Long-Term Clear Graft Survival and Chronic Endothelial Cell Loss Following Posterior Lamellar Keratoplasty

Eye Center, Faculty of Medicine, University Hospital Freiburg, Killianstr. 5, 79106 Freiburg, Germany



Philip Maier

14.1 Introduction

Diseases of the corneal endothelium (Fuchs endothelial dystrophy, bullous keratopathy, and endothelial failure following penetrating keratoplasty) count for the most frequent indications for corneal transplantation. Instead of replacing all five layers of the cornea (epithelium, Bowman layer, stroma, Descemet membrane, endothelium) by penetrating keratoplasty (PK), which has first been performed by Eduard Zirm [13] in 1905, nowadays techniques for the selective replacement of the diseased corneal endothelium are preferred. This has first been suggested by Tillet in 1956 [4] to avoid characteristical complications following PK (e.g. high irregular astigmatism and wound healing problems). The technical principle of this kind of posterior lamellar keratoplasty or endothelial keratoplasty (EK) was to excise the host’s posterior stroma including Descemet membrane and endothelium and to replace it by identical lamellar graft prepared from the donor lamella. However, due to interface irregularities between host and donor stroma the visual results were not satisfying for the patients. One important technical step in improving posterior lamellar keratoplasty was the introduction of descemetorhexis by Melles et al. [5], where the stroma of the recipients cornea is not manipulated and only the Descemet membrane including the diseased endothelial cells is removed completely [6]. By avoiding manipulations of the recipient’s posterior stroma the clinical results could be improved as the lamellar graft could be attached to a smooth surface of the posterior stroma. In case of Descemet stripping endothelial keratoplasty (DSEK) the lamellar graft is manually prepared and consists of posterior stroma, Descemet membrane, and endothelium of the donor. It is transferred into the anterior chamber via a corneoscleral or clear cornea incision using special surgical instruments, unfolded and sutureless attached to the posterior stroma of the recipient’s cornea by an air bubble. Regarding DSEK one problem is the manual preparation process of the lamellar graft leading to graft preparation failures in some cases [7]. To overcome these problems of complete manual graft preparation the separation of the stromal lamellae can be performed using a microkeratome, by which a 400–500 μm thick layer of the anterior part of the donor cornea can be cut leaving a 80–150 μm thick lamella consisting of posterior stroma, Descemet membrane, and endothelium (Descemet stripping automated endothelial keratoplasty, DSAEK). The characteristic advantages of DSEK/DSAEK compared to PK are a suture-free graft adaptation and a faster improvement of visual acuity without a change of the refractive status. On the other hand, due to more intensive graft manipulations during graft preparation, implantation, and adaptation, the early endothelial cell loss following DSEK/DSAEK seems to be higher than that after PK. Furthermore, patients following DSEK/DSAEK often do not reach the maximum possible visual acuity which may be explained by optical phenomena caused by graft-host-interface reactions [8]. To reach the respective advantages of DSEK/DSAEK but to avoid graft-host-interface problems, Descemet membrane endothelial keratoplasty (DMEK) has been developed where the corneal graft consists only of Descemet membrane and endothelial cells without corneal stroma [9]. For DMEK graft preparation Descemet membrane including the endothelium is manually peeled off the donor’s posterior stroma using fine forceps. Following complete peeling off Descemet membrane forms a roll with the endothelial cells on the exterior. This graft roll can then be implanted into the anterior chamber through a small clear cornea incision and attached to the posterior recipient’s stroma by using an air bubble after the graft has been carefully unfolded. As no stromal tissue of the donor is grafted the primary anatomical situation is restored by DMEK leading to the maximum reachable visual acuity. However, one disadvantage of DMEK is that macroscopically invisible collagen fibres of the donor stroma can insert into Descemet membrane causing tears in the thin graft leading to graft preparation failures [10]. Furthermore, the unfolding of the graft may also be much more difficult than in DSAEK potentially leading to more intraoperative endothelial cell damage and more primary graft failures [3, 11].

The following chapter will focus on the long-term follow-up after DSEK/DSAEK/DMEK compared to each other and to PK with respect to clear graft survival and chronic endothelial cell loss.

14.2 Graft Failure

14.2.1 Primary Graft Failure

Primary graft failures (failure within 6 months following surgery) are not associated with endothelial graft rejections but are caused by endothelial insufficiency from insufficient tissue quality, complicated recipient factors (inflammation, interface problems, infection, flat anterior chamber, previous surgeries, etc.) or problems with the surgical technique. Primary graft failure following posterior lamellar keratoplasty can also be due to primary malfunction of the donor endothelium, but it more often is explained by a high endothelial cell loss due to intensive manipulation of the graft during surgery. Therefore, in some reports this kind of failure is named as iatrogenic primary graft failure. Various comparative cohort studies show that the risk for early/primary graft failure is lower after PKP (0–3 %) than after DSAEK (0–29 %) as well as after DMEK (0–9 %, see Table 14.2). Besides these results of various cohort and case control studies, Monnereau et al. [12] give an overview on real-life data when they published the results from 18 surgeons performing DMEK with different techniques. They included in total 431 cases of DMEK and found 1.2 % intraoperative failures and 10 % surgery associated primary graft failures.

Table 14.1
Factors associated and not associated with an increased endothelial cell loss following DSAEK

Factors associated with increased cell loss

Factors not associated with increased cell loss

Smaller incisions [51]

Donor tissue storage time [25, 52]

Longer incisions [53]

Graft diameter [25, 54]

Larger area of tissue compression by forceps [53]

Preoperative ECD [55]

Donor tissue dislocation [53]

Manual versus microkeratome preparation of the donor tissue [53]
Use of pre-cut versus surgeon-cut tissue [56]
Use of graft delivery devices [57, 58]

14.2.2 Midterm and Long-Term Clear Graft Survival

In the Cornea Donor Study clear graft survival following PK for endothelial insufficiency was 86 % after 5 years [13], and the clear graft survival rate was better for patients with Fuchs dystrophy (93 %) than for patients with pseudophakic or aphakic corneal oedema [14]. For PK the 10-year failure rate for Fuchs dystrophy ranges between 10 and 20 % [1517]. From the Irish keratoplasty registry a clear graft survival rate of 93 % after 1 year was reported [18], and in the Swedish Corneal Transplant Register a graft survival rate of 90 % after 2 years was found [19]. Thompson et al. [17] reported on 90 % clear graft survival 5 years after PK. Regarding late endothelial graft failure the recipient bed endothelial reservoir seems to play an important role on the long-term prognosis following PK as, for example patients with keratoconus (92 %) have a much better prognosis regarding 15-year long-term clear graft survival than patients with bullous keratopathy (67 %) [20].

For posterior lamellar keratoplasty such long-term survival data are not available. However, large register data from the United Kingdom [21] and from Australia [22] show that early and midterm to long-term graft survival following posterior lamellar keratoplasty is worse compared to PK. These register data seem to mirror the real-life situation including a lot of learning curves from different surgeons. In the analysis of the Australian registry [22] the authors included the clinical course after 13920 penetrating keratoplasties (2965 were performed in Fuchs endothelial dystrophy or bullous keratopathy) and 2287 endothelial keratoplasties performed between 1996 and 2013. Primary (within 1 month after surgery) as well as overall graft survival was worse following posterior lamellar keratoplasty compared to PK. If primary graft failures were excluded from the analysis, as they may be correlated to surgical effects, grafts following endothelial keratoplasty in patients with Fuchs endothelial dystrophy still had worse survival rates than grafts following PK. The analysis included data from 74 surgeons performing endothelial keratoplasty and found that graft survival was better for more experienced surgeons that had performed 100 or more endothelial keratoplasties. A similar surgeon’s experience effect regarding endothelial keratoplasty was found in the analysis of the registry data from the United Kingdom [21] where 2074 cases of endothelial keratoplasty performed between 2006 and 2011 were compared to 2622 cases of penetrating keratoplasty performed between 2000 and 2005 for Fuchs endothelial dystrophy or bullous keratopathy. The authors also found an overall worse survival rate after 2 years for endothelial keratoplasty compared to PK. Besides donor’s endothelial cell count, donor age, and glaucoma at the time of surgery the surgical experience was also a statistically significant factor influencing clear graft survival following endothelial keratoplasty. Akanda et al. [23] performed a systematic review on the available literature regarding graft rejection and graft failure following PK compared to EK. For posterior lamellar procedures they included seven studies and found that the pooled odds ratio for graft failure of PK versus posterior lamellar procedures was 2.09 (95 % CI: 0.57–7.59) demonstrating a statistically non-significant difference. From various cohort studies, it was reported that clear graft survival following DSAEK with follow-ups from 12 to 24 months ranges from 55 to 100 % (for literature see Table 14.1). Up to now the study with the longest follow-up after DSAEK comprises 5 years. Ratanasit and Gorovoy found that 92 % of the 51 included DSAEK grafts were clear and Price et al. reported on 93 % cumulative 5-year clear graft survival [24, 25]. As for PK Price et al. found a better clear graft survival for patients with Fuchs dystrophy (95 % cumulative 5-year clear graft survival) compared to pseudophakic bullous keratopathy (76 % cumulative 5-year clear graft survival) and also as for PK [26, 27] they found pre-existing glaucoma as a significant risk factor for graft failure. In their study 3.6 % of graft failed due to endothelial decompensation and 3 % of patients were regrafted for unsatisfactory corrected distance visual acuity [25].

Following DMEK only data on midterm clear graft survival with 12 months follow-up are available where graft survival rates were between 99.3 and 100 % [28, 29]. Regarding the data from the Australian register it is important to notice that the number of included DMEK cases is very small (129 out of 1214 cases of endothelial keratoplasty). As it could be shown the success rates of DMEK increase with more experience [30], these follow-up data on DMEK have to be interpreted with caution as the learning curves of different surgeons are included in this small number of cases. Our own experience shows that during the first 2 years following surgery the graft failure rate was higher following EK than following PK, where PK (2 %) had the lowest graft failure rate followed by DMEK (7 %) and DSAEK (20 %) [31]. However, graft failure was also defined as repeat keratoplasty due to unsatisfying visual results following EK, which was higher following DSAEK compared to DMEK.

14.3 Endothelial Cell Loss

14.3.1 Physiologic Endothelial Cell Loss

Human corneal endothelial cells may be able to proliferate in vitro but they rarely show mitotic activity in vivo [32, 33]. Therefore, during lifetime a yearly chronic endothelial cells loss of about 0.6 % can be observed [34]. This cell loss does not lead to malfunction of a healthy endothelium if there is no additional condition causing a decrease in endothelial cell density like Fuchs dystrophy, trauma, inflammation or iatrogenic cell damage [3537].

14.3.2 Endothelial Cell Loss Following Penetrating Keratoplasty

The yearly decrease of endothelial cells in the first year following PK is in the mean 4–16 % [15, 20, 38], but this chronic endothelial cell loss slows down during long-term follow-up. However, the individual endothelial cell loss varies widely [39, 40]. One reason for this may be the migration capability of the recipient’s peripheral endothelium as Reinhard et al. found a higher yearly endothelial cell loss for patients with bullous keratopathy (29.4 ± 17.6 %) compared to patients with Fuchs dystrophy (17.0 ± 19.1 %) or keratoconus (14.0 ± 19.0 %) [41]. The measurement of endothelial cell densities in the first years following PK can be more useful as a predictor for late endothelial graft failure as the preoperative endothelial cell counts [20, 40, 42]. The overall endothelial cell loss 5 years following PK lies in the range of 59–70 % [43, 44].

14.3.3 Endothelial Cell Loss Following Endothelial Keratoplasty

In the age of posterior lamellar keratoplasty it seems important to differentiate between the early endothelial cell loss that can be associated to the surgical trauma and the chronic endothelial cell loss in long-term follow-up.

14.3.4 Early Endothelial Cell Loss up to Six Months Following Posterior Lamellar Keratoplasty

In the early post-operative follow-up the endothelial cell decrease is higher following EK than after PK, which can be associated to more intensive surgical manipulations of the graft during EK. In a systematic review of the literature including 19 trials Lee et al. [45] reported on an early endothelial cell loss within 6 months between 25 and 54 % after DSAEK with a mean endothelial cell loss of 37 %. When comparing DSAEK and PK, Price et al. [46] found a statistically significant lower endothelial cell loss 6 months following PK compared to DSAEK (11 ± 20 % vs. 34 ± 22 %, respectively, p < 0.001). For their multi-centre case series Monnereau et al. [12] found an overall decrease in ECD of 47 % (SD 20 %) 6 months following DMEK. In a large single centre series on 500 DMEK cases Rodríguez-Calvo-de-Mora et al. [30] report on an average endothelial cell loss 6 months following DMEK of 37 ± 18 %. In a direct comparison of DSAEK and DMEK comparable rates of early endothelial cell loss within 6 months following surgery were found (41 % for DSAEK and 39 % for DMEK) [47]. Regarding the early, most probably surgery associated endothelial cell loss, it has to be considered that the extent of the early endothelial cell loss may depend on the learning curve and the experience of the surgeon [48]. One other important factor regarding the early endothelial cell loss following posterior lamellar keratoplasty is the dislocation of the lamellar graft which is the most frequent post-surgical complication following endothelial keratoplasty. Graft dislocations are treated by a repeat air fill of the anterior chamber (so called rebubbling), which may be required in 1–82 % of cases following DSAEK [49] and even more often after DMEK (33–81 %) [28, 47]. The risk of graft dislocation seems to be lower using grafts that have been stored in organ culture, which is the case in most European eye banks, than using grafts stored in short-term culture, which is the case in most American eye banks [50]. Finally, for all studies comparing pre- vs. post-operative endothelial cell loss the reported figures have to be interpreted with caution as in many cases the technique of measuring the endothelial cell density of the graft in the eye bank and of the recipient following surgery varies so that there might be a systematic error in these comparisons.

Several factors have been identified that are associated with an increased early endothelial cell loss following DSAEK. These factors are summarised in Table 14.1.

14.3.5 Chronic Endothelial Cell Loss Following Posterior Lamellar Keratoplasty

The high endothelial cell loss in the early post-operative period following EK seems to stabilise over time with longer follow-up what further supports the important influence of the surgical manipulations regarding the early endothelial cell loss. When looking at the systematic review of Lee at al. [45] they reported on an endothelial cell loss at 1 year after DSAEK from 24 to 61 %, with an average of 41 % which was only a slight increase compared to the endothelial cell loss after 6 months (mean 37 %). Similar data were found by Price et al. [46] with an endothelial cell loss of 38 ± 22 % 1 year after DSAEK compared to 20 ± 23 % after PK where the increase of cell loss within 6–12 months after surgery was higher in the PK group compared to the DSAEK group. This observation could also be found 3 years following keratoplasty, when the mean endothelial cell loss was 39–46 % after DSAEK and 47–51 % after PKP [59, 60]. Van Dooren et al. [61] presented their 3 years follow-up data of 39 DSAEK versus 33 PK patients and calculated a regression model for the decrease in endothelial cell density. They found in their statistical model that the early endothelial cell loss was higher following DSAEK compared to PK but is slower in the long run so that the endothelial cell densities seem to converge after a certain time of follow-up. It seems as if the chronic endothelial cell loss, which is a characteristical property of PK grafts, may be slower after DSAEK. This hypothesis can be strengthened by the longest available follow-up data for 51 patients 5 years following DSAEK where the mean endothelial cell density decreased from 2700 (2500–3000) cells mm2 to 1312 (374–2927) cells/mm2 which is a mean cell loss of about 49 % [24]. Price et al. also reported on 5-year follow-up after DSAEK and found a mean endothelial cell loss of 53 % [25]. Comparing these results to the data of the Cornea Donor Study shows that the endothelial cell loss 5 years after DSAEK (49–53 %) was lower than that 5 years after PK (70 %) [24, 25, 44].

For DMEK less information regarding the chronic endothelial cell loss is available as the technique is evolutionary younger. After 12 months the endothelial cell loss varies between 19 and 40 % [29, 62]. However, it seems that in the long run the chronic endothelial cell loss after DMEK is comparable to that after DSAEK and therefore slower than after PK, so that the endothelial cell densities may also converge about 2 years after surgery [63]. Up to date the longest follow-up period following DMEK is 5 years where the loss of endothelial cells was high early after DMEK with 38 % after 1 year but 55 % after 5 years meaning a mean decrease of endothelial cells of around 7 % per year after the first year [64].

Figure 14.1 shows a comparison of the chronic endothelial cell loss up to 5 years following PK (data extracted from the younger cohort (<66 years of age) of the Cornea Donor Study Investigator Group), DSAEK (data extracted from Price et al. [25]), and DMEK (data extracted from Baydoun et al. [64]) demonstrating the high endothelial cell loss early after endothelial keratoplasty whereas the chronic endothelial cell loss seems to slow down so that in the long run the endothelial cell density may be higher after endothelial keratoplasty compared to the endothelial cell density following PK. Furthermore, Quilendrino et al. [65] created a mathematical model for the estimation of the increase in posterior corneal surface area associated with post-operative corneal deturgescence and its effect on ECD measurements. From this theoretical model they found that the endothelial cell loss following DMEK may be overestimated by about 8 % due to the increased posterior corneal surface area associated with post-operative corneal deturgescence. However, the reason for the slower chronic endothelial cell loss following either DSAEK or DMEK compared to PK is not fully understood. One explanation could be the reduced risk of endothelial immune reactions following DMEK, which still is the most important cause of graft failure after PK [19]. In the available literature the following risks of endothelial graft rejection within 2 years after transplantation have been reported: 0–23 % for PK, 0–14 % for DSAEK, and 1–3 % for DMEK (Table 14.2). In their systematic review Akanda et al. [23] included seven publications on the risk of graft rejection comparing PK and EK and found that a pooled odds ratio for rejection of PK over EK was 1.52 (95 % CI: 1.00–2.32) demonstrating a statistically significantly reduced risk of graft rejection following posterior lamellar keratoplasty. However, data from the registry of the United Kingdom [21] could not demonstrate a difference in the risk of endothelial immune reactions comparing 1961 PK versus 1525 EK for either Fuchs endothelial dystrophy or bullous keratopathy. However, in this study there was no differentiation between DSAEK and DMEK for EK, and it seems obvious that most included cases were DSAEK and not DMEK. As the risk for endothelial rejections following DSAEK seems to be in the same range as for PK, it does not seem surprising that no difference in the risk of immune reaction could be found. The reason for the reduced risk of rejection following DMEK is not completely understood; however, it seems that the risk increases with the amount of grafted tissue. Further evidence for this observation can be found in the mouse experiments from Hori et al. where they found that alloimmunogenicity of the normal cornea resides within its epithelial and stromal layers, whereas immune privilege arises from the endothelium [66, 67]. Besides the corneal epithelium, which seems to play the major role in allograft sensitisation [68], it has also been found that corneal stroma cells are able to inhibit the production of anti-inflammatory cytokines [69]. This experimental data may help to explain the clinically observed reduced risk of endothelial immune reaction following DMEK compared to DSAEK and PK.


Fig. 14.1
Decrease of the median absolute endothelial cell density (a) respectively the ratio of endothelial cell survival (b) as a function of time following penetrating keratoplasty (PK) extracted from the younger donor cohort (<66 years of age) of the Cornea Donor Study Investigator Group [44] versus Descemet stripping endothelial keratoplasty (DSAEK) extracted from Price et al. [25] versus Descemet membrane endothelial keratoplasty (DMEK) extracted from Baydoun et al. [64]

Table 14.2
Data from the literature on primary graft failure, mid-to long-term clear graft survival, chronic endothelial cell loss, and the risk of endothelial immune reactions for PK, DSAEK, and DMEK, respectively




Primary graft failure

0–3 % [15, 30]

0–29 % [15, 19, 27, 30, 34, 38, 40, 71, 7376]

Only gold members can continue reading. Log In or Register to continue

Jun 27, 2017 | Posted by in OPHTHALMOLOGY | Comments Off on Long-Term Clear Graft Survival and Chronic Endothelial Cell Loss Following Posterior Lamellar Keratoplasty
Premium Wordpress Themes by UFO Themes