- Email: [email protected]
Predicted Long-term Outcome of Corneal Transplantation
Predicted Long-term Outcome of Corneal Transplantation Vincent M. Borderie, MD, PhD,1 Pierre-Yves Boëlle, PhD,2 Olivier Touzeau, MD, PhD,1 Cécile Allouch, MD, PhD,1 Sandrine Boutboul, MD, PhD,1 Laurent Laroche, MD1 Objective: To analyze graft survival and the outcome of the corneal endothelium after corneal transplantation in a single model to predict the long-term prognosis of these grafts. Design: Cohort study. Data were recorded prospectively and then analyzed retrospectively. Participants: One thousand one hundred forty-four consecutive eyes of 1144 patients who underwent corneal transplantation between 1992 and 2006. Interventions: Penetrating keratoplasty and deep anterior lamellar keratoplasty. Main Outcome Measures: Slit-lamp examination and wide-field specular microscopy results. A joint analysis of endothelial cell loss and time to graft failure was undertaken. From midterm simultaneous analysis of graft survival and endothelial cell loss, long-term graft survival was predicted. Results: The observed 5- and 10-year graft survival estimates were, respectively, 74% and 64%. The average endothelial cell density (cell loss) was 2270 cells/mm2 before surgery, 1058 cells/mm2 (–53%) during the sixth postoperative year, and 865 cells/mm2 (– 61%) during the 10th postoperative year. Overall, the predicted graft survival estimate was 27% at 20 years and 2% at 30 years. Both observed and predicted graft survival were higher in patients who had undergone lamellar keratoplasty than in patients who had undergone penetrating keratoplasty and had normal recipient endothelium and higher in patients who had undergone penetrating keratoplasty and had normal recipient endothelium than in patients who had undergone penetrating keratoplasty and had impaired recipient endothelium. Conclusions: For corneal diseases involving the endothelium, penetrating keratoplasty seems to be a good therapeutic approach in elderly patients because the graft life-span may be similar to the patient life expectancy. Conversely, for younger patients, penetrating keratoplasty is only a midterm therapeutic approach. For corneal diseases not involving the endothelium, deep anterior lamellar keratoplasty seems to be a promising therapeutic approach with higher long-term expected survival. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Ophthalmology 2009;116:2354 –2360 © 2009 by the American Academy of Ophthalmology.
Corneal transplantation is a 100-year-old surgical procedure that is performed routinely in a large number of countries. The outcomes at 5 years are well established from studies including a large number of patients with relevant follow up.1–5 The life-span of a corneal graft depends mainly on the corneal endothelium that maintains a stromal hydration level compatible with corneal transparency. Indeed, this monolayer of nondividing highly differentiated cells loses cells at an accelerated rate after transplantation, effectively limiting the graft life-span.6 – 8 In this respect, deep anterior lamellar keratoplasty, which preserves the recipient corneal endothelium, may have a better prognosis than penetrating keratoplasty, the reference technique. The objective of the present study was to analyze graft survival and the outcome of the corneal endothelium after corneal transplantation in a single model to predict the long-term prognosis of these grafts.
Patients and Methods Study Design One thousand three hundred consecutive penetrating keratoplasties and deep anterior lamellar keratoplasties carried out in 1144 pa-
2354
© 2009 by the American Academy of Ophthalmology Published by Elsevier Inc.
tients between December 1992 and September 2006 were studied. Data were recorded prospectively and then were analyzed retrospectively. In the 156 patients (12%) who received 2 grafts in the same or contralateral eye during the study period, only the first graft was included in the study, yielding a total of 1144 procedures in 1144 patients. Between 1992 and 2001, all patients underwent penetrating keratoplasty. From 2002 through 2006, patients with corneal diseases not or moderately involving the corneal endothelium underwent deep anterior lamellar keratoplasty when technically possible, whereas patients with corneal diseases involving the corneal endothelium underwent penetrating keratoplasty. In accordance with French law, institutional review board or ethics committee approval was not required for this study because no modifications to French standards of treatment and follow-up were made.
Donor Corneas Donor corneas were stored as described previously.9 The average donor age was 69.9⫾13.9 years (mean⫾standard deviation), and the average preoperative endothelial cell density (ECD) was 2264⫾295 cells/mm2. ISSN 0161-6420/09/$–see front matter doi:10.1016/j.ophtha.2009.05.009
Borderie et al 䡠 Long-term Outcome of Corneal Transplantation Surgical and Medical Treatment
Statistical Analysis
All transplantations were performed at a single institution. The average graft diameter was 8.21⫾0.24 mm. Surgical techniques and patient postoperative treatments have been reported elsewhere.10 –12
Graft survival was analyzed using the Cox proportional hazard regression model. Association of variables with survival was tested by the log-rank test. Five-year graft survival was computed using the Kaplan-Meier method. A multivariable Cox proportional hazard regression then was carried out, including variables that were significant at a univariate level (P⬍0.05). The following were studied: donor variables (i.e., donor age, ECD after preservation, death-to-preservation time, preservation time, deswelling time) and recipient and surgical variables (i.e., recipient age, preoperative diagnosis, recipient rejection status, trephination size, preoperative intraocular pressure, lens status, and combined procedures, which include intraocular lens removal, anterior chamber intraocular lens insertion, vitrectomy, trabeculectomy, cataract surgery, and posterior chamber intraocular lens insertion). For preoperative ECD, the graft ECD after preservation was used for patients who underwent penetrating keratoplasty, and the recipient preoperative ECD was used when available for patients who underwent deep anterior lamellar keratoplasty. To predict long-term graft survival, a joint regression model was fitted on survival and ECD to individuals with at least 2 endothelial density measurements in time (n ⫽ 771).13 The logarithm of endothelial density was described as a mixed piecewise linear model in time, with a change in slope 2 years after surgery, and the time to graft failure was described by a lognormal distribution. A multinormal distribution was assumed for individual parameters (intercept and slopes) and logged survival. The estimated correlation between ECD change and graft survival was tested using the log likelihood ratio test. Model-predicted parameters for endothelial density and survival were obtained for each patient and were used in a Kaplan-Meier plot for predicted graft survival. Correlation between model-predicted quantities was tested using the Spearman correlation test. Differences in predicted survivals were tested with the Wilcoxon rank-sum test.
Recipients and Transplantation Outcome The mean age of the patients was 58.4⫾21.8 years. High-risk recipients were defined as having a vascularized cornea (2 or more quadrants of corneal vascularization) or a history of irreversible corneal allograft rejection in the operated eye. Only grafts intended for optical indication were included in the study. Based on the type of surgery and the recipient corneal endothelium condition, patients were divided in 3 groups as follows: penetrating keratoplasty performed in a recipient eye with impaired corneal endothelium (n ⫽ 701), penetrating keratoplasty performed in a recipient eye with normal corneal endothelium (n ⫽ 361), and deep anterior lamellar keratoplasty performed in a recipient eye with normal corneal endothelium (n ⫽ 82). Patient characteristics are shown in Table 1. Patients were examined at 1 and 2 weeks; at 1, 3, 6, 9, 12, 18, 24, and 36 months; and at 4, 5, 6, 8, 10, and 15 years after surgery. Slit-lamp findings and intraocular pressure were recorded at each examination. The central corneal ECD was assessed by means of wide-field specular microscopy (Konan Keeler Pocklington, Hyogo, Japan; Topcon SP 3000P, Tokyo, Japan) during the second (16⫾4 months after surgery), fourth (41⫾7 months after surgery), sixth (69⫾10 months after surgery), and 10th years (110⫾12 months after surgery), and at each examination afterward. Images of the central corneal endothelium with at least 100 cells available for analysis were used for assessing ECD, which was measured using the fixed frame method. The criteria for graft failure were irreversible graft stromal edema or corneal opacification.
Table 1. Characteristics of the Recipients at the Time of Transplantation
Preoperative diagnosis Keratoconus Bullous keratopathy Endothelial dystrophies Stromal dystrophies Corneal scar after infectious keratitis Trauma Regraft Postoperative lens status Phakic Posterior chamber intraocular lens Anterior chamber intraocular lens Aphakic Recipient rejection status Low risk High risk Preoperative IOP No glaucoma and IOP ⱕ20 mmHg Glaucoma or IOP ⬎ 20 mmHg
Overall (n ⴝ 1144)
Penetrating Keratoplasty and Impaired Recipient Endothelium (n ⴝ 701)
Penetrating Keratoplasty and Normal Recipient Endothelium (n ⴝ 361)
Deep Anterior Lamellar Keratoplasty (n ⴝ 82)
258 (22%) 449 (39%) 121 (11%) 32 (3%) 151 (13%) 46 (4%) 87 (8%)
0 (0%) 449 (64%) 121 (17%) 0 (0%) 0 (0%) 44 (6%) 87 (13%)
218 (60%) 0 (0%) 0 (0%) 27 (8%) 116 (32%) 0 (0%) 0 (0%)
40 (49%) 0 (0%) 0 (0%) 5 (6%) 35 (43%) 2 (2%) 0 (0%)
408 (36%) 331 (29%) 311 (27%) 94 (8%)
56 (8%) 254 (36%) 309 (44%) 82 (12%)
281 (78%) 67 (18%) 2 (1%) 11 (3%)
71 (87%) 10 (12%) 0 (0%) 1 (1%)
884 (77%) 260 (23%)
547 (78%) 154 (22%)
274 (76%) 87 (24%)
63 (77%) 19 (23%)
896 (78%) 248 (22%)
471 (67%) 230 (33%)
348 (96%) 13 (4%)
77 (94%) 5 (6%)
P Value ⬍0.001
⬍0.001
0.73 ⬍0.001
IOP ⫽ intraocular pressure. N ⫽ 1144 (1 graft per patient was included).
2355
Ophthalmology Volume 116, Number 12, December 2009 Table 2. Follow-up of Patients Date of Surgery Total number of patients Graft failure during the study period Deceased during the study period with clear graft Follow-up time* Mean Range
1992–1995 229 (100%) 70 (31%) 11 (5%)
1996–1998 295 (100%) 65 (22%) 21 (7%)
1999–2002 343 (100%) 69 (20%) 12 (3%)
2003–2006 277 (100%) 25 (9%) 1 (0.3%)
45 months 0.5-171 months
48 months 0.5-143 months
40 months 0.5-107 months
29 months 0.5-66 months
n ⫽ 1144 (one graft per patient was included). *From surgery to failure for unsuccessful grafts or from surgery to the date of last visit for successful grafts.
Results Overall Results The average follow-up time (from surgery to failure for unsuccessful grafts or from surgery to the date of last visit for successful grafts) was 40.5⫾32.1 months (Table 2). The percentage of patients lost to follow-up or deceased was 7% (81/1144) at 12 months after surgery, 29% (318/1090) at 36 months, 45% (409/ 910) at 60 months, and 60% (289/484) at 120 months. The observed 5- and 10-year graft survival estimates were, respectively, 74.3% and 64.4% (Fig 1). The average ECD (and endothelial cell loss) was 2270 cells/mm2 before surgery, 1604 cells/mm2 (–29.7%) at 1 year after surgery, 1321 cells/mm2 (– 42.4%) at 3 years, 1058 cells/mm2 (–52.7%) at 5 years, and 865 cells/mm2 (60.7%) at 10 years.
Factors Influencing Graft Survival Because group and preoperative diagnosis were not independent variables, only one of both was used in the Cox model. Classification of patients in 3 groups (or preoperative diagnosis), graft preservation time, recipient rejection status, preoperative intraocular pressure, and lens status significantly influenced graft survi-
Figure 1. Kaplan-Meier survival curve showing overall observed (solid line) and predicted (dashed line) graft survival of 1144 consecutive penetrating and deep anterior lamellar keratoplasties (second grafts excluded). Dotted lines indicate the 95% confidence interval for the observed graft survival. The numbers of eyes at risk at 12, 36, 60, and 120 months are shown above the survival curve.
2356
val (Table 3; Fig 2). When postoperative events (i.e., rejection episodes, rise in intraocular pressure, epithelial complications, trauma, infectious keratitis, and miscellaneous events) were introduced in the multivariate Cox model, donor, recipient, and surgical variables gave the same results. Rejection episodes, postoperative infectious keratitis, postoperative trauma, and miscellaneous events significantly influenced graft survival (Table 4). The observed graft survival estimates at 5 and 10 years were, respectively, 62.1⫾2.0% and 47.1⫾5.0% in patients who underwent penetrating keratoplasty and had impaired recipient endothelium, 91.1⫾2.0% and 86.6⫾3.0% in patients who underwent penetrating keratoplasty and had normal recipient endothelium, and 97.2⫾ 2.0% (5 years) in patients who underwent deep anterior lamellar keratoplasty. For all patients who underwent penetrating keratoplasty, these figures were 73.0⫾2.0% and 63.3⫾3.0%.
Exponential Model Used to Predict the Long-term Postoperative Endothelial Cell Density The simultaneous regression model fitted the data significantly better when a correlation was allowed between ECD and time to graft failure (P⬍0.001). The median predicted time to graft failure was 175 months overall (95% confidence interval [CI], 168 –184 months), with a significant change according to the group (P⬍0.001). It was 139 months (95% CI, 124 –149 months) in patients who underwent penetrating keratoplasty and had impaired recipient endothelium, 221 months (95% CI, 210 –235 months) in patients who underwent penetrating keratoplasty and had normal recipient endothelium, and 276 months (95% CI, 254 –295 months) in patients who underwent deep anterior lamellar keratoplasty. In the statistical model, the early phase of the postoperative endothelial cell loss corresponded to the first postoperative 2 years and the late phase corresponded to the subsequent postoperative years. The overall predicted loss in endothelial cells was 41⫾24% over the first 2 years (early phase), followed by a 10⫾6% yearly decrease afterward (late phase). In patients who underwent penetrating keratoplasty and had impaired recipient endothelium, the loss in the first 2 years was the largest (50⫾22%). It was 33⫾22% in those who underwent penetrating keratoplasty and had normal recipient endothelium and 17⫾22% in patients who underwent deep anterior lamellar keratoplasty. The late phase–predicted yearly rate of change was less variable according to group (9%, 10%, and 11%, respectively), with no significant differences between the 3 groups. A larger initial cell density was associated with increased predicted time to graft failure (Spearman correlation coefficient, r ⫽ 0.70; P⬍0.0001). A larger percentage of decrease in cell density over the first 2 years was found to lead to a shorter predicted time to graft failure (r ⫽ 0.88; P⬍0.0001). Using Cox regression, it was verified that patients with large initial decrease in ECD in the first year were indeed more prone to graft failure (data not shown).
Borderie et al 䡠 Long-term Outcome of Corneal Transplantation Table 3. Influence of Donor, Recipient, and Surgical Variables on Graft Survival
Multivariable Level
Predicted Graft Survival
Hazard Ratio (95%CI)
Hazard Ratio (95%CI)
20-year Graft Survival
Observed Graft Survival Univariable Level Variable
N
5-year Graft Survival
Preservation time (weeks) 2–3 4–5
303 841
75% 71%
1 1.4 (1.1–1.8) (P ⫽ 0.02)
1 1.4 (1.1–1.9) (P ⫽ 0.004)
28.4% 24.3% (P ⫽ 0.003)
Recipient rejection status High-risk Low-risk
260 884
60.0% 78.3%
2.3 (1.7–3.0) 1 (P⬍0.001)
2.5 (1.9–3.3) 1 (P⬍0.001)
19.3% 29.5% (P⬍0.001)
Preoperative diagnosis Regraft Bullous keratopathy Trauma Stromal dystrophies Corneal scar after infectious keratitis Endothelial dystrophies Keratoconus
87 449 46 32 151 121 258
49.8% 58.1% 59.8% 82.6% 83.6% 93.0% 97.8%
1.2 (0.8–1.8) 1 1.0 (0.6–1.6) 0.4 (0.2–1.0) 0.4 (0.3–0.6) 0.2 (0.1–0.3) 0.04 (0.02–0.1) (P⬍0.001)
Preoperative intraocular pressure Glaucoma or intraocular pressure ⬎ 20 mmHg No glaucoma and intraocular pressure ⱕ 20 mmHg
248 896
51.8% 81.3%
2.8 (2.1–3.6) 1 (P⬍0.001)
1.6 (1.2–2.2) 1 (P⬍0.001)
12.0% 31.6% (P⬍0.001)
Postoperative lens status Ant chamber intraocular lens Aphakic Post chamber intraocular lens Phakic
311 94 331 408
52.8% 58.1% 75.0% 93.8%
8.1 (5.2–12.7) 7.7 (4.5–13.1) 4.0 (2.5–6.3) 1 (P⬍0.001)
4.5 (2.6–8.1) 4.0 (2.2–7.6) 2.4 (1.4–4.2) 1 (P⬍0.001)
11.8% 13.6% 17.0% 48.5% (P⬍0.001)
Group Penetrating keratoplasty and impaired recipient endothelium Penetrating keratoplasty and normal recipient endothelium Deep anterior lamellar keratoplasty
701 361 82
62.1% 91.1% 97.2%
4.2 (2.9–6.2) 1 0.4 (0.1–1.6) (P⬍0.001)
1.6 (1.0–2.6) 1 0.5 (0.1–1.9) (P ⫽ 0.04)
14.9% 40.6% 63.2% (P⬍0.001)
11.6% 11.0% 33.5% 25.4% 28.4% 25.4% 57.7% (P⬍0.001)
CI ⫽ confidence interval. Note: Only variables significant at a multivariate level are shown. As group and preoperative diagnosis were strongly linked variables only group was used for multivariate analysis. When the Cox model was set up with preoperative diagnosis (in spite of group) similar results were obtained.
Predicted Long-term Graft Survival Overall, the predicted graft survival estimate was 27.2⫾2.0% at 20 years and 2.0⫾0.5% at 30 years (Fig 1). Significant differences were found among the predicted graft survival in the 3 groups of patients (P⬍0.001). It was 14.9⫾2.0% at 20 years and 0.6⫾0.3% at 30 years in patients who underwent penetrating keratoplasty and had impaired recipient endothelium, 40.6⫾3.0% and 2.5⫾1.0% in patients who underwent penetrating keratoplasty and had normal recipient endothelium, and 63.2⫾6.0% and 10.5⫾4.0% in patients who underwent deep anterior lamellar keratoplasty (Fig 2). For all patients who underwent penetrating keratoplasty, these figures were 23.9⫾2.0% and 1.2⫾0.4%.
Discussion In the present study, the overall observed graft survival after corneal transplantation was 74% at 5 years and 64% at 10
years, and the predicted graft survival was 27% at 20 years and 2% at 30 years. Both observed graft survival and predicted graft survival were higher in patients who underwent deep anterior lamellar keratoplasty than in patients who underwent penetrating keratoplasty and had normal recipient endothelium and were higher in patients who underwent penetrating keratoplasty and had normal recipient endothelium than in patients who underwent penetrating keratoplasty and had impaired recipient endothelium. Graft survival at 5 years after penetrating keratoplasty has been reported in studies including a large number of patients with the following survival estimates: 86% in the Cornea Donor Study (1090 eyes), 69% in the Singapore Corneal Transplant Study (901 eyes), 62% in the Australian study (10 952 eyes), and 90% for first-time grafts and 53% for initial regrafts in the study reported by Thompson et al (3992 eyes).1–5 For the 10-year survival, the following estimates have been reported: 72% in the Japanese study reported by Inoue et al14 (396 eyes) and 82% for first-time
2357
Ophthalmology Volume 116, Number 12, December 2009
Figure 2. Kaplan-Meier survival curve showing observed and predicted graft survival in patients who underwent penetrating keratoplasty and had impaired recipient endothelium (green), who underwent penetrating keratoplasty and had normal recipient endothelium (blue), or who had deep anterior lamellar keratoplasty (red). Log-rank for observed survival: P⬍0.001.
grafts and 41% for initial regrafts in the study reported by Thompson et al.5 The survival estimates of the present study (overall, 74% at 5 years and 64% at 10 years; penetrating keratoplasty, 73% at 5 years and 63% at 10 years) seem to be in good accordance with those previously reported. Corneal graft survival also seems similar to patient survival after renal transplantation, which has been reported to be in the range of 60% to 65% 15 years after surgery.15
Indication for corneal transplantation is a major prognosis factor for graft survival, with 5- and 10-year survival of more than 90% in patients with keratoconus or Fuchs endothelial dystrophy, 50% to 70% for patients with bullous keratopathy, 50% and less for regrafts.3–5 The same significant differences in graft survival were found among indications. The outcome of the corneal endothelium 15 years after penetrating keratoplasty has been reported in a series of 67 eyes.16 The average endothelial cell loss from preoperative donor levels was 71%, and the average ECD was 872 cells/mm2. In a series of 15 eyes that underwent penetrating keratoplasty, the ECD averaged 998 cells/mm2 10 years after surgery and 852 cells/mm2 20 years after surgery.17 In the Cornea Donor Study, the median density of 347 eyes with clear grafts was 778 cells/mm2 at 5 years (median cell loss, 70%).1 Ten-year postoperative data from 72 patients who underwent penetrating keratoplasty showed an average ECD of 958 cells/mm2 (– 67%).18 In the present study, the average ECDs at 5 and 10 years were 1058 cells/mm2 (–53%) and 865 cells/mm2 (– 61%), respectively. For anterior lamellar keratoplasty, only short-term data are available. The reported ECDs after anterior lamellar keratoplasty are the following: 2417 cells/mm2 (average postoperative time, 3 years; n ⫽ 20), 2220 cells/mm2 (average postoperative time, 1 year; n ⫽ 24), and 1937 cells/mm2 (average postoperative time, 2 years; n ⫽ 113).19 –21 These figures are in accordance with the current data. Long-term graft survival typically is difficult to assess because of competitive mortality and loss to follow-up in the long term. For postoperative time points of more than 10 years, no studies with a large number of patients and documented follow-up are available. Mathematical models may be useful to help predict the long-term survival based on the observed midterm survival.22 For example, Armitage et al23
Table 4. Influence of Postoperative Variables on Graft Survival
Multivariable Level
Predicted Graft Survival
Observed Graft Survival Univariable Level N
5-year Graft Survival
Hazard Ratio (95% confidence interval)
Hazard Ratio (95% confidence interval)
20-year Graft Survival
917 227
84.8% 44.4%
1 4.8 (3.7–6.2) (P⬍0.001)
1 4.1 (3.1–5.3) (P⬍0.001)
31.5% 13.0% (P⬍0.001)
Postoperative infectious keratitis None At least 1
1115 29
76.8% 37.6%
1 3.3 (2.1–5.4) (P⬍0.001)
1 2.5 (1.5–4.1) (P⬍0.001)
27.5% 16.0% (P⬍0.001)
Postoperative trauma None At least 1
1120 24
75.9% 49.0%
1 2.5 (1.4–4.4) (P ⫽ 0.001)
1 2.7 (1.5–4.8) (P⬍0.001)
27.5% 26.0% (P ⫽ 0.43)
Miscellaneous postoperative events None At least 1
1091 53
76.0% 58.0%
1 1.9 (1.2–3.0) (P ⫽ 0.008)
1 3.0 (1.8–4.9) (P⬍0.001)
27.0% 25.0% (P⬍0.001)
Variable Rejection episodes No rejection event At least 1 rejection event
Only variables significant at a multivariate level are shown.
2358
Borderie et al 䡠 Long-term Outcome of Corneal Transplantation used a biexponential model to describe the postoperative endothelial cell loss, because it is the primary cause of graft failure after the first 5 postoperative years.16 From this model, penetrating grafts with uncomplicated outcome were predicted to have a maximum life-span of 3 decades: the average time to reach a postoperative ECD of 500 cells/ mm2 was 21 years for a graft with a preoperative ECD of 2000 cells/mm2 and 28 years for a graft with 2500 cells/ mm2. However, this analysis might have underestimated postoperative endothelial cell loss because patients whose graft had failed were censored progressively from the analysis, leading to the observation of patients maintaining large cell densities with time. This in turn might have lead to overestimate graft survival. Herein, a joint analysis of endothelial cell loss and time to graft failure was undertaken, allowing explicit account of the informative censoring at work. The mixed-model formulation also allowed for prediction of individual time to graft failure conditional on follow-up and time-dependent profile of ECD. A fully parametric formulation of the model was used to allow for graft survival prediction. It was found that a biphasic decrease for endothelial cell loss fitted the data well, as reported by Armitage et al.23 A log normal distribution was used to describe time to graft failure, because this formulation describes well the variability of biologic processes. The fit of the model was fair compared with observed survivals, although it may lead to slight overestimates the survival of impaired recipients. The model-predicted median time to graft failure was 15 years. The overall observed midterm graft survival was 74% at 5 years and 64% at 10 years, and the predicted long-term graft survival was 48% at 15 years, 27% at 20 years, and 2% at 30 years, which shows limitation of corneal graft lifespan to 2 to 3 decades. These figures are in accordance with the data from the Australian Corneal Graft Registry. In fact, the reported observed graft survival was 60% at 10 years and 46% at 15 years in the Australian study.24 Major differences were found among surgical groups. Patients who underwent anterior lamellar keratoplasty had significantly lower endothelial cell loss and significantly higher predicted graft survival than patients who underwent penetrating keratoplasty and had either impaired or normal recipient endothelium. The observed graft survival was higher in lamellar keratoplasty patients, but the difference did not reach statistical significance. The better outcome of lamellar grafts may originate from absence of donor endothelium, which implies absence of endothelial rejection, and from lack of corneal endothelium surgical wound and scar. From these data, it is still difficult to predict precisely the life-span of a lamellar corneal graft, because follow-up is fewer than 5 years in most patients. In the group of lamellar keratoplasties, the model that predicts postoperative endothelial cell loss was set up with 2 (preoperative and 1-year data) or 3 (preoperative, 1-year, and 3-year data) time points in most cases, whereas in patients who underwent penetrating keratoplasty, data obtained at 2 to 6 time points were available for analysis. This implies that the late phase of endothelial cell loss was predicted less accurately after lamellar keratoplasty than in eyes that underwent penetrating keratoplasty. For a few patients who underwent lamellar keratoplasty and for whom 5-year data were available, the
endothelial cell loss observed between 3 and 5 years was very low, which implies that the model may overestimate the late-phase cell loss in eyes that underwent anterior lamellar keratoplasty. Longer follow-up of patients who underwent anterior lamellar keratoplasty is needed to investigate whether the late phase of endothelial cell loss is similar to that observed after penetrating keratoplasty or if it returns to the normal age-related endothelial cell loss. If the latter hypothesis is conformed, the predicted long-term graft survival after anterior lamellar keratoplasty will be higher than that reported in the present study. Patients who underwent penetrating keratoplasty did better when the recipient endothelium was normal. After penetrating corneal transplantation, 58% of the corneal endothelium (peripheral endothelium) are from recipient origin and 42% (graft endothelium) are from donor origin. If the preoperative peripheral ECD is very low, the total number of corneal endothelial cells available after transplantation will be lower than with a normal preoperative peripheral ECD. In conclusion, for corneal diseases involving the endothelium, penetrating keratoplasty seems to be a good therapeutic approach in elderly patients because the graft lifespan may be similar to the patient life expectancy. Conversely, for young and middle-aged patients, penetrating keratoplasty is only a midterm therapeutic approach. For corneal diseases not involving the endothelium, deep anterior lamellar keratoplasty seems to be a promising therapeutic approach with higher long-term expected survival.
References 1. Cornea Donor Study Investigator Group. The effect of donor age on corneal transplantation outcome: results of the Cornea Donor Study. Ophthalmology 2008;115:620 – 6. 2. Cornea Donor Study Investigator Group. Donor age and corneal endothelial cell loss 5 years after successful corneal transplantation: specular microscopy ancillary study results. Ophthalmology 2008;115:627–32. 3. Tan DT, Janardhanan P, Zhou H, et al. Penetrating keratoplasty in Asian eyes: the Singapore Corneal Transplant Study. Ophthalmology 2008;115:975– 82. 4. Williams KA, Esterman AJ, Bartlett C, et al. How effective is penetrating corneal transplantation? Factors influencing longterm outcome in multivariate analysis. Transplantation 2006; 81:896 –901. 5. Thompson RW Jr, Price MO, Bowers PJ, Price FW Jr. Longterm graft survival after penetrating keratoplasty. Ophthalmology 2003;110:1396 – 402. 6. Maurice DM. The location of the fluid pump in the cornea. J Physiol 1972;221:43–54. 7. Mergler S, Pleyer U. The human corneal endothelium: new insights into electrophysiology and ion channels. Prog Retin Eye Res 2007;26:359 –78. 8. Bourne WM, McLaren JW. Clinical responses of the corneal endothelium. Exp Eye Res 2004;78:561–72. 9. Borderie VM, Scheer S, Touzeau O, et al. Donor organ cultured corneal tissue selection before penetrating keratoplasty. Br J Ophthalmol 1998;82:382– 8. 10. Borderie VM, Touzeau O, Allouch C, et al. The results of successful penetrating keratoplasty using donor organcultured corneal tissue. Transplantation 1999;67:1433– 8.
2359
Ophthalmology Volume 116, Number 12, December 2009 11. Borderie VM, Touzeau O, Bourcier T, et al. Graft reepithelialization after penetrating keratoplasty using organ-cultured donor tissue. Ophthalmology 2006;113:2181– 6. 12. Borderie VM, Werthel AL, Touzeau O, et al. Comparison of techniques used for removing the recipient stroma in anterior lamellar keratoplasty. Arch Ophthalmol 2008;126:31–7. 13. Schluchter MD. Methods for the analysis of informatively censored longitudinal data. Stat Med 1992;11:1861–70. 14. Inoue K, Amano S, Kimura C, et al. Long-term effects of topical cyclosporine A treatment after penetrating keratoplasty. Jpn J Ophthalmol 2000;44:302–5. 15. Joss N, Rodger RS, McMillan MA, Junor BJ. Randomized study comparing cyclosporine with azathioprine one year after renal transplantation-15-year outcome data. Transplantation 2007;83:582–7. 16. Patel SV, Hodge DO, Bourne WM. Corneal endothelium and postoperative outcomes 15 years after penetrating keratoplasty. Am J Ophthalmol 2005;139:311–9. 17. Inoue K, Kimura C, Amano S, et al. Corneal endothelial cell changes twenty years after penetrating keratoplasty. Jpn J Ophthalmol 2002;46:189 –92.
18. Ing JJ, Ing HH, Nelson LR, et al. Ten-year postoperative results of penetrating keratoplasty. Ophthalmology 1998;105: 1855– 65. 19. Morris E, Kirwan JF, Sujatha S, Rostron CK. Corneal endothelial specular microscopy following deep lamellar keratoplasty with lyophilized tissue. Eye 1998;12:619 –22. 20. Panda A, Bageshwar LM, Ray M, et al. Deep lamellar keratoplasty versus penetrating keratoplasty for corneal lesions. Cornea 1999;18:172–5. 21. Sugita J, Kondo J. Deep lamellar keratoplasty with complete removal of pathological stroma for vision improvement. Br J Ophthalmol 1997;81:184 – 8. 22. Remuzzi G, Cravedi P, Perna A, et al, Dual Kidney Transplant Group. Long-term outcome of renal transplantation from older donors. N Engl J Med 2006;354:343–52. 23. Armitage WJ, Dick AD, Bourne WM. Predicting endothelial cell loss and long-term corneal graft survival. Invest Ophthalmol Vis Sci 2003;44:3326 –31. 24. Williams KA, Lowe M, Bartlett C, et al, All Contributors. Risk factors for human corneal graft failure within the Australian corneal graft registry. Transplantation 2008;86: 1720 – 4.
Footnotes and Financial Disclosures Originally received: January 26, 2009. Final revision: April 2, 2009. Accepted: May 7, 2009. Available online: October 7, 2009.
Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Manuscript no. 2009-115.
1
Department of Ophthalmology, Centre Hospitalier National d’Ophtalmologie des XV-XX, UPMC University of Paris, Paris, France. 2
Department of Public Health, INSERM U707, Hôpital Saint Antoine, Assistance Publique—Hôpitaux de Paris, UPMC University of Paris 06, Paris, France.
2360
Supported by UPMC University of Paris 06, Paris, France. Correspondence: Vincent M. Borderie, MD, PhD, Service V, Centre Hospitalier National d’Ophtalmologie des XV-XX, 28 rue de Charenton, 75012 Paris, France. E-mail: [email protected].
Corneal transplantation is a 100-year-old surgical procedure that is performed routinely in a large number of countries. The outcomes at 5 years are well established from studies including a large number of patients with relevant follow up.1–5 The life-span of a corneal graft depends mainly on the corneal endothelium that maintains a stromal hydration level compatible with corneal transparency. Indeed, this monolayer of nondividing highly differentiated cells loses cells at an accelerated rate after transplantation, effectively limiting the graft life-span.6 – 8 In this respect, deep anterior lamellar keratoplasty, which preserves the recipient corneal endothelium, may have a better prognosis than penetrating keratoplasty, the reference technique. The objective of the present study was to analyze graft survival and the outcome of the corneal endothelium after corneal transplantation in a single model to predict the long-term prognosis of these grafts.
Patients and Methods Study Design One thousand three hundred consecutive penetrating keratoplasties and deep anterior lamellar keratoplasties carried out in 1144 pa-
2354
© 2009 by the American Academy of Ophthalmology Published by Elsevier Inc.
tients between December 1992 and September 2006 were studied. Data were recorded prospectively and then were analyzed retrospectively. In the 156 patients (12%) who received 2 grafts in the same or contralateral eye during the study period, only the first graft was included in the study, yielding a total of 1144 procedures in 1144 patients. Between 1992 and 2001, all patients underwent penetrating keratoplasty. From 2002 through 2006, patients with corneal diseases not or moderately involving the corneal endothelium underwent deep anterior lamellar keratoplasty when technically possible, whereas patients with corneal diseases involving the corneal endothelium underwent penetrating keratoplasty. In accordance with French law, institutional review board or ethics committee approval was not required for this study because no modifications to French standards of treatment and follow-up were made.
Donor Corneas Donor corneas were stored as described previously.9 The average donor age was 69.9⫾13.9 years (mean⫾standard deviation), and the average preoperative endothelial cell density (ECD) was 2264⫾295 cells/mm2. ISSN 0161-6420/09/$–see front matter doi:10.1016/j.ophtha.2009.05.009
Borderie et al 䡠 Long-term Outcome of Corneal Transplantation Surgical and Medical Treatment
Statistical Analysis
All transplantations were performed at a single institution. The average graft diameter was 8.21⫾0.24 mm. Surgical techniques and patient postoperative treatments have been reported elsewhere.10 –12
Graft survival was analyzed using the Cox proportional hazard regression model. Association of variables with survival was tested by the log-rank test. Five-year graft survival was computed using the Kaplan-Meier method. A multivariable Cox proportional hazard regression then was carried out, including variables that were significant at a univariate level (P⬍0.05). The following were studied: donor variables (i.e., donor age, ECD after preservation, death-to-preservation time, preservation time, deswelling time) and recipient and surgical variables (i.e., recipient age, preoperative diagnosis, recipient rejection status, trephination size, preoperative intraocular pressure, lens status, and combined procedures, which include intraocular lens removal, anterior chamber intraocular lens insertion, vitrectomy, trabeculectomy, cataract surgery, and posterior chamber intraocular lens insertion). For preoperative ECD, the graft ECD after preservation was used for patients who underwent penetrating keratoplasty, and the recipient preoperative ECD was used when available for patients who underwent deep anterior lamellar keratoplasty. To predict long-term graft survival, a joint regression model was fitted on survival and ECD to individuals with at least 2 endothelial density measurements in time (n ⫽ 771).13 The logarithm of endothelial density was described as a mixed piecewise linear model in time, with a change in slope 2 years after surgery, and the time to graft failure was described by a lognormal distribution. A multinormal distribution was assumed for individual parameters (intercept and slopes) and logged survival. The estimated correlation between ECD change and graft survival was tested using the log likelihood ratio test. Model-predicted parameters for endothelial density and survival were obtained for each patient and were used in a Kaplan-Meier plot for predicted graft survival. Correlation between model-predicted quantities was tested using the Spearman correlation test. Differences in predicted survivals were tested with the Wilcoxon rank-sum test.
Recipients and Transplantation Outcome The mean age of the patients was 58.4⫾21.8 years. High-risk recipients were defined as having a vascularized cornea (2 or more quadrants of corneal vascularization) or a history of irreversible corneal allograft rejection in the operated eye. Only grafts intended for optical indication were included in the study. Based on the type of surgery and the recipient corneal endothelium condition, patients were divided in 3 groups as follows: penetrating keratoplasty performed in a recipient eye with impaired corneal endothelium (n ⫽ 701), penetrating keratoplasty performed in a recipient eye with normal corneal endothelium (n ⫽ 361), and deep anterior lamellar keratoplasty performed in a recipient eye with normal corneal endothelium (n ⫽ 82). Patient characteristics are shown in Table 1. Patients were examined at 1 and 2 weeks; at 1, 3, 6, 9, 12, 18, 24, and 36 months; and at 4, 5, 6, 8, 10, and 15 years after surgery. Slit-lamp findings and intraocular pressure were recorded at each examination. The central corneal ECD was assessed by means of wide-field specular microscopy (Konan Keeler Pocklington, Hyogo, Japan; Topcon SP 3000P, Tokyo, Japan) during the second (16⫾4 months after surgery), fourth (41⫾7 months after surgery), sixth (69⫾10 months after surgery), and 10th years (110⫾12 months after surgery), and at each examination afterward. Images of the central corneal endothelium with at least 100 cells available for analysis were used for assessing ECD, which was measured using the fixed frame method. The criteria for graft failure were irreversible graft stromal edema or corneal opacification.
Table 1. Characteristics of the Recipients at the Time of Transplantation
Preoperative diagnosis Keratoconus Bullous keratopathy Endothelial dystrophies Stromal dystrophies Corneal scar after infectious keratitis Trauma Regraft Postoperative lens status Phakic Posterior chamber intraocular lens Anterior chamber intraocular lens Aphakic Recipient rejection status Low risk High risk Preoperative IOP No glaucoma and IOP ⱕ20 mmHg Glaucoma or IOP ⬎ 20 mmHg
Overall (n ⴝ 1144)
Penetrating Keratoplasty and Impaired Recipient Endothelium (n ⴝ 701)
Penetrating Keratoplasty and Normal Recipient Endothelium (n ⴝ 361)
Deep Anterior Lamellar Keratoplasty (n ⴝ 82)
258 (22%) 449 (39%) 121 (11%) 32 (3%) 151 (13%) 46 (4%) 87 (8%)
0 (0%) 449 (64%) 121 (17%) 0 (0%) 0 (0%) 44 (6%) 87 (13%)
218 (60%) 0 (0%) 0 (0%) 27 (8%) 116 (32%) 0 (0%) 0 (0%)
40 (49%) 0 (0%) 0 (0%) 5 (6%) 35 (43%) 2 (2%) 0 (0%)
408 (36%) 331 (29%) 311 (27%) 94 (8%)
56 (8%) 254 (36%) 309 (44%) 82 (12%)
281 (78%) 67 (18%) 2 (1%) 11 (3%)
71 (87%) 10 (12%) 0 (0%) 1 (1%)
884 (77%) 260 (23%)
547 (78%) 154 (22%)
274 (76%) 87 (24%)
63 (77%) 19 (23%)
896 (78%) 248 (22%)
471 (67%) 230 (33%)
348 (96%) 13 (4%)
77 (94%) 5 (6%)
P Value ⬍0.001
⬍0.001
0.73 ⬍0.001
IOP ⫽ intraocular pressure. N ⫽ 1144 (1 graft per patient was included).
2355
Ophthalmology Volume 116, Number 12, December 2009 Table 2. Follow-up of Patients Date of Surgery Total number of patients Graft failure during the study period Deceased during the study period with clear graft Follow-up time* Mean Range
1992–1995 229 (100%) 70 (31%) 11 (5%)
1996–1998 295 (100%) 65 (22%) 21 (7%)
1999–2002 343 (100%) 69 (20%) 12 (3%)
2003–2006 277 (100%) 25 (9%) 1 (0.3%)
45 months 0.5-171 months
48 months 0.5-143 months
40 months 0.5-107 months
29 months 0.5-66 months
n ⫽ 1144 (one graft per patient was included). *From surgery to failure for unsuccessful grafts or from surgery to the date of last visit for successful grafts.
Results Overall Results The average follow-up time (from surgery to failure for unsuccessful grafts or from surgery to the date of last visit for successful grafts) was 40.5⫾32.1 months (Table 2). The percentage of patients lost to follow-up or deceased was 7% (81/1144) at 12 months after surgery, 29% (318/1090) at 36 months, 45% (409/ 910) at 60 months, and 60% (289/484) at 120 months. The observed 5- and 10-year graft survival estimates were, respectively, 74.3% and 64.4% (Fig 1). The average ECD (and endothelial cell loss) was 2270 cells/mm2 before surgery, 1604 cells/mm2 (–29.7%) at 1 year after surgery, 1321 cells/mm2 (– 42.4%) at 3 years, 1058 cells/mm2 (–52.7%) at 5 years, and 865 cells/mm2 (60.7%) at 10 years.
Factors Influencing Graft Survival Because group and preoperative diagnosis were not independent variables, only one of both was used in the Cox model. Classification of patients in 3 groups (or preoperative diagnosis), graft preservation time, recipient rejection status, preoperative intraocular pressure, and lens status significantly influenced graft survi-
Figure 1. Kaplan-Meier survival curve showing overall observed (solid line) and predicted (dashed line) graft survival of 1144 consecutive penetrating and deep anterior lamellar keratoplasties (second grafts excluded). Dotted lines indicate the 95% confidence interval for the observed graft survival. The numbers of eyes at risk at 12, 36, 60, and 120 months are shown above the survival curve.
2356
val (Table 3; Fig 2). When postoperative events (i.e., rejection episodes, rise in intraocular pressure, epithelial complications, trauma, infectious keratitis, and miscellaneous events) were introduced in the multivariate Cox model, donor, recipient, and surgical variables gave the same results. Rejection episodes, postoperative infectious keratitis, postoperative trauma, and miscellaneous events significantly influenced graft survival (Table 4). The observed graft survival estimates at 5 and 10 years were, respectively, 62.1⫾2.0% and 47.1⫾5.0% in patients who underwent penetrating keratoplasty and had impaired recipient endothelium, 91.1⫾2.0% and 86.6⫾3.0% in patients who underwent penetrating keratoplasty and had normal recipient endothelium, and 97.2⫾ 2.0% (5 years) in patients who underwent deep anterior lamellar keratoplasty. For all patients who underwent penetrating keratoplasty, these figures were 73.0⫾2.0% and 63.3⫾3.0%.
Exponential Model Used to Predict the Long-term Postoperative Endothelial Cell Density The simultaneous regression model fitted the data significantly better when a correlation was allowed between ECD and time to graft failure (P⬍0.001). The median predicted time to graft failure was 175 months overall (95% confidence interval [CI], 168 –184 months), with a significant change according to the group (P⬍0.001). It was 139 months (95% CI, 124 –149 months) in patients who underwent penetrating keratoplasty and had impaired recipient endothelium, 221 months (95% CI, 210 –235 months) in patients who underwent penetrating keratoplasty and had normal recipient endothelium, and 276 months (95% CI, 254 –295 months) in patients who underwent deep anterior lamellar keratoplasty. In the statistical model, the early phase of the postoperative endothelial cell loss corresponded to the first postoperative 2 years and the late phase corresponded to the subsequent postoperative years. The overall predicted loss in endothelial cells was 41⫾24% over the first 2 years (early phase), followed by a 10⫾6% yearly decrease afterward (late phase). In patients who underwent penetrating keratoplasty and had impaired recipient endothelium, the loss in the first 2 years was the largest (50⫾22%). It was 33⫾22% in those who underwent penetrating keratoplasty and had normal recipient endothelium and 17⫾22% in patients who underwent deep anterior lamellar keratoplasty. The late phase–predicted yearly rate of change was less variable according to group (9%, 10%, and 11%, respectively), with no significant differences between the 3 groups. A larger initial cell density was associated with increased predicted time to graft failure (Spearman correlation coefficient, r ⫽ 0.70; P⬍0.0001). A larger percentage of decrease in cell density over the first 2 years was found to lead to a shorter predicted time to graft failure (r ⫽ 0.88; P⬍0.0001). Using Cox regression, it was verified that patients with large initial decrease in ECD in the first year were indeed more prone to graft failure (data not shown).
Borderie et al 䡠 Long-term Outcome of Corneal Transplantation Table 3. Influence of Donor, Recipient, and Surgical Variables on Graft Survival
Multivariable Level
Predicted Graft Survival
Hazard Ratio (95%CI)
Hazard Ratio (95%CI)
20-year Graft Survival
Observed Graft Survival Univariable Level Variable
N
5-year Graft Survival
Preservation time (weeks) 2–3 4–5
303 841
75% 71%
1 1.4 (1.1–1.8) (P ⫽ 0.02)
1 1.4 (1.1–1.9) (P ⫽ 0.004)
28.4% 24.3% (P ⫽ 0.003)
Recipient rejection status High-risk Low-risk
260 884
60.0% 78.3%
2.3 (1.7–3.0) 1 (P⬍0.001)
2.5 (1.9–3.3) 1 (P⬍0.001)
19.3% 29.5% (P⬍0.001)
Preoperative diagnosis Regraft Bullous keratopathy Trauma Stromal dystrophies Corneal scar after infectious keratitis Endothelial dystrophies Keratoconus
87 449 46 32 151 121 258
49.8% 58.1% 59.8% 82.6% 83.6% 93.0% 97.8%
1.2 (0.8–1.8) 1 1.0 (0.6–1.6) 0.4 (0.2–1.0) 0.4 (0.3–0.6) 0.2 (0.1–0.3) 0.04 (0.02–0.1) (P⬍0.001)
Preoperative intraocular pressure Glaucoma or intraocular pressure ⬎ 20 mmHg No glaucoma and intraocular pressure ⱕ 20 mmHg
248 896
51.8% 81.3%
2.8 (2.1–3.6) 1 (P⬍0.001)
1.6 (1.2–2.2) 1 (P⬍0.001)
12.0% 31.6% (P⬍0.001)
Postoperative lens status Ant chamber intraocular lens Aphakic Post chamber intraocular lens Phakic
311 94 331 408
52.8% 58.1% 75.0% 93.8%
8.1 (5.2–12.7) 7.7 (4.5–13.1) 4.0 (2.5–6.3) 1 (P⬍0.001)
4.5 (2.6–8.1) 4.0 (2.2–7.6) 2.4 (1.4–4.2) 1 (P⬍0.001)
11.8% 13.6% 17.0% 48.5% (P⬍0.001)
Group Penetrating keratoplasty and impaired recipient endothelium Penetrating keratoplasty and normal recipient endothelium Deep anterior lamellar keratoplasty
701 361 82
62.1% 91.1% 97.2%
4.2 (2.9–6.2) 1 0.4 (0.1–1.6) (P⬍0.001)
1.6 (1.0–2.6) 1 0.5 (0.1–1.9) (P ⫽ 0.04)
14.9% 40.6% 63.2% (P⬍0.001)
11.6% 11.0% 33.5% 25.4% 28.4% 25.4% 57.7% (P⬍0.001)
CI ⫽ confidence interval. Note: Only variables significant at a multivariate level are shown. As group and preoperative diagnosis were strongly linked variables only group was used for multivariate analysis. When the Cox model was set up with preoperative diagnosis (in spite of group) similar results were obtained.
Predicted Long-term Graft Survival Overall, the predicted graft survival estimate was 27.2⫾2.0% at 20 years and 2.0⫾0.5% at 30 years (Fig 1). Significant differences were found among the predicted graft survival in the 3 groups of patients (P⬍0.001). It was 14.9⫾2.0% at 20 years and 0.6⫾0.3% at 30 years in patients who underwent penetrating keratoplasty and had impaired recipient endothelium, 40.6⫾3.0% and 2.5⫾1.0% in patients who underwent penetrating keratoplasty and had normal recipient endothelium, and 63.2⫾6.0% and 10.5⫾4.0% in patients who underwent deep anterior lamellar keratoplasty (Fig 2). For all patients who underwent penetrating keratoplasty, these figures were 23.9⫾2.0% and 1.2⫾0.4%.
Discussion In the present study, the overall observed graft survival after corneal transplantation was 74% at 5 years and 64% at 10
years, and the predicted graft survival was 27% at 20 years and 2% at 30 years. Both observed graft survival and predicted graft survival were higher in patients who underwent deep anterior lamellar keratoplasty than in patients who underwent penetrating keratoplasty and had normal recipient endothelium and were higher in patients who underwent penetrating keratoplasty and had normal recipient endothelium than in patients who underwent penetrating keratoplasty and had impaired recipient endothelium. Graft survival at 5 years after penetrating keratoplasty has been reported in studies including a large number of patients with the following survival estimates: 86% in the Cornea Donor Study (1090 eyes), 69% in the Singapore Corneal Transplant Study (901 eyes), 62% in the Australian study (10 952 eyes), and 90% for first-time grafts and 53% for initial regrafts in the study reported by Thompson et al (3992 eyes).1–5 For the 10-year survival, the following estimates have been reported: 72% in the Japanese study reported by Inoue et al14 (396 eyes) and 82% for first-time
2357
Ophthalmology Volume 116, Number 12, December 2009
Figure 2. Kaplan-Meier survival curve showing observed and predicted graft survival in patients who underwent penetrating keratoplasty and had impaired recipient endothelium (green), who underwent penetrating keratoplasty and had normal recipient endothelium (blue), or who had deep anterior lamellar keratoplasty (red). Log-rank for observed survival: P⬍0.001.
grafts and 41% for initial regrafts in the study reported by Thompson et al.5 The survival estimates of the present study (overall, 74% at 5 years and 64% at 10 years; penetrating keratoplasty, 73% at 5 years and 63% at 10 years) seem to be in good accordance with those previously reported. Corneal graft survival also seems similar to patient survival after renal transplantation, which has been reported to be in the range of 60% to 65% 15 years after surgery.15
Indication for corneal transplantation is a major prognosis factor for graft survival, with 5- and 10-year survival of more than 90% in patients with keratoconus or Fuchs endothelial dystrophy, 50% to 70% for patients with bullous keratopathy, 50% and less for regrafts.3–5 The same significant differences in graft survival were found among indications. The outcome of the corneal endothelium 15 years after penetrating keratoplasty has been reported in a series of 67 eyes.16 The average endothelial cell loss from preoperative donor levels was 71%, and the average ECD was 872 cells/mm2. In a series of 15 eyes that underwent penetrating keratoplasty, the ECD averaged 998 cells/mm2 10 years after surgery and 852 cells/mm2 20 years after surgery.17 In the Cornea Donor Study, the median density of 347 eyes with clear grafts was 778 cells/mm2 at 5 years (median cell loss, 70%).1 Ten-year postoperative data from 72 patients who underwent penetrating keratoplasty showed an average ECD of 958 cells/mm2 (– 67%).18 In the present study, the average ECDs at 5 and 10 years were 1058 cells/mm2 (–53%) and 865 cells/mm2 (– 61%), respectively. For anterior lamellar keratoplasty, only short-term data are available. The reported ECDs after anterior lamellar keratoplasty are the following: 2417 cells/mm2 (average postoperative time, 3 years; n ⫽ 20), 2220 cells/mm2 (average postoperative time, 1 year; n ⫽ 24), and 1937 cells/mm2 (average postoperative time, 2 years; n ⫽ 113).19 –21 These figures are in accordance with the current data. Long-term graft survival typically is difficult to assess because of competitive mortality and loss to follow-up in the long term. For postoperative time points of more than 10 years, no studies with a large number of patients and documented follow-up are available. Mathematical models may be useful to help predict the long-term survival based on the observed midterm survival.22 For example, Armitage et al23
Table 4. Influence of Postoperative Variables on Graft Survival
Multivariable Level
Predicted Graft Survival
Observed Graft Survival Univariable Level N
5-year Graft Survival
Hazard Ratio (95% confidence interval)
Hazard Ratio (95% confidence interval)
20-year Graft Survival
917 227
84.8% 44.4%
1 4.8 (3.7–6.2) (P⬍0.001)
1 4.1 (3.1–5.3) (P⬍0.001)
31.5% 13.0% (P⬍0.001)
Postoperative infectious keratitis None At least 1
1115 29
76.8% 37.6%
1 3.3 (2.1–5.4) (P⬍0.001)
1 2.5 (1.5–4.1) (P⬍0.001)
27.5% 16.0% (P⬍0.001)
Postoperative trauma None At least 1
1120 24
75.9% 49.0%
1 2.5 (1.4–4.4) (P ⫽ 0.001)
1 2.7 (1.5–4.8) (P⬍0.001)
27.5% 26.0% (P ⫽ 0.43)
Miscellaneous postoperative events None At least 1
1091 53
76.0% 58.0%
1 1.9 (1.2–3.0) (P ⫽ 0.008)
1 3.0 (1.8–4.9) (P⬍0.001)
27.0% 25.0% (P⬍0.001)
Variable Rejection episodes No rejection event At least 1 rejection event
Only variables significant at a multivariate level are shown.
2358
Borderie et al 䡠 Long-term Outcome of Corneal Transplantation used a biexponential model to describe the postoperative endothelial cell loss, because it is the primary cause of graft failure after the first 5 postoperative years.16 From this model, penetrating grafts with uncomplicated outcome were predicted to have a maximum life-span of 3 decades: the average time to reach a postoperative ECD of 500 cells/ mm2 was 21 years for a graft with a preoperative ECD of 2000 cells/mm2 and 28 years for a graft with 2500 cells/ mm2. However, this analysis might have underestimated postoperative endothelial cell loss because patients whose graft had failed were censored progressively from the analysis, leading to the observation of patients maintaining large cell densities with time. This in turn might have lead to overestimate graft survival. Herein, a joint analysis of endothelial cell loss and time to graft failure was undertaken, allowing explicit account of the informative censoring at work. The mixed-model formulation also allowed for prediction of individual time to graft failure conditional on follow-up and time-dependent profile of ECD. A fully parametric formulation of the model was used to allow for graft survival prediction. It was found that a biphasic decrease for endothelial cell loss fitted the data well, as reported by Armitage et al.23 A log normal distribution was used to describe time to graft failure, because this formulation describes well the variability of biologic processes. The fit of the model was fair compared with observed survivals, although it may lead to slight overestimates the survival of impaired recipients. The model-predicted median time to graft failure was 15 years. The overall observed midterm graft survival was 74% at 5 years and 64% at 10 years, and the predicted long-term graft survival was 48% at 15 years, 27% at 20 years, and 2% at 30 years, which shows limitation of corneal graft lifespan to 2 to 3 decades. These figures are in accordance with the data from the Australian Corneal Graft Registry. In fact, the reported observed graft survival was 60% at 10 years and 46% at 15 years in the Australian study.24 Major differences were found among surgical groups. Patients who underwent anterior lamellar keratoplasty had significantly lower endothelial cell loss and significantly higher predicted graft survival than patients who underwent penetrating keratoplasty and had either impaired or normal recipient endothelium. The observed graft survival was higher in lamellar keratoplasty patients, but the difference did not reach statistical significance. The better outcome of lamellar grafts may originate from absence of donor endothelium, which implies absence of endothelial rejection, and from lack of corneal endothelium surgical wound and scar. From these data, it is still difficult to predict precisely the life-span of a lamellar corneal graft, because follow-up is fewer than 5 years in most patients. In the group of lamellar keratoplasties, the model that predicts postoperative endothelial cell loss was set up with 2 (preoperative and 1-year data) or 3 (preoperative, 1-year, and 3-year data) time points in most cases, whereas in patients who underwent penetrating keratoplasty, data obtained at 2 to 6 time points were available for analysis. This implies that the late phase of endothelial cell loss was predicted less accurately after lamellar keratoplasty than in eyes that underwent penetrating keratoplasty. For a few patients who underwent lamellar keratoplasty and for whom 5-year data were available, the
endothelial cell loss observed between 3 and 5 years was very low, which implies that the model may overestimate the late-phase cell loss in eyes that underwent anterior lamellar keratoplasty. Longer follow-up of patients who underwent anterior lamellar keratoplasty is needed to investigate whether the late phase of endothelial cell loss is similar to that observed after penetrating keratoplasty or if it returns to the normal age-related endothelial cell loss. If the latter hypothesis is conformed, the predicted long-term graft survival after anterior lamellar keratoplasty will be higher than that reported in the present study. Patients who underwent penetrating keratoplasty did better when the recipient endothelium was normal. After penetrating corneal transplantation, 58% of the corneal endothelium (peripheral endothelium) are from recipient origin and 42% (graft endothelium) are from donor origin. If the preoperative peripheral ECD is very low, the total number of corneal endothelial cells available after transplantation will be lower than with a normal preoperative peripheral ECD. In conclusion, for corneal diseases involving the endothelium, penetrating keratoplasty seems to be a good therapeutic approach in elderly patients because the graft lifespan may be similar to the patient life expectancy. Conversely, for young and middle-aged patients, penetrating keratoplasty is only a midterm therapeutic approach. For corneal diseases not involving the endothelium, deep anterior lamellar keratoplasty seems to be a promising therapeutic approach with higher long-term expected survival.
References 1. Cornea Donor Study Investigator Group. The effect of donor age on corneal transplantation outcome: results of the Cornea Donor Study. Ophthalmology 2008;115:620 – 6. 2. Cornea Donor Study Investigator Group. Donor age and corneal endothelial cell loss 5 years after successful corneal transplantation: specular microscopy ancillary study results. Ophthalmology 2008;115:627–32. 3. Tan DT, Janardhanan P, Zhou H, et al. Penetrating keratoplasty in Asian eyes: the Singapore Corneal Transplant Study. Ophthalmology 2008;115:975– 82. 4. Williams KA, Esterman AJ, Bartlett C, et al. How effective is penetrating corneal transplantation? Factors influencing longterm outcome in multivariate analysis. Transplantation 2006; 81:896 –901. 5. Thompson RW Jr, Price MO, Bowers PJ, Price FW Jr. Longterm graft survival after penetrating keratoplasty. Ophthalmology 2003;110:1396 – 402. 6. Maurice DM. The location of the fluid pump in the cornea. J Physiol 1972;221:43–54. 7. Mergler S, Pleyer U. The human corneal endothelium: new insights into electrophysiology and ion channels. Prog Retin Eye Res 2007;26:359 –78. 8. Bourne WM, McLaren JW. Clinical responses of the corneal endothelium. Exp Eye Res 2004;78:561–72. 9. Borderie VM, Scheer S, Touzeau O, et al. Donor organ cultured corneal tissue selection before penetrating keratoplasty. Br J Ophthalmol 1998;82:382– 8. 10. Borderie VM, Touzeau O, Allouch C, et al. The results of successful penetrating keratoplasty using donor organcultured corneal tissue. Transplantation 1999;67:1433– 8.
2359
Ophthalmology Volume 116, Number 12, December 2009 11. Borderie VM, Touzeau O, Bourcier T, et al. Graft reepithelialization after penetrating keratoplasty using organ-cultured donor tissue. Ophthalmology 2006;113:2181– 6. 12. Borderie VM, Werthel AL, Touzeau O, et al. Comparison of techniques used for removing the recipient stroma in anterior lamellar keratoplasty. Arch Ophthalmol 2008;126:31–7. 13. Schluchter MD. Methods for the analysis of informatively censored longitudinal data. Stat Med 1992;11:1861–70. 14. Inoue K, Amano S, Kimura C, et al. Long-term effects of topical cyclosporine A treatment after penetrating keratoplasty. Jpn J Ophthalmol 2000;44:302–5. 15. Joss N, Rodger RS, McMillan MA, Junor BJ. Randomized study comparing cyclosporine with azathioprine one year after renal transplantation-15-year outcome data. Transplantation 2007;83:582–7. 16. Patel SV, Hodge DO, Bourne WM. Corneal endothelium and postoperative outcomes 15 years after penetrating keratoplasty. Am J Ophthalmol 2005;139:311–9. 17. Inoue K, Kimura C, Amano S, et al. Corneal endothelial cell changes twenty years after penetrating keratoplasty. Jpn J Ophthalmol 2002;46:189 –92.
18. Ing JJ, Ing HH, Nelson LR, et al. Ten-year postoperative results of penetrating keratoplasty. Ophthalmology 1998;105: 1855– 65. 19. Morris E, Kirwan JF, Sujatha S, Rostron CK. Corneal endothelial specular microscopy following deep lamellar keratoplasty with lyophilized tissue. Eye 1998;12:619 –22. 20. Panda A, Bageshwar LM, Ray M, et al. Deep lamellar keratoplasty versus penetrating keratoplasty for corneal lesions. Cornea 1999;18:172–5. 21. Sugita J, Kondo J. Deep lamellar keratoplasty with complete removal of pathological stroma for vision improvement. Br J Ophthalmol 1997;81:184 – 8. 22. Remuzzi G, Cravedi P, Perna A, et al, Dual Kidney Transplant Group. Long-term outcome of renal transplantation from older donors. N Engl J Med 2006;354:343–52. 23. Armitage WJ, Dick AD, Bourne WM. Predicting endothelial cell loss and long-term corneal graft survival. Invest Ophthalmol Vis Sci 2003;44:3326 –31. 24. Williams KA, Lowe M, Bartlett C, et al, All Contributors. Risk factors for human corneal graft failure within the Australian corneal graft registry. Transplantation 2008;86: 1720 – 4.
Footnotes and Financial Disclosures Originally received: January 26, 2009. Final revision: April 2, 2009. Accepted: May 7, 2009. Available online: October 7, 2009.
Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Manuscript no. 2009-115.
1
Department of Ophthalmology, Centre Hospitalier National d’Ophtalmologie des XV-XX, UPMC University of Paris, Paris, France. 2
Department of Public Health, INSERM U707, Hôpital Saint Antoine, Assistance Publique—Hôpitaux de Paris, UPMC University of Paris 06, Paris, France.
2360
Supported by UPMC University of Paris 06, Paris, France. Correspondence: Vincent M. Borderie, MD, PhD, Service V, Centre Hospitalier National d’Ophtalmologie des XV-XX, 28 rue de Charenton, 75012 Paris, France. E-mail: [email protected].