DMEK, DSAEK, and repeat PKP for endothelial failure after PKP: systematic review and meta-analysis with a graphical synthesis of clinical decision factors.

purposeTo compare outcomes of Descemet membrane endothelial keratoplasty (DMEK), Descemet stripping automated endothelial keratoplasty (DSAEK), and repeat penetrating keratoplasty (re-PKP) performed for endothelial failure after penetrating keratoplasty (PKP), and to summarize the clinical factors that appear to influence procedure selection.

methodsA review and meta-analysis was conducted using a preregistered protocol. Studies reporting outcomes of DMEK, DSAEK, or re-PKP for post-PKP endothelial failure were included. A systematic search was conducted in MEDLINE (via PubMed), Embase, and the Cochrane Library from inception. Pooled proportions were calculated for graft failure, rejection, and detachment. Eligible studies included original clinical studies of adult eyes with endothelial failure after prior PKP undergoing DMEK, DSAEK, or re-PKP, with at least five post-PKP eyes in a relevant treatment arm and reporting at least one outcome of interest. Visual acuity (VA) was synthesized as weighted means with bootstrap confidence intervals at predefined timepoints (preoperative, 3-5 months, 6 months, 12-13 months). Clinical decision factors were summarized descriptively and graphically. Risk of bias was assessed using the Joanna Briggs Institute checklist for case series and ROBINS-I V2 for comparative non-randomized studies.

resultsThirty-two studies met inclusion criteria, consisting predominantly of retrospective case series, with a smaller number of prospective or comparative observational studies. Pooled failure rates were 0.22 for DMEK, 0.19 for DSAEK, and 0.29 for re-PKP. Rejection was lowest after DMEK (0.07) and higher after DSAEK (0.14) and re-PKP (0.15). Detachment occurred more frequently after DMEK (0.37) than DSAEK (0.18). All three procedures improved VA relative to baseline. DMEK showed numerically greater and more sustained improvement through 12-13 months, whereas DSAEK and re-PKP improved early and then stabilized; however, these findings are descriptive and should be interpreted with caution given differences in baseline visual acuity and follow-up across studies. Several clinical factors-including the initial PKP indication, pre-existing glaucoma, and prior glaucoma surgery-were variably associated with procedure choice, but reporting was inconsistent.

conclusionsDMEK, DSAEK, and re-PKP all improve vision after endothelial failure of a prior PKP but differ in complication profiles and in the trajectory of visual recovery. DMEK showed numerically greater visual acuity improvement and the lowest pooled rejection proportion, albeit with a higher pooled detachment proportion. Current evidence is limited by incomplete reporting of preoperative and graft-related variables, the predominance of retrospective non-randomized studies, and variable risk of bias across included studies, constraining the ability to derive a consistent decision framework.