Abstract
Objective:
To evaluate donor, recipient, and operative factors associated with endothelial cell density (ECD) at 1 year after Descemet membrane endothelial keratoplasty (DMEK) in the Diabetes Endothelial Keratoplasty Study (DEKS).
Design:
This prospective cohort study was a pre-defined secondary analysis of the DEKS multicenter, double-masked, randomized clinical trial.
Participants:
Subjects undergoing DMEK without graft failure and with analyzable endothelial images at 1 year.
Methods:
A central, masked image analysis reading center determined ECD from central endothelial images of corneal grafts from donors without or with diabetes preoperatively and 1 year after DMEK. Associations between ECD at 1 year and donor, recipient, and operative factors were assessed using multivariable linear mixed-effect models adjusting for surgeon (random effect) and accounting for correlation between fellow donor and recipient eyes.
Main outcome measure:
Post-DMEK ECD at 1 year.
Results:
Of 1421 eyes undergoing DMEK, 1274 eyes were included in this analysis; 724 (57%) were in women, 1215 (95%) were in white individuals, and 1229 (96%) had Fuchs endothelial corneal dystrophy (FECD). Lower ECD (estimated effect [99% CI]) at 1 year was associated with lower donor preoperative ECD (−392 [−447 to −336] cells/mm2 per 500 cells/mm2 preoperatively), eye bank-reported donor endothelial cell damage (−161 [−281 to −40] cells/mm2) for 10–20% cell damage by trypan blue staining post lenticule preparation vs. <5% damage, operative complications (−308 [−450 to −166] cells/mm2), and recipient indication other than FECD (−191 [−346 to −36] cells/mm2). One-year endothelial cell loss from preoperative was 26% to 30% across preoperative ECD groups (<2500, 2500 to <2750, 2750 to <3000, ≥3000 cells/mm2), was 36% when there was 10–20% endothelial cell damage after lenticule preparation vs. 29% for <5% cell damage, was 28% when the preoperative indication was FECD vs. 40% when it was not, and was 42% when intraoperative complications occurred vs. 27% when they did not.
Conclusions:
Higher endothelial cell loss at 1 year after DMEK is associated with endothelial cell damage following lenticule preparation, recipient indication other than FECD, and intraoperative complications. Estimating donor endothelial cell damage by trypan blue staining was an independent assessment of donor endothelial health that complemented ECD.
Keywords: Descemet membrane endothelial keratoplasty, endothelial cell density, endothelial cell loss, corneal endothelium
Précis
In a pre-defined secondary analysis of the Diabetes Endothelial Keratoplasty Study, modifiable risk factors for higher endothelial cell loss at 1 year after DMEK were intraoperative complications and donor endothelial cell damage following lenticule preparation.
Introduction
Endothelial cell loss (ECL) at 6 months after Descemet stripping automated endothelial keratoplasty (DSAEK) was associated with subsequent late endothelial graft failure in the Cornea Preservation Time Study (CPTS),1 indicating the importance of understanding, and potentially mitigating, risk factors for ECL. Risk factors for ECL after DSAEK2 were different from those after penetrating keratoplasty (PK),3–5 reflecting differences in donor tissue preparation and surgical manipulation of the graft between these procedures. With further variation in how donor tissue is handled for Descemet membrane endothelial keratoplasty (DMEK), understanding the risk factors for ECL after DMEK could help optimize DMEK graft survival.
The Diabetes Endothelial Keratoplasty Study (DEKS) was a large, multicenter, prospective, randomized trial that investigated the effect of diabetes in the donor on DMEK graft survival and ECL.6 The study found that diabetes in the donor did not affect graft survival or ECL at 1 year after DMEK.7,8 The purpose of the present pre-defined secondary analysis of the DEKS was to investigate donor, donor tissue, recipient, and operative factors associated with ECL after DMEK by determining changes in endothelial cell density (ECD) from before to after DMEK. This analysis complements a recent study of risk factors for graft success in the DEKS.9
Methods
The detailed methods for the DEKS and the primary outcome results have been published previously.6–8 Eyes undergoing DMEK were randomized to receive donor tissue from subjects without or with diabetes, as defined by donor hemoglobin A1c level (≥6.5% indicated diabetes was present).6 The DEKS was a National Eye Institute-funded multicenter clinical trial, was approved by the Jaeb Center for Health Research Institutional Review Board, and adhered to the tenets of the Declaration of Helsinki. Written informed consent was obtained from all participants. The study was registered as a clinical trial at ClinicalTrials.gov (NCT05134480).
All recipients had corneal endothelial dysfunction and were enrolled between February 2022 and July 2024. Exclusion criteria included high-risk recipient eyes, e.g. those with PK failure or glaucoma drainage devices.6 Donor corneas had a minimum endothelial cell density (ECD) of 2300 cells/mm2 as determined by eye bank measurements. Multiple donor, donor tissue, recipient, and operative factors were collected prospectively and analyzed with respect to ECD at 1 year after DMEK.
Endothelial Cell Imaging and Analysis
Endothelial cell image analysis followed well established protocols for image acquisition (including image quality and calibration) and analysis, and was overseen by the Corneal Image Analysis Reading Center (CIARC) at Case Western Reserve University and University Hospitals Eye Institute (Cleveland, OH).6,8,10 Eye banks provided 1 to 3 endothelial images for preoperative ECD measurement by the CIARC. At 1 and 12 months after DMEK, clinical sites submitted 3 postoperative endothelial images to the CIARC for analysis. Endothelial images were analyzed by using the “center” method11,12 (Konan Medical, Irvine, CA) to derive endothelial cell morphometry (i.e., coefficient of variation of cell size and % of hexagonal cells) in addition to ECD.8 Because we previously found little change in morphometry that was unrelated to donor diabetes or diabetes severity at 1 year in the DEKS,8 the present analysis was confined to factors associated with ECD.
Endothelial Cell Damage
Endothelial cell damage was assessed after DMEK lenticule preparation had been completed and prior to loading by staining donor corneas with trypan blue 0.06% for 30 seconds followed by rinsing with balanced salt solution.13 Trained eye bank technicians and surgeons preparing their own tissue assessed the percentage of stained areas relative to the intended graft area. Training was required and included strategies for the visual assessment and comparisons between visual and automated assessments of the same images (automated image processing for segmentation and quantification of the stained area). Technicians and surgeons preparing their own tissue were examined to be within 5% accuracy of objective assessment of the same images using image analysis software and to correctly identify grafts with >20% cell damage, which were disqualified for the DEKS.6
Statistical Analysis
Only eyes without graft failure and with at least one analyzable endothelial image preoperatively and at 1 year after DMEK were included for analysis.8 Mean and standard deviation of ECD were tabulated by each factor level at screening, 1 month, and 1 year. Summary statistics for ECL, which was the change in ECD after DMEK compared to donor preoperative ECD (%), were tabulated at 1 month and 1 year, and the association between ECL at 1 month and 1 year was assessed by the Pearson correlation coefficient. Longitudinal linear mixed-effects models were used to evaluate associations between candidate factors and ECD over time. Each model included ECD as the outcome and adjusted for screening ECD, donor diabetes status, and recipient diagnosis as fixed effects, with surgeon, recipient, and donor included as random effects. Statistical significance was defined as α = 0.01. Initially, each factor was evaluated in a separate model. Factors meeting the significance threshold were then included together in a multivariable model with ECD as the outcome, and adjusted P-values and 99% confidence Intervals (CIs) were reported. Continuous covariates were modeled continuously but categorized for presentation in tables to aid interpretation.
Results
Of 1421 eyes undergoing DMEK in the DEKS, 1274 eyes included in this analysis had ≥1 analyzable postoperative endothelial image at either 1 month or 1 year and were without graft failure at 1 year.8 At 1 year, 1103 eyes had analyzable endothelial images and were without graft failure.8 Of the 1274 eyes included for analysis, 724 (57%) were in women, 1215 (95%) were in white individuals, and 1229 (96%) had Fuchs endothelial corneal dystrophy (FECD).
Endothelial cell density before DMEK and at 1 month and 1 year after DMEK is shown for each donor, donor tissue, recipient, and operative factor (Tables 1–4). There was an association between ECL at 1 month and 1 year (r=0.83, P<0.001). Independent risk factors for lower ECD at 1 year in the final multivariable model are shown in Table 5. Lower preoperative donor ECD by 500 cells/mm2 was associated with a lower ECD (estimated effect [99% CI]) at 1 year by 392 [336 to 447] cells/mm2 (P<0.001). Eye bank-reported endothelial cell damage by trypan blue staining after lenticule preparation was associated with lower ECD at 1 year after DMEK (−161 [−281 to −40] cells/mm2 lower when there was 10–20% endothelial cell damage compared to <5% endothelial cell damage, P<0.001). Lower ECD at 1 year was also associated with intraoperative complications (lower by 308 [−450 to −166] cells/mm2, P<0.001) and any preoperative indication other than Fuchs dystrophy (lower by 191 [−346 to −36] cells/mm2, P=0.001). There were no associations between ECD at 1 year and numerous other factors (Tables 1–4), including donor age, donor diabetes status, timing of lenticule preparation, graft diameter, preoperative and postoperative lens status, anterior chamber air/gas volume, complete air/gas fill duration, trainee participation, or whether surgery was performed at a private or academic site.
Table 1.
Donor Factors Affecting ECD at 1 Year
| Screening | 1 Month | 1 Year | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Risk Factor | N | ECD Mean ± SD |
N | ECD Mean ± SD |
% ECL Mean ± SD |
N | ECD Mean ± SD |
% ECL Mean ± SD |
P-Value1 |
| Overall | 1274 | 2674 ± 289 | 1164 | 2039 ± 458 | 23.6 ± 15.2 | 1103 | 1925 ± 497 | 28.2 ± 16.4 | |
| Age2 | 0.83 | ||||||||
| 50 to 64 | 618 | 2682 ± 288 | 567 | 2032 ± 464 | 24.1 ± 15.4 | 534 | 1926 ± 505 | 28.4 ± 16.6 | |
| 65 to 75 | 656 | 2667 ± 290 | 597 | 2046 ± 453 | 23.1 ± 15.1 | 569 | 1924 ± 490 | 28.0 ± 16.3 | |
| Sex | 0.03 | ||||||||
| Female | 529 | 2642 ± 277 | 478 | 1987 ± 460 | 24.7 ± 15.7 | 458 | 1892 ± 505 | 28.6 ± 17.1 | |
| Male | 745 | 2697 ± 295 | 686 | 2076 ± 454 | 22.9 ± 14.8 | 645 | 1948 ± 491 | 27.9 ± 15.9 | |
| Race/ethnicity (N=1068)3 | 0.04 | ||||||||
| Non-Hispanic White | 960 | 2668 ± 290 | 873 | 2003 ± 467 | 24.8 ± 15.6 | 830 | 1886 ± 497 | 29.4 ± 16.5 | |
| Other4 | 108 | 2775 ± 281 | 100 | 2124 ± 416 | 23.2 ± 13.7 | 89 | 2036 ± 471 | 26.7 ± 15.4 | |
| Diabetes status | 0.62 | ||||||||
| Without diabetes | 816 | 2676 ± 290 | 745 | 2045 ± 461 | 23.5 ± 15.0 | 710 | 1927 ± 498 | 28.3 ± 16.1 | |
| With diabetes | 458 | 2671 ± 286 | 419 | 2030 ± 453 | 23.8 ± 15.6 | 393 | 1920 ± 496 | 28.0 ± 17.0 | |
| Pre-Operative CIARC-measured ECD2,5 | <0.001 | ||||||||
| <2500 | 399 | 2355 ± 110 | 368 | 1798 ± 393 | 23.7 ± 16.3 | 340 | 1651 ± 415 | 29.9 ± 17.1 | |
| 2500–<2750 | 389 | 2626 ± 70 | 358 | 2017 ± 389 | 23.1 ± 14.8 | 342 | 1891 ± 427 | 28.0 ± 16.3 | |
| 2750–<3000 | 308 | 2870 ± 72 | 278 | 2180 ± 430 | 24.0 ± 14.9 | 268 | 2114 ± 440 | 26.4 ± 15.2 | |
| ≥3000 | 178 | 3156 ± 135 | 160 | 2400 ± 465 | 23.9 ± 14.3 | 153 | 2277 ± 549 | 28.0 ± 17.0 | |
| Severity of donor folds | 0.51 | ||||||||
| None/mild | 1124 | 2676 ± 290 | 1024 | 2038 ± 461 | 23.7 ± 15.3 | 968 | 1925 ± 502 | 28.2 ± 16.6 | |
| Moderate/severe | 150 | 2664 ± 281 | 140 | 2050 ± 442 | 22.9 ± 14.4 | 135 | 1922 ± 464 | 28.1 ± 15.0 | |
| Cause of death | 0.77 | ||||||||
| Heart disease | 464 | 2706 ± 311 | 426 | 2049 ± 453 | 24.3 ± 14.6 | 401 | 1933 ± 496 | 28.8 ± 15.8 | |
| Cancer | 398 | 2638 ± 275 | 359 | 2038 ± 457 | 22.4 ± 15.7 | 356 | 1925 ± 498 | 27.2 ± 17.1 | |
| Respiratory disease | 143 | 2710 ± 287 | 130 | 2010 ± 466 | 25.8 ± 14.9 | 119 | 1886 ± 512 | 31.1 ± 16.6 | |
| CVA | 102 | 2652 ± 263 | 97 | 1999 ± 478 | 25.0 ± 14.7 | 91 | 1877 ± 521 | 29.6 ± 16.6 | |
| Trauma | 36 | 2650 ± 262 | 31 | 1931 ± 407 | 26.5 ± 16.5 | 30 | 1908 ± 453 | 27.5 ± 15.0 | |
| Other | 131 | 2657 ± 262 | 121 | 2102 ± 465 | 20.7 ± 15.8 | 106 | 1983 ± 477 | 24.9 ± 15.9 | |
| Death to preservation time2 | 0.27 | ||||||||
| <7 hours | 195 | 2639 ± 264 | 181 | 2011 ± 471 | 23.6 ± 15.7 | 162 | 1896 ± 524 | 28.6 ± 18.0 | |
| 7–<10 hours | 329 | 2646 ± 280 | 297 | 2033 ± 473 | 23.2 ± 15.4 | 289 | 1932 ± 484 | 27.1 ± 15.7 | |
| 10–<14 hours | 312 | 2696 ± 290 | 280 | 2034 ± 457 | 24.2 ± 15.8 | 273 | 1908 ± 509 | 29.2 ± 17.1 | |
| ≥14 hours | 438 | 2696 ± 302 | 406 | 2061 ± 443 | 23.5 ± 14.5 | 379 | 1944 ± 487 | 28.1 ± 15.8 | |
| Death to surgery time2 | 0.33 | ||||||||
| <7 days | 693 | 2697 ± 287 | 633 | 2080 ± 470 | 22.8 ± 15.5 | 605 | 1971 ± 501 | 27.2 ± 16.5 | |
| ≥7 days | 581 | 2647 ± 289 | 531 | 1991 ± 439 | 24.6 ± 14.9 | 498 | 1869 ± 487 | 29.4 ± 16.3 | |
P-value is testing for ECD at 1 year vs. screening, not ECL. Model adjusts for pre-operative ECD, donor diabetes status, and recipient diagnosis as fixed effects and surgeon as a random effect, and accounts for the correlations from recipients and donors having both eyes in the study. No adjustment for multiple comparisons has been made. Following the analysis of the CPTS study, the threshold for statistical significance is α=0.01.
Variable was treated as continuous in the model. It has been categorized for display purposes only.
Excludes 206 tissues with either race or ethnicity unknown.
Other races or ethnicities include: Black/African American (n=83), American Indian/Alaskan Native (n=5), Asian (n=4), Native Hawaiian/other Pacific Islander (n=1)—all non-Hispanic, and White Hispanic (n=15).
Pre-operative CIARC–measured ECD used when available. For n=2 tissues, the post-operative CIARC–measured ECD was used. For n=51 tissues, the eye bank–measured cell density was used. CIARC, Cornea Image Analysis Reading Center.
Table 4.
Operative Factors Affecting ECD at 1 Year
| Screening | 1 Month | 1 Year | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Risk Factor | N | ECD Mean ± SD |
N | ECD Mean ± SD |
% ECL Mean ± SD |
N | ECD Mean ± SD |
% ECL Mean ± SD |
P-Value1 |
| Overall | 1274 | 2674 ± 289 | 1164 | 2039 ± 458 | 23.6 ± 15.2 | 1103 | 1925 ± 497 | 28.2 ± 16.4 | |
| Gas used | 0.67 | ||||||||
| Air | 230 | 2644 ± 281 | 213 | 2169 ± 489 | 18.1 ± 16.0 | 199 | 2113 ± 516 | 20.3 ± 16.8 | |
| Gas | 1044 | 2681 ± 290 | 951 | 2011 ± 446 | 24.9 ± 14.8 | 904 | 1883 ± 483 | 29.9 ± 15.8 | |
| % AC filled with gas (N=683)2,3 | 0.95 | ||||||||
| <60% | 49 | 2647 ± 269 | 43 | 2047 ± 490 | 22.2 ± 17.0 | 44 | 1912 ± 510 | 27.8 ± 18.1 | |
| 60% to 75% | 302 | 2641 ± 288 | 275 | 2011 ± 431 | 23.5 ± 14.5 | 267 | 1884 ± 472 | 29.0 ± 15.7 | |
| >75% | 332 | 2697 ± 274 | 296 | 2001 ± 476 | 25.7 ± 16.0 | 267 | 1919 ± 517 | 29.1 ± 17.4 | |
| Complete air/gas fill duration2 | 0.97 | ||||||||
| 0 minutes | 535 | 2676 ± 295 | 491 | 2051 ± 481 | 23.3 ± 15.9 | 435 | 1979 ± 518 | 26.1 ± 17.4 | |
| 1 to 10 minutes | 710 | 2676 ± 283 | 649 | 2037 ± 438 | 23.7 ± 14.7 | 649 | 1896 ± 476 | 29.3 ± 15.5 | |
| 11 to 60 minutes | 29 | 2612 ± 318 | 24 | 1853 ± 491 | 29.1 ± 15.1 | 19 | 1677 ± 593 | 38.4 ± 18.6 | |
| Use of AC maintainer | 0.97 | ||||||||
| No | 1054 | 2680 ± 291 | 956 | 2009 ± 443 | 24.9 ± 14.8 | 921 | 1876 ± 482 | 30.1 ± 15.9 | |
| Yes | 220 | 2645 ± 275 | 208 | 2180 ± 499 | 17.7 ± 15.9 | 182 | 2172 ± 498 | 18.3 ± 15.8 | |
| Was viscoelastic used in AC? | 0.72 | ||||||||
| No | 107 | 2650 ± 293 | 94 | 2178 ± 453 | 17.8 ± 13.7 | 88 | 2146 ± 478 | 19.4 ± 14.8 | |
| Yes | 1167 | 2677 ± 288 | 1070 | 2027 ± 457 | 24.1 ± 15.2 | 1015 | 1906 ± 494 | 28.9 ± 16.3 | |
| Lens surgery | 0.05 | ||||||||
| No | 725 | 2672 ± 299 | 659 | 1994 ± 476 | 25.3 ± 16.0 | 621 | 1877 ± 520 | 29.9 ± 17.3 | |
| Yes | 549 | 2677 ± 276 | 505 | 2098 ± 427 | 21.5 ± 13.9 | 482 | 1986 ± 460 | 25.9 ± 15.0 | |
| Was a resident or fellow involved in preparation of the donor? | 0.36 | ||||||||
| No | 1134 | 2681 ± 285 | 1034 | 2042 ± 441 | 23.6 ± 14.7 | 985 | 1920 ± 488 | 28.5 ± 16.1 | |
| Yes | 140 | 2622 ± 316 | 130 | 2017 ± 578 | 23.6 ± 19.0 | 118 | 1968 ± 569 | 25.4 ± 18.7 | |
| Was a resident or fellow involved in preparation of the recipient? | 0.69 | ||||||||
| No | 1055 | 2673 ± 286 | 958 | 2035 ± 440 | 23.7 ± 14.7 | 917 | 1913 ± 489 | 28.5 ± 16.3 | |
| Yes | 219 | 2682 ± 302 | 206 | 2062 ± 537 | 23.3 ± 17.6 | 186 | 1982 ± 534 | 26.7 ± 17.1 | |
| Was a resident or fellow involved in insertion and positioning of donor? | 0.46 | ||||||||
| No | 1020 | 2669 ± 287 | 931 | 2043 ± 446 | 23.3 ± 14.8 | 888 | 1922 ± 493 | 28.1 ± 16.3 | |
| Yes | 254 | 2696 ± 296 | 233 | 2024 ± 506 | 24.9 ± 16.8 | 215 | 1936 ± 515 | 28.6 ± 17.1 | |
| Insertion method (N=1060)4 | 0.11 | ||||||||
| Straiko modified Jones tube | 769 | 2687 ± 284 | 703 | 2013 ± 445 | 25.0 ± 14.9 | 702 | 1876 ± 479 | 30.2 ± 15.7 | |
| LEITR modified Jones tube | 162 | 2671 ± 328 | 153 | 2008 ± 406 | 24.5 ± 13.3 | 142 | 1870 ± 480 | 29.7 ± 16.3 | |
| Other5 | 129 | 2652 ± 278 | 112 | 1905 ± 552 | 27.7 ± 18.9 | 83 | 1847 ± 577 | 31.6 ± 19.2 | |
| Incision location | 0.08 | ||||||||
| Temporal | 1192 | 2673 ± 289 | 1091 | 2045 ± 457 | 23.4 ± 15.1 | 1033 | 1931 ± 497 | 27.9 ± 16.3 | |
| Other6 | 82 | 2693 ± 281 | 73 | 1950 ± 466 | 26.7 ± 17.1 | 70 | 1836 ± 495 | 31.8 ± 17.7 | |
| Incision site | 0.78 | ||||||||
| Cornea | 1178 | 2677 ± 288 | 1077 | 2024 ± 456 | 24.2 ± 15.2 | 1022 | 1902 ± 494 | 29.1 ± 16.3 | |
| Scleral tunnel | 96 | 2646 ± 295 | 87 | 2225 ± 452 | 16.0 ± 12.9 | 81 | 2215 ± 448 | 16.8 ± 12.8 | |
| Incision size (mm) | 0.05 | ||||||||
| 2.0 ± 0.2 | 158 | 2656 ± 321 | 145 | 2088 ± 475 | 21.4 ± 14.5 | 138 | 1996 ± 534 | 25.0 ± 17.3 | |
| 2.5 ± 0.2 | 668 | 2679 ± 290 | 607 | 1974 ± 467 | 26.2 ± 15.8 | 553 | 1848 ± 490 | 31.1 ± 16.3 | |
| 3.0 or greater | 448 | 2675 ± 275 | 412 | 2119 ± 424 | 20.6 ± 13.9 | 412 | 2004 ± 479 | 25.3 ± 15.5 | |
| Donor tissue unscroll time (N=1219)2,7 | 0.002 | ||||||||
| ≤2 minutes | 641 | 2681 ± 291 | 587 | 2103 ± 426 | 21.4 ± 13.5 | 554 | 1967 ± 486 | 26.7 ± 15.6 | |
| 3 to 5 minutes | 435 | 2660 ± 288 | 397 | 1962 ± 476 | 26.1 ± 16.7 | 372 | 1859 ± 494 | 30.3 ± 17.0 | |
| >5 minutes | 143 | 2680 ± 287 | 134 | 1965 ± 509 | 26.9 ± 16.4 | 125 | 1892 ± 524 | 30.0 ± 16.6 | |
| Graft diameter2 | 0.12 | ||||||||
| 7.0 to <8.0 mm | 733 | 2684 ± 295 | 680 | 2057 ± 421 | 23.2 ± 13.7 | 658 | 1919 ± 469 | 28.6 ± 15.2 | |
| 8.0 to 8.5 mm | 541 | 2661 ± 280 | 484 | 2014 ± 506 | 24.2 ± 17.1 | 445 | 1933 ± 537 | 27.6 ± 18.0 | |
| Use of intracameral lidocaine | 0.46 | ||||||||
| No | 730 | 2690 ± 297 | 660 | 2040 ± 430 | 24.0 ± 14.2 | 658 | 1897 ± 481 | 29.5 ± 15.7 | |
| Yes | 544 | 2654 ± 276 | 504 | 2039 ± 493 | 23.2 ± 16.5 | 445 | 1966 ± 517 | 26.3 ± 17.3 | |
| Use of intracameral antimicrobial | 0.93 | ||||||||
| No | 1232 | 2674 ± 290 | 1125 | 2047 ± 460 | 23.3 ± 15.2 | 1069 | 1931 ± 498 | 28.0 ± 16.4 | |
| Yes | 42 | 2696 ± 265 | 39 | 1821 ± 363 | 32.1 ± 13.0 | 34 | 1731 ± 417 | 35.1 ± 15.1 | |
| Use of intracameral steroid | NA8 | ||||||||
| No | 1273 | 2674 ± 289 | 1163 | 2039 ± 458 | 23.6 ± 15.2 | 1102 | 1924 ± 497 | 28.2 ± 16.4 | |
| Yes | 1 | 3047 ± NA | 1 | 2520 ± NA | 17.3 ± NA | 1 | 2447 ± NA | 19.7 ± NA | |
| Was pupil constricted using miotic in AC? | 0.17 | ||||||||
| No | 289 | 2679 ± 290 | 252 | 2006 ± 469 | 24.9 ± 15.6 | 228 | 1919 ± 517 | 28.4 ± 17.6 | |
| Yes | 985 | 2673 ± 289 | 912 | 2049 ± 455 | 23.3 ± 15.1 | 875 | 1926 ± 492 | 28.1 ± 16.1 | |
| Any operative complications?9 | <0.001 | ||||||||
| No | 1208 | 2677 ± 289 | 1105 | 2056 ± 444 | 23.1 ± 14.6 | 1046 | 1946 ± 482 | 27.4 ± 15.8 | |
| Yes | 66 | 2629 ± 290 | 59 | 1733 ± 595 | 34.1 ± 21.7 | 57 | 1537 ± 603 | 41.9 ± 21.1 | |
P-value is testing for ECD at 1 year vs. screening, not ECL. Model adjusts for pre-operative ECD, donor diabetes status, and recipient diagnosis as fixed effects and surgeon as a random effect, and accounts for the correlations from recipients and donors having both eyes in the study. No adjustment for multiple comparisons has been made. Following the analysis of the CPTS study, the threshold for statistical significance is α=0.01.
Variable was treated as continuous in the model. It has been categorized for display purposes only.
Excludes 591 tissues for which the percentage of the anterior chamber filled with gas was not assessed.
Excludes 214 tissues prepared by the surgeon. All of these tissues were inserted with an IOL insertor.
Other insertion methods include: Gueder tube (n=37), Stephen cannula (n=35), Descepro pull-through (n=25), DORC tube (n=12), Tan DMEK Endoglide (n=6), Gunther Weiss glass cannulas (n=6), 2.4mm micro Jones tube (n=5), Jones tube (n=2), older model Straiko modified Jones tube (n=1).
Other incision locations include: Superotemporal (n=42), Inferotemporal (n=35), Superonasal (n=3), Nasal (n=1), Superior (n=1).
Excludes 55 tissues with unknown scroll time.
P-value was not calculated because n=1 in one of the groups.
Operative complications are: difficult unfolding and positioning without use of positioning hook, difficult unfolding and positioning with use of positioning hook, difficult air/gas fill and retention in positioning, reinsertion of donor after partial extrusion, reinsertion of donor after complete extrusion, inverted donor tissue, significant fibrin in anterior chamber, significant hyphemia, vitreous loss, other write-ins.
Table 5.
Results of Multivariate Model Containing Results from the Previous Tables Significant at α=0.01
| Effect | Adjusted Difference from Reference Group in ECD at 12 Months (99% CI) | P-Value1 |
|---|---|---|
| Pre-Operative CIARC-measured ECD2,3 | <0.001 | |
| <2500 | 0 Reference | |
| 2500–<2750 | 216 (138, 294) | |
| 2750–<3000 | 415 (329, 500) | |
| ≥3000 | 554 (452, 656) | |
| Eye bank–reported % cell damage by trypan blue staining | <0.001 | |
| Little to none (<5%) | 0 Reference | |
| Mild (5–10%) | −73 (−143, −3) | |
| Moderate/severe (10–20%) | −161 (−281, −40) | |
| Donor tissue unscroll time2 | 0.43 | |
| ≤2 minutes | 0 Reference | |
| 3 to 5 minutes | −58 (−133, 17) | |
| >5 minutes | −15 (−131, 101) | |
| Any operative complications | <0.001 | |
| No | 0 Reference | |
| Yes | −308 (−450, −166) | |
| Indication for DMEK | 0.001 | |
| FECD | 0 Reference | |
| Other | −191 (−346, −36) |
The model adjusts for surgeon as a random effect and accounts for the correlations from recipients and donors having both eyes in the study. It was fit using n=1005 eye bank–prepared tissues that were not missing any covariate data. Threshold for statistical significance is α=0.01
Variable was treated as continuous in the model. It has been categorized for display purposes only.
Pre-operative CIARC–measured ECD used when available. For n=2 tissues, the post-operative CIARC–measured ECD was used. For n=41 tissues, the eye bank–measured cell density was used. CIARC, Cornea Image Analysis Reading Center.
Mean ECL at 1 year ranged from 26% to 30% across preoperative categorical ECD groups (Table 1). Mean ECL at 1 year was 36% when there was 10–20% endothelial cell damage after lenticule preparation (Table 2), 29% when there was <5% endothelial cell damage after lenticule preparation (Table 2), 40% when the preoperative indication was not FECD (Table 3), 28% when the preoperative indication was FECD (Table 3), 42% when intraoperative complications occurred (Table 4), and 27% when intraoperative complications did not occur (Table 4).
Table 2.
Donor Lenticule Preparation Parameters Affecting ECD at 1 Year
| Screening | 1 Month | 1 Year | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Risk Factor | N | ECD Mean ± SD |
N | ECD Mean ± SD |
% ECL Mean ± SD |
N | ECD Mean ± SD |
% ECL Mean ± SD |
P-Value1 |
| Overall | 1274 | 2674 ± 289 | 1164 | 2039 ± 458 | 23.6 ± 15.2 | 1103 | 1925 ± 497 | 28.2 ± 16.4 | |
| Prepared by | 0.04 | ||||||||
| Eye bank or combination of eye bank and surgeon | 1060 | 2680 ± 290 | 968 | 2000 ± 453 | 25.2 ± 15.2 | 927 | 1873 ± 488 | 30.3 ± 16.1 | |
| Surgeon | 214 | 2647 ± 280 | 196 | 2234 ± 433 | 15.6 ± 12.7 | 176 | 2200 ± 451 | 17.1 ± 13.6 | |
| Timing of tissue preparation (N=1060)2 | 0.09 | ||||||||
| Day of Surgery | 2 | 2460 ± 149 | 2 | 1629 ± 320 | 34.1 ± 9.0 | 2 | 1648 ± 188 | 33.1 ± 3.6 | |
| 1 day before surgery | 704 | 2680 ± 289 | 644 | 2027 ± 446 | 24.2 ± 14.7 | 614 | 1886 ± 485 | 29.8 ± 15.9 | |
| 2 days before surgery | 339 | 2688 ± 293 | 308 | 1946 ± 467 | 27.5 ± 16.0 | 301 | 1844 ± 499 | 31.5 ± 16.6 | |
| 3 days before surgery | 15 | 2535 ± 263 | 14 | 1992 ± 404 | 20.9 ± 11.8 | 10 | 1966 ± 367 | 24.0 ± 10.2 | |
| Separation method (N=1059)2,3 | 0.21 | ||||||||
| Submerged peeling | 821 | 2680 ± 292 | 754 | 1983 ± 450 | 25.9 ± 15.1 | 716 | 1847 ± 487 | 31.2 ± 16.1 | |
| Non-submerged peeling | 238 | 2682 ± 286 | 214 | 2062 ± 459 | 22.8 ± 15.2 | 210 | 1961 ± 485 | 27.1 ± 15.6 | |
| Original storage solution | 0.04 | ||||||||
| Optisol GS | 996 | 2674 ± 294 | 909 | 2031 ± 468 | 24.0 ± 15.4 | 852 | 1915 ± 506 | 28.5 ± 16.7 | |
| Eusol C | 164 | 2677 ± 272 | 145 | 2082 ± 400 | 21.3 ± 14.1 | 147 | 1995 ± 433 | 25.6 ± 14.4 | |
| Kerasave | 66 | 2738 ± 267 | 64 | 2111 ± 418 | 23.1 ± 13.9 | 61 | 1967 ± 475 | 28.5 ± 14.8 | |
| Life 4°C | 48 | 2591 ± 251 | 46 | 1976 ± 494 | 24.0 ± 16.4 | 43 | 1820 ± 538 | 29.9 ± 18.2 | |
| Change in storage solution (N=1060)2 | 0.61 | ||||||||
| Original | 378 | 2734 ± 286 | 356 | 2029 ± 464 | 25.6 ± 15.4 | 340 | 1917 ± 469 | 30.0 ± 14.6 | |
| Fresh | 682 | 2650 ± 289 | 612 | 1983 ± 446 | 25.0 ± 15.1 | 587 | 1847 ± 497 | 30.4 ± 16.9 | |
| Antifungal added after preparation? (N=682)2,4 | 0.96 | ||||||||
| No | 661 | 2653 ± 287 | 595 | 1990 ± 444 | 24.8 ± 15.0 | 578 | 1853 ± 494 | 30.2 ± 16.8 | |
| Yes | 21 | 2570 ± 328 | 17 | 1748 ± 477 | 33.1 ± 15.3 | 9 | 1463 ± 589 | 44.6 ± 19.8 | |
| Tissue preloaded? (N=1060)2 | 0.23 | ||||||||
| No | 297 | 2692 ± 291 | 276 | 1933 ± 508 | 28.4 ± 17.0 | 275 | 1832 ± 503 | 32.2 ± 16.1 | |
| Yes | 763 | 2676 ± 290 | 692 | 2027 ± 427 | 24.0 ± 14.2 | 652 | 1889 ± 481 | 29.5 ± 16.0 | |
| Eye bank–reported separation difficulty: tears (N=1060)2 | 0.75 | ||||||||
| None | 883 | 2682 ± 288 | 807 | 2005 ± 444 | 25.1 ± 14.8 | 774 | 1874 ± 480 | 30.2 ± 15.9 | |
| Few/moderate | 177 | 2669 ± 302 | 161 | 1976 ± 499 | 25.8 ± 17.2 | 153 | 1867 ± 530 | 30.6 ± 17.2 | |
| Eye bank–reported separation difficulty: adhesions (N=1060)2 | 0.47 | ||||||||
| None | 714 | 2681 ± 299 | 653 | 2000 ± 439 | 25.3 ± 14.5 | 619 | 1867 ± 485 | 30.5 ± 15.9 | |
| Mild | 276 | 2673 ± 263 | 253 | 2020 ± 471 | 24.3 ± 16.2 | 245 | 1888 ± 485 | 29.6 ± 16.5 | |
| Moderate/Severe | 70 | 2697 ± 310 | 62 | 1919 ± 526 | 28.7 ± 17.5 | 63 | 1869 ± 535 | 30.9 ± 17.0 | |
| Eye bank–reported % cell damage by trypan blue staining (N=1060)2 | <0.001 | ||||||||
| Little to none (<5%) | 611 | 2691 ± 290 | 559 | 2043 ± 443 | 23.9 ± 14.7 | 534 | 1911 ± 485 | 29.0 ± 15.9 | |
| Mild (5–10%) | 364 | 2666 ± 291 | 336 | 1964 ± 452 | 26.2 ± 15.2 | 317 | 1846 ± 485 | 31.1 ± 15.9 | |
| Moderate/severe (10–20%) | 85 | 2665 ± 287 | 73 | 1841 ± 495 | 31.3 ± 16.9 | 76 | 1712 ± 492 | 36.1 ± 16.7 | |
| Any eye bank observations? (N=1060)2,5 | 0.02 | ||||||||
| No | 914 | 2679 ± 290 | 840 | 2016 ± 447 | 24.6 ± 14.9 | 794 | 1885 ± 482 | 29.8 ± 15.9 | |
| Yes | 146 | 2685 ± 296 | 128 | 1898 ± 481 | 29.2 ± 16.3 | 133 | 1797 ± 519 | 33.2 ± 17.0 | |
P-value is testing for ECD at 1 year vs. screening, not ECL. Model adjusts for pre-operative ECD, donor diabetes status, and recipient diagnosis as fixed effects and surgeon as a random effect, and accounts for the correlations from recipients and donors having both eyes in the study. No adjustment for multiple comparisons has been made. Following the analysis of the CPTS study, the threshold for statistical significance is α=0.01.
Excludes 214 tissues prepared by the surgeon.
Excludes 1 tissue prepared by hydrodissection.
Excludes 378 tissues that were not transferred into a fresh storage solution.
Eye bank observations are defined as the observation of tears, adhesions, or cell damage at or above the “moderate” level.
Table 3.
Recipient Factors Affecting ECD at 1 Year
| Screening | 1 Month | 1 Year | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Risk Factor | N | ECD Mean ± SD |
N | ECD Mean ± SD |
% ECL Mean ± SD |
N | ECD Mean ± SD |
% ECL Mean ± SD |
P-Value1 |
| Overall | 1274 | 2674 ± 289 | 1164 | 2039 ± 458 | 23.6 ± 15.2 | 1103 | 1925 ± 497 | 28.2 ± 16.4 | |
| Age2 | 0.26 | ||||||||
| 39 to 70 | 672 | 2671 ± 288 | 619 | 2037 ± 467 | 23.7 ± 15.5 | 583 | 1917 ± 505 | 28.4 ± 16.8 | |
| 71 to 90 | 602 | 2678 ± 290 | 545 | 2042 ± 448 | 23.6 ± 14.9 | 520 | 1934 ± 489 | 27.9 ± 16.0 | |
| Sex | 0.83 | ||||||||
| Female | 724 | 2683 ± 299 | 664 | 2045 ± 464 | 23.8 ± 15.3 | 626 | 1917 ± 496 | 28.6 ± 16.3 | |
| Male | 550 | 2663 ± 274 | 500 | 2032 ± 451 | 23.4 ± 15.2 | 477 | 1936 ± 499 | 27.6 ± 16.6 | |
| Race/ethnicity | 0.07 | ||||||||
| Non-Hispanic White | 1215 | 2674 ± 289 | 1109 | 2049 ± 451 | 23.2 ± 14.9 | 1052 | 1932 ± 493 | 27.9 ± 16.2 | |
| Other3 | 59 | 2694 ± 277 | 55 | 1853 ± 556 | 31.3 ± 19.1 | 51 | 1782 ± 558 | 34.1 ± 19.4 | |
| Diabetes4 | 0.35 | ||||||||
| Without diabetes | 982 | 2679 ± 291 | 903 | 2040 ± 468 | 23.7 ± 15.5 | 863 | 1929 ± 502 | 28.2 ± 16.6 | |
| With diabetes | 292 | 2658 ± 280 | 261 | 2037 ± 425 | 23.2 ± 14.1 | 240 | 1910 ± 480 | 28.3 ± 15.8 | |
| Indication for DMEK | <0.001 | ||||||||
| FECD | 1229 | 2671 ± 290 | 1124 | 2044 ± 458 | 23.4 ± 15.1 | 1068 | 1934 ± 492 | 27.8 ± 16.1 | |
| Other5 | 45 | 2760 ± 229 | 40 | 1922 ± 461 | 30.2 ± 16.6 | 35 | 1643 ± 579 | 40.2 ± 20.9 | |
| Pre-operative lens status | 0.17 | ||||||||
| Phakic | 705 | 2667 ± 285 | 643 | 2066 ± 443 | 22.4 ± 14.5 | 615 | 1947 ± 477 | 27.2 ± 15.7 | |
| Pseudophakic | 569 | 2683 ± 294 | 521 | 2007 ± 475 | 25.1 ± 16.0 | 488 | 1897 ± 520 | 29.4 ± 17.2 | |
| Pre-operative/post-operative lens status (N=1229)6 | 0.16 | ||||||||
| Phakic/Phakic | 154 | 2633 ± 315 | 136 | 1947 ± 480 | 26.0 ± 15.9 | 131 | 1807 ± 515 | 31.8 ± 17.5 | |
| Phakic/Pseudophakic | 549 | 2677 ± 276 | 505 | 2098 ± 427 | 21.5 ± 13.9 | 482 | 1986 ± 460 | 25.9 ± 15.0 | |
| Pseudophakic/Pseudophakic | 526 | 2676 ± 297 | 483 | 2014 ± 476 | 24.7 ± 15.9 | 455 | 1915 ± 511 | 28.6 ± 16.6 | |
| Presence of pre-operative focal epithelial edema (N=1229)6 | 0.99 | ||||||||
| No | 992 | 2674 ± 291 | 906 | 2071 ± 454 | 22.5 ± 14.7 | 864 | 1956 ± 489 | 27.0 ± 15.9 | |
| Yes | 237 | 2660 ± 289 | 218 | 1932 ± 458 | 27.1 ± 16.2 | 204 | 1839 ± 492 | 31.0 ± 16.8 | |
| Evidence of corneal abnormality other than primary diagnosis7 | 0.46 | ||||||||
| No | 1113 | 2676 ± 289 | 1017 | 2045 ± 449 | 23.5 ± 14.9 | 955 | 1930 ± 490 | 28.0 ± 16.2 | |
| Yes | 161 | 2666 ± 285 | 147 | 2003 ± 520 | 24.8 ± 17.2 | 148 | 1892 ± 542 | 29.2 ± 18.0 | |
| Baseline IOP2 | 0.89 | ||||||||
| <20 | 1203 | 2675 ± 291 | 1100 | 2045 ± 458 | 23.4 ± 15.2 | 1040 | 1929 ± 500 | 28.0 ± 16.5 | |
| ≥20 | 71 | 2672 ± 246 | 64 | 1951 ± 451 | 26.9 ± 15.6 | 63 | 1857 ± 446 | 30.8 ± 14.9 | |
| Glaucoma medications at enrollment | 0.16 | ||||||||
| No | 1203 | 2672 ± 291 | 1098 | 2042 ± 458 | 23.5 ± 15.2 | 1046 | 1931 ± 494 | 27.9 ± 16.3 | |
| Yes | 71 | 2711 ± 253 | 66 | 1997 ± 462 | 26.4 ± 15.5 | 57 | 1806 ± 542 | 32.6 ± 18.6 | |
| Previous glaucoma surgery | 0.58 | ||||||||
| No | 1246 | 2673 ± 289 | 1136 | 2042 ± 458 | 23.5 ± 15.2 | 1080 | 1928 ± 497 | 28.1 ± 16.3 | |
| Yes | 28 | 2745 ± 283 | 28 | 1932 ± 457 | 29.1 ± 16.7 | 23 | 1761 ± 505 | 34.3 ± 19.3 | |
| Previous glaucoma surgery or medications | 0.32 | ||||||||
| No | 1192 | 2672 ± 291 | 1087 | 2043 ± 458 | 23.4 ± 15.2 | 1036 | 1932 ± 496 | 27.9 ± 16.3 | |
| Yes | 82 | 2708 ± 260 | 77 | 1990 ± 459 | 26.5 ± 15.4 | 67 | 1819 ± 513 | 31.9 ± 17.8 | |
| Current smoker | 0.04 | ||||||||
| No | 1217 | 2669 ± 287 | 1110 | 2037 ± 454 | 23.6 ± 15.1 | 1059 | 1926 ± 492 | 28.0 ± 16.2 | |
| Yes | 57 | 2781 ± 309 | 54 | 2092 ± 549 | 25.1 ± 17.3 | 44 | 1898 ± 607 | 31.5 ± 20.4 | |
P-value is testing for ECD at 1 year vs. screening, not ECL. Model adjusts for pre-operative ECD, donor diabetes status, and recipient diagnosis as fixed effects and surgeon as a random effect, and accounts for the correlations from recipients and donors having both eyes in the study. No adjustment for multiple comparisons has been made. Following the analysis of the CPTS study, the threshold for statistical significance is α=0.01.
Variable was treated as continuous in the model. It has been categorized for display purposes only.
Other races or ethnicities include: Black/African American (n=33), American Indian/Alaskan Native (n=2), Asian (n=2), Native Hawaiian/other Pacific Islander (n=1), more than one race (n=5)—all non-Hispanic, White Hispanic (n=14), and Black/African American Hispanic (n=2).
Recipients with diabetes include those with a history of diabetes at baseline, or HbA1c at baseline ≥6.5.
Other diagnoses: Failed DMEK (n=9), Failed DSAEK (n=16), PBK (n=20).
Excludes 45 tissues without an FECD diagnosis.
Corneal abnormalities include: map-dot-fingerprint, subepithelial scarring, stromal scarring, LASIK, keratoconus, Salzmann nodule, band keratopathy, PBK.
Discussion
The DEKS found that diabetes in the donor did not affect graft survival or ECL at 1 year after DMEK, and that ECL from preoperative was approximately 28% for all eyes with analyzable endothelial images.7,8 In this pre-defined secondary DEKS analysis, ECD at 1 year after DMEK was associated with preoperative donor ECD, endothelial cell damage after lenticule preparation, recipient diagnosis, and operative complications.
Donor Factors
Lower donor preoperative ECD was associated with lower ECD at 1 year after DMEK, similar to after DSAEK in the CPTS,2 though there was little variation in ECL from preoperative at 1 year (26–30%) across categorical groups of preoperative ECD (<2500, 2500 to <2750, 2750 to <3000, ≥3000 cells/mm2; Table 1). In the parallel DEKS secondary analysis examining risk factors for graft success, preoperative donor ECD was not associated with graft survival at 1 year.9 The Cornea Donor Study (CDS) and the CPTS found that graft survival after PK and DSAEK, respectively, was not associated with preoperative ECD but was associated with postoperative ECD at 6 months,1,14 indicating that early ECL is more important than preoperative ECD (when >2300 cells/mm2). In fact, the highest ECL after endothelial keratoplasty, including in the DEKS, occurs by 1 month after surgery,8,15 and this is probably explained by endothelial cell damage during donor lenticule preparation and surgical manipulation.
The DEKS also assessed the viability of the donor endothelium at the end of lenticule preparation by visually assessing the percentage of endothelial cell damage using trypan blue staining.6,13 In contrast to specular or confocal microscopy images that typically represent <0.5% of the total endothelial surface area, staining donor corneas with trypan blue assesses the entire donor endothelial area. Donor tissue with 10–20% endothelial cell damage was associated with higher ECL at 1 year after DMEK (36%) compared to donor tissue with <5% damage (29%). Donor tissues with >20% endothelial cell damage by this method were excluded from the DEKS. Because donor preoperative ECD and donor cell damage by trypan blue staining were independent risk factors for lower ECD at 1 year after DMEK in the multivariable model (Table 5), the DEKS has shown that estimating endothelial cell damage by trypan blue staining is an important complementary assessment to measuring donor ECD. Cutoff criteria for both parameters are therefore important when selecting suitable donor tissue for keratoplasty, and data for both should also be provided to surgeons when accepting donor tissue. Automating the analysis of trypan blue staining of donor lenticules16 would help promote standardization and efficiency for eye banks. In fact, the DEKS Area of Cell Damage (ACD) ancillary study will investigate objective and automated methods of quantifying cell damage in a subset of eyes enrolled in the DEKS (personal communication with Dr. J, Hou, Gift of Sight, Minneapolis, MN, on July 11, 2026).16
Donor factors not associated with higher ECL at 1 year after DMEK included donor age, donor sex, cause of death, death to preservation time, and death to surgery time. Donor age was associated with higher ECL at 5 and 10 years after PK in the CDS3,4 but not at 3 years after DSAEK in the CPTS.2 While female donors were associated with lower ECL at 5 years after PK,4 the same relationship was not present after endothelial keratoplasty in the DEKS or the CPTS.2 Donor tissue preparation factors not associated with higher ECL included the timing of lenticule preparation relative to surgery,17 separation method or difficulty, storage solution,18 and whether the tissue was preloaded or not. There was a trend (Table 2) towards higher ECL when the eye bank made tissue observations (of tears, adhesions, endothelial cell damage, etc.), but this became non-significant in the final statistical model. Of note, while ECL appeared to be lower for surgeon-prepared tissue (17%) vs. eye bank and combined eye bank/surgeon preparation (30%), the difference was not statistically significant (statistical models were adjusted for surgeon effect because outcomes from the same surgeon were correlated, i.e. they were not independent data; only 3 surgeons at 2 sites prepared their own tissue).
Recipient Factors
A recipient diagnosis other than FECD was associated with higher ECL (40% vs. 28%) at 1 year after DMEK. While only 45 eyes (<4%) in this analysis (and the overall DEKS cohort) had a recipient diagnosis of pseudophakic corneal edema or previous failed endothelial keratoplasty, the findings are consistent with other series after DMEK19–21 and with the CPTS, which found higher ECL at 3 years after DSAEK for eyes with pseudophakic/aphakic corneal edema compared to FECD.2 Lower ECL in the setting of FECD might be explained by the healthy peripheral endothelial cell reserve in FECD, which migrate centrally after Descemet stripping only procedures resulting in resolution of corneal edema.22,23 Healthier peripheral endothelial cells in FECD might help to slow ECL after DMEK compared to indications other than FECD in which peripheral endothelial cells are depleted or dysfunctional. Of note, while operative complications were also more frequent in eyes without FECD (13%) compared to eyes with FECD (5%, P=0.007), recipient diagnosis and operative complications were independent risk factors for ECL in the final statistical model. Although the parallel DEKS analysis did not find an association between recipient diagnosis and graft survival at 1 year after DMEK,9 longer follow-up is required study to better assess this relationship. Indeed, with longer follow-up in the CPTS and CDS, recipient diagnosis was associated with graft failure after DSAEK and PK, respectively.24,25
Recipient factors not associated with higher ECL included preoperative and postoperative lens status, baseline intraocular pressure, use of glaucoma medications, or previous glaucoma surgery. Only 28 eyes had had previous glaucoma surgery, all of which were either laser trabeculoplasty or minimally invasive glaucoma surgery (MIGS), and the DEKS excluded eyes with a glaucoma drainage device, which has been associated with higher ECL and graft failure after endothelial keratoplasty.26–28
Operative factors
The occurrence of an intraoperative complication, as reported by the surgeon, was the only operative risk factor associated with higher ECL at 1 year after DMEK, and was the only risk factor for graft failure in the parallel 1-year graft success study.9 Operative complications included complications from concomitant surgery (e.g. posterior capsule rupture or vitreous loss with combined cataract surgery) and donor tissue manipulation (including difficult positioning, extrusion, and inversion). In other series, higher ECL has been associated with difficulty unfolding the DMEK graft29,30 and with longer operative time.31 In the DEKS, when intraoperative complications were removed from the final multivariable model, the time taken to unscroll the DMEK graft (defined as from the start of tissue unscrolling to injection of the air/gas tamponade) became a significant risk factor for higher ECL, indicating its strong correlation with intraoperative complications. In the CPTS, operative complications were also associated with higher ECL and graft failure after DSAEK,2,24 and were an independent risk factor (in addition to ECL) for late endothelial graft failure.1 The findings reflect the additional donor tissue surgical manipulation associated with endothelial keratoplasty compared to PK. Because the DEKS excluded high-risk eyes (e.g. with previous failed PK or glaucoma drainage devices) that may be associated with increased difficulty unfolding the DMEK graft, higher ECL is expected in these eyes compared to that found in the DEKS.27
Operative factors not associated with higher ECL included the percentage of anterior chamber air/gas fill, complete air/gas fill duration, combined lens surgery, insertion method and location, incision site and size, graft diameter, trainee participation in donor or recipient preparation, and trainee participation in graft insertion and positioning. Although ECL appeared lower when using air (20%) compared to gas (30%), the difference was not significant after adjusting for surgeon effect (79% of the 230 cases that used air were by 2 surgeons at 1 clinical site who also prepared their own tissue), consistent with a meta-analysis that found no difference in ECL between air and 20% sulfur hexafluoride.32
Strengths and Limitations
This analysis has several strengths, including the large sample size with prospective follow-up, and rigorous, standardized endothelial image analysis methods. Associated donor and recipient data were collected in a standardized manner enabling assessment of multiple candidate factors associated with ECL. With only three failed grafts at 1 year that were not primary or early failures,7 the potential for bias in the endothelial cell analysis due to attrition was minimal. Similarly, while unanalyzable postoperative images might have been associated with lower postoperative ECD, any bias introduced would have been minimal with only 60 eyes and 12 eyes having unanalyzable images at 1 month and 12 months, respectively. Although the visual assessment of endothelial cell damage from trypan blue staining was subjective, all observers were trained and examined on their interpretations to promote standardization and ensure accuracy within 5%. While subjective, this assessment, which is currently used by eye banks in clinical settings to determine tissue suitability, was shown to be important and warrants the development of automated analyses to eliminate subjectivity.16 The results of this study are applicable to most DMEK cases that use intermediate-term cold storage of donor tissue, but because the DEKS excluded higher-risk eyes, including those with failed PK, glaucoma drainage devices, and uveitis, the results will not apply to more complex cases of DMEK. Finally, the current analysis did not investigate postoperative factors, such as the need for secondary procedures for graft dislocation, that may also affect ECL (postoperative events will be reported separately).33,34
Conclusions
The modifiable risk factors for higher ECL at 1 year after DMEK were intraoperative complications and donor endothelial cell damage after lenticule preparation. Operative complications were also a risk factor for graft failure at 1 year after DMEK,9 indicating the importance of minimizing them for improved longer-term graft survival. Comparing longer-term ECL in the setting of operative complications between DMEK and DSAEK will be interesting as this might help guide case selection for each procedure. Such ECL data at 3 and 5 years will become available in the National Eye Institute-funded extension to the DEKS, which will include a subset of the DEKS cohort.35 Donor endothelial cell damage after lenticule preparation emerged as an important risk factor for higher ECL in the DEKS, and it presents an opportunity for eye banks to benchmark their lenticule preparation outcomes in the context of clinical outcomes. While donor endothelial cell damage did not emerge as a risk factor for graft failure at 1 year,9 extended follow-up in the DEKS will determine its association with longer-term graft failure.35
Funding/Support:
This work was supported by cooperative agreements with the National Eye Institute, National Institutes of Health, Department of Health and Human Services (EY030030 and EY030039). Sponsors: Eye Bank Association of America, The Cornea Society, Cleveland Eye Bank Foundation, Beauty of Sight (formerly Florida Lions Eye Bank), CorneaGen, Eversight, Eye Bank for Sight Restoration, Eye Bank of Kentucky, Gift of Sight, Iowa Lions Eye Bank, Kenneth Maverick MD, Lions Eye Bank of Wisconsin, San Antonio Eye Bank, San Diego Eye Bank, Utah Lions Eye Bank, VisionFirst, and VisionGift.
Abbreviations:
- CDS
Cornea Donor Study
- CI
confidence interval
- CPTS
Cornea Preservation Time Study
- DEKS
Diabetes Endothelial Keratoplasty Study
- DMEK
Descemet membrane endothelial keratoplasty
- DSAEK
Descemet stripping automated endothelial keratoplasty
- ECD
endothelial cell density
- ECL
endothelial cell loss
- FECD
Fuchs endothelial cornea dystrophy
- PK
penetrating keratoplasty
Diabetes Endothelial Keratoplasty Study Group
| Francis W. | Price Jr. | MD | Price Vision Group | Indianapolis, IN |
| Anjulie | Gang | MD | Price Vision Group | Indianapolis, IN |
| David | Verdier | MD | Verdier Eye Center | Grand Rapids, MI |
| Keith H. | Baratz | MD | Mayo Clinic | Rochester, MN |
| Shahzad I. | Mian | MD | University of Michigan | Ann Arbor, MI |
| Christopher | Hood | MD | University of Michigan | Ann Arbor, MI |
| Nambi | Nallasamy | MD | University of Michigan | Ann Arbor, MI |
| Angela | Verkade | MD | University of Michigan | Ann Arbor, MI |
| Sudeep | Pramanik | MD | Mid-Atlantic Cornea Consultants | Baltimore, MD |
| Mark A. | Greiner | MD | University of Iowa Hospials and Clinics | Iowa City, IA |
| Christopher | Sales | MD | University of Iowa Hospials and Clinics | Iowa City, IA |
| Lawrence | Tenkman | MD | Bennett & Bloom Eye Centers | Louisville, KY |
| Christopher | Ketcherside | MD | Kansas City Eye Clinic | Overland Park, KS |
| Evan | Schoenberg | MD | Georgia Eye Partners | Atlanta, GA |
| James | Huffman | MD | Huffman and Huffman PSC | Lexington, KY |
| Paul | Phillips | MD | Sightline Ophthalmic Associates | Pittsburgh, PA |
| Divya | Srikumaran | MD | The Wilmer Eye Institute/Johns Hopkins | Baltimore, MD |
| Yassine | Daoud | MD | The Wilmer Eye Institute/Johns Hopkins | Baltimore, MD |
| Allen | Eghrari | MD | The Wilmer Eye Institute/Johns Hopkins | Baltimore, MD |
| Albert | Jun | MD PhD | The Wilmer Eye Institute/Johns Hopkins | Baltimore, MD |
| Yong | Kam | MD | Moyes Eye Center | Kansas City, MO |
| Mark | Terry | MD | Legacy Devers Eye Institute | Portland, OR |
| Jessica | Chen | MD | Legacy Devers Eye Institute | Portland, OR |
| Michael | Straiko | MD | Legacy Devers Eye Institute | Portland, OR |
| Peter | Veldman | MD | University of Chicago | Chicago, IL |
| Parag | Majmudar | MD | Chicago Cornea Consultants | Chicago, IL |
| Rachel | Epstein | MD | Chicago Cornea Consultants | Chicago, IL |
| Neal | Vaidya | MD | Chicago Cornea Consultants | Chicago, IL |
| Ula | Jurkunas | MD | Massachusetts Eye and Ear | Boston, MA |
| Nandini | Venkateswaran | MD | Massachusetts Eye and Ear | Boston, MA |
| Ari | Brandsdorfer | MD | Consulting Ophthalmologists PC | Glastonbury, CT |
| David | Hardten | MD | Minnesota Eye Consultants | Bloomington, MN |
| Sherman | Reeves | MD | Minnesota Eye Consultants | Bloomington, MN |
| Ellen | Koo | MD | Bascom Palmer Eye Institute | Miami, FL |
| Michael | Greenwood | MD | Vance Thomson Vision | West Fargo, ND |
| Jamie | Alexander | MD | Cornea Associates of Texas | Dallas, TX |
| John | Bokosky | MD | Eye Care of San Diego | San Diego, CA |
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Trial Registration: NCT05134480, ClinicalTrials.gov
Meeting Presentation: The material within this manuscript has been submitted for presentation at the American Academy of Ophthalmology Annual Meeting, 2026.
Conflict of Interest: None of the authors have relevant conflicting relationships.
References
- 1.Patel SV, Lass JH, Benetz BA, et al. Postoperative endothelial cell density is associated with late endothelial graft failure after Descemet stripping automated endothelial keratoplasty. Ophthalmology 2019;126(8):1076–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Lass JH, Benetz BA, Patel SV, et al. Donor, recipient, and operative factors associated with increased endothelial cell loss in the Cornea Preservation Time Study. JAMA Ophthalmol 2019;137(2):185–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lass JH, Benetz BA, Gal RL, et al. Donor age and factors related to endothelial cell loss 10 years after penetrating keratoplasty: Specular Microscopy Ancillary Study. Ophthalmology 2013;120(12):2428–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lass JH, Beck RW, Benetz BA, et al. Baseline factors related to endothelial cell loss following penetrating keratoplasty. Arch Ophthalmol 2011;129(9):1149–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Cornea Donor Study Investigator Group, Lass JH, Gal RL, et al. Donor age and corneal endothelial cell loss 5 years after successful corneal transplantation. Specular Microscopy Ancillary Study results. Ophthalmology 2008;115(4):627–32.e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Price MO, Szczotka-Flynn LB, Bauza CE, et al. Diabetes Endothelial Keratoplasty Study: Methods and impact on the use of corneas from donors with diabetes for Descemet membrane endothelial keratoplasty. Cornea 2026;45(1):312–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Price FW Jr., Szczotka-Flynn LB, Price MO, et al. Donor diabetes and 1-year Descemet membrane endothelial keratoplasty success rate: A randomized clinical trial. JAMA Ophthalmol 2025;143(12):1043–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lass JH, Benetz BA, Verdier DD, et al. Endothelial cell loss 1 year after successful DMEK in the Diabetes Endothelial Keratoplasty Study: A randomized clinical trial. JAMA Ophthalmol 2025;143(12):1053–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Price MO, Szczotka-Flynn LB, Reed ZW, et al. Factors associated with successful Descemet membrane endothelial keratoplasty in the Diabetes Endothelial Keratoplasty Study. Ophthalmology 2026;In Preparation. [DOI] [PubMed] [Google Scholar]
- 10.Benetz BA, Gal RL, Ruedy KJ, et al. Specular Microscopy Ancillary Study methods for donor endothelial cell density determination of Cornea Donor Study images. Current Eye Research 2006;31(4):319–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Patel SV, McLaren JW, Bachman LA, Bourne WM. Comparison of flex-center, center, and corner methods of corneal endothelial cell analysis. Cornea 2010;29(9):1042–7. [DOI] [PubMed] [Google Scholar]
- 12.Ohno K, Nelson LR, McLaren JW, et al. Comparison of recording systems and analysis methods in specular microscopy. Cornea 1999;18(4):416–23. [DOI] [PubMed] [Google Scholar]
- 13.Szkarlat MR, Hicks N, Titus MS, Sawant OB. Corneal endothelium viability assay using trypan blue dye after preloaded Descemet membrane endothelial keratoplasty graft preparation. Cornea 2024;43(6):771–6. [DOI] [PubMed] [Google Scholar]
- 14.Lass JH, Sugar A, Benetz BA, et al. Endothelial cell density to predict endothelial graft failure after penetrating keratoplasty. Arch Ophthalmol 2010;128(1):63–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wacker K, Baratz KH, Maguire LJ, et al. Descemet stripping endothelial keratoplasty for Fuchs’ endothelial corneal dystrophy: Five-year results of a prospective study. Ophthalmology 2016;123(1):154–60. [DOI] [PubMed] [Google Scholar]
- 16.Bedard P, Justin JJ, Hou JH. Trypan-assisted automated endothelial cell loss measurements compared with specular microscopy. Cornea 2021;40(8):1031–5. [DOI] [PubMed] [Google Scholar]
- 17.Feng MT, Burkhart ZN, Price FW Jr., Price MO. Effect of donor preparation-to-use times on Descemet membrane endothelial keratoplasty outcomes. Cornea 2013;32(8):1080–2. [DOI] [PubMed] [Google Scholar]
- 18.Price MO, Knight OJ, Benetz BA, et al. Randomized, prospective, single-masked clinical trial of endothelial keratoplasty performance with 2 donor cornea 4°C storage solutions and associated chambers. Cornea 2015;34(3):253–6. [DOI] [PubMed] [Google Scholar]
- 19.Peraza-Nieves J, Baydoun L, Dapena I, et al. Two-year clinical outcome of 500 consecutive cases undergoing Descemet membrane endothelial keratoplasty. Cornea 2017;36(6):655–60. [DOI] [PubMed] [Google Scholar]
- 20.Birbal RS, Baydoun L, Ham L, et al. Effect of surgical indication and preoperative lens status on Descemet membrane endothelial keratoplasty outcomes. Am J Ophthalmol 2020;212:79–87. [DOI] [PubMed] [Google Scholar]
- 21.Baydoun L, Ham L, Borderie V, et al. Endothelial survival after Descemet membrane endothelial keratoplasty: Effect of surgical indication and graft adherence status. JAMA Ophthalmol 2015;133(11):1277–85. [DOI] [PubMed] [Google Scholar]
- 22.Moloney G, Petsoglou C, Ball M, et al. Descemetorhexis without grafting for Fuchs endothelial dystrophy-supplementation with topical ripasudil. Cornea 2017;36(6):642–8. [DOI] [PubMed] [Google Scholar]
- 23.Borkar DS, Veldman P, Colby KA. Treatment of Fuchs endothelial dystrophy by Descemet stripping without endothelial keratoplasty. Cornea 2016;35(10):1267–73. [DOI] [PubMed] [Google Scholar]
- 24.Terry MA, Aldave AJ, Szczotka-Flynn LB, et al. Donor, recipient, and operative factors associated with graft success in the Cornea Preservation Time Study. Ophthalmology 2018;125(11):1700–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Sugar A, Gal RL, Kollman C, et al. Factors associated with corneal graft survival in the Cornea Donor Study. JAMA ophthalmology 2015;133(3):246–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Birbal RS, Tong CM, Dapena I, et al. Clinical outcomes of Descemet membrane endothelial keratoplasty in eyes with a glaucoma drainage device. Am J Ophthalmol 2019;199:150–8. [DOI] [PubMed] [Google Scholar]
- 27.Sorkin N, Mimouni M, Kisilevsky E, et al. Four-year survival of Descemet membrane endothelial keratoplasty in patients with previous glaucoma surgery. Am J Ophthalmol 2020;218:7–16. [DOI] [PubMed] [Google Scholar]
- 28.Patel SV, Hodge DO, Winkler NS, et al. Graft survival, graft rejection, and glaucoma in a consecutive series of Descemet stripping endothelial keratoplasty. Cornea 2023;42(5):549–56. [DOI] [PubMed] [Google Scholar]
- 29.Maier AK, Gundlach E, Schroeter J, et al. Influence of the difficulty of graft unfolding and attachment on the outcome in Descemet membrane endothelial keratoplasty. Graefes Arch Clin Exp Ophthalmol 2015;253(6):895–900. [DOI] [PubMed] [Google Scholar]
- 30.Gundlach E, Spiller N, Pilger D, et al. Impact of difficult unfolding and attachment of the graft lamella on the long-term outcome after Descemet membrane endothelial keratoplasty. Graefes Arch Clin Exp Ophthalmol 2020;258(11):2459–65. [DOI] [PubMed] [Google Scholar]
- 31.Leterrier J, Mastrangelo L, Goetz C, et al. Preoperative and perioperative factors that predict endothelial cell loss 1 year after uncomplicated Descemet membrane endothelial keratoplasty. PLoS One 2025;20(12):e0339346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Marques RE, Guerra PS, Sousa DC, et al. Sulfur hexafluoride 20% versus air 100% for anterior chamber tamponade in DMEK: A meta-analysis. Cornea 2018;37(6):691–7. [DOI] [PubMed] [Google Scholar]
- 33.Dunker S, Winkens B, van den Biggelaar F, et al. Rebubbling and graft failure in Descemet membrane endothelial keratoplasty: A prospective Dutch registry study. Br J Ophthalmol 2023;107(1):17–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Tapley JL, Hill JR, Bauer AJ, et al. Rate of endothelial cell loss and graft survival in Descemet membrane endothelial keratoplasty in eyes requiring a rebubble. Cornea 2023;42(8):934–9. [DOI] [PubMed] [Google Scholar]
- 35.Lass JH. Effect of donor diabetes and other factors on corneal transplant endothelial cell loss and success at 5 years (DEKS). 2026, https://clinicaltrials.gov/study/NCT07217249
