0.75% 95%CI(0.74-0.77) for OC, corresponding to a 39% reduction in daily cell loss. When vECD values were compared at M3, the difference was even greater: 1773 ± 250 in ASM versus 1217 ± 299 cells/mm2 in OC (P < 0.001). The mean difference of 556 cells/mm295%CI(356-756) corresponded to 53% 95%CI(28-79) more viable cells in ASM (mean of the individual differences between paired corneas). Results are shown in Figure 1.
In addition, the course of corneal quality status is shown in Figure 2. Notably, 33% of corneas stayed in the standard category in ASM (0% in OC) and 59% in the emergency category (33% in OC) (P = 0.005, Fisher exact test).
Corneas Stored in ASM Are Thinner and More Transparent Than in OC
IOP restoration rapidly reduced corneal thickness by D2 with 679 ± 48 in ASM versus 974±127 μm in OC (P < 0.001). Thereafter, thickness slightly decreased in ASM (630 ± 51 μm, P = 0.001 compared with day 2) unlike OC, where it kept increasing to reach 1114 ± 111 μm, P = 0.018 compared with D2). Notably, cornea thickness range was narrower in ASM than in OC, at all time points. After 48-hour deswelling, OC corneas were significantly thinner than ASM (nondeswelled) corneas (481 ± 74 versus 630 ± 51 μm, P < 0.001).
Expression of Na+/K+ ATPase and of Cox IV in ECs Is Significantly Higher in ASM
Stronger staining of Na+/K+ ATPase and of mitochondria by Cox IV was observed in ECs of corneas stored in ASM than in OC. After protein quantification by WB, expression of Na+/K+ ATPase was 2.9 ± 0.8 (1.7, 3.6) times higher in ASM than in OC (P < 0.001), and expression of Cox IV was 1.5 ± 0.3 (1.2, 1.8) higher in ASM than in OC (P = 0.019) (Figure 3).
Subcellular Structures of ECs Are Better Preserved in ASM Than in OC
For all pairs of corneas studied, the hexagonal organization of the apical surface of ECs, fully outlined by ZO-1 expression, appeared more uniform in ASM than in OC, suggesting less cellular stress in ASM. In addition, ECs of ASM-stored corneas presented more complex interdigitation of their lateral membranes, as revealed by N-CAM and by CD166 expression. These 2 features are characteristic of normal in vivo ECs (Figure 4).
The Epithelium Remains Multilayered and Differentiated in ASM
For all pairs of corneas studied, the labeling of CK3/CK12 and of E-cadherin showed better organization of epithelium in ASM than in OC, with more fully differentiated superficial CK3/CK12+ cells and well-defined E-cadherin expression in plasma membranes of superficial and wing cells (intermediate epithelial layers) (Figure 5). On confocal images, epithelial thickness was 23 ± 4 µm (18, 28) in ASM versus 8 ± 2 µm (4, 10) in OC (P < 0.043), corresponding to 3–6 cell layers in ASM versus 1–3 layers in OC. This was consistent with the hematoxylin eosin saffron-stained cross-sections, which showed an epithelium better preserved in ASM than in OC.
In this study, we show that restoration of IOP and renewal of the nutrient medium, thanks to a new device called ASM, extends corneal storage to 3 months in unprecedented conditions. This duration was chosen to differ clearly from the maximum 5 weeks of OC, while remaining realistic for eye banks.
This study further confirms what we demonstrated for 1-month storage2: compared to passive OC in the same medium (simple immersion), ASM significantly increases EC survival (+53% at 3 mo), prevents constant massive stromal edema in OC and also better preserves epithelial stratification. This benefit is significant for eye banking as the ASM would still deliver one-third of corneas for standard grafts and more than half for emergency surgery, versus 0% and 33%, respectively, in OC.
Despite this important breakthrough, the systematic storage of all corneas in ASM for the very long term is clearly not of interest under these conditions. Only ad integrum preservation would be helpful. Notably, corneal cryopreservation, which is assumed to be the only method allowing proper storage with no time limit,18 has never been transferred routinely: it causes endothelial lesions more severe than those because of conventional storage methods and requires a very specific and costly infrastructure. On the other hand, thanks to the ASM’s new possibilities, being able to prolong the storage of selected corneas up to 3 months could optimize banks’ throughput by: eliminating destruction because of shelf-life; allowing permanent availability, for emergency use, of grafts that keep enough ECs, and facilitating interbank exchanges and exports to countries with severe shortages without significant loss of quality. Another advantage of ASM is that it provides 2 transparent windows, on either side of the cornea, to allow the inspections necessary for graft qualification (EC quantification, transparency and thickness) to be performed at any time and noninvasively. It will thus be easy to determine, through simple periodic checks, the optimal duration for each graft to further minimize losses.
As at 1 month, ECs in ASM, compared to OC, retain at 3 months a higher amount of Na+/K+ ATPase, the main enzyme pumps involved in regulating corneal hydration. While we found no difference in Cox IV expression (reflecting the mitochondrial content of ECs) after 1 month, our results suggest that after 3 months the ECs in OC contain fewer mitochondria than in ASM. The 3 markers of cellular structures analyzed, ZO-1 (apical junctions), NCAM, and CD166 (laterobasal walls),19 confirm what we showed after 1 month: ECs in ASM keep a structure closer to their physiological state than in OC, with more even tight junctions without stress patterns, and a greater complexity of laterobasal walls showing larger exchange surfaces. The clinical importance of this better cellular preservation is unknown but could allow faster recovery of corneal deswelling function with corneas stored in ASM, even in prolonged storage.
Only 1 ex vivo study studied OC for up to 3 months. It reported massive endothelial necrosis after 8 weeks at 31°C in 100 mL of nonrenewed modified minimum essential medium supplemented with 10% fetal calf serum.20 To prevent medium degradation, most banks renew it every 2–3 weeks, as in this study. Only 3 clinical series report keratoplasties performed with corneas stored in OC for a prolonged period but far <3 months: 20 penetrating keratoplasties with corneas of 36 ± 7 days (maximum 48)7; 6 penetrating keratoplasties with corneas of 42–49 days,8 444 lamellar and penetrating keratoplasty with corneas of 38 ± 7 days (maximum 72 d).9 All 3 show results comparable to those of OC series shorter than 4 weeks. The 3 authors stress the importance of being able to prolong OC to facilitate proper graft allocation.
In the present study, significantly, the postdeswelling thickness of corneas in OC was abnormally low (481 ± 74 μm), particularly compared to what we measured after 1 month (608 ± 53 μm, n = 50, P < 0.001, mean difference 127 μm 95%CI(90-164)). Corneal thickness in ASM (630 ± 51 μm, no deswelling step) was also significantly lower than what we measured after 1 month (684 ± 52 μm, n = 50, P = 0.002, mean difference 54 μm 95%CI(20-87).2 These stromal changes, which appear to be less pronounced in ASM than in OC, are probably caused by the loss of keratocytes and glycosaminoglycans in this medium.21
The ASM’s limitations were identified previously2: (1) EC loss remains higher than its physiological level.22,23 We believe the main factor is the suboptimal nutrient medium, but we needed to use it in this comparative test to isolate the gain due only to the ASM, all other things being equal; (2) corneal thickness remains slightly higher than normal.24 Five factors may contribute to this: (a) endothelial pumps limited by a nonoptimal nutrient medium; (b) stromal accumulation of lactate produced by keratocytes and epithelial cells in relative hypoxia25; (c) corneal retraction because of loss of scleral tension; (d) persistence of small endothelial breaches, allowing water pushed by the restored IOP to pass through; (e) the constant and nonphysiological immersion of the epithelial surface in the medium. There is thus still a big opportunity to improve cell viability and stromal thickness in ASM and achieve storage even closer to physiological conditions.
In summary, the ASM allows corneas to be stored for up to 3 months with unparalleled safety and quality, while allowing logistical optimization and flexibility for eye banks. Other perspectives must now be explored, in particular the storage of precut endothelial grafts, the proportion of which is increasing in banks. The 2 major advantages of the ASM (long-term storage and the possibility of repeated noninvasive controls) also make it possible to consider other innovative interventions on corneas such as transfection, cell therapy, and stimulation of endothelial proliferation.
Beyond eye banking, the very prolonged ex vivo survival of human corneas allowed by the ASM opens up new experimental perspectives, especially as the present results suggest that 3 months can be exceeded. The ASM could facilitate preclinical research to measure the long-term effects (on the 3 corneal layers and on the structure and transparency) of new drugs, of advanced therapy medicinal products, medical devices, and physical processes designed to pass through or modify the cornea. This research bench is supported by the “French Food and Drug Administration,” the Agence Nationale de la Sécurité du Médicament et des Produits de Santé.
We are grateful to those who donated their corneas to science and to their families. We thank the hospital team for coordinating organ and tissue procurement (JL. Pugniet, T. Peyragrosse, M Barallon, and F. Rogues) and for their invaluable expertise during family interviews. We thank Yuenu Wu for her technical support with WBs. In addition, we also thank the French Biomedical Agency for its institutional support and authorization.
1. Guindolet D, Crouzet E, He Z, et al. Storage of porcine cornea in an innovative bioreactor. Invest Ophthalmol Vis Sci. 2017; 58135907–5917
2. Garcin T, Gauthier AS, Crouzet E, et al. Innovative corneal active storage machine for long-term eye banking. Am J Transplant. 2019; 1961641–1651
3. Gain P, Jullienne R, He Z, et al. Global survey of corneal transplantation and eye banking. JAMA Ophthalmol. 2016; 1342167–173
4. Agence de la BiomédecineAnnual Report. Review of Year 2018 Activities. 2019Saint-Denis, FranceAgence de la Biomédecine
5. Rosenwasser GO, Szczotka-Flynn LB, Ayala AR, et al.; Cornea Preservation Time Study GroupEffect of cornea preservation time on success of Descemet stripping automated endothelial keratoplasty: a randomized clinical trial. JAMA Ophthalmol. 2017; 135121401–1409
6. Lass JH, Szczotka-Flynn LB, Ayala AR, et al.; Writing Committee for the Cornea Preservation Time Study GroupCornea preservation time study: methods and potential impact on the cornea donor pool in the United States. Cornea. 2015; 346601–608
7. Frueh BE, Böhnke M. Corneal grafting of donor tissue preserved for longer than 4 weeks in organ-culture medium. Cornea. 1995; 145463–466
8. Ehlers H, Ehlers N, Hjortdal JO. Corneal transplantation with donor tissue kept in organ culture for 7 weeks. Acta Ophthalmol Scand. 1999; 773277–278
9. Madzak A, Hjortdal J. Outcome of human donor corneas stored for more than 4 weeks. Cornea. 2018; 37101232–1236
10. Bernard A, Campolmi N, He Z, et al. CorneaJ: an imageJ Plugin for semi-automated measurement of corneal endothelial cell viability. Cornea. 2014; 336604–609
11. Pipparelli A, Thuret G, Toubeau D, et al. Pan-corneal endothelial viability assessment: application to endothelial grafts predissected by eye banks. Invest Ophthalmol Vis Sci. 2011; 5286018–6025
12. Acquart S, Campolmi N, He Z, et al. Non-invasive measurement of transparency, arcus senilis, and scleral rim diameter of corneas during eye banking. Cell Tissue Bank. 2014; 153471–482
13. Brubaker RF. Flow of aqueous humor in humans [the Friedenwald lecture]. Invest Ophthalmol Vis Sci. 1991; 32133145–3166
14. Gain P, Thuret G, Chiquet C, et al. Use of a pair of blood culture bottles for sterility testing of corneal organ culture media. Br J Ophthalmol. 2001; 85101158–1162
15. Thuret G, Carricajo A, Chiquet C, et al. Sensitivity and rapidity of blood culture bottles in the detection of cornea organ culture media contamination by bacteria and fungi. Br J Ophthalmol. 2002; 86121422–1427
16. Thuret G, Carricajo A, Vautrin AC, et al. Efficiency of blood culture bottles for the fungal sterility testing of corneal organ culture media. Br J Ophthalmol. 2005; 895586–590
17. Gauderman WJ, Barlow WE. Sample size calculations for ophthalmologic studies. Arch Ophthalmol. 1992; 1105690–692
18. Armitage J. Cryopreservation for corneal storage. Dev Ophthalmol. 2009; 43:63–69
19. He Z, Forest F, Gain P, et al. 3D map of the human corneal endothelial cell. Sci Rep. 2016; 6:29047
20. Redbrake C, Salla S, Frantz A. Changes in human donor corneas preserved for longer than 4 weeks. Cornea. 1998; 17162–65
21. Møller-Pedersen T, Møller HJ. Viability of human corneal keratocytes during organ culture. Acta Ophthalmol Scand. 1996; 745449–455
22. Møller-Pedersen T. A comparative study of human corneal keratocyte and endothelial cell density during aging. Cornea. 1997; 163333–338
23. Bourne WM, Nelson LR, Hodge DO. Central corneal endothelial cell changes over a ten-year period. Invest Ophthalmol Vis Sci. 1997; 383779–782
24. Aghaian E, Choe JE, Lin S, et al. Central corneal thickness of Caucasians, Chinese, Hispanics, Filipinos, African Americans, and Japanese in a glaucoma clinic. Ophthalmology. 2004; 111122211–2219
25. Li S, Kim E, Bonanno JA. Fluid transport by the cornea endothelium is dependent on buffering lactic acid efflux. Am J Physiol Cell Physiol. 2016; 3111C116–C126