Article Text

Download PDFPDF

Descemet's membrane endothelial keratoplasty (DMEK): first UK prospective study of 1-year visual outcomes, graft survival and endothelial cell count
  1. Robert A Reid,
  2. Ewan A Craig,
  3. Hanif Suleman
  1. Department of Ophthalmology, Shrewsbury and Telford Hospitals NHS Trust, Shrewsbury, UK
  1. Correspondence to Robert A Reid, Department of Ophthalmology, Shrewsbury and Telford Hospitals NHS Trust, Mytton Oak Road, Shrewsbury SY3 8XQ, UK; rareid67{at}


Aim To evaluate the clinical outcomes of Descemet's membrane endothelial keratoplasty (DMEK) in the treatment of patients with Fuchs’ endothelial dystrophy.

Methods This prospective study involved 16 consecutive patients who had DMEK done and who were evaluated over a year. Measurements included best corrected visual acuity (BCVA), endothelial cell count, and central corneal thickness.

Results Two patients had failed grafts. Of the remaining 14 patients with successful grafts, after 12 months, 79% had BCVA of 6/6 or better, and all patients had a BCVA of 6/9 or better. Median endothelial cell count was 1567 cells per mm2 (range=900–2359) representing a 40% reduction compared with preoperative counts. Median central corneal thickness was 498 µm (range 445–567 µm) compared with a median of 649 µm (range 548–740 µm) preoperatively. All patients attained total visual rehabilitation without further surgical intervention.

Conclusions In our experience, DMEK has the potential to become the primary procedure for treating Fuchs’ endothelial dystrophy and endothelial disease as it produces rapid total visual rehabilitation with few complications, and an easy follow-up and management regimen. Donor preparation and graft insertion, however, remain important challenges.

  • Cornea
  • Dystrophy
  • Anterior chamber
  • Vision
  • Clinical Trial

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


In the past, penetrating keratoplasty (PK), Descemet's stripping endothelial keratoplasty (DSEK), and Descemet's stripping automated endothelial keratoplasty (DSAEK) have been used to address endothelial disease.1 ,2

PK involves removal of all corneal layers from the patient and replacing it with all corneal layers from a donor. It is associated with a significant number of complications, including graft rejection, infection, irregular astigmatism, and graft failure, among others. It also requires numerous clinic visits, ongoing active suture management. Its follow-up of patients is clinic-intense and resource-intense.3

DSEK attempts to address this problem by grafting the endothelium, Descemet's membrane and inner stromal layers of the cornea, thus reducing the amount of tissue in the graft and reducing the complications associated with PK. However, DSEK is often compromised by deposits, opacities and blurring in the interface between the donor sclera and the recipient sclera.4

DSAEK seeks to reduce this complication by microkeratome dissection, which generally produces a smoother plane, is easier than manual dissection and can be done by eye-bank technicians.

Melles et al introduced the technique of Descemet's membrane endothelial keratoplasty (DMEK) as an alternative to PK, DSEK and DSAEK as a surgical procedure for visual rehabilitation of patients suffering from endothelial dystrophy.3 ,4 Price et al also published data on their experience with DMEK and modifications to Melles’ technique.2 ,5 ,6

This report describes our experience with DMEK at the Royal Shrewsbury Hospital including the 1-year outcomes of 16 patients with Fuch's endothelial dystrophy (FED).

Materials and methods


Sixteen consecutive patients with FED who underwent DMEK are included in this report. The clinical data collected preoperatively included patient age, gender, affected eye, best corrective visual acuity (BCVA) as measured in our afternoon clinics, donor endothelial cell count, central corneal thickness (CCT), donor graft size and their main clinical diagnosis. The demographics of these patients and some clinical characteristics are shown in table 1. Preoperative BCVA, endothelial cell count, and CCT shall be presented along with their results.

Table 1

Patient demographics

DMEK was performed by a single surgeon (EAC) between September 2008 and November 2010 using a modification of the technique developed by Melles et al.7 During that period, the surgeon performed a total of 22 DMEK procedures; six were excluded from this analysis because they had other preoperative ocular pathology causing visual impairment. The excluded patients had the following in the operated eye:

  • Previous central retinal vein occlusion and choroidal neovascular membrane.

  • Irido corneal endothelial syndrome, amblyopia, corneal decompensation with recurrent corneal erosion.

  • Eye trauma with iridodialysis and macular hole.

  • A failed PK.

  • Congenital glaucoma, buphthalmos with corneal decompensation.

  • Buphthalmos, exotropia, retinal detachment repair, corneal decompensation.

In this series, DMEK was done on one phakic patient with a clear lens, five were already pseudophakic at presentation, and 16 had combined DMEK/phaco (bimanual micro incision cataract surgery).

Preparation of graft

Donor grafts were obtained from the Corneal Transplant Service Eye Bank and delivered to the operating theatre in dextran medium on the day before surgery. One hour before the operation, the corneoscleral rim of the donor was excised using scissors. The corneal donor was then fixed to a holder. An 8.0–9.0 mm disc of Descemet's membrane was excised using a trephine. A button of Descemet's membrane was harvested by gently detaching the donor endothelium attached to Descemet's membrane starting from the periphery and proceeding toward the centre of the donor cornea. The resulting roll of Descemet's membrane was stored in normal saline until the patient's receptor cornea was prepared.1 ,3 ,7

Preparation of recipient

Under general anaesthesia, one 2.85 mm self-sealing clear corneal incision and four paracenteses were made into the host cornea. A cystotome was used to make a small incision down to the plane between Descemet's membrane and stroma. A reverse circular incision like a reverse capsulorrhexis was made. This descemetorrhexis was sized to the same diameter as that of the donor graft. An Iris repositor and air bubble were used to detach the Descemet's membrane and endothelium. The resulting detached tissue was removed using forceps.

Descemet's membrane implantation

The Descemet's membrane roll was placed in a galley pot with Vision Blue. The Descemet's membrane roll was then aspirated into a round-ended glass pipette. The roll was injected into the anterior chamber, and then unrolled. It was then aligned to the base stroma of the recipient so that the donor Descemet's membrane was in contact with the recipient stroma, with the endothelial cells orientated towards the anterior chamber. Air tamponade was used to hold the graft in place for 5–7 min. Intracameral Cefuroxime was injected and the eye was padded.

Follow-up regime

Postoperative follow-up was done using the following schedule: patients were seen on the day of surgery, after their operation. They were then seen 1 day, 1 week, 1 month, 3 month postoperatively, and every 3 months thereafter. If a complication developed, the follow-up schedule was modified as dictated by the nature and severity of the problem.

Snellen BCVA, as measured in our afternoon clinics, CCT, endothelial cell count, simulated K values (SIM K), and complications were recorded at each visit. Any other relevant findings were also noted.

Drug regime

The patients were started on Dexamethasone 0.1% preservative-free drops, one drop eight times per day for the first week. This was reduced to four times per day for the next 3 months, then reduced by one drop per day at each subsequent 3-month visit. This regimen was modified if the patient developed either a reaction to the drops or developed elevated intraocular pressure (IOP).


The results of DMEK in 16 patients over a 12-month period are reported.

Visual acuity

After 1 year, 14 patients (88%) who had a successful DMEK procedure experienced improvement in their BCVA and sustained a clear corneal graft that was attached and stable (table 2).

Table 2

Preoperative and postoperative Snellen BCVA

Clear, stable vision was achieved without further surgical intervention, astigmatic correction or contact lenses. A simple steroid regimen was followed, and in the case of two patients Yag Capsulotomy was required.

Graft attachment and clearing

In two patients, the DMEK failed due to total graft detachment. In one case, a repeat DMEK was successfully done after 3 months; the repeat graft was completely attached from the outset and the patient continues to be monitored. In the other, a second attempt at DMEK after 8 months also resulted in detachment, so a DSEK graft procedure was done successfully thereafter.

In both cases, the graft was correctly oriented. In one patient, the graft was very difficult to unroll once harvested and seemed to have reverted to a rolled position soon after surgery. In the second, the donor tissue was noted to be ‘friable and very floppy’. Both were found detached in the anterior chamber.

Seven patients (50%) had a portion of the graft detached after 1 week, estimated between 5% and 50%, with a median of 10% detachment. At 12 months, just under half the partial detachments noted in the first week had settled and were completely attached. Four patients had a portion of the graft still detached at 12 months, estimates ranging from 1% to 10%, with a median of 5% detachment. This partial detachment appeared not to affect graft viability or visual rehabilitation.

Only one patient needed additional air tamponade, as 60% detachment was noted after 1 month. This completely resolved the detachment, and at 12 months, the graft was still completely attached and clear.

All subsequent results are based on the 14 patients with successful DMEK procedures.

Donor endothelial cell count and central corneal thickness

There was a 40% reduction in endothelial cell count and a 23% reduction in CCT between the preoperative measurement and the 12-month follow-up visit (table 3).

Table 3

Preoperative and postoperative, cell count, and CCT

Corneal oedema

Nine patients (64%) had postoperative corneal oedema. Eight of these patients had preoperative oedema; one seemed to have developed oedema as a result of the DMEK procedure, and her oedema cleared by the 1-month visit. In eight of the nine patients, the oedema had resolved completely by 1 month postoperatively. For the remaining patient (who had oedema preoperatively) the oedema cleared after 9 months. Two patients had central corneal oedema; seven had peripheral oedema. Five patients had anatomically clear corneas with no oedema from the first postoperative day.

Elevated intraocular pressure

Seven patients (50%) had elevated IOP. These seven patients fell into two groups. The first group had high IOP immediately postoperatively due to air tamponade; this consisted of two patients whose IOP ranged from 40 mm Hg to 60 mm Hg. This was corrected by releasing the anterior chamber air bubble. No IOP problems developed thereafter. The second group of five patients had IOP ranging from 21 mm Hg to 30 mm Hg. The elevated IOP was first observed after 3–9 months. IOP was lowered by switching them from Dexamethasone drops to Rimexolone 1% drops, with no need for further antihypertensive treatment.

Postoperative inflammation

One patient had postoperative inflammation with graft oedema and fine Keratic precipitates (KP) on the graft after 1 week. These may have been early signs of rejection. This potential rejection episode was dealt with by increasing the frequency of Dexamethasone from eight times per day to hourly applications. The patient was reviewed twice per week until the cornea cleared at 1 month postoperative.

Graft folds, rolls and wrinkles

Eight patients (57%) had peripheral folds or rolled or wrinkled graft edges. This involved less than 5% of the graft area. They remained stable and unchanged during the course of the study and appeared not to affect graft stability or visual outcome.

Bare area

Four patients had a bare area of corneal stroma which was not covered by the corneal graft. However, these areas were completely clear with no associated corneal oedema.

Sim K

There appeared to be no notable change in patients’ corneal astigmatism as measured by Sim K. Any variation in corneal astigmatism did not affect patients’ BCVA.


This is the first UK prospective study with 1-year follow-up results of a cohort of patients. DMEK resulted in improved visual outcomes, quick visual recovery usually within 3 months postoperatively, and few serious complications. This compares favourably with others’ results.2 ,4 ,6 ,811 Some patients may develop small graft detachments, but these appear to have little effect on graft stability and visual outcome, which concurs with others’ findings.2 ,6 ,8 ,12 Clear, stable vision was achieved with a simple treatment and follow-up regimen, with interventions consisting primarily of modifying the type and frequency of drops. Three patients had less than 6/6 BCVA at 12 months. Of these, two had central corneal oedema and one had graft oedema and KPs immediately after surgery. In all three cases, the oedema had cleared by month 9 and the resulting BCVA was 6/9.

The extent of the improvement to BCVA by DMEK is an underestimate, as preoperative BCVA was consistently measured in afternoon clinics, when most patients reported improvement in their sight relative to the severe blurring they experienced in the mornings before they had surgery. It also does not take into account the improved quality of sight most patients reported, as they now had stable, clear vision all day.

Endothelial cell survival is one of the crucial determining factors of the health and viability of donor tissue in the long term. In our study, endothelial cell loss at 1 year was calculated by comparing the endothelial cell count of the donor (as recorded by the Corneal Transplant Service Eye Bank using their Leica light microscope) with the endothelial cell count of the graft as measured in vivo postoperatively by our non-contact specular microscope.

Potential inaccuracies exist based on the way in which we determined endothelial cell loss. The endothelial cell count provided by the Corneal Transplant Service Eye Bank at the time of harvesting the corneas was used as our estimate of preoperative cell count. This may overestimate the actual cell count of the implanted donor tissue as it does not take into account cell loss due to the process of harvesting the cornea, storage of the cornea in dextran medium instead of organ culture, as suggested by Melles,1 transportation of the tissue, the process of harvesting the endothelium, and the process of insertion, orientation, and implantation of the graft. Furthermore, our specular microscope was unable to deliver accurate readings in conditions such as irregular corneal surface, and mild corneal oedema or dryness.

Our measure of 40% endothelial cell loss is higher than the findings reported by Bachmann et al (37% cell loss at 1 month),13 Melles et al (34% cell loss at 12 months),10 Price et al (32%±20% cell loss at 3 months),2 and Ham et al (24% cell loss at 12 months).14

In our study, the cell loss is within the range reported by Lee et al for DSEK.15 It should be noted, however, that these results are from the initial cases during the surgeon's (EAC) early learning curve.

Some patients developed elevated IOP between 3 and 9 months postoperatively, possibly due to the use of intensive steroids. Having corrected this, when detected, by switching from Dexamethasone to Rimexolone, we could consider routinely taking this approach for all patients after the first month, to avoid this complication.

Other complications such as corneal oedema occurred at the periphery of the graft and, therefore, did not affect visual outcome and rehabilitation. Bare areas and crescents at the edge of the graft also seemed not to affect central clarity, visual outcome, or graft viability, and were not associated with corneal oedema. This could possibly be due to seeding or migration of endothelial cells to cover these bare areas.16

Typically, the current practice is that DSEK is used as a back-up procedure in the event of graft failure with DMEK.1 ,17 However, in one of the two patients with graft detachment, a repeat DMEK procedure was successfully done, as was also done by Price et al.2

At present, DMEK is a relatively new procedure, performed by a small but growing number of practitioners, proving its viability compared with well-established procedures being practised by a large number of practitioners. Many of these practitioners may consider the outcomes of DSEK and PK to be ‘good enough’ in light of the steep learning curve associated with DMEK. Our study will contribute to a body of work that provides information on DMEK as a viable first-line option for treating FED. It will also provide data to assist in counselling patients on the type and nature of their surgery, expected visual outcomes, visual rehabilitation, complication rates, and follow-up and management regimens.

In conclusion, in our experience, DMEK is a safe procedure for treating patients with primary endothelial disease. It delivers good visual outcomes, corneal clarity, and complete visual recovery, with few interventions, a simpler management regimen, and few complications. We propose that DMEK becomes the primary surgical procedure for treating endothelial disease where appropriate.


We would like to acknowledge the assistance of administrative staff including Annette Owen, Caroline Hale and Sian Meagre as well as the ophthalmology technicians and nursing staff.



  • Collaborators Chris James Wenham ensured that the appropriate tests were done for each patient so that appropriate data would be available for each patient.

  • Contributors EAC and RAR planned and designed the prospective study. EAC conducted all DMEK operations on the patients. RAR designed the data sheets, collected the data, and analysed the data with the assistance of HS. RAR wrote the first several drafts of the article, which was then reviewed and amended by both EAC and HS. EAC is the guarantor of this paper.

  • Competing interests None.

  • Ethics approval Shrewsbury and Telford Hospitals NHS Trust Local Ethics Committee, Research and Development Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • At a glance
    Keith Barton James Chodosh Jost Jonas