Purpose
To compare graft survival between 3 10-year periods and according to surgical techniques performed in the last years.
Design
Cohort study.
Methods
setting : Regional center (Besançon University Hospital, France). patients : All 1132 patients operated on between 1983 and 2014. Graft and patient baseline characteristics, risk factors for failure, surgical procedures, and postoperative corneal status were collected. main outcome measures : Five-year survival rate in the whole cohort; 1-year and 3-year survival rates, respectively, among 88 patients with endothelial dystrophy (ED) or postoperative bullous keratopathy (PBK) operated on using endothelial lamellar keratoplasty (ELK) or penetrating keratoplasty (PK), and among 56 patients with keratoconus operated on using anterior lamellar keratoplasty (ALK) or PK.
Results
Between the 1983–1993 and the 2004–2014 periods, overall 5-year graft survival rate increased from 61.4% to 76.5% ( P = .0004). The main prognostic factors were preoperative diagnosis, graft endothelial density, and postoperative lens status. After adjusting for these factors, difference in survival rates was no longer significant (hazard ratio 0.90 for the second and 1.17 for the third period, compared to the first, P = .4191). Only 1 graft failure, after PK, occurred among the 56 patients with keratoconus. Among the 88 patients with ED or PBK, the 1-year graft survival was higher with PK (90.6%) than with ELK (60.8%) ( P = .0025) but no significance remained after adjustment (hazard ratio 3.22, P = .1304).
Conclusions
Despite numerous changes in graft procedures and surgical techniques, no noticeable improvement in graft survival was found during the last 30 years while taking into account other prognostic factors.
Over the last 110 years of keratoplasty surgeons have been closely observing outcomes, especially graft lifespan. Beyond changing patterns of patients and medical practices, over the last 15 years new surgical techniques have grown in popularity worldwide, thanks to more selective tissue transplantation. Instead of full-thickness transplantation (penetrating keratoplasty, PK), anterior lamellar keratoplasty (ALK) is mainly performed for keratoconus and endothelial lamellar keratoplasty (EK) for Fuchs endothelial dystrophy (FECD) and pseudophakic bullous keratopathy (PBK).
Although excellent visual and survival results in the short term have been reported for both ALK and EK, these outcomes were observed in limited case series published by experienced pioneering surgeons. Recently, a debate has been launched following several multicenter studies that compared PK with lamellar keratoplasty (LK) outcomes in a “real-world” context. In the United Kingdom, no improvement in graft survival was observed between the period 2000–2005, when only PK was performed, and the period 2006–2011, when EK became the mainstream procedure for FECD and PBK. In an Australian study based on data from the national population-based registry, graft survival rates were poorer after LK than after PK, for patients with keratoconus operated on using ALK, as well as for patients with FECD or PBK operated on using EK. Furthermore, 2 Cochrane reviews published in 2014, which compared ALK and EK with PK, concluded that there was no strong evidence of any difference. As a result, several authors emphasized the need for new comparative data from multiple settings and populations.
In this context of controversy, we focused on the “BELIEvE” cohort of patients who received corneal grafts in the Ophthalmology Department of the Besançon Regional University Hospital since 1983. Data from this cohort were analyzed in order to (1) assess the evolution over time of 5-year graft survival, after adjusting for the main known prognosis factors; and (2) look for a difference in survival rates between LK and PK among patient groups with homogeneous indications, by controlling for the main confounding factors.
Methods
The Ophthalmology Department of the Besançon University Hospital provides keratoplasty for any patient living in the region of Franche-Comté (1 178 000 inhabitants, 2% of the French population according to the 2012 census [French National Institute for Statistics and Economic Research (INSEE)]). This observational cohort study was conducted from the data of the “BELIEvE” cohort. “BELIEvE” is the acronym for “Besançon university hospital, an epidemiological survey from survival data of corneal transplant patients” (in French: “ cohorte Bisontine pour l’Estimation de la survie des greffons cornéens au Long terme à partir des données de suivi Internes au CHRU de Besançon – EValuation Epidemiologique Clinique ”). The cohort is based on the data related to all consecutive keratoplasties performed in the Besançon University Hospital between January 1, 1983 and December 31, 2014. These data were collected prospectively in the context of patient care monitoring in the center. The regional Ethics Committee (East II Institutional Review Board, Besançon, France) approved the data collection and subsequent analysis, which complied with the tenets of the Declaration of Helsinki. The data recording and management was also reported to and approved by the French Data Protection Agency (CNIL). All of the information was anonymized before the analysis.
Data Collection in the Cohort and Verification
All of the data were prospectively collected with a standardized form by a single specialized orthoptist, who worked in the department throughout the study period. For each keratoplasty, the following baseline characteristics were collected from medical files and surgical reports: patients’ characteristics at the time of surgery (sex; age; preoperative corneal diagnosis classified according to the International Classification of Diseases, Ninth Edition, Clinical Modification [United States Department of Health and Human Services]; and risk factors for rejection [intraocular pressure >21 mm Hg and/or corneal stromal neovascularization ≥2 corneal quadrants, reported the day before keratoplasty]), graft characteristics (endothelial density of the corneal donor tissue after preservation, measured by the regional eye bank, and graft diameter), surgical procedures (technique—PK, ALK, or EK [including Descemet stripping automated endothelial keratoplasty (DSAEK) and Descemet membrane endothelial keratoplasty (DMEK)] and potential combined surgery [cataract, vitrectomy, or glaucoma surgery at the time of keratoplasty]), and postoperative lens status. The postoperative events were also recorded. Follow-up data were extracted from the medical files completed by the treating surgeon at each follow-up visit. Patients were, as a rule, examined at 1 week; 1, 3, 6, 9, 12, 18, and 24 months; and then yearly after surgery (patient care protocol after keratoplasty is detailed in the Appendix ; Supplemental Material available at AJO.com ). A slit-lamp inspection was performed, intraocular pressure was systematically measured, and in some cases central corneal thickness and endothelial density were measured by specular microscopy.
All data were validated by an expert clinician and when data were missing from the database, the medical files were searched.
Graft and Surgical Management in the Center
Most of the corneal grafts were stored in organ culture as in other French or European centers. Only 10.7% of all the grafts (120 out of 1121, data missing for 11 grafts) were stored in an alternative way between 1983 and 1992 (cold storage). On average, 2–3 corneal senior surgeons carried out the keratoplasties in the center at each time. All were qualified specialists with training in experienced centers. Up to December 2004, PK was the only surgical technique used. LK gradually outnumbered PK from this date onward. ALK was performed starting in 2005 (except for 1 in 1999) and the first EK in 2008. For the EK, the surgical techniques were in line with the procedures first described by Melles and associates and Price and Price and for ALK by Melles and associates.
Study Design: Cohort Study
Patients
Between January 1983 and December 2014, 1168 patients receiving 1621 keratoplasties were registered in the “BELIEvE” cohort. In the 453 patients (38.8%) who received 2 grafts in the same or contralateral eye during the study period, only the first graft was included in the study. Patients who received a first graft in another center were excluded. As a result, 1132 patients were included in the analysis (1132 first keratoplasties performed in the center between 1983 and 2014).
Variables and Outcome
Variables taken into account in the analysis were patients’ baseline characteristics, graft characteristics, surgical procedures (keratoplasty technique and combined surgery), postoperative lens status, and year of the surgery. Preoperative diagnoses were grouped into the 5 following categories: keratoconus, FECD, PBK, infectious corneal diseases, and others (including anterior or stromal dystrophy, trauma, ulcerative complications of a dry ocular surface, undetermined or rare unclassified conditions). High-risk recipients were defined as patients with, at baseline, corneal stromal neovascularization observed in at least 2 quadrants and/or an intraocular pressure greater than 21 mm Hg reported the day before surgery. Infant recipients were not taken into account as “high-risk recipients.” Surgical procedures were classified into 3 groups: PK, ALK, and EK (including DSAEK and DMEK). The study period was divided into 3 10-year intervals: 1983–1993, 1994–2003, and 2004–2014.
Transplant outcomes retained for the analysis were the occurrence of graft failure together with the date of diagnosis. Graft failure was defined as a second graft in the same eye or as an irremediable loss of central clarity for a minimum of 3 consecutive months documented in the medical file by the treating surgeon. This definition is in agreement with the Collaborative Corneal Transplantation Studies. The event date was, respectively, the date of the regraft decision or the first date the patient was seen with an irreversibly cloudy cornea. The closing date to determine graft status was December 31, 2014.
Statistical Analysis
Quantitative variables were categorized in tertiles (patient’s age) or dichotomized around the median (endothelial density, graft diameter). Qualitative variables were expressed as frequency (%) and compared using the Pearson χ 2 or the Fisher test as appropriate. Missing data were treated as a separate category when their frequency was higher than 5%. Graft survival rates were estimated at 1, 3, and 5 years after keratoplasty using the Kaplan-Meier method. For surviving grafts, trial time was calculated as the time between the date of the graft and the date on which the patient was last seen. For failed grafts, trial time was calculated as the time between the date of the graft and the date of failure. The proportional hazard Cox model was used to estimate both unadjusted and adjusted hazard ratios (HR) with their 95% confidence intervals (95% CI) for the different covariates. Covariates with a P value < .20 were retained for multivariable analysis. The proportional hazard assumption was tested using graphical methods.
To assess the evolution over time in the cohort, we first compared baseline characteristics of patients operated on during the 3 10-year periods. Then graft survival rates were estimated for all patients, as well as according to the 3 time periods and to the different statuses of each characteristic. Prognostic values of the different covariates were assessed by using both unadjusted and adjusted HR with 95% CI. For this first analysis, the surgical technique was not taken into account in the multivariable model owing to collinearity with both the time period and the preoperative diagnosis.
To compare survival rates between LK and PK, we then performed similar analyses in the following 2 subgroups: (1) patients operated on for keratoconus between 2005 and 2014, and (2) patients operated on for FECD or PBK between 2008 and 2014. For this last analysis, DSAEK or DMEK were considered a single EK class and survival rates were only estimated at 1 year because of the small numbers of patients/failures.
All reported P values were 2-sided and a P value < .05 was considered statistically significant. Inflation of the alpha risk was accounted in the subgroup analyses using the Bonferroni method. The significance threshold was corrected according to the number of comparisons. Data management and statistical analyses were performed with SAS, version 9.4 for Windows (SAS Institute Inc, Cary, North Carolina, USA).
Results
Evolution of Baseline Characteristics Over Time
The characteristics of patients, surgical procedures, and grafts in the present cohort are described in Table 1 . There were marked changes between the first period (1983–1993) and the 2 other periods (1994–2003 and 2004–2014) for recipient’s age, diagnosis, lens status, and graft characteristics. The proportion of younger patients (less than 46 years old) dropped from 38.6% in the first to 27.4% in the third period ( P = .003). Although the clinical diagnosis was mainly PBK (35%) or keratoconus (27.8%) on the whole, the proportion of PBK decreased noticeably, from 40.1% in the first to 33.1% in the third period ( P < .0001). In contrast, the proportion of FECD increased from 1.4% to 15.5%. In parallel, the proportion of patients with postgraft pseudophakia (posterior chamber intraocular lens) increased sharply, from 23.6% to 43.7% ( P < .0001). Finally, endothelial density of the donor corneal tissue improved, with high-density tissue (≥2240 cells/mm 2 ) increasing from 33.4% in the first to 57.2% in the second period but decreasing in the third period to 22.8%, whereas the proportion of “not reported” density decreased sharply from 41.6% to 1.8% from the first to the last period ( P < .0001).
| Period | P Value a | ||||||
|---|---|---|---|---|---|---|---|
| [1983–1993] (N = 365) | [1994–2003] (N = 486) | [2004–2014] (N = 281) | |||||
| n | % | n | % | n | % | ||
| Patient characteristics | |||||||
| Sex (n = 1117) | .8183 | ||||||
| Male | 187 | 53.3 | 302 | 62.3 | 150 | 53.4 | |
| Female | 264 | 48.7 | 183 | 37.7 | 131 | 46.6 | |
| Age (y) | .0030 | ||||||
| ≤46 | 141 | 38.6 | 201 | 41.4 | 77 | 27.4 | |
| 47–70 | 114 | 31.2 | 146 | 30.0 | 110 | 39.2 | |
| ≥71 | 110 | 30.1 | 139 | 28.6 | 94 | 33.5 | |
| Diagnosis b (n = 1128) | <.0001 | ||||||
| Keratoconus | 77 | 21.2 | 165 | 34.0 | 72 | 25.9 | |
| FECD | 5 | 1.4 | 17 | 3.6 | 43 | 15.5 | |
| PBK | 146 | 40.1 | 162 | 33.3 | 87 | 31.3 | |
| Infectious corneal conditions | 53 | 14.6 | 41 | 8.4 | 20 | 7.2 | |
| Other | 83 | 22.8 | 101 | 20.8 | 56 | 20.1 | |
| High-risk recipient c | .0210 | ||||||
| 63 | 17.3 | 67 | 22.0 | 40 | 14.2 | ||
| Graft characteristics | |||||||
| Endothelial density d | <.0001 | ||||||
| ≥2240 | 122 | 33.4 | 278 | 57.2 | 64 | 22.8 | |
| <2240 | 91 | 24.9 | 204 | 42.0 | 212 | 75.4 | |
| Not reported | 152 | 41.6 | 4 | 0.8 | 5 | 1.8 | |
| Graft diameter (mm) | <.0001 | ||||||
| ≤8 | 173 | 47.4 | 277 | 57.0 | 148 | 52.7 | |
| >8 | 96 | 23.3 | 184 | 37.9 | 100 | 35.6 | |
| Not reported | 96 | 23.3 | 25 | 5.1 | 33 | 11.7 | |
| Surgical procedures | |||||||
| Graft technique (n = 1128) | <.0001 | ||||||
| PK | 364 | 100.0 | 484 | 99.8 | 206 | 73.8 | |
| ALK | 0 | 0.0 | 1 | 0.2 | 17 | 6.1 | |
| EK | 0 | 0.0 | 0 | 0 | 56 | 20.1 | |
| Combined surgery (n = 1087) | .0460 | ||||||
| 90 | 27.2 | 96 | 19.9 | 59 | 21.5 | ||
| Postoperative lens status (n = 1099) | <.0001 | ||||||
| Phakic | 146 | 43.1 | 249 | 51.6 | 112 | 40.4 | |
| Anterior chamber IOL | 91 | 26.8 | 87 | 18 | 42 | 15.2 | |
| Posterior chamber IOL | 80 | 23.6 | 123 | 25.5 | 121 | 43.7 | |
| Aphakic | 22 | 6.5 | 24 | 5.0 | 2 | 0.7 | |
| Outcome | |||||||
| Graft failure in the 5-year postgraft period | <.0001 | ||||||
| 115 | 31.5 | 98 | 20.2 | 44 | 15.7 | ||
a Comparisons were performed with χ 2 or Fisher test as appropriate.
b Preoperative clinical diagnosis.
c Intraocular pressure >21 mm Hg and/or corneal neovascularization ≥2 corneal quadrants, reported the day before keratoplasty.
d Endothelial density of the corneal donor tissue after preservation (cells/mm 2 ).
Follow-up and Graft Survival
The average trial time was 74.7 months (standard deviation [SD]: 86.0, range 0–329) for the 365 patients grafted in the first period, 71.7 months (SD 69.3, range 0–246) for the 486 patients grafted in the second period, and 30.3 months (SD 33.9, range 0–127) for the 291 patients grafted in the third period. In the whole cohort, 257 patients (22.7%) had graft failure during the 5 years post graft. As shown in Table 1 , the proportion of patients with graft failure in the 5-year follow-up dropped by about 10% between the first and the second period (from 31.5% to 20.2%) and further declined to 15.7% in the third period ( P < .0001).
Graft survival rates were 87.8% at 1 year, 77% at 3 years, and 70% at 5 years in the whole cohort, but varied according to the date of surgery ( Table 2 ). The overall 5-year graft survival significantly improved period by period, from 61.4% in the first to 76.5% in the third period ( P = .0004) ( Figure 1 ). Highly significant differences in 5-year survival rates were also observed according to many baseline characteristics listed in Table 2 , apart from recipient sex.
| SR a 1 Year | SR a 5 Years | Unadjusted HR b | |||||
|---|---|---|---|---|---|---|---|
| % | SD | % | SD | HR | 95% CI | P Value | |
| Period of transplant | .0004 | ||||||
| 1983–1993 | 84.3 | 2.0 | 61.4 | 2.9 | 1.00 | ||
| 1994–2003 | 90.2 | 1.4 | 74.2 | 2.3 | 0.60 | 0.46–0.78 | |
| 2004–2014 | 88.3 | 2.1 | 76.5 | 3.5 | 0.64 | 0.45–0.91 | |
| Patient characteristics | |||||||
| Sex | .1002 | ||||||
| male | 88.9 | 1.3 | 72.0 | 2.1 | 1.00 | ||
| female | 86.4 | 1.7 | 66.6 | 2.6 | 1.23 | 0.96–1.57 | |
| Age (y) | <.0001 | ||||||
| ≤46 | 91.6 | 1.4 | 81.7 | 2.2 | 1.00 | ||
| [47–70] | 87.6 | 1.8 | 71.5 | 2.7 | 1.69 | 1.21–2.34 | |
| ≥71 | 83.0 | 2.2 | 49.7 | 3.7 | 3.00 | 2.19–4.12 | |
| Diagnosis c | <.0001 | ||||||
| Keratoconus | 97.9 | 0.9 | 93.1 | 1.7 | 1.00 | ||
| FECD | 89.1 | 4.2 | 72.4 | 7.3 | 5.10 | 2.60–11.65 | |
| PBK | 82.5 | 2.0 | 48.8 | 3.2 | 9.52 | 5.68–15.98 | |
| Infectious corneal diseases | 84.3 | 3.6 | 68.0 | 5.2 | 5.78 | 3.13–10.68 | |
| Others | 84.3 | 2.5 | 71.5 | 3.2 | 5.42 | 3.12–9.43 | |
| High-risk recipient d | .0013 | ||||||
| No | 88.1 | 1.1 | 72.9 | 1.7 | 1.00 | ||
| Yes | 86.2 | 2.5 | 57.1 | 4.2 | 1.58 | 1.20–2.09 | |
| Graft characteristics | |||||||
| Endothelial density e | <.0001 | ||||||
| ≥2240 | 92.1 | 1.3 | 77.7 | 2.3 | 1.00 | ||
| <2240 | 86.7 | 1.7 | 67.9 | 2.6 | 1.52 | 1.14–2.03 | |
| Not reported | 77.8 | 3.4 | 54.2 | 4.3 | 2.59 | 1.87–3.59 | |
| Graft diameter (mm) | .0169 | ||||||
| ≤8 | 88.5 | 1.4 | 71.3 | 2.2 | 1.00 | ||
| >8 | 88.5 | 1.7 | 71.6 | 2.8 | 0.97 | 0.73–1.28 | |
| Not reported | 83.0 | 3.2 | 60.3 | 4.7 | 1.57 | 1.13–2.20 | |
| Surgical procedures | |||||||
| Graft technique (1999–2014) | <.0001 | ||||||
| PK | 93.1 | 1.3 | 79.1 | 2.6 | 1.00 | ||
| ALK | 90.0 | 9.5 | 90.0 | 9.5 | 0.51 | 0.07–3.70 | |
| EK | 66.7 | 6.7 | 56.6 | 7.9 | 4.72 | 2.80–7.94 | |
| Combined surgery | .0006 | ||||||
| No | 89.5 | 1.1 | 73.0 | 1.8 | 1.00 | ||
| Yes | 83.7 | 2.5 | 59.1 | 4.0 | 1.61 | 1.22–2.12 | |
| Postoperative lens status | <.0001 | ||||||
| Phakic | 94.0 | 1.1 | 86.5 | 1.7 | 1.00 | ||
| Anterior chamber IOL | 82.3 | 2.8 | 43.5 | 4.4 | 4.92 | 3.50–6.91 | |
| Posterior chamber IOL | 82.9 | 2.2 | 60.3 | 3.5 | 3.52 | 2.52–4.93 | |
| Aphakic | 83.8 | 5.6 | 51.0 | 8.9 | 3.95 | 2.29–6.81 | |
a Graft survival rates estimated at 1, 3, and 5 years post graft by the Kaplan-Meier method.
b HR estimated from univariable Cox model.
c Preoperative clinical diagnosis.
d Intraocular pressure >21 mm Hg and/or corneal neovascularization ≥2 corneal quadrants, reported the day before keratoplasty.
e Endothelial density of the corneal donor tissue after preservation (cells/mm 2 ).
Factors Affecting Graft Survival in Multivariable Analysis
After adjusting for other significant factors, the difference in survival rates between periods was no longer significant (adjusted HR [95% CI] 0.90 [0.63–1.28] for the second period and 1.17 [0.76–1.83] for the third period, compared to the first, P = .4191) ( Table 3 ). Recipient sex, combined surgery, and graft diameter were not significant either. However, patient age, diagnosis, high-risk status, endothelial density of the corneal donor tissue after preservation, and postoperative lens status remained significant. Interestingly, the prognostic effect of endothelial density was related to the 161 corneal donor tissues with a “not reported” density (HR = 2.55, 95% CI 1.65–3.96, compared to tissues with a density ≥2240 cells/mm 2 , P = .0002). Among the 161 corresponding patients, 152 (94.4%) were transplanted during the first period.
| Adjusted HR | 95% CI | P Value a | |
|---|---|---|---|
| Period of transplant | .4191 | ||
| 1983–1993 | 1.00 | ||
| 1994–2003 | 0.90 | 0.63–1.28 | |
| 2004–2014 | 1.17 | 0.76–1.83 | |
| Patient characteristics | |||
| Sex | .3841 | ||
| Male | 1.00 | ||
| Female | 1.12 | 0.86–1.46 | |
| Age (y) | .0194 | ||
| ≤46 | 1.00 | ||
| 47–70 | 0.60 | 0.40–0.88 | |
| ≥71 | 0.80 | 0.53–1.20 | |
| Diagnosis b | .0008 | ||
| Keratoconus | 1.00 | ||
| FECD | 3.97 | 1.64–9.62 | |
| PBK | 4.31 | 2.10–8.86 | |
| Infectious corneal diseases | 3.93 | 1.94–7.96 | |
| Others | 3.94 | 2.06–7.54 | |
| High-risk recipient c | .0232 | ||
| No | 1.00 | ||
| Yes | 1.42 | 1.05–1.92 | |
| Graft characteristics | |||
| Endothelial density d | .0002 | ||
| ≥2240 | 1.00 | ||
| <2240 | 1.35 | 0.98–1.86 | |
| Not reported | 2.55 | 1.65–3.96 | |
| Graft diameter (mm) | .9314 | ||
| ≤8 | 1.00 | ||
| >8 | 0.99 | 0.74–1.32 | |
| Not reported | 1.07 | 0.72–1.59 | |
| Surgical procedures | |||
| Combined surgery | .1110 | ||
| No | 1.00 | ||
| Yes | 0.77 | 0.57–1.06 | |
| Postoperative lens status | .0006 | ||
| Phakic | 1.00 | ||
| Anterior chamber IOL | 3.22 | 1.84–5.64 | |
| Posterior chamber IOL | 2.53 | 1.56–4.10 | |
| Aphakic | 2.30 | 1.18–4.48 |
b Preoperative clinical diagnosis.
c Intraocular pressure >21 mm Hg and/or corneal neovascularization ≥2 corneal quadrants, reported the day before keratoplasty.
d Endothelial density of the corneal donor tissue after preservation (cells/mm 2 ).
Comparison of Graft Survival According to Surgical Technique in Patients With Keratoconus
Among the 56 patients operated on for keratoconus between 2005 and 2014, only 1 graft failure was observed at 5 years. No patients had preoperative clinical risk factors and there were no significant differences in patients’ characteristics between the PK (n = 39) and the ALK groups (n = 17) (data not shown). As there was only 1 event, no graft survival analysis was performed in this group. The failure occurred at 14.5 months in a 42-year-old man grafted by PK during combined cataract surgery. The endothelial density of the corneal donor tissue after preservation was 2200 cells/mm 2 and the graft diameter measured 8.5 mm. The cataract surgery might have been the most damaging factor.
Comparison of Graft Survival According to Surgical Technique in Patients With Fuchs Endothelial Dystrophy/Pseudophakic Bullous Keratopathy
Among the 88 patients operated on for FECD or PBK between 2008 and 2014, 23 (26.1%) had graft failure during the 5-year follow-up. Among the 23 patients with failure, 16 (69.6%) were operated on using EK. Eighteen failures (78.3%) occurred during the first year post surgery.
Several baseline characteristics significantly differed between the EK and PK groups ( Table 4 ): no patients vs 6 (14.3%) were classified in the “high-risk” group in the EK and PK groups, respectively ( P = .0097), the sex ratio was reversed (EK 0.4, PK 1.8, P = .0015), and in the EK group graft diameter was more often lower than 8 mm ( P < .0001). Because of collinearity between corneal diagnosis and lens status, both variables were combined in a new composite variable. With this last variable, the proportion of patients with PBK tended to be lower in the EK (52.2%) than in the PK group (76.2%), but the difference did not reach significance ( P = .0576).
