Modeling Endothelial Cell Loss After Descemet Stripping Endothelial Keratoplasty: Data From 5 Years of Follow-up

Purpose: To report 5-year outcomes for graft survival and endothelial cell survival after Descemet stripping endothelial keratoplasty (DSEK) including regression modeling for cell survival over time. Methods: This is a single-institution, retrospective, consecutive interventional series of 210 primary DSEK grafts operated for low-to-moderate risk indications, specifically Fuchs dystrophy and bullous keratopathy. Primary outcomes were cumulative graft survival and % endothelial cell loss from 3 months through 5 years; the secondary outcome was to trend endothelial cell density over time by least-squares and mixed nonlinear modeling. Results: Cumulative graft survival was high at 99%, 98%, and 94% at 1, 3, and 5 years, respectively. Mean endothelial cell loss was (mean ± SD) 44% ± 16%, 56% ± 17%, and 67% ± 13% at 1, 3, and 5 years. The trend in mean endothelial cell density over time by least square regression was accurately and parsimoniously described by a straight line taking earliest values (3 mo) through to 5 years, with a rate of cell loss of 148 ± 13 cells·mm−2·yr−1. Higher-order polynomial and exponential models did not provide a closer regression fit. Mixed nonlinear modeling using exponential decay equations confirmed a relatively stable rate of cell loss for DSEK from 3 months through to 5 years, in contrast to penetrating keratoplasty models using similar techniques in previous studies. Conclusions: DSEK graft survival is high in our series through 5 years. Endothelial cell loss occurs at a relatively constant albeit low-grade rate from the earliest postoperative measurements through 5 years.