Aim To evaluate the outcomes of the type-I and type-II Boston keratoprostheses in a single Irish centre.
Methods A retrospective chart review of keratoprosthesis implantations carried out in our institution from November 2002 to March 2014 was performed. All procedures were performed by a single surgeon (WP).
Results Thirty-four keratoprosthesis implantations were carried out in 31 patients with a mean follow-up of 42±31 months (range 2–110 months). Seventeen patients were female (54.8%) and 14 were male (45.2%). The majority of keratoprostheses implanted were type-I (31/34, 91.2%), and three were type-II (3/34, 8.8%). Twenty-nine patients (85.3%) had an improvement in distance best-corrected visual acuity (BCVA) from baseline. Fifty per cent (17/34) of patients had a best-ever BCVA of at least 6/12. Eighteen patients (64.3%) retained a BCVA of at least 6/60 at 1 year. Over the course of follow-up, six keratoprostheses were explanted from six eyes of five patients, one of which was a type-II keratoprosthesis. Twenty-six patients (76.5%) developed postoperative complications. Complications included retroprosthetic membrane (18 patients, 52.9%), an exacerbation or new diagnosis of glaucoma (6 patients, 17.6%), endophthalmitis (5 patients, 14.7%) and retinal detachment (2 patients, 5.9%).
Conclusions These data demonstrate excellent visual acuity and retention outcomes in a cohort with a long follow-up period in a single centre. Complications remain a considerable source of morbidity. These outcomes provide further evidence for the long-term stability of type-I and type-II Boston keratoprostheses in the management of patients in whom a traditional graft is likely to fail.
- Treatment Surgery
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
The Boston keratoprosthesis is an implanted corneal device that can restore vision in patients with corneal blindness, which may otherwise be untreatable. The Boston type-I device is the most commonly used corneal prosthesis worldwide.1 The type-I device is implanted within a donor corneal transplant but does not depend on the donor graft for optical clarity. The most common indications for the type-I device are corneal graft failure and chemical injury.1 The type-II device is modified with an elongated optic designed to protrude through a complete tarsorrhaphy. The type-II device is used where the pathology makes type-I device survival unlikely, most commonly ocular cicatricial pemphigoid and Stevens-Johnson syndrome (SJS).2
Cost-effectiveness of these keratoprostheses has been demonstrated, and they result in significant improvements in functional ability and mental health.3–5 The type-I device was licensed in 1992 by the US Food and Drug Administration. Subsequently, several design advancements have improved clinical outcomes.6 Internationally, more than 1000 devices were implanted yearly from 2009 to 2013 and more than 8000 devices have been implanted to date.7 European approval (Conformitée Européenne mark) was granted in 2014 and this will likely expand the use of the device.8
This study's purpose was to retrospectively assess all patients who underwent Boston keratoprosthesis surgery in an Irish centre.
This study is a retrospective case series of patients who underwent Boston type-I or type-II keratoprosthesis insertion between November 2002 and March 2014 at our institution. One surgeon (WP) performed all procedures. The tenets of the Declaration of Helsinki were honoured.
Patients were selected as type-I keratoprosthesis candidates if they were deemed unlikely to maintain a successful standard penetrating keratoplasty, with or without a history of previous failed corneal grafts. A type-II device was placed where advanced keratinisation of the ocular surface precluded placement of a type-I device. Elderly patients unsuitable for a penetrating keratoplasty due to reduced life expectancy were also considered. The surgery was performed in the standard manner described by Dohlman et al.9
Patients were reviewed on the first postoperative day, at 1 week, 1 month, 3 months and every 2–3 months thereafter. Initial treatment included topical prednisolone acetate (1%) every 2 h and chloramphenicol (0.5%) four times a day. Chloramphenicol was tapered to twice daily and the steroid used either once or twice daily in the long term. Some patients’ maintenance antibiotic was changed from chloramphenicol (0.5%) to vancomycin (0.5%) if suspicious signs of an infectious process, such as conjunctivitis, developed. For type-I devices, bandage contact lenses (PureVision2, Bausch and Lomb, Rochester, USA) were used and replaced every 1–3 months. Glaucoma medications were administered based on tactile estimations of intraocular pressure (IOP), optic disc changes or visual field deterioration. Further interventions were instigated as required.
Patient age, gender, surgery date, previous ocular surgeries, ocular comorbidities and surgical indication were obtained from the clinical records. Intraoperative information noted included type of keratoprosthesis, size of corneal trephine and donor button and concurrent surgeries. Initial postoperative best-corrected visual acuity (BCVA), 6-weekly (±2 weeks) and 6-monthly (±2 months) BCVA and manifest refraction were recorded. Follow-up duration was defined as the length of time between keratoprosthesis implantation and device failure, death of patient or time of data collection. Complications and procedures following implantation were documented. Primary outcome measures included BCVA, retention of the device and complications. Device removal or extrusion was defined as retention failure.
Kaplan-Meier survival curves were used to assess duration of device retention and similar curves were used to illustrate how visual acuity of at least 6/60 and 6/12 was maintained over the follow-up period. Data was compiled in Microsoft Excel and statistical analysis was performed using PrismStats V.4.0 (GraphPad Software).
Thirty-four keratoprostheses were implanted in 31 patients; three patients underwent bilateral surgery. Mean follow-up was 42±31 months (range 2–110). Ten patients (29.4%) died with a keratoprosthesis in situ, with average retention of 36±30 months before death (range 2–90).
There were 14 men and 17 women with a mean age of 69±15 years (range 37–96). Thirty-one type-I and three type-II keratoprostheses were implanted. Twenty-three implantations represented the primary vision-restoring procedure and 11 followed one or multiple corneal grafts (table 1). The patients’ original diagnoses are listed in table 2.
Comorbidities, the most common of which was glaucoma, are listed in table 3. Preoperative BCVA ranged from 6/36 to perception of light (PL). Twenty-seven patients (79.4%) had BCVA of counting fingers or worse. Contralateral BCVA ranged from 6/5 to no PL, including three prosthetic eyes.
Devices were obtained from the Massachusetts Eye and Ear Infirmary (Boston, Massachusetts, USA). Most implants had 8.50 mm polymethylmethacrylate back plates, while one had a titanium back plate. Recipient corneal trephines ranged from 8.00 mm to 9.00 mm and the allograft donor button from 8.25 mm to 9.50 mm, with 0.25–0.50 mm overlap. Six (17.6%) keratoprostheses were pseudophakic and 28 (82.4%) aphakic. No intraoperative complications were noted and no concomitant glaucoma procedures were performed.
Six weeks postoperatively, 28 eyes (82.4%) had improved visual acuity. Of these, 25 (73.5%) reached BCVA of at least 6/60, 21 (61.8%) achieved at least 6/48, 18 (52.9%) had at least 6/30 and 11 (32.4%) achieved at least 6/12. There were six (17.6%) patients that had not improved at 6 weeks; two had end-stage glaucoma, one of which was related to iridocorneal epithelial syndrome, two patients had had multiple retinal surgeries, one of which had a history of Stickler's syndrome, one patient had macular scarring associated with high myopia and the final patient experienced postoperative corneal melt but subsequently improved when this resolved.
Subjective refraction ranged from −22.00 to +3.00 dioptres sphere. The patient with −22.00 dioptres sphere had an aphakic keratoprosthesis erroneously placed in a pseudophakic eye. The majority of patients were myopic with a mean refraction of −3.50±2.8 dioptres sphere, excluding the outlier. Two patients were hyperopic and both measured +3.00 dioptres sphere. The patient with the −22.00 dioptres sphere refractive error was fitted with a −14.00 dioptres sphere contact lens and achieved BCVA of 6/12.
The majority of eyes (19/34; 55.9%) reached their best-ever BCVA by 6 weeks postoperatively. Time to best-ever vision ranged from 1 day to 30 months (mean 6±8 months). Best-ever vision was maintained for an average of 25±24 months (range 2–78). Best-ever BCVA of at least 6/60 was achieved in 26 (76.5%) eyes. Four of eight eyes (50%) with at least 6 years of follow-up have retained their best-ever vision (range 6/7.5–6/12) for more than 6 years. Eighteen patients (18/28, 64.3%) retained BCVA of at least 6/60 at 1 year.
At the last follow-up, 16 (47%) patients maintained BCVA of at least 6/60. In patients with longer follow-up, 70% maintained at least 6/60 at 5 years (7/10 patients) and 62.5% at 6 years (5/8 patients). Two patients had orbital prostheses at last follow-up, having undergone evisceration at 48 months and 84 months. In the six eyes that ultimately lost PL including two that were eviscerated, the cause of visual deterioration included two cases of endophthalmitis, one retinal detachment, one case of corneal necrosis and extrusion, one phthisical eye due to sympathetic ophthalmia and one eye with Stickler's syndrome that was found to have advanced glaucomatous optic neuropathy immediately after keratoprosthesis implantation which progressed after surgery.
Of the 34 keratoprostheses implanted, 6 were removed from six eyes of five patients, indicating an 82.4% survival rate with average follow-up of 42±31 months (range 2–110). There were five type-I and one type-II devices removed. One of the type-I devices, explanted due to a Pseudomonas aeruginosa abscess, was cleaned and again implanted in a new donor button. This patient was excluded from the long-term visual acuity data set, due to the device's failure, but he maintained BCVA of 6/12 at last follow-up, 72 months later. The remaining four type-I device explantations were due to two cases of persistent leak with hypotony and two cases of corneal necrosis. They were managed by undertaking penetrating keratoplasty and allowing the graft to epithelise. One of these patients’ eyes failed to re-epithelise and was ultimately eviscerated. The type-II device, which was explanted due to skin retraction and persistent leak, was also cleaned and implanted again within a new corneal graft. This revision was maintained for a further 30 months when an episode of endophthalmitis resulted in device extrusion and the eye was eviscerated. This patient previously had a failed type-I device in the contralateral eye. The mean time to device failure was 26±19 months (range 2–48).
The most common complication was retroprosthetic membrane (RPM) formation, affecting 18 eyes (52.9%). This was treated with Nd:YAG laser membranotomy in 17 cases; an average of 2.7±2.6 procedures (range 1–11) were required. One patient developed a thick membrane following endophthalmitis requiring surgical membranectomy. Orbital floor triamcinolone was used to treat low-grade inflammation including anterior uveitis and mild vitritis, in an attempt to prevent RPM formation in 21 patients (61.7%) with a mean of 1.5±1.1 injections (range 1–4). Two patients with pseudophakic eyes (5.9%) underwent Nd:YAG posterior capsulotomy.
A new diagnosis or exacerbation of glaucoma occurred in six eyes (17.6%), two of which had known preoperative glaucoma. Topical treatment was sufficient in four cases and two patients underwent Ahmed-valve insertion; the glaucoma surgery rate was 5.9%. At last follow-up, four patients needed three topical IOP-lowering agents, one was prescribed latanoprost alone and the remaining patients’ type-I keratoprosthesis had failed.
Other complications included corneal melt (5/34 patients; 14.7%), sterile vitritis (4/34 patients; 11.8%) and endophthalmitis (5/34 patients; 14.7%). Of the five endophthalmitis cases, only one patient grew an organism from vitreous culture, Staphylococcus cohnii. Posterior-segment complications included retinal detachment (2/34 patients; 5.9%), hypotony (2/34 patients; 5.9%), neovascular age-related macular degeneration (2/34 patients; 5.9%), vitreous haemorrhage (1/34 patients, 2.9%) and epiretinal membrane (1/34 patients; 2.9%). With regards to the two retinal detachments, one patient underwent retinal detachment repair with the device in situ but ultimately developed inoperable re-detachment. The other patient's detachment was deemed inoperable. Both of the patients with neovascular age-related macular degeneration had intravitreal bevacizumab. Two patients with type-II keratoprosthesis, who developed skin retraction, underwent surgical revision. Three patients developed Charles Bonnet syndrome.
This paper reports the results of 34 type-I and type-II keratoprosthesis procedures in 31 Irish patients. Classically, the Boston keratoprosthesis is placed in cases of repeated graft failure or where failure is predicted, but indications are slowly expanding due to demonstrated efficacy. A recent study proposes the use of keratoprosthesis as an alternative to corneal graft in the management of active infectious keratitis in select cases.10 In this centre, keratoprosthesis has also been used to treat corneal disease in older patients for whom rapid visual recovery is required.11
A seminal article by Yaghouti et al12 describes outcomes of 63 patients with type-I keratoprosthesis due to various underlying pathologies. SJS carried the worst prognosis with 33% (n=7) of eyes maintaining a visual acuity of at least 6/60 after 2 years, while 64% (n=17) of patients with chemical burn and 72% (n=20) of patients with ocular cicatricial pemphigoid maintained at least 6/60 vision; patients with non-cicatricial disease (n=19) had the best outcomes, with 83% maintaining at least 6/60 vision. Other studies have also shown that non-autoimmune aetiologies fare better than those with autoimmune disorders.13 ,14 In our series, non-autoimmune cases fared slightly worse with 58.8% maintaining vision of at least 6/60 at 2 years versus 75% of those with autoimmune diagnoses; this probably reflects the small sample size.
Due to its poor prognosis, Zerbe et al15 advise against using type-I keratoprosthesis in SJS, opting in favour of type-II devices. We have two cases of SJS; one patient has a type-II keratoprosthesis and maintains vision of 6/15 after 66 months and the second patient has a type-I keratoprosthesis and maintains vision of 6/18 after 12 months. Two small series have reported two further patients with SJS who fared well with type-I keratoprosthesis.16 ,17 With careful selection, patients with SJS can do well with type-I devices.
Visual outcomes and retention
The majority of patients had improved vision over the course of this study (29/34 patients, 85.3%). Maintenance of BCVA of at least 6/60 over time was impressive, with the percentage of patients having this level of acuity at 1 year, 5 years and 6 years equal to 64.3% (n=28), 70% (n=10) and 62.5% (n=8), respectively (figure 1). Our 1-year figures are comparable to Zerbe et al15 where 56% (n=62) of patients had at least 6/60 vision and Aldave et al1 who describe an international series of 113 patients and a local Californian series of 110 patients where maintenance of at least 6/60 at 1 year was 68% and 63%, respectively. Our follow-up is longer than most in the literature with an average of 42 months (range 2–110 months).
Retention in this series was in keeping with other studies (figure 2); the retention rate of 82.4% with 42 months average follow-up measures favourably with other long-term studies such as Greiner et al who had a retention rate of 80.5% (n=40) after 34 months average follow-up, and Aldave et al who had a retention rate of 80% (n=223) at 24 months average follow-up.1 ,18 We had only three type-II devices, one of which failed after 48 months. Type-II device failure rates are high; the largest series to date of 29 patients reports a retention failure rate of 41.4% with 45 month average follow-up.2
Similar to previous reports, the most frequent complication in our series was RPM, encountered in 52.9% of eyes. Rates as low as 14% and as high as 77% have been reported previously.1 ,19 Their aetiology is uncertain, but may be due to corneal stromal downgrowth and metaplastic lens epithelium migration.20 Nd:YAG laser membranotomy is a commonly performed procedure postoperatively with rates between 9.9% and 25% previously reported; it was performed in 50% of eyes in this series.1 ,18
Glaucoma is a significant problem in patients with keratoprosthesis and is frequently the reason visual acuity does not improve postoperatively.21 In our series, 17.6% of patients experienced elevated IOP postoperatively, with two requiring glaucoma surgery (5.9%). Other series describe rates of raised IOP of 8.7–50%.2 ,18 It has been proposed by Kamyar et al19 that aggressive surgical preoperative or perioperative management of known glaucoma in patients with keratoprosthesis can reduce glaucomatous morbidity at a later stage. This is particularly meaningful given the extended lifetime of modern keratoprosthesis and the difficulties posed by the monitoring of intraocular pressure once the device is in situ.
Our rate of retinal detachment in type-I patients (2/31 patients, 6.5%) is in keeping with previous reports (0–18.6%).7 ,17 One patient's detachment was operable but ultimately detached again. There were no retinal detachments in the type-II group although a rate of detachment of 27.6% has been described by Pujari et al.2
There were five cases of endophthalmitis (5/34 patients, 14.7%). This was high relative to most series but similar to the 15.5% rate reported by Srikumaran et al7 in their series of 139 patients. Of these five patients, four went on to have keratoprosthesis failure. Only one patient's culture was positive and grew Staphylococcus cohnii, a relatively rare coagulase-negative staphylococci, which are the most common cause of postoperative endophthalmitis, and often exist as lid commensal organisms.22 Behlau et al,23 in their thorough worldwide analysis of endophthalmitis in Boston keratoprosthesis between 1990 and 2010, found that the majority of cases were attributed to Gram-positive organisms, but advised broad-spectrum antibiotic cover at all times to anticipate those few cases caused by Gram-negative organisms. Some of our patients were treated with vancomycin alone postoperatively and this may have contributed to the higher incidence of endophthalmitis in this series as vancomycin does not provide Gram-negative coverage.
This study provides long-term, real-life data in a series performed by a single surgeon, which is significant, as more independent surgeons take up this practice worldwide. The study highlights the ability of keratoprostheses to provide excellent vision for long periods in those who would otherwise be untreatable. However, complications can occur at any time, and tend to increase cumulatively as time passes.14 ,24 These data contribute to the evidence of efficacy and long-term stability of type-I and type-II Boston keratoprostheses.
The authors thank Mr James Byrne, Medical Records Department, Royal Victoria Eye and Ear Hospital, for his contribution.
Contributors ESD and SND prepared the manuscript. All authors reviewed the manuscript for important intellectual content. WP supervised the preparation of the manuscript and revised successive drafts of the manuscript prior to publication.
Competing interests None declared.
Ethics approval Royal Victoria Eye and Ear Hospital Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.