Article Text

Download PDFPDF

Descemet membrane endothelial keratoplasty with a stromal rim (DMEK-S)
  1. Pavel Studeny1,2,3,
  2. Andrej Farkas1,
  3. Magdalena Vokrojova2,
  4. Petra Liskova3,
  5. Katerina Jirsova3,4
  1. 1Department of Ophthalmology, Sokolov Hospital, Sokolov, Czech Republic
  2. 2Department of Ophthalmology, Teaching Hospital Kralovske Vinohrady, Charles University, Prague, Czech Republic
  3. 3Laboratory of the Biology and Pathology of the Eye, Institute of Inherited Metabolic Disorders, General Teaching Hospital, Charles University, Prague, Czech Republic
  4. 4Ocular Tissue Bank, General Teaching Hospital, Prague, Czech Republic
  1. Correspondence to Dr Pavel Studeny, Department of Ophthalmology, Sokolov Hospital, 35601 Sokolov, Czech Republic; studenypavel{at}


Aim To describe a novel technique for the preparation and transplantation of posterior corneal lamellae consisting of endothelium and bare Descemet membrane with a stromal supporting rim.

Methods Posterior lamellar discs for Descemet membrane endothelial keratoplasty with a stromal rim (DMEK-S) were prepared manually immediately before surgery using the big bubble technique. The retrospective case series that underwent DMEK-S comprised 20 eyes of 18 patients with endothelial dysfunction. Best-corrected visual acuity (BCVA) and endothelial cell density (ECD) were measured preoperatively and 12–24 months after DMEK-S.

Results At the end of the follow-up, 10 out of 18 eyes achieved a BCVA of 1.0 or better, and 17 reached 0.5 or better. Primary graft failure occurred in two eyes. The average ECD at 1 year was 1608 (±503) cells/mm2, that is, a mean cell loss from preoperative values of 44%. Partial early postoperative graft detachment (12 of 20 eyes) was treated by injecting an air bubble into the anterior chamber in all cases. The loss of donor corneas during preparation decreased from approximately 10 to 5% as more experience was acquired with the procedure.

Conclusions Preliminary outcomes show that DMEK-S may be used in the treatment of corneal endothelial dysfunction. As this approach is entirely manual, and no special surgical equipment is needed, it has the potential to become widely adopted.

  • Posterior lamellar keratoplasty
  • Descemet membrane endothelial keratoplasty with a stromal rim
  • DMEK-S
  • cornea
  • vision
  • dystrophy
  • treatment of surgery
  • eye (tissue) banking

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.


Posterior lamellar keratoplasty (PLK) is becoming a widely used technique in the treatment of corneal endothelial failure.1–3

Two different types of lamellar donor discs are being used in PLK. The first type is composed of the posterior stroma, Descemet membrane (DM) and the endothelium.4–6 In such lamellae, the stroma works as a carrier for the 15–20 μm thick DM-endothelium layer. The recipient's diseased endothelium and DM are either removed together with the underlying stroma after a manual midstromal pocket dissection (deep lamellar endothelial keratoplasty (DLEK))7 8 or stripped using a custom-made scraper (Descemet stripping endothelial keratoplasty (DSEK)).9 10

In the second type of lamellae, no stromal tissue is present. The donor discs are prepared from the corneoscleral donor button by stripping a circular portion of DM, and the transplantation technique has been referred to as Descemet membrane endothelial keratoplasty (DMEK).11–13 The recipient corneal bed preparation is the same as in DSEK.9

In this study, we describe the preparation of a new type of lamella consisting of bare DM-endothelium with a supporting stromal rim as well as the transplantation technique used, which we refer to as Descemet membrane endothelial keratoplasty with a stromal rim (DMEK-S). The clinical outcome of 20 eyes followed for at least 12 months is also reported.

Patients and methods


The study was approved by the Ethics Committee of Karlovy Vary District Hospital and conformed to the Declaration of Helsinki.

Twenty eyes of 18 Czech patients (11 women and seven men) are reviewed in this retrospective non-randomised study. All surgeries, including the lamellae dissection, were performed by the same experienced surgeon (PS) between May 2006 and November 2007. Patients' ages at the time of surgery ranged from 53 to 82 years (67.2±10.8 years). Fourteen patients (16 eyes) suffered from Fuchs endothelial corneal dystrophy (FECD), including nine patients with FECD (11 eyes) who had prior cataract surgery, and four patients (four eyes) sustained endothelial trauma during intraocular lens implantation. Preoperatively, every patient had either bullae or microcystic epithelial oedema in at least part of the corneal surface. Eyes with a known history of ocular disorders causing visual impairment are not reported.

Fifteen eyes had DMEK-S surgery only, while five underwent DMEK-S and simultaneous cataract surgery. In all cases that did not show primary graft failure, the follow-up periods ranged from 12 to 24 months (mean 15.7±months) (table 1). During the time frame of this series, the surgeon performed 26 DMEK-S in total - ie, six eyes were excluded from analysis because of other known ocular pathology. The first 10 cases were consecutive in all but one eye operated on between case 1 and 2. Primary graft failure was defined as a permanent corneal oedema from the first postoperative day without the identification of a secondary cause.14

Table 1

Details of patient history, preoperative examination, donor and concomitant surgery performed with Descemet membrane endothelial keratoplasty with a stromal rim

Preparation of donor tissue

Donor corneoscleral discs 17 mm in diameter were obtained from three Czech tissue banks. The corneas were stored under hypothermic conditions in Optisol-GS (Bausch & Lomb Surgical, Rochester, New York) or Eusol-C (Alchimia SRL, Treviso, Italy) at 4°C or in tissue cultures in E-MEM (Minimum Essential Medium with Earle's salts) l-glutamine, 25 mM HEPES (AppliChem, Darmstadt, Germany) with 2% fetal bovine serum (Invitrogen/Gibco, Paisley, UK) at 31°C. The quantitative and qualitative parameters of the endothelium were assessed using an automatic or semiautomatic system, respectively (Konan, Eye Bank KeratoAnalyzer, Konan, Japan; Lucia analysis system, Laboratory Imaging, Prague, Czech Republic) (table 1).

Each lamella was prepared immediately before grafting in the operating room. Corneoscleral discs were mounted endothelial side up in a Barron artificial anterior chamber (Katena, Denville, New Jersey). The separation of DM and stroma was performed by the big bubble technique using a 1 cm3 insulin syringe (B. Braun, Melsungen, Germany).15 16 We placed the needle in the far periphery of the donor disc in order to minimise endothelial cell damage. The needle bevel was oriented towards the DM and endothelium.

Subsequently, the endothelium was covered with a thin coat of dispersive viscoelastic (Viscoat, Alcon, Fort Worth, Texas), and the corneoscleral disc was turned up. Approximately 80% of the upper thickness of the donor cornea was cut off using a crescent knife (Huco Vision SA, Saint-Blaise, Switzerland). The central round part, 6 mm in diameter, was marked by a Free Hand Corneal Trephine (PMS, Tuttlingen, Germany), and the letter S was written using a Skin Marker (Kendall, Mansfield, Massachusetts, USA) onto the stromal part of the peripheral lamellae. Next, the stoma in the central area was perforated by a cornea diamond knife (Huco) to reach the big bubble, then a cyclospatula (Jezek, Prague, Czech Republic), as an alternative to a grooved spatula, was inserted between the stroma and DM, and the stroma incised along its body. Subsequently, the rest of the stroma in the central 6.0 mm area was cut off using curved corneal scissors (Huco). Then, the whole 8.0 mm donor disc with the endothelial side up was trephinated by a Barron disposable punch (Katena). Preparation of the donor button typically took 20–30 min in total. The main steps of the lamellae preparation are shown in figure 1 and the supplementary video file.

Figure 1

Donor tissue preparation. A corneoscleral disc is mounted endothelial side up in a Barron artificial anterior chamber, and a big bubble is introduced to separate the Descemet membrane and stroma (A, B). The disc is turned over, and approximately 80% of the depth of the stromal tissue is removed (C, D). A circle 6 mm in diameter is marked in the centre, and the letter S is written on the stromal rim (E). The rest of the stroma is removed after entering the bubble (F).

Surgical technique

Pupillary dilatation was achieved using 0.5% tropicamide drops preoperatively. The patient was approached from the 12 o'clock position after subconjunctival anaesthesia with 0.1 ml of 1% lidocain combined with topical anaesthesia using several drops of 0.5% bupivacain. After incision of the conjunctiva, a 4.75 mm wide scleral tunnel incision extending 1.0 mm into the peripheral cornea was made with a crescent knife (Huco) followed by two 0.9 mm corneal paracenteses at 9 and 3 o'clock. The endothelium was stained with VisionBlue® (trypan blue ophthalmic solution) 0.06% (D.O.R.C. International, Zuidland, The Netherlands), and the anterior chamber was filled with cohesive viscoelastic (Provisc; Alcon). An 8.0 mm diameter trephination epithelial mark was made to outline the area of DM excision (Free Hand Corneal Trephine; PMS). Continuous curvilinear descemetorhexis following the contour of the superficial mark was performed using a hook inserted through the sclerocorneal tunnel. The hook was custom-made by bending upward the tip of a cyclospatula (Jezek). The peripheral part of the recipient bed, approximately 1 mm in width, was roughened by this hook. The cohesive viscoelastic and the entire membrane peeled off from the posterior cornea were removed by the Irrigation and Aspiration hand piece of a phaco device (Katena), and the anterior chamber was filled with Ringer solution (Fresenius Kabi, Verona, Italy).

The endothelium of the donor tissue was coated with dispersive viscoelastic (Alcon). The lamella was then overfolded 40:60 using the taco technique,17 grasped with a corneal forceps, placed onto a plastic glide (IOL Glide; BD Visitec, Sarasota, Florida) and inserted into the anterior chamber using this glide and a cyclospatula. Inside the anterior chamber, it was allowed to unfold spontaneously under irrigation from a cannula using the temporal side port. When necessary, the graft was centred by tapping on the corneal surface. Fixation of the lamella against the recipient posterior stroma was achieved by air injection to approximately three-quarters the depth of the anterior chamber through a temporal or nasal side entry. The air bubble had to surpass the edge of the lamella in a supine position, and it remained in place until its spontaneous resorption, commonly in 1–2 days. The conjunctiva was closed with two interrupted 8-0 Vicryl sutures (Ethicon, Somerville, New Jersey). At the end of the surgery, the patient was administered an antibiotic-corticosteroid drop (tobramycin and dexamethasone) and positioned in a supine position for several hours. The total time required for the whole DMEK-S procedure was similar to that required for other PLK techniques such as DSEK. The main steps of the surgery are shown in figure 2 and the supplementary video file.

Figure 2

Intraoperative photographs showing Descemet membrane endothelial keratoplasty with a stromal rim. A sclerocorneal tunnel is made at the 12 o'clock position (A). An 8.0 mm diameter centred threpination epithelial mark is made to outline the area of DM excision after staining the endothelium with Trypan Blue solution (B). Insertion of the donor tissue folded endothelial side inward into the anterior chamber using a plastic glide (C). Final position of the posterior lamellar graft; the outer stromal surface is marked with the letter S to allow correct orientation of the lamella (D).

Any simultaneous cataract surgery was performed before the Descemet stripping through the same sclerocorneal tunnel incision. In cases of severe corneal oedema, the epithelium was scraped off to obtain a clearer view; this was necessary in four eyes. After surgery, an occlusive plastic dressing was kept in place overnight, and the patients were instructed to lie on their backs. They were discharged 2–4 days postoperatively. In those with a removed epithelium bandage, contact lenses were worn until the cornea epithelialised.


Anterior segment examination and best-corrected visual acuity (BCVA) measurements (Snellen charts) were carried out at 1 week and at 1, 3, 6, 12, 18 and 24 months after the surgery. Graft detachment was treated on the 7th postoperative day by injecting an air bubble under microscopical control in the operating room. Approximately three-quarters of the anterior chamber was filled with air. Immediately after the injection, the patient was left in a supine position for an hour, then discharged and asked to remain in a supine position at home for the next several hours. If reattachment was unsuccessful, the procedure was repeated on day 14.

The combined administration of corticosteroid and antibiotic eye-drops (tobramycin and dexamethasone) was performed five times a day for the first month after the surgery. Steroid drops were then continued three times a day for the next 5 months. From the seventh month until 1 year postoperatively, non-steroidal anti-inflammatory drops (indomethacin) were applied three times a day.

The assessment of ECD was performed in the central cornea using a Topcon SP3000p non-contact autofocus specular microscope with Topcon IMAGEnet 2000 software (Topcon, Tokyo, Japan) or a contact endothelial camera (Heidelberg retina tomograph II Rostock corneal module; HRT II; Heidelberg Engineering, Heidelberg, Germany). A single reading was taken. With the Topcon, we used a ‘centre-dot’ method, that is, a machine-based algorithm in which at least 20 cells need to be marked. With the HRT II, the endothelial cell counting was also performed using built-in image-analysis software after manual marking of 60 cells as a minimum. Postoperative cell loss was calculated as a percentage of the preoperative value declared by eye banks.


The results of DMEK-S in 20 eyes of 18 patients are reported herein. Data collected at the postoperative examinations are shown in table 2. BCVA at the end of follow-up improved in 18 (90%) eyes. The remaining two eyes (the first cases that had been grafted) had primary graft failure with persisting corneal oedema from the first postoperative day until at least 3 months after DMEK-S, when a repeat corneal graft surgery was performed. One case underwent DLEK, whereas the second case had another DMEK-S (supplementary material). The results of the second graft surgery, including details of the donor graft, are shown in table 3. At final follow-up, all but one eye reached a BCVA better than 0.5 (including the two eyes that were reoperated), and 10 eyes had a BCVA of 1.0. A BCVA lower than 0.8 was attributed to opacification of the intraocular lens in case 13, suspected macular oedema and amblyopia in cases 4 and 15, respectively. One eye was lost to follow-up at 12 months because of the patient's death.

Table 2

Outcome of Descemet membrane endothelial keratoplasty with a stromal rim

Table 3

Outcome of repeat corneal graft surgery due to primary graft failure in two eyes that underwent Descemet membrane endothelial keratoplasty with a stromal rim

The preoperative ECD was 2888 (±265) cells/mm2 (n=18, ie, those eyes in which the ECD could be assessed after 12 months). The postoperative ECD at 1 year for these eyes ranged from 781 to 2381 and averaged 1608 (±503) cells/mm2, that is, an average decrease of 44% from the preoperative values.

No significant intraoperative complications such as suprachoroidal haemorrhage or expulsive haemorrhage were observed. Minor technical problems that were occasionally experienced included iris prolapse, problems with the initial centring of the donor disc, incorrect injection of the fixative air bubble between the lamella and the recipient cornea requiring gentle aspiration of the air, repositioning of the lamella and reinjection of the air bubble. All of these technical difficulties were dealt with appropriately during the surgery, and they did not result in any postoperative problems.

Postoperative complications in this series were only related to partial graft detachment, which was observed in 12 eyes and was successfully treated with air injection into the anterior chamber, most often in one session, and only three eyes required air to be applied repeatedly. Total dislocation (ie, mobile or resting on the iris) was never seen. No pupillary block was documented. There were no episodes of endothelial rejection in any of the cases. Intraocular pressure measurements were normal upon check-ups.


Herein, we have described a novel approach to the manual preparation and transplantation of a posterior lamellar disc consisting of two areas of different thickness. The central round part, 6.0 mm in diameter, is composed of the DM-endothelium layer lined by a peripheral ring about 100 μm thick and 1.0 mm wide that, in addition to the DM endothelium, also comprises the posterior stroma. The stromal rim fixes the thin, fragile central part, helping it to keep its shape and preventing the scrolling of DM. Such lamellar discs offer the following advantages: the central part provides the excellent optical features of the DM-endothelium layer, and the outer part makes the disc more stable for manipulation before and during the implantation. Moreover, the stromal part allows marking of the anterio-posterior orientation of the lamellae. This enables the surgeon to know exactly on which side the endothelium is when manipulating the lamella as well as its orientation in the anterior chamber, thus avoiding unnecessary primary graft failure as reported by Ham et al.18 Prior to the first case described in this study, the surgeon had been performing DLEK and DSEK surgeries on a routine basis.

The width of the tunnel incision used in DMEK-S is somewhat larger than in classical DMEK (4.75 vs 3.50 mm);19 20 however, as it is still sutureless, this approach maintains the advantages of reducing the number of postoperative follow-ups and the lack of suture-related complications.

A disadvantage of our method is the rate of loss of donor corneas. Initially, in approximately the first 100 cases, about 10% of donor corneas were damaged during preparation. This rate has improved after gaining experience to about a 5% loss attributable mainly to bubble rupture.

It has been shown that the thinner the stromal tissue on the graft, the better the visual outcome,21 22 which is in accordance with the BCVA results obtained in this study where no stromal tissue was present in the visual axis. At the end of the follow-up period (12–24 months), 55.6% (10 of 18) of eyes achieved a BCVA of 1.0 or better and 99.4% (17 of 18) 0.5 or better. The worst BCVA obtained was 0.4, including eyes with repeat surgery. These data indicate that the minimum period before drawing any conclusions regarding final BCVA following DMEK-S should be at least 1 year. We find our BCVA outcome to be comparable with the results of Melles et al19 and Ham et al20–23 using DMEK and better than those reported by Tappin.12

The postoperative ECD in our study showed great variation with a mean of 1608 (±503) cells/mm2 1 year postoperatively, that is a 44% decrease. Three of the grafts had an ECD lower than 1000 at the final evaluation, but none of the eyes developed signs of endothelial decompensation. Our ECD values obtained 1 year postoperatively appear to be comparable with those reported in other studies. Ham et al20 reported an average ECD of 2623 before and 1815 cells/mm2 6 months after DMEK in 40 patients.

We expect that the usage of precut discs for DMEK-S with an ECD higher than 2500 cells/mm2 will lead to further improvement of the postoperative results and shortening of the time spent in the operating theatre.

The most frequently encountered complication of DMEK is a persistent complete or partial detachment of the lamellae leading to graft failure.19 20 Although our early postoperative partial dislocation rate was quite high (60%) it was fully manageable with repeated air injection into the anterior chamber. Treatment of graft detachment on day 7 was chosen based on our previous unpublished experiences when dealing with complications in other PLK methods. We hypothesise that later management of graft detachment may be more successful due to the complete reparation of the endothelial lesions induced at surgery and finished by day 5,24 complemented by the complete absorbance of possible viscoelastic residuals from the anterior chamber. Two patients suffering from primary graft failure were subsequently treated with another DMEK-S or DLEK, respectively, with good results. As both of the patients were the earliest cases treated with DMEK-S, the failures have been attributed to a learning curve.

Our modified technique for PLK appears to be surgically easier than DMEK, mainly because it avoids transplanting a rolled DM and enables easy orientation of the position of the endothelium. Moreover, neither the lamellae preparation nor the actual surgery requires any special instrumentation that would limit its widespread use. We report here the outcome of DMEK-S in 20 eyes, the largest case series with follow-up longer than 1 year using DM-endothelium transplants to date. We show that DMEK-S provides excellent optical features and fast visual rehabilitation, and that it has the potential to become widely adopted.



  • Funding This work was supported by the Czech Ministry of Education, Youth and Sports research project 0021620806/20610011.

  • Competing interests None.

  • Ethics approval The study was approved by the Ethics Committee of Karlovy Vary District Hospital.

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

Linked Articles

  • At a glance
    Harminder S Dua Arun D Singh