Graft survival and clinical outcomes of Descemet membrane endothelial keratoplasty: long-term results

The aim is to report long-term graft survival rates and clinical outcomes of Descemet membrane endothelial keratoplasty (DMEK). In this study 150 eyes that underwent DMEK whether for Fuchs endothelial corneal dystrophy (FECD) or for bullous keratopathy (BK) surveilled for 7 years at 6 time points to evaluate graft survival rates and clinical outcomes of post-corneal transplantation. Overall, the estimated survival probability of 95% confidence interval at 7 years of post-DMEK was 0.58 (0.72–0.77), and the survival probabilities of eyes operated for FECD (0.53) were higher than eyes operated for BK (0.42) (log rank 26.87, [p = 0.197]). Post-transplant eyes with FECD achieved better visual acuity levels than eyes with BK (p = 0.006). Primary graft failure occurred in 11.3% eyes. Secondary graft failure rate was 9.3%, and allograft rejection rate was 4.7%. Although DMEK is effective and safe in long term, visual results and graft survival rates are better in cases with FECD.


Introduction
Melles introduced Descemet membrane endothelial keratoplasty (DMEK) for patients with corneal endothelial dysfunctions for the first time in 2006 [1]. Since then, it has become a preferable therapy with expedited visual recovery and better refractive and ocular surface integrity results [2,3]. Compared to penetrating keratoplasty (PK), DMEK is a minimally invasive surgery and provides better anatomic restoration of the corneal surfaces than Descemet stripping endothelial keratoplasty (DSEK), which is another type of lamellar endothelial keratoplasty [4,5].
One of the disadvantages of DMEK is probability of corneal damage during removal of the graft from the donor, where loss of endothelial tissue is possible [6,7]. In addition, during the surgery, there might be difficulty in obtaining the graft from an incision in the anterior chamber, which leads to greater tissue manipulation, resulting in damaged endothelial cells and increased graft detachment [8][9][10]. However, assessment of the learning curve for graft preparation and implantation could reduce the complications of DMEK and enhance the success of the procedure by establishing a good standard of care [11].
There are plenty of short-term clinical outcomes reported for the transplantation in the literature [5], but there are limited data from long-term post-surgery follow-up. One recent example is the study by Vasiliauskait_ e et al. [12], who suggested that the 10-year graft survival rate is 0.79. Therefore, in the present study, we aimed to evaluate the long-term clinical outcomes and graft survival rates of a group of post-DMEK cases.

Materials and methods
Study design, patient selection, and data collection Between 2013 and 2019, 137 consecutive patients underwent DMEK for Fuchs endothelial corneal dystrophy ([FECD], 61 eyes, 40.7%) and bullous keratopathy ([BK], 89 eyes, 59.3%). In total, 150 eyes of such patients were operated (Table 1) and evaluated retrospectively. The present study was performed in conformance with the principles of the Declaration of Helsinki after obtaining approval from the Local Ethics Committee of Okmeydani Training and Research Hospital, Istanbul, Turkey. Written informed consent was also obtained from all patients before the procedure.
Patients were excluded from the study if they had conditions that might affect their vision, such as retinal detachment, recurrent uveitis, trauma, proliferative diabetic retinopathy, macular holes, glaucoma, retinal vascular occlusion, age-related macular degeneration, etc. Eyes were also excluded if they had less than 6 months of follow-up. This mono-centre study also included three surgeons' learning curves and early adoption techniques for the standardization of the implantation. The data recorded during the study period included preoperative demographic and medical data (sex, age at date of visit, surgical indications, lens situation, visual acuity), intraoperative data (number of donor endothelial cells, graft size, complications), and postoperative data (best corrected visual acuity [BCVA], endothelial cell density [ECD], presence of graft/detachment of graft, primary and secondary graft failure, and other postoperative complications).
Corneal allograft rejection was defined as an endothelial rejection line or inflammation with keratic precipitates, cells in the stroma, or an increase in aqueous cells from a previous visit, with or without any clinically apparent change in recipient stromal thickness or clarity [13].
Based on the study by Lass et al. [13], graft failures were classified as primary graft failure (PGF) if the graft was cloudy on the first postoperative day and did not clear up or if it required a re-graft within 8 weeks. Secondary graft failure (SGF) was considered to occur when the cornea was initially clear but later became cloudy in 3 months. A Snellen letter chart was employed to evaluate BCVA. For statistical analyses, the visual acuity measurements were converted into logMAR values. A preoperative number of donor endothelial cells were obtained from the Beyoglu Eye Bank donor documentation as counted with a specular microscope (Cell Check, Konan Medical Inc, USA). Patients' endothelial cell density (ECD) was visualized under a noncontact auto-focus Nidek Cem-530 specular microscope (Nidek CO., Gamagori, Japan) at 6 months and the 1st, 3rd, 5th, and 7th years post-surgery. Furthermore, central corneal thickness (CCT) was measured with Sirius corneal topography (Costruzione Strumenti Oftalmici, Italy), and graft attachment/detachment was evaluated by anterior segment optical coherence tomography (OCT) (Carl Zeiss Meditec. Inc., Dublin, USA).

Surgical procedure
For all DMEK procedures, grafts were obtained from donor globes within less than 36 h post-mortem. From the ocular globes, corneo-scleral buttons were excised, and the organ culture was stored in modified minimum essential medium (EMEM) at 318C [14]. The average donor age was 53.34 ± 12.75 years, and the donors' average endothelial cell count was 2572 ± 250 cells/ mm 2 until the time of transplantation. The mean graft storage time was 10.58 ± 1.92 days ( Table 1).
The Beyoglu Eye Training and Research Hospital Cornea Bank provided us with all corneas, and before the surgery, Descemet membranes (DMs) were separated from the donor corneas with a 3608 manual dissection. DMs were detached and cut using a 7.75mm punch. Immediately after, they were stained with trypan blue dye (0.06%). A single peripheral triangular mark was made with a micro-scissor, and the DM was completely separated from the donor cornea [15]. Finally, they were re-stained with trypan blue dye and stored in EMEM solution until the day of DMEK.
We performed a similar surgical technique to the one developed by Melles et al. [1]. Briefly, under subtenon's anaesthesia, 3 side-port incisions were created in the superotemporal, superonasal, and inferotemporal quadrants, respectively. Right after administering 0.01% carbachol (Miostat; Alcon Laboratories Inc.) to the anterior chamber, a peripheral iridectomy was performed through a vitrectomy. After filling the anterior chamber with air (35 mmHg), descemetorhexis was performed, and the host DM was removed.
We assembled a device with an Alcon Monarch C cartridge, along with serum set. After entering the temporal quadrant with a 2.75-mm scalpel as the main incision of the procedure, the DM was taken into the cartridge using the injector's negative pressure, and the donor's DM was placed in the anterior chamber through the main incision using the cartridge once more. The anterior chamber and side incisions were flattened, and a tapping technique was used to unfold and position the graft. As a tamponade, an air bubble or sulphur hexafluoride (SF6-gas) was injected underneath the graft to position it against the recipient's posterior stroma, and then the main incision was closed up with 10/0 nylon suture. In all phakic cases, standard phacoemulsification and intraocular lens implantation were performed prior to DMEK surgery.
All patients were kept strictly in a supine position throughout the first 24 h after the surgery. Postoperatively, dexamethasone (0.1%, 4x/day) was administered for 4 weeks and then changed to fluorometholone (0.1%, 4x/day) at the first-month visit, which was gradually decreased to once a day at the 12-month visit. Additionally, topical antibiotics were prescribed for approximately 2 weeks along with lubricant eyes drops (5 times a day).
At the first day post-operation, all recipients' anterior segment underwent OCT imaging to see whether the graft was properly attached. If the DM was more than 1/3 detached, the eyes were re-bubbled in 24-36 h. If the DM was less than 1/3 detached and there was cornea edema, patients were followed up for 2-3 weeks. When edema was persistent, re-bubbling was injected. In cases of graft failure, DMEK was repeated or PK was performed.

Statistical analyses
A descriptive analysis was carried out for all eyes using SPSS 17.0 (SPSS Inc., Chicago, IL). Kaplan-Meier survival analysis was performed to estimate the cumulative success probability of graft survival. All primary and secondary graft failures and re-transplantations were considered as failures in the survival analysis. Log-rank tests were used to compare the survival time distributions of different subgroups.
Linear mixed models with a random intercept were used for the analysis of the influence of parameters such as the patients' age, sex, lens status, surgery indication, graft storage time, adherence status, donors' cause of death, and age on BCVA, ECD, and CCT. Analyses were repeated for tests comparing groups at multiple time points, and p values were Bonferroni adjusted (p \ 0.01).

Endothelial cell density (ECD)
Preoperatively, the patients' mean ECD was 2572 ± 250 cells/mm 2 , while postoperatively, it decreased gradually to 1793 ± 636 cells/mm 2 at the 6th month, 1688 ± 626 cells/mm 2 at the 1st year, 1532 ± 608 cells/mm 2 at the 3rd year, 1310 ± 525 cells/mm 2 at the 5th year, and 1190 ± 441 cells/mm 2 at the 7th year. ECD declined by 37% (± 20.75), 42.58% (± 19.04), 44.75% (± 17.83), 47.49% (± 13.60), and 55.03% (± 13.41), respectively, compared to preoperative ECD values ( Table 3). The decrease in ECD was significant at all visits from 6 months to the 5th year post-DMEK (p \ 0.001). When the mean ECD between 5th and 7th years was compared, it was statistically significantly lower in 7th year (p = 0.01). Although there was a negative correlation between ECD and graft storage time, it was not statistically significant (p = 0.091), and no correlation was found between donor age, cause of death, and ECD (p [ 0.05). Female patients had a positive correlation with ECD (p = 0.035), and there was a negative correlation between patients' age and ECD (p = 0.034). There was also a negative relationship between eyes with minor (p = 0.031) and major (p\0.01) detachment and ECD values (Table 4), but no statistically significant differences were found with eyes that underwent re-bubbling and those that did not, and ECD declined over the years (p [ 0.05).
Eyes with FECD attained better visual acuity levels than eyes with BK with an average of 0.18 on the logMAR scale (p = 0.006). There was a correlation between visual acuity at 7 years and age, and older recipients had better logMAR scores (p = 0.009; Table 4). No correlation was found between major and minor detachments and 7-year logMAR measurements (Table 4). Furthermore, there was no correlation between donor age and cause of death, graft storage time, and BCVA logMAR scores (p [ 0.05). There were no significant differences between logMAR measurements of cases that underwent re-bubbling and those that did not (p = 0.316).

Corneal pachymetry
The mean CCT in preoperative eyes was 665 ± 126 lm, and the mean CCT in postoperative eyes was 568 ± 102 lm at the 6th month, 572 ± 93 lm at the 1st year, 572 ± 91 lm at the 3rd year, 575 ± 94 lm at the 5th year, and 578 ± 54 lm at the 7th year (Table 3). Changes in corneal thickness were statistically significant between preoperative and postoperative measurements for the 6th month (p \ 0.01), while the other repeated measures were similar (p [ 0.05). There was a negative correlation between ECD and pachymetry results (r = -0.04, p = 0.036), but pachymetry had a positive correlation with graft storage time (r = 15.48, p = 0.008). No association was found of CCT outcomes with patients' age, sex, lens status, indication of DMEK, and status of graft attachment (Table 4).

Postoperative complications and retransplantations
In the first 6 months, post-surgery, 29 (19.3%) eyes had graft detachment; 19 (12.7%) of them underwent re-bubbling. Only one of 11 eyes with minor graft detachment was re-bubbled and graft re-attachment was achieved. All 18 eyes with major graft detachment were re-bubbled; nevertheless, graft re-attachment was achieved in 11 (7.3%) of them. Two eyes underwent PK; 5 eyes underwent re-DMEK. There was no statistically significant difference between the rate of minor and major graft detachment development in patients who underwent phacoemulsification and those who did not. (p = 0.378 and p = 0.216, respectively). PGF occurred in 11.3% (17/150) of the eyes, 8 of them had PK, and for 9 eyes, DMEK was repeated. The SGF rate was 9.3%, and all SGF occurred after 1-year follow-up of DMEK, with 8 eyes (57.1%) occurring between 1 and 3 years and 6 (42.9%) eyes occurring between 3 and 7 years. There were 4 eyes that developed SGF and also received a second graft, 5 eyes underwent PK, and 5 eyes rejected a second surgical operation. The postoperative allograft rejection rate was 4.7% (7/150), and 3 of those were treated with corticosteroid, while 4 eyes that was not responder to corticosteroid treatment was undergone PK (Table 5).

Discussion
In the last decade, DMEK has become a popular technique for the treatment of corneal endothelial dystrophies [16,17]. A number of cases are needed to gain experience on graft preparation and unrolling time for corneal surgeons to achieve a good standard of care. Nevertheless, DMEK has better graft survival rates than DSEK and PK [18]. According to Birbal et al. [4], the 5-year graft survival success of DMEK is 90%. Moreover, the cumulative mid-term graft survival rate was reported to be as high as 96%, which corroborates the 5-year graft survival rate reported by Schlögl et al. [3] and Baydoun et al. [19]. The results of Vasiliauskait_ e et al. [12] suggested a 79% of graft survival rate of posttransplantation in 10 years. In this long-term study, we evaluated graft survival rates of 150 eyes and analysed risk factors that affected the outcomes. Overall, the graft survival rates of DMEK were as follows: 96% at the 6th month, 94% at the 3rd year, 88% at the 5th year, and 58% at the 7th year. Compared to the studies mentioned, our findings suggested a lower survival rate of corneal endothelial grafts. The reason for this might be that we included learning-curve cases in our study, and 59.3% of the study population was diagnosed with BK.
Many publications that generated short-and longterm outcomes for DMEK reported a significant improvement in visual acuity for the majority of cases. For example, Birbal et al. [4] achieved excellent visual outcomes within 6 months of transplantation. In a year, 73% of cases' BCVA scores improved by C 2 Snellen lines, and by five years, the improvement was maintained [4,14]. According to Vasiliauskait_ e et al. [12], even though there were differences in BCVA results in 1st, 5th, and 10th years post-transplantation, there was a significant increase in BCVA compared with the preoperational results. In our study, 62.6% of eyes had an increased BCVA by C 2 Snellen lines in a year, but the long-term postoperative visual acuity results were lower than in the other studies. The reasons for this might be patients' delayed visits to the hospital before the surgery (despite their need for treatment) and patients' bad preoperative visual acuity. The delayed preoperative visit could affect the potential of the stroma to clear up without haze or scar after DMEK and so postoperative visual acuity may not increase.
Recent studies have shown a continued BCVA improvement in visual acuity from 6 months to 3 years postoperatively, but on the other hand, there might be selection bias due to geriatric patients, who have lower BCVA in general and withdraw from continuous follow-up periods. In contrast, younger patients have more adherence to follow-up visits [11,18]. We also observed a positive correlation with BCVA (logMAR) and patients' age. In the 6th-month visit, there was quite an increase in visual acuity, which was maintained for up to 3 years. Similar to the other studies, younger patients attended post-DMEK surveillance more than elderly recipients.
Postoperative ECD was negatively associated with DMEK, DSEK, and PK [4,12], and the highest ECD reduction was observed in 6th-month check-ups [18]. After DMEK, eyes operated on for FECD had less endothelial cell loss than those with BK and other corneal dysfunctions because of the donor endothelial cells' regenerative capacity [1,19,20]. During DMEK, difficulties in unrolling endothelial lamella and greater tissue manipulation could cause damage to cells and detachment of the graft [10]. In the case of graft detachment, air injection one time had no effect on ECD, and two injections and more re-bubbling procedures might reduce ECD [15,[21][22][23][24][25][26]. A negative correlation between graft storage time and ECD was reported by Birbal et al. [4], and ECD declined 37% in the first 6 months. However, this result had no statistical significance. In this study, ECD numbers had a negative correlation with the age of the patient and minor and major detachment, whereas re-bubbling had no significant effect on postoperative ECD.
The most important complication of DMEK in the first 6 months is graft detachment, which might have rates of up to 82% [9,23]. Minor graft detachment was seen twice as much as major detachment and five times more than full detachment [17]. Surgeons' experience with DMEK could reduce the occurrence of graft detachment, and decentred grafts and overlapped host-donor DMs may correlate with higher re-bubbling rates [17,[23][24][25]. Upside-down grafts are one of the causes of detachment, and preoperative stamping is helpful to show the correct position of the graft (right side up) to avoid this complication of DMEK [15,26].
In this study, we used a single peripheral triangular marking as the graft marking technique, which is useful in cases with poor anterior chamber visibility. When endothelial cells are damaged during re-bubbling or peripheral partial detachment occurs, waiting for a while for spontaneous re-attachment might overcome the problem. However, if the partial detachment is major and at the centre, or full detachment happens, then re-bubbling is required [21]. Re-bubbling increases the graft survival rate [6] and decreases re-transplantation rates [19,21,27], although there is no consensus on its timing. A delayed procedure might cause scarring on the DM [21].
Eyes that underwent re-bubbling once and primarily attached eyes had similar postoperative visual outcomes [27]. Minor and major detachment cases were re-bubbled from 1 day to 13 weeks [12]. Like in previous studies, in the present study, eyes that underwent re-bubbling once and primarily attached eyes had no significant differences in ECD and visual acuity outcomes. Eyes with minor detachment were followed for a mean of 6.92 ± 3.63 (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14) days, and if the edema did not subside, they were re-bubbled. Rebubbled eyes with major detachment were injected with an air bubble underneath the graft in the first 24-36 h, and as a result, 61.1% of them benefited from the process, whereas 38.9% needed another DMEK or a PK. Primary graft failure is an early post-DMEK complication. It is defined as corneal edema for a duration of 3-4 weeks [11,23,28]. SGF and allograft rejection constitute more severe complications that occur later after transplantation [11,23]. Despite graft rejections being associated with increased ECD loss, ECD loss is not a risk factor for graft failure [5]. Usually, SGF occurs after allograft rejection or persistent graft detachment [18]. Eyes that developed SGF had surgery-related endothelial cell loss. Although eyes retained sufficient ECD, they had failure, and the reasons might be DM storage conditions, manipulation-related functioning problems of the corneal graft, or endothelial cell shock [15].
SGF was observed in 6% of the study group and became the major complication in the long-term follow-up of corneal post-transplantation along with decreased ECD. Re-transplantation was performed for the management of SGF and graft detachment, and half of the cases were re-transplanted in a year [12,18]. We observed the progression of SGF in 9.3% of eyes and followed-up 57.14% of those for the first year. We performed a second transplantation on 64.3% of them.
Despite the results of the study, its limitations are its retrospective design and the volunteer visits by the patients during the study period, which led to a lack of long-term data. In conclusion, DMEK is a corneal transplant option for patients with endothelial dysfunctions for long-term follow-ups. The reasons are its lower complication risk, better graft survival rate, and its therapeutic safety and efficacy. However, postoperative visual acuity might be affected by the patients' preoperative level of vision.