CorneaFebruary 2026Comparative Study

Long-term Outcomes of Descemet Stripping Automated Endothelial Keratoplasty Versus Descemet Membrane Endothelial Keratoplasty Under a Failed Penetrating Keratoplasty Graft.

Glaucoma Surgery

Summary

The only significant difference between DSAEK and DMEK cohorts was rebubble rate.

Abstract

PURPOSE

To compare long-term outcomes of Descemet membrane endothelial keratoplasty (DMEK) versus Descemet stripping automated endothelial keratoplasty (DSAEK) under a failed penetrating keratoplasty (PKP) graft.

METHODS

We retrospectively reviewed consecutive surgeries (Febuary 2007-September 2023) that received a DSAEK graft (n = 52) or DMEK graft (n = 51) under a failed PKP. Postoperative complication rates of primary graft failure (PGF), late endothelial graft failure, graft rejection, rebubble, and steroid response were recorded. Endothelial cell loss (ECL) at 6-month, 2-year, and 5-year postoperative visits were calculated. Graft survival time and survival probability at 5 years was determined using a Kaplan-Meier analysis. Twelve DMEK and 17 DSAEK eyes had 5-year follow-up. Eyes with a glaucoma filtration tube, trabeculectomy, or anterior chamber intraocular lens were excluded.

RESULTS

Rebubble rate was significant between DSAEK and DMEK cohorts (5.8%, 29.4%, P = 0.0017*). Rate of rejection, late endothelial graft failure, steroid response, 6-month ECL, 2-year ECL, 5-year ECL, survival probability, and 5-year mean survival time were not significant between DSAEK and DMEK cohorts [(9.6%, 5.9%, P = 0.7155), (13.5%, 9.8%, P = 0.7602), (9.6%, 11.8%, P = 0.7602), (32.0% ± 21.0%, 41.65% ± 25.6%, P = 0.1949), (55.8% ± 21.6%, 62.5% ± 19.8%, P = 0.4463), (59.5% ± 35.0%, 72.5% ± 5.7%, P = 0.3468), (88.5%, 84.3%, P = 0.483), (53.9 ± 2.4 confidence interval, 49.1-58.6 months, 52.7 ± 2.9 confidence interval, 47.0-58.5 months)]. PGF occurred in 0/52 (0%) and 3/51 (5.9%) eyes in the DSAEK and DMEK cohorts with nonsignificant trends favoring DSAEK ( P = 0.1178).

CONCLUSIONS

The only significant difference between DSAEK and DMEK cohorts was rebubble rate. However, nonsignificant trends favored PGF and ECL in the DSAEK cohort. Further investigation is warranted to determine if DSAEK provides better short-term and similar long-term outcomes compared with DMEK under a failed PKP.

Keywords

Descemet membrane endothelial keratoplastyDescemet stripping automated endothelial keratoplastycorneaendothelial keratoplastypenetrating keratoplasty graft failuretransplant

In the Knowledge Library

Discussion

Comments and discussion will appear here in a future update.