Penetrating Keratoplasty for With and Without Resolved Corneal Hydrops: Long-term Results

JAY J. MEYER, AKILESH GOKUL, ALEXANDRA Z. CRAWFORD, AND CHARLES N.J. MCGHEE

PURPOSE: To evaluate the long-term risk of endothelial hydrops is frequently seen in eyes with keratoconus and rejection, graft survival, and associated factors following has been reported to occur in approximately 3% of patients penetrating keratoplasty (PK) for keratoconus, with and with this condition, but also occurs in other ectatic condi- without prior resolved corneal hydrops. tions such as pellucid marginal degeneration and keratoglo- 1,2 DESIGN: Retrospective cohort study. bus. The natural course of hydrops typically involves the METHODS: Primary outcome measures were endothelial spontaneous resolution of corneal edema over 2– rejection–free survival and failure-free survival. Secondary 4 months.3,4 Upon resolution of the edema, patients may outcome measures were corneal neovascularization have minimal to severe vision loss depending on the following hydrops and complications following PK. degree of residual scarring. RESULTS: A total of 245 eyes underwent PK for kera- Following resolution of hydrops, up to 60% of eyes may toconus with mean follow-up of 5.6 ± 3.6 years. Eyes require keratoplasty for visual rehabilitation.1 Although with prior hydrops (n [ 74) had lower endothelial rejec- penetrating keratoplasty (PK) performed for keratoconus tion–free survival rates compared with eyes without prior carries a favorable overall prognosis, endothelial graft hydrops: 86.5% ± 4.0% vs 86.5% ± 2.6% at 1 year, rejection may result in graft failure.5 Numerous studies 61.0% ± 6.2% vs 76.9% ± 3.3% at 5 years, and 45.8% have reported the long-term outcomes of PK for keratoco- ± 10.1% vs 70.9% ± 4.3% at 10 years, respectively nus and factors associated with increased risk of rejection (P [ .023). Multivariate analysis identified factors asso- and/or failure.5–19 However, only a few studies have ciated with endothelial rejection as age £25 years (P [ specifically assessed the outcomes of PK for eyes with .017), corneal neovascularization (P [ .001), donor prior resolved hydrops, with conflicting results.1,20,21 A trephination size >8mm(P [ .017), and poor clinic study by Tuft and associates1 found a higher risk of endo- attendance (P [ .015). There was no difference in the thelial rejection following PK in 87 eyes with prior failure-free survival rates with and without prior resolved hydrops, although other associated and poten- hydrops: 98.6% ± 1.3% vs 97.1% ± 1.3% at 1 year, tially confounding factors were not evaluated. Basu and 97.3% ± 1.9% vs 95.1% ± 1.9% at 5 years, and 97.3% associates20 also found a higher rate of endothelial rejec- ± 1.9% vs 92.2% ± 2.7% at 10 years, respectively tion in their study of 32 eyes with prior hydrops and (P [ .42). Corneal neovascularization was present at subanalysis identified longer duration of corneal hydrops the time of PK in 44.6% of eyes with prior hydrops and (>3 months) and ocular allergies as significant factors. 7.6% without prior hydrops (P < .001). However, Akova and associates21 did not find an CONCLUSIONS: Corneal neovascularization, a frequent increased rate of endothelial rejection in 35 eyes with complication of corneal hydrops, was associated with prior hydrops. increased risk of endothelial rejection following PK. The current study assessed the long-term outcomes of However, allograft survival was similar in eyes with and penetrating keratoplasty for keratoconus in a large series without prior hydrops. (Am J Ophthalmol 2016;169: of eyes with and without prior resolved corneal hydrops 282–289. Ó 2016 Elsevier Inc. All rights reserved.) to evaluate the long-term risk of endothelial rejection, graft survival, and associated factors. CUTE CORNEAL HYDROPS IS THE DEVELOPMENT OF marked corneal edema owing to a break in the A Descemet membrane what allows aqueous humor to enter the corneal stroma and epithelium. Corneal METHODS

WE CONDUCTED A RETROSPECTIVE COHORT STUDY OF ALL Accepted for publication Jul 1, 2016. From the Department of Ophthalmology, New Zealand National Eye patients who underwent PK for keratoconus between Centre, Faculty of Medical and Health Sciences, University of January 2000 and June 2013 in the Ophthalmology Depart- Auckland, Auckland, New Zealand. ment, Greenlane Clinical Centre, Auckland, which is the Inquiries to Jay J. Meyer, Department of Ophthalmology, Private Bag 92019, University of Auckland, Auckland, New Zealand; e-mail: major tertiary center for corneal transplantation in New [email protected] Zealand. The study was approved by the University of

282 © 2016 ELSEVIER INC.ALL RIGHTS RESERVED. 0002-9394/$36.00 http://dx.doi.org/10.1016/j.ajo.2016.07.001 Auckland Ethics Committee (#013907) and followed the required for each transplant procedure. Corneal neovascu- tenets of the Declaration of Helsinki. larization was recorded as the number of quadrants of Exclusion criteria included post-PK follow-up of less neovascularization present and the number of quadrants than 2 years, repeat PK, prior hydrops of unknown duration of neovascularization crossing the trephination. or treatment, hydrops complicated by microbial , Endothelial rejection was defined clinically as the acute PK performed for nonoptical reasons, and PK performed onset of anterior with overlying graft edema. Endo- before clinical resolution of hydrops. thelial rejection episodes were all initially treated with topical corticosteroids (prednisolone acetate 1%) every DATA COLLECTION: The medical records of all eligible hour. Patients treated with intravenous (IV) methylpred- patients were reviewed by 1 experienced clinical investi- nisolone received 500–1000 mg daily for 1–3 days. Graft gator (J.M.). Acute corneal hydrops was defined clinically failure was defined clinically as the loss of transparency as the onset of corneal edema in eyes with keratoconus for at least 3 months with associated decline in vision. with or without an identifiable break in Descemet mem- Compliance with clinic appointments was calculated as brane and without signs of an alternative diagnosis. The the number of appointments attended divided by the num- duration of hydrops was calculated as the date of onset of ber of appointments scheduled, during the first 2 years symptoms to the date of complete disappearance of corneal following PK. edema by slit-lamp examination. Atopy (nonocular) was defined as a history of asthma, ANALYSIS: The primary outcome measures were (1) atopic dermatitis/eczema, or allergic rhinitis and was self- endothelial rejection–free allograft survival and (2) reported by the patient and/or was identified by review of pa- failure-free allograft survival. The rejection-free survival tients’ nonophthalmologic electronic medical record or elec- time was calculated as the date of the PK to the date of tronic pharmacy records for medications used to treat these the first episode of endothelial rejection. Secondary ana- conditions. Ocular allergy was diagnosed based on a history lyses were degree of corneal neovascularization in eyes of symptoms of ocular itching and ocular findings of a papil- with prior hydrops and complications in all eyes following lary reaction or hyperemia of the , or if topical PK. medications for allergic were prescribed. All statistical analyses were performed using SPSS 19.0 The treatments of episodes of hydrops were recorded, for Windows (SPSS; IBM, Chicago, Illinois, USA). Preop- including type, frequency, and duration of topical cortico- erative features were compared using the x2 test. Primary steroid, if prescribed. All episodes of acute corneal hydrops analyses involved Kaplan-Meier survival analysis to esti- were treated with topical medications at the discretion of mate the cumulative probability (reported as percentage the clinician. with standard error) of endothelial rejection–free allograft All PK procedures were performed using standard survival and failure-free allograft survival; univariate anal- techniques under general anesthesia. Donor were ysis of potential risk factors for endothelial rejection, stored at 34 C in New Zealand Eye Bank organ culture me- performed using the log-rank test; and multivariate analysis dium. The donor corneal button was typically oversized by of factors with significant univariate associations, 0.25 mm and sutured using 10-0 nylon in an interrupted, performed using a Cox proportional hazards model to esti- continuous, or combined interrupted and continuous mate the relative effect of variables on allograft survival. fashion. Patients were examined 1 day, 1 week, 1 month, Secondary analyses were focused on potential risk fac- and 3 months postoperatively, and then scheduled for exam- tors for endothelial rejection specific to eyes with prior inations approximately every 3 months. All patients received hydrops only, and on risk factors for the development a topical antibiotic (chloramphenicol 0.5% or ciprofloxacin of corneal neovascularization prior to undergoing PK. 0.3%) 3–4 times daily and corticosteroid (prednisolone ace- Univariate associations were determined using the x2 tate 1%) every 1–2 hours for the first week. Thereafter, the test, log-rank test, and independent samples t test. A P corticosteroid was typically reduced to 4 times daily and value of less than .05 was considered statistically signifi- tapered by 1 drop every 3 months to 1 drop daily and was cant. continued until suture removal. Loose or broken sutures were removed upon presentation. Intact sutures were routinely removed in one sitting in the operating theatre at 18–24 months post-PK using aseptic technique, followed RESULTS by topical antibiotic coverage for 2 weeks and topical corticosteroid with g. prednisolone acetate 1% 4 times a OF THE 410 EYES THAT UNDERWENT PK FOR KERATOCONUS day for 1 month and discontinued thereafter if no previous in the study period, 245 were included in the study (74 rejection or significant risk factors for rejection. with prior hydrops, 171 without prior hydrops), with a Corneal neovascularization was recorded by the surgeon mean follow-up of 5.6 6 3.6 years. Excluded eyes were as at the time of PK and was reported to the New Zealand follows: 96 (23.4%) repeat PK, 23 eyes (5.6%) that showed National Eye Bank on a standardized pro forma, as is evidence of prior hydrops but had not been seen in the

VOL. 169 PENETRATING KERATOPLASTY WITH AND WITHOUT RESOLVED HYDROPS 283 TABLE 1. Baseline Characteristics of Eyes That Underwent Penetrating Keratoplasty for Keratoconus With and Without Prior Hydrops

Characteristics Total, n (%) Hydrops, n (%) No Hydrops, n (%) P Value

Sex Male 138 (56.3) 39 (52.7) 99 (57.9) .270 Female 107 (43.7) 35 (47.3) 72 (42.1) Age at PK (y) <_25 91 (37.1) 34 (45.9) 57 (33.3) .042* >25 154 (62.9) 40 (54.1) 114 (66.7) Laterality of operated eye Right 112 (45.7) 37 (50) 75 (43.9) .228 Left 133 (54.3) 37 (50) 96 (56.1) Ocular allergy Present 44 (18.0) 25 (33.8) 19 (11.1) <.001* Absent 201 (82.0) 49 (66.2) 152 (88.9) Atopy Yes 159 (64.9) 51 (68.9) 108 (63.2) .236 No 86 (35.1) 23 (31.1) 63 (36.8) FIGURE 1. Comparison of Kaplan-Meier curves showing Corneal neovascularization endothelial rejection–free allograft survival for eyes following Present 46 (18.8) 33 (44.6) 13 (7.6) penetrating keratoplasty with and without previous hydrops. Absent 199 (81.2) 41 (55.4) 158 (92.4) <.001* Vessels crossing trephination Yes 39 (15.9) 29 (39.2) 10 (5.8) <.001* No 7 (2.9) 4 (5.4) 3 (1.8) hydrops was 24.8 6 7.1 years. The mean duration of Absent 199 (81.2) 41 (55.4) 158 (92.4) corneal hydrops was 3.2 6 2.6 months, and the duration Donor trephine size (mm) between resolution of hydrops and PK was 17.3 6 <_8 117 (47.8) 23 (31.1) 94 (55.0) <.001* 24.9 months. The cohort of eyes with prior hydrops had a >8 128 (52.2) 51 (68.9) 77 (45.0) higher percentage of patients with age <25 years at the Recipient trephine size (mm) time of PK, greater frequency of ocular allergy, greater de- <_ 8 229 (93.5) 67 (90.5) 162 (94.7) .173 gree of corneal neovascularization (both overall and >8 16 (6.5) 7 (9.5) 9 (5.3) 2 crossing the trephination), and a higher percentage of Donor ECD (cells/mm ) > <_3000 87 (35.5) 24 (32.4) 63 (36.8) .304 donor button size 8mm(Table 1). >3000 158 (64.5) 50 (67.6) 108 (63.2) Appointments attendeda OUTCOMES OF PRIMARY ANALYSES: The overall <_80% 94 (38.4) 26 (35.1) 68 (39.8) rejection-free survival rates were 86.5% 6 2.2% at 1 year, >80% 151 (61.6) 48 (64.9) 103 (60.2) .295 72.1% 6 3.0% at 5 years, and 64.1% 6 4.0% at 10 years. Suture complicationb An endothelial rejection episode occurred in 29 of the 74 Yes 69 17 (23.0) 52 (30.4) eyes (39.2%) with prior hydrops and 42 of 171 eyes No 176 57 (77.0) 119 (69.6) .151 (24.5%) without prior hydrops (P ¼ .031). There was a sig- ¼ ECD ¼ endothelial cell density; PK ¼ penetrating keratoplasty. nificant difference (P .023) in the endothelial rejection– Values marked with an asterisk indicate statistically significant free allograft survival rates for the eyes with and without 6 6 results. prior hydrops: 86.5% 4.0% vs 86.5% 2.6% at 1 year, aPercentage attended during first 2 years following PK. 61.0% 6 6.2% vs 76.9% 6 3.3% at 5 years, and 45.8% 6 bPostoperative loose and/or broken suture(s) and/or suture in- 10.1% vs 70.9 6 4.3% at 10 years, respectively (Figure 1). filtrate(s). On univariate analysis, significant associations with decreased 5-year endothelial rejection–free survival were as follows: female sex, age <_25 years, previous episode of service acutely, 34 eyes (8.3%) with inadequate (<2 years) hydrops, presence of corneal neovascularization, corneal post-PK follow-up, 9 eyes that underwent nonoptical PK, 2 neovascularization crossing trephination, donor trephine eyes with PK prior to complete resolution of hydrops, and 1 size >8 mm, and attendance of <_80% of clinic appoint- eye with microbial keratitis at the time of hydrops. ments in the first 2 years following PK (Table 2). Female The mean age of the patients at the time of PK was 29.6 subjects had lower mean compliance with clinic appoint- 6 9.9 years (26.6 6 7.0 years with prior hydrops, 30.9 6 ments (78.3% 6 23.3%) compared to male subjects 10.6 years without hydrops; P ¼ .002). In the 74 eyes (82.98% 6 18.37%), though this difference was not signif- with hydrops, the mean age of patients at the onset of acute icant (P ¼ .086).

284 AMERICAN JOURNAL OF OPHTHALMOLOGY SEPTEMBER 2016 TABLE 2. Univariate Associations of Endothelial Rejection– Free Survival Rates 5 Years After Penetrating Keratoplasty for Keratoconus

Endothelial Rejection–Free Survival at 5 Factor Total Rejection Years 6 SE (%) P Value

Gender Male 138 33 75.3 6 3.9 .043* Female 107 38 67.9 6 4.7 Age at PKP (y) <_25 91 34 65.5 6 5.0 .013* >25 154 37 76.1 6 3.7 Laterality of operated eye Right 112 27 76.0 6 4.3 .201 Left 133 44 69.0 6 4.2 Ocular allergy Present 44 14 71.7 6 3.4 .812 Absent 201 57 74.0 6 6.8 Atopy FIGURE 2. Comparison of Kaplan-Meier survival curves Present 159 53 68.7 6 3.8 .061 demonstrating failure-free allograft survival for eyes following Absent 86 18 78.3 6 5.0 keratoplasty with and without previous hydrops. Prior episode of hydrops Present 74 29 61.0 6 6.2 .023* 6 Absent 171 42 76.9 3.3 Multivariate analysis showed a significantly higher risk Duration of hydrops (mo) of endothelial rejection at 5 years in eyes of patients <_25 No hydrops 171 42 76.9 6 3.3 .038* <_3 47 17 63.7 6 7.7 years of age at PK (hazard ratio 1.82; 95% confidence inter- ¼ >3 27 12 58.8 6 9.6 val 1.11–2.98; P .017), presence of corneal neovasculari- Corneal neovascularization zation (hazard ratio 2.99; 95% confidence interval Present 46 46 45.8 6 8.0 <.001* 1.58–5.66; P ¼ .001), donor trephination size >8mm Absent 199 25 78.2 6 3.1 (hazard ratio 1.86; 95% confidence interval 1.12–3.08; Vessels crossing trephination P ¼ .017), and attendance of <_80% of clinic appointments Yes 39 22 43.8 6 8.9 <0.001* in the first 2 years following PK (hazard ratio 1.80; 95% No 7 3 53.6 6 20.1 confidence interval 1.12–2.89; P ¼ .015). Absent 199 46 78.2 6 3.1 All keratoplasties that failed had at least 1 prior episode Donor trephine size (mm) of endothelial rejection in the study period. Graft failure <_8 117 26 81.6 6 3.8 .007* occurred in 2 of 74 eyes (2.7%) with prior hydrops and 9 >8 128 45 63.6 6 4.5 Recipient trephine size (mm) of 171 eyes (5.3%) without prior hydrops during the study ¼ <_8 229 65 72.8 6 3.1 .470 period. There was no difference (P .42) in the failure- >8 16 6 62.5 6 12.1 free survival rates of eyes with and without prior hydrops: Donor ECD (cells/mm2) 98.6% 6 1.3% vs 97.1% 6 1.3% at 1 year, 97.3% 6 1.9% <_3000 87 29 67.0 6 5.3 .212 vs 95.1% 6 1.9% at 5 years, and 97.3% 6 1.9% vs 92.2% >3000 158 42 75.0 6 3.6 6 2.7% at 10 years, respectively (Figure 2). Univariate anal- Appointments attendeda ysis did not identify any factors associated with graft failure. <_80% 94 36 64.6 6 5.1 .010* Subgroup analysis of all 71 eyes that experienced a rejec- >80% 151 35 76.9 6 3.7 b tion event did not show any association between adminis- Suture complication tration of IV methylprednisolone at the time of rejection Present 69 23 68.6 6 5.7 .415 and graft survival, as 3 of 21 (14.3%) grafts of patients Absent 176 48 73.5 6 3.5 that did not receive IV methylprednisolone eventually ECD ¼ endothelial cell density; PK ¼ penetrating keratoplasty; failed, compared with 8 of 50 (16.0%) that received IV SE ¼ standard error. methylprednisolone (P ¼ .58). Values marked with an asterisk indicate statistically significant results. OUTCOMES OF SECONDARY ANALYSES: Corneal aPercentage attended during first 2 years following PK. b neovascularization. Overall, corneal neovascularization Postoperative loose and/or broken suture(s) and/or suture was present in 44.6% of eyes following hydrops and infiltrate(s). 7.6% of eyes without prior hydrops at the time of PK

VOL. 169 PENETRATING KERATOPLASTY WITH AND WITHOUT RESOLVED HYDROPS 285 TABLE 3. Univariate Associations for Development of Corneal Neovascularization Prior to Undergoing Penetrating Keratoplasty for Keratoconus in Eyes With Prior Hydrops Only and All Eyes That Underwent Penetrating Keratoplasty for Keratoconus

Eyes With Prior Hydrops That Underwent PK for Keratoconus

Neovascularization Neovascularization

Factor Total, n (%) Present, n (%) Absent, n (%) P Value

Steroids initiated at presentation of hydrops Yes 39 (52.7) 19 (57.6) 20 (48.8) .302 No 35 (47.3) 14 (42.4) 21 (51.2) Ocular allergy Present 25 (33.8) 15 (36.6) 10 (30.3) .375 Absent 49 (66.2) 26 (63.4) 23 (69.7) Atopy Present 51 (68.9) 28 (68.3) 23 (69.7) .550 Absent 23 (31.1) 13 (31.7) 10 (30.3)

All Eyes That Underwent PK for Keratoconus

Neovascularization Neovascularization

Factor Total, n (%) Present, n (%) Absent, n (%) P Value

Ocular allergy Yes 44 (18.0) 13 (28.3) 31 (15.6) .039* No 201 (82.0) 33 (71.7) 168 (84.4) Atopy Yes 159 (64.9) 34 (73.9) 125 (62.8) .104 No 86 (35.1) 12 (26.1) 74 (37.2)

PK ¼ penetrating keratoplasty. Values marked with an asterisk indicate statistically significant results.

(P < .001). In eyes with prior hydrops, corneal neovascula- topical corticosteroids on presentation with acute hydrops rization was present at the time of PK in 19 eyes (57.6%) (n ¼ 39, 13 with rejection events) were compared with eyes treated with topical steroids and in 20 eyes (48.8%) not that did not receive corticosteroids (n ¼ 35, 16 with rejec- treated with topical steroids upon presentation (P ¼ tion events), there was no statistically significant effect on .302) (Table 3). There was no difference in the mean age the endothelial rejection–free allograft survival rates at 5 of patients at the onset of hydrops for eyes with corneal years (P ¼ .10): 68.6% 6 7.5% and 49.8% 6 10.3%, neovascularization compared with those without respectively. neovascularization: 25.2 6 7.6 years and 24.5 6 6.7 years Eyes that underwent PK <_12 months following onset of (P ¼ .67), respectively. Presence of prior hydrops and acute hydrops did not have a significantly different endo- ocular allergy were associated with the presence of thelial rejection–free survival at 5 years compared with corneal neovascularization at the time of PK in all eyes eyes that underwent PK >12 months following hydrops that underwent PK for keratoconus (Tables 1 and 3). onset: 65.9% 6 7.9% and 55.0% 6 9.6%, respectively After excluding all eyes with neovascularization at the (P ¼ .36). Similarly, eyes that underwent PK <_9 months time of PK (with and without prior hydrops), there was following the resolution of hydrops had endothelial rejec- no difference (P ¼ .78) in the endothelial rejection–free tion–free allograft survival rates at 5 years that were similar allograft survival rates at 5 years for eyes with prior hydrops to eyes that received PK >9 months following the resolu- (n ¼ 41, 10 with rejection events) and those without prior tion of hydrops (P ¼ .37): 66.6% 6 8.1% and 55.0% 6 hydrops (n ¼ 158, 36 with rejection events): 74.6% 6 9.3%, respectively. 7.6% and 78.9% 6 3.3%, respectively. COMPLICATIONS: Loose suture(s) requiring removal ¼ EFFECTS OF CORTICOSTEROID TREATMENT ON EYES developed in 22.9% of eyes (n 56). Suture-related infil- WITH PRIOR HYDROPS: Acute episodes of hydrops were trates (n ¼ 6) or abscesses (n ¼ 7) developed in 5.3% of treated with topical corticosteroid (fluorometholone eyes. Raised intraocular pressure requiring treatment (related 0.1%, prednisolone acetate 1%, or 0.1%) or unrelated to topical corticosteroid) developed in 3.7% of 2–6 times daily for a median duration of 84 days (range eyes (n ¼ 9). Urrets-Zavalia syndrome occurred in 1.2% of 14–621 days). When eyes that received treatment with eyes (n ¼ 3). There were no cases of endophthalmitis.

286 AMERICAN JOURNAL OF OPHTHALMOLOGY SEPTEMBER 2016 DISCUSSION endothelial rejection may not be specifically attributable to the occurrence of hydrops, but rather to its association PENETRATING KERATOPLASTY PERFORMED FOR KERATOCO- with neovascularization. nus has a favorable prognosis compared with other indica- Historically, eyes with hydrops often underwent corneal tions, with 10-year survival rates of 89%–99% transplantation acutely, and in 1 small study outcomes reported.16,22 The occurrence of a rejection episode in were no worse in these eyes compared with eyes without eyes grafted for keratoconus is a risk factor for graft failure, prior hydrops.28 As earlier transplantation may prevent and 90% of rejection episodes have been reported to the formation of peripheral neovascularization, this obser- occur within the first 4 years following transplantation.5 vation may further support the hypothesis that neovascula- We found a lower endothelial rejection–free survival in rization is the main predisposition to increased graft eyes with prior hydrops compared with eyes without prior rejection following hydrops. hydrops. Based on the present study and 2 previous Corticosteroids have been shown to inhibit inflammatory studies,1,20 eyes with prior hydrops are at a higher risk for corneal angiogenesis and subsequent neovascularization.29,30 the development of endothelial rejection. Although eyes Eyes with hydrops in this study were commonly treated with with prior hydrops had a significantly higher incidence of topical corticosteroids of varying compositions/potency, allograft rejection, allograft survival was comparable and frequency of application, and duration of treatment. excellent in keratoconic eyes with or without prior hydrops. Overall, rates of neovascularization appeared to be similar Possible mechanisms for increased endothelial rejection in eyes that received and did not receive topical after prior hydrops include chronic inflammation, associ- corticosteroids. Because topical steroids may have been ated corneal neovascularization, and associated ocular al- used in more severe cases of hydrops and not in milder lergy. As these conditions may be interrelated, clinically cases, it is impossible to draw any conclusions on the basis a combination of these factors may be involved for many of the data we report. However, treatment with patients. corticosteroids did not appear to completely eliminate the Active intraocular inflammation at the time of trans- risk of neovascularization based on the observation that plantation is associated with an increased risk of graft rejec- some eyes treated with prednisolone acetate 1% 4 times tion and failure.23 Whereas corneal hydrops typically daily still developed neovascularization while receiving the resolves clinically within a few months, chronic inflamma- topical corticosteroid. Treatment of hydrops with topical tion could persist much longer. Despite traditionally being corticosteroid on presentation also did not significantly considered a noninflammatory corneal condition, immu- alter the subsequent risk of endothelial graft rejection in nohistochemical studies of keratoconic corneas have those eyes. recently demonstrated extensive inflammatory cells. How- Poor attendance of clinic appointments was found to be ever, these studies did not demonstrate a more pronounced associated with endothelial rejection, possibly because inflammatory response in corneal buttons following these patients were less compliant with topical corticoste- hydrops compared with keratoconic buttons without prior roid usage. In New Zealand, medications can only be hydrops.24 dispensed for a maximum quantity to last 3 months, so pre- Ocular allergy may also predispose to endothelial rejec- scriptions for topical corticosteroid must be obtained every tion,20,25 possibly owing to altered or elevated immune 3 months at the planned clinic appointment or from the pa- responses and/or associated neovascularization. Elevation tient’s general practitioner. As a result, poor compliance of inflammatory mediators has been found in the tear film with clinic attendance may also be considered as a surro- of individuals with allergic .26 We chose to sepa- gate for patient compliance with topical corticosteroid us- rate nonocular and ocular atopy for the purposes of analysis, age. as in the study by Basu and associates,20 who identified a Larger donor trephination size (>8 mm) was found to be higher risk of endothelial rejection in eyes with active a risk factor for rejection. Larger graft size is associated with ocular allergies. A similar association was not identified increased risk of rejection, likely owing to increased prox- in our study. Ocular allergy was, however, associated with imity to the limbal blood vessels.12,23,31 However, host corneal neovascularization among all eyes. trephination size >8 mm was not a significant risk factor Corneal neovascularization has been considered the in the present study. The transplantation of an increased most important host factor predisposing to immune allo- number of endothelial cells, and potentially greater graft rejection.27 A higher rate of neovascularization was antigenic load, in larger donor buttons is another identified in corneas with prior hydrops (44.6%) compared potential mechanism for increased endothelial rejection. with those without prior hydrops (7.6%), and this likely Although hydrops is associated with an increased risk of contributed to the higher rate of endothelial rejection in endothelial rejection, nearly all episodes of graft rejection the hydrops cohort. When all eyes with corneal neovascu- were reversed with aggressive medical treatment. The larization (both with and without prior hydrops) were failure-free survival of eyes with prior hydrops was excellent excluded, hydrops per se was no longer a risk factor at 97.3% 6 1.9% at 10 years, compared with 92.2% 6 2.7% for endothelial rejection. Thus, the increased risk of at 10 years in the nonhydrops cohort. In contrast, the

VOL. 169 PENETRATING KERATOPLASTY WITH AND WITHOUT RESOLVED HYDROPS 287 Australian corneal graft registry found a slightly lower graft geographically isolated country, any episodes of rejection survival rate in eyes with prior hydrops compared with are generally referred back to our center. This may have keratoconus alone, although further detailed evaluation increased our overall rates of rejection, as those patients of these data to explore this association was not lost to follow-up or with poor follow-up and without rejec- performed.32 tion may be less likely to return for evaluation. The strengths of this study include cohort size, strict in- In summary, keratoconus complicated by hydrops and clusion and exclusion criteria, and the quality of documen- subsequent neovascularization may increase the rate of tation; however, limitations include its retrospective endothelial graft rejection following PK. However, with nature. Also, eyes that were excluded owing to being lost timely management of any endothelial graft rejection, to follow-up or were followed up elsewhere owing to long the prognosis for survival in these eyes is excellent, and is distances of travel may have introduced some selection similar to eyes with severe keratoconus without prior bias. Because ours is the major tertiary referral center in a hydrops.

FUNDING/SUPPORT: ALEXANDRA Z. CRAWFORD WAS SUPPORTED IN PART BY AN UNRESTRICTED RESEARCH FELLOWSHIP FROM the Maurice and Phyllis Paykel Trust and Akilesh Gokul was supported in part by a University of Auckland Doctoral Scholarship. Financial disclosures: The following authors have no financial disclosures: Jay J. Meyer, Akilesh Gokul, Alexandra Z. Crawford, and Charles N.J. McGhee. All authors attest that they meet the current ICMJE criteria for authorship. The authors would like to express their sincere thanks to the staff of the New Zealand National Eye Bank for allowing access to their data.

REFERENCES 12. Boisjoly HM, Bernard PM, Dube´ I, Laughrea PA, Bazin R, Bernier J. Effect of factors unrelated to tissue matching on 1. Tuft SJ, Gregory WM, Buckley RJ. Acute corneal hydrops in corneal transplant endothelial rejection. Am J Ophthalmol keratoconus. Ophthalmology 1994;101(10):1738–1744. 1989;107(6):647–654. 2. Amsler MM. Quelques donne´s du problem du ke´ratocoˆne. 13. Jonas JB, Rank RM, Budde WM. Immunologic graft reactions Bull Soc Belge Ophthalmol 1961;128:331–354. after allogenic penetrating keratoplasty. Am J Ophthalmol 3. Grewal S, Laibson PR, Cohen EJ, Rapuano CJ. Acute 2002;133(4):437–443. hydrops in the corneal ectasias: associated factors and out- 14. Zadok D, Schwars S, Marcovich A, et al. Penetrating kerato- comes. Trans Am Ophthalmol Soc 1999;97:187–198. discussion plasty for keratoconus: long-term results. 2005;24(8): 198–203. 959–961. 4. Lockington D, Fan Gaskin JC, McGhee CN, Patel DV. A 15. Olson RJ, Pingree M, Ridges R, Lundergan ML, prospective study of acute corneal hydrops by in vivo confocal Alldredge C Jr, Clinch TE. Penetrating keratoplasty for kera- microscopy in a New Zealand population with keratoconus. toconus: a long-term review of results and complications. J Br J Ophthalmol 2014;98(9):1296–1302. Refract Surg 2000;26(7):987–991. 5. Kelly TL, Williams KA, Coster DJ, et al. Corneal transplan- 16. Fukuoka S, Honda N, Ono K, Mimura T, Usui T, Amano S. tation for keratoconus: a registry study. Arch Ophthalmol 2011; Extended long-term results of penetrating keratoplasty for 129(6):691–697. keratoconus. Cornea 2010;29(5):528–530. 6. Thompson RW Jr, Price MO, Bowers PJ, Price FW Jr. Long- 17. Lim L, Pesudovs K, Coster DJ. Penetrating keratoplasty for term graft survival after penetrating keratoplasty. Ophthal- keratoconus: visual outcome and success. Ophthalmology mology 2003;110(7):1396–1402. 2000;107(6):1125–1131. 7. Inoue K, Amano S, Oshika T, Tsuru T. Risk factors for 18. Pramanik S, Musch DC, Sutphin JE, Farjo AA. Extended corneal graft failure and rejection in penetrating keratoplasty. long-term outcomes of penetrating keratoplasty for keratoco- Acta Ophthalmol Scand 2001;79(3):251–255. nus. Ophthalmology 2006;113(9):1633–1638. 8. Yamagami S, Suzuki Y, Tsuru T. Risk factors for graft failure 19. Sharif KW, Casey TA. Penetrating keratoplasty for keratoco- in penetrating keratoplasty. Acta Ophthalmol Scand 1996; nus: complications and long-term success. Br J Ophthalmol 74(6):584–588. 1991;75(3):142–146. 9. Paglen PG, Fine M, Abbott RL, Webster RG Jr. The prog- 20. Basu S, Reddy JC, Vaddavalli PK, Vemuganti GK, nosis for keratoplasty in keratoconus. Ophthalmology 1982; Sangwan VS. Long-term outcomes of penetrating kerato- 89(6):651–654. plasty for keratoconus with resolved corneal hydrops. Cornea 10. Muraine M, Sanchez C, Watt L, Retout A, Brasseur G. Long- 2012;31(6):615–620. term results of penetrating keratoplasty. A 10-year-plus retro- 21. Akova YA, Dabil H, Kavalcioglu O, Duman S. Clinical fea- spective study. Graefes Arch Clin Exp Ophthalmol 2003; tures and keratoplasty results in keratoconus complicated by 241(7):571–576. acute hydrops. Ocul Immunol Inflamm 2000;8(2):101–109. 11. Maguire MG, Stark WJ, Gottsch JD, et al. Risk factors for 22. Williams KA, Lowe MT, Keane MC, et al. The Australian corneal graft failure and rejection in the collaborative corneal Corneal Graft Registry 2012 Report. Adelaide, Australia: transplantation studies. Collaborative Corneal Transplanta- Snap Printing; 2012. tion Studies Research Group. Ophthalmology 1994;101(9): 23. Williams KA, Roder D, Esterman A, et al. Factors pre- 1536–1547. dictive of corneal graft survival. Report from the

288 AMERICAN JOURNAL OF OPHTHALMOLOGY SEPTEMBER 2016 Australian Corneal Graft Registry. Ophthalmology 1992; 28. Alberth B. Therapy of acute keratoconus. Klin Monatsbl 99(3):403–414. Augenheilkd 1979;174:466–470. 24. FanGaskinJC,LohIP,McGheeCN,SherwinT.An 29. Mirabelli P, Peebo BB, Xeroudaki M, Koulikovska M, immunohistochemical study of inflammatory changes Lagali N. Early effects of dexamethasone and anti-VEGF and matrix remodeling with and without acute hydrops therapy in an inflammatory corneal neovascularization in keratoconus. Invest Ophthalmol Vis Sci 2015;56(10): model. Exp Eye Res 2014;125:118–127. 5831–5837. 30. Nadao S, Zandi S, Lara-Castillo N, Taher M, Ishibashi T, 25. Yildiz EH, Erdurmus M, Hammersmith KM, Rapuano CJ, Hafezi-Moghadam A. Larger therapeutic window for steroid Laibson PR, Cohen EJ. Comparative study of graft rejection versus VEGF-A inhibitor in inflammatory angiogenesis: sur- in keratoconus patients with and without self-reported atopy. prisingly similar impact on leukocyte infiltration. Invest Cornea 2009;28(8):846–850. Ophthalmol Vis Sci 2012;53(7):3296–3302. 26. Stern ME, Siemasko KF, Gao J, Calonge M, Niederkorn JY, 31. Vo¨lker-Dieben HJM. The effect of immunological and non- Pflugfelder SC. Evaluation of ocular surface inflammation in immunological factors on corneal graft survival: a single cen- the presence of dry eye and allergic conjunctival disease. ter study. Doc Ophthalmol 1984;57(1):1–151. Ocul Surf 2005;3(4 Suppl):S161–S164. 32. Williams KA, Lowe MT, Bartlett CM, Kelly L, Coster DJ. 27. Panda A, Vanathi M, Kumar A, Dash Y, Priya S. Corneal The Australian Corneal Graft Registry Report 2007. graft rejection. Surv Ophthalmol 2007;52(4):375–396. Adelaide: Flinders University Press; 2008:1–194.

VOL. 169 PENETRATING KERATOPLASTY WITH AND WITHOUT RESOLVED HYDROPS 289