Impact of graft thickness on visual acuity after Descemet's stripping endothelial keratoplasty
- Anthony J Shinton1,
- Michael Tsatsos2,
- Aris Konstantopoulos2,
- Srini Goverdhan2,
- Ahmad F Elsahn2,
- David F Anderson2,
- Parwez Hossain1,2
- 1Division of Infection, Inflammation and Immunity, Faculty of Medicine, University of Southampton, Southampton, UK
- 2Southampton Eye Unit, Southampton General Hospital, Southampton, UK
- Correspondence to Parwez Hossain, University of Southampton, Eye Unit, MP104, Southampton General Hospital, Tremona Road, Southampton SO16 6YD, UK;
Contributors AS: data collection, statistical analysis, conception and design, wrote manuscript, revised manuscript. MT: conception and design, revised manuscript. AK: conception and design, ethics application, revised manuscript. SG: conception and design, revised manuscript. AFE: conception and design, revised manuscript. DFA: conception and design, revised manuscript. PH: conception and design, revised manuscript.
- Accepted 2 October 2011
- Published Online First 25 October 2011
Objective To evaluate the relationship, over time, between central graft thickness and visual acuity following Descemet's stripping endothelial keratoplasty (DSEK).
Methods A retrospective analysis of 70 consecutive cases of DSEK. All donor lenticules were dissected manually. Serial postoperative measurements of central graft and total corneal thicknesses were made using anterior segment optical coherence tomography. Visual acuity, refraction and patient demographics were collected from case notes. The correlation between central graft thickness and visual acuity at serial time points was calculated.
Results The median age at surgery was 75 years (lower quartile (LQ) 66, upper quartile (UQ) 83, range 36–90 years). Nineteen eyes were excluded from statistical analysis, leaving 51 eyes of 46 patients remaining. Last follow-up occurred a median of 12 months postoperatively (LQ 6, UQ 23, range 4–38 months). The median preoperative visual acuity was 0.71 logarithm of the minimum angle of resolution (logMAR), improving to 0.34 logMAR postoperatively (p<0.001, n=43). Median graft thickness decreased from 209 μm at day 1 to 142 μm at last follow-up (p<0.001). No statistically significant correlation was found between central total corneal thickness and visual acuity at any time point. Except for a single time point, no statistically significant correlation was found between central graft thickness and visual acuity.
Conclusion There is no clear association between central graft, or total corneal, thickness and visual acuity following DSEK.
- corneal transplantation
- Descemet stripping endothelial keratoplasty
- Fuchs' endothelial dystrophy
- ocular surface
- optical coherence tomography
- sclera and episclera
- stem cells
Penetrating keratoplasty has been largely superseded by selective endothelial keratoplasty procedures for the treatment of corneal endothelial dysfunction. Advantages of this new approach include earlier visual recovery, reduced postoperative discomfort and greater refractive stability.1–3 The surgical technique has evolved considerably since the first description of posterior lamellar keratoplasty in 1998.4 Posterior lamellar keratoplasty was soon succeeded by deep lamellar endothelial keratoplasty.5 Descemet's stripping endothelial keratoplasty (DSEK), with manual dissection of the graft tissue, was established in 2004.6 7 This is the technique currently used at Southampton Eye Unit. The host corneal endothelial layer is removed and replaced with a lenticule of corneal tissue containing the posterior corneal stroma and endothelial layer.
More recent modifications to the technique have focused on using thinner donor lenticules. In Descemet's membrane endothelial keratoplasty (DMEK) the graft does not include any stromal tissue.8 DMEK therefore represents pure anatomical restoration. Another proposed modification is ultrathin Descemet's stripping automated endothelial keratoplasty (DSAEK). This involves automatic dissection of the graft to a thickness of less than 100 μm. The main hypothesis for these developments is that a thinner graft lenticule leads to an improved visual outcome. The relationship between graft thickness and visual acuity, however, remains controversial. Several studies investigating this relationship have produced conflicting evidence.9–14 Studies using manually dissected grafts are especially limited. Our goal was to add further weight to this important debate.
Patients and methods
Ethics committee approval was obtained. The study adhered to the tenets of the Declaration of Helsinki. Outcomes from Southampton Eye Unit's initial 70 consecutive cases of DSEK (August 2007 to February 2011) were analysed retrospectively. Theatre records were used to identify patients who had undergone DSEK. The case notes of all patients were reviewed.
Visual and refractive data
Refractive error data and patient demographics were collected. Visual acuity (VA) was recorded preoperatively, at 1, 3, 6 and 9 months postoperatively, and at last follow-up. VA, rather than best corrected visual acuity (BCVA), was used due to the availability of data at regular time intervals. A single BCVA value, measured at manifest refraction at least 4 months postoperatively, was also collected from the case notes. When VA and BCVA were measured in Snellen, they were converted to a logarithm of the minimum angle of resolution (logMAR) to facilitate statistical analysis. Counting fingers, hand movements and perception of light were assigned the arbitrary values of 1.60, 1.90 and 2.20 logMAR, respectively. The same assumptions have been made by others.15 16
Anterior segment optical coherence tomography (Visante AS-OCT; Carl Zeiss Meditec, Dublin, California, USA) scans were performed by ophthalmic imagers at the patients' follow-up appointments. The Visante AS-OCT has an optical axial resolution of up to 18 μm. High-resolution corneal scans were taken at the horizontal meridian. Scans were centered on the vertex reflection of the central cornea. The date of surgery and the scan date were used to assign each scan to the appropriate time category. Central graft and central total corneal thickness measurements were made using the flap tool available on the Visante AS-OCT software (version 1.1.2). Thickness measurements were taken when the graft–host cornea interface was clearly visible, as close to the apical centre of the cornea as possible. All measurements were made by a single trained operator.
Tissue preparation and surgical technique
Donor corneas were provided by the UK Corneal Transplant Service. The donor cornea was placed on an artificial anterior chamber (Moria, Doylestown, Pennsylvania, USA). Pachymetry measurements (SP-3000 Pachymeter; Tomey Corporation, Japan) of the central corneal thickness were taken and used as a guideline for the corneal dissection. The donor cornea was incised at the limbus either at the same depth as the average of the central corneal thickness measurements (DFA) or at 90% of the depth (PNH). The whole donor cornea was dissected following the same lamellar plane using the Moria lamellar dissector. The donor lenticule was then punched using an 8.50, 8.75, 9.00, 9.25, or 9.50 mm trephine (Barron Precision Instruments, Michigan, USA). Descemetorhexis of the host cornea was performed using a reverse Sinsky hook or a nuclear splitter. In cases performed before April 2010, Goosey forceps (Moria) were used to introduce the graft into the anterior chamber. An EndoSaver (SD Healthcare Ltd, Manchester, UK) was used by DFA in 21 subsequent cases. In 12 cases, DSEK was combined with phacoemulsification and intraocular lens implantation as a triple procedure.
The data were checked for normality using the Shapiro–Wilk test (SPSS for Mac version 18.0) before statistical evaluation. Normally distributed data were evaluated using the paired Student's t test. Descriptive statistics for non-normally distributed variables are reported as the median, upper quartile (UQ), lower quartile (LQ) and range. Non-normally distributed data were analysed with the Wilcoxon signed rank test, Mann–Whitney U test and Spearman's rank correlation (rs). Statistical analysis was performed using SPSS for Mac V.18.0. Median differences, and their CI, were calculated using CI analysis software (version 2.2, University of Southampton, Southampton, UK). A p value less than 0.05 was considered to be statistically significant.
Seventy eyes of 61 patients underwent DSEK at our institution between August 2007 and February 2011. Nineteen eyes were excluded: graft failure (n=11), pre-existing retinal pathology (n=5); deceased (n=3). The rate of graft failure was 15.7%. Statistical analysis was performed on the remaining 51 eyes of 46 patients. DSEK was performed by one of three surgeons: DFA (n=37), PNH (n=12), JAM (n=2). Thirty-eight eyes were pseudophakic at the time of DSEK. In 12 eyes, DSEK was combined with phacoemulsification and intraocular lens implantation as a triple procedure. One eye remained phakic following DSEK, and lens opacities were minimal in this eye at 1 year postoperatively. Patient demographics are shown in table 1. The median age of corneal donors was 76.4 years (LQ 67.5, UQ 82.3, range 49.1–92.7 years). The median endothelial cell count of donor corneas, as measured by the eye bank, was 2600 cells mm−2 (range 2000–3000 cells mm−2).
Descriptive statistics for VA at serial time points are presented in table 2. LogMAR VA was significantly lower at 1 month postoperatively than preoperatively (p=0.036, n=47). LogMAR VA was significantly lower at 3 months than at 1 month (p=0.001, n=42). LogMAR VA was significantly lower at 6 months than at 3 months (p<0.001, n=34). There was no statistically significant difference in VA between 6 and 9 months (p=0.301, n=20). Last follow-up (≥4 months postoperatively) occurred a median of 12 months postoperatively (LQ 6, UQ 23, range 4–38 months). There was no statistically significant difference in VA between 9 months and the final measurement (p=0.691, n=23). The median difference between VA preoperatively and at last follow-up was 0.39 logMAR (95% CI 0.24 to 0.56, p<0.001, n=43).
BCVA was measured a median of 13 months postoperatively (Q1 6, Q3 20, range 4–38 months). The median postoperative BCVA was 0.32 logMAR (LQ 0.28, UQ 0.46, range 0.02–0.70, n=36).
The mean preoperative refractive cylinder was 1.17±1.13 diopter (range 0.00–4.25, n=34). The mean postoperative refractive cylinder in those eyes was 1.45±1.00 diopter (range 0.00–4.50). This was not a statistically significant difference (p=0.270, paired Student's t test). The mean preoperative spherical equivalent was 0.06±2.20 diopter (range −6.25–3.88, n=34). The mean postoperative spherical equivalent in those eyes was 0.11±1.36 diopter (range −4.13–3.00). This was not a statistically significant difference (p=0.887, paired Student's t test).
Graft and corneal thickness
Analysis of serial central graft thickness measurements is presented in table 3. Graft thickness was significantly lower at week 1 than at day 1 postoperatively (p<0.001, n=21). Graft thickness was significantly lower at month 1 than at week 1 postoperatively (p<0.001, n=25). There was no statistically significant reduction in graft thickness between two consecutive time points beyond month 1. Last follow-up thickness measurements (≥4 months postoperatively) were made a median of 14 months postoperatively (LQ 6, UQ 22, range 4–38 months). The median difference between graft thickness at day 1 and last follow-up was 60 μm (95% CI −38 to −88, p<0.001, n=24). This was a reduction of 29% from graft thickness at day 1.
Median total corneal thickness at day 1 was 821 μm (LQ 742, UQ 984, range 548–1265, n=32). Median total corneal thickness at last follow-up was 618 μm (LQ 583, UQ 685, range 524–918, n=39). The median difference between total corneal thickness at day 1 and at last follow-up was 180 μm (95% CI 134 to 237, n=43). This was a reduction of 22% from total corneal thickness at day 1.
Analysis of correlation between VA and central graft or total corneal thickness is shown in table 4. At 1 month postoperatively, the correlation between graft thickness and VA was very weak (rs=0.196). At 6 months postoperatively, there was a statistically significant correlation of moderate strength between graft thickness and VA (rs=0.460, p=0.041). For each eye, the last measurement of VA and the last AS-OCT scan occurred a similar length of time postoperatively (median difference 0.0, LQ 0.0, UQ 1.5 months). At last follow-up, the correlation between central graft thickness and VA was very weak (rs=0.010). No statistically significant correlation was found between total corneal thickness and VA at any time point.
The eyes were divided into two groups depending on their central graft thickness at last follow-up. The VA at last follow-up of these two groups was compared using the Mann–Whitney U test. When 100 μm (the definition of ‘ultrathin’ DSEK) was used as the cut-off, no statistically significant difference existed (p=0.502). When 142 μm (the median graft thickness at last follow-up) was used as the cut-off instead, a significant difference was still not found (p=0.944).
We saw no significant correlation between graft thickness and VA at any time point except 6 months. The statistically significant finding at 6 months may represent a type I error due to repeated statistical testing. Furthermore, only 20 pairs of data were available at 6 months, compared with 39 at last follow-up. The correlation between graft, or total corneal, thickness and VA at last follow-up was very weak. To our knowledge, this is the first study to investigate the correlation between graft thickness and VA at multiple time points. Furthermore, ours is the largest study of manually dissected graft thickness and VA.
The influence of graft thickness on VA has been the subject of several studies, yet remains controversial.9–14 We are aware of two studies that have reported a significant relationship between graft thickness and VA after DSAEK.11 14 Neff et al14 retrospectively analysed 33 cases of DSAEK using precut tissue. The eyes were divided into thin and thick graft groups, either side of the median postoperative graft thickness (131 μm). The thin graft group had a significantly better postoperative best spectacle-corrected visual acuity (BSCVA) than the thick graft group (p<0.01). When we compared eyes either side of the median graft thickness at last follow-up (142 μm), no such difference in VA was found.
Pogorelov et al11 measured central corneal and graft thickness 6 months after DSAEK in 15 eyes. They concluded that BSCVA correlates significantly with both central corneal and graft thickness (Pearson correlations −0.745 and −0.589, respectively, p<0.05).11 Chen et al2 measured corneal thickness with ultrasonic pachymetry 6 months after DSAEK. They found a weak, but statistically significant, correlation with postoperative BSCVA (r2=0.117, p=0.001).2
A larger number of studies have reported no significant correlation between graft thickness and VA after endothelial keratoplasty.9–13 17 Terry et al17 measured preoperative graft thickness in 65 cases of DSAEK. Graft thickness did not correlate significantly with improvement in vision (Pearson correlation =0.114, p=0.439). Neff et al14 noted that the mean graft thickness in those studies was higher than that of their study. This was considered to be a potential reason why earlier studies had not demonstrated a correlation. The median graft thickness of our eyes at last follow-up (142 μm) was only marginally higher than that of Neff et al14 (131 μm).
Price and Price3 conducted a study comparing the outcomes of microkeratome versus manual dissection of the graft. In those 200 cases, no significant correlation was found between central corneal thickness and BSCVA at 6 months postoperatively (p=0.25).3 Corneal thickness does not influence VA following deep lamellar endothelial keratoplasty either (r2=0.03).18
When the donor lenticule includes no stromal tissue, as in DMEK, visual advantages have been demonstrated.19 Unfortunately, DMEK is technically demanding and DMEK grafts are prone to tears during harvesting.19 The use of ultrathin grafts, of 100 μm thickness or less, has been proposed as an easier alternative procedure. There are currently limited published data regarding grafts of this thickness. Fourteen of the 51 grafts in our study (27.5%) were thinner than 100 μm at last follow-up, but this subgroup did not achieve a better final VA.
To avoid distorting any relationship, we excluded 16 eyes whose VA was impaired by either graft failure or pre-existing retinal pathology. We report a graft failure rate of 15.7%. While this is higher than some published figures, it should be remembered that this cohort represents the surgeons' early cases. It may therefore be related to an initial surgical learning curve. A systematic review of the safety and outcomes of DSEK and DSAEK found a mean primary graft failure rate of 5% (range 0–29%), and a mean graft rejection rate of 10% (range 0–45.5%).1 The relationship between graft thickness and VA may be confounded by graft viability.
Our mean postoperative BCVA (0.36 logMAR or 20/46 Snellen, n=36), measured between 4 and 38 months after DSEK, compares favourably with other series using manually dissected grafts.3 13 While there is evidence that DSAEK has superior visual outcomes, it requires an expensive microkeratome.3 17 Pre-cut tissue is not currently available in the UK. In our study, the median logMAR VA improved at each successive time point. There was no statistically significant difference, however, between 6 and 9 months. This supports published data that show visual rehabilitation occurs within 6 months postoperatively.3 7 Central graft thickness stabilised by 1 month postoperatively, much earlier than VA. Factors other than falling graft thickness must contribute to the improvement in VA.
Price and Price3 reported that DSEK induced a significant hyperopic shift, whereas DSAEK did not. We found no significant difference between pre and postoperative values for spherical equivalent or refractive cylinder. Our results support the conclusion of Rice et al:13 a refractive-neutral result can be achieved with manual dissection of the graft. We recognise, however, that measurement of refraction in eyes with corneal decompensation is unreliable.
The major limitation of this study is its retrospective nature. Preoperative and serial postoperative BCVA were often not available. The use of VA instead is undeniably less reliable, but should not have affected our conclusions. For measurements of both VA and thicknesses, the number of data points falls progressively at longer postoperative time points. This is partly because patients had not yet reached these time points, and partly due to missed follow-up.
The central thickness of manually dissected donor grafts does not appear to affect VA. Similarly, no significant correlation was found between central corneal thickness and VA after DSEK. Dissection of thinner lenticules of graft tissue may be unnecessary.
Competing interests None.
Ethics approval Ethics committee approval was obtained from Southampton and South West Hampshire Local Research Ethics Committee. The study adhered to the tenets of the Declaration of Helsinki.
Provenance and peer review Not commissioned; externally peer reviewed.