Figure 1. Graph displaying the best-corrected visual acuity (BCVA) before (Preop) and at 1, 3, and 6 months after Descemet membrane endothelial keratoplasty (DMEK) surgery.
Figure 2. Topographic corneal power maps of the anterior corneal curvature (A-C), posterior corneal curvature (D-F), and central corneal thickness (G-I) before Descemet membrane endothelial keratoplasty (Preop) (A, D, and G) and 6 months after Descemet membrane endothelial keratoplasty (Postop) (B, E, and H) and the corresponding difference maps (C, F, and I). The anterior corneal curvature is stable, but the posterior curvature does show a change of approximately 1.0 diopter. Compare with Table 7.
Figure 3. Scatterplot displaying the difference in preoperative to 6 months' postoperative central corneal thickness and spherical equivalent, which revealed a significant correlation. D indicates diopters.
Figure 4. Slitlamp and Scheimpflug images of an eye 1 month after successful Descemet membrane endothelial keratoplasty (DMEK) surgery (A and B), and another eye 1 and 6 months after DMEK surgery complicated by graft detachment (C-F). Note the near normal corneal anatomy after successful DMEK surgery, while the other eye initially showed corneal decompensation (C and D) (arrows) followed by “spontaneous clearance” of the transplanted cornea despite persistent graft detachment (E and F) (arrows).
Figure 5. Slitlamp photographs of an eye 1 month after Descemet membrane endothelial keratoplasty. Note the recipient Descemet membrane remnant (arrows) in the paracentral cornea (A), potentially interfering with the visual acuity (B).
Figure 6. Diagram displaying the number and sequence of secondary interventions after the 200 Descemet membrane endothelial keratoplasty (DMEK) surgeries performed in this study. DSEK indicates Descemet stripping endothelial keratoplasty.
Dirisamer M, Ham L, Dapena I, Moutsouris K, Droutsas K, van Dijk K, Frank LE, Oellerich S, Melles GRJ. Efficacy of Descemet Membrane Endothelial KeratoplastyClinical Outcome of 200 Consecutive Cases After a Learning Curve of 25 Cases. Arch Ophthalmol. 2011;129(11):1435-1443. doi:10.1001/archophthalmol.2011.195
Author Affiliations: Netherlands Institute for Innovative Ocular Surgery (Drs Dirisamer, Dapena, Moutsouris, Droutsas, Oellerich, and Melles and Mss Ham and van Dijk), Melles Cornea Clinic Rotterdam (Drs Dirisamer, Dapena, Moutsouris, Droutsas, Oellerich, and Melles and Mss Ham and van Dijk), and Amnitrans Eyebank Rotterdam (Dr Melles and Ms Ham), Rotterdam, and Department of Methodology and Statistics, Utrecht University, Utrecht (Dr Frank), the Netherlands; and Ophthalmology Department, AKh Linz, Linz, Austria (Dr Dirisamer).
Objective To evaluate Descemet membrane endothelial keratoplasty for management of corneal endothelial disorders.
Methods Descemet membrane endothelial keratoplasty was performed in 200 patients with Fuchs endothelial dystrophy or bullous keratopathy. Best-corrected visual acuity, subjective and objective refractive outcome and stability, and endothelial cell density were evaluated at 1, 3, and 6 months postoperatively, and intraoperative and postoperative complications were documented.
Results At 6 months, 94% reached a best-corrected visual acuity of 20/40 or better (≥0.5); 77%, 20/25 or better (≥0.8); 47%, 20/20 or better (≥1.0), and 16%, 20/17 or better (≥1.2) (n = 159). The preoperative to 6 months' postoperative spherical equivalent showed a mean (SD) +0.38 (1.2) diopter hyperopic shift (P = .001) that correlated with a decrease in central corneal thickness (n = 143) (P = .047). Two-thirds of eyes showed refractive stability at 3 months. Donor endothelial cell density showed a decrease from mean (SD) 2560 (186) cells/mm2 preoperatively to 1690 (520) cells/mm2 at 6 months after surgery (n = 173) (P < .001). Graft detachment was the main complication and occurred in 18 eyes (9%). Recipient Descemet membrane remnants were present in 12 eyes (6%). Secondary glaucoma was seen in 8 eyes (4%), of which 4 showed air-bubble dislocation behind the iris. In 2 of 33 phakic eyes (6%), a secondary cataract developed requiring phacoemulsification.
Conclusions Descemet membrane endothelial keratoplasty may offer complete visual rehabilitation within 1 to 6 months after surgery in a majority of eyes. Similar to earlier keratoplasty techniques, Descemet membrane endothelial keratoplasty may be associated with a one-third decrease in donor endothelial cell density in the early postoperative phase. Incidence of (partial) graft detachment stabilized at about 5% but could be further reduced by patient selection and/or technique modification.
Trial Registration clinicaltrials.gov Identifier: NCT00521898
Since 1998, we have introduced various techniques for treatment of corneal endothelial disorders, popularized as deep lamellar endothelial keratoplasty and Descemet stripping (automated) endothelial keratoplasty (DSEK/DSAEK).1- 5 Recently, we described a technique for selective transplant of the Descemet membrane (DM), referred to as Descemet membrane endothelial keratoplasty (DMEK).1,2
Our initial results showed that if successful, DMEK may enable significantly better visual outcomes than deep lamellar endothelial keratoplasty and DSEK/DSAEK6 and faster visual rehabilitation.7 These results have been confirmed by others,8,9 who also reported a best-corrected visual acuity (BCVA) of 20/40 or better (≥0.5) in about 95% of patients and 20/25 or better (≥0.8) in about 60% at 6 months postoperatively. All of these reports included the first DMEK cases performed by these surgeons, so the overall outcome may have been negatively biased by a “learning-curve” effect.
In our initial cases, DMEK was relatively frequently complicated by graft detachment.10,11 Hence, extensive in vitro studies were performed to minimize the risk of detachment and to further standardize the technique to allow its efficacy.12 Hence, our study aimed to evaluate the surgical protocol for standardized “no-touch” DMEK by documenting the clinical outcome of 200 consecutive DMEK cases performed after a first series of 25 learning-curve cases.
Our prospective study included 200 consecutive eyes of 166 patients with Fuchs endothelial dystrophy or bullous keratopathy who underwent DMEK (Table 1). The 200 cases evaluated were cases 26 to 225 from the first 225 consecutive DMEK surgeries performed by our group, after an initial learning curve of 25 DMEK eyes. The 200 surgeries were (partially) performed by 5 surgeons (M.D., I.D., K.M., K.D., and G.R.J.M.), and in 51 cases, the entire surgery was conducted by the fellow alone. After institutional review board review of the study protocol, all patients signed an institutional review board–approved informed consent; the study was conducted according to the Declaration of Helsinki and registered at www.clinicaltrials.gov (NCT00521898).
A procedure for harvesting of the DM graft has been previously described.13 In short, from donor globes obtained 36 hours or less post mortem, corneoscleral buttons were excised and stored by organ culture at 31°C. After 1 week of culture, endothelial cell morphology and viability were evaluated and the corneoscleral buttons were mounted endothelial side up on a custom-made holder. Being submerged in saline, a 9.5-mm-diameter DM sheet with its endothelium was stripped from the posterior stroma. Because of the elastic tissue properties, a “Descemet roll” formed spontaneously, with the endothelium at the outer side. Each Descemet roll was then stored for 5 to 10 days in organ culture medium until the time of transplant.13
All eyes were operated on under local anesthesia (4 mL of ropivacaine hydrochloride, 1%, with 150 IE of Hyason), followed by ocular massage and application of a Honan balloon for 10 minutes, and the patient was positioned in an anti-Trendelenburg position.
Surgeries were performed as previously described.12 A 3.0-mm tunnel incision was made at the limbus, the anterior chamber was filled with air, and a circular portion of DM was scored with an inversed Sinskey hook (DORC International, Zuidland, the Netherlands) and stripped from the posterior stroma so that a 9.0-mm-diameter descemetorhexis was created.14
The donor Descemet roll was stained with a trypan blue solution, 0.06% (VisionBlue; DORC International), and sucked into a custom-made injector (DMEK inserter; DORC International) to inject the Descemet roll into the recipient anterior chamber and the graft was oriented endothelial side down (donor DM facing recipient posterior stroma) by indirect manipulation with air and balanced salt solution.12 The graft was then gently spread out over the iris, and an air bubble was injected underneath the graft to position it onto the recipient posterior stroma.12 The anterior chamber was left completely filled with air for 30 to 60 minutes followed by an air-liquid exchange to pressurize the eye. Each surgical procedure was recorded on DVD (DVR-RT601H-S; Pioneer, Tokyo, Japan).
In all eyes, a YAG laser peripheral iridotomy was made at the 12-o’clock position 1 to 2 weeks before the surgery to reduce the potential risk of pupillary block glaucoma after surgery due to the 30% to 50% air bubble in the anterior chamber. Because patients were requested to lie in a supine position after surgery (with the Bell phenomenon rotating the eye upward on eye closure), the iridotomy was made at the 12-o’clock position.
Donor endothelial cell density (ECD) was evaluated in vitro (Axiovert 40 inverted light microscope; Zeiss, Göttingen, Germany) and photographed (PixeLINK PL-A662; Zeiss).13 In patient eyes, ECD was evaluated in vivo using a Topcon SP3000p noncontact autofocus specular microscope (Topcon Medical Europe BV, Capelle a/d IJssel, the Netherlands). Images of the central corneal window were manually corrected and 3 measurements were averaged.
All recipient eyes were examined before and after DMEK at 1, 3, and 6 months with biomicroscopy, Pentacam imaging (Oculus, Wetzlar, Germany), noncontact specular microscopy, and slitlamp photography (Topcon Medical Europe BV). Best-corrected visual acuity, ECD, and intraoperative and postoperative complications were recorded in a database.
For all comparisons, 2-sided paired-sample t tests were performed (SPSS version 18.0; IBM SPSS, Chicago, Illinois). P values for the Pentacam and refractive data were corrected with the Benjamini and Hochberg correction (multiple tests increase false positives).15 After correction, all P values <.05 represented statistical significance. The relation between the change in spherical equivalent (SE) and central corneal thickness (induced hyperopic shift) was estimated with the Pearson correlation.
The first 200 consecutive DMEK cases were evaluated. Hence, surgical cases 26 to 225 were evaluated at 1, 3, and 6 months after surgery, 33 (17%) of them with phakic eyes (Table 1). In patients with phakic eyes referred to us for combined cataract extraction and DMEK, phacoemulsification was performed 1 to 2 months prior to the transplant. Postphacoemulsification measurements were used as preoperative DMEK refractive data to avoid bias related to the margin of error in the refractive effect after intraocular lens implantation.
From a total of 200 eyes, 41 were excluded from visual acuity analysis: 23 had low visual potential, 12 had secondary DSEK performed after DMEK failure, 4 showed spontaneous corneal clearance despite graft detachment,16 and 2 had incomplete measurements (Table 2).
At 6 months, all but 9 eyes (94%) reached a BCVA of 20/40 or better (≥0.5); 77%, 20/25 or better (≥0.8); 47%, 20/20 or better (≥1.0); and 16%, 20/17 or better (≥1.2) (n = 159) (Figure 1) (Table 3). At 1 month, these percentages were 85%, 53%, 22%, and 3%, respectively, and at 3 months, 92%, 65%, 36%, and 6% (Figure 1) (Table 3). In 23 patients under the 20/25 (0.8) level, the BCVA changed 2 or more lines from the 3 to the 6 months' examination: 21 eyes (91%) improved and 2 (9%) deteriorated. The latter 2 had a decrease from 20/28 (0.7) to 20/50 (0.4) and 20/20 (1.0) to 20/28 (0.7) without an apparent cause.
From 143 patients with a preoperative BCVA of at least counting fingers, refractive data were available at all follow-up intervals (Table 4). Because the reliability of refraction may be compromised by low visual acuity, we performed the same analysis for the 115 eyes with a preoperative BCVA of 20/60 or better (≥0.3), but similar correlations were found (Table 5).
For the whole group (n = 143), the mean (SD) preoperative to postoperative change in SE (hyperopic and myopic shifts in corneal power averaged) was +0.49 (1.2) diopter (D) (P < .001) at 3 months and +0.38 (1.2) D (P = .001) at 6 months (Table 4). The mean (SD) preoperative to postoperative absolute change in SE (absolute change in corneal power) was 0.96 (0.8) D at 3 months and 0.90 (0.8) D at 6 months (Table 4). For the group of 115 eyes with a preoperative BCVA of 20/60 or better (≥0.3), the data are presented in Table 5.
For the whole group (n = 143), the mean (SD) preoperative to postoperative change in refractive cylinder (hyperopic and myopic shifts in cylindrical power averaged) was −0.36 (1.2) D (P = .002) at 3 months and −0.29 (1.1) D (P = .006) at 6 months (Table 4). The mean (SD) preoperative to postoperative absolute change in refractive cylinder (absolute change in cylindrical power) was 0.93 (0.8) D at 3 months and 0.91 (0.7) D at 6 months (Table 4). For the group of 115 eyes with a preoperative BCVA of 20/60 or better (≥0.3), the data are presented in Table 5.
Refractive stability was analyzed by comparing preoperative with postoperative refraction, as well as 3-month with 6-month postoperative refractions (n = 143). The change in SE before and at 3 months after surgery was 0.5 D or less in 38% and 1.0 D or less in 62% of eyes and at 6 months, 0.5 D or less in 38% and 1.0 D or less in 69% (Table 6). The change in cylindrical error before and at 3 months after surgery was 1.0 D or less in 66% and 2.0 D or less in 94% of eyes and at 6 months, 1.0 D or less in 64% and 2.0 D or less in 95% (Table 6).
From 3 to 6 months, no significant change in SE (P = .09) or cylindrical error (P = .35) was found (Table 4). Patients were fitted with glasses if desired at 3 months or continued wearing their preoperative glasses. All patients obtained full binocular vision, except for 1 case (monoculus).
Using topographic maps (n = 150), the change in true net power mean (SD) keratometric values before surgery and at 3 months was −1.4 (0.7) D and at 6 months, −1.2 (0.7) D (P < .001) (Table 7). The change in anterior corneal curvature before surgery and at 3 months was −0.5 (0.4) D and at six months, −0.3 (0.4) D (both P < .001) (Figure 2) (Table 7), and the change in posterior corneal curvature before surgery and at 3 months was +0.8 (0.4) D and at 6 months, +0.7 (0.4) D (P < .001) (Figure 2) (Table 7).
Although there was a change from before surgery to 3 months postoperatively, no significant difference in topographic astigmatism was found at the 6-month interval (P = .40) (Table 7).
Central corneal thickness decreased from mean (SD) 675 (94) μm before surgery to 526 (46) μm at 3 months and 527 (50) μm at 6 months (n = 150), ie, a decrease of 149 (48) μm and 148 (44) μm, respectively (P < .001) (Figure 2) (Table 8).
A significant negative correlation was found between the preoperative and 6-month central corneal thickness and SE values (n = 126) (r2 = 0.032; P = .047), which represents a small effect (Figure 3).
Of the DMEK eyes with an attached Descemet graft, preoperative and postoperative ECD measurements were available in 173 eyes. Mean (SD) donor ECD was 2560 (186) cells/mm2 preoperatively and 1690 (520) cells/mm2 at 6 months after surgery (P < .001) (Table 9).
In all eyes in which graft attachment was obtained, the transplanted cornea cleared within 1 to 12 weeks (Figure 4A and B). No clearance was seen in the presence of a complete graft detachment, eg, a Descemet roll in the recipient anterior chamber (Figure 4C and D), although “spontaneous clearance” despite graft detachment was seen in 4 eyes (Figure 4E and F) (Table 2).
Failure to unfold the Descemet graft in the anterior chamber occurred in 1 case (0.5%) and vitreous pressure was present in 15 cases (7.5%) (Table 10).
In 4 patients with pseudophakic eyes (2%), an intraoperative hemorrhage, originating from the iris root, was caused by traction on the peripupillary iris while positioning an air cannula underneath the unfolded Descemet graft prior to lifting the graft toward the recipient posterior cornea (Table 10).
None of the transplanted corneas failed to clear in the presence of an attached Descemet graft; in other words, a failure of the transplanted cornea to clear was only seen in the presence of graft detachment. One eye showed a secondary graft failure at 10 months after surgery (Table 10).
Graft detachment was defined as a lack of adherence between the Descemet graft and the recipient posterior stroma (frequently seen as a free-floating Descemet roll in the recipient anterior chamber within the first postoperative week) or a partial lack of adherence requiring secondary surgical intervention (rebubbling or regraft). Overall, 9% (18 of 200) of eyes showed a graft detachment, 4% (8 of 200) a complete and 5% (10 of 200) a partial detachment (Figure 4C and D) (Table 10). Small peripheral detachments occurred in 7% (14 of 200).
Twelve eyes (6%) had recipient DM remnants at the donor-to-host interface after surgery (Figure 5).17 Three eyes (1.5%) of 3 patients developed an allograft rejection. One patient noticed discomfort and a reduced visual acuity, but no subjective complaints were experienced by the other patients, who both had discontinued their steroid medication prematurely (Table 10).
At 3 months after DMEK, 1 eye (0.5%) presented with a small peripheral infiltrate in an area with remnant peripheral corneal edema that resolved with topical antibiotics (Table 10).
Because a 50% air fill of the anterior chamber was maintained at the end of the surgery, the operated-on DMEK eyes were considered at risk to develop pupillary block glaucoma. Hence, a YAG laser iridotomy was made at the 12-o’clock position prior to surgery. Although no true pupillary block glaucoma was observed, secondary glaucoma due to air-bubble dislocation behind the iris and/or mechanical forward displacement of the iris diaphragm occurred in 2% (4 of 200) of eyes (Table 10).
An additional 4 eyes developed another type of secondary glaucoma after DMEK surgery (2% [4 of 200]) (Table 10). One patient with bilateral DMEK developed a steroid-induced glaucoma in both eyes within the first postoperative month. One case with preexisting open-angle glaucoma developed 40–mm Hg spikes after surgery that required additional topical antiglaucoma medication. One eye developed peripheral anterior synechiae resulting in recurrent intraocular pressure elevations, eventually necessitating glaucoma surgery. One eye that had undergone phacoemulsification, penetrating keratoplasty (PK), and vitrectomies showed hypotonia for several weeks after the DMEK (Table 10).
In phakic eyes, mild anterior crystalline lens opacities were sometimes observed after DMEK and usually faded within months. However, in 2 of 33 phakic eyes (6%), the induced lens opacities required phacoemulsification (Table 10).18
Cystoid macular edema developed in 1 patient (0.5% [1 of 200]) after creating a YAG laser iridotomy prior to DMEK (Table 10). One high-myopic eye (0.5% [1 of 200]) presented with a retinal detachment at 2 months after surgery requiring vitrectomy (Table 10). One eye developed a macular hole (0.5% [1 of 200]) and 2 eyes, a macular pucker (1% [2 of 200]) within the first months after surgery (Table 10).
A rebubbling procedure was performed in 7 eyes (3.5%), 12 eyes (6%) underwent a second DSEK, and 5 eyes (2.5%) underwent a second DMEK (Figure 6) (Table 10). All of the secondary DSEK and DMEK procedures were successful, and visual outcomes were similar to primary DSEK or DMEK procedures.19
To our knowledge, this is the first prospective study on the efficacy of DMEK, ie, an analysis of its outcomes in a large patient series, unbiased by a learning-curve effect. In past years, the donor preparation protocol was validated,13 and the surgical technique could be standardized as a completely “no-touch” technique.12 Hence, the present data may allow a new baseline for clinical outcome in keratoplasty surgery.
About 80% of cases reached a BCVA of 20/25 or better (≥0.8) at 6 months after DMEK, with about 50% reaching 20/20 or better (≥1.0). These visual outcomes may compare favorably with any earlier keratoplasty technique. Historic studies on PK for Fuchs endothelial dystrophy reported a visual outcome of 20/40 or better (≥0.5) at 1 year in 40% to 50% of patients.20 Descemet stripping endothelial keratoplasty/DSAEK may surpass PK, with visual acuities up to 20/40 (0.5) in most cases but with only small percentages reaching 20/25 or better (≥0.8).1- 5,21 Clinical observation suggests that (cultured) donor posterior stroma in DSEK/DSAEK grafts degrades the optical quality of a transplanted cornea.2,22,23
Furthermore, the rate of visual rehabilitation may also be faster after DMEK, with most patients reaching their maximal visual potential in 1 to 3 months,7 compared with 6 to 12 months following DSEK/DSAEK and PK.3- 5,20,24,25
After DMEK, both the SE and the cylindrical error were within 1.0 D from the preoperative refractive error. Pachymetry and refractive data suggested that the transplanted cornea stabilizes approximately 3 months after DMEK, so new glasses could commonly be prescribed at this point, until which most patients were able to continue wearing their “own” glasses.
Unexpectedly, our study revealed about a +0.4-D hyperopic shift after DMEK. In DSEK/DSAEK, a +1.5-D shift may result from the “negative-lenticle” effect of the stroma carried by the endothelial transplant (being thinner centrally than at its peripheral flange). Since only an isolated donor DM is transplanted in DMEK, the hyperopic shift in DMEK cannot be explained by the same mechanism. The correlation found between the SE and the decrease in central corneal thickness (Figure 3) may indicate that the refractive shift in DMEK results from the preoperative to postoperative difference in recipient corneal hydration and the associated posterior corneal curvature change.8,26 The change in anterior corneal power is only about −0.3 D and falls within the margin of error in intraocular lens power calculation, so established nomograms may be used to calculate the intraocular lens power for cataract surgery at any time prior to DMEK.26
In the first 6 months after DMEK surgery, the donor ECD decreased about 34% compared with the preoperative counts, similar to the 31% to 34% decrease in ECD after DSEK/DSAEK.27- 29 Longer-term follow-up may reveal how the decline in ECD after DMEK compares with PK and DSEK/DSAEK. Since DMEK graft diameters (9.0-10.0 mm) exceed those in PK (7.0-8.0 mm) and DSEK/DSAEK (8.0-9.0 mm), more endothelium is transplanted, potentially providing longer graft survival in DMEK.
(Partial) graft detachment, the most frequent complication in endothelial keratoplasty,2- 4,11,30 occurred in 18 eyes (9%): 13 in the first 100 cases and 5 in the second 100 cases (Table 10) (Figure 6), despite general precautions, such as a 60-minute air fill of the anterior chamber at the end of the surgery to support the graft and avoiding the use of plastic and/or viscoelastic materials.10 Three additional risk factors were identified: intraoperative vitreous pressure, improper graft positioning, and postoperative ocular hypotonia. Hence, it may be advocated to obtain a soft eye before surgery.12 Inward folds (causing partial detachments by the graft “springing away” from the recipient stroma) may be managed by “bubble bumping,” ie, applying intermittent pressure on the corneal surface to completely unfold the graft.12 Upside-down positioning may be avoided by inserting the Descemet graft as a “double roll” and checking its upward orientation.12 Eyes at risk of developing postoperative hypotonia (eyes with aphakia, sector iridectomy, shallow anterior chamber, glaucoma shunt tube, or preceding posterior segment surgery) may be managed with a modified surgical technique.12,30
Secondary glaucoma due to air bubble misdirection may be avoided by leaving a 20% to 30% (instead of 50%) air bubble in phakic eyes.30 Performing the descemetorhexis “under air” may avoid the risk of remnant recipient DM fragments at the donor-to-host interface, potentially interfering with the optical performance of the transplanted cornea.17 All other complications may have been coincidental.
In conclusion, standardized “no-touch” DMEK may provide complete visual rehabilitation in a far majority of eyes, with a decrease in donor ECD similar to earlier keratoplasty techniques and with 5% to 9% (partial) graft detachment as the most frequent complication.
Correspondence: Gerrit R. J. Melles, MD, PhD, Netherlands Institute for Innovative Ocular Surgery, Rotterdam, the Netherlands (email@example.com).
Submitted for Publication: February 22, 2011; final revision received April 15, 2011; accepted April 18, 2011.
Published Online: July 11, 2011. doi:10.1001/archophthalmol.2011.195
Financial Disclosure: Dr Melles is a consultant for DORC International/Dutch Ophthalmic USA.
Additional Information: A video of the surgical technique is available at www.niios.com.