Recurrent Bacterial Keratitis

Abstract

Purpose.: To investigate the rate of recurrent bacterial keratitis, associated bacteria, and surgical intervention.

Methods.: Patients with suspected bacterial keratitis were identified from microbiological requests over a 16-year period between 1995 and 2010. Recurrences and number of surgical interventions were analyzed according to bacterial type.

Results.: A total of 2418 patients were included, of whom 2124 (87.84%) had only one episode of keratitis, 294 (12.15%) at least two, 88 (3.63%) at least three, 40 (1.65%) at least four, and 22 (0.91%) five or more episodes. The bacterial isolation rate was 35.74% (SD 9.41%), increasing to 56.01% in patients with two or more episodes. There was an increase in the isolation of Staphylococcus aureus with increasing number of episodes (P = 0.008), and S. aureus occurred more commonly in patients with recurrent disease due to the same bacterial group (P = 0.04). Patients whose recurrent keratitis was associated with S. aureus had a higher rate of requiring subsequent corneal transplantation (7 of 10) compared to those with Enterobacteriaceae (2 of 7), Pseudomonas aeruginosa (2 of 4), streptococci (2 of 5), or coagulase-negative staphylococci (none of 8) (P = 0.02).

Conclusions.: S. aureus is particularly associated with recurrent keratitis. Identification and treatment of the possible source of the infection may be necessary to reduce the risk of recurrent disease. The potential for the autocthonous S. aureus colonizing the nasopharynx or conjunctiva or lid margin to be a reservoir for recurrent keratitis suggests that decolonization of S. aureus could be considered as a potential intervention in those patients with recurrent disease.

Introduction

Bacterial keratitis is a major cause of loss of vision worldwide. ^1–3 Each episode leads to further corneal scarring and visual impairment. In the United Kingdom, bacterial keratitis accounts for 8% of corneal transplants undertaken per year. Treatment is dependent on prescribing a topical antimicrobial, management of any associated ocular surface disease, and, when evident, removal of a known risk factor such as continued contact lens wear. Clinical outcome is influenced by the health of the host's ocular surface and immune system, type and virulence of the infecting bacteria, and susceptibility to the prescribed antimicrobial. ^4–8 Although treatment of an endogenous source of the infection might be considered important to reduce the risk of further disease, it is seldom identified. In particular, an entity that has received little attention is disease recurrence. There is also little information, for example, on the incidence, outcome, and source or type of bacteria associated with recurrent keratitis. The type of bacteria associated with recurrent disease may indicate the possible source of the infection and enable treatment aimed at eradication of the source, with a possible reduction in recurrent disease, and ultimately reduce visual loss. In order, therefore, to assess if recurrent bacterial keratitis is a significant problem and to test the hypothesis that it is associated with particular bacterial groups, we analyzed data from patients with clinically diagnosed bacterial keratitis over a 16-year period to determine the incidence of recurrent disease in this population and associated bacteria. 

Methods

A retrospective case series analysis was undertaken of the microbiological data of patients presenting with clinically suspected bacterial keratitis to the Royal Liverpool University Hospital over a 16-year period between 1995 and 2010. Patients were identified from the microbiological requests, which were also used to determine number of episodes and time periods between each episode of keratitis. Patients had corneal samples collected from the corneal ulcer at presentation, which were cultured on agar culture plates and into enrichment culture broth (brain heart infusion). The plates were incubated overnight at 37°C under both aerobic and enriched carbon dioxide (5%) atmospheric conditions. Per protocol, patients received a topical fluoroquinolone as initial treatment unless Staphylococcus aureus or a Streptococcus was suspected or detected on a Gram-stained smear, in which case an additional antimicrobial such as teicoplanin 1% or equivalent was prescribed. Subsequent treatment was according to the bacterial antimicrobial susceptibility and clinical course. 

Patients were classified by the number of episodes and the type of bacteria (bacterial group) associated with each episode. A recurrent episode was defined as a further episode of clinically suspected bacterial keratitis occurring in the same patient either more than 3 months after the preceding episode, or within 3 months of the preceding episode if there was documentation that the corneal ulcer had healed on review of the case notes. Those patients who had repeated sampling of the corneal ulcer prior to healing were classified as having had one episode. The surgical database was cross referenced for those patients who had undergone corneal transplantation following the keratitis. The case records of those patients who had had three or more episodes of keratitis were reviewed for recognized risk factors such as contact lens wear or ocular surface disease as previously described. ⁵ Bacteria were grouped as coagulase-negative staphylococci (CNS), S. aureus , Enterobacteriaceae, Pseudomonas aeruginosa , streptococci, other Gram-negative bacteria, or other Gram-positive bacteria. ^5,6 Nonparametric tests (Kruskal-Wallis and Mann-Whitney tests; SPSS version 20, IBM SPSS Statistics, Chicago, IL) were used to test for differences between bacterial type, risk factors, and number of episodes using a significance level of P < 0.05. The study and data accumulation were in conformity with UK laws, and the study adhered to the tenets of the Declaration of Helsinki. 

Results

Clinical and microbiological data for 2418 patients with clinically suspected bacterial keratitis were included. There was an increase in the number of patients presenting with bacterial keratitis, from 154 in 1995 to 317 in 2010 (mean 182 per year, SD 53). While the majority of patients (2124, 87.84%) over this time period had only one episode, 12.15% (294) had at least two episodes, 3.63% (88) at least three, 1.65% (40) at least four, and 0.91% (22) five or more episodes. The mean, median, and mode time in months between recurrent episodes was 25.06 (SD 31.48), 11 months, and 1 month, respectively, with a minimum of 0.81 and a maximum of 127 months. 

Isolation Rates and Bacterial Group

The overall bacterial isolation rate for all episodes was 35.74% (SD 9.41%). There was an increase in the isolation rate for patients with recurrent disease to 56.01% for those with two or more episodes. Bacterial groups isolated over the 16-year period were CNS (26.32%, SD 10.7%), S. aureus (10.3%, SD 5.7%), Enterobacteriaceae (15.31%, SD 6.4%), P. aeruginosa (11.24%, SD 5.0%), streptococci (13.88%, SD 5.2%), other Gram-negative (8.61%, SD 4.6%), and other Gram-positive bacteria (11.96%, SD 4.4%), with little change in the annual spectrum (0.10 < P < 0.91). Polymicrobial isolations occurred in 9.0% (SD 4.7%) of cases, and CNS was almost always present as the coisolate. 

Recurrences

There were differences in the mean number of episodes (recurrence rate) in patients with the same, different, or no bacterial isolate. Patients with recurrent disease with the same group of bacteria had the highest number of episodes (3.54, SD 2.05), compared to patients with recurrent disease from whom different bacterial groups had been isolated (2.62, SD 1.18) (P = 0.01) or patients from whom no bacteria had been isolated (2.35, SD 0.63) (P < 0.01). 

No apparent risk factors were identified that were peculiar to patients with recurrent disease of more than three episodes. Patients whose recurrent keratitis was associated with S. aureus had a significantly higher rate of requiring a penetrating keratoplasty, that is, 7 of 10 (70%) patients compared to 2 of 7 (29%) when associated with Enterobacteriaceae, 2 of 4 (50%) with P. aeruginosa , 2 of 5 (40%) with streptococci, and none of the 8 patients with CNS (P = 0.02). 

Bacterial Groups in Patients With and Without Recurrent Disease

The percentage distribution of the bacterial groups varied according to the number of episodes of keratitis. In patients with up to two episodes of bacterial keratitis, CNS were the most common bacteria isolated, whereas in patients with three or more episodes, S. aureus was the most common group (Table 1). There was a significant increase in the isolation rate of S. aureus with increasing number of episodes (P = 0.008). 

Table 1

 

Patients With and Without Recurrent Disease and Associated Bacterial Groups; Percentage Distribution of Bacterial Groups of Patients Who Were Culture Positive According to the Number of Episodes

No. Episodes	ENT, %	G-ve, %	CNS, %	SA, %	G+ve, %	STP, %	PS, %
1	16.58	6.95	29.41	8.56	12.83	14.44	11.23
≥1	15.31	8.61	26.32	12.68	11.96	13.88	11.24
≥2	14.29	9.96	23.81	16.02	11.26	13.42	11.26
≥3	9.91	12.61	17.12	18.92	14.41	16.22	10.81
≥4	6.06	6.06	21.21	33.33	18.18	9.09	6.06
≥5	8.70	8.70	21.74	21.74	21.74	13.04	4.35
≥6	7.14	7.14	21.43	35.71	21.43	7.14	0.00

 ENT, Enterobacteriaceae; G-ve, other Gram negative; SA, S. aureus ; G+ve, other Gram positive; STP, streptococci; PS, P. aeruginosa .

 Bacterial groups associated with seven or more episodes are not shown due to the small number of patients with those numbers of episodes.

In order to compare the frequency of occurrence of the same bacterial groups in those patients with recurrent disease, the number of times the same group of bacteria was isolated in a particular patient was determined. This was then expressed as a ratio of the possible number of events in which that bacterial group could have been isolated. For example, for one patient with three episodes, the number of times a particular bacterial group could occur only once is three (either the first, second, or third occasion), the number of times it could occur only twice is three (first and second, first and third, or second and third), and the number of times it could occur three times is one. Therefore, if a particular bacterial group was isolated on two occasions each in 4 of 10 patients, this would give a percentage of 13% (4/30 × 100). Of those patients with recurrent disease due to the same bacterial type (Table 2), S. aureus recurred significantly more frequently in comparison to the other bacterial groups (P = 0.04). For example, in patients with two or more episodes, S. aureus was isolated on two or more occasions in the same patient in 19% of events compared to between 2% and 7% for the other bacterial groups (Table 2). 

Table 2

 

Bacterial Group Isolated on One or More Occasions in the Same Patient, Expressed as a Percent of the Possible Number of Combinations for That Number of Episodes

No. Episodes	ENT, %	G-ve, %	CNS, %	SA, %	G+ve, %	STP, %	PS, %
1	42	23	53	32	35	32	21
≥1	46	25	58	52	41	35	24
≥2	4	2	5	19	7	3	3
≥3	0	0	0	4	3	0	0.60
≥4	0	0	0	0.40	0	0	0

Discussion

The risk of developing bacterial keratitis and the severity of the disease are largely dependent on both the identity of the infecting organism and the condition of the ocular surface. ⁵ Identification, however, of other factors that may contribute toward recurrent disease, such as infection from endogenous host sites (e.g., eyelid margin, conjunctiva, and nose), may be of importance for developing interventions to prevent further episodes and so reduce loss of vision. 

There is some evidence that the source of infection is a signature for the bacteria associated with the keratitis. For example, P. aeruginosa is frequently associated with contact lens wear. ⁴ We recently reported that a subpopulation of P. aeruginosa better able to survive in environmental water is associated with keratitis infections. ^9,10 This association suggests that there is a link between a subset of bacteria with particular phenotypic properties and their possible transmission routes, in this instance through water, and that these are associated with contact lens use. In contrast to P. aeruginosa , the source of S. aureus responsible for keratitis may be autocthonous, as nasopharyngeal colonization with S. aureus is recognized as a risk for certain systemic infections. ^11–15 Approximately 20% of the healthy human population are persistently colonized with S. aureus , and the same strain can persist over months or years. ^16–18 A key feature of nonocular S. aureus infection is its recurrence, which is seen in approximately 30% of all cases. ¹⁹ Persistent carriage might predispose these individuals to recurrent S. aureus keratitis, especially those with underlying ocular surface disease. It is not known, however, whether treatment of patients who have S. aureus keratitis with antibacterials (topical, nasal, and systemic) would concomitantly eliminate nasopharyngeal carriage. 

It is also possible that recurrent S. aureus keratitis may be due to persistence of the bacteria at or around (conjunctiva or lid margin) the infection site as small colony variants. ²⁰ These bacterial mutants develop in certain types of infection, often during prolonged treatment with antibacterials, and are associated with persistent and recurrent infections. ²¹  

It is of interest, therefore, that S. aureus was the predominant organism associated with increasing recurrent keratitis. In addition, compared to other bacterial groups associated with recurrent keratitis, whose distribution was similar to that in previous reports, ^5–8 a significantly higher rate of S. aureus recurrence was associated with an increased risk of undergoing a penetrating keratoplasty. 

Although CNS was isolated more frequently for single episodes, the number of recurrences due to CNS was significantly fewer in comparison to S. aureus or the other bacterial groups. This raises the issue of whether CNS is associated with minimal pathology due to its lesser virulence relative to S. aureus and whether CNS and S. aureus are in competition in the anterior nares and/or ocular surface. ²² We have shown that in contrast to more pathogenic organisms ( S. aureus , P. aeruginosa , or the Enterobacteriaceae) where the efficacy of treatment was related to the minimum inhibitory concentration of the topical antimicrobial applied, no such relationship was evident for CNS-associated keratitis. ⁸  

At the outset of our study we hypothesized that identification of the scale of recurrent bacterial keratitis and the identity of bacteria responsible for it might allow for new therapeutic or preventative strategies. The association between S. aureus and recurrent keratitis underlines the need to determine the source of infection. Since nasal carriage is a known risk factor for S. aureus disease, ¹² future studies should identify whether carriage of S. aureus is a risk factor for keratitis and its recurrence. Concomitantly, decolonization of S. aureus could be studied to examine its impact upon recurrent keratitis particularly given its effect on infections elsewhere in the body. ²³ Exogenous sources of S. aureus are also likely to be important in keratitis, and a study of these will aid in determining the key factors responsible for recurrent infection to help reduce the morbidity associated with the disease. 

Acknowledgments

We thank Jayendra Shankar and Sarah Aldwinckle for help with collection of data. 

Disclosure: R. Kaye, None; A. Kaye, None; H. Sueke, None; T. Neal, None; C. Winstanley, None; M. Horsburgh, None; S. Kaye, None