The graph shows the preoperative best-corrected visual acuity (BCVA) (n = 275), the pooled postoperative BCVA to 6 months (n = 275), and the BCVA for the eyes that reached the 6-month postoperative follow-up (n = 176).
eFigure 1. World map of participating surgeons and their location
eFigure 2. Slitlamp and optical coherence tomography (OCT) pictures showing various degrees of graft detachment after Descemet membrane endothelial keratoplasty (DMEK)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Monnereau C, Quilendrino R, Dapena I, et al. Multicenter Study of Descemet Membrane Endothelial Keratoplasty: First Case Series of 18 Surgeons. JAMA Ophthalmol. 2014;132(10):1192–1198. doi:10.1001/jamaophthalmol.2014.1710
Surgeons starting to perform Descemet membrane endothelial keratoplasty (DMEK) should be informed about the learning curve and experience of others.
To document the clinical outcome of standardized “no-touch” DMEK and its complications during the learning curves of experienced surgeons.
Design, Setting, and Participants
Retrospective multicenter study. A total of 431 eyes from 401 patients with Fuchs endothelial dystrophy (68.2%) and bullous keratopathy (31.8%) underwent DMEK performed by 18 surgeons in 11 countries.
Descemet membrane endothelial keratoplasty.
Main Outcomes and Measures
Best-corrected visual acuity (BCVA), endothelial cell density, and intraoperative and postoperative complications.
Of 275 eyes available for BCVA pooled analysis, BCVA improved in 258 eyes (93.8%), remained unchanged in 12 (4.4%), and deteriorated in 5 (1.8%). Two hundred seventeen eyes (78.9%) reached a BCVA of at least 20/40 (≥0.5), 117 (42.5%) at least 20/25 (≥0.8), and 61 (22.2%) at least 20/20 (≥1.0). Eyes with at least 6 months of follow-up (n = 176) reached similar BCVA outcomes. Mean (SD) decrease in endothelial cell density at 6 months was 47% (20%) (n = 133 [P = .02]). Intraoperative complications were rare, including difficulties in inserting, unfolding, or positioning of the graft (1.2%) and intraoperative hemorrhage (0.5%). The main postoperative complication was graft detachment (34.6%); 20.4% underwent a single rebubbling procedure, occasionally requiring a second (2.6%) and a third rebubbling (0.7%), and 17.6% underwent a second keratoplasty.
Conclusions and Relevance
Our multicenter study showed that the standardized no-touch DMEK technique was feasible in most hands. The main challenges for surgeons starting to perform the procedure may be (1) to decide whether graft preparation is outsourced or performed during surgery, (2) to limit the number of graft detachments and secondary procedures, and (3) to obtain organ cultured donor corneal tissue.
Quiz Ref IDSince 1998, the Netherlands Institute for Innovative Ocular Surgery (NIIOS) introduced various techniques for endothelial keratoplasty, currently referred to as deep lamellar endothelial keratoplasty, Descemet stripping endothelial keratoplasty (DSEK)/automated DSEK (DSAEK), and Descemet membrane endothelial keratoplasty (DMEK).1-6 In DMEK, only the donor Descemet membrane (DM) and its endothelium are transplanted, allowing for better outcomes than all other keratoplasty techniques currently available, with 94% of eyes reaching a best (spectacle)-corrected visual acuity (BCVA) of at least 20/40 (≥0.5), 77% reaching at least 20/25 (≥0.8), and 47% reaching at least 20/20 (≥1.0) within 6 months.7,8
With deep lamellar endothelial keratoplasty and DSEK/DSAEK, we noticed that surgeons were sometimes unable to start these techniques successfully, owing to difficulties with donor tissue preparation and/or a lack of technique standardization. With DMEK, we therefore designed the technique for preparing the donor DM and the surgery itself as standardized “no-touch” procedures.9,10 The first DMEK outcomes of former NIIOS course participants were collected to document their experiences in the initial performance of DMEK and to evaluate the procedure’s clinical outcomes.
Because recognition of the problems and complications associated with commencing a new procedure may enable further technique improvements, recommendations, and/or logistic support, the aim of our study was to evaluate the clinical outcome of 431 eyes undergoing DMEK in the first clinical series of 18 different surgeons, located in 11 different countries.
A total of 431 eyes (of 401 patients) that underwent DMEK for endothelial disorders underwent retrospective analysis (Table 1). The surgical procedures were performed by 18 corneal surgeons in 11 countries (eFigure 1 in the Supplement). The mean (SD) number of DMEK procedures performed per surgeon was 24 (20 [range, 2-74]). In 38 eyes (8.8%), DMEK was combined with phacoemulsification surgery. For eyes that underwent a second DMEK, only the primary procedure was included for analysis in our study. Patients signed informed consent forms approved by the local institutional review boards in the various countries of the participating surgeons. The study was conducted in accordance with the Declaration of Helsinki.
Preparation of an isolated DM graft was performed by stripping the DM from a corneoscleral rim (NIIOS technique) by 15 surgeons in 385 of the 431 cases (89.3%).10 One surgeon used the “submerged cornea using backgrounds away” technique for 5 grafts (1.2%); 1 surgeon, a combination of both techniques for 25 grafts (5.8%); and 1 surgeon, an air bubble to separate the DM from a corneoscleral rim for 16 grafts (3.7%).11-13 Twelve surgeons prepared 282 of 431 grafts (65.4%) themselves from just before to as many as 6 days before surgery; 2 surgeons (42 of 431 grafts [9.7%]) used grafts prepared by an eye bank as long as 1 week before surgery; and 4 surgeons (107 of 431 grafts [24.8%]) used self-prepared or bank-supplied grafts for their procedures but did not document how many grafts of each type were used (Table 2). Preparation of the DMEK graft performed by the surgeon proved successful in 92.9% of cases. Seven surgeons routinely had a backup cornea in case the preparation failed.
Grafts underwent cold storage in preservative medium (Optisol-GS [Bausch & Lomb Surgical] or Eusol-C [Alchimia]) or storage in organ culture medium (modified minimum essential medium [CorneaMax; Eurobio AbCys] or Tissue-C [Alchimia]) (Table 3). Six surgeons used organ culture (113 of 431 grafts [26.2%]), 7 used cold storage (104 of 431 [24.1%]), 4 used both methods (140 of 431 [32.5%]), and 1 surgeon used freshly prepared grafts from a whole globe (74 of 431 [17.2%]).
All surgeons used the standardized no-touch DMEK technique with various modifications (Table 2).9 Ten surgeons (179 of 431 grafts [41.5%]) inserted the graft into the recipient anterior chamber with a curved glass pipet (DORC International); 5 surgeons (72 of 431 [16.7%]) used an intraocular lens injector (MSI-PF; Canon Staar, Bausch & Lomb), 1 surgeon (37 of 431 [8.6%]) used a glass cannula (modified Szurman; Geuder AG); and 1 surgeon (74 of 431 [17.2%]) used an 18-gauge intravenous catheter (Braunüle; Braun Melsungen AG). For the remaining 69 grafts (16.0%), the surgeon did not specify the manner of graft insertion.
Two surgeons marked the edges of the graft to visualize its orientation in the recipient anterior chamber, whereas the other surgeons used the Moutsouris sign.9 The size of the graft and duration of complete anterior chamber air fill varied among surgeons (Table 2). All surgeons left a 50% to 100% air fill at the end of the procedure, except for two, of whom one completed air-fluid exchange and the other did not specify the extent of air left in the anterior chamber.
From a total of 431 eyes, 156 were excluded from the BCVA analysis, including 45 because of low visual potential due to concomitant eye diseases unrelated to the corneal transplant, 79 that underwent a second surgery within the study period, 6 with graft failure or allograft rejection but not yet scheduled for reoperation, and 26 with incomplete data. Hence, 275 eyes were available for BCVA analysis, of which 176 had 6 months of follow-up, 43 had 3 months, and 56 had 1 month.
Preoperative donor endothelial cell density (ECD) was measured by providing eye banks, using specular microscopy or inverted light or phase-contrast microscopy (Noncon Robo [Konan Medical Inc]; Kerato Analyzer [Konan Medical Inc]; inverted microscope [Nikon]; Topcon SP1000 [Topcon Medical Europe BV]; and Axiovert inverted light microscope [Zeiss]). Postoperative ECD was measured as long as 6 months after the procedure using specular microscopy (Noncon Robo; Topcon SP2000/SP3000 [Topcon Medical Europe BV]; Bon Optic EM-2 specular microscope [Carleton Ltd]; CSO specular microscope [CSO]; and EM 3000 [Tomey GmbH]). Complete preoperative and postoperative ECD measurements were available for 133 eyes.
Statistical analysis was performed with commercially available software (SPSS, version 15.0 for Windows; SPSS, Inc). Analysis of variance and the paired t test were used to compare ECD between different subgroups and different times. Regression analysis was performed to evaluate the potential effect of different factors on the outcome of the surgical procedures. Statistical significance was determined as P < .05.
A total of 431 consecutive eyes underwent standardized no-touch DMEK with various modifications (Table 2).9 A total of 18 surgeons performed the procedures in 401 patients.
Of the 431 eyes, 156 were excluded from BCVA analysis, leaving 275 eyes available. Preoperative BCVA was less than 20/40 (<0.5) for 235 eyes (85.5%), less than 20/100 (<0.2) for 114 (41.5%), and less than 20/200 (<0.1) for 57 (20.7%). Postoperative BCVA improved by 2 or more Snellen lines in 258 of 275 eyes (93.8%), remained unchanged in 12 (4.4%), and deteriorated in 5 (1.8%). For all eyes with a follow-up of 1 to 6 months, 217 of 275 (78.9%) reached a BCVA of at least 20/40 (≥0.5), 117 (42.5%) at least 20/25 (≥0.8), 61 (22.2%) at least 20/20 (≥1.0), and 8 (2.9%) at least 24/20 (≥1.2) (Figure). If limited to the 176 eyes that reached the 6-month follow-up, 144 (81.8%) reached a BCVA of at least 20/40 (≥0.5), 77 (43.8%) reached at least 20/25 (≥0.8), 33 (18.8%) reached at least 20/20 (≥1.0), and 3 (1.7%) reached at least 24/20 (≥1.2) (Figure).
Exclusion of the first 25 procedures on the learning curve for each surgeon resulted in 45 cases of 5 different surgeons that reached the 6-month of follow-up. In this group, 39 of 45 eyes (87%) reached a BCVA of at least 20/40 (≥0.5), 27 (60%) reached at least 20/25 (≥0.8), and 12 (27%) reached at least 20/20 (≥1.0).
Complete preoperative and 6-month postoperative ECD measurements were available for 133 eyes. Mean (SD) donor ECD was 2625 (333) cells/mm2 before and 1399 (533) cells/mm2 at 6 months after surgery, representing an overall decrease in ECD of 47% (20%) (P = .02) from the preoperative to the 6-month postoperative measurements. Owing to missing values, linear regression analysis did not have enough statistical power to determine possible subfactors affecting ECD decrease, such as lens status, type of inserter (plastic or glass), air-bubble time and size, donor storage medium, surgeon- or eye bank–prepared grafts, or the individual surgeon (P = .44).
Intraoperative complications included difficulties during insertion, unfolding, and/or positioning of the Descemet graft and were observed in 5 eyes (1.2%) (Table 4). In 2 cases (0.5%) a small intraoperative hemorrhage occurred (Table 4). No other intraoperative complications were reported.
Quiz Ref IDThe most frequent postoperative complication was partial graft detachment, which occurred in 124 of 431 eyes (28.8%) (Table 4). Detachment was no more than one-third of the graft surface area in 80 eyes (18.6%) and more than one-third of the surface area in 31 (7.2%). For 13 cases (3.0%), the size of the detachment was not specified (eFigure 2 in the Supplement). Complete detachment, that is, a roll in the anterior chamber, occurred in 18 of 431 eyes (4.2%), and in 7 cases (1.6%), the graft was positioned upside down. Of the partially detached grafts, 43 of 124 (34.7%) required a second transplant (Table 4). Regression analysis showed that the use of plastic inserters (P = .005), cold storage medium (P = .005), and a short air-bubble time (<1 hour) during surgery (P = .02) correlated positively with the incidence and extent of postoperative graft detachment. In addition, cold storage correlated with more extensive detachments than grafts stored using organ culture (P = .01).
Other postoperative complications included primary graft failure owing to endothelial insufficiency (10 of 431 cases [2.3%]), secondary graft failure (27 of 431 [6.3%]), allograft rejection (16 of 431 [3.7%]), epithelial defects and/or erosion (13 of 431 [3.0%]), secondary glaucoma (12 of 431 [2.8%]; of which 5 cases responded to corticosteroid therapy), and significant folds or wrinkles of the Descemet graft (8 of 431 [1.9%]). All other complications were incidental (≤1%) and included cystoid macular edema, anterior synechiae, hypotonia, pupillary block, dendritic keratitis/endothelitis, intraocular lens vitreous luxation, cataract, melting ulcus corneae, subepithelial haze, macular pucker, and interface pigment deposits (Table 4).
Quiz Ref IDA total of 102 of 431 eyes (23.7%) required a rebubbling procedure because of partial graft detachment. A single rebubbling procedure was performed for 88 eyes, whereas 11 needed a secondary and 3 needed a third procedure (Table 4). Rebubbling proved successful in 83 of 102 eyes (81.4%), whereas the remaining 19 eyes needed a second transplant. Overall, 79 of 431 eyes (18.3%) needed a secondary corneal procedure; of these, 46 obtained a secondary DMEK, 2 needed a third DMEK, and 1 needed a fifth DMEK. Fifteen eyes underwent a secondary DSEK/DSAEK and 15 underwent a secondary penetrating keratoplasty (Table 4). After a secondary DMEK, preoperative and postoperative BCVA were available for 20 eyes, showing an increase in BCVA in 15 (75%) from less than 20/40 (<0.5) to at least 20/40 (≥0.5). Seven eyes (35%) obtained a BCVA of at least 20/25 (≥0.8) and 3 (15%) of at least 20/20 (≥1.0).
In the present multicenter study, we evaluated the first series of DMEK procedures performed by 18 corneal surgeons in various clinical settings and different countries to document the logistic and technical problems, the clinical outcomes, and the complications encountered when surgeons start performing DMEK. As such, our findings may assist other surgeons in making the switch from DSEK/DSAEK to DMEK, in choosing the best approach for their specific setting, and in shortening their learning curves. Overall, the first experience with DMEK was such that 17 of the 18 participating surgeons continued to perform the procedure. One surgeon felt more comfortable continuing with DSAEK. All surgeons had long-term experience with penetrating keratoplasty and/or DSEK/DSAEK, so that in this study group the clinical results after DMEK justified making the switch from DSEK/DSAEK to DMEK for all but 1 surgeon.
Within the first 6 months after DMEK, 78.9% of eyes reached a BCVA of at least 20/40 (≥0.5); 42.5%, at least 20/25 (≥0.8); and 22.2%, at least 20/20 (≥1.0) (Figure). Hence, during the learning curve, most patients may already obtain a BCVA that allows them to perform their daily activities and to obtain or keep a driver’s license. In a previous study,14 the learning curve did not seem to influence the BCVA or the ECD, but the number of functional (attached) grafts increased with surgical experience. In the present study, visual outcome seemed to vary with surgical experience. Thirteen of 18 surgeons performed fewer than 25 DMEK procedures at the time of data collection, which may explain the visual outcome being lower than in large series on DMEK, but still favorable compared with that reported for DSEK/DSAEK.7,8,15-17 Because not all surgeons had performed 25 DMEK procedures yet, a statistical analysis of the first 25 cases only per surgeon was not feasible. Our study showed a mean ECD decrease of 47% when postoperative values were compared with preoperative values. Other investigators have reported a mean ECD decrease of 34% to 41% at the same postoperative interval.7,17-19 These numbers may be interpreted with some caution, because 50% of surgeons in previous reports prepared the donor tissue themselves immediately before surgery. As a result, the decrease in ECD may have been overestimated, that is, ECD before DM stripping was used for comparison with postoperative ECD values. We had anticipated that during the surgeon’s learning curve, DM stripping would also be more traumatic than when tissue prepared by an eye bank was used. However, in this study, the decrease in ECD did not prove to differ statistically between the surgeon- and eye bank–prepared grafts.
Intraoperative complications during the DMEK learning curve were relatively rare (1.6%), which agrees with previous findings.13,20-22 The relatively low number of intraoperative complications may in part be explained by the fact that all surgeons participated in a Wetlab instruction course at the NIIOS, suggesting that such training sessions are effective in avoiding the most common pitfalls (Table 4).
As in earlier studies, by far the most frequent postoperative complication after DMEK was partial graft detachment.7,14,23 Of the 149 graft detachments reported in our study, 80 (53.7%) consisted of relatively small, peripheral detachments (one-third or less of the graft surface area). Such detachments are usually clinically insignificant, because they spare the visual axis. If no graft adherence is obtained, the recipient cornea overlying the detachment tends to clear with time, owing to endothelial repopulation of the denuded recipient posterior stroma.24 Larger detachments (31 of 149 detached grafts [20.8%]) were often managed by a rebubbling procedures, whereas a second DSEK/DSAEK or DMEK was used in the event of complete detachment (18 of 149 detached grafts [12.1%]) or when the graft was positioned upside down (7 of 149 detached grafts [4.7%]) (Table 4). Hence, about 18% of cases (45.6% of the detached grafts) may require a secondary surgical intervention to manage graft detachment.
Quiz Ref IDOur study showed that 3 main factors may contribute to inadvertent graft detachment, all of which may be relatively easy to manage or avoid: the use of cold storage media, the type of inserter used, and the air-bubble time during surgery. In 2011, Laaser et al25 reported a difference in detachments between DSAEK grafts stored in cold preservative medium (Optisol-GS) and organ culture medium (CorneaMax medium); grafts stored in cold preservative medium more often required rebubbling than those stored in organ culture medium. Our study seemed to confirm this trend with 34.6% detachments with cold preservative medium vs 26.5% with organ culture (Table 3). Also, plastic (instead of glass) graft inserters correlated with higher postoperative graft detachment rates. This finding agrees with results of our initial in vitro experiments (G.R.J.M., unpublished data; 2004), which showed more variable endothelial cell damage with plastic than with glass inserters and a somewhat different behavior of the graft during insertion and unfolding, possibly owing to electrostatic forces induced by plastics. In all surgical procedures performed with an injector made of plastic material, the air-fill time was 1 hour or longer; subsequently, no correlation between plastic injectors and short air-fill time could be identified. A short air-fill time (<1 hour) was reported in 69 DMEK procedures. In all of these procedures, an inserter made of glass material had been used. Subsequently, the factor of short air-fill time was independent of a possible positive correlation with a plastic injector. Finally, leaving the patient in a supine position with a complete air fill of the anterior chamber for at least 1 hour may be advocated to minimize the risk for graft detachment. In other words, a mean air-bubble time of less than 1 hour may be insufficient to obtain complete graft attachment. After reducing the air-bubble size at termination of the surgery, the actual size of the air bubble left in the anterior chamber may have little effect on the final graft detachment rate.
Secondary graft failure within the first 6 months occurred in 6.3% of eyes and may be attributed to endothelial damage during donor tissue preparation and/or manipulation of the tissue during or after insertion of the graft into the anterior chamber; in addition, it may be attributed to postoperative events. Compared with an incidence of 5% to 12% after DSEK/DSAEK, a lower allograft rejection rate has been reported for DMEK, varying from 1% to 5% within the first postoperative year.18,26-31 In the present study, 3.7% of eyes showed an allograft rejection; 8 surgeons reported rejections in 1 or more cases, whereas others observed none. The risk for rejection may vary considerably among study populations; for example, African American patients may show a stronger immune response to corneal grafts and require higher doses of corticosteroid regimes than white patients.26 All other reported complications seemed incidental and not specifically related to DMEK or its learning curve (Table 4).
Most of the secondary procedures were performed to manage partial graft detachment after DMEK, in particular rebubbling in 23.7% (Table 4). This procedure proved successful in 81.4%, whereas the remaining eyes required retransplant. Hence, rebubbling was quite effective in the present study.16 However, previous experience suggested that, especially with small, peripheral detachments over a limited graft surface area, spontaneous corneal clearance or reattachment may be awaited.7 Repeated DMEK procedures were also effective because 75% of these eyes had a final BCVA of at least 20/40 (≥0.5).
Our multicenter study showed that the standardized no-touch DMEK technique was feasible for most surgeons. The main challenges for surgeons starting to perform the procedure may be (1) to decide whether graft preparation is outsourced or performed during surgery, (2) to limit the number of graft detachments and second procedures, and (3) to obtain organ cultured donor corneal tissue.
Submitted for Publication: September 11, 2013; final revision received January 16, 2014; accepted March 6, 2014.
Corresponding Author: Gerrit R. J. Melles, MD, PhD, Netherlands Institute for Innovative Ocular Surgery, Laan op Zuid 88, 3071 AA Rotterdam, the Netherlands (firstname.lastname@example.org).
Published Online: July 3, 2014. doi:10.1001/jamaophthalmol.2014.1710.
Author Contributions: Dr Melles had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Monnereau, Dapena, Alfonso, Böhnke, Hashemi, Naveiras, Melles.
Acquisition, analysis, or interpretation of data: Monnereau, Quilendrino, Liarakos, Alfonso, Arnalich-Montiel, Böhnke, Pereira, Dirisamer, Parker, Droutsas, Geerling, Gerten, Hashemi, Kobayashi, Naveiras, Oganesyan, Orduña Domingo, Priglinger, Stodulka, Torrano Silva, Venzano, Vetter, Yiu, Melles.
Drafting of the manuscript: Monnereau, Quilendrino, Liarakos, Alfonso, Gerten, Naveiras, Orduña Domingo, Torrano Silva, Venzano, Melles.
Critical revision of the manuscript for important intellectual content: Monnereau, Quilendrino, Dapena, Alfonso, Arnalich-Montiel, Böhnke, Pereira, Dirisamer, Parker, Droutsas, Geerling, Hashemi, Kobayashi, Naveiras, Oganesyan, Priglinger, Stodulka, Vetter, Yiu, Melles.
Statistical analysis: Quilendrino, Liarakos, Torrano Silva.
Administrative, technical, or material support: Monnereau, Alfonso, Böhnke, Pereira, Parker, Droutsas, Hashemi, Naveiras.
Study supervision: Dapena, Arnalich-Montiel, Dirisamer, Priglinger, Stodulka, Vetter, Melles.
Conflict of Interest Disclosures: Dr Geerling is a consultant for Bausch & Lomb, Alcon, Allergan, Santen, and Thea Pharma; is a board member of Health Care Goringer Publishers; and has grants or grants pending for the Deutsche Forschungsgemeinschaft. Dr Stodulka is a consultant for Bausch & Lomb. Dr Venzano has grants or grants pending and travel/accommodation/meeting expenses unrelated to the activities listed. Dr Melles is a consultant for DORC International/Dutch Ophthalmic USA. No other disclosures were reported.