Microbial Keratitis and Endophthalmitis After the Boston Type 1 Keratoprosthesis

CLINICAL SCIENCE

Michael D. Wagoner, MD, PhD,*† Jeffrey D. Welder, MD,* Kenneth M. Goins, MD,* and Mark A. Greiner, MD*†

Key Words: keratoprosthesis, corneal transplant, microbial keratitis, Purpose: To determine the incidence, ocular surface disease endophthalmitis associations, microbiological proﬁle, and clinical course of postoperative infections after implantation of the Boston type 1 (Cornea 2016;35:486–493) keratoprosthesis (KPro-1).

Methods: A retrospective chart review was conducted of all cases he Boston type 1 keratoprosthesis (KPro-1) offers an of the KPro-1 performed at a tertiary eye care center between Talternative to traditional corneal transplantation in eyes January 1, 2008, and December 31, 2014. with a poor prognosis for primary or repeat keratoplasty.1–4 Results: Seventy-five KPro-1 procedures were included in the The procedure has a distinct advantage of bypassing analysis. Postoperative infections occurred in 13 eyes (17.3%) immunologic-mediated endothelial rejection, a complication with an incidence of 0.064 cases per eye-year. The Kaplan–Meier that may compromise graft clarity after penetrating kerato- 5 probability of an infection-free graft was 0.96, 0.90, 0.88, 0.87, plasty (PKP). The KPro-1 optic does not depend on fi and 0.86 at years 1 through 5, respectively. The rate of infections a healthy corneal epithelium or precorneal tear lm to was greater in eyes with ocular surface disease (26.3% vs. 8.1%, provide satisfactory visual acuity, a feature that is particu- P=0.06), especially in those with a history of chemical or larly beneficial in eyes with chronic ocular surface disor- 6,7 thermal injury or herpes zoster keratopathy (P=0.001). There ders. However, these visual benefits can be compromised were 8 cases (10.7%) of microbial keratitis due to either fungi (5 by ocular surface–related issues that contribute to the 8–10 cases) or bacteria (3 cases). There were 7 cases (9.3%) of development of microbial keratitis and/or endophthalmi- 11–15 endophthalmitis due to bacteria (5 cases) or intraocular extension tis and the resultant potential for device extrusion and 16 of fungal keratitis (2 cases). The incidence of microbial keratitis irreversible ocular injury. and endophthalmitis was 0.034 and 0.03 cases per eye-year, Multiple management strategies have been introduced respectively. Therapeutic management of the infected eyes to reduce vision-threatening postoperative infections after – required graft and device removal in 7 eyes (53.8%). After keratoprosthesis implantation.17 21 Soft contact lenses have completion of microbiologic treatment, 7 eyes (53.8%) had lost been used to reduce the risk of epithelial-related microbial – more than 2 lines of the best-corrected preinfection visual acuity, infections.17 20 Systemic doxycycline and topical progesta- including 5 eyes with endophthalmitis that had hand motion tional corticosteroids have been used to reduce inflammation vision or worse. nonmicrobial ulceration and help prevent secondary infections due to graft–host and graft–device dehiscence.17–20 Conclusions: Postoperative infections are a serious issue that Prophylactic antibiotics, antiseptics, and antifungal agents compromises device retention and visual outcomes after have been used to reduce microbial colonization of the ocular keratoprosthesis implantation. surface and the associated risk of microbial keratitis and/or endophthalmitis.21–25 There is no evidence-based consensus on the com- Received for publication October 9, 2015; revision received November 4, position or duration of an optimal postkeratoprosthesis 2015; accepted November 11, 2015. Published online ahead of print prophylactic regimen.5,17–25 Theprolongeduseofsoft January 14, 2016. contact lenses may be associated with an increased risk of From the *Department of Ophthalmology and Visual Sciences, University of 26,27 Iowa Hospitals and Clinics, University of Iowa Carver College of gram-negative keratitis, especially in this population Medicine, Iowa City, IA; and †Cornea Research Center, Stephen A. Wynn with a high burden of ocular surface disease (OSD). Institute for Vision Research, Department of Ophthalmology and Visual Extended use topical vancomycin, fluoroquinolones, and Sciences, University of Iowa Carver College of Medicine, Iowa City, IA. steroids may be associated with an overall increased risk Supported in part by the Beulah and Florence Chair in Cornea/External Disease and Refractive Surgery, the UIHC Cornea Research Fund, and of postoperative infections, especially those caused by the M. D. Wagoner and M. A. Greiner Cornea Excellence Fund. drug-resistant organisms and fungi.9,10,15,28,29 In this The authors have no conflicts of interest to disclose. study, we have endeavored to determine the incidence, Reprints: Michael D. Wagoner, MD, PhD, Cornea Research Center, Stephen OSD association, microbiological profile, and clinical A. Wynn Institute for Vision Research, Room 4184 MERF, 375 Newton Rd, Iowa City, IA 52242-1091 (e-mail: [email protected]). course of postkeratoprosthesis microbial keratitis and/ Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved. or endophthalmitis.

486 | www.corneajrnl.com Cornea Volume 35, Number 4, April 2016

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Cornea Volume 35, Number 4, April 2016 Microbial Keratitis and Endophthalmitis

PATIENTS AND METHODS were provided for all new surgical cases, as well as for This study was approved by the Institutional Review existing postoperative cases. Board of the University of Iowa Hospitals and Clinics In eyes with a corneal infiltrate, a complete microbio- (UIHC) and conformed to the requirements of the United logical evaluation was performed as per previously described 5 States Health Insurance Portability and Accountability Act methods. Intensive topical antibiotic therapy was initiated with and the tenets of the Declaration of Helsinki. A retrospective fortified vancomycin (25 mg/mL) and, depending on the chart review was performed of all cases of KPro-1 implan- preference of the treating ophthalmologist, with fortified tation at UIHC between January 1, 2008, and December 31, ceftazidime (50 mg/mL), fortified tobramycin (16 mg/mL), or fl 2014. Procedures that were done in eyes in which a previous a fourth-generation uoroquinolone. The antibiotic regimen fi graft failed or in which the KPro-1 was the primary corneal was adjusted, as necessary, on the basis of the identi cation of fi procedure were included in the statistical analysis. Eyes with the bacterial isolate, the susceptibility pro le, and the clinical fi previous keratoprosthesis procedures were excluded from the course. If a fungal isolate was identi ed, intensive topical – analysis. Follow-up data were obtained, when available, therapy was initiated with topical amphotericin B (1.5 5.0 mg/ through June 30, 2015. mL) and voriconazole (10 mg/mL), as well as oral voricona- The main outcome measure was postoperative infec- zole (200 mg). The antifungal regimen was adjusted, as fi tions with microbial keratitis and/or endophthalmitis. These necessary, on the basis of the susceptibility pro le and clinical events were defined by characteristic clinical findings of an course. When necessary, the device was removed and replaced infectious process combined with microbiological confirma- with another KPro-1, if possible. tion with a positive corneal or vitreous isolate, respectively. In eyes with suspected microbial endophthalmitis, Cases of clinically suspected microbial keratitis or endoph- a diagnostic vitreous tap was performed, followed by an thalmitis that had a negative microbiological evaluation were intravitreal injection with vancomycin (1 mg/0.1 mL) and, not included in the analysis. Secondary outcome measures depending on the preference of the treating ophthalmologist, either ceftazidime (2.25 mg/0.1 mL) or amikacin (0.4 mg/0.1 included the microbiological profile, association of infections mL). In cases with suspected intraocular extension of with OSD, changes in the incidence of infections during the a confirmed case of fungal keratitis, an intravitreal injection course of the study, and the clinical course of infected eyes, with amphotericin (5 mg/0.1 mL) and voriconazole (100 mg/ including the need for device removal and replacement and 0.1 mL) was performed at the time of diagnostic tap and final visual outcome. topical antifungal therapy was continued as per the clinical The KPro-1 procedures were performed by experienced course. When necessary, the device was removed and UIHC corneal surgeons using previously described techni- 30,31 replaced with a tectonic PKP. ques. A polymethylmethacrylate back-plate was used in All data were entered using Microsoft Office Excel all cases performed before February 2012. Thereafter, a tita- (Microsoft Corporation, Redmond, WA). Data that were nium back-plate was used. A glaucoma drainage device collected from the patient medical record included patient (GDD) was implanted by an experienced UIHC glaucoma age, surgical diagnosis, antimicrobial prophylaxis regimen, surgeon in every case without a GDD. In cases with a previous follow-up intervals (surgery to most recent examination or to GDD that was not functioning adequately, a new GDD was development of microbial keratitis and/or endophthalmitis, implanted or a revision of the existing device was done. whichever came first), postoperative infections and the Postoperative examinations were performed at 1 day, 1 corresponding microbiological evaluation (isolates, ocular week, 2 weeks, 1 month, and quarterly thereafter. Every eye location, susceptibility profile), medical and surgical therapy, fi was treated inde nitely with a 16.0-mm-diameter bandage graft and device removal and replacement, and best-corrected soft contact lens (Kontur Kontact Lens Co, Hercules, CA) visual acuity (BCVA), as documented at the visit immediately that was replaced every 1 to 3 months. Daily doxycycline before infection, after completion of treatment of the 100 mg was initiated and continued indefinitely. Topically, infection, and at the most recent visit. prednisolone acetate 1% and medroxyprogesterone 1% were Surgical diagnosis was defined as the original indication used 4 times a day for 3 months, 3 times a day for 3 months, for the first corneal transplantation (in eyes with failed grafts) or 2 times a day for 3 months, and then once a day thereafter. the primary KPro-1 procedure (in eyes without previous From January 1, 2008, until February 2011, every eye keratoplasty). The original surgical indication was subclassified received a standard antibacterial prophylactic regimen that into indications that are usually associated with or not associated included topical vancomycin (16 or 25 mg/mL) and a fourth- with OSD (OSD+ and OSD2, respectively). Indications that generation fluoroquinolone antibiotic (either gatifloxacin were classified as OSD+ included aniridia, chemical or thermal 0.3% or moxifloxacin 0.5%). These were administered 4 injury, herpes simplex or zoster viral keratitis, microbial times a day for 3 months, 3 times a day for 3 months, 2 times keratitis, and cicatricial conjunctival disorders. Indications that a day for 3 months, and then once a day thereafter. were classified as OSD2 included endothelial dystrophy or Beginning in February 2011, a fornix rinse with povi- degeneration, pediatric glaucoma, trauma, iridocorneal endothe- done–iodine 5% was performed every 3 months at the time lial syndrome, and keratoconus. Comparisons between both of contact lens cleaning or replacement. Beginning in groups were made using the 2-tailed Fisher exact test. P # 0.05 February 2012, a 7-day course of topical amphotericin B was considered statistically significant. (1.5 mg/mL) 4 times daily was prescribed after completion Three clinical eras were defined by the contemporary of the fornix rinse. The “upgraded” prophylactic regimens prophylactic regimens. These were era A (vancomycin +

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Wagoner et al Cornea Volume 35, Number 4, April 2016

fluoroquinolone) from January 1, 2008, to February 1, 2011; was an OSD2 disorder (Table 2). From era A to era C, era B (vancomycin + fluoroquinolone + povidone–iodine) there was a progressive decrease in the OSD+/OSD2 for from February 1, 2011, to February 1, 2012; and era C both new surgical cases and prophylactically upgraded (vancomycin + fluoroquinolone + povidone–iodine + am- extant cases. photericin B) from February 1, 2012, to June 30, 2015. The overall rate of postoperative infections was higher Infection data that occurred during each clinical era were in OSD+ eyes than in OSD2 eyes (26.3% vs. 8.1%; P = tabulated using Kaplan–Meier (K–M) probabilities of 0.06). Eyes with a history of herpes zoster keratopathy or infection-free survival, incidence (cases per eye-year), and chemical or thermal injury were significantly more likely to rates (percentage of infected eyes). This included infections develop postoperative infections than did other eyes with that occurred in surgical cases performed within the era, as other surgical indications (P = 0.001). OSD+ eyes had well as previously performed cases for which the post- a higher rate of microbial keratitis (18.4% vs. 2.7%; P= operative regimen had been upgraded to the newer protocol. 0.06) and endophthalmitis (13.2% vs. 5.4%; P = 0.43) than Comparisons of differences in K–M infection-free probabil- did OSD2 eyes. ities, and the incidence and rate of infections, between the groups were made with a log-rank test, 2-tailed Z-test, and 2-tailed Fisher exact test, respectively. P # 0.05 was Microbiological Profile considered statistically significant. There were a total of 8 bacterial infections. This included infections with 6 gram-positive organisms, of which coagulase-negative Staphylococcus was the most common RESULTS isolate. There were a total of 5 fungal infections. Candida Between January 1, 2008, and December 31, 2014, species accounted for all but 1 isolate. Candida parapsilosis a total of 79 initial KPro-1 procedures were performed. was the most common isolate. All gram-positive bacteria and Seventy-five procedures that were performed in 52 eyes of 51 fungal isolates were sensitive to vancomycin and amphoter- men and 23 eyes of 23 women were included in the analysis icin B, respectively. (Table 1). Four cases with previous keratoprosthesis procedures were excluded from the analysis. There were 56 surgical cases performed during era A, Clinical Course 12 cases during era B, and 7 cases during era C. The mean A comprehensive summary of the clinical course of the patient age at the time of surgery was 58 years (range 0.5–91 13 infected eyes is provided in Table 3. Therapeutic years). Infection data were available for 56 eyes with management required graft and device removal in 7 eyes postoperative follow-up in era A, 57 eyes in era B, and 55 (53.8%) and replacement with either another KPro-1 (2 eyes) eyes in era C. or tectonic PKP (5 eyes). After successful treatment, 7 eyes (53.8%) lost more than 2 lines of the preinfection BCVA, including 5 eyes with endophthalmitis that had vision of hand Incidence motions or worse. Postoperative infections occurred in 13 eyes (17.3%), with an incidence of 0.064 cases per eye-year. The K–M probability of an infection-free graft was 0.96, 0.90, 0.88, Microbial Keratitis Alone 0.87, and 0.86 at years 1 through 5, respectively. The mean Bacterial and fungal isolates each accounted for 3 patient age at the time of infection was 54.1 years (range, 4– cases of localized microbial keratitis. All 6 cases presented 78). The median interval from surgery to infection was 11 at a scheduled quarterly visit in an asymptomatic patient months (range, 1–60 months). with a corneal epithelial defect and associated There were 8 cases (10.7%) of microbial keratitis due stromal infiltrate. to either fungi (5 cases) or bacteria (3 cases). There were 7 The 3 cases of bacterial keratitis had infiltrates at the cases (9.3%) of endophthalmitis due to bacteria (5 cases) or graft–device junction (case 5), the graft–host junction (case intraocular extension of fungal keratitis (2 cases). The 1), or in the midgraft (case 3). One case of a-hemolytic incidence of microbial keratitis and endophthalmitis was streptococcal keratitis (case 5) occurred at the graft–device 0.034 and 0.03 cases per eye-year, respectively. Four cases junction in an eye with a chronically (.6 months) exposed of culture-negative cornealulcerationand3casesof GDD tube that was due for a quarterly povidone–iodine culture-negative vitritis were not included in the analysis. rinse and amphotericin B pulse. All 3 eyes were managed Between era A and era C, there was a statistically with medical therapy. After successful treatment, 2 eyes significant increase in K–M infection-free survival probability (cases 1 and 3) recovered the preinfection BCVA, whereas (P = 0.003), an 82% reduced incidence (P = 0.04), and a 48% the other eye (case 5) lost more than 2 lines of reduced rate (P=0.36) of infection, respectively. Snellen acuity. In all 3 cases of fungal keratitis, the infiltrate was at the graft–device junction. The eyes that were on quarterly OSD Association povidone–iodine rinses alone (case 4) or with the additional There were 38 eyes in which the original surgical amphotericin B pulse (case 6) were due or past due for their indication was an OSD+ disorderand37eyesinwhichit quarterly interventions. Only 1 eye (case 4) required

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Cornea Volume 35, Number 4, April 2016 Microbial Keratitis and Endophthalmitis

TABLE 1. Microbial Keratitis and Endophthalmitis After KPro-1 Implantation Prophylactic Regimen Cumulative V+F V+F+PI V+F+PI+A KPro-1 cases All cases 75 56 57 55 OSD+/OSD2 1.03 1.15 1.03 0.90 New surgical cases, n 56 12 7 OSD+/OSD2 1.15 0.71 0.75 Cases with upgraded regimen, n 45 48 OSD+/OSD2 1.14 0.92 Interval Dates Surgery January 2008–December January 2008–January January 2008–January 2012 January 2008–December 2014 2011 2014 Follow-up January 2008–June 2015 January 2008–January February 2011–January February 2012–June 2015 2011 2012 Follow-up Mean (mo) 37.7 11.5 10.1 29.2 Range (mo) 0.9–72.8 0.8–30.5 0.2–12.0 4.1–39.6 Cumulative (y) 235.5 53.7 47.9 133.9 K–M survival* All infection cases 6 mo 0.98 (0.92–1.0) 0.94 (0.83–0.98) 0.96 (0.86–0.99) 1.0 (0.91–1.0) 12 mo 0.96 (0.89–0.98) 0.92 (0.80–0.97) 0.92 (0.80–0.97) 0.98 (0.88–1.0) 18 mo 0.90 (0.82–0.95) 0.79 (0.65–0.89) 0.98 (0.88–1.0) 24 mo 0.90 (0.82–0.95) 0.79 (0.65–0.89) 0.98 (0.88–0.99) 36 mo 0.88 (0.79–0.93) 0.96 (0.85–0.99) 48 mo 0.87 (0.79–0.93) 60 mo 0.86 (0.78–0.92) Incidence (cases per eye-y) All infection cases Total 0.06 0.13 0.10 0.02 Bacterial 0.03 0.07 0.04 0.02 Fungal 0.03 0.06 0.06 0.01 Microbial keratitis Total 0.03 0.07 0.04 0.02 Bacterial 0.01 0.04 0 0.01 Fungal 0.02 0.04 0.04 0.01 Endophthalmitis Total 0.03 0.06 0.06 0.01 Bacterial 0.02 0.04 0.04 0.01 Fungal 0.01 0.02 0.02 0 Infections (eyes/regimen) All infected eyes Total, n (%)† 13 (17.3) 6 (10.7) 4 (7.0) 3 (5.5) Bacterial, n 8 4 2 2 Fungal, n 5 2 2 1 Microbial keratitis Total, n (%) 8 (10.7) 4 (7.1) 2 (3.5) 2 (3.6) Bacterial, n 3 2 0 1 Fungal, n 5 2 2 1 Endophthalmitis Total, n (%) 7 (9.3) 3 (5.4) 3 (5.3) 1 (1.8) Bacterial, n 5 2 2 1 Fungal, n 2 1 1 0

A, amphotericin B; F, ﬂuoroquinolone; PI, povidone–iodine; V, vancomycin. *% Infection-free survival probability (95% conﬁdence interval). †Two eyes had both fungal keratitis and endophthalmitis.

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Wagoner et al Cornea Volume 35, Number 4, April 2016

TABLE 2. Surgical Indication Versus Microbial Keratitis and Endophthalmitis After KPro-1 Implantation Microbial Keratitis + Surgical Cases, Any, Eyes, Microbial Keratitis, Endophthalmitis, Endophthalmitis, Surgical Indication Eyes, n n (%) Eyes, n (%) Eyes, n (%) Eyes, n (%) OSD+ Aniridia 10 0 0 0 0 Chemical/thermal injury 9 4 (44.4) 2 (22.2) 0 2 (22.2) Herpes simplex virus 5 1 (20.0) 1 (20.0) 0 0 Herpes zoster virus 4 3 (75.0) 0 2 (66.7) 1 (33.3) Microbial keratitis 4 1 (25.0) 1 (25.0) 0 0 Cicatricial conjunctiva 4 1 (25.0) 1 (25.0) 0 0 Other 2 0 0 0 0 All OSD+ 38 10 (26.3) 5 (13.1) 2 (5.3) 3 (7.9) OSD2 Endothelial dystrophy 13 0 0 0 0 Pediatric glaucoma 9 1 (11.1) 0 0 1 (11.1) Trauma 8 1 (12.5) 0 0 1 (11.1) Iridocorneal endothelial syndrome 4 0 0 0 0 Keratoconus 3 1 (33.3) 1 (33.3) 0 0 All OSD2 37 3 (8.1) 1 (2.7) 0 2 (5.4) Total 75 13 (17.3) 6 (8.0) 2 (2.7) 5 (6.7) medical intervention, whereas the other 2 eyes (cases 2 and 2) recovered the preinfection BCVA, 1 eye (case 4) lost 2 6) were managed with the additional removal and replace- lines of Snellen acuity, and 1 eye (case 6) lost more than 2 ment of the KPro-1. After successful treatment, 1 eye (case lines of Snellen acuity.

TABLE 3. Microbial Keratitis and Endophthalmitis After KPro-1 Implantation: Case Summaries Prophylaxis Surgery to Site Susceptibility BCVA Age Infection Case Diagnosis (y) V F PI A (mo) CV Organism V F A Preinfection Posttreatment Final 1 Chemical injury 75 + + 2 + Corynebacterium + 6/6 6/6* 6/6 species 2 Keratoconus 43 + + 5 + Candida parapsilosis + 6/9 6/9† 6/30 3 Herpes simplex 78 + + 6 + Staphylococcus + + 6/9 6/9* 6/12 (coagulase2) 4 Chemical injury 47 + + + 14 + C. parapsilosis + 6/6 6/9* 6/9 5‡ Microbial keratitis 70 + + + + 29 + Streptococcus + 6/12 6/30* 6/60 (a-hemolytic) 6 Cicatricial 58 + + + + 60 + Scedosporium + 6/12 6/30† 6/60 conjunctiva apiospermum 7 Herpes zoster 66 + + + 5 + + Candida albicans + 6/12 HM§ LP 8 Herpes zoster 65 + + 15 + + C. parapsilosis + 6/12 HM§ LP 9 Herpes zoster 51 + + 1 + Streptococcus + 6/12 NLP§ NLP (a-hemolytic) 10 Chemical injury 56 + + + 7 + Staphylococcus + + 6/9 6/9* 6/9 (coagulase2) 11 Thermal injury 44 + + + 11 + Moraxella osloensis + 6/12 LP§ NLP 12k Trauma 46 + + 12 + Staphylococcus + + 6/60 6/60* LP¶ (coagulase2) 13‡ Pediatric 4++++ 30 +Pseudomonas CF LP§ NLP glaucoma ﬂuorescens

A, amphotericin B; C, cornea; CF, counting ﬁngers; F, ﬂuoroquinolone; HM, hand motion; LP, light perception; NLP, no light perception; PI, povidone–iodine; V, vancomycin; V, vitreous. *Medical therapy only. †Medical therapy + KPro-1 replacement. ‡GDD erosion before endophthalmitis. §Medical therapy + tectonic PKP. kPatient admitted to noncompliance with all topical medications before infection. ¶Due to late-onset retinal detachment.

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Cornea Volume 35, Number 4, April 2016 Microbial Keratitis and Endophthalmitis

Microbial Keratitis + Endophthalmitis comprehensive polymicrobial prophylaxis in the latter – Fungal isolates were associated with 2 cases of era.32 34 The majority of patients were compliant with the microbial keratitis with intraocular extension and secondary outpatient regimen, as confirmed by specific questions that endophthalmitis. Both cases presented in asymptomatic were asked at the time of each outpatient visit. Every patient patients at a scheduled quarterly visit. The eye (case 7) on with a postoperative infection had been compliant with the quarterly povidone–iodine prophylaxis was 2 months follow-up schedule, although several were at, or just beyond, removed from the most recent rinse. On initial examination, the 3-month follow-up interval when presenting with asymp- both eyes had corneal epithelial defects at the graft–device tomatic infections. Only 1 patient admitted to noncompliance junction, stromal infiltrates that extended to the stem and into with the outpatient regimen. the anterior chamber, and severe intraocular inflammation. Microbial keratitis or endophthalmitis each occurred in Both eyes were managed with medical therapy and removal approximately 10% of the surgical eyes. In the latter era of the of the KPro-1 and replacement with tectonic PKPs, but study, there was a statistically significant improvement in the suffered severe vision loss to hand motions due to rapid K–M infection-free survival probability and a significant graft opacification. reduction in the incidence of infections. The small number of latter-era infections that did occur may have been due to the interval between the quarterly polymicrobial treatments, as Endophthalmitis Alone well as mitigating anatomic circumstances, rather than a lack Bacterial isolates accounted for all 5 cases of non– of therapeutic efficacy. Among 7 eyes that became infected keratitis-related endophthalmitis. Two patients (cases 9 and after the implementation of the povidone–iodine rinses, 6 10) presented with pain and decreased vision of less than 2 presented with asymptomatic infections when the eye was days. Three patients (cases 11–13) were asymptomatic at the either due or overdue for the quarterly rinse. Two of these time of scheduled quarterly visit. eyes that were overdue for a quarterly rinse also had Three eyes on povidone–iodine prophylaxis were 2 chronically exposed GDD tubes. Both these eyes were months (case 10) or more than 3 months (cases 11 and 13) infected with virulent bacterial pathogens that are atypical beyond the most recent rinse. One patient (case 12) for postkeratoprosthesis isolates. It should be emphasized that admitted to noncompliance with his topical vancomycin the statistical comparisons in this study are compromised by and fluoroquinolone prophylactic regimen. One eye (case potential favorable skewing of latter-era infection data by 13), with a chronically (.6 months) exposed GDD tube, factors such as a lower burden of OSD and a higher was infected with vancomycin- and fluoroquinolone- percentage of eyes that were beyond the first 2 postoperative resistant Pseudomonas fluorescens. years, when infection risk is greatest.9 Furthermore, no Four eyes presented with infiltrates on the back-plate causality can be implied with respect to the contribution of (case 9) or stem (cases 11–13) of the device in association the more comprehensive prophylactic regimen to better with corneal epithelial defects at the graft–device junction. infection outcomes due to other contributing variables, such There were no stromal infiltrates. Two eyes (case 9 and 13) as improved surgical experience and postoperative decision were Seidel-positive. One eye (case 10), with an exposed making, better patient selection, and improvements in the corneal suture, had severe vitreous inflammation without an keratoprosthesis design. associated stromal infiltrate. The final common pathway for a majority of the Two eyes (cases 10 and 12) with coagulase-negative infections was the development of an epithelial defect. The staphylococci that were successfully treated with medical most common location of epithelial breakdown was at the therapy regained the preinfection BCVA. Three eyes (cases 9, graft–device junction, a factor that predisposed to intraocular 11, and 12) with more virulent organisms (a-hemolytic microbial contamination in all but 1 case of endophthalmitis. Streptococcus, Moraxella osloensis, and P. fluorescens, Among indications associated with OSD, infections occurred respectively) were successfully treated with additional in over one quarter of the surgical eyes, compared with less removal of the KPro-1 and replacement with tectonic PKPs, than one tenth of those without this risk factor. This was but suffered severe visual loss to light perception or worse especially true for eyes with a history of chemical or thermal due to rapid graft opacification. injuries or herpes zoster keratopathy, in which more than one half of the surgical eyes developed infections. When present, bacterial infections were almost invariably due to gram- DISCUSSION positive organisms, especially coagulase-positive Staphylo- Our study is confirmatory of previous studies with coccus. Fungal infections were predominantly due to respect to the incidence, ocular surface associations, micro- Candida species, especially C. parapsilosis. All gram- biological profile, and clinical course of postoperative positive organisms and fungi were sensitive to vancomycin infections after the KPro-1.9–15,17–20 The series is sufficiently and amphotericin B, respectively. large to document and compare the incidence of infections The prognostic importance of postoperative infections that occurred during 3 consecutive clinical eras, as defined by is manifest by the adverse tectonic and visual consequences the contemporaneous antimicrobial regimen. Key features of associated with these events. The virulence and location of the study were the provision of the same, continuous the infecting organisms, as well as the ability to eradicate the prophylactic regimen for every patient, with progression from infection with medical therapy alone versus the need to resort antibacterial prophylaxis alone in the first era to more to surgical intervention, influenced the recovery of

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Wagoner et al Cornea Volume 35, Number 4, April 2016 preinfection vision. The prognosis for visual recovery was respect to case selection, operative technique, and provision excellent among the cases of bacterial keratitis or endoph- of continuous polymicrobial prophylaxis. thalmitis that were successfully treated with medical therapy alone, especially if the bacterial isolate was coagulase- REFERENCES negative Staphylococcus. However, successful microbiologic 1. Aldave AJ, Kamal KM, Vo RC, et al. The Boston type I keratoprosthesis: and tectonic management of the globe required removal of improving outcomes and expanding indications. Ophthalmology. 2009; – more than half of the grafts and replacement with either 116:640 651. 2. Colby KA, Koo EB. Expanding indications for the Boston keratopros- another KPro-1 or PKP. The indications for surgical inter- thesis. Curr Opin Ophthalmol. 2011;22:267–273. vention were concerns about intraocular extension of fungal 3. Zerbe BL, Belin MW, Ciolino JB, et al. Results from the multicenter keratitis or evidence of microbial contamination of the device. Boston type 1 keratoprosthesis study. Ophthalmology. 2007;125:745–749. The visual prognosis was reasonably good in the cases of 4. Aldave AJ, Sangwan VS, Basu S, et al. International results with the Boston type 1 keratoprosthesis. Ophthalmology. 2012;119:1530–1538. fungal keratitis that were successfully treated with KPro-1 5. Wagoner MD, Ba-Abbad R, Sutphin JE, et al. Corneal transplant survival removal and replacement before intraocular extension. How- after onset of severe endothelial rejection. Ophthalmology. 2007;114: ever, the visual prognosis was dismal for eyes with either 1630–1636. secondary fungal or primary bacterial endophthalmitis that 6. Rixen JJ, Cohen AW, Kitzmann AS, et al. Treatment of aniridia with – required KPro-1 replacement with a PKP. In these cases, the Boston type 1 keratoprosthesis. Cornea. 2013;32:947 950. 7. Phillips DL, Hager JL, Goins KM, et al. Boston type 1 keratoprosthesis primary reason for failure to recover premorbid vision was for chemical and thermal injury. Cornea. 2014;33:905–999. rapid loss of clarity of the tectonic graft. 8. Chan CC, Holland EJ. Infectious keratitis after Boston type 1 keratopros- Based on our experience and consistent with a commit- thesis implantation. Cornea. 2012;31:1128–1134. ment to the continued use of a comprehensive prophylactic 9. Kim MJ, Yu F, Aldave AJ. Microbial keratitis after Boston type I fi keratoprosthesis: incidence, organisms, risk factors, and outcomes. program, we recommend several additional modi cations of Ophthalmology. 2013;120:2209–2216. our KPro-1 clinical practice. For all cases, gram-positive 10. Jain V, Mhatre K, Shome D, et al. Fungal keratitis with the type 1 Boston prophylaxis should be maintained at a minimum dosing keratoprosthesis: early Indian experience. Cornea. 2012;31:841843. frequency of twice daily. For OSD+ eyes, a higher dosing 11. Nouri M, Terada H, Alfonso EC, et al. Endophthalmitis after keratopros- schedule can be considered on a case-by-case basis. The thesis: incidence, bacterial causes, and risk factors. Arch Ophthalmol. – 2001;119:484–489. povidone iodine rinse and weekly amphotericin pulse should 12. Fintelmann RE, Maguire JI, Ho AC, et al. Characteristics of endoph- be administered as often as every 6 to 8 weeks, especially in thalmitis in patients with the Boston keratoprosthesis. Cornea. 2009;28: high-risk cases. Efforts should be made to discontinue topical 877–878. corticosteroids after 6 to 12 months, if adequate control of 13. Ramchandran RS, Diloreto DA Jr, Chung MM, et al. Infectious fl endophthalmitis in adult eyes receiving Boston type 1 keratoprosthesis. ocular in ammation can be achieved with progestational Ophthalmology. 2012;119:674–681. steroids alone. Caution is recommended in performing 14. Chan CC, Holland EJ. Infectious endophthalmitis after Boston type 1 surgery in eyes with OSD, especially in those with a history keratoprosthesis implantation. Cornea. 2012;31:346–349. of chemical or thermal injury or zoster keratopathy. If case- 15. Greiner MA, Li JHY, Mannis MJ. Longer-term vision outcomes and specific circumstances justify proceeding with a keratopros- complications with the Boston type I keratoprosthesis at the University of California, Davis. Ophthalmology. 2011;118:1543–1550. thesis, preoperative and postoperative surface optimization is 16. Ciolino JB, Belin MW, Todani A, et al. Retention of the Boston keratopros- mandatory. Eyes with chemical or thermal injuries should be thesis type 1: multicenter study results. Ophthalmology. 2013;120:1195–1200. treated preoperatively with appropriate ocular surface reha- 17. Khan BF, Harissa-Dagher M, Khan DM, et al. Advances in Boston 35 keratoprosthesis: enhancing retention and prevention of infection and bilitation and eyelid procedures. Based on favorable fl – personal experience with the management of neurotrophic in ammation. Int Ophthalmol Clin. 2007;113:61 71. 36 18. Traish AS, Chodosh J. Expanding application of the Boston type 1 keratopathy with autologous serum eye drops, their postkeratoprosthesis due to advances in design and improved post-operative operative use should be considered in eyes with zoster therapeutic strategies. Semin Ophthalmol. 2010;25:239–243. keratopathy and other neurotrophic disorders. Exposure of 19. Dohlman CH, Dudenhoefer EJ, Khan BF, et al. Protection of the ocular surface after keratoprosthesis surgery: the role of soft contact lenses. GDD tubes should be treated as a high-risk factor that – 37 CLAO J. 2002;28:72 74. mandates surgical repair, whenever possible. 20. Kammerdiencer LL, Speiser JL, Aquavella JV, et al. Protective effect In conclusion, postoperative infections remain a seri- of soft contact lenses after Boston keratoprosthesis. Br J Ophthalmol. 2015 ous issue that compromises device retention and visual Aug 14. pii: bjophthalmol-2014-306396. doi: 10.1136/bjophthalmol-2014- outcomes after keratoprosthesis implantation. The presence 306396. [Epub head of print]. 21. Durand ML, Dohlman CH. Successful prevention of bacterial endophthal- of multiple contributory variables and the relatively small mitis in eyes with the Boston keratoprosthesis. Cornea. 2009;28:896–901. size of our study preclude the ability to provide definitive 22. Brown CR, Wagoner MD, Welder JD, et al. Boston type 1 keratopros- statistical proof as to whether or not our comprehensive thesis for herpes simplex and herpes zoster keratopathy. Cornea. 2014; prophylactic regimen, or any of its individual components, 33:801–805. were solely or partially responsible or noncontributory to 23. Hager JL, Phillips DL, Goins KM, et al. Boston type 1 keratoprosthesis for failed keratoplasty. Int Ophthalmol. 2015 May 16 [Epub ahead of print]. improved infection outcomes. Nonetheless, we are reluctant 24. Haugsdal JM, Goins KM, Greiner MA, et al. Boston type 1 keratopros- to abandon our current practice to pursue a randomized thesis for primary open angle glaucoma. Br J Ophthalmol. 2015 July 23. controlled trial because it is impractical to conduct a suffi- pii: bjophthalmol-2015-307190. doi: 10.1136/bjophthalmol-2015-307190. ciently large study to obtain a definitive answer. In addition, [Epub ahead of print]. 25. Phillips DL, Goins KM, Greiner MA, et al. Boston type 1 keratoprosthesis we strongly believe that withholding potential therapeutic for iridocorneal endothelial syndromes. Cornea. 2015;34:1383–1386. benefits in control patients raises ethical concerns. Instead, 26. Schein OD, Glynn RJ, Poggio EC, et al. The relative risk of ulcerative we plan to continue our current KPro-1 clinical practice with keratitis among users of daily-wear and extended-wear soft contact

Copyright Ó 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Cornea Volume 35, Number 4, April 2016 Microbial Keratitis and Endophthalmitis

lenses. A case-control study. Microbial Keratitis Study Group. N Eng J 32. Pelletier JS, Miller D, Liang B, et al. In vitro efﬁcacy of a povidone- Med. 1989;321:773–778. iodine 0.4% and dexamethasone 0.1% suspension against ocular 27. Schein OD, Beuhler P, Stamler JF, et al. The impact of overnight wear on pathogens. J Cataract Refract Surg. 2011;37:763–766. the risk of contact lens-associated ulcerative keratitis. Arch Ophthalmol. 33. Magalhaes FP, do Nascimento HM, Ecker DJ, et al. Microbiota evaluation 1994;112:186–190. of patients with a Boston type 1 keratoprosthesis treated with topical 0.5% 28. Barnes SD, Dohlman CH, Durand ML. Fungal colonization and infection moxiﬂoxacin and 5% povidone-iodine. Cornea. 2013;32:407–411. in Boston keratoprosthesis. Cornea. 2007;26:9–15. 34. Amphotericin B, EP. Susceptibility and Resistance Data. Bellingham, 29. Robert MC, Eid EP, Saint-Antoine P, et al. Microbial colonization and WA: Toku-E; 2014. Available at: https://www.tokue.com/Assets/MIC/ antibacterial resistance patterns after Boston type 1 keratoprosthesis. Amphotericin%20B%20EP.pdf. Accessed April 18, 2014. Ophthalmology. 2013;120:1521–1528. 35. Wagoner MD. Chemical injuries of the eye: current concepts in 30. Aquavella JV, Qian Y, McCormick GJ, et al. Keratoprosthesis: the pathophysiology and therapy. Surv Ophthalmol. 1997;41:275–313. Dohlman-Doane device. Am J Ophthalmol. 2005;140:1032–1038. 36. Goins KM. New insights into the diagnosis and treatment of neurotrophic 31. Birkholz ES, Goins KM. Boston keratoprosthesis: an option for patients keratopathy. Ocul Surf. 2005;3:96–110. with multiple failed corneal grafts. 2012. Available at: http://webeye. 37. Li JY, Greiner MA, Brandt JD, et al. Long-term complications associated ophth.uiowa.edu/eyeforum/cases/94-Boston-Keratoprosthesis-Failed- with glaucoma drainage devices and Boston keratoprosthesis. Am J Corneal-Grafts.htm. Accessed August 31, 2015. Ophthalmol. 2011;152:209–218.