Purpose To evaluate graft size on outcome following Descemet stripping automated endothelial keratoplasty (DSAEK)
Methods Consecutive patients who had undergone a DSAEK for Fuchs endothelial dystrophy (FED) and pseudophakic bullous keratopathy (PBK) with at least 1 year of follow-up. Patients were divided into three groups according to the size of the donor trephine: <9, 9 and 9.5 mm. Main outcomes were postoperative best corrected visual acuity (BCVA) and graft failure. Grafts were prepared using an automated microkeratome. For larger grafts (≥9 mm), a manual dissection of the residual peripheral ring of anterior lamella was performed before trephination. Donor age, endothelial cell density (ECD) and postmortem times; recipient details including risk factors, comorbidity, surgical complications and postoperative BCVA and graft survival were analysed.
Results Of 174 patients, 131 were included: 84 (64%) with FED and 47 (36%) with PBK. Mean preoperative and postoperative BCVA were 1.01±0.76 and 0.2±0.2 logMAR, respectively, at 12 months with 80.5% achieving 20/40 or better. Postoperative BCVA was significantly associated with ECD (p=0.005), PBK or FED (p=0.004), risk factors (p=0.007) and comorbidity (p=0.016). Eleven patients (8.40%) experienced endothelial graft failure; 17.86% for <9 mm, 7.69% for 9 mm and 3.84% for 9.5 mm trephine sized grafts. Graft failure was significantly associated with ECD (p=0.039) and graft trephine size (p=0.04).
Conclusions Larger grafts occupy a smaller chord length in the eye than the trephine size and are expected to provide 10%–20% more endothelial cells. Increased graft size and donor ECD is significantly associated with a reduced graft failure rate.
Statistics from Altmetric.com
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.
Corneal transplant survival and visual outcome are dependent on endothelial cell density (ECD) and function. Central ECD is the principal criterion for the selection of donor corneas for transplantation. ECD, however, is not uniform across the cornea. Schimmelpfenning1 and Daus and colleagues2 reported a significant increase in ECD in the peripheral cornea. Amann et al3 also demonstrated an increase in ECD from the central to the peripheral regions of the human cornea, specifically an increase of 10% beyond 9 mm.3 In addition to the increase in ECD, the proliferative capacity of human corneal endothelial cells is also increased in the peripheral cornea.4–8 These findings would suggest that larger donor endothelial graft in Descemet stripping automated endothelial keratoplasty (DSAEK), that is, greater than 9 mm, may lead to an increase in the number, migratory and proliferative potential of endothelial cells, and in doing so, may lead to improved graft survival.
One of the potential advantages of DSAEK, as opposed to penetrating keratoplasty (PK), is the ability to transplant a larger graft regardless of the proximity to the limbal vasculature. In the UK, most surgeons use an average donor trephine size of 8.5 mm for DSAEK.9 It is unclear whether this is based on an assumption that size of the trephine is the size (chord length) that a graft achieves in the eye and that an 8.5 mm graft would avoid encroachment on the angle of the anterior chamber (AC). In addition, peripheral graft thickness may increase with larger trephine sizes, particularly if the trephine is not well centred, increasing the risk or peripheral anterior synechiae.
Previous reports on DSAEK based on graft sizes of 8.5 mm up to 9 mm, however,10 ,11 did not find an association between graft size and graft failure or endothelial cell survival.4 ,11 In contrast, Lass et al12 recently reported that following PK, a greater preoperative ECD and a larger donor graft size were associated with higher ECD at 10 years from surgery. Given the significant increase in ECD beyond 9 mm,10 we wished to determine for patients undergoing DSAEK, whether it is possible to prepare and insert grafts of 9.5 mm trephine size and whether there was an influence of graft size on graft survival.
All consecutive patients who had undergone a DSAEK for Fuchs endothelial dystrophy (FED) or pseudophakic bullous keratopathy (PBK) by one surgeon with at least 1-year follow-up were included in the analysis. In the UK, following informed consent, all corneal transplant donor and recipient data for corneas issued by the corneal transplant service eye banks are maintained on a national database under supervision of National Health Service Blood and Transplant. Donor details such as age, ECD, death to enucleation and total postmortem time (death to processing of eye in eyebank) and recipient details, including preoperative risk factors (presence of inflammation, vascularisation, ocular surface disease, glaucoma) and comorbidity (retinal disease such as macular degeneration), peroperative data and surgical complications, and postoperative outcome data including best corrected visual acuity (BCVA) and graft survival recorded with at least 1 year of follow-up were analysed. Anterior segment optical coherence tomography (AS-OCT, Topcon Corp, Tokyo, Japan) was performed postoperatively in a subgroup of patients to measure graft size and thickness. The study was conducted in accord with The Declaration of Helsinki.
Donor corneo–scleral discs were placed onto the Moria Artificial Anterior Chamber (Moria Surgical, Antony, France). The donor corneal epithelium was removed and the anterior surface repeatedly dried using a Merocel sponge (Alcon Laboratories, Fort Worth, Texas, USA) to reduce corneal thickness. A Moria automated microkeratome (Moria Surgical, Antony, France) with a 350 μm head was then used to remove an anterior lamella cap. Manual dissection of the residual peripheral anterior lamella similar to the technique proposed by Lichtinger et al13 was then performed using a Mani 1.25 mm crescent blade (Altomed, Boldon, UK) (see online supplementary video). A disc of chosen trephine diameter was punched out from the endothelial side after transfer onto a Barron vacuum donor punch (Katena Products, New Jersey, USA) (figure 1). Patients with a horizontal white-to-white corneal diameter of ≥11.50 mm received either 9 mm or 9.5 mm grafts, those with <11.50 mm received 8.5–9 mm trephine size grafts. After donor graft preparation, the host was prepared as described in detail by Price.14 Briefly, a 5 mm sclera–corneal incision was made and the recipient Descemet membrane was stripped. The donor graft was transferred onto a Moria Busin glide (Moria Surgical, Antony, France) endothelial side up and a small amount of ophthalmic viscoelastic device (Healon, Abbott Laboratories, Illinois, USA) placed on the endothelium. After removal of the host Descemet's membrane, the disc was then introduced into the AC through a 5 mm section with a single-use 28-gauge forceps using the Busin ‘pull-through’ technique.15 Once the graft was unfolded and positioned centrally, a full air fill of the AC was maintained for 10 min. The corneal incision was then closed with interrupted 10–0 nylon sutures. Enough air was then removed and replaced with balanced salt solution to avoid blockage of the pupil. Postoperative corticosteroid regimen consisted of 1% prednisolone acetate eye drops instilled four times daily for the first 6 months, tapered by one drop per month to once-daily use and of chloramphenicol eye drops instilled four times a day for the first week only.
A generalised linear binary logistic model was used to test for variables associated with graft failure. Donor age, ECD, postmortem time and the size of the endothelial keratoplasty (EK) were tested as covariates, and indication (FED or PBK), risk factors, intraoperative complications and comorbidity as factors. A general linear model was used to test for variables associated with postoperative BCVA. Donor age, ECD, postmortem time, size of the EK and preoperative BCVA were tested as covariates, and indication (FED, PBK), risk factors, intraoperative complications and comorbidity as factors. A probability value (p) of <0.05 was considered statistically significant. A Bonferonni correction was made for multiple comparisons. Analysis was made using SPSS statistics software V.20.0 (IBM, Armonk, New York, USA).
A total of 131 of 174 patients were included. Forty-three patients were excluded, the first 30 due to a change in technique and also to allow for a learning curve, and the last 13 as they had not yet reached 1 year of follow-up. Of the 131 patients who were included, 84 (64%) had FED and 47 (36%) had PBK. Donor age was between 32 and 96 years (69.4±11.6 years) with a mean ECD of 2775.6 cells/mm2±196.4 cells/mm2. Death to enucleation time and postmortem time were, respectively, 6.9±121.7 and 33.2±8.0 h. Patients were classified into three groups according to the size of the donor trephine: <9 mm (8.5–8.75 mm) (n=28), 9 mm (n=52) and 9.5 mm (n=51). Risk factors were present in 32 (24%) of patients: 8 (29%) in <9 mm group, 11(22%) in 9 mm group and 13 (25%) in 9.5 mm group. Comorbidity factors were present in 25 (19%) of patients.
Mean preoperative and postoperative BCVA were 1.01±0.76 and 0.2±0.2 logMAR, respectively, at 12 months, with 80.5% achieving 20/40 or better visual acuity. There were no significant differences between preoperative BCVA (0.89±0.38, 0.88±0.60 and 1.1±0.92) and postoperative BCVA (0.28±0.38, 0.27±0.27 and 0.25±0.23) among the three trephine size groups. Postoperative BCVA was significantly associated with ECD (p=0.005), PBK or FED (p=0.004), presence of risk factors (p=0.007) or comorbidity (p=0.016) but not with donor age (p=0.47), postmortem time (p=0.47), complications (p=0.42), graft trephine size (0.98) or preoperative BCVA (p=0.13).
Eleven patients (8.40%) experienced endothelial graft failure: 17.86% for <9 mm, 7.69% for 9 mm and 3.84% for 9.5 mm trephine sized grafts. Graft failure was significantly associated with ECD (p=0.039) and graft trephine size (p=0.04), but not with postmortem time (p=0.68), donor age (p=0.35), presence or absence of risk factors (p=0.1), complications (p=0.55) and comorbidity (p=0.28), or diagnosis (FED or PBK p=0.24). There was no significant association between donor ECD and graft size (p=0.80). There was, however, an increased association with graft failure using a combination of ECD and graft size (p=0.005). The follow-up period tended to be longer for the <9 mm grafts of 676 days (SD 231), compared with 552 days (SD 183) days for 9 mm grafts and 578 days (SD 175) for 9.5 mm sized grafts. There was no significant difference in the time to failure between donor graft trephine size, that is, 648 days (SD 380) (n=5) for <9 mm, 604 days (SD 180) (n=4) for 9 mm and 352 days (n=2) for 9.5 mm trephine sized grafts (p=0.83 and p=0.16).
Adverse events and reactions
Although subjective, no difficulty was encountered in unfolding 9.5 mm trephine size grafts in the eye and these grafts centred well with almost no additional centration manoeuvres. In three of the donor corneas, evidence of previous cataract surgery was noted while preparing the donor, which required suturing of the cataract incision site in order to maintain intraocular pressure to allow automated dissection. The dissection was completed manually and 9.5 mm grafts prepared. All patients had a well-attached endothelial graft postoperatively. One patient who had 9 mm EK required graft rebubbling. In four patients, two of whom had 9 mm and two 9.5 mm grafts, the dissection of the peripheral corneal rim was incomplete, leaving a portion of thickened periphery with the development of peripheral inferior anterior synechiae in two patients. Two patients (one who received a 9 mm and the other a 9.5 mm sized graft) had reversible graft rejection without the development of graft failure at 2 years follow-up.
Graft size in the eye
The expected size of the graft in the eye was approximated as follows (figure 2). The radius of curvature of the trephine block according to the manufacturer was 9.5 mm. The circumferential arc length of the trephined graft is then the product of the trephine size (Y) and angle (radians). For example, for a trephine of 9.5 mm on a trephine base of radius R1 of 9.5 mm, the arc length is 9.5mm×α1 where α1 can be calculated from the function, . Therefore, for a 9.5 mm sized graft, or 1.05 radians (60°), giving a circumferential arc length C of 9.95 mm. If the graft assumes a radius of curvature on the posterior corneal surface R2 in the eye of 5.70 mm, then or 1.75 radians (100°). The expected chord length (Y2) of the graft in the eye is therefore or or 8.73 mm. The predicted maximum and measured maximum chord lengths of the posterior cornea are shown in table 1 for four eyes with differing posterior radius of curvatures. The mean posterior radius of curvature was calculated by measuring the sag heights for several chord lengths in each eye. In 12 patients, AS-OCT (Topcon, Tokyo, Japan) confirmed adhesion of the graft with clearance from the iris and angles (figure 3).
There has been a rapid uptake of EK for FED and PBK, predominantly DSAEK, although the Descemet membrane endothelial keratoplasty (DMEK) is quickly gaining recognition. For DSAEK, most surgeons use an average donor trephine size of 8.5 mm9 possibly based on the assumption that this is the size that graft achieves in the eye and to reduce the risk of peripheral anterior synechiae. In many centres, the donor posterior lenticule is prepared by an eye bank, which limits the size of the graft that can be inserted into the eye. Graft failure rate following DSAEK is almost exclusively endothelial failure and in a small number of cases endothelial failure following rejection.9 Although the low incidence of rejection in this study would suggest that graft size is not a risk factor, a larger data set would be required to address this. Although there are many factors that contribute to graft endothelial failure, the health of the donor endothelium is likely to be pivotal. It is reasonable to suggest, therefore, that transplanting healthier endothelial cells may lead to improved graft survival.
ECD increases towards the paracentral and peripheral regions of the cornea.1 ,3 Schimmelpfennig showed a non-uniform distribution of endothelial cells across the cornea with a peripheral ECD of 3632±592 cells/mm2 compared with a central ECD of 2778±284 cells/mm2.1 Amann et al3 compared ECD using four different methods: two in vivo specular microscopy measurements and two in vitro staining methods. The paracentral region was defined as 2.2–3.3 mm from the centre of the cornea and the peripheral cornea as between 4.2 and 5.3 mm from the central cornea, depending on whether non-contact specular microscopy (ROBO SP 8000) contact (Keeler-Konan SP-1 contact-specular microscope) method or whole mount and histological sections were used. Using these measurements, a graft size of 8 mm incorporates the paracentric region, which has an increased ECD of at least 5.8% compared with the central cornea.3 In contrast, the ECD can increase in the peripheral region by as much as 22.9%. Even across a wide age range of donors (3–75 years) ECD appears to decrease at the slowest rate in the peripheral region (reflected by increased telomerase activity) signifying proliferation and greater regeneration potential.7 ,8 Hence, if graft size is increased to 9.5 mm, more of the periphery will be included and may help in maintaining an adequate number of functional endothelial cells in the long term.7 ,8 ,16 Using the method proposed by Unterlauft et al17 for estimating the area of the endothelial corneal surface, grafts of size 8.5, 9 and 9.5 mm will have areas of approximately 128.7, 142.3 and 156.9 mm2.3 If the preoperative ECD of the donor corneo–scleral disc was 2700 cells/mm2, an increment from 8.5 mm diameter to a 9.5 mm diameter graft would result in transplanting of approximately 424 000 cells rather than 350 000 cells. Although speculative, this may potentially translate to a higher graft survival, as it has also been supposed by Anshu et al.11
Two previous studies, however, did not find a difference in graft survival with graft size following DSAEK. In these studies, graft size comparisons were limited to between 8.0 –8.5 mm in one10 and 8.5–9.0 mm in another.11 Extending the graft size beyond 9 mm may be pivotal, and this is the first report comparing survival between grafts of <9.0 and –9.5 mm. Despite the limitations of this study, there appears to be a significant association between graft size and ECD and the rate of graft failure; that is, increased graft size and donor ECD was significantly associated with a reduced graft failure rate. The overall rate of endothelial graft failure of 7.64% in this study is comparable with that reported by Dapena et al18 (0.5%–29%). Although the failure rate of 17.86% for <9 mm grafts was high, this is still in keeping with the UK national average of graft survival at one year of 86% (95% CI 84 to 88) for FED and 80% (95% CI 78 to 82) for PBK.9 In contrast, the failure rates for the 9 mm and, in particular, 9.5 mm trephine sized grafts were much lower at 6.56% and 3.64%, respectively. The main limitation of this study, however, is that the minimum follow-up was at least 1 year and it is unclear whether difference in graft survival with size and ECD will be maintained at 5 years or longer. Although the period of follow-up was longer for patients who had <9 mm sized grafts, there was no significant difference in the time to failure between graft size. In addition, although we did not measure central postoperative ECD, this is not uniform across the human cornea, and central ECD measurements would not distinguish between smaller or larger DSAEK. Nevertheless, it would be of interest to measure changes in both central and peripheral ECD and endothelial morphology according to graft size, with increasing postoperative time. In 80.5% of patients, postoperative BCVA was ≥20/40, which is in keeping with previous reports.18 Postoperative BCVA was associated with ECD, risk factors, comorbidity and the indication for DSAEK (PBK or FED). The association with donor ECD with postoperative BCVA is of interest, but we do not have sufficient data to determine if there was an association between central corneal thickness, donor ECD and postoperative ECD. Similarly, we do not have data to determine whether the time to achieve BCVA varied among the three groups. Although we did not find an association between donor age, postmortem time and graft survival, this may reflect the high prevalence of older donors (mean age of 69.4 years±11.6) and long mean postmortem time (33.2±8.0 h) with very few young donors or short postmortem times.
Although the technique used to remove the peripheral rim of the anterior donor corneal stroma13 is fairly straightforward, in four patients the periphery of the graft had not been adequately thinned by the dissection and two of the four patients developed inferior anterior synechiae but with no further related complications. It is noteworthy that the size of the graft in the eye is approximately 0.5–0.75 mm less than the trephine size due to the difference between the radius of curvature of the trephine block and host cornea. This reduction in the actual size of the graft in the eye, together with a thinned graft periphery, allows the surgeon the option of transplanting a larger graft without increasing the risk of peripheral anterior synechiae and engagement of the iris–corneal angle (at least in the pseudophakic eye). Using OCT in a small number of patients there was good clearance of the graft edge from the iris and angle. In addition, although subjective, there appeared to be easier handling and centration of the 9.5 mm trephine size grafts. This may not be the case in the presence of a shallow AC.
Despite the limitations of this study, there appears to be an effect of graft size (up to 9.5 mm) and ECD on graft survival following DSAEK. This report shows that using a modified technique suggested by Lichtinger et al13 it is possible to prepare a larger graft without compromising the anterior segment. By creating a large lamellar donor graft it is possible to transplant a greater number of cells and potentially increase long-term graft survival without compromising the rate of perioperative and postoperative complications. It will be of interest to determine whether graft survival following DMEK similarly increases with larger-sized grafts.
Twitter Follow Sajjad Ahmad at @s2ahm
Contributors VR, AT, SBK, SA, CW, MB: involved in design and study. CB, NMEH, VR, AT, SK, SA, CW, MB: involved in management, statistical analysis and interpretation of the data, and preparation of the manuscript.
Competing interests None.
Patient consent Obtained.
Ethics approval National Health Service Blood and Transplant.
Provenance and peer review Not commissioned; externally peer reviewed.