Purpose
To investigate graft survival and surgical experience on clinical outcome following deep anterior lamellar keratoplasty (DALK).
Design
Multicenter cohort study.
Methods
The United Kingdom Transplant Database was used to identify patients who had undergone a first DALK or penetrating keratoplasty (PKP) for keratoconus. Data were collected at the time of surgery and at 1, 2, and 5 years postoperatively. Graft survival, best-corrected visual acuity, and refractive error were analyzed for 3 consecutive time periods. DALK outcomes were analyzed according to surgeon experience.
Results
A total of 4521 patients were included. Graft survival was 92% (95% CI: 90–92) for PKP and 90% (95% CI: 88–92) for DALK ( P = .09). For corneal transplants undertaken in the periods 1999–2002, 2002–2005, and 2005–2007, graft survival was 90%, 92%, and 88% following DALK, and 93%, 91%, and 92% following PKP, respectively. There was no evidence of a difference between surgeons in terms of case mix ( P = .4) or outcome ( P = .2). Surgeon experience, in terms of the number of previous DALK undertaken, had no significant effect on outcome. A donor recipient trephine size disparity of 0.5 mm was associated with an increased risk of graft failure for both DALK ( P = .03) and PKP ( P = .002), whereas ocular surface disease was a significant risk factor for DALK ( P = .04) but not PKP.
Conclusions
There has been little change in graft survival for DALK and PKP over the past decade. Ocular surface disease is an important risk factor for graft failure following DALK. A surgical learning curve for DALK could not be demonstrated in terms of clinical outcome.
Penetrating keratoplasty (PKP) has been the predominant corneal transplant procedure for keratoconus. Since the introduction of deep anterior lamellar keratoplasty (DALK) in 1997, however, there has been an increase in its uptake. The proportion of patients undergoing DALK in the United Kingdom (UK) increased from 12% in 1999 to 43% in 2010, however, since 2010 the uptake of DALK has reached a plateau (data from UK Transplant Registry). It is unclear whether the lack of any further uptake is attributable to surgeons’ reservations regarding clinical outcomes, complexity of the DALK surgical technique, or surgical learning curve.
Although there have been many published comparative analyses between DALK and PKP for keratoconus, reported outcomes vary. Graft survival and outcome is likely to be influenced by a variety of factors, such as host and donor factors, developments in surgical technique, and possibly a surgeon learning effect. Several variations and combinations of the 2 widely recognized methods proposed by Melles and Anwar have been and continue to be developed, each with its relative merits. It is, however, unclear if outcomes are changing and whether this is influencing uptake. Moreover, in view of the relative complexity of the technique compared to PKP, it is uncertain whether this is influenced by surgeon experience. Based on 3-year graft survival within the period 1999–2005, we previously reported that the increased failure rate following DALK was associated with a higher rate of early failures and suggested that this might reflect a lack of surgical experience. To address these questions, we investigated 5-year graft survival in 3 consecutive patient cohorts and the effect of surgeon experience on graft survival, visual acuity, and refractive error following DALK.
Methods
Patients
All patients registered on the UK Transplant Registry of National Health Service Blood and Transplant (NHSBT) who had undergone a PKP or DALK as a first transplant for keratoconus between April 1, 1999 and March 31, 2010 were included. The described research methods and analysis plan adhered to the tenets of the Declaration of Helsinki, and Institutional Review Board of NHSBT approval was obtained. Data were collected from the transplant surgery record and follow-up forms at 1, 2, and 5 years following corneal transplant. All transplants in the UK are registered with NHSBT and completion of transplant record and follow-up forms are a requirement for all surgeons undertaking corneal transplantation. Clinical information collected includes: (1) preoperative data such as the presence of inflammation, ocular surface disease (specifically atopic disease and corneal vascularization), and glaucoma; (2) perioperative events including donor and recipient characteristics and complications; and (3) postoperative outcome (visual acuity, complications, refractive data, rejection episodes, graft survival, and reasons for graft failure).
Surgeons
Corneal transplant outcomes are analyzed and reported annually for individual surgeons and collectively for individual transplant centers in the UK. In order to identify a possible surgical outcome learning curve for DALK, graft survival for all surgeons who had undertaken at least 50 PKP prior to undertaking DALK was analyzed according to the number of DALK procedures a surgeon had previously undertaken, grouped in increments of 10 procedures (eg, 1–10 to 120–131 and >131 DALK). This was first investigated collectively, stratifying all surgeons within these intervals, and then for individual surgeons who had performed <50, 50–100, or >100 DALK. In addition to graft survival, visual acuity and refractive error were also analyzed with respect to surgeon experience, taking into account factors such as donor and recipient trephine size and the donor-recipient trephine size disparity. A surgeon’s experience for DALK was stratified according to all DALK he or she had performed for all indications (indications other than and including keratoconus), but survival and clinical outcome was only determined for first DALK grafts performed for keratoconus.
Analysis
The following outcome variables were analyzed: graft failure, rejection episodes, best-corrected visual acuity (BCVA), and refractive error. The number of DALK performed at the time of each transplant was calculated based on data collected in the UK Transplant Registry. A data set was created and ordered by date for each surgeon. This enabled the generation of a DALK number that, when merged into the study data set, was used to categorize the number of DALK that a surgeon had previously undertaken for each first DALK undertaken for keratoconus. An additional comparative 5-year graft survival analysis for DALK and PKP was undertaken for 3 patient cohorts who had undergone a PKP or DALK in (April to March) (1) 1999–2002, (2) 2002–2005, and (3) 2005–2007. All statistical analyses were performed with SAS v9.3 software (SAS Institute Inc, Cary, North Carolina, USA). Differences in categorical and continuous variables between the DALK and PKP groups were tested using χ 2 tests and Student t tests. Kaplan-Meier survival curves and the log-rank test were used to compare univariate differences in graft survival. A Bonferroni correction was made for multiple tests when analyzing individual time periods whereby statistical significance was regarded as P < .01 to correct for multiple hypothesis testing. Cox regression was used to investigate the relative risk of graft failure for each type of graft. Significant factors affecting graft survival for both DALK and PKP selected within the Cox regression model were determined using forward selection methods with significance determined using the Wald test. Visual acuity and refractive outcome data are presented for grafts that were surviving at the time of the reported follow-up. The impact on graft survival of factors such as the size difference between donor and recipient corneal trephines, additional surgical procedures, suturing method (interrupted, continuous, or both), rejection episodes, donor age, and ocular surface disease were also investigated. Refractive data were analyzed using the methods of Long, Kaye, and Harris.
Results
Patient and Surgical Characteristics
A total of 4521 patients had received a first corneal transplant for keratoconus between April 1999 and March 2010; 3297 had undergone a PKP and 1224 a DALK. Follow-up data were reported for 95% of all PKP (n = 3124) performed and 89% of DALK (n = 1086). There were no significant differences in the preoperative characteristics between patients undergoing DALK and PKP, except for BCVA. Mean BCVA was better in those in the DALK (logMAR 1.22) than in the PKP (logMAR 1.38) cohorts ( P < .01), with 43% (DALK) and 36% (PKP) of patients having a preoperative BCVA of better than 6/60 ( P < .0001). A similar number of patients in both groups had ocular surface disease; 42/1086 (3.9%) of DALK and 94/3030 (3.1%) of PKP patients ( P = .17). Surgeons working in tertiary centers or university teaching hospitals perform the majority of all DALK procedures undertaken in the UK. In an assessment of unit activity levels, we found that over the reported time period 92% of DALK were undertaken in such centers.
Donor-host size disparities of 0 mm, 0.25 mm, and 0.5 mm were present in both DALK and PKP groups. There were, however, more patients who had undergone a PKP than a DALK with donor-host trephine difference of 0.5 mm ( P < .001). In addition, a continuous suture was used more frequently for DALK than for PKP ( P = .02). In terms of the 3 cohorts 1999–2002, 2002–2005, and 2005–2007, for DALK, there were fewer cases of ocular surface disease in the 2005–2007 cohort (2%) compared to the 1999–2002 cohort (6%) ( P = .03). A continuous suture was used more often in the 2005–2007 cohort (39%), compared to 25% in the earlier cohorts, where interrupted sutures (46% and 54%) were used more commonly ( P = .001, P = .002). There was a decrease in donor-to-recipient trephine size disparity in the 2005–2007 cohort (46% no disparity), compared to 35% (no disparity) in the 1999–2002 cohort ( P = .02).
Graft Survival
There was no significant difference in the overall 5-year graft survival between DALK and PKP; 92% (95% CI: 90–92) for PKP and 90% (95% CI: 88–92) for DALK ( P = .09). Although there was a lower 5-year ( P = .016) graft survival rate for DALK than for PKP in the 2005–2007 cohort, graft survival for both PKP and DALK was comparatively lower for this cohort; P = .04 ( Figures 1 and 2 ). There were no significant differences in the mean time to graft failure for those grafts that failed within 5 years following PKP (mean 534 days, median 378 days) compared to DALK (mean 436 days, median 357 days); P = .12.
Causes and Factors Associated With Graft Failure
Factors affecting PKP and DALK survival in the Cox proportional hazards model (Wald test) are shown in Tables 1 and 2 , respectively. The presence of ocular surface disease was significantly associated with an increased risk of graft failure following DALK ( P = .05) but not PKP ( P = .07). Graft survival, however, was similar between PKP and DALK, whether ocular surface disease was present ( P = .16) or absent ( P = .38). There was no significant difference comparing graft survival between DALK and PKP in patients with ( P = .16) or without ( P = .38) ocular surface disease.
| Factor | Number at Risk | Relative Risk | 95% CI | P a |
|---|---|---|---|---|
| Other surgical procedures ( P < .0001) | ||||
| No | 3014 | 1.0 | – | |
| Yes | 110 | 2.9 | 1.8–4.7 | <.0001 |
| Trephine difference ( P = .002) | ||||
| 0 mm | 765 | 1.0 | – | |
| 0.25 mm | 2045 | 0.95 | 0.7–1.3 | .8 |
| 0.5 mm | 314 | 1.8 | 1.2–2.7 | .009 |
| Rejection episodes ( P < .0001) | ||||
| No | 2655 | 1.0 | – | |
| Yes | 469 | 3.4 | 2.6–4.6 | <.0001 |
| Factor | Number at Risk | Relative Risk | 95% CI | P a |
|---|---|---|---|---|
| Trephine difference ( P = .07) | ||||
| 0 mm | 454 | 1.0 | – | |
| 0.25 mm | 599 | 1.4 | 0.86–2.3 | .18 |
| 0.5 mm | 33 | 3.0 | 1.1–7.8 | .03 |
| Ocular surface disease ( P = .05) | ||||
| No | 1026 | 1.0 | – | |
| Yes | 42 | 2.4 | 1.0–5.5 | .04 |
| Unknown | 18 | 2.5 | 0.78–8.0 | .13 |
| Rejection episodes ( P = .0002) | ||||
| No | 984 | 1.0 | – | |
| Yes | 102 | 2.8 | 1.6–4.8 | .0002 |
The causes of failure following DALK were reported as rejection (13%; 10/77 patients), infection (3.9%; 3/77 patients), and preoperative surgical complications (7.8%; 6/77 patients). All rejection episodes in DALK patients were reported as stromal rejection. In 41.6% of cases (32/77 patients) the causes of failure included poor visual outcome attributable to interface opacity, poor graft clarity and scarring, trauma, and high refractive error and irregular astigmatism. Cases where the cause of failure was either unknown or not reported accounted for 33.8% of DALK failures (26/77 grafts). For PKP, the main causes of failure were endothelial or both endothelial and stromal rejection (21%; 41/197), primary graft failure (11%; 22/197), infection (7%; 14/197), and endothelial failure (9%; 18/197).
Visual Acuity and Refractive Outcome
There were no significant differences in mean BCVA at 5 years for all surviving grafts; 0.25 logMAR for both PKP and DALK ( P = .8). At 5 years, in the PKP cohort 39% achieved a BCVA of 6/6 Snellen acuity or better, compared to 29% in the DALK cohort ( P = .002). Conversely, there were no significant differences in the proportion of patients with no improvement in their BCVA ( P > .05) between the DALK and PKP groups. There was no evidence to suggest a difference in the proportion of patients with a BCVA of 6/60 or worse at 5 years: 4.9% of DALK and 4.7% of PKP ( P = .9).
There were significant differences in the mean postoperative refractive error between PKP and DALK, with patients who had undergone DALK having a more myopic mean refractive error. At 2 years follow-up mean refractive error for PKP was −2.20/+0.40×6: lower confidence bound (LCI) −14.0/6.5×43; upper confidence bound (UCI) 2.7/6.3×137. At 2 years follow-up the mean refractive error for DALK was −3.60/0.40×102: LCI −14.1/5.7×48; UCI 1.40/6.00×135 ( P < .0001). Similarly, at 5 years mean refractive error was −2.9/0.4×13 (LCI −14.20/6.30×44, UCI 2.70/6.00×138) for PKP and −3.3/0.3×90 (LCI −13.5/5.6×50, UCI 1.7/5.5×137) for DALK ( P = .007). The variation in refractive error was high in both PKP and DALK groups, demonstrated by the wide 95% confidence intervals.
The mean donor and recipient trephine sizes in the PKP and DALK groups were 7.94 mm and 7.74 mm, and 8.0 mm and 7.88 mm, respectively. There was no significant difference in the mean donor or recipient trephine sizes between PKP and DALK patients ( P > .4). Trephine size disparity did not appear to influence the magnitude or variation in refractive error; P > .1 between cohorts (0 mm vs 0.25 mm, 0.25 vs 0.5 mm, 0 mm vs 0.5 mm) ( Table 3 ).
| Corneal Donor-to-Host Trephine Size Difference | Type of Transplant | Patients (n) | Mean | 95% CI |
|---|---|---|---|---|
| 0.00 mm | PKP | 279 | −2.2/0.4×14 | LCI −13.0/5.8×44 UCI 3.3/5.5×138 |
| DALK | 129 | −3.4/0.5×110 | LCI −12.9/5.1×52 UCI 1.4/5.5×138 | |
| 0.25 mm | PKP | 667 | −3.1/0.4×14 | LCI −14.5/6.6×45 UCI 2.2/6.3×138 |
| DALK | 129 | −3.3/0.3×77 | LCI −14.0/5.9×48 UCI 2.1/5.6×135 | |
| 0.50 mm | PKP | 108 | −3.4/0.6×8 | LCI −15.5/5.7×42 UCI 3.8/5.4×137 |
| DALK | 12 | −3.6/1.7×38 | LCI −12.8/6.8×62 UCI 1.2/5.2×166 |
Effect of Surgical Experience on Outcome Following Deep Anterior Lamellar Keratoplasty
There were 355 surgeons in total, of whom 111 performed DALK. All surgeons undertaking DALK were experienced in PKP and had undertaken at least 50 PKP prior to undertaking their first DALK. All surgeons continued to perform PKP, as well as DALK, during the course of the study. The number of surgeons who had undertaken fewer than 10 DALK procedures remained relatively stable over the time of the data collection. In contrast, there was a cumulative increase in the number of surgeons who had performed more than 30 DALK procedures, accounting for the majority of operating surgeons with advancing time.
Surgeon experience and 5-year DALK Kaplan-Meier survival analysis for 3 consecutive time periods, 1999–2002, 2002–2005, and 2005–2007 ( Figure 2 ), showed no evidence of a significant trend ( P = .2, log-rank test). There was no surgeon effect for DALK graft survival ( P = .2) or evidence of a differential selection of case mix (risk factors) between surgeons ( P = .4). For all DALK undertaken within the study period, there was no apparent improvement in graft survival as surgeons became more experienced ( Tables 4 and 5 ).
| Surgeon Experience a | Number of Surgeons | Number of First DALK for Keratoconus | Percentage Survival (95% CI) |
|---|---|---|---|
| 1–10 | 105 | 334 | 92 (88–94) |
| 11–20 | 57 | 218 | 85 (76–92) |
| 21–30 | 40 | 143 | 90 (80–96) |
| 31–40 | 29 | 107 | 94 (80–98) |
| 41–50 | 24 | 86 | 92 (80–96) |
| 51–60 | 18 | 57 | 97 (78–100) |
| 61–70 | 13 | 30 | 88 (58–96) |
| 71–80 | 7 | 23 | 85 (48–96) |
| 81–90 | 7 | 16 | 88 (58–96) |
| 91–100 | 5 | 19 | 79 (36–94) |
| 101–110 | 4 | 18 | 81 (38–96) |
| 111–120 | 3 | 10 | 86 (34–98) |
| 121–130 | 3 | 14 | 100 b |
| 131+ | 3 | 11 | 89 (44–98) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
