Article Text

Download PDFPDF

Original article
Endothelial keratoplasty without Descemet's stripping in eyes with previous penetrating corneal transplants
  1. Jennifer M Nottage1,2,
  2. Verinder S Nirankari1,3
  1. 1Department of Ophthalmology, University of Maryland, Maryland, USA
  2. 2Cornea Service, Wills Eye Institute, Philadelphia, Pennsylvania, USA
  3. 3Cornea Service, Eye Consultants of Maryland, Owings Mills, Maryland, USA
  1. Correspondence to Dr Verinder S Nirankari, Cornea Service, Eye Consultants of Maryland, 21 Crossroads Drive, Suite 425, Owings Mills, MD 21117, USA; vnirankari{at}


Aims To evaluate techniques, outcomes and complications of endothelial keratoplasty (EK) without Descemet's stripping in eyes with previous penetrating keratoplasty (PK).

Methods A retrospective, consecutive analysis of patients who underwent EK in eyes with previous PK.

Results 33 eyes of 31 consecutive patients underwent EK without Descemet's stripping in eyes with previous PK. All 33 eyes were followed for an average of 14.2±4.4 months. The reason for EK was either graft rejection (n=9) or endothelial failure (n=24). 32/33 (97.0%) eyes remained clear at the last follow-up visit. Visual acuity improved in 28/33 (84.8%) patients with 19/28 (67.9%) of these patients having more than three lines of visual acuity improvement. The average postoperative vision in this group was 20/60 (range 20/25–20/100). Pre-cut donor tissue was used for all cases with an average thickness of 127 μm. Endothelial cell density declined by 35% at 3 months and was stable at 1 year. 2/33 (6%) eyes had graft dislocations. One eye was successfully treated by a re-bubble. The other case failed EK revision and required a repeat PK. No patients developed EK or PK graft failure/rejection.

Conclusion EK without Descemet's membrane stripping in eyes with previous PK graft failure/rejection is a valuable alternative to a repeat full thickness PK.

  • Cornea
  • treatment surgery

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.


The Eye Bank Association of America (EBAA) statistical report on Eye Banking Activity for 2008 reported that there were 41 652 grafts transplanted in the USA.1 Re-grafts usually constitute between 10% and 40% of the grafts performed each year.2 3 Previously, all patients received repeat penetrating keratoplasty (PK) for treatment of endothelial failure or graft rejection. With recent advances in lamellar corneal surgery, the option of endothelial keratoplasty (EK) can now be considered for post-PK patients with failed grafts if they had a previously clear graft with a stable refraction.4–9

EK has been studied extensively and is comparable to the gold standard: PK. EK allows the surgeon to replace only the damaged tissue while maintaining the structural integrity of the eye. EK has been proven to be a safe procedure that offers a rapid visual recovery, with minimal induced astigmatism and mild predictable hyperopic shift in refraction postoperatively.10–12 The risk of expulsive intra-operative suprachoroidal haemorrhage is minimised because a 5 mm sclera tunnel incision can be closed quickly.13 Furthermore, the best corrected visual acuity can usually be obtained with glasses, whereas PK that may require rigid gas-permeable contact lenses for best vision. Because of the sutureless nature of the procedure, fewer postoperative visits are required. Transplant rejection is less frequent and may be less severe after EK than PK.14 EK is more technically challenging and the success rate is dependent on the surgeon's experience level.11 15

Descemet's stripping can be more challenging in patients who have had PK. There is a high risk of opening the graft–host junction during the stripping process. If patients do not have guttata and the Descemet's membrane (DM) is clear before the corneal decompensation, one can consider non-Descemet's stripping endothelial keratoplasty (non-DSEK). Non-DSEK has been performed as a primary therapy for endothelial dysfunction and has shown favourable results comparable to DSEK.16 17

In a previous study by Price and Price7 of EK after failed PK, 5/7 cases did not receive DM stripping with favourable results. In a study by Covert and Koenig,8 DM stripping was performed on all seven cases and there were three graft dislocation and two patients with iatrogenic graft failure. Furthermore, Terry and colleagues9 reported 17 cases with DM stripping. They reported a 5.9% dislocation rate without any cases of PK wound dehiscence. Because of the risk of graft–host interface iatrogenic dehiscence and clear DM before corneal decompensation, DM was not stripped in the present study. The purpose of the present study was to evaluate techniques, outcomes and complications of EK without Descemet's stripping in eyes with previous PK.

Materials and methods

A list of all patients who underwent DSEK after failed PK between November 2008 and April 2010 at a single referral-based institution were compiled. Outcomes from 33 consecutive cases performed by a single surgeon (VSN) were analysed retrospectively to assess the rate of complication and types of complications. The surgeon has performed PK for 25 years and DSEK for 5 years. The medical records were reviewed for demographics, baseline preoperative best spectacle-corrected visual acuity, postoperative best spectacle-corrected visual acuity, date and indication for PK, date and indication for DSEK, surgical technique, intra-operative/postoperative complications, subsequent surgical procedures and date of last follow-up. Average PK graft size with the initial surgery was 7.5 mm. The initial PK indications included Fuchs' dystrophy, corneal scar, bullous keratopathy, keratoconus and corneal dystrophy. Indications for EK included graft failure or graft rejection. Additional procedures, such as vitrectomy or synechialysis, may have been performed at the time of the EK procedure.

Vision was measured using the Snellen chart and converted to logarithm of the minimum angle of resolution (LogMAR) equivalents. No patient had greater than 3 D of keratometric astigmatism or required a contact lens for best corrected vision prior to graft oedema. The visual recovery, graft survival, intraoperative and postoperative complications were then compared with previously published results.

Intraocular pressure was measured by pneumotonometry; whenever a lower or higher than normal pressure was obtained, the measurement was repeated and the mean recorded. New-onset glaucoma was defined as an increased intraocular pressure requiring treatment in a patient without prior history of glaucoma. Fluorescein angiography was performed to confirm cystoid macula oedema when suspected. Endothelial cell counts were measured using the Tomey EM-3000 endothelial camera (a non-contact automated specular microscope; Tomey, Phoenix, AZ, USA). An allograft rejection was defined as corneal clouding in association with an epithelial or endothelial rejection line, keratic precipitates and/or anterior chamber cells. Iatrogenic primary graft failure was defined as any graft that had to be replaced with a new graft within 1 year of surgery. Graft dislocation was defined as any transplant requiring additional injection of air to ensure graft adhesion. A graft with interface fluid after surgery, but in position, was followed without procedural intervention.

Detailed outcome analyses were conducted for all 33 eyes of 31 patients with a mean follow-up of 14.2 months.

Surgical technique

All surgeries were performed by a single surgeon (VSN) in a uniform manner. First, a 5 mm scleral tunnel was created 1 mm behind the superior limbus. The pre-cut corneal button was placed endothelial side up and trephinated with an 8 mm Hessburg trephine. Using the operating microscope, the donor rim was examined to ensure the punch was centred. Then attention was turned back to the patient. A paracentesis was created and balanced salt solution (Alcon, Fort Worth, Texas, USA) was placed into the anterior chamber. A 2.75 mm keratome was used to enter the anterior chamber and expand the opening to 5 mm. No DM stripping was done in any cases. Pre-cut cornea was then placed endothelial side up on the surgical field. A thin strip of Provisc (Alcon) was placed onto the endothelium. The cornea was folded using a 60/40 taco technique, grasped gently with Ogawa forceps (M.S. Surgical, Chennai, Tamil Nadu, India) and placed into the anterior chamber. Three interrupted 10-0 nylon sutures were placed at the main incision. An air bubble was inserted into the anterior chamber to unfold the donor button. Paracentral venting incisions were created. The graft was centred and any residual interface fluid was removed using a 30-gauge needle and a Lindstrom roller (BD Visitec, Franklin Lakes, NJ, USA). Then a 100% air fill was obtained and left in place for 10 min to promote attachment of the graft. Homatropine drops were administered. After 10 min, 50% of air was removed and replaced with BSS. Next, subconjunctival antibiotics and steroids were injected. Patients remained flat for 30 min after the conclusion of the procedure to allow the air bubble to push the donor tissue against the recipient cornea. They were instructed to remain in the supine position as an outpatient as much as reasonably possible for 24–48 h. Postoperatively, patients were placed on a standard prednisolone taper every 2 h for 1 week, 4× for 1 month, 3× for 1 month, 2× for 1 month, then daily dosing.


A total of 33 eyes of 31 patients underwent EK. The mean age was 74.7±10.5 (range 49.3 to 89.1)  years. There were 24 (26 eyes) women and seven men (seven eyes). The reason for the failed PK was graft rejection in nine eyes (27.3%) and endothelial failure in 24 eyes (72.7%). The length of time before the most recent PK failure/rejection was 13.7±9.2 years. The length of follow-up after EK was 14.2±4.4 months. The reasons for the initial PK included Fuch's dystrophy n=15 (45.5%), corneal scar n=8 (24.2%), keratoconus n=5 (15.2%), pseudophakic bullous keratopathy n=4 (12.1%) and stromal corneal dystrophy n=1 (3.0%). 8/33 (24.2%) had more than one previous PK. Preoperative comorbidities included age-related macular degeneration n=4 (12.1%) with three dry and one wet, glaucoma n=4 (12.1%), cystoid macular oedema n=2 (6.1%), diabetic retinopathy n=2 (6.1%), uveitis n=1 (3.0%), strabismus/amblyopia n=1 (3.0%) and choroidal nevus n=1 (3.0%) (table 1).

Table 1

Characteristics of patients

A large proportion of patients had had more than one PK (8/33, 24.2%). In addition, the study eyes also had history of multiple surgeries for their other ocular comorbidities. Most patients 32/33 (97%) had cataract extraction prior to EK. Cataract extraction (CE) was performed at the time of PK in 10/32 (31.3%) with two of these requiring trans-scleral sutured posterior chamber intraocular lenses. One patient underwent combined EK with CE/PCIOL (cataract extraction with intraocular lens implantation). Furthermore, three eyes had previous trabeculectomy, four eyes had astigmatic keratotomy, three eyes had corneal laser photocoagulation using the yellow dye laser for deep stromal vascularisation, 13 eyes had yag capsulotomy, one eye underwent phototherapeutic keratectomy, one eye underwent lamellar keratectomy and one eye had pterygium excision. One patient had multiple previous graft failure, requiring placement of an investigational ciclosporine implant to prevent further rejection.

All donor tissue was pre-cut by the eye bank. The mean pre-cut central corneal thickness was 516±94 μm. The cell density was 2943±164 cells/mm2. The endothelial donor thickness was 127±26 μm.

Preoperatively, the average LogMAR VA was 1.435±0.40 (range 0.48–2.3). Postoperatively, the average LogMAR VA was 0.723±0.42 (range 0.1–1.86). Visual acuity improved in 28/33 (84.8%), stayed the same in 4/33 (12.1%) and declined in 1/33 (3.0%) eyes. Of the patients with visual acuity improvement, 19/28 (67.9%) of these patients had more than three Snellen lines of visual acuity improvement. The average postoperative vision in this group was 0.433±0.05 (range 0.1–0.7) figure 1.

Figure 1

Visual results of EK after PK. EK, endothelial keratoplasty; LogMAR, logarithm of the minimum angle of resolution; PK, penetrating keratoplasty.

The endothelial cell density (ECD) was measured in all but one patient at 3 months with an average density of 1920.5. This declined by 34.8% from the preoperative measurement (p<0.0001). Only 21 patients had ECD measured 12 months postoperatively with an average of 1775.5. The ECD declined by 39.7% at 12 months compared with the preoperative measurement (p<0.0001). ECD only declined by 4.9% (p=0.2795) from 3 to 12 months figure 2.

Figure 2

Post-operative endothelial cell loss.

The central corneal thickness was measured in 30/33 patients at 3 months. The average thickness was 615.1±71.4 μm.

Two patients (6.1%) had graft dislocations. The first dislocation occurred 1 week postoperatively after having successful attachment prior to dislocation with successful treatment with re-bubble in the operating room. The second patient had graft failure that required a new EK graft. The second EK graft also subsequently failed requiring PK to improve vision.

Postoperatively 32/33 (97.0%) of grafts remained clear at the last follow-up appointment (figures 3 and 4). No patients had significant clinical interface haze. No patients developed EK or PK rejection or PK failure.

Figure 3

Failed penetrating keratoplasty.

Figure 4

Successful endothelial keratoplasty after penetrating keratoplasty.

Patients in this study had the following postoperative newly diagnosed conditions: glaucoma n=2 (6.1%); cystoid macular oedema n=1 (3.0%).


Because of the published advantages of EK over PK, EK is fast becoming the preferred primary procedure of choice for endothelial dysfunction for the surgeon and patient alike. Several author have suggested that EK can even be performed on patients with PK failure and have published short case series reporting good visual results.7–9 Most of these published cases included DM stripping. Because of the risk of graft–host interface iatrogenic dehiscence and clear DM before corneal decompensation, we decided to evaluate the outcome of EK without DM stripping in eyes with previous PK.

Kobayash et al16 and Wu et al17 have shown that EK donor tissue can adhere directly over recipient endothelium and DM. This technique is called non-DSEK and both studies report favourable results. Descemet's can be more challenging to strip in patients with previous PK because the Descemet is thinner and more friable. In addition, there is a risk of opening the graft–host junction while attempting to strip the fragile tissue. Before graft failure, this patient population had a clear DM and did not normally have guttata. Because of these risks, we did not strip Descemet's in this study and had favourable outcomes.

Many studies have tried to link residual Descemet's with primary graft failure or graft dislocation. Individual case reports by Kymionis et al18 and Mondloch et al19 have shown residual fragment of Descemet's and residual fetal layer of Descemet's after stripping Descemet in failed grafts. Furthermore, two studies looking at the histology of eyes with primary graft failure only found 2/16 eyes20 and 5/19 eyes21 had retained Descemet's membrane. Both of these studies had a higher number of atrophic corneal endothelium on examination with 12/16 eyes and 16/19 eyes, respectively.20 21 Attrition of endothelial cells due to iatrogenic cell loss may play a larger role in the graft failure for these patients. In the above histological evaluations of the retained Descemet's, they are reported as residual fragments with irregular edges. These fragments of Descemet's can cause an irregular surface limiting donor tissue adhesion, which could contribute to endothelial cell loss and graft failure. By not disturbing Descemet's and leaving a smooth surface for graft adhesion, it may increase EK graft survival in patients with failed PK. This may explain why there was a higher graft failure/dislocation in the study of Covert and Koenig,8 but does not explain why the study of Terry and colleagues did not find a higher rate.9

EK has an independent risk factor, from PK, of graft dislocation, with an incidence between 4% and 35%.10 22 Dislocation rates of EK in patients with previous failed PK have been reported at 1/7,7 3/78 and 1/17.9 Our study only had 2/33 (6.1%) patients with graft dislocation. One patient was successfully treated with re-bubbling and maintained a clear graft at the last postoperative visit. The second patient failed re-bubbling and EK re-graft, requiring a traditional PK to improve the visual quality. The dislocation rate for this study is comparable to current peer-review published data at 6.1% (range 4–35%).10 22

One concern with leaving host Descemet's and endothelium is creating additional interfaces that could limit vision. Postoperatively 32/33 (97%) eyes had clear grafts without interface haze. 1/33 (3.0%) patients had iatrogenic primary graft failure that failed EK re-graft and required PK. Our study showed favourable visual results even with the residual DM. Visual acuity improved in 28/33 (84.8%) patients with 19/28 (67.9%) of these patients having more than three lines of visual acuity improvement. The average postoperative vision in this group was 20/60 (range 20/25–20/100). Because patients were not selected for EK after PK if they had significant astigmatism or irregular astigmatism, most patients achieved best-corrected vision with glasses. When factoring the other ocular comorbidities, such as age-related macular degeneration and cystoid macular oedema, this patient population had a significant improvement in vision with few complications.

Current long-term trends show that EK may have a lower rate of endothelial cell loss than PK.12 Mean PK endothelial cell loss at 12 months in cohorts of mainly low-risk Fuchs' dystrophy was between 15% and 37%.6 23 24 Endothelial cell loss after DSEK averages 34% at 6 months, 38–39% at 12 months, 42% at 2 years25 and 53% at 5 years.12 The mean ECD declined by 34.8% at 3 months and 39.7% at 12 months. These numbers are comparable with the results for primary DSEK and slightly higher than patients with primary PK. The small graft size (8 mm) and insertion technique (60/40 taco without an anterior chamber maintainer) may contribute to a higher ECD. We used a different automated specular microscope pre- and postoperatively and this could influence the changes in preoperative to postoperative measurements of ECD.


EK without Descemet's stripping in eyes with previous PK is a valuable alternative to a repeat full thickness corneal transplant.



  • Funding None

  • Competing interests None.

  • Patient consent This was a retrospective chart review and no patient consent was obtained. IRB approval was obtained from Union Memorial Hospital, Baltimore, Maryland, USA.

  • Ethics approval This study was conducted with the approval of the Union Memorial Hospital.

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