To compare national outcomes of endothelial keratoplasty (EK) and penetrating keratoplasty (PK) during comparable 6-year periods.
Prospective cohort study of national registry data.
setting : United Kingdom National Transplant Registry, 2000 through 2011, inclusive. patient population : All United Kingdom patients undergoing first EK (n = 2074) for Fuchs endothelial dystrophy or pseudophakic bullous keratopathy from January 2006 through December 2011. Comparison cohort of patients undergoing first PK (n = 2622, same indications, January 2000 through December 2005). observation procedure : Year of surgery, surgeon and center experience, corneal diagnosis, donor factors, patient factors, and surgical risk factors were analyzed against graft survival and visual outcomes.
For both Fuchs endothelial dystrophy and pseudophakic bullous keratopathy, EK achieved better average best-corrected acuity and lower refractive error. For both groups, graft failure was significantly higher for EK than PK. EK failure in Fuchs endothelial dystrophy was associated with center experience (hazard ratio [HR], 2.3; P < .0001), donor endothelial density (HR, 1.8; P = .01), glaucoma at time of surgery (HR, 2.1; P = .003), and donor age older than 75 years (HR, 1.3; P = .05). EK failure in pseudophakic bullous keratopathy was associated with center experience of fewer than 15 cases (HR, 2.0; P < .0001) and glaucoma at time of surgery (HR, 1.7; P = .002).
Prospective national registry data for EK showed higher graft failure than is seen in PK or in retrospective case series of EK. Higher failure rates may be acceptable given established benefits of the procedure, including lower refractive error, structural globe integrity, and faster visual recovery. Center experience influenced EK survival more than surgeon experience, and overall surgical outcomes may be improved by standardized techniques and support within experienced units.
Endothelial keratoplasty (EK) currently is the preferred treatment for irreversible corneal endothelial decompensation. Attempts to replace endothelium selectively extend back to the 1950s; however, it was Melles’ original work in the 1990s with subsequent stepwise developments that resulted in the most popular current techniques: Descemet stripping endothelial keratoplasty (DSEK) and Descemet stripping automated endothelial keratoplasty (DSAEK).
Recognized advantages of EK over penetrating keratoplasty (PK) include earlier visual recovery, better refractive outcomes, greater structural globe integrity, and comparable or possibly reduced graft rejection. In addition, longitudinal series from experienced surgeons suggest that EK can achieve similar long-term graft survival and endothelial cell loss to PK. A report in 2009 cited an average iatrogenic primary graft failure rate of 5% (range, 0% to 29%), 1-year graft survival rate (including cases of iatrogenic primary graft failure) of 94% (range, 55% to 100%), rejection episode rate of 10% (range, 0% to 45%), and endothelial cell loss rate of 42% at 12 months (range, 16% to 45%). The authors concluded that EK was effective and safe with graft survival, complication rates, surgical risks, visual acuity, and endothelial cell loss similar to that of PK.
Importantly, literature on clinical outcomes in EK still derives mostly from case series of high-volume centers and experienced single surgeons. Of the 13 individual centers included in the 2009 report, all were regional referral centers for the procedure. Furthermore, frequently cited EK articles often are drawn from serial publications (sometimes with overlapping patient inclusion) of a small number of individual units with excellent results. Although these outcomes have been cited as being average results for the procedure, in reality, they are averages drawn from a relatively small number of high-volume centers. In fact, it is not yet known whether equivalent results are achieved in smaller units or by less experienced surgeons, although the procedure is performed widely across a wide range of settings, Surgical outcomes from these settings have been reported infrequently and probably remain underrepresented in the literature to date. A recent study from the Australian graft registry concluded that the outcomes of EK overall, when evaluated on nationwide data, were much worse than the case series literature has suggested, and resulted in increased donor tissue requirements and potentially increased patient morbidity.
We previously examined the effect of surgeon experience and center case volume on PK outcomes in the United Kingdom for patients receiving first transplants for a range of indications. We found no significant difference in long-term (1- and 2-year) graft survival and minor differences in acuity and refractive outcomes, regardless of surgeon experience and center volume. EK commonly is regarded as more experience dependent and technique dependent than PK, with evidence in support of this including high rates of early, iatrogenic primary graft failure during initial case series, even among experienced corneal surgeons. Although it is likely that late (1- and 2-year) graft survival also is influenced by surgeon experience and possibly center experience with the procedure, this question has not been addressed widely to date.
We therefore set out to compare effects of surgeon and center experience on 1- and 2- year EK outcomes from nationwide United Kingdom data between 2006 through 2011. Historical PK outcomes (2000 through 2005) were used to compare more recent EK data with PK data at a time when PK was still the mainstream primary treatment for endothelial failure. For further standardization, we limited analysis to eyes in which the indication for transplant was either Fuchs endothelial dystrophy or pseudophakic bullous keratopathy (PBK).
The study data were drawn from prospective data collected for the United Kingdom transplant registry. The transplant registry maintains a national database of all solid organ and tissue donors, transplant recipients, and procedures performed in the United Kingdom. The described research methods and analysis plan adhered to the tenets of the Declaration of Helsinki and United Kingdom law in relation to data collection and medical research. Institutional review board approval was provided by the National Health Service Blood and Transplant Ocular Tissue Advisory Group, which provided approval for ongoing data collection in the national transplant register and subsequent analysis.
Patients and Transplants
Analysis cohorts included (1) all United Kingdom patients receiving a first PK for Fuchs endothelial dystrophy or PBK between January 2000 and December 2005 or (2) a first endothelial keratoplasty for Fuchs endothelial dystrophy or PBK between January 2006 and December 2011. In this report, the term endothelial keratoplasty was used regardless of donor dissection technique and included DSEK and DSAEK cases. Transplants without reported follow-up were excluded from analysis.
In the United Kingdom, the National Health Service Blood and Transplant collects corneal transplant data prospectively at the time of surgery and at 1, 2, and 5 years after surgery. Provision of data by surgeons is a requirement for all corneal transplantation in the United Kingdom. Standardized data forms, including corneal diagnosis, previous surgery, risk factors for graft failure, surgical technique, and surgeon and center details are completed at initial surgery and each postoperative interval until graft failure, patient death, or loss to follow-up. The United Kingdom Transplant Registry collects and maintains nationwide data on all solid organ and tissue donation.
From the data available, the following independent variables were used in the analysis: center experience and surgeon experience (both according to the number of previous EK procedures undertaken before the referenced graft: fewer than 15 cases, 15 to 29 cases, or 30 cases or more); primary corneal diagnosis (limited to Fuchs endothelial dystrophy and PBK); donor age; donor gender; donor endothelial cell density (ECD); recipient age; recipient gender; year of transplantation; and preoperative risk factors for PK graft failure, including perioperative inflammation, host neovascularization, diagnosed glaucoma in the operated eye (before surgery), and concurrent surgical procedures at the time of grafting.
Primary outcome measures included the interval to graft failure, summarized in 1- and 2-year graft failure rates, with failed grafts defined as those for which the treating surgeon documented failure on follow-up forms or there was a subsequent record in the database of a replacement graft in the same eye.
Secondary outcome measures included measures of graft function, including best-corrected visual acuity and day-to-day acuity. Day-to-day acuity was the acuity recorded in the treated eye wearing the preferred correction by spectacle lens or contact lens or unaided. Best-corrected visual acuity before transplant was compared with best-corrected visual acuity and day-to-day acuity at 2 years (range, 18 to 30 months). At this interval, most transplant recipients have had suture removal, astigmatism management, and a refractive correction provided where necessary. Snellen measurements were converted to logarithm of the minimal angle of resolution equivalents for purposes of analysis.
Statistical analyses were performed with SAS software version 9.3 (SAS Institute, Inc, Cary, North Carolina, USA). Kaplan-Meier survival curves were used to compare univariate differences in graft survival between (1) EK and PK, (2) concurrent 2-year periods during which transplants were undertaken, (3) strata for center and surgeon volume, and (4) donor ECD. A Cox proportional hazards regression model was fitted to investigate the influence of surgeon and center activity levels on graft survival while adjusting for other known significant factors. Variables were tested in a stepwise multivariate analysis and were omitted where they did not contribute to the final Cox model. Ninety-five percent confidence intervals (CIs) are shown. Visual acuity and refractive outcomes, as indicators of transplant function, were compared between the PK and EK groups and between the Fuchs endothelial dystrophy and PBK groups using χ 2 tests. There was no attempt to exclude or adjust for contralateral eyes of a single patient. Refractive data were analyzed according to the methods described by Long and by Kaye and Harris. These methods allow an analysis of the total refractive error and are not subject to statistical flaws in analysis of the cylinder or sphere as independent variables. Multivariate analyses of variance techniques were used to investigate differences in mean refractive error between PK and EK groups. Differences in variance of the refractive errors between the groups were analyzed using F tests.
Patient summary data, indication for surgery (Fuchs endothelial dystrophy or PBK), and procedure (PK or EK) are shown in Table 1 . Patients receiving grafts in the 2 periods were similar in terms of demographic data.
|Recipient age (y)|
|0 to 59||185||13.5||137||10.8|
|60 to 69||336||24.5||346||27.1|
|70 to 79||531||38.7||477||37.4|
|80 to 89||304||22.1||292||22.9|
|Recipient age (y)|
|0 to 59||105||8.4||67||8.4|
|60 to 69||176||14.2||108||13.5|
|70 to 79||497||40.0||302||37.8|
|80 to 89||411||33.1||289||36.2|
Factors Conferring Risk of Graft Failure
The proportion of eyes with established risk factors for graft failure after PK was lower in the EK than in the PK group. In the EK group, there was less recipient corneal vascularization at surgery, better preoperative acuity, and a lower proportion of eyes having additional noncataract procedures at the time of surgery. In Fuchs endothelial dystrophy eyes undergoing PK and EK, similar proportions had combined keratoplasty and cataract surgery (data not shown).
Surgical Technique by Year
From 2000 to 2012, an increasing proportion of first transplants in Fuchs endothelial dystrophy and PBK were EK procedures ( Figure 1 ). The 6-year intervals selected (January 2000 through December 2005 for PK and January 2006 through December 2011 for EK) were chosen subjectively based on several factors. Before and including 2005, nearly all EK procedures were deep lamellar EK, which is performed rarely and is of limited relevance to modern practice. Second, during this period, the preferred technique for the large majority of surgeons managing endothelial failure (in the United Kingdom) was still PK. Finally, after 2006 almost all EK procedures were DSEK or DSAEK, which rapidly supplanted PK as preferred management by the end of 2007. No Descemet’s membrane EK procedures were recorded in the registry during the studied period.
Graft Survival after Penetrating Keratoplasty Compared with Endothelial Keratoplasty
Two-year graft survival rates for PK were significantly higher than EK for Fuchs endothelial dystrophy and PBK patients ( Figure 2 ). Survival rates in Fuchs were 94% (95% CI, 92% to 96%) for PK and 81% (95% CI, 78% to 84%) for EK; in PBK, survival rates were 79% (95% CI, 76% to 82%) for PK and 70% (95% CI, 66% to 74%) for EK.
Endothelial Keratoplasty Survival by Year of Transplant
To determine improvement in EK outcomes over the period studied, we compared graft survival rates stratified by 2-year intervals for date of surgery ( Figure 3 ). Two-year graft survival rates for Fuchs endothelial dystrophy were 78% (95% CI, 72% to 84%) for 2006 through 2007, were 79% (95% CI, 74% to 82%) for 2008 through 2009, and improved to 84% (95% CI, 80% to 88%) for 2010 through 2011. In PBK, 2-year graft survival rates were 72% (95% CI, 64% to 80%) for 2006 through 2007, were 66% (95% CI, 60% to 72%) for 2008 through 2009, and improved to 71% (95% CI, 66% to 76%) for 2010 through 2011. Although there was a trend toward improvement over the 6 years, EK survival rates did not exceed PK survival rates for any examined period.
Endothelial Keratoplasty Survival by Preoperative Donor Corneal Endothelial Cell Density
Graft survival rates were significantly lower ( P = .006) at 2 years in eyes with Fuchs endothelial dystrophy where donor ECD was 2400 cells/mm 2 or fewer: 72% (95% CI, 64 to 80) compared with 82% (95% CI, 80 to 84) where ECD was more than 2400 cells/mm 2 ( Figure 4 ). Two-year graft survival rates also were lower for PBK eyes where donor ECD was 2400 cells/mm 2 or less, but the difference was not statistically significant: 64% (95% CI, 54 to 74) compared with 71% (95% CI, 66 to 74; P = .3 for ECD of more than 2400 cells/mm 2 ).
Endothelial Keratoplasty Survival by Donor Age
Donor age older than 75 years was associated with increased graft failure for patients undergoing surgery for Fuchs endothelial dystrophy ( P = .05). This finding was independent of ECD.
Endothelial Keratoplasty Survival by Surgeon Experience
Patients with Fuchs endothelial dystrophy whose surgeons performed fewer than 15 previous EK procedures had significantly lower ( P < .0001) graft survival at 2 years: 69% (95% CI, 64 to 74) compared with 85% (95% CI, 80 to 88) for 15 to 29 procedures and 87% (95% CI, 84 to 90) for 30 procedures or more ( Figure 5 ). Two-year graft survival rates also were significantly lower ( P = .0001) for PBK if the surgeon had performed fewer than 15 EKs previously: 61% (95% CI, 54 to 68) compared with 73% (95% CI, 64 to 80) for 15 to 29 procedures and 74% (95% CI, 68 to 80) for 30 procedures or more. After center experience was included in the analysis, surgeon experience was not found to be significant, and therefore it was not included in the final Cox model.
Endothelial Keratoplasty Survival by Center Experience
In centers with fewer than 15 previous EK procedures, 2-year graft survival rates for eyes with Fuchs endothelial dystrophy were significantly lower ( P < .0001): 69% (95% CI, 64% to 74%) compared with 81% (95% CI, 74% to 86%) for 15 to 29 procedures and 87% (95% CI, 84% to 90%) for 30 procedures or more ( Figure 6 ). Two-year graft survival rates also were significantly lower ( P = .0001) for PBK in centers with fewer than 15 previous EKs: 60% (95% CI, 52% to 66%) compared with 71% (95% CI, 62% to 78%) for 15 to 29 procedures and 74% (95% CI, 68% to 78%) for 30 procedures or more.
Factors Affecting Graft Survival
Individual factors tested for inclusion in the Cox model are listed in Table 2 . Those factors that were found to influence graft survival on multivariate analysis were included in the final model, shown in Table 3 .
|Risk Factor||P Value|
|Donor age a||.05||.14|
|Endothelial cell count b||.01||.6|
|Year of transplant c||.3||.03|
|Preoperative risk factors d||.9||.8|
|Inflammation (at time of transplant)||.4||.2|
|Indication for graft: optical only||.4||.6|
|Other procedures at time of graft||.2||.2|
|Glaucoma at time of graft e||.003||.002|
|Visual acuity before keratoplasty f||.09||.2|
d Any preoperative risk factors for graft failure (in penetrating keratoplasty, including inflammation, corneal vascularization, indication for graft, other procedures at time of graft, glaucoma, and preoperative visual acuity).
|Risk Factor||No.||Hazard Ratio||95% CI||P Value|
|Center EK experience ( P < .0001)|
|<15||322||2.9||2.2 to 4.0||<.0001|
|15 to 29||233||1.5||1.0 to 2.2||.03|
|Endothelial cell density ( P = .01), cells/mm 2|
|≤2400||156||1.6||1.1 to 2.2||.01|
|Glaucoma ( P = .003) a|
|Yes||57||2.1||1.3 to 3.4||.003|
|Donor gender ( P = .02)|
|Male||761||1.4||1.1 to 1.9||.02|
|Donor age ( P = .05), y|
|≥75||615||1.3||1.0 to 1.7||.05|
|Preoperative acuity ( P = .09)|
|Better than counting fingers||994||1.0|
|Counting fingers or worse||230||1.4||1.0 to 1.9||.05|
|Not reported||51||0.7||0.3 to 1.6||.4|
|Center EK experience ( P = .0001)|
|<15||226||2.2||1.6 to 3.0||<.0001|
|15 to 29||137||1.3||0.9 to 1.9||.2|
|Glaucoma ( P = .002) a|
|Yes||150||1.7||1.2 to 2.3||.002|
|Year of transplant ( P = .03)|
|2006 through 2007||155||1.0||—|
|2008 through 2009||284||1.7||1.1 to 2.5||<.0001|
|2010 through 2011||360||1.4||0.9 to 2.1||.13|
|Donor gender ( P = .07)|
|Male||458||1.3||0.98 to 1.7||.07|