- Email: [email protected]
Longer-Term Vision Outcomes and Complications with the Boston Type 1 Keratoprosthesis at the University of California, Davis
Longer-Term Vision Outcomes and Complications with the Boston Type 1 Keratoprosthesis at the University of California, Davis Mark A. Greiner, MD, Jennifer Y. Li, MD, Mark J. Mannis, MD Purpose: To evaluate retention of visual acuity and development of complications after Boston type 1 keratoprosthesis implantation over a longer follow-up period than previously reported. Design: Cohort study. Participants: Forty eyes of 35 patients who underwent Boston type 1 keratoprosthesis surgery at the University of California, Davis, between 2004 and 2010. Methods: Preoperative, intraoperative, and postoperative parameters were collected and analyzed. Main Outcome Measures: Best-corrected visual acuity (BCVA) and postoperative complications. Results: Preoperative visual acuity ranged from 20/150 to light perception and was ⱕ20/400 in 38 eyes (95%). Preoperative diagnoses included failed corneal transplants (19 eyes, 47.5%), chemical injury (10 eyes, 25%), and aniridia (5 eyes, 12.5%). Mean follow-up duration was 33.6 months (range, 5–72 months). Of 36 eyes followed for ⱖ1 year, 32 eyes (89%) achieved postoperative BCVA ⱖ20/200. Of eyes that achieved BCVA ⱖ20/200, at last follow-up, 19 of 32 eyes (59%) followed for ⱖ1 year retained BCVA ⱖ20/200; 16 of 27 eyes (59%) followed for ⱖ2 years retained BCVA ⱖ20/200; 7 of 14 eyes (50%) followed for ⱖ3 years retained BCVA ⱖ20/200; and 2 of 7 eyes (29%) followed for ⱖ4 years retained BCVA ⱖ20/200. End-stage glaucoma most commonly caused vision loss (7 of 13 eyes, 54%) when BCVA ⱖ20/200 was not retained (follow-up ⱖ1 year). Glaucoma was newly diagnosed in 11 eyes (27.5%); progression was noted in 9 eyes (22.5%). Glaucoma drainage device erosion occurred in 9 eyes (22.5%). Retroprosthetic membrane formed in 22 eyes (55%), 5 eyes (12.5%) developed endophthalmitis, 6 eyes (15%) developed corneal melt, 7 eyes (17.5%) underwent keratoprosthesis replacement, and 23 eyes (57.5%) required major surgery to treat postoperative complications. The initial keratoprosthesis was retained in 32 eyes (80%). Conclusions: Keratoprosthesis implantation remains a viable option for salvaging vision. A significant number of patients lost vision over the postoperative course. Glaucoma and complications related to glaucoma surgery are significant challenges to maintaining good vision after keratoprosthesis surgery. Our study highlights the need for long-term follow-up and a team approach to management, and points to a more guarded long-term visual prognosis after surgery. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Ophthalmology 2011;118:1543–1550 © 2011 by the American Academy of Ophthalmology.
The modern keratoprosthesis was developed to treat patients with severe ocular surface disease and a poor prognosis for traditional penetrating keratoplasty. Specific indications for this implant include graft failure, chemical burns, stem cell deficiencies, and cicatrizing ocular immune conditions, including Stevens–Johnson syndrome (SJS) and ocular cicatricial pemphigoid (OCP). The device is typically implanted after multiple penetrating keratoplasties have been attempted, and in this context, keratoprosthesis surgery represents a last-chance surgical intervention to salvage functional vision. The model most commonly used in the United States is the Boston type 1 keratoprosthesis, developed by Dohlman et al1 © 2011 by the American Academy of Ophthalmology Published by Elsevier Inc.
in the 1970s. This model features a donor cornea sandwiched between a central stem and a back plate. Recent updates to this model include holes in the back plate that allow nutrients to bathe the donor button and a titanium ring to prevent dislocation of the back plate. Postoperative management involves chronic treatment with topical antibiotics, corticosteroids, and a bandage contact lens to help prevent epithelial defects. The back plate design, chronic soft contact lens use, and topical vancomycin administration featured in this style of keratoprosthesis are credited with increased retention and reduced rates of corneal melt and endophthalmitis.2–5 Keratoprosthesis implantation has become a viable surgical intervention to rehabilitate eyes with severe corneal ISSN 0161-6420/11/$–see front matter doi:10.1016/j.ophtha.2010.12.032
1543
Ophthalmology
Volume 118, Number 8, August 2011
and surface disease, but proclamations of success have been based on early follow-up data. The multicenter Boston type 1 keratoprosthesis study reported by Zerbe et al6 represents the most comprehensive report to date, but at 8.5 months, the average follow-up was brief. Reports from Chew et al7 and Aldave et al,8 and from Bradley et al9 at U.C. Davis Health System Eye Center, represent case series reports that demonstrate significant improvements in vision at mean follow-up periods of 16, 17, and 19 months, respectively. However, the average follow-up periods presented in each of these reports remain too brief to determine long-term visual acuity outcomes and prognosis. Maintenance of useful vision after keratoprosthesis surgery has been complicated at U.C. Davis Health System Eye Center by significant postoperative adverse events, especially those related to glaucoma, that lead to vision loss after initial gains and high rates of surgical intervention postoperatively. We conducted a review of this institution’s collective experience with the Boston type 1 keratoprosthesis to investigate retention of visual acuity and determine postoperative complications over a longer period of follow-up than reported previously.
Materials and Methods
Figure 1. Comparison of preoperative BCVA (dark gray) versus postoperative BCVA (light gray). One eye with preoperative BCVA of fix and follow has been coded as hand motions; postoperative BCVA for this patient was light perception. The number of eyes with BCVA of ⱖ20/200 increased from 2 (5%) before keratoprosthesis surgery to 20 (50%) at the last follow-up visit recorded (average follow-up of 33.6 months). Fifteen of 40 eyes (37.5%) achieved BCVA of ⱖ20/100. BCVA ⫽ best-corrected visual acuity; CF ⫽ counting fingers; HM ⫽ hand motions; LP ⫽ light perception; NLP ⫽ no light perception.
Surgical Procedure The Boston type 1 keratoprosthesis is obtained from the Massachusetts Eye and Ear Infirmary (Boston, MA). Details of the device and its surgical implantation have been described in detail,6 and the surgical technique and postoperative management at our institution do not differ significantly from that of the Boston Type 1 Keratoprosthesis Study Group.
Data Collection and Analysis We collected and analyzed data by chart review on a retrospective basis. All Boston type 1 keratoprosthesis surgeries performed at the U.C. Davis Health System Eye Center between May 2004 and May 2010 were included for analysis. Visual acuity and postoperative complications were tracked at postoperative months 1, 3, 6, 9, and 12, at annual intervals thereafter, and at the last follow-up visit recorded. Thirty eyes were included for analysis in the initial report on keratoprosthesis surgery at U.C. Davis.9 Approval from the U.C. Davis Institutional Review Board was obtained.
acuity ranged from 20/150 to light perception (median, count fingers), and was ⱕ20/400 in 38 eyes (95%) (Fig 1). Thirty-seven of 40 eyes were examined within 1 year of the conclusion of the study period. Of the 3 eyes not examined within 1 year of the study’s conclusion, 1 eye never achieved an improvement in vision with keratoprosthesis surgery and the patient was known to have moved away from the area, and 2 eyes had achieved and maintained an improvement in visual acuity to ⱖ20/200 at the last visit of record.
Concomitant Procedures Procedures performed at the time of keratoprosthesis surgery are shown in Table 2 (available at http://aaojournal.org) and include cataract extraction with intraocular lens implantation, tarsorrhaphy, anterior vitrectomy, glaucoma drainage device (GDD) implantation, and amniotic membrane graft. No intraoperative complications were noted.
Preoperative Glaucoma
Results Patient Demographics Fifty-seven percent of patients were male and 43% were female, and the mean patient age was 52.9 years (range, 2– 86 years). Each eye had undergone an average of 2.05 prior corneal transplants (range, 0 – 6 prior grafts; standard deviation [SD], 1.4 prior grafts; mode, 2 prior grafts). The most common preoperative corneal diagnosis was graft failure (19 eyes, 47.5%), followed by chemical injury (10 eyes, 25%), aniridia (5 eyes, 12.5%), congenital hereditary endothelial dystrophy (2 eyes, 5%), Peters anomaly (1 eye, 2.5%), and 1 eye (2.5%) each with SJS, OCP, and rheumatoid arthritis. Details of the preoperative corneal diagnoses are listed in Table 1 (available at http://aaojournal.org). Preoperative visual
1544
Twenty-three eyes (57.5%) carried a preoperative diagnosis of glaucoma, defined as requiring antihypertensive medications at the time of keratoprosthesis surgery or having undergone prior incisional glaucoma surgery or cyclodestructive procedures. On average, 1.52 antihypertensive drops per eye were used by patients with glaucoma before keratoprosthesis placement. Fourteen eyes (35%) had prior glaucoma surgery, including 11 eyes (27.5%) with GDDs, 5 eyes (12.5%) with trabeculectomies, and 4 eyes (10%) with transscleral diode cyclophotocoagulation.
Visual Acuity Outcomes Keratoprosthesis surgery resulted in significant and often dramatic improvements in vision for patients in our series (Fig 1). The
Greiner et al 䡠 Longer-Term Vision with the Boston KPro Table 3. Visual and Surgical Outcomes by Original Diagnosis Preoperative Diagnosis
Eyes Achieving BCVA >20/200*
Eyes Maintaining BCVA >20/200*
Eyes Requiring Major Surgery
Retention of Initial Kpro
Graft failure (n ⫽ 19) Chemical injury (n ⫽ 10) Aniridia (n ⫽ 5) Other (n ⫽ 6) Total (n ⫽ 40)
16 9 4 3 32
10 (63%) 7 (78%) 2 (50%) 0 19 (59%)
12 (63%) 2 (20%) 5 (100%) 4 (67%) 23 (58%)
16 (84%) 10 (100%) 4 (80%) 4 (67%) 32 (80)%
Kpro ⫽ keratoprosthesis. Average follow-up: 33.6 mos (range, 5–72 mos; SD, 16.5 mos; median, 33.0 mos). All 40 eyes included in the tabulation of eyes requiring major surgery. See for a listing of procedures. Preoperative diagnoses labeled “Other” include congenital hereditary endothelial dystrophy (2 eyes) and Peters anomaly, OCP, SJS, and rheumatoid arthritis (1 eye each). *Eyes never achieving BCVA ⱖ20/200 or followed for ⬍1 year are not included in the visual acuity outcomes but are included in the surgical complication rate.
number of eyes with a best-corrected visual acuity (BCVA) of ⱖ20/200 increased from 2 (5%) before keratoprosthesis surgery to 20 (50%) at the last follow-up visit recorded, over an average follow-up period of 33.6 months (range, 5–72 months; SD, 16.5 months; median, 33.0 months). Fifteen of 40 eyes (37.5%) achieved BCVA of ⱖ20/100 at last follow-up. In the subgroup of 36 eyes followed for at least 1 year after keratoprosthesis implantation (mean follow-up, 37.2 months; range, 13–72 months; SD, 14.3 months; median, 33.8 months), 32 eyes (89%) achieved BCVA of ⱖ20/200 at some point in the postoperative period. Of the 32 eyes that achieved ⱖ20/200 vision, 19 of these 32 eyes (59%) retained BCVA of ⱖ20/200 and 15 eyes (47%) retained BCVA ⱖ20/100 at the time of the last follow-up visit. Four eyes (10%) of the 36 followed for a minimum of 1 year failed to achieve 20/200 vision. Of these 4 eyes, 1 eye underwent keratoprosthesis surgery for graft failures, was likely prephthisical at implantation, had a GDD erosion, and developed frank phthisis; a second eye (Peters’ anomaly, age 2 years at implantation) developed microbial keratitis leading to endophthalmitis and removal of the keratoprosthesis; a third eye had surgery for graft failures and developed a retroprosthetic membrane refractory to surgery; and the fourth eye, with a history of acute retinal necrosis and failed grafts, likely had a central retinal artery occlusion preoperatively and developed a prephthisical state postoperatively. Three of these 4 patients carried a preoperative diagnosis of glaucoma that required incisional surgery. The subgroup of 36 eyes followed for 1 year or more was analyzed further. Eighteen of 36 eyes (50%) gained 3 or more lines of acuity on Snellen chart testing with keratoprosthesis surgery. This cohort includes 7 of 9 eyes with a preoperative diagnosis of chemical injury and 3 of 4 eyes with herpes keratitis. The remaining 18 eyes (50%) followed for 1 year or more did not have a significant change in vision, lost 3 or more lines of acuity, or lost light perception altogether. This cohort without significant improvement includes 2 of 4 patients with preoperative aniridia and the patients with SJS, OCP, and rheumatoid arthritis (1 each). Subgroup analyses of visual outcomes, surgical complications, and anatomic retention based on preoperative diagnoses are shown in Table 3. In the subgroup of 29 eyes followed for 2 years or more after keratoprosthesis implantation (mean follow-up, 40.8 months; range, 24 –72 months; SD, 13.0 months; median, 35.4 months), 27 eyes (93%) achieved BCVA of ⱖ20/200 in the postoperative period, and 16 of these 24 eyes (59%) retained BCVA of ⱖ20/200 at the time of the last follow-up visit. In the subgroup of 14 eyes followed for 3 years or more after keratoprosthesis implantation (mean follow-up, 51.4 months;
range, 39 –72 months; SD, 10.9 months; median, 48.8 months), all 14 eyes (100%) achieved BCVA of ⱖ20/200 postoperatively, and 7 of these 14 eyes (50%) retained BCVA of ⱖ20/200 at the time of the last follow-up visit. In the subgroup of 7 eyes followed for 4 years or more after keratoprosthesis surgery (mean follow-up, 60.4 months; range, 50 –72 months; SD, 7.4 months; median, 59.8 months), all 7 eyes (100%) achieved BCVA of ⱖ20/200 in the postoperative period, but only 2 of these 7 eyes (29%) retained BCVA of ⱖ20/200 at the time of the last follow-up visit.
Vision Loss after Initial Improvement We plotted the percentage of eyes that retained BCVA of ⱖ20/200 after achieving this level of vision by year of postoperative follow-up (Fig 2). As this plot indicates, after an initial plateau in the first and second postoperative years, a decreasing percentage of patients retain BCVA as the time from implantation increases. Causes of vision loss were investigated in patients with a minimum follow-up of 1 year. Of the 36 eyes followed for at least 1 year after keratoprosthesis surgery, 13 eyes had achieved but failed to retain BCVA of 20/200 at the time of the last visit of record (Table 4, available at http://aaojournal.org). As indicated in Table 4,
Figure 2. Percentage of eyes that retained BCVA of 20/200 over time. After an initial plateau in the first and second postoperative years, a decreasing percentage of patients retain BCVA as the time from implantation increases. Progression to end-stage glaucoma was causal in the majority of cases. BCVA ⫽ best-corrected visual acuity.
1545
Ophthalmology
Volume 118, Number 8, August 2011
progression to end-stage glaucoma— defined as severe vision loss (BCVA ⬍20/400) with documented end-stage glaucomatous optic neuropathy or field loss—was implicated as causal in the majority of cases (7 of 13 eyes, 54%). Other causes of vision loss after achieving an initial improvement in BCVA to ⱖ20/200 are listed in Table 4 (available at http://aaojournal.org). In total, 6 complications leading to vision loss occurred earlier in the postoperative course (at or before 9 months), and 7 occurred later in the course (23 months and beyond). Comparisons between the group of eyes that retained ⱖ20/200 visual acuity and the group of eyes that failed to retain that level of vision revealed no significant differences in the number of eyes with preoperative glaucoma, preoperative glaucoma surgery, or GDD placed at any point before, during, or after keratoprosthesis surgery (data not shown).
Postoperative Complications Retroprosthetic membrane formation was the most common postoperative complication observed, occurring in more than half of all eyes (22 eyes, 55%) (Table 5). Most eyes did not require treatment. Membranotomy with yttrium-aluminum-garnet (YAG) laser was required in 10 eyes (25%). However, retroprosthetic membranes were refractory to YAG laser membranotomy in 5 eyes (12.5%), and surgical membranectomy was performed in these cases. Glaucoma and related complications affected a significant number of eyes. Thirty-four eyes (85%) were being treated for glaucoma, including 23 eyes (57.5%) with a preoperative glaucoma diagnosis and 11 eyes (27.5%) in which a new glaucoma diagnosis was made on the basis of glaucomatous optic neuropathy, characteristic field loss, or elevated intraocular pressure. Postoperatively, elevated intraocular pressure by tactile measurement was noted in 16 eyes (40%), and an average of 1.22 glaucoma drops per eye was used to lower intraocular pressure. Glaucoma progression was noted in 9 eyes (22.5%), including 4 eyes with previously implanted GDDs. Glaucoma drainage device erosion
Table 5. Postoperative Complications after Boston Type 1 Keratoprosthesis Implant Postoperative Complications
No. Eyes
% Eyes
Retroprosthetic membrane YAG membranotomy Surgical membranectomy Glaucoma Elevated intraocular pressure New diagnosis postoperatively Progression End-stage GDD erosion Endocyclophotocoagulation Transscleral diode cyclophotocoagulation New GDD implant Endophthalmitis Corneal melt Keratoprosthesis extrusion Keratoprosthesis replacement Keratoprosthesis removal
22 10 5 34 16 11 9 7 9 3 2 2 5 6 6 7* 4*
55 25 12.5 85 40 27.5 22.5 17.5 22.5 7.5 5 5 12.5 15 15 17.5 10
GDD ⫽ glaucoma drainage device; YAG ⫽ yttrium-aluminum-garnet laser. *One patient with corneal melt with keratoprosthesis extrusion chose explantation rather than replacement.
1546
occurred in 9 eyes (22.5%) and was associated with endophthalmitis in 2 cases. In 1 case (patient 5, Table 4, available at http:// aaojournal.org), cultures positive for Pseudomonas aeruginosa were isolated from vitreous after a first GDD became exposed, and 2 months later from a second GDD that also became exposed. In a second case (patient 8, Table 4, available at http://aaojournal. org), the initial GDD erosion was associated with periprosthetic infiltrate and keratoprosthesis extrusion (cultures positive for methicillin-sensitive Staphylococcus aureus [MSSA]), and 9 months after repair, the same device became exposed again and led to development of endophthalmitis (cultures negative). Glaucoma drainage device erosions required 15 additional surgeries to revise or remove the device, and required explantation in 5 eyes (12.5%). Fifteen percent of all eyes required a surgical or laser procedure to treat uncontrolled intraocular pressure, including 3 eyes (7.5%) that underwent endocyclophotocoagulation, 2 eyes (5%) that underwent cyclophotocoagulation, and 2 eyes (5%) that underwent GDD implantation. Endophthalmitis was documented in 5 eyes (12.5%). All patients were using topical vancomycin (50 mg/ml) at the time of presentation. Three of the 5 cases developed from microbial keratitis of the periprosthetic corneal graft, and the other 2 cases developed from GDD erosions (patients 5 and 8, Table 4, available at http://aaojournal.org). The causative organisms isolated included P. aeruginosa (1 eye), Proteus mirabilis (1 eye), Haemophilus influenzae (1 eye), and Candida parapsilosis (1 eye). As mentioned previously, cultures were negative in 1 case of clinical endophthalmitis associated with GDD reexposure (patient 8, Table 4, available at http://aaojournal.org). Periprosthetic infiltrates not leading to endophthalmitis occurred in 2 eyes (5%), with isolates of MSSA and Streptococcus viridans recovered. Retention of the initial keratoprosthesis was achieved in 32 eyes (80%) over an average follow-up of 33.6 months (range, 5–72 months; SD, 16.5 months; median, 33.0 months) (Table 3). Six eyes (15%) developed corneal melt leading to keratoprosthesis extrusion; 5 of these eyes required surgery to reimplant the device, and 1 patient chose explantation rather than replacement. Two additional eyes (5%) required keratoprosthesis replacement in conjunction with other complications (surgical membranectomy and GDD exposure, 1 eye each). Overall, 16 keratoprosthesis replacement surgeries were performed in the 8 eyes that failed to retain the initial keratoprosthesis (mean, 2.13 surgeries; range, 1– 4 surgeries; SD, 1.5 surgeries; median, 2.0 surgeries). Of the 6 eyes that had corneal melts, 5 eyes (12.5%) developed noninfectious corneal melts, and 1 eye (Patient 8, Table 4, available at http://aaojournal. org) developed extrusion in association with microbial keratitis (MSSA) and GDD erosion. Underlying diagnoses of the eyes that experienced corneal melts included SJS, OCP, rheumatoid arthritis, aniridia, Peters anomaly, and failed grafts (1 eye each). Keratoprosthesis replacement failed and required explantation in 3 eyes (7.5%), including 1 case of recurrent corneal melting related to rheumatoid arthritis in which 4 attempts were made to replace the device. At last follow-up, 36 eyes (90%) retained a keratoprosthesis device. The incidence of complications requiring surgery in the postoperative period has been tabulated on a per eye, per annum basis (Table 6, available at http://aaojournal.org). In total, 23 of 40 eyes (57.5%) underwent at least 1 major surgical procedure in the postoperative period, including all 5 eyes (100%) with aniridia, 12 of 19 eyes (63%) with failed grafts, and each of the eyes with Peters’ anomaly, SJS, and rheumatoid arthritis (Table 3). Only 2 of 10 eyes (20%) that underwent keratoprosthesis surgery for chemical injury developed a complication that required major surgery.
Greiner et al 䡠 Longer-Term Vision with the Boston KPro
Discussion The Boston type 1 keratoprosthesis has been developed to salvage vision in eyes with severe corneal and ocular surface disease. The device is implanted in eyes with the highest probability of traditional graft failure and continued corneal blindness and the lowest probability of regaining useful vision by any other means. Successful surgery can yield rapid improvement in vision10 and effect positive transformations on the lives of patients and their families. Our study confirms that good visual outcomes are achieved with keratoprosthesis surgery but puts into question the long-term sustainability of vision improvements.
Visual Acuity Outcomes Keratoprosthesis surgery in our study was performed on patients with poor preoperative vision (preoperative visual acuity ⱕ20/400 in 95% of eyes) and a high percentage of preoperative glaucoma (57.5%). These figures are comparable to the Multicenter Boston Type 1 Keratoprosthesis Study (94% with preoperative visual acuity ⱕ20/400, 52% with preoperative glaucoma) and to the series by Aldave et al8 (90% with preoperative visual acuity ⱕ20/400, 76% with preoperative glaucoma), Chew et al7 (86% with preoperative visual acuity ⱕ20/400, preoperative glaucoma in 73%), and Bradley et al9 (86% with preoperative visual acuity ⱕ20/400, preoperative glaucoma in 60%). In our study, the majority of patients underwent keratoprosthesis surgery for a history of multiple failed grafts or chemical burns, and as in the major studies mentioned our series, a minority percentage of patients had limbal stem cell deficiency or autoimmune conditions. We also show visual acuity outcomes at a minimum of 1 year of follow-up (89% achieving BCVA ⱖ20/200) that are comparable to these studies (Zerbe et al6 with 57% achieving BCVA ⱖ20/200 at an average of 8.5 months, Aldave et al8 with 75% achieving BCVA ⱖ20/100 after 1 year, Chew et al7 with 89% achieving BCVA ⱖ20/200 after 14 months, and Bradley et al9 with 75% achieving BCVA ⱖ20/200 at an average of 28 months). Variance in the percentage of patients achieving improved visual acuity after surgery may be due to small differences in the number of patients treated with poorer prognosis conditions, and in the case of the multicenter study, lack of standardization in surgical technique among a large pool of surgeons. Each of the major studies of the Boston type 1 keratoprosthesis to date had average follow-up periods that were significantly shorter than ours. With an average follow-up of 33.6 months postimplantation, we were able to investigate matters related to the retention of visual acuity, and not simply the early postoperative achievement of good visual acuity, which has been a major focus of previous studies. After gaining vision initially, a significant number of our patients lost vision over the postoperative course. Vision loss events in the postoperative course were evenly distributed between 2 distinct time periods: early (6 events, at or before 9 months) and later (7 events, ⬃2 years and beyond) in the postoperative course. Although the majority of patients who achieved acuity of ⱖ20/200 retained that level of
vision up to 2 years after surgery, declines in retaining the vision gained became evident at 3 years of follow-up and continued through 4 years of follow-up postimplantation, suggesting that our study’s longer follow-up period allowed detection of later postoperative vision-loss events. Endstage glaucomatous optic neuropathy also tended to occur early in our series (5 eyes) and in other reports,7,10 which suggests that keratoprosthesis surgery itself may further compromise borderline optic nerve function and lead to early glaucoma-related vision loss. Later vision loss due to end-stage glaucoma occurred in 2 eyes, both of which had disease progression despite GDD placement before keratoprosthesis surgery, which suggests poor intraocular pressure control may lead to progressive vision loss despite preoperative optimization. Hypotony, vascular occlusion, and endophthalmitis were causes of later-onset vision loss. Severe complications in keratoprosthesis surgery can still strike at any time postoperatively, despite implant design and postoperative treatment modifications made since Yaghouti et al11 reached the same conclusion in their long-term analysis of type 1 and 2 implants approximately a decade ago. Overall, patients who underwent keratoprosthesis surgery for a history of chemical injury had the best outcomes in our series (highest BCVA and implant retention rates, lowest postoperative surgical complication rate; Table 3). This finding stands apart from previous articles that cite poorer outcomes for this subgroup undergoing keratoprosthesis implantation.11 In addition, our patients with aniridia, SJS, and OCP generally had poor outcomes (lowest BCVA and implant retention rates, highest postoperative surgical complication rate; Table 3). This finding stands in contrast with recent multicenter reports suggesting good outcomes for patients with aniridia and SJS,12,13 but echoes previous concerns about performing keratoprosthesis surgery in these very high-risk patients.11
Postoperative Complications We found that the number of eyes requiring major surgical invention ranged between 29% and 50% annually (Table 6, available at http://aaojournal.org). The implication—that with each postoperative year, approximately 1 of every 2 to 3 eyes with a keratoprosthesis may develop a complication requiring surgery—is sobering when considered along with the difficulties in retaining BCVA over time. We chose to analyze the annual incidence of major postoperative surgery per eye, because it is a useful indicator of ongoing postoperative morbidity and will help frame an expectation that additional surgery may be needed to maintain vision and implant function as the keratoprosthesis ages. This calculation likely overestimates the actual number of patients who will require postoperative surgery, because some eyes developed multiple complications. However, the actual number of eyes requiring at least 1 major surgery after keratoprosthesis implantation in our series was significant (23 eyes, 58%) and confirms a similar figure reported by Yaghouti et al11 (23 eyes, 37%) in their series of 63 eyes with type 1 and type 2 keratoprostheses. In this context, it may be useful to investigate the psychology related to visual rehabilitation in keratoprosthesis surgery, including patient and
1547
Ophthalmology
Volume 118, Number 8, August 2011
provider preoperative expectations and postoperative satisfaction, given heightened anticipation for restored vision. In our study, retroprosthetic membrane was the most common complication (22 eyes, 55%). Generally, these were not visually significant complications, and the low morbidity likely reflects heightened awareness of this treatable complication. When they did require treatment (10 eyes, 25%), the response to YAG laser membranotomy was generally favorable, with only 1 patient losing vision after achieving a BCVA of 20/40 (patient 3, Table 4, available at http://aaojournal.org), as the result of a refractory membrane. Rates of visually significant retroprosthetic membranes requiring treatment with YAG laser are conserved across studies (range, 23%–37%).6 – 8,11,14 If the initial attempt was unsuccessful, repeat YAG laser treatments were attempted. We were able to avoid surgical membranectomy in all but 5 eyes (12.5%), which is comparable to the 11% of patients who required this additional surgical procedure in the Multicenter Boston Type 1 Keratoprosthesis Study. No clear cause for these membranes has been identified (Chew et al7 summarize several possibilities). In lieu of a clear cause and in light of favorable treatment response with a standard laser, early detection, early treatment, and appropriate retreatment in the outpatient setting seem to be the most rational approach and the one adopted by most surgeons. Although retroprosthetic membrane formation may be the most commonly observed postoperative complication in our series and in others,6 – 8,14 glaucoma clearly represents the most significant for its impact on vision and resistance to treatment. Elevated intraocular pressure was noted in fewer patients than the total number with glaucoma progression and acquired glaucoma in our series and, we believe, reflects the inaccuracy of tactile pressure measurements in keratoprosthesis eyes.15 Glaucoma progression even occurred in eyes with functional GDDs and appropriate medical management in our series. In conjunction with the alarming rate of GDD erosions, these findings suggest that GDDs may not provide adequate control of intraocular pressure and seem to carry significant risks in this population. Further analysis of our keratoprosthesis cases with glaucoma and GDD erosions has been conducted and will be published separately. Given the significant risks of GDDs in the context of keratoprosthesis surgery, alternatives such as transscleral or endoscopic cyclophotocoagulation should be considered carefully in glaucoma management.16 Our own experience with cyclophotocoagulation has been limited to 4 eyes that underwent transscleral or endoscopic cyclophotocoagulation to control intraocular pressure after GDD removal was required for complications due to endophthalmitis (2 eyes) or in advanced glaucoma (2 eyes). Repeat cyclophotocoagulation treatments were required in 2 patients, suggesting the need for continued surveillance and possibly additional treatment to achieve acceptable intraocular pressures. Endophthalmitis occurred in 5 eyes (12.5%) in our series, including 3 gram-negative and 1 fungal infection. This rate is consistent with other reports7,17 and suggests that the nil incidence of endophthalmitis in the Multicenter Boston Type 1 Keratoprosthesis Study is an artifact of an average
1548
follow-up period too short to detect this clinically significant complication. These infections occurred in patients who underwent keratoprosthesis surgery for failed grafts (4 eyes) and aniridia (1 eye). This trend is in contradistinction to prior reports of higher rates of endophthalmitis in eyes with SJS, OCP, and chemical burns.17 In addition, the spectrum of infectious organisms in our study departs from others. Endophthalmitis after keratoprosthesis surgery typically occurs with gram-positive pathogens; thus, our antibiotic prophylaxis has been focused on daily topical vancomycin administration to cover against typical and resistant gram-positive organisms. Bacterial endophthalmitis occurred only with gram-positive bacteria, and only grampositive skin flora were found on surveillance cultures in the study by Nouri et al.17 Chew et al7 reported a prominence of gram-positive species in their cases of endophthalmitis and related this to discontinuation of vancomycin prophylaxis. Aldave et al8 reported no episodes of endophthalmitis but noted a predominance of fungus in corneal infiltrates. However, all of our patients were using vancomycin eye drops (50 mg/ml) at the time of infection, and developed predominantly gram-negative infections. Our study suggests the need for long-term treatment with antibiotics that cover for gram-negative pathogens in addition to gram-positive organisms. Our findings also suggest that vancomycin prophylaxis may somehow encourage selection of gram-negative infections in eyes with a keratoprosthesis, or at least a segment of this population. Four of the 5 patients with endophthalmitis underwent glaucoma surgery before keratoprosthesis placement (3 with GDDs, 1 with trabeculectomy). Given that the conjunctiva had been violated surgically, and hardware was present in several eyes at the time of keratoprosthesis surgery, there may have been a greater risk for infection with virulent organisms that do not typically grow on the ocular surface. Alternatively, the ocular surface flora may be altered directly by chronic exposure to vancomycin in these eyes. Of note, Barnes et al18 showed that chronic vancomycin administration increases the incidence of fungal infections without changing rates of fungal colonization significantly, suggesting that vancomycin itself may alter conjunctival barrier function in keratoprosthesis cases. The question of gram-negative bacteria selection cannot be addressed without further study of the ocular surface flora in this era of chronic vancomycin administration and soft contact lens wear. What remains clear is that there is a lifetime risk for infection because of the presence of hardware in the eye. Corneal melting was generally an inflammatory rather than an infectious complication in our series. Preoperative diagnoses in these 6 cases were mostly typical, including SJS, OCP, and aniridia, which are well-described conditions predisposing to an inhospitable ocular surface environment and poor prognosis after keratoprosthesis implantation.11,19 Rheumatoid arthritis is not widely reported as a subset of patients who undergo keratoprosthesis implantation. With corneal melting as a clinical hallmark of this disease, the poor outcome we report in our 1 patient with this disease also may be unsurprising. What is remarkable about this case (patient 7, Table 4, available at http://aaojournal.org) is the maintenance of 20/100 acuity after recurrent melts and
Greiner et al 䡠 Longer-Term Vision with the Boston KPro multiple keratoprosthesis replacements; vision loss occurred with removal of the keratoprosthesis, which was required for lack of sufficient tissue integrity to support the implant itself. Of the other 3 eyes with corneal melts that achieved but could not maintain 20/200 vision, 2 eyes lost vision because of other causes late in the postoperative course (patient 3 with recurrent retroprosthetic membrane at 56 months, patient 8 with hypotony and choroidal effusions at 23 months), and 1 eye (patient 13) had no significant ocular comorbidities to limit visual recovery if the patient had elected to pursue replacement of the keratoprosthesis. As these cases suggest, corneal melting in our series did lead to significant postoperative morbidity but did not lead directly to vision loss, and when treated did allow extended maintenance of useful vision. Alternatives to keratoprosthesis surgery carry substantial risks and may not be viable options in the context of severe ocular surface disease. Keratolimbal allograft with or without subsequent keratoplasty has been shown to restore the phenotypic corneal epithelium, but progressive and substantial declines in postoperative graft survival and ambulatory vision (ⱖ20/200) after 2 to 3 years have been reported, and outcomes are comparable to those of keratoprosthesis surgery.20,21 Survival of ambulatory vision after keratolimbal allograft was lowest in patients with SJS in the study by Solomon et al,20 which underscores the challenges raised by our study and others11 of performing visual rehabilitation surgery in this patient population. Keratolimbal allograft surgery usually requires chronic systemic immunosuppressive agents to prolong graft survival but does not prevent acute rejection episodes21 and may limit regimen adherence and graft survival because of systemic toxicity. In addition, the preoperative presence of severe limbal stem cell deficiency, conjunctival scarring, an unstable tear film, and factors that limit a patient’s candidacy for systemic immunosuppression make keratoprosthesis surgery a more viable option in certain cases. Improved outcomes after keratolimbal allograft may be achieved with administration of combined immunosuppressive agents22 and delay of penetrating keratoplasty until ocular surface inflammation subsides.20 Infection, glaucoma, and posterior segment complications are significant postoperative risks common to keratolimbal allograft and keratoprosthesis surgery. Other alternatives to keratoprosthesis exist but may be less viable than staged keratolimbal allograft, including keratolimbal autograft, which is contraindicated in bilateral disease and conjunctival grafting techniques, which may not provide sufficient limbal stem cells to support a clear visual axis. The Boston type 1 keratoprosthesis remains an important and viable option for salvaging vision after multiple keratoplasty failures and in patients with a high risk of failure with traditional grafting methods. Good vision can be achieved with keratoprosthesis implantation. However, after achieving an initial improvement in vision, a significant number of our patients lost vision over the extended postoperative course. Glaucoma and complications related to glaucoma surgery, including GDD erosions, continue to represent a significant challenge in achieving and maintaining good long-term vision outcomes in keratoprosthesis surgery. The authors emphasize the need for a team ap-
proach that begins in the preoperative period, including a meticulous evaluation by both glaucoma and retina specialists. Moreover, consideration should be given to surgical prophylaxis of glaucoma before or at the time of keratoprosthesis surgery in an effort to thwart the progression of glaucoma in the long term. In conclusion, there is a significant need for major surgery in the postoperative period to address complications related to the prosthesis, which surgeons and patients must anticipate. Also, with nearly ubiquitous use of vancomycin in the postoperative course, gram-negative and fungal endophthalmitis cases such as those reported in this article may represent a burgeoning problem and warrant a heightened awareness of this devastating complication. With the longest follow-up period reported to date, our study highlights the need for continued long-term follow-up in patients undergoing keratoprosthesis surgery and points to a more guarded long-term visual prognosis after surgery.
References 1. Dohlman CH, Schneider HA, Doane MG. Prosthokeratoplasty. Am J Ophthalmol 1974;77:694 –700. 2. Harissi-Dagher M, Khan BF, Schaumberg DA, Dohlman CH. Importance of nutrition to corneal grafts when used as a carrier of the Boston keratoprosthesis. Cornea 2007;26:564 – 8. 3. Dohlman CH, Dudenhoefer EJ, Khan BF, Morneault S. Protection of the ocular surface after keratoprosthesis surgery: the role of soft contact lenses. CLAO J 2002;28:72– 4. 4. Khan BF, Harissi-Dagher M, Khan DM, Dohlman CH. Advances in Boston keratoprosthesis: enhancing retention and prevention of infection and inflammation. Int Ophthalmol Clin 2007;47:61–71. 5. Durand ML, Dohlman CH. Successful prevention of bacterial endophthalmitis in eyes with the Boston keratoprosthesis. Cornea 2009;28:896 –901. 6. Zerbe BL, Belin MW, Ciolino JB, Boston Type 1 Keratoprosthesis Study Group. Results from the multicenter Boston Type 1 Keratoprosthesis Study. Ophthalmology 2006; 113:1779 – 84. 7. Chew HF, Ayres BD, Hammersmith KM, et al. Boston keratoprosthesis outcomes and complications. Cornea 2009;28: 989 –96. 8. Aldave AJ, Kamal KM, Vo RC, Yu F. The Boston type I keratoprosthesis: improving outcomes and expanding indications. Ophthalmology 2009;116:640 –51. 9. Bradley JC, Hernandez EG, Schwab IR, Mannis MJ. Boston type 1 keratoprosthesis: the University of California Davis experience. Cornea 2009;28:321–7. 10. Dunlap K, Chak G, Aquavella JV, et al. Short-term visual outcomes of Boston type 1 keratoprosthesis implantation. Ophthalmology 2010;117:687–92. 11. Yaghouti F, Nouri M, Abad JC, et al. Keratoprosthesis: preoperative prognostic categories. Cornea 2001;20:19 –23. 12. Akpek EK, Harissi-Dagher M, Petrarca R, et al. Outcomes of Boston keratoprosthesis in aniridia: a retrospective multicenter study. Am J Ophthalmol 2007;144:227–31. 13. Sayegh RR, Ang LP, Foster CS, Dohlman CH. The Boston keratoprosthesis in Stevens-Johnson syndrome. Am J Ophthalmol 2008;145:438 – 44.
1549
Ophthalmology
Volume 118, Number 8, August 2011
14. Aquavella JV, Gearinger MD, Akpek EK, McCormick GJ. Pediatric keratoprosthesis. Ophthalmology 2007;114:989 – 94. 15. Netland PA, Terada H, Dohlman CH. Glaucoma associated with keratoprosthesis. Ophthalmology 1998;105:751–7. 16. Rivier D, Paula JS, Kim E, et al. Glaucoma and keratoprosthesis surgery: role of adjunctive cyclophotocoagulation. J Glaucoma 2009;18:321– 4. 17. Nouri M, Terada H, Alfonso EC, et al. Endophthalmitis after keratoprosthesis: incidence, bacterial causes, and risk factors. Arch Ophthalmol 2001;119:484 –9. 18. Barnes SD, Dohlman CH, Durand ML. Fungal colonization and infection in Boston keratoprosthesis. Cornea 2007;26:9–15.
19. Khan B, Dudenhoefer EJ, Dohlman CH. Keratoprosthesis: an update. Curr Opin Ophthalmol 2001;12:282–7. 20. Solomon A, Ellies P, Anderson DF, et al. Allograft with or without penetrating keratoplasty for total limbal stem cell deficiency. Ophthalmology 2002;109:1159 – 66. 21. Ilari L, Daya SM. Long-term outcomes of keratolimbal allograft for the treatment of severe ocular surface disorders. Ophthalmology 2002;109:1278 – 84. 22. Liang L, Sheha H, Tseng SC. Long-term outcomes of keratolimbal allograft for total limbal stem cell deficiency using combined immunosuppressive agents and correction of ocular surface deficits. Arch Ophthalmol 2009;127: 1428 –34.
Footnotes and Financial Disclosures Originally received: July 24, 2010. Final revision: December 26, 2010. Accepted: December 30, 2010. Available online: March 21, 2011.
Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Manuscript no. 2010-1024.
U.C. Davis Health System Eye Center, University of California at Davis, Sacramento, California.
1550
Correspondence: Mark J. Mannis, MD, Chair, Department of Ophthalmology and Vision Science, U.C. Davis Health System Eye Center, 4860 Y Street, Suite 2400, Sacramento, CA 95817. E-mail: [email protected].
The modern keratoprosthesis was developed to treat patients with severe ocular surface disease and a poor prognosis for traditional penetrating keratoplasty. Specific indications for this implant include graft failure, chemical burns, stem cell deficiencies, and cicatrizing ocular immune conditions, including Stevens–Johnson syndrome (SJS) and ocular cicatricial pemphigoid (OCP). The device is typically implanted after multiple penetrating keratoplasties have been attempted, and in this context, keratoprosthesis surgery represents a last-chance surgical intervention to salvage functional vision. The model most commonly used in the United States is the Boston type 1 keratoprosthesis, developed by Dohlman et al1 © 2011 by the American Academy of Ophthalmology Published by Elsevier Inc.
in the 1970s. This model features a donor cornea sandwiched between a central stem and a back plate. Recent updates to this model include holes in the back plate that allow nutrients to bathe the donor button and a titanium ring to prevent dislocation of the back plate. Postoperative management involves chronic treatment with topical antibiotics, corticosteroids, and a bandage contact lens to help prevent epithelial defects. The back plate design, chronic soft contact lens use, and topical vancomycin administration featured in this style of keratoprosthesis are credited with increased retention and reduced rates of corneal melt and endophthalmitis.2–5 Keratoprosthesis implantation has become a viable surgical intervention to rehabilitate eyes with severe corneal ISSN 0161-6420/11/$–see front matter doi:10.1016/j.ophtha.2010.12.032
1543
Ophthalmology
Volume 118, Number 8, August 2011
and surface disease, but proclamations of success have been based on early follow-up data. The multicenter Boston type 1 keratoprosthesis study reported by Zerbe et al6 represents the most comprehensive report to date, but at 8.5 months, the average follow-up was brief. Reports from Chew et al7 and Aldave et al,8 and from Bradley et al9 at U.C. Davis Health System Eye Center, represent case series reports that demonstrate significant improvements in vision at mean follow-up periods of 16, 17, and 19 months, respectively. However, the average follow-up periods presented in each of these reports remain too brief to determine long-term visual acuity outcomes and prognosis. Maintenance of useful vision after keratoprosthesis surgery has been complicated at U.C. Davis Health System Eye Center by significant postoperative adverse events, especially those related to glaucoma, that lead to vision loss after initial gains and high rates of surgical intervention postoperatively. We conducted a review of this institution’s collective experience with the Boston type 1 keratoprosthesis to investigate retention of visual acuity and determine postoperative complications over a longer period of follow-up than reported previously.
Materials and Methods
Figure 1. Comparison of preoperative BCVA (dark gray) versus postoperative BCVA (light gray). One eye with preoperative BCVA of fix and follow has been coded as hand motions; postoperative BCVA for this patient was light perception. The number of eyes with BCVA of ⱖ20/200 increased from 2 (5%) before keratoprosthesis surgery to 20 (50%) at the last follow-up visit recorded (average follow-up of 33.6 months). Fifteen of 40 eyes (37.5%) achieved BCVA of ⱖ20/100. BCVA ⫽ best-corrected visual acuity; CF ⫽ counting fingers; HM ⫽ hand motions; LP ⫽ light perception; NLP ⫽ no light perception.
Surgical Procedure The Boston type 1 keratoprosthesis is obtained from the Massachusetts Eye and Ear Infirmary (Boston, MA). Details of the device and its surgical implantation have been described in detail,6 and the surgical technique and postoperative management at our institution do not differ significantly from that of the Boston Type 1 Keratoprosthesis Study Group.
Data Collection and Analysis We collected and analyzed data by chart review on a retrospective basis. All Boston type 1 keratoprosthesis surgeries performed at the U.C. Davis Health System Eye Center between May 2004 and May 2010 were included for analysis. Visual acuity and postoperative complications were tracked at postoperative months 1, 3, 6, 9, and 12, at annual intervals thereafter, and at the last follow-up visit recorded. Thirty eyes were included for analysis in the initial report on keratoprosthesis surgery at U.C. Davis.9 Approval from the U.C. Davis Institutional Review Board was obtained.
acuity ranged from 20/150 to light perception (median, count fingers), and was ⱕ20/400 in 38 eyes (95%) (Fig 1). Thirty-seven of 40 eyes were examined within 1 year of the conclusion of the study period. Of the 3 eyes not examined within 1 year of the study’s conclusion, 1 eye never achieved an improvement in vision with keratoprosthesis surgery and the patient was known to have moved away from the area, and 2 eyes had achieved and maintained an improvement in visual acuity to ⱖ20/200 at the last visit of record.
Concomitant Procedures Procedures performed at the time of keratoprosthesis surgery are shown in Table 2 (available at http://aaojournal.org) and include cataract extraction with intraocular lens implantation, tarsorrhaphy, anterior vitrectomy, glaucoma drainage device (GDD) implantation, and amniotic membrane graft. No intraoperative complications were noted.
Preoperative Glaucoma
Results Patient Demographics Fifty-seven percent of patients were male and 43% were female, and the mean patient age was 52.9 years (range, 2– 86 years). Each eye had undergone an average of 2.05 prior corneal transplants (range, 0 – 6 prior grafts; standard deviation [SD], 1.4 prior grafts; mode, 2 prior grafts). The most common preoperative corneal diagnosis was graft failure (19 eyes, 47.5%), followed by chemical injury (10 eyes, 25%), aniridia (5 eyes, 12.5%), congenital hereditary endothelial dystrophy (2 eyes, 5%), Peters anomaly (1 eye, 2.5%), and 1 eye (2.5%) each with SJS, OCP, and rheumatoid arthritis. Details of the preoperative corneal diagnoses are listed in Table 1 (available at http://aaojournal.org). Preoperative visual
1544
Twenty-three eyes (57.5%) carried a preoperative diagnosis of glaucoma, defined as requiring antihypertensive medications at the time of keratoprosthesis surgery or having undergone prior incisional glaucoma surgery or cyclodestructive procedures. On average, 1.52 antihypertensive drops per eye were used by patients with glaucoma before keratoprosthesis placement. Fourteen eyes (35%) had prior glaucoma surgery, including 11 eyes (27.5%) with GDDs, 5 eyes (12.5%) with trabeculectomies, and 4 eyes (10%) with transscleral diode cyclophotocoagulation.
Visual Acuity Outcomes Keratoprosthesis surgery resulted in significant and often dramatic improvements in vision for patients in our series (Fig 1). The
Greiner et al 䡠 Longer-Term Vision with the Boston KPro Table 3. Visual and Surgical Outcomes by Original Diagnosis Preoperative Diagnosis
Eyes Achieving BCVA >20/200*
Eyes Maintaining BCVA >20/200*
Eyes Requiring Major Surgery
Retention of Initial Kpro
Graft failure (n ⫽ 19) Chemical injury (n ⫽ 10) Aniridia (n ⫽ 5) Other (n ⫽ 6) Total (n ⫽ 40)
16 9 4 3 32
10 (63%) 7 (78%) 2 (50%) 0 19 (59%)
12 (63%) 2 (20%) 5 (100%) 4 (67%) 23 (58%)
16 (84%) 10 (100%) 4 (80%) 4 (67%) 32 (80)%
Kpro ⫽ keratoprosthesis. Average follow-up: 33.6 mos (range, 5–72 mos; SD, 16.5 mos; median, 33.0 mos). All 40 eyes included in the tabulation of eyes requiring major surgery. See for a listing of procedures. Preoperative diagnoses labeled “Other” include congenital hereditary endothelial dystrophy (2 eyes) and Peters anomaly, OCP, SJS, and rheumatoid arthritis (1 eye each). *Eyes never achieving BCVA ⱖ20/200 or followed for ⬍1 year are not included in the visual acuity outcomes but are included in the surgical complication rate.
number of eyes with a best-corrected visual acuity (BCVA) of ⱖ20/200 increased from 2 (5%) before keratoprosthesis surgery to 20 (50%) at the last follow-up visit recorded, over an average follow-up period of 33.6 months (range, 5–72 months; SD, 16.5 months; median, 33.0 months). Fifteen of 40 eyes (37.5%) achieved BCVA of ⱖ20/100 at last follow-up. In the subgroup of 36 eyes followed for at least 1 year after keratoprosthesis implantation (mean follow-up, 37.2 months; range, 13–72 months; SD, 14.3 months; median, 33.8 months), 32 eyes (89%) achieved BCVA of ⱖ20/200 at some point in the postoperative period. Of the 32 eyes that achieved ⱖ20/200 vision, 19 of these 32 eyes (59%) retained BCVA of ⱖ20/200 and 15 eyes (47%) retained BCVA ⱖ20/100 at the time of the last follow-up visit. Four eyes (10%) of the 36 followed for a minimum of 1 year failed to achieve 20/200 vision. Of these 4 eyes, 1 eye underwent keratoprosthesis surgery for graft failures, was likely prephthisical at implantation, had a GDD erosion, and developed frank phthisis; a second eye (Peters’ anomaly, age 2 years at implantation) developed microbial keratitis leading to endophthalmitis and removal of the keratoprosthesis; a third eye had surgery for graft failures and developed a retroprosthetic membrane refractory to surgery; and the fourth eye, with a history of acute retinal necrosis and failed grafts, likely had a central retinal artery occlusion preoperatively and developed a prephthisical state postoperatively. Three of these 4 patients carried a preoperative diagnosis of glaucoma that required incisional surgery. The subgroup of 36 eyes followed for 1 year or more was analyzed further. Eighteen of 36 eyes (50%) gained 3 or more lines of acuity on Snellen chart testing with keratoprosthesis surgery. This cohort includes 7 of 9 eyes with a preoperative diagnosis of chemical injury and 3 of 4 eyes with herpes keratitis. The remaining 18 eyes (50%) followed for 1 year or more did not have a significant change in vision, lost 3 or more lines of acuity, or lost light perception altogether. This cohort without significant improvement includes 2 of 4 patients with preoperative aniridia and the patients with SJS, OCP, and rheumatoid arthritis (1 each). Subgroup analyses of visual outcomes, surgical complications, and anatomic retention based on preoperative diagnoses are shown in Table 3. In the subgroup of 29 eyes followed for 2 years or more after keratoprosthesis implantation (mean follow-up, 40.8 months; range, 24 –72 months; SD, 13.0 months; median, 35.4 months), 27 eyes (93%) achieved BCVA of ⱖ20/200 in the postoperative period, and 16 of these 24 eyes (59%) retained BCVA of ⱖ20/200 at the time of the last follow-up visit. In the subgroup of 14 eyes followed for 3 years or more after keratoprosthesis implantation (mean follow-up, 51.4 months;
range, 39 –72 months; SD, 10.9 months; median, 48.8 months), all 14 eyes (100%) achieved BCVA of ⱖ20/200 postoperatively, and 7 of these 14 eyes (50%) retained BCVA of ⱖ20/200 at the time of the last follow-up visit. In the subgroup of 7 eyes followed for 4 years or more after keratoprosthesis surgery (mean follow-up, 60.4 months; range, 50 –72 months; SD, 7.4 months; median, 59.8 months), all 7 eyes (100%) achieved BCVA of ⱖ20/200 in the postoperative period, but only 2 of these 7 eyes (29%) retained BCVA of ⱖ20/200 at the time of the last follow-up visit.
Vision Loss after Initial Improvement We plotted the percentage of eyes that retained BCVA of ⱖ20/200 after achieving this level of vision by year of postoperative follow-up (Fig 2). As this plot indicates, after an initial plateau in the first and second postoperative years, a decreasing percentage of patients retain BCVA as the time from implantation increases. Causes of vision loss were investigated in patients with a minimum follow-up of 1 year. Of the 36 eyes followed for at least 1 year after keratoprosthesis surgery, 13 eyes had achieved but failed to retain BCVA of 20/200 at the time of the last visit of record (Table 4, available at http://aaojournal.org). As indicated in Table 4,
Figure 2. Percentage of eyes that retained BCVA of 20/200 over time. After an initial plateau in the first and second postoperative years, a decreasing percentage of patients retain BCVA as the time from implantation increases. Progression to end-stage glaucoma was causal in the majority of cases. BCVA ⫽ best-corrected visual acuity.
1545
Ophthalmology
Volume 118, Number 8, August 2011
progression to end-stage glaucoma— defined as severe vision loss (BCVA ⬍20/400) with documented end-stage glaucomatous optic neuropathy or field loss—was implicated as causal in the majority of cases (7 of 13 eyes, 54%). Other causes of vision loss after achieving an initial improvement in BCVA to ⱖ20/200 are listed in Table 4 (available at http://aaojournal.org). In total, 6 complications leading to vision loss occurred earlier in the postoperative course (at or before 9 months), and 7 occurred later in the course (23 months and beyond). Comparisons between the group of eyes that retained ⱖ20/200 visual acuity and the group of eyes that failed to retain that level of vision revealed no significant differences in the number of eyes with preoperative glaucoma, preoperative glaucoma surgery, or GDD placed at any point before, during, or after keratoprosthesis surgery (data not shown).
Postoperative Complications Retroprosthetic membrane formation was the most common postoperative complication observed, occurring in more than half of all eyes (22 eyes, 55%) (Table 5). Most eyes did not require treatment. Membranotomy with yttrium-aluminum-garnet (YAG) laser was required in 10 eyes (25%). However, retroprosthetic membranes were refractory to YAG laser membranotomy in 5 eyes (12.5%), and surgical membranectomy was performed in these cases. Glaucoma and related complications affected a significant number of eyes. Thirty-four eyes (85%) were being treated for glaucoma, including 23 eyes (57.5%) with a preoperative glaucoma diagnosis and 11 eyes (27.5%) in which a new glaucoma diagnosis was made on the basis of glaucomatous optic neuropathy, characteristic field loss, or elevated intraocular pressure. Postoperatively, elevated intraocular pressure by tactile measurement was noted in 16 eyes (40%), and an average of 1.22 glaucoma drops per eye was used to lower intraocular pressure. Glaucoma progression was noted in 9 eyes (22.5%), including 4 eyes with previously implanted GDDs. Glaucoma drainage device erosion
Table 5. Postoperative Complications after Boston Type 1 Keratoprosthesis Implant Postoperative Complications
No. Eyes
% Eyes
Retroprosthetic membrane YAG membranotomy Surgical membranectomy Glaucoma Elevated intraocular pressure New diagnosis postoperatively Progression End-stage GDD erosion Endocyclophotocoagulation Transscleral diode cyclophotocoagulation New GDD implant Endophthalmitis Corneal melt Keratoprosthesis extrusion Keratoprosthesis replacement Keratoprosthesis removal
22 10 5 34 16 11 9 7 9 3 2 2 5 6 6 7* 4*
55 25 12.5 85 40 27.5 22.5 17.5 22.5 7.5 5 5 12.5 15 15 17.5 10
GDD ⫽ glaucoma drainage device; YAG ⫽ yttrium-aluminum-garnet laser. *One patient with corneal melt with keratoprosthesis extrusion chose explantation rather than replacement.
1546
occurred in 9 eyes (22.5%) and was associated with endophthalmitis in 2 cases. In 1 case (patient 5, Table 4, available at http:// aaojournal.org), cultures positive for Pseudomonas aeruginosa were isolated from vitreous after a first GDD became exposed, and 2 months later from a second GDD that also became exposed. In a second case (patient 8, Table 4, available at http://aaojournal. org), the initial GDD erosion was associated with periprosthetic infiltrate and keratoprosthesis extrusion (cultures positive for methicillin-sensitive Staphylococcus aureus [MSSA]), and 9 months after repair, the same device became exposed again and led to development of endophthalmitis (cultures negative). Glaucoma drainage device erosions required 15 additional surgeries to revise or remove the device, and required explantation in 5 eyes (12.5%). Fifteen percent of all eyes required a surgical or laser procedure to treat uncontrolled intraocular pressure, including 3 eyes (7.5%) that underwent endocyclophotocoagulation, 2 eyes (5%) that underwent cyclophotocoagulation, and 2 eyes (5%) that underwent GDD implantation. Endophthalmitis was documented in 5 eyes (12.5%). All patients were using topical vancomycin (50 mg/ml) at the time of presentation. Three of the 5 cases developed from microbial keratitis of the periprosthetic corneal graft, and the other 2 cases developed from GDD erosions (patients 5 and 8, Table 4, available at http://aaojournal.org). The causative organisms isolated included P. aeruginosa (1 eye), Proteus mirabilis (1 eye), Haemophilus influenzae (1 eye), and Candida parapsilosis (1 eye). As mentioned previously, cultures were negative in 1 case of clinical endophthalmitis associated with GDD reexposure (patient 8, Table 4, available at http://aaojournal.org). Periprosthetic infiltrates not leading to endophthalmitis occurred in 2 eyes (5%), with isolates of MSSA and Streptococcus viridans recovered. Retention of the initial keratoprosthesis was achieved in 32 eyes (80%) over an average follow-up of 33.6 months (range, 5–72 months; SD, 16.5 months; median, 33.0 months) (Table 3). Six eyes (15%) developed corneal melt leading to keratoprosthesis extrusion; 5 of these eyes required surgery to reimplant the device, and 1 patient chose explantation rather than replacement. Two additional eyes (5%) required keratoprosthesis replacement in conjunction with other complications (surgical membranectomy and GDD exposure, 1 eye each). Overall, 16 keratoprosthesis replacement surgeries were performed in the 8 eyes that failed to retain the initial keratoprosthesis (mean, 2.13 surgeries; range, 1– 4 surgeries; SD, 1.5 surgeries; median, 2.0 surgeries). Of the 6 eyes that had corneal melts, 5 eyes (12.5%) developed noninfectious corneal melts, and 1 eye (Patient 8, Table 4, available at http://aaojournal. org) developed extrusion in association with microbial keratitis (MSSA) and GDD erosion. Underlying diagnoses of the eyes that experienced corneal melts included SJS, OCP, rheumatoid arthritis, aniridia, Peters anomaly, and failed grafts (1 eye each). Keratoprosthesis replacement failed and required explantation in 3 eyes (7.5%), including 1 case of recurrent corneal melting related to rheumatoid arthritis in which 4 attempts were made to replace the device. At last follow-up, 36 eyes (90%) retained a keratoprosthesis device. The incidence of complications requiring surgery in the postoperative period has been tabulated on a per eye, per annum basis (Table 6, available at http://aaojournal.org). In total, 23 of 40 eyes (57.5%) underwent at least 1 major surgical procedure in the postoperative period, including all 5 eyes (100%) with aniridia, 12 of 19 eyes (63%) with failed grafts, and each of the eyes with Peters’ anomaly, SJS, and rheumatoid arthritis (Table 3). Only 2 of 10 eyes (20%) that underwent keratoprosthesis surgery for chemical injury developed a complication that required major surgery.
Greiner et al 䡠 Longer-Term Vision with the Boston KPro
Discussion The Boston type 1 keratoprosthesis has been developed to salvage vision in eyes with severe corneal and ocular surface disease. The device is implanted in eyes with the highest probability of traditional graft failure and continued corneal blindness and the lowest probability of regaining useful vision by any other means. Successful surgery can yield rapid improvement in vision10 and effect positive transformations on the lives of patients and their families. Our study confirms that good visual outcomes are achieved with keratoprosthesis surgery but puts into question the long-term sustainability of vision improvements.
Visual Acuity Outcomes Keratoprosthesis surgery in our study was performed on patients with poor preoperative vision (preoperative visual acuity ⱕ20/400 in 95% of eyes) and a high percentage of preoperative glaucoma (57.5%). These figures are comparable to the Multicenter Boston Type 1 Keratoprosthesis Study (94% with preoperative visual acuity ⱕ20/400, 52% with preoperative glaucoma) and to the series by Aldave et al8 (90% with preoperative visual acuity ⱕ20/400, 76% with preoperative glaucoma), Chew et al7 (86% with preoperative visual acuity ⱕ20/400, preoperative glaucoma in 73%), and Bradley et al9 (86% with preoperative visual acuity ⱕ20/400, preoperative glaucoma in 60%). In our study, the majority of patients underwent keratoprosthesis surgery for a history of multiple failed grafts or chemical burns, and as in the major studies mentioned our series, a minority percentage of patients had limbal stem cell deficiency or autoimmune conditions. We also show visual acuity outcomes at a minimum of 1 year of follow-up (89% achieving BCVA ⱖ20/200) that are comparable to these studies (Zerbe et al6 with 57% achieving BCVA ⱖ20/200 at an average of 8.5 months, Aldave et al8 with 75% achieving BCVA ⱖ20/100 after 1 year, Chew et al7 with 89% achieving BCVA ⱖ20/200 after 14 months, and Bradley et al9 with 75% achieving BCVA ⱖ20/200 at an average of 28 months). Variance in the percentage of patients achieving improved visual acuity after surgery may be due to small differences in the number of patients treated with poorer prognosis conditions, and in the case of the multicenter study, lack of standardization in surgical technique among a large pool of surgeons. Each of the major studies of the Boston type 1 keratoprosthesis to date had average follow-up periods that were significantly shorter than ours. With an average follow-up of 33.6 months postimplantation, we were able to investigate matters related to the retention of visual acuity, and not simply the early postoperative achievement of good visual acuity, which has been a major focus of previous studies. After gaining vision initially, a significant number of our patients lost vision over the postoperative course. Vision loss events in the postoperative course were evenly distributed between 2 distinct time periods: early (6 events, at or before 9 months) and later (7 events, ⬃2 years and beyond) in the postoperative course. Although the majority of patients who achieved acuity of ⱖ20/200 retained that level of
vision up to 2 years after surgery, declines in retaining the vision gained became evident at 3 years of follow-up and continued through 4 years of follow-up postimplantation, suggesting that our study’s longer follow-up period allowed detection of later postoperative vision-loss events. Endstage glaucomatous optic neuropathy also tended to occur early in our series (5 eyes) and in other reports,7,10 which suggests that keratoprosthesis surgery itself may further compromise borderline optic nerve function and lead to early glaucoma-related vision loss. Later vision loss due to end-stage glaucoma occurred in 2 eyes, both of which had disease progression despite GDD placement before keratoprosthesis surgery, which suggests poor intraocular pressure control may lead to progressive vision loss despite preoperative optimization. Hypotony, vascular occlusion, and endophthalmitis were causes of later-onset vision loss. Severe complications in keratoprosthesis surgery can still strike at any time postoperatively, despite implant design and postoperative treatment modifications made since Yaghouti et al11 reached the same conclusion in their long-term analysis of type 1 and 2 implants approximately a decade ago. Overall, patients who underwent keratoprosthesis surgery for a history of chemical injury had the best outcomes in our series (highest BCVA and implant retention rates, lowest postoperative surgical complication rate; Table 3). This finding stands apart from previous articles that cite poorer outcomes for this subgroup undergoing keratoprosthesis implantation.11 In addition, our patients with aniridia, SJS, and OCP generally had poor outcomes (lowest BCVA and implant retention rates, highest postoperative surgical complication rate; Table 3). This finding stands in contrast with recent multicenter reports suggesting good outcomes for patients with aniridia and SJS,12,13 but echoes previous concerns about performing keratoprosthesis surgery in these very high-risk patients.11
Postoperative Complications We found that the number of eyes requiring major surgical invention ranged between 29% and 50% annually (Table 6, available at http://aaojournal.org). The implication—that with each postoperative year, approximately 1 of every 2 to 3 eyes with a keratoprosthesis may develop a complication requiring surgery—is sobering when considered along with the difficulties in retaining BCVA over time. We chose to analyze the annual incidence of major postoperative surgery per eye, because it is a useful indicator of ongoing postoperative morbidity and will help frame an expectation that additional surgery may be needed to maintain vision and implant function as the keratoprosthesis ages. This calculation likely overestimates the actual number of patients who will require postoperative surgery, because some eyes developed multiple complications. However, the actual number of eyes requiring at least 1 major surgery after keratoprosthesis implantation in our series was significant (23 eyes, 58%) and confirms a similar figure reported by Yaghouti et al11 (23 eyes, 37%) in their series of 63 eyes with type 1 and type 2 keratoprostheses. In this context, it may be useful to investigate the psychology related to visual rehabilitation in keratoprosthesis surgery, including patient and
1547
Ophthalmology
Volume 118, Number 8, August 2011
provider preoperative expectations and postoperative satisfaction, given heightened anticipation for restored vision. In our study, retroprosthetic membrane was the most common complication (22 eyes, 55%). Generally, these were not visually significant complications, and the low morbidity likely reflects heightened awareness of this treatable complication. When they did require treatment (10 eyes, 25%), the response to YAG laser membranotomy was generally favorable, with only 1 patient losing vision after achieving a BCVA of 20/40 (patient 3, Table 4, available at http://aaojournal.org), as the result of a refractory membrane. Rates of visually significant retroprosthetic membranes requiring treatment with YAG laser are conserved across studies (range, 23%–37%).6 – 8,11,14 If the initial attempt was unsuccessful, repeat YAG laser treatments were attempted. We were able to avoid surgical membranectomy in all but 5 eyes (12.5%), which is comparable to the 11% of patients who required this additional surgical procedure in the Multicenter Boston Type 1 Keratoprosthesis Study. No clear cause for these membranes has been identified (Chew et al7 summarize several possibilities). In lieu of a clear cause and in light of favorable treatment response with a standard laser, early detection, early treatment, and appropriate retreatment in the outpatient setting seem to be the most rational approach and the one adopted by most surgeons. Although retroprosthetic membrane formation may be the most commonly observed postoperative complication in our series and in others,6 – 8,14 glaucoma clearly represents the most significant for its impact on vision and resistance to treatment. Elevated intraocular pressure was noted in fewer patients than the total number with glaucoma progression and acquired glaucoma in our series and, we believe, reflects the inaccuracy of tactile pressure measurements in keratoprosthesis eyes.15 Glaucoma progression even occurred in eyes with functional GDDs and appropriate medical management in our series. In conjunction with the alarming rate of GDD erosions, these findings suggest that GDDs may not provide adequate control of intraocular pressure and seem to carry significant risks in this population. Further analysis of our keratoprosthesis cases with glaucoma and GDD erosions has been conducted and will be published separately. Given the significant risks of GDDs in the context of keratoprosthesis surgery, alternatives such as transscleral or endoscopic cyclophotocoagulation should be considered carefully in glaucoma management.16 Our own experience with cyclophotocoagulation has been limited to 4 eyes that underwent transscleral or endoscopic cyclophotocoagulation to control intraocular pressure after GDD removal was required for complications due to endophthalmitis (2 eyes) or in advanced glaucoma (2 eyes). Repeat cyclophotocoagulation treatments were required in 2 patients, suggesting the need for continued surveillance and possibly additional treatment to achieve acceptable intraocular pressures. Endophthalmitis occurred in 5 eyes (12.5%) in our series, including 3 gram-negative and 1 fungal infection. This rate is consistent with other reports7,17 and suggests that the nil incidence of endophthalmitis in the Multicenter Boston Type 1 Keratoprosthesis Study is an artifact of an average
1548
follow-up period too short to detect this clinically significant complication. These infections occurred in patients who underwent keratoprosthesis surgery for failed grafts (4 eyes) and aniridia (1 eye). This trend is in contradistinction to prior reports of higher rates of endophthalmitis in eyes with SJS, OCP, and chemical burns.17 In addition, the spectrum of infectious organisms in our study departs from others. Endophthalmitis after keratoprosthesis surgery typically occurs with gram-positive pathogens; thus, our antibiotic prophylaxis has been focused on daily topical vancomycin administration to cover against typical and resistant gram-positive organisms. Bacterial endophthalmitis occurred only with gram-positive bacteria, and only grampositive skin flora were found on surveillance cultures in the study by Nouri et al.17 Chew et al7 reported a prominence of gram-positive species in their cases of endophthalmitis and related this to discontinuation of vancomycin prophylaxis. Aldave et al8 reported no episodes of endophthalmitis but noted a predominance of fungus in corneal infiltrates. However, all of our patients were using vancomycin eye drops (50 mg/ml) at the time of infection, and developed predominantly gram-negative infections. Our study suggests the need for long-term treatment with antibiotics that cover for gram-negative pathogens in addition to gram-positive organisms. Our findings also suggest that vancomycin prophylaxis may somehow encourage selection of gram-negative infections in eyes with a keratoprosthesis, or at least a segment of this population. Four of the 5 patients with endophthalmitis underwent glaucoma surgery before keratoprosthesis placement (3 with GDDs, 1 with trabeculectomy). Given that the conjunctiva had been violated surgically, and hardware was present in several eyes at the time of keratoprosthesis surgery, there may have been a greater risk for infection with virulent organisms that do not typically grow on the ocular surface. Alternatively, the ocular surface flora may be altered directly by chronic exposure to vancomycin in these eyes. Of note, Barnes et al18 showed that chronic vancomycin administration increases the incidence of fungal infections without changing rates of fungal colonization significantly, suggesting that vancomycin itself may alter conjunctival barrier function in keratoprosthesis cases. The question of gram-negative bacteria selection cannot be addressed without further study of the ocular surface flora in this era of chronic vancomycin administration and soft contact lens wear. What remains clear is that there is a lifetime risk for infection because of the presence of hardware in the eye. Corneal melting was generally an inflammatory rather than an infectious complication in our series. Preoperative diagnoses in these 6 cases were mostly typical, including SJS, OCP, and aniridia, which are well-described conditions predisposing to an inhospitable ocular surface environment and poor prognosis after keratoprosthesis implantation.11,19 Rheumatoid arthritis is not widely reported as a subset of patients who undergo keratoprosthesis implantation. With corneal melting as a clinical hallmark of this disease, the poor outcome we report in our 1 patient with this disease also may be unsurprising. What is remarkable about this case (patient 7, Table 4, available at http://aaojournal.org) is the maintenance of 20/100 acuity after recurrent melts and
Greiner et al 䡠 Longer-Term Vision with the Boston KPro multiple keratoprosthesis replacements; vision loss occurred with removal of the keratoprosthesis, which was required for lack of sufficient tissue integrity to support the implant itself. Of the other 3 eyes with corneal melts that achieved but could not maintain 20/200 vision, 2 eyes lost vision because of other causes late in the postoperative course (patient 3 with recurrent retroprosthetic membrane at 56 months, patient 8 with hypotony and choroidal effusions at 23 months), and 1 eye (patient 13) had no significant ocular comorbidities to limit visual recovery if the patient had elected to pursue replacement of the keratoprosthesis. As these cases suggest, corneal melting in our series did lead to significant postoperative morbidity but did not lead directly to vision loss, and when treated did allow extended maintenance of useful vision. Alternatives to keratoprosthesis surgery carry substantial risks and may not be viable options in the context of severe ocular surface disease. Keratolimbal allograft with or without subsequent keratoplasty has been shown to restore the phenotypic corneal epithelium, but progressive and substantial declines in postoperative graft survival and ambulatory vision (ⱖ20/200) after 2 to 3 years have been reported, and outcomes are comparable to those of keratoprosthesis surgery.20,21 Survival of ambulatory vision after keratolimbal allograft was lowest in patients with SJS in the study by Solomon et al,20 which underscores the challenges raised by our study and others11 of performing visual rehabilitation surgery in this patient population. Keratolimbal allograft surgery usually requires chronic systemic immunosuppressive agents to prolong graft survival but does not prevent acute rejection episodes21 and may limit regimen adherence and graft survival because of systemic toxicity. In addition, the preoperative presence of severe limbal stem cell deficiency, conjunctival scarring, an unstable tear film, and factors that limit a patient’s candidacy for systemic immunosuppression make keratoprosthesis surgery a more viable option in certain cases. Improved outcomes after keratolimbal allograft may be achieved with administration of combined immunosuppressive agents22 and delay of penetrating keratoplasty until ocular surface inflammation subsides.20 Infection, glaucoma, and posterior segment complications are significant postoperative risks common to keratolimbal allograft and keratoprosthesis surgery. Other alternatives to keratoprosthesis exist but may be less viable than staged keratolimbal allograft, including keratolimbal autograft, which is contraindicated in bilateral disease and conjunctival grafting techniques, which may not provide sufficient limbal stem cells to support a clear visual axis. The Boston type 1 keratoprosthesis remains an important and viable option for salvaging vision after multiple keratoplasty failures and in patients with a high risk of failure with traditional grafting methods. Good vision can be achieved with keratoprosthesis implantation. However, after achieving an initial improvement in vision, a significant number of our patients lost vision over the extended postoperative course. Glaucoma and complications related to glaucoma surgery, including GDD erosions, continue to represent a significant challenge in achieving and maintaining good long-term vision outcomes in keratoprosthesis surgery. The authors emphasize the need for a team ap-
proach that begins in the preoperative period, including a meticulous evaluation by both glaucoma and retina specialists. Moreover, consideration should be given to surgical prophylaxis of glaucoma before or at the time of keratoprosthesis surgery in an effort to thwart the progression of glaucoma in the long term. In conclusion, there is a significant need for major surgery in the postoperative period to address complications related to the prosthesis, which surgeons and patients must anticipate. Also, with nearly ubiquitous use of vancomycin in the postoperative course, gram-negative and fungal endophthalmitis cases such as those reported in this article may represent a burgeoning problem and warrant a heightened awareness of this devastating complication. With the longest follow-up period reported to date, our study highlights the need for continued long-term follow-up in patients undergoing keratoprosthesis surgery and points to a more guarded long-term visual prognosis after surgery.
References 1. Dohlman CH, Schneider HA, Doane MG. Prosthokeratoplasty. Am J Ophthalmol 1974;77:694 –700. 2. Harissi-Dagher M, Khan BF, Schaumberg DA, Dohlman CH. Importance of nutrition to corneal grafts when used as a carrier of the Boston keratoprosthesis. Cornea 2007;26:564 – 8. 3. Dohlman CH, Dudenhoefer EJ, Khan BF, Morneault S. Protection of the ocular surface after keratoprosthesis surgery: the role of soft contact lenses. CLAO J 2002;28:72– 4. 4. Khan BF, Harissi-Dagher M, Khan DM, Dohlman CH. Advances in Boston keratoprosthesis: enhancing retention and prevention of infection and inflammation. Int Ophthalmol Clin 2007;47:61–71. 5. Durand ML, Dohlman CH. Successful prevention of bacterial endophthalmitis in eyes with the Boston keratoprosthesis. Cornea 2009;28:896 –901. 6. Zerbe BL, Belin MW, Ciolino JB, Boston Type 1 Keratoprosthesis Study Group. Results from the multicenter Boston Type 1 Keratoprosthesis Study. Ophthalmology 2006; 113:1779 – 84. 7. Chew HF, Ayres BD, Hammersmith KM, et al. Boston keratoprosthesis outcomes and complications. Cornea 2009;28: 989 –96. 8. Aldave AJ, Kamal KM, Vo RC, Yu F. The Boston type I keratoprosthesis: improving outcomes and expanding indications. Ophthalmology 2009;116:640 –51. 9. Bradley JC, Hernandez EG, Schwab IR, Mannis MJ. Boston type 1 keratoprosthesis: the University of California Davis experience. Cornea 2009;28:321–7. 10. Dunlap K, Chak G, Aquavella JV, et al. Short-term visual outcomes of Boston type 1 keratoprosthesis implantation. Ophthalmology 2010;117:687–92. 11. Yaghouti F, Nouri M, Abad JC, et al. Keratoprosthesis: preoperative prognostic categories. Cornea 2001;20:19 –23. 12. Akpek EK, Harissi-Dagher M, Petrarca R, et al. Outcomes of Boston keratoprosthesis in aniridia: a retrospective multicenter study. Am J Ophthalmol 2007;144:227–31. 13. Sayegh RR, Ang LP, Foster CS, Dohlman CH. The Boston keratoprosthesis in Stevens-Johnson syndrome. Am J Ophthalmol 2008;145:438 – 44.
1549
Ophthalmology
Volume 118, Number 8, August 2011
14. Aquavella JV, Gearinger MD, Akpek EK, McCormick GJ. Pediatric keratoprosthesis. Ophthalmology 2007;114:989 – 94. 15. Netland PA, Terada H, Dohlman CH. Glaucoma associated with keratoprosthesis. Ophthalmology 1998;105:751–7. 16. Rivier D, Paula JS, Kim E, et al. Glaucoma and keratoprosthesis surgery: role of adjunctive cyclophotocoagulation. J Glaucoma 2009;18:321– 4. 17. Nouri M, Terada H, Alfonso EC, et al. Endophthalmitis after keratoprosthesis: incidence, bacterial causes, and risk factors. Arch Ophthalmol 2001;119:484 –9. 18. Barnes SD, Dohlman CH, Durand ML. Fungal colonization and infection in Boston keratoprosthesis. Cornea 2007;26:9–15.
19. Khan B, Dudenhoefer EJ, Dohlman CH. Keratoprosthesis: an update. Curr Opin Ophthalmol 2001;12:282–7. 20. Solomon A, Ellies P, Anderson DF, et al. Allograft with or without penetrating keratoplasty for total limbal stem cell deficiency. Ophthalmology 2002;109:1159 – 66. 21. Ilari L, Daya SM. Long-term outcomes of keratolimbal allograft for the treatment of severe ocular surface disorders. Ophthalmology 2002;109:1278 – 84. 22. Liang L, Sheha H, Tseng SC. Long-term outcomes of keratolimbal allograft for total limbal stem cell deficiency using combined immunosuppressive agents and correction of ocular surface deficits. Arch Ophthalmol 2009;127: 1428 –34.
Footnotes and Financial Disclosures Originally received: July 24, 2010. Final revision: December 26, 2010. Accepted: December 30, 2010. Available online: March 21, 2011.
Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Manuscript no. 2010-1024.
U.C. Davis Health System Eye Center, University of California at Davis, Sacramento, California.
1550
Correspondence: Mark J. Mannis, MD, Chair, Department of Ophthalmology and Vision Science, U.C. Davis Health System Eye Center, 4860 Y Street, Suite 2400, Sacramento, CA 95817. E-mail: [email protected].