Factors Associated with Successful Descemet membrane endothelial keratoplasty in the Diabetes Endothelial Keratoplasty Study.

Abstract

To identify donor, lenticule preparation, recipient, and operative factors associated with 1-year Descemet membrane endothelial keratoplasty (DMEK) success in the Diabetes Endothelial Keratoplasty Study (DEKS). This prospective cohort study was a pre-specified secondary analysis of the DEKS randomized, double-masked, multi-center clinical trial. Individuals undergoing DMEK to treat corneal endothelial dysfunction at 28 clinical sites. Eyes undergoing DMEK were randomized to receive a cornea from a donor without or with diabetes in a 2:1 ratio. Donor, tissue preparation, recipient and operative factors were recorded prospectively. The study excluded eyes with a prior penetrating keratoplasty or glaucoma tube shunt. Graft failures were classified as follows. If the recipient stroma was cloudy in the first postoperative week and did not clear or required a regraft within 8 weeks, it was classified as primary donor failure in the absence of operative complications, or as early failure associated with operative complications. If the recipient stroma cleared in the first postoperative week or cleared after being cloudy in the first postoperative week and later required a regraft or became cloudy without clearing for 90 days, it was classified as late failure. Proportional hazards and logistic regression were used to estimate risk ratios. DMEK success rate at 1 year RESULTS: Of 1421 DMEK grafts, 1374 (97%) were clear at 1 year, 32 were primary donor failures, 12 were early failures, and 3 were late failures. The 78 cases (5.5%) with operative complications had a reduced success rate (88%; hazard ratio for graft failure 4.0, 95% CI 1.9-8.6, p<0.001). Three additional factors were associated with slightly higher graft success rates in multivariate analysis without reaching the defined statistical significance threshold of P<.01: baseline donor endothelial cell density ≥ 2500 vs. <2500 cells/mm² (97% vs. 95% success, P=.03), tissue peeled by surgeon vs. eye bank (99% vs. 96% success, P=.02), and recipient diabetes status: no vs. yes (97% vs. 94% success, P=.02). DMEK was highly successful across a wide range of donor, lenticule preparation, recipient, and surgical technique variations. Not surprisingly, graft success was lower in cases with operative complications.