Alastair J Stuart, Vito Romano, Gianni Virgili, Alex J Shortt

Corneal endothelial transplantation has become the gold standard for the treatment of corneal endothelial dysfunctions, replacing full thickness transplantation, known as penetrating keratoplasty. Corneal endothelial transplantation has been described using two different techniques: Descemet's membrane endothelial keratoplasty (DMEK) and Descemet's stripping automated endothelial keratoplasty (DSAEK). Both are still performed worldwide. To compare the effectiveness and safety of Descemet's membrane endothelial keratoplasty (DMEK) versus Descemet's stripping automated endothelial keratoplasty (DSAEK) for the treatment of corneal endothelial failure in people with Fuch's endothelial dystropy (FED) and pseudophakic bullous keratopathy (PBK). We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (which contains the Cochrane Eyes and Vision Trials Register) (2017, Issue 7); MEDLINE Ovid; Embase Ovid; LILACS BIREME; the ISRCTN registry; ClinicalTrials.gov and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP). The date of the search was 11 August 2017. We included randomised controlled trials (RCTs) and non-randomised paired, contralateral-eye studies in any setting where DMEK was compared with DSAEK to treat people with corneal endothelial failure. Two review authors independently screened the search results, assessed trial quality and extracted data using the standard methodological procedures expected by Cochrane. Our primary outcome was best corrected visual acuity (BCVA) measured in logarithm of the Minimum Angle of Resolution (logMAR). Secondary outcomes were endothelial cell count, graft rejection, primary graft failure and graft dislocation. We graded the risk of bias of non-randomised studies (NRSs) using ROBINS-I. We did not identify any RCTs but found four non-randomised studies (NRSs) including 72 participants (144 eyes), who had received DSAEK in the first eye followed by DMEK in the fellow eye. All the studies included adult participants where there was evidence of FED and endothelial failure requiring a corneal transplant for the treatment of visual impairment. We did not find any studies that included PBK. The trials were published between 2011 and 2015, and we assessed them as high risk of bias due to potential unknown confounding factors since DSAEK preceded DMEK in all participants. Two studies reported results at 12 months, one at 6 months, and one between 6 and 24 months. At one year, using DMEK in cases of endothelial failure may result in better BCVA compared with DSAEK (mean difference (MD) -0.14, 95% confidence interval (CI) -0.18 to -0.10 logMAR, 4 studies, 140 eyes, low-certainty evidence). None of the participants had severe visual loss (BCVA of 1.0 logMAR or more; very low-certainty evidence). Regarding endothelial cell count data (4 studies, 134 eyes) it is hard to draw any conclusions since two studies suggested no difference and the other two reported that DMEK provides a higher cell density at one year (very low-certainty evidence). No primary graft failure and only one graft rejection were recorded over four studies (144 eyes) (very low-certainty evidence). The most common complications reported were graft dislocations, which were recorded in one or two out of 100 participants with DSAEK but were more common using DMEK, although this difference could not be precisely estimated (risk ratio (RR) 5.40, 95% CI 1.51 to 19.3; 4 studies, 144 eyes, very low-certainty evidence). This review included studies conducted on people with corneal endothelium failure due to FED for whom both DMEK and DSAEK can be considered, and found low-certainty evidence that DMEK provides some advantage in terms of final BCVA, at the cost of more graft dislocations needing 're-bubbling' (very low-certainty of evidence).