DMEK results in superior visual acuity outcomes compared to UT-DSAEK. This difference is most pronounced in the initial 3 months post-operatively, highlighting its faster visual rehabilitation, but it remains significantly better after 1 year follow-up. This is likely due to the improved restoration of anatomy in the DMEK procedure because of the use of a thinner graft with no lamellae and lack of stroma-to-stroma graft interface. Thinner grafts have been shown to induce fewer higher order aberrations and less hyperopic shift which may explain DMEK’s visual acuity superiority of UT-DSAEK (21–23). DMEK procedures also utilise smaller incisions, 2.2-3.5mm for DMEK versus 3.5-4.5mm for UT-DSAEK, which results in less surgically induced astigmatism (24). Both Mencucci et al. and Torras-Sanvicens et al. performed retrospective fellow eye comparisons of patients who underwent UT-DSAEK in one eye and DMEK in the other (16, 20). Although, they found similar BCVA at one year follow-up, DMEK performed better in terms of contrast sensitivity, posterior corneal aberrations and overall patient satisfaction.
Some discordance in visual acuity outcomes exists between our randomised controlled trials (8, 14, 15, 17). All three studies used strict eligibility criteria, only analysing patients with endothelial disease, primarily Fuchs’ endothelial dystrophy, in the absence of other vision-limiting pathologies. Dunker et al. found no significant differences in visual acuity between the two procedures as early as 3 months post-operatively and up to 1 year (14). They assessed outcomes in pre-operatively pseudophakic eyes to isolate the effect of keratoplasty alone on visual outcome. Although this doesn’t reflect the heterogeneity of the patient population attending for endothelial keratoplasty, it does suggest that the outcomes analysed are primarily attributable to the procedure alone. They did however report that a higher percentage of patients attained 0.1 LogMAR in the DMEK arm (66% of 29 eyes) compared to the UT-DSAEK arm (33% of 25 eyes). The DETECT and Matsou studies showed superior visual acuity in the DMEK group. 70% of patients in the DETECT cohort had a triple procedure. This, however, was done in equal proportion between both procedures, 68% in the UT-DSAEK group and 72% in the DMEK group. Although this may affect pre-operative BCVA, all patients analysed across all studies were pseudophakic at the end of their keratoplasty allowing comparable analysis of post-operative BCVA at one year.
The variability in outcomes between the studies may also relate to an inherent limitation in UT-DSAEK that is the lack of graft regularity. Graft regularity is an important parameter in the quality of an UT-DSAEK graft as and remains difficult to standardize, even with eye bank-prepared tissues (21). Despite this, the DETECT study and Matsou et al. found no difference in patient-reported functional vision as assessed by vision related quality of life (15, 25). This is in line with Matsou and Dunker’s findings of no significant difference in mean spherical equivalent, posterior and anterior corneal astigmatism between the two cohorts at any time point (14, 15).
ECD is a major factor in long-term graft survival (26). The Cornea Preservation Time Study evaluated factors influencing graft success and they found higher success in cases of Fuchs’ endothelial dystrophy, cases without any intra-operative complications and in which the donor did not have diabetes (27). In our analysis, we found ECD was comparable between the procedures. This is in line with multiple RCTs looking at ECD between DMEK and traditional DSAEK which have found no significant difference in the early post-operative period (6, 28, 29).
The rate of total complications is markedly higher after DMEK than after UT-DSAEK. The most reported complication in each cohort was graft detachment requiring re-bubbling (anterior chamber tamponade re-injections). This is a significant concern following DMEK (18). Romano et al. reported that re-bubbling was also higher in eyes undergoing triple procedure, occurring in 3/11 eyes having UT-DSAEK-triple and 8/15 undergoing DMEK-triple. However, contradictory results in other case series have shown that combining cataract procedures with DSAEK (30) or DMEK (31) has no impact on rates of detachment. The reported rates of re-bubbling following DMEK vary widely, ranging from 2–82% but on average are considered to occur in just under one-third of patients (32, 33). Repeated re-bubblings are associated with endothelial cell loss, raised intraocular pressure and graft rejections (6). Re-bubbling also poses additional logistic issues. It may have to be performed in theatre where access can often be an issue and may take precedence over other elective ophthalmic procedures. However, DMEK re-bubbling may be more frequently performed in clinic, compared to UT-DSAEK which often requires management in theatre owing to the higher pressure required for successful tamponade. Graft detachment rates may be reduced with the use of pre-stamped DMEK tissue (34). Additionally, studies have shown that SF6 gas anterior chamber tamponade may further reduce the risk of detachment to approximately 12% by facilitating adhesion at the graft-host interface (35–37). The DETECT study and Dunker et al. noted that the use of SF6 made no statistical difference to rates of re-bubbling, although they weren’t powered to specifically analyse this (8, 14, 17). Studies have also shown that the use of SF6 tamponade does not affect long-term BCVA or ECD (35–37).
Raised IOP was the second most reported complication in each cohort. It was not clearly stated what percentage of patients went on to develop secondary glaucoma. This raised IOP is likely a result of post-operative steroid use but may also develop from air bubble-related pupillary block (38). Prophylactic peripheral iridotomy may be performed pre- or intra-operatively to reduce the risk of pupillary block in either procedure (39).
Rates of graft failure (2 in DMEK & 1 in UT-DSAEK) and rejection were very low (1 in DMEK & 2 in UT-DSAEK). However, longer follow-up is required to comment on the risk of risk of allograft rejection and graft survival. As less tissue is transplanted in DMEK, there should be a reduced risk of allograft rejection and less reliance on topical steroids. A multi-centre study of 431 patients who underwent DMEK found a 3.7% (16 eyes) graft rejection rate (40). This is significantly lower than the rejection rates commonly reported following traditional DSAEK (9%) (41). UT-DSAEK may have lower rates of rejection compared to its predecessor, DSAEK, with its thinner graft and a large prospective series has shown similar rates of immunologic rejection to DMEK (10).
An important consideration is the use of eye bank versus surgeon prepared endothelial grafts. Romano et al. looked at DMEK graft preparation and found significantly higher adhesion and lower re-bubbling rates with surgeon prepared DMEK grafts compared to eye bank prepared tissue (18). Similar results were seen in a study examining re-bubbling rates in DSAEK procedures, with fewer detachments occurring in the surgeon prepared group (42). The use of surgeon or local technician prepared endothelial grafts, both DMEK and traditional DSAEK, have also been shown to significantly reduce surgical expenses in Canada and the United Kingdom (42, 43). These studies did not specifically analyse surgeon prepared UT-DSAEK grafts. However, both single and double-pass microkeratome techniques for performing have been well described in the literature with reproducible results (44–46). Techniques to improve graft preparation consistency include controlling artificial anterior chamber pressure and drying the corneal surface, described by Romano et al, and stromal swelling with balanced salt solution prior to microkeratome pass, detailed by Farbman et al. Single and double pass techniques for UT-DSAEK graft preparation have been shown to have comparable thickness, ECD and BCVA (47).
In 2015, according to the Eye Bank Association of America, DMEK comprised 15% of all EK procedures in the United States, whereas DSAEK accounted for over 50%. In 2021, their usage was almost equal (DMEK: 14,128 vs DSAEK: 15,935) (48). DMEK’s widespread adoption may have initially been limited by its highly technical nature, at every stage from donor tissue preparation to placement to post-operative graft attachment. Owing to the fragility of the tissue, graft folds occur in approximately 1.9% which may cause optical aberrations (40). Many studies have shown a steep learning curve for DMEK with an inverse relationship between surgeon experience and rates of graft detachment (49–51). Dapena et al examined the outcomes in a series of consecutive cases performed by a single surgeon and found re-bubbling was required in 20% of the first 45 cases, 13.3% in the middle 45 cases and 4.4% in the final 45 cases (49). DMEK is also a more challenging procedure in patients with complex anatomy, such as a very shallow or deep anterior chamber or a history of prior intra-ocular surgery or trauma (52). Visual outcomes and endothelial cell loss following DMEK remain stable however once the surgeon has completed a minimum of 25 cases (50, 51).
Limitations
This study has several limitations and potential biases, including the limitations of the included studies themselves. 4 of the 7 studies were retrospective thus potentiating selection and performance bias. It was not possible to adjust for age, sex or diagnosis. However, there was no significant difference in these demographic variables between the 2 groups. Additionally, the lack of a standardised graft thickness in UT-DSAEK remains a challenge in reporting its outcomes accurately. The overall heterogeneity was relatively low for visual acuity outcomes, at 3 months and 1 year, and complication measures, indicating a degree of consistency between these results in the studies. Nevertheless, high heterogeneity was seen in visual acuity outcomes at 6 months and ECD results, as indicated by the discrepancies in values between the studies.