Article Text

other Versions

Download PDFPDF
Corneal perforation from peripheral ulcerative keratopathy in patients with rheumatoid arthritis: epidemiological findings of the British Ophthalmological Surveillance Unit
  1. Hannah Mary Timlin1,
  2. Hildegard Nikki Hall1,
  3. Barny Foot2,
  4. Peter Koay1
  1. 1Ophthalmology Department, Princess Alexandra Eye Pavilion, Edinburgh, UK
  2. 2Royal College of Ophthalmologists, London, UK
  1. Correspondence to Dr Hannah Mary Timlin, Princess Alexandra Eye Pavilion, Edinburgh EH3 9HA, UK; hannahtimlin{at}


Background/Aims This study quantifies the threat to vision and the survival in patients presenting with peripheral ulcerative keratopathy (PUK) corneal perforation associated with rheumatoid arthritis (RA) in the UK.

Methods New cases of corneal perforation from PUK in patients with RA were prospectively collected from the UK via the British Ophthalmological Surveillance Unit from July 2012 to June 2014. An initial questionnaire collected data on presentation and the first 2 weeks’ management, and a follow-up questionnaire collected 1-year data on ocular morbidity and mortality.

Results 30 eyes of 28 patients were identified over 2 years, estimating a UK incidence of 0.234/million/year. 20/27 (74%) were female, with a median age of 68 years (range 41–84). The most common initial management was cyanoacrylate glue with a bandage contact lens, oral steroids, topical and oral antibiotics, and lubricants. Long-term management included corneal grafting in 12/20 (60%) eyes of patients living at 1 year. The 1-year all-cause mortality was 6/25 (24%), which increased to 1/2 (50%) if both eyes had perforated. In the remaining patients alive at 1-year follow-up, there was a 13/20 (65%) poor visual outcome of less than or equal to counting fingers. 8/25 (40%) patients had bilateral PUK, with 2/25 (8%) having bilateral perforation. 5/19 (26%) patients alive at 1-year follow-up were eligible for sight impairment registration.

Conclusion This study highlights the serious ocular morbidity and high mortality associated with corneal perforation from PUK in patients with RA despite treatment. The mortality doubled if both eyes perforated, which should serve as a harbinger of impending serious medical problems.

  • cornea
  • ocular surface
  • epidemiology

Statistics from

Request Permissions

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.


Peripheral ulcerative keratopathy (PUK) is a peripheral thinning of the corneal stroma caused by the release of proinflammatory mediators from corneal limbal vasculitis.1 2 PUK can lead to severe visual loss, corneal perforation, endophthalmitis and even evisceration, with optimal management not yet proven. The majority of these cases are related to rheumatoid arthritis (RA),2 and this study was specifically limited to this cause of PUK to narrow the spectrum sampled. Patients with systemic rheumatoid vasculitis (RV) have a higher mortality rate than patients with RA alone (adjusted HR 1.26),3 and more than 20 years ago RV mortality rate was reported as 36% at 3 years.4 Whether this high mortality rate is translated into those with corneal perforation from conjunctival limbal vasculitis with PUK is unknown.

Management of corneal perforation in these patients can involve surgical procedures involving corneal glue, conjunctival flaps, amniotic membrane grafts or corneal transplants, as well as medical therapy including steroids, immunosuppressants and antibiotics. Reported visual outcomes in these patients are highly variable, ranging from 6/9 vision to no perception of light (NPL).1 However, recent improvement in access to immunosuppressants including biologics may have improved visual outcomes in more recent years.

This study aims to collect data on the UK incidence, current ophthalmic medical and surgical management in the UK, 1-year visual outcomes and 1-year mortality of these patients. This understanding is necessary to guide our discussion with patients and in particular for prognosis. As these conditions are very rare and are often managed locally by the attending ophthalmologist, very few ophthalmologists will have the experience of managing many cases. While this study is limited in scope, it provides insight into the management of such cases and the outcomes in the UK.

Materials and methods

Case ascertainment

New cases of corneal perforation from PUK in patients with RA were prospectively collected from the UK through active case ascertainment via the British Ophthalmological Surveillance Unit (BOSU).5 BOSU sends reporting cards, listing the conditions being studied, to all ophthalmology consultants in the UK monthly, who return cards with a tick to indicate that a new case was seen or that there were no cases to report. Corneal perforation from PUK in patients with RA was on the BOSU reporting card for 2 years from July 2012 to June 2014.

Data collection

Initial questionnaires (online supplementary appendix 1) were sent to reporters regarding patient demographics, presentation and initial management within the first 2 weeks. Follow-up questionnaires (online supplementary appendix 2) were sent 12 months later to collect medical and surgical management and visual outcomes. Outcomes included whether the globe was retained, was intact, how its integrity was maintained, Snellen visual acuity, factors causing poor visual acuity, sight impairment registration and mortality, and were collected from the data documented at the first clinic appointment after 1 year from perforation. Reporters who did not return questionnaires were sent reminders. Researchers’ queries were communicated with the reporter via telephone, letter or email.

Supplementary file 1

Supplementary file 2

Statistical analysis

Statistical analysis was performed using Excel software (Microsoft Excel, V.2013).

UK population figures

UK incidence calculation used the Office for National Statistics UK population of 64.1 million in mid-2013.6

Subgroup analysis

Subgroups were analysed and compared with successful outcomes, being that the patient was alive at 1 year and having vision of 6/60 or better in the affected eye.


Thirty eyes of 28 patients were reported and confirmed as genuine cases. Two patients had sequential bilateral disease within the 2-year reporting period. Initial questionnaire data were available on 29 cases (29 eyes of 27 patients), as one patient’s notes were lost after being reported. One-year follow-up questionnaires were returned for 27 cases (27 eyes of 25 patients). These were not completed for two cases because one patient did not attend further appointments and the other was transferred to another hospital, from where there was no questionnaire response. Six patients (seven eyes) died over the 1-year follow-up period, leaving complete 1-year data available in 20 cases (20 eyes of 19 patients).

Reports excluded

Sixty-one reporting cards were returned, but 31 were excluded. Fourteen reporters did not respond to questionnaires or reminders, but four had colleagues in the same department who had reported cases within 2 months. Eight reports were duplicates, three ophthalmologists ‘ticked the wrong box’, two patients had a descemetocoele without perforation, two patients did not have RA but osteoarthritis and rosacea, and two patients presented outside the reporting period.

Patient characteristics

Twenty of the 27 (74%) patients were female and seven (26%) were male. Twenty-six of the 27 (96%) patients were of white British ethnicity and 1 (4%) was of black Caribbean ethnicity. The median age was 68 years (range 41–84). Sixteen of 29 (55%) were left eyes and 13 (45%) were right eyes.

Pre-existing ocular conditions

Eighteen of 29 (62%) eyes had an ocular copathology. This consisted of 7/29 (24%) cataracts, 4/29 (14%) age-related macular degeneration, 4/29 (14%) dry eyes, 3/29 (10%) glaucoma, 2/29 (7%) amblyopia, 2/29 (7%) uveitis, 2/29 (7%) previous herpes simplex keratitis, 2/29 (7%) corneal graft, 1/29 (3%) previous polymicrobial keratitis, 1/29 (3%) previous PUK, and 1/29 (3%) had undergone recent cataract surgery which was thought to have precipitated the PUK.

Size of perforation

Thirteen of 29 (45%) perforations were less than 1 mm in size, 14/29 (48%) were between 1 and 3 mm, and 2/29 (7%) were larger than 3 mm.

Inflammatory markers

At the time of perforation, the erythrocyte sedimentation rate was measured in 11 cases, which was elevated in 3/11 (27%), and the C reactive protein was measured in 14 cases, which was elevated in 5/14 (36%).

Medical management within the first 2 weeks


Twenty-eight of 29 (97%) cases were started on topical antibiotics and 15/29 (52%) received oral antibiotics.


Seven of 29 (24%) cases received intravenous steroid, 17/29 (59%) received oral steroids without intravenous treatment, 3/29 (10%) received topical steroids alone (none of whom were on immunosuppression) and 3/29 (10%) received no systemic or topical steroids (all of whom were already on systemic immunosuppression).


Eighteen of 29 (62%) eyes were already on topical lubricants prior to perforation, which increased to 22/29 (76%) after diagnosis.


Fifteen of 29 (52%) cases were already on immunosuppressants prior to perforation. A further 2/29 (7%) of cases were started on immunosuppressants within the first 2 weeks.


Four of 29 (14%) cases were already on biologics prior to perforation, which did not increase during the first 2 weeks.

Surgical management within the first 2 weeks

Fifteen of 29 (52%) eyes had initial surgical treatment with cyanoacrylate glue and bandage contact lens (BCL). Six of 29 (21%) eyes were managed with a BCL alone. Two of 29 (7%) eyes received treatment with a primary corneal graft. Two of 29 (7%) eyes were treated with fibrin glue alone. Two of 29 (7%) eyes had no procedure or BCL.

Medical management at 1 year

By 1 year, 10/19 (53%) of patients were on systemic immunosuppressants, 3/19 (16%) on biologics and 9/19 (47%) on oral steroids. Six of 19 (32%) patients were on no systemic steroid, immunosuppression or biologics at 1 year.

Surgical management at 1 year

Cyanoacrylate glueing was repeated up to a maximum of three times. However, by 1-year follow-up, 12/20 (60%) of the eyes in living patients had undergone a corneal graft.

Bilateral perforation and PUK

Two patients (8%) had both eyes perforate within the reporting 2-year period: one 6 months apart and the other 6 days apart. A further eight (32%) patients had corneal thinning of their fellow eye.

Mortality at 1 year

There was a 6/25 (24%) all-cause mortality within the 1 year following perforation. The patient who had perforation in both eyes within 6 days died 3 weeks later, aged 83 years. However, the other bilateral patient remained alive at 1 year.

Patients who are deceased versus alive at 1 year were similar in the following: age (mean 67 years, range 41–84 vs mean 67 years, range 47–83), majority were female (5/6 (83%) vs 13/19 (68%)), and rate of systemic treatment with immunosuppressants/biologics preperforation (5/6 (83%) vs 16/19 (84%)), at 2 weeks (5/6 (83%) vs 15/19 (79%)) or at any point within the first year (5/6 (83%) vs 16/19 (83%));  however, they had a lower rate of systemic comorbidity (1/6 (17%) vs 9/21 (43%) cardiovascular disease, 0/6 vs 4/21 (19%) diabetes, 0/6 vs 3/21 (14%) malignancy) and a lower rate of biologic use (0/6 vs 4/19 (21%)).

Ocular morbidity at 1 year

In the remaining patients alive at 1-year follow-up, 2/20 (10%) of the eyes were enucleated, 1/20 (5%) had NPL vision, 10/20 (50%) had vision between counting fingers (CF) and perception of light, and 7/20 (35%) had vision between 6/12 and 6/36 in the affected eye.

The causes of poor vision were reported to be corneal scarring in 9/20 (45%) of the eyes, cataract in 5/20 (25%), astigmatism in 3/20 (15%), enucleation in 2/20 (10%), amblyopia in 2/20 (10%), glue in 1/20 (5%), dystrophic epithelium in 1/20 (5%), aphakia in 1/20 (5%) and anterior ischaemic optic neuropathy from giant cell arteritis in 1/20 (5%).

The integrity of the globe was maintained by a penetrating keratoplasty in 5/20 (25%), tectonic corneal graft in 4/20 (20%), glue in 3/20 (15%), corneal scar tissue in 3/20 (15%) and iris plugging in 3/20 (15%). In 2/20 (10%) of cases, the globe integrity was not maintained by 1 year and the globe had been enucleated.

Patients grouped into poor 1-year vision of worse than 6/60 versus better or equal to 6/60 were similar in their rate of immunosuppression/biologics preperforation (9/13 (69%) vs 4/7 (57%)) at 2 weeks (10/13 (77%) vs 5/7 (71%)) and within the first year (10/13 (77%) vs 5/13 (71%)). However, they had a greater rate of diabetes (4/13 (31%) vs 0/7), biologic use (4/13 (31%) vs 0/7), perforations greater than 3 mm in size (2/13 (15%) vs 0/7), and corneal graft within the first 2 weeks (5/13 (38%) vs 0/7) and within the first year (9/13 (69%) vs 3/7 (43%)).

Sight impairment registration at 1 year

Five of 19 (26%) patients living were eligible for sight impairment registration by 1 year, with 1/19 (5%) for severe sight impairment registration and 4/19 (21%) for sight impairment registration.


Summary of findings

In this study, patients with RA with corneal perforation from PUK had a 24% 1-year all-cause mortality rate and a 65% poor visual outcome of less than or equal to CF.


The 1-year all-cause mortality was 24%, which increased to 50% if both eyes perforated. Although there are no other publications of survival in patients with RA who have perforated their corneas from PUK specifically, Stylianides et al7 reported a 58% 5-year mortality in 117 patients with RA-associated corneal ulceration receiving penetrating keratoplasty (from the time of transplant). Additionally, more than 20 years ago, RV mortality rates were reported as 36% at 3 years, 43% at 5 years and 55% at 10 years.4 Voskuyl et al reported the risk of death of patients with RV to be 1.26 times that of patients with RA diagnosed in 1990–1992. Their Kaplan-Meier survival curve shows an approximate 10% 1-year and 33% 5-year mortality rate in patients with RV compared with 4% and 23% in patients with RA.3 Side effects of medication may contribute to this increased risk of death, as they do in patients with RA who have a 1.97 adjusted HR for all-cause mortality with glucocorticoid exposure.8

A single-centre UK study from Bristol covering 2002–2012 found a 15% mortality among their cohort of patients with RA with PUK (n=46) over a mean of 4.6 years follow-up.9 Of this population, only seven had corneal perforation, compared with all of the patients in our study, so the higher mortality of 24% in our group may reflect more severe RV or more poorly controlled disease. These authors noted a trend towards a decrease from the 28% mortality in an earlier study in the same centre from 1987 to 2002 with a mean of 3-year follow-up,10 although no longer statistically significant after Bonferroni correction. This trend may reflect more aggressive systemic management of RA over the later years.

Unfortunately, our present study did not collect the causes of death from reporting clinicians as this information is rarely shared with ophthalmologists. However, in patients with RA the Health Improvement Network has shown greater rates of death from cardiovascular causes, respiratory causes, malignancy, infection and suicide than population controls.11

Visual morbidity

After excluding the quarter of patients who died during the 1-year follow-up period, approximately a quarter (26%) of the remaining patients were eligible for sight impairment or severe sight impairment registration. Additionally, 65% had ‘poor’ vision (less than or equal to CF), no vision or loss of the globe. This severity of visual outcome is confirmed by Saripalli et al12 who reported that 63.6% (21/33) of the eyes with corneal melt (with and without perforation) had vision worse than 6/60,12 and Pleyer et al13 who reported that 69% of patients with corneal melt and penetrating keratoplasty had vision worse than 6/60.

Saripalli et al12 also showed that 57% of patients with RA corneal melt had bilateral disease with follow-up between 3 months and 14 years,12 which is slightly greater than our 40%, probably reflecting our shorter 1-year follow-up.

Most common management

The most common initial management of corneal perforation in these patients was commencement of topical and oral antibiotics, lubricants, oral steroids and cyanoacrylate glue with a BCL. Additionally, long-term management included a corneal graft in 60% of patients within the first year, in comparison with 100% of cases in Bristol, UK, from 2002 to 2012.9

Interestingly, no cases received treatment with an amniotic membrane graft. Amniotic membranes, especially multilayer amniotic membranes, can be useful14 before perforation to protect the corneal that is melting, but when the cornea perforates the use of amniotic membranes becomes limited. This has been borne out in management choices of the surgeons surveyed, where none of the surgeons chose to repair the perforation with amniotic membrane. Furthermore, no cases were reportedly treated with corneolimbal grafts, as described recently by Livny et al.15

More than half of patients were already on immunosuppressants or biologics at the time of diagnosis. Unfortunately, data were not collected on whether the patients were on this due to their RA severity or for ocular indications. There was little change in the number of patients on immunosuppression or biologics after identification of a corneal perforation. However, we did not collect data on the number of immunosuppressants or doses, which may have been increased.

Incidence in the UK population

Over the 2-year data collection period, 30 true cases of corneal perforation from PUK in patients with RA were reported. This gives a UK incidence of 0.234/million/year. Although no other incidence rates of this condition have been published, it is lower than estimates extrapolated from previous data.1 12 16

Comparison with other published incidences

A retrospective study in Sheffield, UK, from 1996 to 19981 described five patients requiring ‘emergency corneal surgery to prevent corneal melt’ from RA. One case progressed to perforation, one did not, and three patients were not described to have perforated or not. Based on a Sheffield city population of 552 698 in 2011,17 their incidence of corneal perforation from PUK in patients with RA is 0.603–2.41/million/year. A study in Moorfields, London and Addenbrooke’s, Cambridge, UK, over a 14-year period from 1980 to 1994 reported 32 cases of corneal perforation from RA.18 However, the source population is difficult to estimate due to the tertiary referral nature of these hospitals and extensive overlap of areas with nearby hospitals. McKibbin et al reported a prospective study in Yorkshire from 1995 to 1997, where 21 patients with RA in an estimated population of 2 993 60018 presented with corneal melting,16 giving an incidence of 2.34/million/year for corneal melt. Unfortunately, the proportion of melts progressing to perforation is unknown, explaining their 10-fold higher incidence than in our study. A study from Danville, Pennsylvania, USA, from 1976 to 200212 reported 21 cases of corneal melt from RA. Again, their RA perforation rate is not reported. However, they do report perforation in 11 out of 27 patients with corneal melt from any underlying systemic rheumatic disease. With a population of 43 059 in Danville City in 2010,19 their estimated incidence of corneal melt from RA is 18.8/million/year and for corneal perforation from all systemic rheumatological diseases is 9.82/million/year. Overall, these incidences may be overestimations, affected by the drift of cases into a hospital’s catchment population, leading to an underestimation of the denominator. This is not an issue in our UK-wide study.

This study reports a minimum incidence and there is likely under-reporting. Possible causes for this include patients not seeing an ophthalmologist, and ophthalmologists not returning the reporting card, not recalling the case when returning the report card or not remembering the patient’s details when receiving the questionnaire. There were 10 reporters who did not respond to questionnaires or reminders, and who did not have a colleague from the same hospital report within 2 months. Potentially, if these were missed true cases, this would have increased the incidence up to 0.312/million/year. Even taking this into account, the incidence is lower than other extrapolated estimates.1 12 16 As well as errors in source population size and under-reporting, other potential explanations for this discrepancy include advances in medical treatment, such as greater availability of immunosuppression or biologics, or earlier detection of PUK signs with initiation of treatment, reducing the rate of progression from melt to perforation.


This study shows a 2.86 times greater incidence of PUK perforation in women than men. This is in keeping with the UK gender incidence of RA being 2.64 times greater in women than men.20

Risk factors for poor outcomes

Bilateral perforation

Mortality rates increased from 24% to 50% once perforation occurred in the fellow eye, suggesting that urgent systemic review by a rheumatologist is required in bilateral cases, with close monitoring of immunosuppression and general health.

Immunosuppressants and biologics

Out of the five patients who did not receive any immunosuppressants or biologics within the first 2 weeks, 1/5 (20%) was deceased at 1 year. Sadly, this patient suffered further ocular vasculitis in the form of giant cell arteritis causing anterior ischaemic optic neuropathy leading to NPL. Similarly, 5/20 (25%) of patients receiving immunosuppressants or biologics had died at 1-year review. However, there were more patients in the 1-year survival group on biologics (4/19) specifically than the deceased group (0/6), suggesting their beneficial effect on survival in these patients. Unfortunately this beneficial effect of biologics did not reflect in the visual outcomes, with all patients on biologics having poor visual outcomes.

Size of perforation

Both the patients with a perforation larger than 3 mm had poor visual outcomes of enucleation and CF vision, suggesting this to be a poor predictive factor.

Rate of corneal graft

There was a greater rate of corneal graft in the poor visual outcome group than the good visual outcome group within the first 2 weeks (38% vs 0%) and by 1 year (69% vs 43%), perhaps reflecting the greater sizes of the perforation defects.


Diabetes was more common in the poor visual outcome group. However, diabetic retinal disease was not documented as contributing to their poor vision.


As discussed, the main limitation of this observational prospective study was the likelihood of under-reporting of cases. BOSU has published a previous good participation rate of ophthalmologists, with a reporting scheme of 68%.5 Unfortunately, although investigated, an alternative method of estimating case ascertainment was not identified for comparison with this study’s capture rate. Other BOSU studies have reported between 72% and 95% case ascertainment rates.5


This observational study highlights the serious ocular morbidity and systemic mortality of developing a corneal perforation from PUK in patients with RA. Future studies could ask for reporting of patients with PUK from RA without perforation in order to identify risk factors for progression from corneal thinning to perforation.


A huge thank you to the UK ophthalmologists who reported cases to this study, including Dr Anna Maimo, Christopher Scott, Alan Fitt, Aravind Reddy, John Dart, Amit Gaur, Jacobus Pauw, James Kersey, Shadrokh Nabili, Bruce Allan, Roger Ellingham, Gerard Ainsworth, Simon Hardman-Lea, Athanassius Tadros, Anju Kadyan, Peter McDonnell, Paul Tesha, Nabil Habib, Gerry McGinnity, Sarah Anderson, Simon Hewick, Matthew Raynor, Stephen Vardy, Robert Murray, Frank Larkin, Louise Downey, Martha Ramirez-Florez, Christine Ellerton, Miles Stanford, Clive Edelsten, Christopher Jenkins, Mark Watts, Jan Van Der Hoek, Achim Nestel, Nicola Topping, Keith Davey, Ewan Craig, Parwez Hossain, Mahesh Ramchandani, Chris Hemmerdinger and Dipak Parmar.



  • Contributors Conception or design of the work: HMT, BF, PK. Data collection: HMT, HNH. Data analysis and interpretation: HMT, HNH. Drafting the article: HMT. Critical revision of the article: HMT, HNH, BF, PK. Final approval of the version to be published: HMT, HNH, BF, PK.

  • Funding This work was supported by the Royal College of Ophthalmologists Ross Award Bursary.

  • Competing interests None declared.

  • Ethics approval NHS Lothian provided full REC support. Ethical approval was sought but deemed not required due to the observational nature and absence of patient identifiable data collected.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • At a glance
    Keith Barton James Chodosh Jost B Jonas