- Email: [email protected]
Immunologic Graft Rejection in Descemet's Stripping Endothelial Keratoplasty and Penetrating Keratoplasty for Endothelial Disease
Immunologic Graft Rejection in Descemet’s Stripping Endothelial Keratoplasty and Penetrating Keratoplasty for Endothelial Disease Isaac Ezon, MD,1 Carolyn Y. Shih, MD, MPH,1 Lisa M. Rosen, ScM,2 Tushar Suthar, MD,1 Ira J. Udell, MD1 Objective: To evaluate and compare the cumulative incidence and risk factors for first-episode immunologic graft rejection in Descemet’s stripping automated endothelial keratoplasty (DSAEK) and penetrating keratoplasty (PK) and to identify potential risk factors for rejection. Design: Retrospective chart review. Participants: All patients who underwent PK or DSAEK for endothelial disease at the Department of Ophthalmology, North Shore LIJ, between January 2004 and June 2010. Methods: One hundred sixty-nine PK cases and 122 DSAEK cases were reviewed. All patients had a minimum of 3 months of follow-up, with median follow-up of 36 months in the PK group and 29 months in the DSAEK group. Main Outcome Measures: Cumulative incidence of first-episode immunologic graft rejection in PK and DSAEK cohorts. Risk factors for graft rejection were reviewed. Results: Cumulative incidence of rejection was not significantly different between the DSAEK and PK cohorts (P⬍0.1324). However, among patients without glaucoma, the risk of rejection in PK was higher than that in DSAEK (hazard ratio [HR], 5.56). Prior incisional glaucoma surgery imparted a 3.15 times greater risk of rejection regardless of transplant type. Phakic patients were more likely to experience rejection than patients with a posterior chamber intraocular lens (HR, 3.23; P⬍0.0266), but not more likely than those with an anterior chamber intraocular lens or who were aphakic. Graft failure occurred within 6 months in 31% of PK rejections and none of the DSAEK rejections. Conclusions: Descemet’s stripping automated endothelial keratoplasty and PK did not show a statistically significant difference in the incidence of rejection; however, among nonglaucomatous eyes, there were significantly fewer rejections in those that underwent DSAEK. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Ophthalmology 2013;120:1360 –1365 © 2013 by the American Academy of Ophthalmology.
Descemet’s stripping automated endothelial keratoplasty (DSAEK) offers several advantages over penetrating keratoplasty (PK), including less intraoperative risk, better postoperative visual acuity outcomes, and faster recovery.1 It has been hypothesized that the reduced amount of transplanted tissue in DSAEK as compared with PK also may result in fewer instances of immunologic rejection.2 There are few studies directly comparing the rejection rates of DSAEK with those of PK.2– 4 This study evaluated the cumulative incidence of first-episode immunologic graft rejection between DSAEK and PK and the risk factors associated with this complication.
Patients and Methods This was a retrospective chart review of all patients who underwent PK between January 1, 2004, and June 30, 2010, or who underwent DSAEK between May 1, 2006, and June 30, 2010, at
1360
© 2013 by the American Academy of Ophthalmology Published by Elsevier Inc.
the Department of Ophthalmology, North Shore LIJ. All surgeries were performed by 2 surgeons (I.J.U. and C.Y.S.). The study was approved by the North Shore LIJ Health System Institution Review Board. Potential subjects were identified by reviewing billing records for transplantation procedures. Inclusion criteria included a primary indication of endothelial related disease (Fuchs’ corneal dystrophy, bullous keratopathy [pseudophakic, phakic, aphakic], iridocorneal endothelial syndrome, or failed prior PK or DSAEK) and at least 3 months of follow-up available in the patient’s record. Exclusion criteria included preoperative diagnoses of herpetic disease, interstitial keratitis, infectious keratitis, keratoconus, trauma, and early graft failure (within 2 weeks of procedure).
Surgical Management and Postoperative Care Penetrating Keratoplasty. Penetrating keratoplasty procedures typically were performed with a donor button that was oversized by 0.25 mm. The recipient bed was prepared by partial trephination with a Hessburg-Barron trephine (Barron Precision InstruISSN 0161-6420/13/$–see front matter http://dx.doi.org/10.1016/j.ophtha.2012.12.036
Ezon et al 䡠 Rejection in DSAEK versus PK ments, Grand Rapids, MI) of the appropriate size. Where necessary, anterior vitrectomy and lysis of synechiae were performed. Multiple 10-0 nylon interrupted sutures or running 10-0 nylon with interrupted sutures were placed. At the completion of the procedure, subconjunctival Solu-Medrol (Pfizer Labs, New York, NY) and antibiotic was injected, and Maxitrol ointment (Alcon, Fort Worth, TX) was applied. In cases of combined pars plana vitrectomy or glaucoma procedure, the corneal transplant process was materially the same. It was performed in sequence with these other procedures, and the anesthesia method varied. Descemet’s Stripping Automated Endothelial Keratoplasty. Descemet’s stripping automated endothelial keratoplasty was performed using the technique similar to that described by Terry et al.5 Seventy donor buttons were cut by the surgeon and 52 were precut by the eye bank. Briefly, a temporal perilimbal scleral tunnel incision was created, Descemet’s membrane was scored using a reverse Sinsky hook under viscoelastic and then removed with a forceps. The host bed was scraped with a Terry scraper. Viscoelastic was removed with irrigation and aspiration. Donor tissue was folded, inserted with Charley forceps or Busin glide (15% of cases), and unfolded with a combination of air and balanced salt solution. The anterior chamber was filled with air completely for 10 to 15 minutes then was replaced partially with balanced salt solution. The eye was pharmacologically dilated, and wounds were sealed. Subconjunctival Solu-Medrol and antibiotics were administered at the end of the case. Maxitrol and a patch were then applied to the eye.
Postoperative Management Postoperative examinations for the PK and DSAEK procedure generally were performed at 1 day, 5 to 7 days, and months 1, 2, 4, and 6 after surgery, then every 3 to 4 months thereafter. In both the PK and DSAEK groups, at postoperative day 1, patients were started on prednisolone acetate 1% (generic; or Pred Forte; Allergan, Irvine, CA) 6 to 8 times per day. This was maintained during the first week of follow-up, and then was tapered to 3 to 4 times daily through postoperative month 1. Topical steroid therapy was tapered further by 1 drop daily every 4 to 6 months if the graft remained clear. All patients were maintained on 1 drop of prednisolone acetate 1% indefinitely thereafter. In phakic eyes or those suspected of having a steroid-sensitive rise in intraocular pressure (IOP), topical steroid use was tapered in frequency or was changed to loteprednol (Lotemax; Bausch & Lomb, Rochester, NY) or fluorometholone (Allergan) with or without the addition of glaucoma medications.
Definition and Management of Rejection Episode At the time of study design, an immunologic rejection episode (henceforth a rejection episode) was defined as the presence of keratic precipitates on the endothelium, a rejection line, or subepithelial infiltrates. If a rejection episode was suspected or confirmed, topical Pred Forte was prescribed hourly. Subconjunctival or systemic steroids were not prescribed for any patient. If the graft rejection seemed to be resolving with 1 week of increased steroid administration, the steroids were gradually tapered to the prerejection drop regimen over the ensuing 6 to 12 months.
Determination of Glaucoma Status Eyes with a known prior history of glaucoma substantiated by visual fields, diagnostic optic nerve appearance, prior incisional glaucoma therapy, or specific diagnosis by the referring ophthalmologist were identified as having glaucoma. Eyes that were receiving IOP-lowering agents before intervention, but without specific diagnosis of glaucoma as above, were identified as glaucoma suspects.
Statistical Analysis Follow-up Time. The median follow-up time for DSAEK and PK was calculated using the reverse Kaplan-Meier method as described in Schemper and Smith.6 Rejection Episodes. Analysis of rejection episode required adjustment for the competing risk of failure (if a graft failure occurred without previous rejection, then a future rejection could not be observed). Therefore, Kaplan-Meier was thought to be inadequate for analysis, and time to rejection therefore was analyzed using appropriate methods for competing risks. The causespecific cumulative incidence function for rejection was used to estimate the probability of rejection by a given point in time. Cumulative incidence functions were compared between the PK and DSAEK groups using Gray’s test.7 Subjects in which neither a rejection nor failure occurred were censored using their last known date of follow-up. To examine the association between glaucoma status, transplant procedure, and time to rejection, the above analysis was repeated and stratified by glaucoma status (no glaucoma vs. glaucoma or glaucoma suspect). On finding a significant result with the Gray’s test, Bonferroni-adjusted pairwise multiple comparisons were conducted to determine which groups of patients differed from each other. More specifically, the study compared DSAEK and PK patients with the same glaucoma status, as well as glaucoma or glaucoma suspect and no glaucoma patients with the same transplant procedure. To determine which factors were associated with time to rejection, univariate screening of each factor was carried out using cause-specific Cox (proportional hazards) regression. For purposes of cause-specific analysis, patients who experienced a failure without a previous rejection were considered to be censored along with patients who neither experienced a failure nor a rejection. Factors of interest are listed in Table 1. All factors that were associated significantly with time to rejection in the univariate screen at the level of P⬍0.10 were included in the multivariate model. Transplant procedure always was included in the multivariate model because it was the main variable of interest. Interaction terms between transplant procedure and each factor included in the multivariate model were explored. The interaction term between glaucoma status and repeat procedure also was of interest. If glaucoma status and repeat procedure had a significant association with time to rejection in the univariate analyses, the interaction between these 2 factors also was included in the multivariate model. The likelihood ratio test was used to compare models with
Table 1. Risk Factors Screened for in Univariate and Multivariate (*P⬍0.10 Included) Time-to-Rejection Analysis Factor
P Value
Transplant Glaucoma status Lens Gender Surgeon Prior incisional glaucoma procedure Repeat procedure Concurrent vitrectomy Peripheral anterior synechiae Age at surgery Indication
⬍0.1225 ⬍0.0889 ⬍0.0013 ⬍0.4353 ⬍0.0703 ⬍0.0044 ⬍0.7943 ⬍0.8974 ⬍0.8443 ⬍0.0209 ⬍0.4499
Values in bold type indicate P⬍0.10 and were included in multivariate analysis.
1361
Ophthalmology Volume 120, Number 7, July 2013 Table 2. Demographics and Indication for Surgery
Procedures (n) Mean age (yrs) % Male Indication for procedure (%) Fuchs’ endothelial dystrophy Pseudophakic bullous keratopathy Fuchs and PBK Failed PK Other* Concurrent vitrectomy, no. (%) Prior vitrectomy, no. (%) Concurrent glaucoma surgery, no. (%) Prior incisional glaucoma surgery, no. (%) PAS, no. (%) Lens status, no. (%) PCIOL Phakic ACIOL Aphakic Repeat procedure, no. (%) No glaucoma, no. (%) Glaucoma suspect, no. (%) Glaucoma, no. (%)
Penetrating Keratoplasty
Descemet Stripping Automated Keratoplasty
169 76.1 34.3
122 75.5 52.5
9.5 39.1 5.3 38.5 7.7 32 (18.9) 16 (9.5) 43 (25.4) 52 (30.8)
26.2 39.3 20.5 9.0 4.9 0 (0) 3 (2.5) 0 (3 revisions) 13 (10.7)
36 (21.3)
3 (2.5)
90 (53.3) 21 (12.4) 50 (29.6) 6 (3.6) 68 (40.2) 62 (36.7) 8 (4.7) 99 (58.6)
115 (94.3) 7 (5.7) 0 0 15 (12.3) 80 (65.6) 8 (6.6) 34 (27.9)
ACIOL ⫽ anterior chamber intraocular lens; PAS ⫽ peripheral anterior synechiae; PBK ⫽ pseudophakic bullous keratopathy; PCIOL ⫽ posterior chamber intraocular lens; PK ⫽ penetrating keratoplasty. *Other includes aphakic bullous keratopathy, phakic bullous keratopathy, iridocorneal endothelial syndrome, failed Descemet stripping automated keratoplasty.
and without interaction terms to select a final model with the best fit. All analysis was conducted with SAS software version 9.2 (SAS Inc, Cary, NC) and R software version 2.8.1 (The R Foundation for Statistical Computing, Vienna, Austria).8
After stratifying by glaucoma status (no glaucoma versus glaucoma or glaucoma suspect), there was a trend toward significance when comparing the cumulative incidence curves for rejection (P⬍0.0518, Gray’s test; Fig 2). Among those without glaucoma, PK transplants had a consistently higher probability of rejection (cumulative incidence function at 24 months, 0.0979) over time as compared with DSAEK transplants (cumulative incidence function at 24 months, 0.0324). There was also a significant interaction between transplant type and glaucoma status in the Cox regression analysis (P⬍0.0206) after adjusting for lens type, surgeon, prior incisional glaucoma surgery, and age. The same relationships between risk of rejection and transplant type with glaucoma status were observed (PK without glaucoma vs. DSAEK without glaucoma: hazard ratio, 5.56; 95% CI, 1.122–5.00). To address possible concern for differences in preoperative diagnoses, subgroup analyses were performed for rejection by major indication. There was no significant difference in cumulative incidence curves for PK and DSAEK procedures when performed for Fuchs’ corneal dystrophy (n ⫽ 48; P⬍0.0642, Gray’s test), pseudophakic bullous keratopathy (n ⫽ 114; P⬍0.7475, Gray’s test), and Fuchs’ corneal dystrophy plus pseudophakic bullous keratopathy (n ⫽ 34; P⬍0.4692, Gray’s test). Among those with failed prior PK, 9 PK transplants had an observed rejection (n ⫽ 67) and 0 DSAEK transplants had an observed rejection (n ⫽ 11). To assess risk factors for first-episode immunologic graft rejection, univariate screening was performed on a series of variables as listed in Table 1. Those variables that were significant at the P⬍0.10 level were included in the final multivariate model. This model identified a 3.15 times greater risk of transplant rejection in patients with prior incisional glaucoma surgery compared with those without such surgery (hazard ratio, 3.15; 95% CI, 1.14 – 8.68; P⬍0.0264). Additionally, phakic eyes were at increased risk of rejection compared with eyes with a posterior chamber intraocular lens (hazard ratio, 3.22; 95% CI, 1.15–9.10; P⬍0.0266). The risk of rejection in eyes with an anterior chamber intraocular lens or aphakic eyes was similar to that of phakic eyes (P⬍0.0951– 0.1842). Surgeon and age were not significantly associated with an increased risk of transplant rejection episode (P⬍0.1092 and
Results One hundred sixty-nine PK and 122 DSAEK procedures were identified that met the criteria listed above. Demographics and indications for surgery are listed in Table 2. Mean follow-up for the PK and DSAEK groups was 35.4 months and 29.0 months, respectively. Median follow-up periods by reverse Kaplan-Meier method were 36.1 months (95% confidence interval [CI]. 30.8 – 41.3) and 29.4 months (95% CI, 24.5–35.2), respectively.
Rejection Episodes No statistically significant difference was found in the cumulative incidence of rejection overall in PK versus DSAEK patients (P⬍0.1324, Gray’s test; Fig 1), although there was a trend toward decreased rejection in DSAEK. At the 2-year follow-up, the cumulative incidence of rejection was 11% (95% CI, 71%– 6%) in PK and 5% (95% CI, 21%–1%) in DSAEK. The cumulative incidence of rejection at other time points is reflected in Table 3. Among patients in which a rejection was observed, the median time until first episode graft rejection was similar in both groups: 21.0 months (DSAEK) versus 14.3 months (PK).
1362
Figure 1. Graph showing the cumulative incidence function of first episode immunologic graft rejection. DSAEK ⫽ Descemet’s stripping automated endothelial keratoplasty; PK ⫽ penetrating keratoplasty.
Ezon et al 䡠 Rejection in DSAEK versus PK Table 3. Cumulative Incidence Estimates and 95% Confidence Intervals at Selected Time Points Penetrating Keratoplasty Incidence Estimate Months (95% Confidence Interval) 12 24 36 48
0.07 (0.04–0.11) 0.11 (0.07–0.17) 0.14 (0.09–0.20) 0.16 (0.10–0.22)
Descemet Stripping Automated Endothelial Keratoplasty
Rejection in the Past Year (n)
Rejection-Free Survival at Time Point (n)*
Incidence Estimate (95% Confidence Interval)
Rejection in the Past Year (n)
Rejection-Free Survival at Time Point (n)*
11 7 3 2
133 104 73 49
0.01 (0.00–0.04) 0.05 (0.02–0.11) 0.07 (0.03–0.13) 0.10 (0.05–0.19)
1 4 1 2
95 70 42 23
*The number of patients who have neither experienced a rejection nor been censored. These are the number of eyes that are still at risk to experience a rejection.
P⬍0.2209, respectively) after adjusting for transplant type, glaucoma status, prior incisional glaucoma surgery, and lens type. There were no significant interactions between transplant procedure type and prior incisional glaucoma surgery, lens type, surgeon, or age. Therefore, these interaction terms were excluded from the final model (P⬍0.4073, likelihood ratio test). Additionally, there were no rejections in the 11 cases of DSAEK performed in eyes with previously failed PK, and primary transplant procedures were just as likely to be rejected as repeat transplant procedures (10.6% vs. 10.7%) during the study period. Evaluation of the procedures demonstrating a graft rejection episode revealed that 4 (17%) of 23 PK rejections and all (8 of 8) DSAEK rejections were in patients who reported topical steroid noncompliance. In 2 additional cases of PK rejection, a milder steroid was prescribed to control IOP, and in 3 (13%) cases, topical steroids had been discontinued previously by the treating physician because the patient was phakic. Of note, in 57% (13/23) of PK eyes that experienced a graft rejection episode, the graft ultimately failed. Eight of these (35% of PK rejections) were deemed rejection-associated failures exhibiting no interval restoration of a clear graft. The average time from rejection to ultimate failure in these 8 cases was 3 months. In comparison, 2 of the DSAEK rejections (25%) went on to graft
Figure 2. Graph showing cumulative incidence function of first episode of immunologic graft rejection by glaucoma status. DSAEK ⫽ Descemet’s stripping automated endothelial keratoplasty; PK ⫽ penetrating keratoplasty.
failure, both at 8 months after rejection and without a period of restored graft clarity.
Discussion Rejection Rate It is difficult to compare the rejection episode rates for DSAEK reported in the literature because follow-up time varies, from 2 months to 2 years or more, and inclusion criteria are nonuniform. At 2 years of follow-up Allen et al3 reported a rejection episode rate of 7.5% in endothelial keratoplasty (DSAEK and deep lamellar endothelial keratoplasty). Bahar et al2 observed 1 rejection (2.2%) over 9.8 months (average), and Shih et al9 identified 1 rejection (0.8%) in a cohort of patients with at least 2 months of postoperative follow-up. With one exception, Lee et al’s10 Ophthalmology Technology Assessment reported DSAEK rejection rates to range from 1% to 18%, with an average rejection rate of 8% to 10%. In the present study, the first-episode rejection rate at 2 years of follow-up of 5% (95% CI, 2%–11%) in DSAEK falls within this range. All rejections were in the setting of topical steroid noncompliance. More recent literature on DSAEK and Descemet’s membrane endothelial keratoplasty rejection rates report DSAEK rejection episode rates of 10.3%,11 12%,12 and 14%13 at 2 years. However, a broader definition of immunologic graft rejection was used in some reports,11,13 including but not limited to the clinical findings of anterior chamber cell or flare without keratic precipitates and corneal edema with conjunctival injection or photophobia and without anterior chamber reaction. Reports on first-episode graft rejection rates for PK also have ranged widely from 2.3% to 68%,14 depending on preoperative indication, potential risk factors, and follow-up time. Alldredge and Krachmer15 reported a 10% to 13% graft rejection rate in PKs performed for Fuchs corneal dystrophy and a 26% to 39% rate in PKs performed for failed PK that were followed up for at least 1 year. The present results (13.5% over 36 months [median]) are compatible over the follow-up period. This study was a direct comparison between DSAEK and PK cohorts with similar preoperative diagnoses of endothelial disease or failure. More PK procedures were combined with other procedures such as glaucoma surgery, vitrec-
1363
Ophthalmology Volume 120, Number 7, July 2013 tomy, or lysis of peripheral anterior synechiae than were DSAEK, although this did not influence the results when adjusted for in the multivariate analysis. We did not find a statistically significant difference in the cumulative incidence of first-episode rejection rate between the 2 groups as a whole, although there was a trend toward higher incidence in the PK group. Allen et al3 concluded that corneal transplant rejections were less frequent in endothelial keratoplasty (7.5%) than PK (13%) followed up for 2 years (P ⫽ 0.035). However, a direct comparison between their study and the present one cannot be made because only 24% of their reported endothelial keratoplasty procedures were DSAEK, most being deep lamellar endothelial keratoplasty, and only eyes with Fuchs’ corneal dystrophy or pseudophakic bullous keratopathy were included. Data are not provided on the subset of patients who underwent DSAEK. Price et al4 reported on 598 DSAEK procedures and identified a 7.6% probability of first-episode graft rejection within 1 year and a 12% probability within 2 years. When compared with a separate group of 30 PK procedures at the same institution with a probability of first-episode graft rejection of 14% within 1 year and of 18% within 2 years, the trend for increased rejection in the PK group was not statistically significant (P ⫽ 0.38). Bahar et al2 reported a 2.2% and 4.2% first-episode graft rejection rate in DSAEK and PK, respectively. However, given the limited follow-up time and small sample size (9.8 months for 45 cases and 15.8 months for 48 cases of DSAEK and PK, respectively) and the unusually low rejection episode rate for PK, statistical conclusions cannot be made from these data.
Rejection Risk Factors A significant difference in the rejection rates of PK and DSAEK eyes was identified only among nonglaucomatous eyes in the multivariate Cox regression model. Specifically, among those without glaucoma, PK had a higher risk of rejection than DSAEK. (There was no increased risk among glaucomatous eyes.) Price et al4 identified preoperative glaucoma or steroid-induced IOP elevation as a risk factor for rejection in DSAEK. The present study did not find a statistically significant increase in the risk of rejection in DSAEK eyes with glaucoma compared with DSAEK eyes without glaucoma. This study identified prior incisional glaucoma surgery and lens status as risk factors for an immunologic graft rejection episode, although the hazards were low. Al-Torbak,16 Kwon et al,17 and Hollander et al18 found a 16% to 31% rejection rate in PKs performed with tube shunts. Wiaux et al19 found no difference in rejection rate of DSAEK in eyes with prior glaucoma tube shunts compared with those without prior tube shunt. It is not known why prior incisional glaucoma surgery increased the risk of rejection episode, whereas prior transplant did not. The impact of prior vitrectomy could not be evaluated statistically per se because rejections occurred only in those with prior vitrectomy. One hypothesis may be that such glaucoma intervention actually is a surrogate for other predictive factors not identified herein. It is plausible that these eyes also were more sensitive to topical steroid-induced elevation in IOP requiring
1364
weaker or tapered steroid doses. Phakic status also may have influenced topical steroid regimens in hopes of avoiding premature cataract formation in these eyes. In fact, 4 (17.4%) of 23 PK rejection episodes were in the setting of topical steroid discontinuation resulting from the patients’ phakic status. There is conflicting published data on DSAEK and on PK implicating age, vitrectomy, repeat procedure, and peripheral anterior synechiae as risk factors for rejection after transplant.20 Cox regression analysis in the present population did not support any of these potential risk factors.
Rejection Observations All of the DSAEK rejection episodes were in the setting of patient-admitted topical steroid noncompliance, and 39% of the PK rejection episodes were in the setting of noncompliance or less-than-ideal dosing of topical steroids. These rejection episodes occurred between 1 and 38 months after surgery. It is possible that if topical steroid compliance had been higher, the incidence of rejection in DSAEK may have been lower. In the publication by Allen et al,3 30% of DSAEK rejection episodes and most PK rejection episodes occurred in patients not using topical immunosuppressants. They, too, identified rejections over at least 23 months after surgery. This would suggest the importance of using topical immunosuppressants in both the acute and long-term postoperative care of corneal transplantation. It is not yet clear from available data whether a rejection episode in DSAEK is more or less likely to respond to aggressive immunosuppression than in PK. The rejectionassociated failure rate in endothelial keratoplasty as reported by Allen et al3 was 6% compared with 16% in PK. In a group of patients who overlapped partially with that of Allen et al, 7% of DSAEK rejection episodes led to failure. Others have reported rejection associated failures in 5% to 37% of PK rejections.21–24 Two of the present 8 DSAEK rejections led to failure 8 months after rejection, whereas 35% of the PK rejections led to failure. If a trend toward better responsiveness of DSAEK rejections than PK rejections were to prove significant and not the result of chance, it may affect management of postoperative follow-up. Jordan et al25 reported that approximately one-third of DSAEK rejection episodes are asymptomatic (and identified on routine follow-up); thus, delayed management of these rejections may not have as significant an impact on graft survival as it would in PK. A limitation of this study included its retrospective nature. As such, patients were not assigned randomly to one procedure or the other, and potential bias was introduced. Another concern is that the follow-up time in the 2 groups differed by approximately 6 months, in large part as a function of when DSAEK was introduced at the authors’ institution. However, this concern was addressed with cumulative incidence function analysis. Analysis of the data for only those procedures performed after May 2006 (when DSAEK was introduced) yielded closer follow-up periods and similar results. Finally, although the sample size was large, the number of uncensored observations was low as a result of competing risk (rejections cannot be observed in
Ezon et al 䡠 Rejection in DSAEK versus PK grafts that have failed already) and the period of follow-up. Although this resulted in loss of power, the small numbers of uncensored observations demonstrate that both procedures allow patients to continue rejection free for several months or years after cornea transplantation. In summary, this single-center retrospective chart review of 291 PK and DSAEK procedures found a non–statistically significant trend toward fewer graft rejection episodes in DSAEK procedures compared with PK procedures performed for endothelial disease. In eyes without prior glaucoma, there was a significantly higher risk of rejection among PK patients. Eyes with prior incisional glaucoma therapy were more likely to experience a graft rejection episode in both transplant procedures than in nonsurgical eyes. Based on these data, the patient’s preoperative glaucoma status should be considered strongly when recommending DSAEK or PK, and DSAEK generally should be preferred in its absence. However, additional longer-term prospective studies are warranted to confirm these findings.
10.
11. 12. 13. 14. 15. 16. 17.
References 1. Hjortdal J, Ehlers N. Descemet’s stripping automated endothelial keratoplasty and penetrating keratoplasty for Fuchs’ endothelial dystrophy. Acta Ophthalmol 2009: 87:310 – 4. 2. Bahar I, Kaiserman I, McAllum P, et al. Comparison of posterior lamellar keratoplasty techniques to penetrating keratoplasty. Ophthalmology 2008;115:1525–33. 3. Allan BD, Terry MA, Price FW Jr, et al. Corneal transplant rejection rate and severity after endothelial keratoplasty. Cornea 2007;26:1039 – 42. 4. Price MO, Jordan CS, Moore G, Price FW Jr. Graft rejection episodes after Descemet stripping with endothelial keratoplasty: part two: the statistical analysis of probability and risk factors. Br J Ophthalmol 2009;93:391–5. 5. Terry MA, Shamie N, Chen ES, et al. Endothelial keratoplasty: a simplified technique to minimize graft dislocation, iatrogenic graft failure, and pupillary block. Ophthalmology 2008;115:1179 – 86. 6. Schemper M, Smith TL. A note on quantifying follow-up in studies of failure time. Control Clin Trials 1996;17:343– 6. 7. Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 1988;16:1141–54. 8. Scrucca L, Santucci A, Aversa F. Competing risk analysis using R: an easy guide for clinicians. Bone Marrow Transplant 2007;40:381–7. 9. Shih CY, Ritterband DC, Rubino S, et al. Visually significant and nonsignificant complications arising from Descemet strip-
18. 19.
20. 21. 22. 23. 24.
25.
ping automated endothelial keratoplasty. Am J Ophthalmol 2009;148:837– 43. Lee WB, Jacobs DS, Musch DC, et al. Descemet’s stripping endothelial keratoplasty: safety and outcomes. A report by the American Academy of Ophthalmology. Ophthalmology 2009; 116:1818 –30. Li JY, Terry MA, Goshe J, et al. Graft rejection after Descemet’s stripping automated endothelial keratoplasty: graft survival and endothelial cell loss. Ophthalmology 2012;119:90–4. Anshu A, Price MO, Price FW Jr. Risk of corneal transplant rejection significantly reduced with Descemet’s membrane endothelial keratoplasty. Ophthalmology 2012;119:536 – 40. Wu EI, Ritterband DC, Yu G, et al. Graft rejection following Descemet stripping automated endothelial keratoplasty: features, risk factors and outcomes. Am J Ophthalmol 2012;153:949–57. Panda A, Vanathi M, Kumar A, et al. Corneal graft rejection. Surv Ophthalmol 2007;52:375–96. Alldredge OC, Krachmer JH. Clinical types of corneal transplant rejection: their manifestations, frequency, preoperative correlates, and treatment. Arch Ophthalmol 1981;99:599–604. Al-Torbak A. Graft survival and glaucoma outcome after simultaneous penetrating keratoplasty and Ahmed glaucoma valve implant. Cornea 2003;22:194 –7. Kwon YH, Taylor JM, Hong S, et al. Long-term results of eyes with penetrating keratoplasty and glaucoma drainage tube implant. Ophthalmology 2001;108:272– 8. Hollander DA, Giaconi JA, Holland GN, et al. Graft failure after penetrating keratoplasty in eyes with Ahmed valves. Am J Ophthalmol 2010;150:169 –78. Wiaux C, Baghdasaryan E, Lee OL, et al. Outcomes after Descemet stripping endothelial keratoplasty in glaucoma patients with previous trabeculectomy and tube shunt implantation. Cornea 2011;30:1304 –11. Inoue K, Amano S, Oshika T, Tsuru T. Risk factors for corneal graft failure and rejection in penetrating keratoplasty. Acta Ophthalmol Scand 2001;79:251–5. Sangwan VS, Ramamurthy B, Shah U, et al. Outcome of corneal transplant rejection: a 10-year study. Clin Experiment Ophthalmol 2005;33:623–7. Hudde T, Minassian DC, Larkin DF. Randomised controlled trial of corticosteroid regimens in endothelial corneal allograft rejection. Br J Ophthalmol 1999;83:1348 –52. Hill JC, Ivey A. Corticosteroids in corneal graft rejection: double versus single pulse therapy. Cornea 1994;13:383– 8. Maguire MG, Stark WJ, Gottsch JD, et al, Collaborative Corneal Transplantation Studies Research Group. Risk factors for corneal graft failure and rejection in the Collaborative Corneal Transplantation Studies. Ophthalmology 1994;101:1536–47. Jordan CS, Price MO, Trespalacios R, Price FW JR. Graft rejection episodes after Descemet stripping with endothelial keratoplasty: part one: clinical signs and symptom. Br J Ophthalmol 2009;93:387–90.
Footnotes and Financial Disclosures Originally received: May 8, 2012. Final revision: November 15, 2012. Accepted: December 14, 2012. Available online: March 24, 2013.
Manuscript no. 2012-694.
1
Department of Ophthalmology, Hofstra North Shore–Long Island Jewish School of Medicine, Great Neck, New York.
2
Biostatistics Unit, Feinstein Institute for Medical Research, Manhasset, New York.
Presented in part at: American Academy of Ophthalmology Annual Meeting, October 2010, Chicago, Illinois. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Correspondence: Isaac Ezon, Department of Ophthalmology, Hofstra North Shore–Long Island Jewish School of Medicine, 600 Northern Boulevard, Suite 107, Great Neck, NY 11021. E-mail: [email protected].
1365
Descemet’s stripping automated endothelial keratoplasty (DSAEK) offers several advantages over penetrating keratoplasty (PK), including less intraoperative risk, better postoperative visual acuity outcomes, and faster recovery.1 It has been hypothesized that the reduced amount of transplanted tissue in DSAEK as compared with PK also may result in fewer instances of immunologic rejection.2 There are few studies directly comparing the rejection rates of DSAEK with those of PK.2– 4 This study evaluated the cumulative incidence of first-episode immunologic graft rejection between DSAEK and PK and the risk factors associated with this complication.
Patients and Methods This was a retrospective chart review of all patients who underwent PK between January 1, 2004, and June 30, 2010, or who underwent DSAEK between May 1, 2006, and June 30, 2010, at
1360
© 2013 by the American Academy of Ophthalmology Published by Elsevier Inc.
the Department of Ophthalmology, North Shore LIJ. All surgeries were performed by 2 surgeons (I.J.U. and C.Y.S.). The study was approved by the North Shore LIJ Health System Institution Review Board. Potential subjects were identified by reviewing billing records for transplantation procedures. Inclusion criteria included a primary indication of endothelial related disease (Fuchs’ corneal dystrophy, bullous keratopathy [pseudophakic, phakic, aphakic], iridocorneal endothelial syndrome, or failed prior PK or DSAEK) and at least 3 months of follow-up available in the patient’s record. Exclusion criteria included preoperative diagnoses of herpetic disease, interstitial keratitis, infectious keratitis, keratoconus, trauma, and early graft failure (within 2 weeks of procedure).
Surgical Management and Postoperative Care Penetrating Keratoplasty. Penetrating keratoplasty procedures typically were performed with a donor button that was oversized by 0.25 mm. The recipient bed was prepared by partial trephination with a Hessburg-Barron trephine (Barron Precision InstruISSN 0161-6420/13/$–see front matter http://dx.doi.org/10.1016/j.ophtha.2012.12.036
Ezon et al 䡠 Rejection in DSAEK versus PK ments, Grand Rapids, MI) of the appropriate size. Where necessary, anterior vitrectomy and lysis of synechiae were performed. Multiple 10-0 nylon interrupted sutures or running 10-0 nylon with interrupted sutures were placed. At the completion of the procedure, subconjunctival Solu-Medrol (Pfizer Labs, New York, NY) and antibiotic was injected, and Maxitrol ointment (Alcon, Fort Worth, TX) was applied. In cases of combined pars plana vitrectomy or glaucoma procedure, the corneal transplant process was materially the same. It was performed in sequence with these other procedures, and the anesthesia method varied. Descemet’s Stripping Automated Endothelial Keratoplasty. Descemet’s stripping automated endothelial keratoplasty was performed using the technique similar to that described by Terry et al.5 Seventy donor buttons were cut by the surgeon and 52 were precut by the eye bank. Briefly, a temporal perilimbal scleral tunnel incision was created, Descemet’s membrane was scored using a reverse Sinsky hook under viscoelastic and then removed with a forceps. The host bed was scraped with a Terry scraper. Viscoelastic was removed with irrigation and aspiration. Donor tissue was folded, inserted with Charley forceps or Busin glide (15% of cases), and unfolded with a combination of air and balanced salt solution. The anterior chamber was filled with air completely for 10 to 15 minutes then was replaced partially with balanced salt solution. The eye was pharmacologically dilated, and wounds were sealed. Subconjunctival Solu-Medrol and antibiotics were administered at the end of the case. Maxitrol and a patch were then applied to the eye.
Postoperative Management Postoperative examinations for the PK and DSAEK procedure generally were performed at 1 day, 5 to 7 days, and months 1, 2, 4, and 6 after surgery, then every 3 to 4 months thereafter. In both the PK and DSAEK groups, at postoperative day 1, patients were started on prednisolone acetate 1% (generic; or Pred Forte; Allergan, Irvine, CA) 6 to 8 times per day. This was maintained during the first week of follow-up, and then was tapered to 3 to 4 times daily through postoperative month 1. Topical steroid therapy was tapered further by 1 drop daily every 4 to 6 months if the graft remained clear. All patients were maintained on 1 drop of prednisolone acetate 1% indefinitely thereafter. In phakic eyes or those suspected of having a steroid-sensitive rise in intraocular pressure (IOP), topical steroid use was tapered in frequency or was changed to loteprednol (Lotemax; Bausch & Lomb, Rochester, NY) or fluorometholone (Allergan) with or without the addition of glaucoma medications.
Definition and Management of Rejection Episode At the time of study design, an immunologic rejection episode (henceforth a rejection episode) was defined as the presence of keratic precipitates on the endothelium, a rejection line, or subepithelial infiltrates. If a rejection episode was suspected or confirmed, topical Pred Forte was prescribed hourly. Subconjunctival or systemic steroids were not prescribed for any patient. If the graft rejection seemed to be resolving with 1 week of increased steroid administration, the steroids were gradually tapered to the prerejection drop regimen over the ensuing 6 to 12 months.
Determination of Glaucoma Status Eyes with a known prior history of glaucoma substantiated by visual fields, diagnostic optic nerve appearance, prior incisional glaucoma therapy, or specific diagnosis by the referring ophthalmologist were identified as having glaucoma. Eyes that were receiving IOP-lowering agents before intervention, but without specific diagnosis of glaucoma as above, were identified as glaucoma suspects.
Statistical Analysis Follow-up Time. The median follow-up time for DSAEK and PK was calculated using the reverse Kaplan-Meier method as described in Schemper and Smith.6 Rejection Episodes. Analysis of rejection episode required adjustment for the competing risk of failure (if a graft failure occurred without previous rejection, then a future rejection could not be observed). Therefore, Kaplan-Meier was thought to be inadequate for analysis, and time to rejection therefore was analyzed using appropriate methods for competing risks. The causespecific cumulative incidence function for rejection was used to estimate the probability of rejection by a given point in time. Cumulative incidence functions were compared between the PK and DSAEK groups using Gray’s test.7 Subjects in which neither a rejection nor failure occurred were censored using their last known date of follow-up. To examine the association between glaucoma status, transplant procedure, and time to rejection, the above analysis was repeated and stratified by glaucoma status (no glaucoma vs. glaucoma or glaucoma suspect). On finding a significant result with the Gray’s test, Bonferroni-adjusted pairwise multiple comparisons were conducted to determine which groups of patients differed from each other. More specifically, the study compared DSAEK and PK patients with the same glaucoma status, as well as glaucoma or glaucoma suspect and no glaucoma patients with the same transplant procedure. To determine which factors were associated with time to rejection, univariate screening of each factor was carried out using cause-specific Cox (proportional hazards) regression. For purposes of cause-specific analysis, patients who experienced a failure without a previous rejection were considered to be censored along with patients who neither experienced a failure nor a rejection. Factors of interest are listed in Table 1. All factors that were associated significantly with time to rejection in the univariate screen at the level of P⬍0.10 were included in the multivariate model. Transplant procedure always was included in the multivariate model because it was the main variable of interest. Interaction terms between transplant procedure and each factor included in the multivariate model were explored. The interaction term between glaucoma status and repeat procedure also was of interest. If glaucoma status and repeat procedure had a significant association with time to rejection in the univariate analyses, the interaction between these 2 factors also was included in the multivariate model. The likelihood ratio test was used to compare models with
Table 1. Risk Factors Screened for in Univariate and Multivariate (*P⬍0.10 Included) Time-to-Rejection Analysis Factor
P Value
Transplant Glaucoma status Lens Gender Surgeon Prior incisional glaucoma procedure Repeat procedure Concurrent vitrectomy Peripheral anterior synechiae Age at surgery Indication
⬍0.1225 ⬍0.0889 ⬍0.0013 ⬍0.4353 ⬍0.0703 ⬍0.0044 ⬍0.7943 ⬍0.8974 ⬍0.8443 ⬍0.0209 ⬍0.4499
Values in bold type indicate P⬍0.10 and were included in multivariate analysis.
1361
Ophthalmology Volume 120, Number 7, July 2013 Table 2. Demographics and Indication for Surgery
Procedures (n) Mean age (yrs) % Male Indication for procedure (%) Fuchs’ endothelial dystrophy Pseudophakic bullous keratopathy Fuchs and PBK Failed PK Other* Concurrent vitrectomy, no. (%) Prior vitrectomy, no. (%) Concurrent glaucoma surgery, no. (%) Prior incisional glaucoma surgery, no. (%) PAS, no. (%) Lens status, no. (%) PCIOL Phakic ACIOL Aphakic Repeat procedure, no. (%) No glaucoma, no. (%) Glaucoma suspect, no. (%) Glaucoma, no. (%)
Penetrating Keratoplasty
Descemet Stripping Automated Keratoplasty
169 76.1 34.3
122 75.5 52.5
9.5 39.1 5.3 38.5 7.7 32 (18.9) 16 (9.5) 43 (25.4) 52 (30.8)
26.2 39.3 20.5 9.0 4.9 0 (0) 3 (2.5) 0 (3 revisions) 13 (10.7)
36 (21.3)
3 (2.5)
90 (53.3) 21 (12.4) 50 (29.6) 6 (3.6) 68 (40.2) 62 (36.7) 8 (4.7) 99 (58.6)
115 (94.3) 7 (5.7) 0 0 15 (12.3) 80 (65.6) 8 (6.6) 34 (27.9)
ACIOL ⫽ anterior chamber intraocular lens; PAS ⫽ peripheral anterior synechiae; PBK ⫽ pseudophakic bullous keratopathy; PCIOL ⫽ posterior chamber intraocular lens; PK ⫽ penetrating keratoplasty. *Other includes aphakic bullous keratopathy, phakic bullous keratopathy, iridocorneal endothelial syndrome, failed Descemet stripping automated keratoplasty.
and without interaction terms to select a final model with the best fit. All analysis was conducted with SAS software version 9.2 (SAS Inc, Cary, NC) and R software version 2.8.1 (The R Foundation for Statistical Computing, Vienna, Austria).8
After stratifying by glaucoma status (no glaucoma versus glaucoma or glaucoma suspect), there was a trend toward significance when comparing the cumulative incidence curves for rejection (P⬍0.0518, Gray’s test; Fig 2). Among those without glaucoma, PK transplants had a consistently higher probability of rejection (cumulative incidence function at 24 months, 0.0979) over time as compared with DSAEK transplants (cumulative incidence function at 24 months, 0.0324). There was also a significant interaction between transplant type and glaucoma status in the Cox regression analysis (P⬍0.0206) after adjusting for lens type, surgeon, prior incisional glaucoma surgery, and age. The same relationships between risk of rejection and transplant type with glaucoma status were observed (PK without glaucoma vs. DSAEK without glaucoma: hazard ratio, 5.56; 95% CI, 1.122–5.00). To address possible concern for differences in preoperative diagnoses, subgroup analyses were performed for rejection by major indication. There was no significant difference in cumulative incidence curves for PK and DSAEK procedures when performed for Fuchs’ corneal dystrophy (n ⫽ 48; P⬍0.0642, Gray’s test), pseudophakic bullous keratopathy (n ⫽ 114; P⬍0.7475, Gray’s test), and Fuchs’ corneal dystrophy plus pseudophakic bullous keratopathy (n ⫽ 34; P⬍0.4692, Gray’s test). Among those with failed prior PK, 9 PK transplants had an observed rejection (n ⫽ 67) and 0 DSAEK transplants had an observed rejection (n ⫽ 11). To assess risk factors for first-episode immunologic graft rejection, univariate screening was performed on a series of variables as listed in Table 1. Those variables that were significant at the P⬍0.10 level were included in the final multivariate model. This model identified a 3.15 times greater risk of transplant rejection in patients with prior incisional glaucoma surgery compared with those without such surgery (hazard ratio, 3.15; 95% CI, 1.14 – 8.68; P⬍0.0264). Additionally, phakic eyes were at increased risk of rejection compared with eyes with a posterior chamber intraocular lens (hazard ratio, 3.22; 95% CI, 1.15–9.10; P⬍0.0266). The risk of rejection in eyes with an anterior chamber intraocular lens or aphakic eyes was similar to that of phakic eyes (P⬍0.0951– 0.1842). Surgeon and age were not significantly associated with an increased risk of transplant rejection episode (P⬍0.1092 and
Results One hundred sixty-nine PK and 122 DSAEK procedures were identified that met the criteria listed above. Demographics and indications for surgery are listed in Table 2. Mean follow-up for the PK and DSAEK groups was 35.4 months and 29.0 months, respectively. Median follow-up periods by reverse Kaplan-Meier method were 36.1 months (95% confidence interval [CI]. 30.8 – 41.3) and 29.4 months (95% CI, 24.5–35.2), respectively.
Rejection Episodes No statistically significant difference was found in the cumulative incidence of rejection overall in PK versus DSAEK patients (P⬍0.1324, Gray’s test; Fig 1), although there was a trend toward decreased rejection in DSAEK. At the 2-year follow-up, the cumulative incidence of rejection was 11% (95% CI, 71%– 6%) in PK and 5% (95% CI, 21%–1%) in DSAEK. The cumulative incidence of rejection at other time points is reflected in Table 3. Among patients in which a rejection was observed, the median time until first episode graft rejection was similar in both groups: 21.0 months (DSAEK) versus 14.3 months (PK).
1362
Figure 1. Graph showing the cumulative incidence function of first episode immunologic graft rejection. DSAEK ⫽ Descemet’s stripping automated endothelial keratoplasty; PK ⫽ penetrating keratoplasty.
Ezon et al 䡠 Rejection in DSAEK versus PK Table 3. Cumulative Incidence Estimates and 95% Confidence Intervals at Selected Time Points Penetrating Keratoplasty Incidence Estimate Months (95% Confidence Interval) 12 24 36 48
0.07 (0.04–0.11) 0.11 (0.07–0.17) 0.14 (0.09–0.20) 0.16 (0.10–0.22)
Descemet Stripping Automated Endothelial Keratoplasty
Rejection in the Past Year (n)
Rejection-Free Survival at Time Point (n)*
Incidence Estimate (95% Confidence Interval)
Rejection in the Past Year (n)
Rejection-Free Survival at Time Point (n)*
11 7 3 2
133 104 73 49
0.01 (0.00–0.04) 0.05 (0.02–0.11) 0.07 (0.03–0.13) 0.10 (0.05–0.19)
1 4 1 2
95 70 42 23
*The number of patients who have neither experienced a rejection nor been censored. These are the number of eyes that are still at risk to experience a rejection.
P⬍0.2209, respectively) after adjusting for transplant type, glaucoma status, prior incisional glaucoma surgery, and lens type. There were no significant interactions between transplant procedure type and prior incisional glaucoma surgery, lens type, surgeon, or age. Therefore, these interaction terms were excluded from the final model (P⬍0.4073, likelihood ratio test). Additionally, there were no rejections in the 11 cases of DSAEK performed in eyes with previously failed PK, and primary transplant procedures were just as likely to be rejected as repeat transplant procedures (10.6% vs. 10.7%) during the study period. Evaluation of the procedures demonstrating a graft rejection episode revealed that 4 (17%) of 23 PK rejections and all (8 of 8) DSAEK rejections were in patients who reported topical steroid noncompliance. In 2 additional cases of PK rejection, a milder steroid was prescribed to control IOP, and in 3 (13%) cases, topical steroids had been discontinued previously by the treating physician because the patient was phakic. Of note, in 57% (13/23) of PK eyes that experienced a graft rejection episode, the graft ultimately failed. Eight of these (35% of PK rejections) were deemed rejection-associated failures exhibiting no interval restoration of a clear graft. The average time from rejection to ultimate failure in these 8 cases was 3 months. In comparison, 2 of the DSAEK rejections (25%) went on to graft
Figure 2. Graph showing cumulative incidence function of first episode of immunologic graft rejection by glaucoma status. DSAEK ⫽ Descemet’s stripping automated endothelial keratoplasty; PK ⫽ penetrating keratoplasty.
failure, both at 8 months after rejection and without a period of restored graft clarity.
Discussion Rejection Rate It is difficult to compare the rejection episode rates for DSAEK reported in the literature because follow-up time varies, from 2 months to 2 years or more, and inclusion criteria are nonuniform. At 2 years of follow-up Allen et al3 reported a rejection episode rate of 7.5% in endothelial keratoplasty (DSAEK and deep lamellar endothelial keratoplasty). Bahar et al2 observed 1 rejection (2.2%) over 9.8 months (average), and Shih et al9 identified 1 rejection (0.8%) in a cohort of patients with at least 2 months of postoperative follow-up. With one exception, Lee et al’s10 Ophthalmology Technology Assessment reported DSAEK rejection rates to range from 1% to 18%, with an average rejection rate of 8% to 10%. In the present study, the first-episode rejection rate at 2 years of follow-up of 5% (95% CI, 2%–11%) in DSAEK falls within this range. All rejections were in the setting of topical steroid noncompliance. More recent literature on DSAEK and Descemet’s membrane endothelial keratoplasty rejection rates report DSAEK rejection episode rates of 10.3%,11 12%,12 and 14%13 at 2 years. However, a broader definition of immunologic graft rejection was used in some reports,11,13 including but not limited to the clinical findings of anterior chamber cell or flare without keratic precipitates and corneal edema with conjunctival injection or photophobia and without anterior chamber reaction. Reports on first-episode graft rejection rates for PK also have ranged widely from 2.3% to 68%,14 depending on preoperative indication, potential risk factors, and follow-up time. Alldredge and Krachmer15 reported a 10% to 13% graft rejection rate in PKs performed for Fuchs corneal dystrophy and a 26% to 39% rate in PKs performed for failed PK that were followed up for at least 1 year. The present results (13.5% over 36 months [median]) are compatible over the follow-up period. This study was a direct comparison between DSAEK and PK cohorts with similar preoperative diagnoses of endothelial disease or failure. More PK procedures were combined with other procedures such as glaucoma surgery, vitrec-
1363
Ophthalmology Volume 120, Number 7, July 2013 tomy, or lysis of peripheral anterior synechiae than were DSAEK, although this did not influence the results when adjusted for in the multivariate analysis. We did not find a statistically significant difference in the cumulative incidence of first-episode rejection rate between the 2 groups as a whole, although there was a trend toward higher incidence in the PK group. Allen et al3 concluded that corneal transplant rejections were less frequent in endothelial keratoplasty (7.5%) than PK (13%) followed up for 2 years (P ⫽ 0.035). However, a direct comparison between their study and the present one cannot be made because only 24% of their reported endothelial keratoplasty procedures were DSAEK, most being deep lamellar endothelial keratoplasty, and only eyes with Fuchs’ corneal dystrophy or pseudophakic bullous keratopathy were included. Data are not provided on the subset of patients who underwent DSAEK. Price et al4 reported on 598 DSAEK procedures and identified a 7.6% probability of first-episode graft rejection within 1 year and a 12% probability within 2 years. When compared with a separate group of 30 PK procedures at the same institution with a probability of first-episode graft rejection of 14% within 1 year and of 18% within 2 years, the trend for increased rejection in the PK group was not statistically significant (P ⫽ 0.38). Bahar et al2 reported a 2.2% and 4.2% first-episode graft rejection rate in DSAEK and PK, respectively. However, given the limited follow-up time and small sample size (9.8 months for 45 cases and 15.8 months for 48 cases of DSAEK and PK, respectively) and the unusually low rejection episode rate for PK, statistical conclusions cannot be made from these data.
Rejection Risk Factors A significant difference in the rejection rates of PK and DSAEK eyes was identified only among nonglaucomatous eyes in the multivariate Cox regression model. Specifically, among those without glaucoma, PK had a higher risk of rejection than DSAEK. (There was no increased risk among glaucomatous eyes.) Price et al4 identified preoperative glaucoma or steroid-induced IOP elevation as a risk factor for rejection in DSAEK. The present study did not find a statistically significant increase in the risk of rejection in DSAEK eyes with glaucoma compared with DSAEK eyes without glaucoma. This study identified prior incisional glaucoma surgery and lens status as risk factors for an immunologic graft rejection episode, although the hazards were low. Al-Torbak,16 Kwon et al,17 and Hollander et al18 found a 16% to 31% rejection rate in PKs performed with tube shunts. Wiaux et al19 found no difference in rejection rate of DSAEK in eyes with prior glaucoma tube shunts compared with those without prior tube shunt. It is not known why prior incisional glaucoma surgery increased the risk of rejection episode, whereas prior transplant did not. The impact of prior vitrectomy could not be evaluated statistically per se because rejections occurred only in those with prior vitrectomy. One hypothesis may be that such glaucoma intervention actually is a surrogate for other predictive factors not identified herein. It is plausible that these eyes also were more sensitive to topical steroid-induced elevation in IOP requiring
1364
weaker or tapered steroid doses. Phakic status also may have influenced topical steroid regimens in hopes of avoiding premature cataract formation in these eyes. In fact, 4 (17.4%) of 23 PK rejection episodes were in the setting of topical steroid discontinuation resulting from the patients’ phakic status. There is conflicting published data on DSAEK and on PK implicating age, vitrectomy, repeat procedure, and peripheral anterior synechiae as risk factors for rejection after transplant.20 Cox regression analysis in the present population did not support any of these potential risk factors.
Rejection Observations All of the DSAEK rejection episodes were in the setting of patient-admitted topical steroid noncompliance, and 39% of the PK rejection episodes were in the setting of noncompliance or less-than-ideal dosing of topical steroids. These rejection episodes occurred between 1 and 38 months after surgery. It is possible that if topical steroid compliance had been higher, the incidence of rejection in DSAEK may have been lower. In the publication by Allen et al,3 30% of DSAEK rejection episodes and most PK rejection episodes occurred in patients not using topical immunosuppressants. They, too, identified rejections over at least 23 months after surgery. This would suggest the importance of using topical immunosuppressants in both the acute and long-term postoperative care of corneal transplantation. It is not yet clear from available data whether a rejection episode in DSAEK is more or less likely to respond to aggressive immunosuppression than in PK. The rejectionassociated failure rate in endothelial keratoplasty as reported by Allen et al3 was 6% compared with 16% in PK. In a group of patients who overlapped partially with that of Allen et al, 7% of DSAEK rejection episodes led to failure. Others have reported rejection associated failures in 5% to 37% of PK rejections.21–24 Two of the present 8 DSAEK rejections led to failure 8 months after rejection, whereas 35% of the PK rejections led to failure. If a trend toward better responsiveness of DSAEK rejections than PK rejections were to prove significant and not the result of chance, it may affect management of postoperative follow-up. Jordan et al25 reported that approximately one-third of DSAEK rejection episodes are asymptomatic (and identified on routine follow-up); thus, delayed management of these rejections may not have as significant an impact on graft survival as it would in PK. A limitation of this study included its retrospective nature. As such, patients were not assigned randomly to one procedure or the other, and potential bias was introduced. Another concern is that the follow-up time in the 2 groups differed by approximately 6 months, in large part as a function of when DSAEK was introduced at the authors’ institution. However, this concern was addressed with cumulative incidence function analysis. Analysis of the data for only those procedures performed after May 2006 (when DSAEK was introduced) yielded closer follow-up periods and similar results. Finally, although the sample size was large, the number of uncensored observations was low as a result of competing risk (rejections cannot be observed in
Ezon et al 䡠 Rejection in DSAEK versus PK grafts that have failed already) and the period of follow-up. Although this resulted in loss of power, the small numbers of uncensored observations demonstrate that both procedures allow patients to continue rejection free for several months or years after cornea transplantation. In summary, this single-center retrospective chart review of 291 PK and DSAEK procedures found a non–statistically significant trend toward fewer graft rejection episodes in DSAEK procedures compared with PK procedures performed for endothelial disease. In eyes without prior glaucoma, there was a significantly higher risk of rejection among PK patients. Eyes with prior incisional glaucoma therapy were more likely to experience a graft rejection episode in both transplant procedures than in nonsurgical eyes. Based on these data, the patient’s preoperative glaucoma status should be considered strongly when recommending DSAEK or PK, and DSAEK generally should be preferred in its absence. However, additional longer-term prospective studies are warranted to confirm these findings.
10.
11. 12. 13. 14. 15. 16. 17.
References 1. Hjortdal J, Ehlers N. Descemet’s stripping automated endothelial keratoplasty and penetrating keratoplasty for Fuchs’ endothelial dystrophy. Acta Ophthalmol 2009: 87:310 – 4. 2. Bahar I, Kaiserman I, McAllum P, et al. Comparison of posterior lamellar keratoplasty techniques to penetrating keratoplasty. Ophthalmology 2008;115:1525–33. 3. Allan BD, Terry MA, Price FW Jr, et al. Corneal transplant rejection rate and severity after endothelial keratoplasty. Cornea 2007;26:1039 – 42. 4. Price MO, Jordan CS, Moore G, Price FW Jr. Graft rejection episodes after Descemet stripping with endothelial keratoplasty: part two: the statistical analysis of probability and risk factors. Br J Ophthalmol 2009;93:391–5. 5. Terry MA, Shamie N, Chen ES, et al. Endothelial keratoplasty: a simplified technique to minimize graft dislocation, iatrogenic graft failure, and pupillary block. Ophthalmology 2008;115:1179 – 86. 6. Schemper M, Smith TL. A note on quantifying follow-up in studies of failure time. Control Clin Trials 1996;17:343– 6. 7. Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 1988;16:1141–54. 8. Scrucca L, Santucci A, Aversa F. Competing risk analysis using R: an easy guide for clinicians. Bone Marrow Transplant 2007;40:381–7. 9. Shih CY, Ritterband DC, Rubino S, et al. Visually significant and nonsignificant complications arising from Descemet strip-
18. 19.
20. 21. 22. 23. 24.
25.
ping automated endothelial keratoplasty. Am J Ophthalmol 2009;148:837– 43. Lee WB, Jacobs DS, Musch DC, et al. Descemet’s stripping endothelial keratoplasty: safety and outcomes. A report by the American Academy of Ophthalmology. Ophthalmology 2009; 116:1818 –30. Li JY, Terry MA, Goshe J, et al. Graft rejection after Descemet’s stripping automated endothelial keratoplasty: graft survival and endothelial cell loss. Ophthalmology 2012;119:90–4. Anshu A, Price MO, Price FW Jr. Risk of corneal transplant rejection significantly reduced with Descemet’s membrane endothelial keratoplasty. Ophthalmology 2012;119:536 – 40. Wu EI, Ritterband DC, Yu G, et al. Graft rejection following Descemet stripping automated endothelial keratoplasty: features, risk factors and outcomes. Am J Ophthalmol 2012;153:949–57. Panda A, Vanathi M, Kumar A, et al. Corneal graft rejection. Surv Ophthalmol 2007;52:375–96. Alldredge OC, Krachmer JH. Clinical types of corneal transplant rejection: their manifestations, frequency, preoperative correlates, and treatment. Arch Ophthalmol 1981;99:599–604. Al-Torbak A. Graft survival and glaucoma outcome after simultaneous penetrating keratoplasty and Ahmed glaucoma valve implant. Cornea 2003;22:194 –7. Kwon YH, Taylor JM, Hong S, et al. Long-term results of eyes with penetrating keratoplasty and glaucoma drainage tube implant. Ophthalmology 2001;108:272– 8. Hollander DA, Giaconi JA, Holland GN, et al. Graft failure after penetrating keratoplasty in eyes with Ahmed valves. Am J Ophthalmol 2010;150:169 –78. Wiaux C, Baghdasaryan E, Lee OL, et al. Outcomes after Descemet stripping endothelial keratoplasty in glaucoma patients with previous trabeculectomy and tube shunt implantation. Cornea 2011;30:1304 –11. Inoue K, Amano S, Oshika T, Tsuru T. Risk factors for corneal graft failure and rejection in penetrating keratoplasty. Acta Ophthalmol Scand 2001;79:251–5. Sangwan VS, Ramamurthy B, Shah U, et al. Outcome of corneal transplant rejection: a 10-year study. Clin Experiment Ophthalmol 2005;33:623–7. Hudde T, Minassian DC, Larkin DF. Randomised controlled trial of corticosteroid regimens in endothelial corneal allograft rejection. Br J Ophthalmol 1999;83:1348 –52. Hill JC, Ivey A. Corticosteroids in corneal graft rejection: double versus single pulse therapy. Cornea 1994;13:383– 8. Maguire MG, Stark WJ, Gottsch JD, et al, Collaborative Corneal Transplantation Studies Research Group. Risk factors for corneal graft failure and rejection in the Collaborative Corneal Transplantation Studies. Ophthalmology 1994;101:1536–47. Jordan CS, Price MO, Trespalacios R, Price FW JR. Graft rejection episodes after Descemet stripping with endothelial keratoplasty: part one: clinical signs and symptom. Br J Ophthalmol 2009;93:387–90.
Footnotes and Financial Disclosures Originally received: May 8, 2012. Final revision: November 15, 2012. Accepted: December 14, 2012. Available online: March 24, 2013.
Manuscript no. 2012-694.
1
Department of Ophthalmology, Hofstra North Shore–Long Island Jewish School of Medicine, Great Neck, New York.
2
Biostatistics Unit, Feinstein Institute for Medical Research, Manhasset, New York.
Presented in part at: American Academy of Ophthalmology Annual Meeting, October 2010, Chicago, Illinois. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Correspondence: Isaac Ezon, Department of Ophthalmology, Hofstra North Shore–Long Island Jewish School of Medicine, 600 Northern Boulevard, Suite 107, Great Neck, NY 11021. E-mail: [email protected].
1365