- Email: [email protected]
The Impact of Topical Corticosteroid Use before Diagnosis on the Outcome of Acanthamoeba Keratitis
The Impact of Topical Corticosteroid Use before Diagnosis on the Outcome of Acanthamoeba Keratitis Dana Robaei, PhD, FRANZCO,2 Nicole Carnt, B.Optom, PhD,2 Darwin C. Minassian, FRCOphth, MSc(Epidem),3 John K. G. Dart, MD, FRCOphth1 Objective: To examine the impact of topical corticosteroid use before the diagnosis of Acanthamoeba keratitis (AK) on final visual outcomes and to determine the prognostic factors predicting poorer outcomes. Design: Cohort study. Participants: A total of 209 eyes of 196 patients with retrievable medical records, diagnosed with AK at Moorfields Eye Hospital, London, between January 1991 and April 2012. One eye was randomly excluded from analysis in the 13 cases of bilateral AK. Methods: Patient demographic, initial clinical examination findings, and management details were collected. The outcomes of patients treated with topical corticosteroids before diagnosis of AK were compared with those not treated with topical corticosteroids before diagnosis. A multivariable logistic model, optimized for prior corticosteroid use, was used to derive the odds ratios (ORs) of a suboptimal visual outcome. Main Outcome Measures: Suboptimal visual outcome was defined as final visual acuity (VA) 20/80, corneal perforation, or need for keratoplasty. Results: Acanthamoeba keratitis was diagnosed on microbiological culture in 94 eyes (48.0%), on histopathologic examination in 27 eyes (13.8%), on confocal microscopy in 38 eyes (19.4%), and on the basis of a typical clinical course and response to treatment in 37 eyes (18.9%). Final VA and prior corticosteroid use data were available for 174 eyes (88.8%). In multivariable analysis, corticosteroid use before diagnosis was associated with suboptimal visual outcome (OR, 3.90; 95% confidence interval [CI], 1.78e8.55), as were disease stage 3 at presentation (OR, 5.62; 95% CI, 1.59e19.80) and older age (60þ years) at diagnosis (OR, 8.97; 95% CI, 2.13e37.79). Conclusions: Corticosteroid use before diagnosis of AK is highly predictive of a poorer visual outcome. This is largely due to the initial misdiagnosis of AK as herpetic keratitis. It is important to include AK in the differential diagnosis of keratitis in all contact lens users with keratitis, particularly before making a diagnosis of herpes keratitis and before the use of topical corticosteroids in the therapy of any indolent keratitis. Ophthalmology 2014;:1e6 ª 2014 by the American Academy of Ophthalmology.
Acanthamoeba keratitis (AK) is an uncommon but severe corneal infection. Since the first report of AK in the literature in 1974,1 there have been improvements in diagnosis and treatment,2,3 although many patients still experience significant morbidity and visual impairment.4 The role of topical corticosteroids in the management of AK remains controversial.5,6 Although some advocate their use for severe corneal inflammation, neovascularization, and the management of scleritis,7 others believe that they potentiate infection and result in a poorer outcome.8 However, more than 70% of ophthalmologists experienced in managing AK use topical corticosteroids for management.9 Likewise, little is known about the impact of topical corticosteroid use before the diagnosis of AK. Although this practice has been widely reported, because of an initial misdiagnosis of AK, usually as herpes keratitis, its impact on final visual outcomes has been uncertain10e13 in the relatively small case series in which this has been evaluated. These studies may have been underpowered to demonstrate an effect and were not optimized to assess this exposure. 2014 by the American Academy of Ophthalmology Published by Elsevier Inc.
Both the misdiagnosis and the use of corticosteroid, before the use of anti-amoebic therapy, are avoidable. If the latter results in poorer outcomes, this is important knowledge, which should alter clinicians’ approach to the diagnosis and treatment of keratitis, hence the rationale for the investigation reported. The purpose of this study is to define the impact of topical corticosteroid use before diagnosis on visual outcomes in AK.
Methods Case Ascertainment A retrospective review of the medical records of patients diagnosed with AK at Moorfields Eye Hospital between January 1991 and April 2012, and with retrievable medical records, was conducted. These cases were identified from our current microbiology laboratory electronic database, which started in the year 2000. An additional 9 cases from a previous database also were included. The study was approved by the Moorfields Eye Hospital Clinical ISSN 0161-6420/14/$ - see front matter http://dx.doi.org/10.1016/j.ophtha.2014.01.031
1
Ophthalmology Volume -, Number -, Month 2014 Research Management and Audit Department, and the Tenets of the Declaration of Helsinki were adhered to. Cases were included in the analyses if they had a positive Acanthamoeba culture or histopathologic confirmation of trophozoites or cysts. Culture-negative cases that were positively identified as having Acanthamoeba cysts on confocal microscopy together with a typical clinical course and response to treatment were included. In the absence of the conditions described, patients with perineural corneal infiltrates or a typical clinical course with a response to anti-amoebic treatment were included in the sample. One eye only of each bilateral case was selected randomly for inclusion in the analysis, by sequential reference to a table of random numbers, using the rule “right eye if random number is even, otherwise left eye.”
Exposure of Primary Interest Patients were classified into 2 main comparison cohorts according to topical corticosteroid use before diagnosis: Those treated with prior corticosteroids formed the exposed cohort, and those with no prior corticosteroids formed the unexposed cohort.
Definition of Visual Outcomes Our primary outcome measure was a suboptimal visual outcome, defined as a final best-available visual acuity (VA) 20/80 (recorded after completion of therapy for the AK episode), corneal perforation, or need for keratoplasty.
Definition of Disease Staging Disease stage at presentation was divided into 3 categories: Stage 1 AK was defined as the presence of corneal epitheliopathy only. Stage 2 AK was defined by the presence of 1 corneal epithelial defects, perineural infiltrate, or stromal infiltrate in addition to stage 1 findings. Stage 3 disease required the presence of a corneal ring infiltrate and 1 or more features of stage 2 disease.
Statistical Analyses Statistical analyses were performed using Stata software version 8 (StataCorp LP, College Station, TX). Analysis of the impact of corticosteroid use on final visual outcomes was restricted to those with available corticosteroid use and final VA data. Patients with significant preexisting disease-limiting visual potential also were excluded.
Univariable Analyses The frequency of demographic and clinical factors in the 2 exposure groups of primary interest (exposed and not exposed to topical corticosteroids before diagnosis) was estimated, using exact procedures for computation of 95% confidence intervals (CIs). The
“crude” association of the exposure of primary interest with other exposures (one at a time) also was examined, using the Fisher exact test for comparison of frequencies and the nonparametric 2-sample Wilcoxon rank-sum (ManneWhitney) test for comparison of means or medians.
Multivariable Analyses Logistic regression models were used to assess the effect of exposures on the visual outcome. The exposure of main interest was topical corticosteroid use before diagnosis, all others being considered as auxiliary variables (potential confounders/effect modifiers). Preliminary cross-tabulations examining the interrelations of the visual outcome with exposures helped to identify auxiliary variables that could be effect modifiers or important confounders, thus candidates for inclusion in the logistic models. The modeling process was designed to arrive at a final logistic model that estimated the odds ratio (OR) for the exposure of main interest, with optimal control of confounding effects. Auxiliary variables included in the final model were those suspected or known a priori to be associated with the outcome and with the exposure of main interest. Also included were the auxiliaries identified in the modeling process as important confounders on statistical grounds. Events that were believed to have occurred as a consequence of prior corticosteroid use and that could increase the risk of a suboptimal outcome were not adjusted for, because doing so would lead to underestimation of the prior corticosteroid effect. Likelihood ratio tests were used to assess effect modification by auxiliary variables. Model performance and validity were assessed through post-fit diagnostics.
Results A total of 196 patients diagnosed with AK between 1991 and 2012 were included. Thirteen patients (6.6%) had bilateral infection, and in this group, 1 eye was randomly excluded from further analysis, leaving 196 eyes of 196 patients. The majority of patients were treated with dual therapy using a biguanide (polyhexanide 0.02, 0.06% or chlorhexidine 0.02, 0.2%) and diamidine (propamidine 0.1% or hexamidine 0.1%). Table 1 outlines the primary criteria for AK diagnosis in all patients and shows the numbers (n ¼ 22) excluded from analysis because of missing data on prior corticosteroid use or on the primary outcome, leaving a total of 174 patients for further analysis. Of these, 87 (50%) had topical corticosteroids before diagnosis of AK. Table 2 compares the demographic factors between those exposed and unexposed to prior corticosteroids. Patients who had received steroids before the diagnosis of AK were significantly younger (mean, 33.6 years) than those who had not (mean, 40.3 years; P<0.005), had significantly more visits to the clinic
Table 1. Acanthamoeba Keratitis Case Ascertainment (n ¼ 196) Excluded from Analysis n [ 22 Basis of Diagnosis Positive culture (i.e., corneal scrape positive) Positive histopathology Positive confocal microscopy and typical clinical course Perineural infiltrates and typical clinical course None of the above: typical clinical course only Total
2
n (%) 94 27 38 20 17 196
(48.0) (13.8) (19.4) (10.2) (8.7) (100)
Missing Data on Prior Corticosteroid Use
Outcome Not Ascertained Because of Preexisting Eye Disease
5 4 2 1 1 13
2 0 5 1 1 9
Robaei et al
Impact of Prior Corticosteroid Use on AK Outcomes
Table 2. Demographic Characteristics of Patients Who Received (n ¼ 87) and Did Not Receive (n ¼ 87) Topical Corticosteroids before Diagnosis of Acanthamoeba Keratitis Corticosteroids Used before Diagnosis (n [ 87) Age at diagnosis Median (IQR) Mean (SD) Gender, n (%) Female Male Affected eye, n (%) Right Left Both* Unknown Contact lens wear, n (%) Yes No Unknown Initially diagnosed aty Moorfields, n (%) Elsewhere Unknown No. of clinic visits Median (IQR) Mean (SD) Duration of followup (mos) Median (IQR) Mean (SD)
Corticosteroids Not Used before Diagnosis (n [ 87) P Value
38 (28e50) 40.3 (15.4)
31 (23e44) 33.6 (12.4)
0.005
45 (51.72) 42 (48.28)
48 (55.17) 39 (44.83)
0.761
45 (51.72) 41 (47.13) 1 (1.15) 0
41 (47.67) 43 (50.00) 2 (2.33) 1
0.775
79 (91.86) 7 (8.14) 1
83 (95.40) 4 (4.60) 0
0.370
52 (60.47) 34 (39.53) 1
59 (71.95) 23 (28.05) 5
0.143
20 (10e43) 28.9 (24.3)
8 (5e17) 15.3 (16.8)
<0.001
18.5 (6.5e44) 35.3 (45.1)
7.5 (4.5e24) 24.2 (38.2)
diagnosed with scleritis versus 16 (18.4%) of those unexposed to corticosteroids. Likewise, a significantly larger proportion of those who received steroids (37 [42.4%]) went on to require keratoplasty compared with 8 patients (9.2%) who did not receive corticosteroids before diagnosis (P<0.001). Corticosteroids were given to 50% of patients before the diagnosis of AK and to 74% of patients after the diagnosis of AK. Patients who received corticosteroids before diagnosis were more likely to use corticosteroids after the diagnosis compared with those not having prior corticosteroids (86% vs. 62%, P<0.001). The risk of a suboptimal outcome was lowest in those not receiving corticosteroids before or after diagnosis (6%, n ¼ 33). Table 4 outlines the relationship between corticosteroid use before AK diagnosis and a suboptimal visual outcome (VA 20/ 80, corneal perforation, or need for keratoplasty) after adjustment for confounding variables. Prior corticosteroid use is associated with an approximately 4-fold increase in the odds of a suboptimal visual outcome (OR, 3.90; CI, 1.78e8.55; P ¼ 0.001). The crude unadjusted OR was 5.83 (CI, 2.97e11.46; P<0.001). Advanced disease stage at diagnosis also is independently associated with an increase in the odds of a suboptimal visual outcome, as is older age (Table 4). Gender, diabetes, human immunodeficiency virus, and immunosuppressive therapy were trivial confounders, reducing the prior corticosteroids OR from 3.90 to 3.81 when included in the final model. Culture-proven bacterial keratitis, thought to be partly a consequence of prior corticosteroid use, was nevertheless assessed in the model building process. Its inclusion in the final model also had a trivial effect, reducing the OR of main interest from 3.90 to 3.82. None of these were significant risk factors.
0.011
Discussion IQR ¼ interquartile range; SD ¼ standard deviation. The IQR is a measure of dispersion, presented here as the 25th and 75th percentiles of the data distribution. *Only 1 eye of each bilateral case, selected randomly, was included in the analysis. y This refers to when the correct diagnosis of AK was made.
(median, 20 vs. 8; P<0.001), and had a significantly longer followup period (median, 18.5 vs. 7.5 months; P ¼ 0.011). There were no significant differences between the patients exposed and unexposed to corticosteroids with regard to gender and contact lens wear, or whether they were initially diagnosed at Moorfields Eye Hospital or were secondarily referred from elsewhere. Table 3 documents the ophthalmic characteristics of those exposed and unexposed to prior corticosteroids. On univariable (crude) analysis, the use of corticosteroids before the diagnosis of AK was significantly associated with a diagnostic delay, with a mean symptom duration of 55.9 days for those receiving steroids before diagnosis compared with 22.3 days for those not receiving corticosteroids (P<0.001). Commensurate with this finding, a significantly higher proportion among those exposed to corticosteroids presented with late-stage AK than those unexposed to corticosteroids, with 27 (31.8%) having a corneal ring infiltrate at diagnosis compared with 15 (15.5%), respectively (P ¼ 0.037). The majority of patients (63 [73.3%]) who had received topical corticosteroids before diagnosis were initially misdiagnosed with herpes keratitis compared with only 28 patients (32.2%) who did not receive steroids before the diagnosis of AK (P<0.001). On univariable analysis, patients with prior exposure to corticosteroids also had significantly more complications, with 44 (50.6%) being
The widespread use of topical corticosteroids in contact lens wearers who then go on to be diagnosed with AK has been widely reported10e14 and was again confirmed in this study, with 50% of our analysis cohort being prescribed steroids before the diagnosis of AK. This practice was associated with an approximately 4-fold increased odds of a poorer visual outcome on multivariable analysis. Yamazoe et al15 previously reported a similar finding, although in their report the impact was statistically significant on univariable analysis only and unconfirmed on adjustment for confounding variables. The latter study and the other studies evaluating this exposure7,10 include relatively small numbers of cases and may not have had the power to demonstrate the effect that we have clearly shown. Tu et al10 did not find a significant association between prior corticosteroid use and final visual outcome, but instead ascribed a significant role to corneal disease stage at presentation. This was also an independent and predictive factor in determining final VA in our study, resulting in stage 3 disease having a 5-fold increased odds of suboptimal visual outcome compared with stage 1 disease. The early clinical signs of AK are highly variable and, in the absence of perineural infiltrates, often resemble the epitheliopathy, dendritiform lesions, with or without stromal and endothelial (disciform) inflammation, that are often encountered in herpes simplex keratitis (HSK).16 Moreover, long-term contact lens wear is associated with decreased
3
Ophthalmology Volume -, Number -, Month 2014 Table 3. Ophthalmic Characteristics of Patients Who Received (n ¼ 87) and Did Not Receive (n ¼ 87) Topical Corticosteroids before Diagnosis of Acanthamoeba Keratitis
Symptom duration (days) Median (IQR) Mean (SD) Disease stage at diagnosis, n (%) 1 (corneal epitheliopathy) 2 (stage 1 þ epithelial defect, perineural infiltrate, or stromal infiltrate) 3 (stage 2 þ corneal ring infiltrate) Unknown Bacterial super-infection, n (%) Treatment for HSV keratitis, n (%) Unknown Anti-amoebic treatment duration (mos) Median (IQR) Mean (SD) Scleritis, n (%) Corticosteroid-sparing immunosuppressive, n (%) Corneal perforation, n (%) Unknown Keratoplasty, n (%)
Corticosteroids Used before Diagnosis (n [ 87)
Corticosteroids Not Used before Diagnosis (n [ 87)
37 (23e72.5) 54.9 (55.0)
14 (7e28) 22.3 (26.9)
<0.001
12 (14.12) 46 (54.12)
18 (21.43) 53 (63.10)
0.037
27 (31.76) 2 20 (22.99) 63 (73.26) 1
13 (15.48) 3 11 (12.64) 28 (32.18) 0
6 8.1 44 16 13
4 7.1 16 8 6
(3.5e12) (6.4) (50.57) (18.39) (15.12) 1 37 (42.53)
(3e8.5) (6.8) (18.39) (9.20) (6.90) 0 8 (9.20)
P Value
0.112 <0.001 0.068 <0.001 0.123 0.094 <0.001
HSV ¼ herpes simplex virus; IQR ¼ interquartile range; SD ¼ standard deviation. The IQR is a measure of dispersion, presented here as the 25th and 75th percentiles of the data distribution.
Table 4. Association between Corticosteroid Use before Diagnosis and Suboptimal Visual Outcome of Acanthamoeba Keratitis (Final Visual Acuity 20/80 or Corneal Perforation or Need for Keratoplasty), from a Multivariable Logistic Regression Model Optimized to Assess the Exposure of Main Interest Exposures Corticosteroid use before diagnosis No (referent) Yes Auxiliary variables Disease stage at diagnosis Stage 1 (referent) Stage 2 Stage 3 Age at diagnosis 15e29 yrs (referent) 30e59 yrs 60þ yrs Year of diagnosis 1991e2000 (referent) 2001e2012 Initially diagnosed at Moorfieldsz Yes (referent) No
n* (%)
Adjustedy OR (95% CI)
P Value
79 (48.5) 84 (51.5)
e 3.90 (1.78e8.55)
e 0.001
29 (17.8) 95 (58.3) 39 (23.9)
e 3.32 (1.07e10.35) 5.62 (1.59e19.80)
e 0.039 0.007
64 (39.3) 85 (52.1) 14 (8.6)
e 4.33 (1.81e10.37) 8.97 (2.13e37.79)
e 0.001 0.003
38 (23.3) 125 (76.7)
e 0.41 (0.17e1.00)
e 0.050
111 (68.1) 52 (31.9)
e 2.21 (0.97e5.04)
e 0.061
CI ¼ confidence interval; OR ¼ odds ratio. *Patients with complete data on primary outcome and on all 5 exposure variables (n ¼ 163). y The OR for any particular variable in the logistic model is adjusted for confounding effects of all 4 remaining exposure variables in the model. z This refers to the correct diagnosis of AK being made at Moorfields Eye Hospital.
4
corneal sensitivity,17,18 which also is incorrectly ascribed to the herpes virus. Pain is often quoted as invariably accompanying early AK,19 but this is not always the case and the disease can be relatively painless.20,21 As a result, a large proportion of patients presenting with early AK are misdiagnosed with HSK and prescribed topical corticosteroids, delaying the prompt institution of anti-amoebic therapy. In our study, 52.3% of all patients, the majority of whom were contact lens wearers, were initially misdiagnosed with HSK. Tu et al10 reported a similar result with 43%, whereas Illingworth et al19 reported an even higher rate of 77%. Whereas the use of corticosteroids given before the diagnosis of AK was wholly or largely due to the wrong (or lack of) diagnosis, their use after diagnosis of AK was limited to those cases developing inflammatory complications only. “Steroid use after diagnosis” was nonetheless examined in the model-building process. When it was added to the multivariable regression model, the impact (OR) of steroid use before diagnosis was reduced (from 3.90 to 3.32), as expected, but remained significant (P ¼ 0.004) independently of steroid use after diagnosis. The latter was considered to be an indicator of increasing disease severity after exposure to prior steroids, and therefore in the causal pathway to a suboptimal outcome. For this reason, it was not included in the final logistic model. Another reason for the misdiagnosis of AK as HSK is that HSK is often diagnosed on the sole basis of clinical findings22,23 without performing the diagnostic investigations that are commonly used when other causes of keratitis are suspected. Even when these investigations have been done, culture and histology are often negative. In our study, 61.7% of patients were diagnosed on the basis of a
Robaei et al
Impact of Prior Corticosteroid Use on AK Outcomes
positive culture or histopathologic examination of corneal tissue, similar to rates previously reported in the literature.10,13 The remaining patients were diagnosed on the basis of confocal microscopy findings, or typical clinical signs (e.g., epitheliopathy, perineural infiltrates), disease course, and response to treatment. This method of case definition has been validated in 2 studies.3,24 However laboratory diagnosis can be increased to >80% for AK when polymerase chain reaction (PCR) is used to identify Acanthamoeba DNA.25,26 Further problems with the misdiagnosis of AK arise when the disease is polymicrobial.24,27 Bacterial infections that are only partially responsive to therapy are often treated with topical corticosteroids but may persist because of concurrent infection with AK. In addition approximately 10% of AK occurs in noncontact lens users, with or without a history of predisposing trauma, and AK must be considered as a cause of progressive (or indolent) microbial keratitis in this group, for whom the differential is fungal, mycobacterial, nocardia, and HSK, among other causes.28 It is disturbing that despite repeated reports of misdiagnosis in the ophthalmological literature, misdiagnosis of AK as HSK is still occurring15,19,29 and has barely improved in our study in the last decade compared with the previous decade. This finding identifies the need for both increased awareness among healthcare professionals and the routine use of PCR and in vivo confocal microscopy by trained observers as more sensitive diagnostic methods, permitting differentiation of Acanthamoeba, HSK, and fungal keratitis. Apart from the association of topical corticosteroids with diagnostic delay, there is some evidence to suggest that their use may result in increased pathogenicity of the Acanthamoeba organism. In an animal model of AK, McClellan et al30 elegantly demonstrated that the addition of topical corticosteroids, even at low doses, promotes an increase in the number of trophozoites, produced by excystment in infected corneal stroma.30 The resultant proliferation of the invasive form of the organism exposes patients to the risk of significantly greater corneal destruction through an increase in organism load, which may be greater than the increased chemotherapy effect on trophozoites compared with the more resistant cysts.
Study Limitations Given the retrospective nature of the study and reliance on accuracy and completeness of the clinical records, our study is limited by potential documentation bias. It is also difficult to completely adjust for potential confounders, such as disease severity at diagnosis. Whereas for this study we have stratified the disease on the basis of the corneal phenotype, this probably results from both differences in the virulence of the Acanthamoeba strain, as well from variations in the host immune response, neither of which can currently be quantified. Analysis of outcomes using more sophisticated disease stratification will be ideal when our understanding of such factors improves. Notwithstanding these limitations, an observational study of this type remains at present the only viable method of studying the role of topical corticosteroids before AK diagnosis.
In conclusion, our study highlights the critical importance of (1) having a high index of suspicion for AK in contact lens wearers and in noncontact lens users who have a persistent keratitis and (2) the need for judicious use of topical corticosteroids when the possibility of AK has not been definitively ruled out, which requires culture, confocal microscopy, and PCR. Older individuals and those with more advanced disease are independently prone to worse outcomes, and particular caution should be exercised in these cases. Prior corticosteroid use is associated with diagnosis delay and culminates in a great deal of ocular morbidity for affected patients, as evidenced by significantly higher rates of scleritis and keratoplasty in our cohort. It also imposes a significant burden on the healthcare system, with the need for a longer duration of follow-up and significantly more presentations to the outpatient department. Acknowledgments. The authors thank Massimo Bonaiti of the Doctor’s Laboratory for preparing the database of patients for whom Acanthamoeba cultures had been requested.
References 1. Naginton J, Watson PG, Playfair TJ, et al. Amoebic infection of the eye. Lancet 1974;2:1537–40. 2. Duguid IG, Dart JK, Morlet N, et al. Outcome of acanthamoeba keratitis treated with polyhexamethyl biguanide and propamidine. Ophthalmology 1997;104:1587–92. 3. Radford CF, Lehmann OJ, Dart JK. Acanthamoeba keratitis: multicentre survey in England 1992-6. National Acanthamoeba Keratitis Study Group. Br J Ophthalmol 1998;82: 1387–92. 4. Radford CF, Minassian DC, Dart JK. Acanthamoeba keratitis in England and Wales: incidence, outcome, and risk factors. Br J Ophthalmol 2002;86:536–42. 5. Osato MS, Robinson NM, Wilhelmus KR, Jones DB. In vitro evaluation of antimicrobial compounds for cysticidal activity against Acanthamoeba. Rev Infect Dis 1991;13(Suppl 5):S431–5. 6. O’Day DM, Head WS. Advances in the management of keratomycosis and Acanthamoeba keratitis. Cornea 2000;19: 681–7. 7. Park DH, Palay DA, Daya SM, et al. The role of topical corticosteroids in the management of Acanthamoeba keratitis. Cornea 1997;16:277–83. 8. Stern GA, Buttross M. Use of corticosteroids in combination with antimicrobial drugs in the treatment of infectious corneal disease. Ophthalmology 1991;98:847–53. 9. Oldenburg CE, Acharya NR, Tu EY, et al. Practice patterns and opinions in the treatment of acanthamoeba keratitis. Cornea 2011;30:1363–8. 10. Tu EY, Joslin CE, Sugar J, et al. Prognostic factors affecting visual outcome in Acanthamoeba keratitis. Ophthalmology 2008;115:1998–2003. 11. Patel DV, Rayner S, McGhee CN. Resurgence of Acanthamoeba keratitis in Auckland, New Zealand: a 7-year review of presentation and outcomes. Clin Experiment Ophthalmol 2010;38:15–20; quiz 87. 12. Kaiserman I, Bahar I, McAllum P, et al. Prognostic factors in Acanthamoeba keratitis. Can J Ophthalmol 2012;47:312–7. 13. Chew HF, Yildiz EH, Hammersmith KM, et al. Clinical outcomes and prognostic factors associated with acanthamoeba keratitis. Cornea 2011;30:435–41. 14. Saeed A, D’Arcy F, Stack J, et al. Risk factors, microbiological findings, and clinical outcomes in cases of microbial
5
Ophthalmology Volume -, Number -, Month 2014
15. 16. 17. 18. 19. 20. 21. 22. 23.
keratitis admitted to a tertiary referral center in Ireland. Cornea 2009;28:285–92. Yamazoe K, Yamamoto Y, Shimazaki-Den S, Shimazaki J. Visual outcome in Japanese patients with Acanthamoeba keratitis. Eye (Lond) 2012;26:517–22. Illingworth CD, Cook SD. Acanthamoeba keratitis. Surv Ophthalmol 1998;42:493–508. Liesegang TJ. Physiologic changes of the cornea with contact lens wear. CLAO J 28:12e27. Murphy PJ, Patel S, Marshall J. The effect of long-term, daily contact lens wear on corneal sensitivity. Cornea 20:264e269. Illingworth CD, Cook SD, Karabatsas CH, Easty DL. Acanthamoeba keratitis: risk factors and outcome. Br J Ophthalmol 1995;79:1078–82. Tabin G, Taylor H, Snibson G, et al. Atypical presentation of Acanthamoeba keratitis. Cornea 2001;20:757–9. Shukla Kent S, Robert MC, Tokarewicz AC, Mather R. Painless Acanthamoeba keratitis. Can J Ophthalmol 2012;47: 383–4. Kabra A, Lalitha P, Mahadevan K, et al. Herpes simplex keratitis and visual impairment: a case series. Indian J Ophthalmol 2006;54:23–7. Inoue Y, Shimomura Y, Fukuda M, et al. Multicentre clinical study of the herpes simplex virus immunochromatographic
24. 25.
26.
27. 28. 29. 30.
assay kit for the diagnosis of herpetic epithelial keratitis. Br J Ophthalmol 2013;97:1108–12. Bacon AS, Frazer DG, Dart JK, et al. A review of 72 consecutive cases of Acanthamoeba keratitis, 1984-1992. Eye (Lond) 1993;7(Pt 6):719–25. Lehmann OJ, Green SM, Morlet N, et al. Polymerase chain reaction analysis of corneal epithelial and tear samples in the diagnosis of Acanthamoeba keratitis. Invest Ophthalmol Vis Sci 1998;39:1261–5. Boggild AK, Martin DS, Lee TY, et al. Laboratory diagnosis of amoebic keratitis: comparison of four diagnostic methods for different types of clinical specimens. J Clin Microbiol 2009;47:1314–8. Cariello AJ, Passos RM, Yu MC, Hofling-Lima AL. Microbial keratitis at a referral center in Brazil. Int Ophthalmol 2011;31: 197–204. Allan BD, Dart JK. Strategies for the management of microbial keratitis. Br J Ophthalmol 1995;79:777–86. Butler TK, Males JJ, Robinson LP, et al. Six-year review of Acanthamoeba keratitis in New South Wales, Australia: 19972002. Clin Experiment Ophthalmol 2005;33:41–6. McClellan K, Howard K, Niederkorn JY, Alizadeh H. Effect of steroids on Acanthamoeba cysts and trophozoites. Invest Ophthalmol Vis Sci 2001;42:2885–93.
Footnotes and Financial Disclosures Originally received: November 22, 2013. Final revision: January 27, 2014. Accepted: January 28, 2014. Available online: ---.
Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Manuscript no. 2013-1948.
1
National Institute of Health Research Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, London, United Kingdom.
2
Moorfields Eye Hospital NHS Foundation Trust, London, United Kingdom.
3
EpiVision Ophthalmic Epidemiology Consultants, Bucks, United Kingdom.
6
Abbreviations and Acronyms: AK ¼ Acanthamoeba keratitis; CI ¼ confidence interval; HSK ¼ herpes simplex keratitis; OR ¼ odds ratio; PCR ¼ polymerase chain reaction; VA ¼ visual acuity. Correspondence: John K. G. Dart, MD, FRCOphth, Moorfields Eye Hospital NHS Foundation Trust, 162 City Rd., London, EC1V 2PD, UK. E-mail: j.dart@ucl. ac.uk.
Acanthamoeba keratitis (AK) is an uncommon but severe corneal infection. Since the first report of AK in the literature in 1974,1 there have been improvements in diagnosis and treatment,2,3 although many patients still experience significant morbidity and visual impairment.4 The role of topical corticosteroids in the management of AK remains controversial.5,6 Although some advocate their use for severe corneal inflammation, neovascularization, and the management of scleritis,7 others believe that they potentiate infection and result in a poorer outcome.8 However, more than 70% of ophthalmologists experienced in managing AK use topical corticosteroids for management.9 Likewise, little is known about the impact of topical corticosteroid use before the diagnosis of AK. Although this practice has been widely reported, because of an initial misdiagnosis of AK, usually as herpes keratitis, its impact on final visual outcomes has been uncertain10e13 in the relatively small case series in which this has been evaluated. These studies may have been underpowered to demonstrate an effect and were not optimized to assess this exposure. 2014 by the American Academy of Ophthalmology Published by Elsevier Inc.
Both the misdiagnosis and the use of corticosteroid, before the use of anti-amoebic therapy, are avoidable. If the latter results in poorer outcomes, this is important knowledge, which should alter clinicians’ approach to the diagnosis and treatment of keratitis, hence the rationale for the investigation reported. The purpose of this study is to define the impact of topical corticosteroid use before diagnosis on visual outcomes in AK.
Methods Case Ascertainment A retrospective review of the medical records of patients diagnosed with AK at Moorfields Eye Hospital between January 1991 and April 2012, and with retrievable medical records, was conducted. These cases were identified from our current microbiology laboratory electronic database, which started in the year 2000. An additional 9 cases from a previous database also were included. The study was approved by the Moorfields Eye Hospital Clinical ISSN 0161-6420/14/$ - see front matter http://dx.doi.org/10.1016/j.ophtha.2014.01.031
1
Ophthalmology Volume -, Number -, Month 2014 Research Management and Audit Department, and the Tenets of the Declaration of Helsinki were adhered to. Cases were included in the analyses if they had a positive Acanthamoeba culture or histopathologic confirmation of trophozoites or cysts. Culture-negative cases that were positively identified as having Acanthamoeba cysts on confocal microscopy together with a typical clinical course and response to treatment were included. In the absence of the conditions described, patients with perineural corneal infiltrates or a typical clinical course with a response to anti-amoebic treatment were included in the sample. One eye only of each bilateral case was selected randomly for inclusion in the analysis, by sequential reference to a table of random numbers, using the rule “right eye if random number is even, otherwise left eye.”
Exposure of Primary Interest Patients were classified into 2 main comparison cohorts according to topical corticosteroid use before diagnosis: Those treated with prior corticosteroids formed the exposed cohort, and those with no prior corticosteroids formed the unexposed cohort.
Definition of Visual Outcomes Our primary outcome measure was a suboptimal visual outcome, defined as a final best-available visual acuity (VA) 20/80 (recorded after completion of therapy for the AK episode), corneal perforation, or need for keratoplasty.
Definition of Disease Staging Disease stage at presentation was divided into 3 categories: Stage 1 AK was defined as the presence of corneal epitheliopathy only. Stage 2 AK was defined by the presence of 1 corneal epithelial defects, perineural infiltrate, or stromal infiltrate in addition to stage 1 findings. Stage 3 disease required the presence of a corneal ring infiltrate and 1 or more features of stage 2 disease.
Statistical Analyses Statistical analyses were performed using Stata software version 8 (StataCorp LP, College Station, TX). Analysis of the impact of corticosteroid use on final visual outcomes was restricted to those with available corticosteroid use and final VA data. Patients with significant preexisting disease-limiting visual potential also were excluded.
Univariable Analyses The frequency of demographic and clinical factors in the 2 exposure groups of primary interest (exposed and not exposed to topical corticosteroids before diagnosis) was estimated, using exact procedures for computation of 95% confidence intervals (CIs). The
“crude” association of the exposure of primary interest with other exposures (one at a time) also was examined, using the Fisher exact test for comparison of frequencies and the nonparametric 2-sample Wilcoxon rank-sum (ManneWhitney) test for comparison of means or medians.
Multivariable Analyses Logistic regression models were used to assess the effect of exposures on the visual outcome. The exposure of main interest was topical corticosteroid use before diagnosis, all others being considered as auxiliary variables (potential confounders/effect modifiers). Preliminary cross-tabulations examining the interrelations of the visual outcome with exposures helped to identify auxiliary variables that could be effect modifiers or important confounders, thus candidates for inclusion in the logistic models. The modeling process was designed to arrive at a final logistic model that estimated the odds ratio (OR) for the exposure of main interest, with optimal control of confounding effects. Auxiliary variables included in the final model were those suspected or known a priori to be associated with the outcome and with the exposure of main interest. Also included were the auxiliaries identified in the modeling process as important confounders on statistical grounds. Events that were believed to have occurred as a consequence of prior corticosteroid use and that could increase the risk of a suboptimal outcome were not adjusted for, because doing so would lead to underestimation of the prior corticosteroid effect. Likelihood ratio tests were used to assess effect modification by auxiliary variables. Model performance and validity were assessed through post-fit diagnostics.
Results A total of 196 patients diagnosed with AK between 1991 and 2012 were included. Thirteen patients (6.6%) had bilateral infection, and in this group, 1 eye was randomly excluded from further analysis, leaving 196 eyes of 196 patients. The majority of patients were treated with dual therapy using a biguanide (polyhexanide 0.02, 0.06% or chlorhexidine 0.02, 0.2%) and diamidine (propamidine 0.1% or hexamidine 0.1%). Table 1 outlines the primary criteria for AK diagnosis in all patients and shows the numbers (n ¼ 22) excluded from analysis because of missing data on prior corticosteroid use or on the primary outcome, leaving a total of 174 patients for further analysis. Of these, 87 (50%) had topical corticosteroids before diagnosis of AK. Table 2 compares the demographic factors between those exposed and unexposed to prior corticosteroids. Patients who had received steroids before the diagnosis of AK were significantly younger (mean, 33.6 years) than those who had not (mean, 40.3 years; P<0.005), had significantly more visits to the clinic
Table 1. Acanthamoeba Keratitis Case Ascertainment (n ¼ 196) Excluded from Analysis n [ 22 Basis of Diagnosis Positive culture (i.e., corneal scrape positive) Positive histopathology Positive confocal microscopy and typical clinical course Perineural infiltrates and typical clinical course None of the above: typical clinical course only Total
2
n (%) 94 27 38 20 17 196
(48.0) (13.8) (19.4) (10.2) (8.7) (100)
Missing Data on Prior Corticosteroid Use
Outcome Not Ascertained Because of Preexisting Eye Disease
5 4 2 1 1 13
2 0 5 1 1 9
Robaei et al
Impact of Prior Corticosteroid Use on AK Outcomes
Table 2. Demographic Characteristics of Patients Who Received (n ¼ 87) and Did Not Receive (n ¼ 87) Topical Corticosteroids before Diagnosis of Acanthamoeba Keratitis Corticosteroids Used before Diagnosis (n [ 87) Age at diagnosis Median (IQR) Mean (SD) Gender, n (%) Female Male Affected eye, n (%) Right Left Both* Unknown Contact lens wear, n (%) Yes No Unknown Initially diagnosed aty Moorfields, n (%) Elsewhere Unknown No. of clinic visits Median (IQR) Mean (SD) Duration of followup (mos) Median (IQR) Mean (SD)
Corticosteroids Not Used before Diagnosis (n [ 87) P Value
38 (28e50) 40.3 (15.4)
31 (23e44) 33.6 (12.4)
0.005
45 (51.72) 42 (48.28)
48 (55.17) 39 (44.83)
0.761
45 (51.72) 41 (47.13) 1 (1.15) 0
41 (47.67) 43 (50.00) 2 (2.33) 1
0.775
79 (91.86) 7 (8.14) 1
83 (95.40) 4 (4.60) 0
0.370
52 (60.47) 34 (39.53) 1
59 (71.95) 23 (28.05) 5
0.143
20 (10e43) 28.9 (24.3)
8 (5e17) 15.3 (16.8)
<0.001
18.5 (6.5e44) 35.3 (45.1)
7.5 (4.5e24) 24.2 (38.2)
diagnosed with scleritis versus 16 (18.4%) of those unexposed to corticosteroids. Likewise, a significantly larger proportion of those who received steroids (37 [42.4%]) went on to require keratoplasty compared with 8 patients (9.2%) who did not receive corticosteroids before diagnosis (P<0.001). Corticosteroids were given to 50% of patients before the diagnosis of AK and to 74% of patients after the diagnosis of AK. Patients who received corticosteroids before diagnosis were more likely to use corticosteroids after the diagnosis compared with those not having prior corticosteroids (86% vs. 62%, P<0.001). The risk of a suboptimal outcome was lowest in those not receiving corticosteroids before or after diagnosis (6%, n ¼ 33). Table 4 outlines the relationship between corticosteroid use before AK diagnosis and a suboptimal visual outcome (VA 20/ 80, corneal perforation, or need for keratoplasty) after adjustment for confounding variables. Prior corticosteroid use is associated with an approximately 4-fold increase in the odds of a suboptimal visual outcome (OR, 3.90; CI, 1.78e8.55; P ¼ 0.001). The crude unadjusted OR was 5.83 (CI, 2.97e11.46; P<0.001). Advanced disease stage at diagnosis also is independently associated with an increase in the odds of a suboptimal visual outcome, as is older age (Table 4). Gender, diabetes, human immunodeficiency virus, and immunosuppressive therapy were trivial confounders, reducing the prior corticosteroids OR from 3.90 to 3.81 when included in the final model. Culture-proven bacterial keratitis, thought to be partly a consequence of prior corticosteroid use, was nevertheless assessed in the model building process. Its inclusion in the final model also had a trivial effect, reducing the OR of main interest from 3.90 to 3.82. None of these were significant risk factors.
0.011
Discussion IQR ¼ interquartile range; SD ¼ standard deviation. The IQR is a measure of dispersion, presented here as the 25th and 75th percentiles of the data distribution. *Only 1 eye of each bilateral case, selected randomly, was included in the analysis. y This refers to when the correct diagnosis of AK was made.
(median, 20 vs. 8; P<0.001), and had a significantly longer followup period (median, 18.5 vs. 7.5 months; P ¼ 0.011). There were no significant differences between the patients exposed and unexposed to corticosteroids with regard to gender and contact lens wear, or whether they were initially diagnosed at Moorfields Eye Hospital or were secondarily referred from elsewhere. Table 3 documents the ophthalmic characteristics of those exposed and unexposed to prior corticosteroids. On univariable (crude) analysis, the use of corticosteroids before the diagnosis of AK was significantly associated with a diagnostic delay, with a mean symptom duration of 55.9 days for those receiving steroids before diagnosis compared with 22.3 days for those not receiving corticosteroids (P<0.001). Commensurate with this finding, a significantly higher proportion among those exposed to corticosteroids presented with late-stage AK than those unexposed to corticosteroids, with 27 (31.8%) having a corneal ring infiltrate at diagnosis compared with 15 (15.5%), respectively (P ¼ 0.037). The majority of patients (63 [73.3%]) who had received topical corticosteroids before diagnosis were initially misdiagnosed with herpes keratitis compared with only 28 patients (32.2%) who did not receive steroids before the diagnosis of AK (P<0.001). On univariable analysis, patients with prior exposure to corticosteroids also had significantly more complications, with 44 (50.6%) being
The widespread use of topical corticosteroids in contact lens wearers who then go on to be diagnosed with AK has been widely reported10e14 and was again confirmed in this study, with 50% of our analysis cohort being prescribed steroids before the diagnosis of AK. This practice was associated with an approximately 4-fold increased odds of a poorer visual outcome on multivariable analysis. Yamazoe et al15 previously reported a similar finding, although in their report the impact was statistically significant on univariable analysis only and unconfirmed on adjustment for confounding variables. The latter study and the other studies evaluating this exposure7,10 include relatively small numbers of cases and may not have had the power to demonstrate the effect that we have clearly shown. Tu et al10 did not find a significant association between prior corticosteroid use and final visual outcome, but instead ascribed a significant role to corneal disease stage at presentation. This was also an independent and predictive factor in determining final VA in our study, resulting in stage 3 disease having a 5-fold increased odds of suboptimal visual outcome compared with stage 1 disease. The early clinical signs of AK are highly variable and, in the absence of perineural infiltrates, often resemble the epitheliopathy, dendritiform lesions, with or without stromal and endothelial (disciform) inflammation, that are often encountered in herpes simplex keratitis (HSK).16 Moreover, long-term contact lens wear is associated with decreased
3
Ophthalmology Volume -, Number -, Month 2014 Table 3. Ophthalmic Characteristics of Patients Who Received (n ¼ 87) and Did Not Receive (n ¼ 87) Topical Corticosteroids before Diagnosis of Acanthamoeba Keratitis
Symptom duration (days) Median (IQR) Mean (SD) Disease stage at diagnosis, n (%) 1 (corneal epitheliopathy) 2 (stage 1 þ epithelial defect, perineural infiltrate, or stromal infiltrate) 3 (stage 2 þ corneal ring infiltrate) Unknown Bacterial super-infection, n (%) Treatment for HSV keratitis, n (%) Unknown Anti-amoebic treatment duration (mos) Median (IQR) Mean (SD) Scleritis, n (%) Corticosteroid-sparing immunosuppressive, n (%) Corneal perforation, n (%) Unknown Keratoplasty, n (%)
Corticosteroids Used before Diagnosis (n [ 87)
Corticosteroids Not Used before Diagnosis (n [ 87)
37 (23e72.5) 54.9 (55.0)
14 (7e28) 22.3 (26.9)
<0.001
12 (14.12) 46 (54.12)
18 (21.43) 53 (63.10)
0.037
27 (31.76) 2 20 (22.99) 63 (73.26) 1
13 (15.48) 3 11 (12.64) 28 (32.18) 0
6 8.1 44 16 13
4 7.1 16 8 6
(3.5e12) (6.4) (50.57) (18.39) (15.12) 1 37 (42.53)
(3e8.5) (6.8) (18.39) (9.20) (6.90) 0 8 (9.20)
P Value
0.112 <0.001 0.068 <0.001 0.123 0.094 <0.001
HSV ¼ herpes simplex virus; IQR ¼ interquartile range; SD ¼ standard deviation. The IQR is a measure of dispersion, presented here as the 25th and 75th percentiles of the data distribution.
Table 4. Association between Corticosteroid Use before Diagnosis and Suboptimal Visual Outcome of Acanthamoeba Keratitis (Final Visual Acuity 20/80 or Corneal Perforation or Need for Keratoplasty), from a Multivariable Logistic Regression Model Optimized to Assess the Exposure of Main Interest Exposures Corticosteroid use before diagnosis No (referent) Yes Auxiliary variables Disease stage at diagnosis Stage 1 (referent) Stage 2 Stage 3 Age at diagnosis 15e29 yrs (referent) 30e59 yrs 60þ yrs Year of diagnosis 1991e2000 (referent) 2001e2012 Initially diagnosed at Moorfieldsz Yes (referent) No
n* (%)
Adjustedy OR (95% CI)
P Value
79 (48.5) 84 (51.5)
e 3.90 (1.78e8.55)
e 0.001
29 (17.8) 95 (58.3) 39 (23.9)
e 3.32 (1.07e10.35) 5.62 (1.59e19.80)
e 0.039 0.007
64 (39.3) 85 (52.1) 14 (8.6)
e 4.33 (1.81e10.37) 8.97 (2.13e37.79)
e 0.001 0.003
38 (23.3) 125 (76.7)
e 0.41 (0.17e1.00)
e 0.050
111 (68.1) 52 (31.9)
e 2.21 (0.97e5.04)
e 0.061
CI ¼ confidence interval; OR ¼ odds ratio. *Patients with complete data on primary outcome and on all 5 exposure variables (n ¼ 163). y The OR for any particular variable in the logistic model is adjusted for confounding effects of all 4 remaining exposure variables in the model. z This refers to the correct diagnosis of AK being made at Moorfields Eye Hospital.
4
corneal sensitivity,17,18 which also is incorrectly ascribed to the herpes virus. Pain is often quoted as invariably accompanying early AK,19 but this is not always the case and the disease can be relatively painless.20,21 As a result, a large proportion of patients presenting with early AK are misdiagnosed with HSK and prescribed topical corticosteroids, delaying the prompt institution of anti-amoebic therapy. In our study, 52.3% of all patients, the majority of whom were contact lens wearers, were initially misdiagnosed with HSK. Tu et al10 reported a similar result with 43%, whereas Illingworth et al19 reported an even higher rate of 77%. Whereas the use of corticosteroids given before the diagnosis of AK was wholly or largely due to the wrong (or lack of) diagnosis, their use after diagnosis of AK was limited to those cases developing inflammatory complications only. “Steroid use after diagnosis” was nonetheless examined in the model-building process. When it was added to the multivariable regression model, the impact (OR) of steroid use before diagnosis was reduced (from 3.90 to 3.32), as expected, but remained significant (P ¼ 0.004) independently of steroid use after diagnosis. The latter was considered to be an indicator of increasing disease severity after exposure to prior steroids, and therefore in the causal pathway to a suboptimal outcome. For this reason, it was not included in the final logistic model. Another reason for the misdiagnosis of AK as HSK is that HSK is often diagnosed on the sole basis of clinical findings22,23 without performing the diagnostic investigations that are commonly used when other causes of keratitis are suspected. Even when these investigations have been done, culture and histology are often negative. In our study, 61.7% of patients were diagnosed on the basis of a
Robaei et al
Impact of Prior Corticosteroid Use on AK Outcomes
positive culture or histopathologic examination of corneal tissue, similar to rates previously reported in the literature.10,13 The remaining patients were diagnosed on the basis of confocal microscopy findings, or typical clinical signs (e.g., epitheliopathy, perineural infiltrates), disease course, and response to treatment. This method of case definition has been validated in 2 studies.3,24 However laboratory diagnosis can be increased to >80% for AK when polymerase chain reaction (PCR) is used to identify Acanthamoeba DNA.25,26 Further problems with the misdiagnosis of AK arise when the disease is polymicrobial.24,27 Bacterial infections that are only partially responsive to therapy are often treated with topical corticosteroids but may persist because of concurrent infection with AK. In addition approximately 10% of AK occurs in noncontact lens users, with or without a history of predisposing trauma, and AK must be considered as a cause of progressive (or indolent) microbial keratitis in this group, for whom the differential is fungal, mycobacterial, nocardia, and HSK, among other causes.28 It is disturbing that despite repeated reports of misdiagnosis in the ophthalmological literature, misdiagnosis of AK as HSK is still occurring15,19,29 and has barely improved in our study in the last decade compared with the previous decade. This finding identifies the need for both increased awareness among healthcare professionals and the routine use of PCR and in vivo confocal microscopy by trained observers as more sensitive diagnostic methods, permitting differentiation of Acanthamoeba, HSK, and fungal keratitis. Apart from the association of topical corticosteroids with diagnostic delay, there is some evidence to suggest that their use may result in increased pathogenicity of the Acanthamoeba organism. In an animal model of AK, McClellan et al30 elegantly demonstrated that the addition of topical corticosteroids, even at low doses, promotes an increase in the number of trophozoites, produced by excystment in infected corneal stroma.30 The resultant proliferation of the invasive form of the organism exposes patients to the risk of significantly greater corneal destruction through an increase in organism load, which may be greater than the increased chemotherapy effect on trophozoites compared with the more resistant cysts.
Study Limitations Given the retrospective nature of the study and reliance on accuracy and completeness of the clinical records, our study is limited by potential documentation bias. It is also difficult to completely adjust for potential confounders, such as disease severity at diagnosis. Whereas for this study we have stratified the disease on the basis of the corneal phenotype, this probably results from both differences in the virulence of the Acanthamoeba strain, as well from variations in the host immune response, neither of which can currently be quantified. Analysis of outcomes using more sophisticated disease stratification will be ideal when our understanding of such factors improves. Notwithstanding these limitations, an observational study of this type remains at present the only viable method of studying the role of topical corticosteroids before AK diagnosis.
In conclusion, our study highlights the critical importance of (1) having a high index of suspicion for AK in contact lens wearers and in noncontact lens users who have a persistent keratitis and (2) the need for judicious use of topical corticosteroids when the possibility of AK has not been definitively ruled out, which requires culture, confocal microscopy, and PCR. Older individuals and those with more advanced disease are independently prone to worse outcomes, and particular caution should be exercised in these cases. Prior corticosteroid use is associated with diagnosis delay and culminates in a great deal of ocular morbidity for affected patients, as evidenced by significantly higher rates of scleritis and keratoplasty in our cohort. It also imposes a significant burden on the healthcare system, with the need for a longer duration of follow-up and significantly more presentations to the outpatient department. Acknowledgments. The authors thank Massimo Bonaiti of the Doctor’s Laboratory for preparing the database of patients for whom Acanthamoeba cultures had been requested.
References 1. Naginton J, Watson PG, Playfair TJ, et al. Amoebic infection of the eye. Lancet 1974;2:1537–40. 2. Duguid IG, Dart JK, Morlet N, et al. Outcome of acanthamoeba keratitis treated with polyhexamethyl biguanide and propamidine. Ophthalmology 1997;104:1587–92. 3. Radford CF, Lehmann OJ, Dart JK. Acanthamoeba keratitis: multicentre survey in England 1992-6. National Acanthamoeba Keratitis Study Group. Br J Ophthalmol 1998;82: 1387–92. 4. Radford CF, Minassian DC, Dart JK. Acanthamoeba keratitis in England and Wales: incidence, outcome, and risk factors. Br J Ophthalmol 2002;86:536–42. 5. Osato MS, Robinson NM, Wilhelmus KR, Jones DB. In vitro evaluation of antimicrobial compounds for cysticidal activity against Acanthamoeba. Rev Infect Dis 1991;13(Suppl 5):S431–5. 6. O’Day DM, Head WS. Advances in the management of keratomycosis and Acanthamoeba keratitis. Cornea 2000;19: 681–7. 7. Park DH, Palay DA, Daya SM, et al. The role of topical corticosteroids in the management of Acanthamoeba keratitis. Cornea 1997;16:277–83. 8. Stern GA, Buttross M. Use of corticosteroids in combination with antimicrobial drugs in the treatment of infectious corneal disease. Ophthalmology 1991;98:847–53. 9. Oldenburg CE, Acharya NR, Tu EY, et al. Practice patterns and opinions in the treatment of acanthamoeba keratitis. Cornea 2011;30:1363–8. 10. Tu EY, Joslin CE, Sugar J, et al. Prognostic factors affecting visual outcome in Acanthamoeba keratitis. Ophthalmology 2008;115:1998–2003. 11. Patel DV, Rayner S, McGhee CN. Resurgence of Acanthamoeba keratitis in Auckland, New Zealand: a 7-year review of presentation and outcomes. Clin Experiment Ophthalmol 2010;38:15–20; quiz 87. 12. Kaiserman I, Bahar I, McAllum P, et al. Prognostic factors in Acanthamoeba keratitis. Can J Ophthalmol 2012;47:312–7. 13. Chew HF, Yildiz EH, Hammersmith KM, et al. Clinical outcomes and prognostic factors associated with acanthamoeba keratitis. Cornea 2011;30:435–41. 14. Saeed A, D’Arcy F, Stack J, et al. Risk factors, microbiological findings, and clinical outcomes in cases of microbial
5
Ophthalmology Volume -, Number -, Month 2014
15. 16. 17. 18. 19. 20. 21. 22. 23.
keratitis admitted to a tertiary referral center in Ireland. Cornea 2009;28:285–92. Yamazoe K, Yamamoto Y, Shimazaki-Den S, Shimazaki J. Visual outcome in Japanese patients with Acanthamoeba keratitis. Eye (Lond) 2012;26:517–22. Illingworth CD, Cook SD. Acanthamoeba keratitis. Surv Ophthalmol 1998;42:493–508. Liesegang TJ. Physiologic changes of the cornea with contact lens wear. CLAO J 28:12e27. Murphy PJ, Patel S, Marshall J. The effect of long-term, daily contact lens wear on corneal sensitivity. Cornea 20:264e269. Illingworth CD, Cook SD, Karabatsas CH, Easty DL. Acanthamoeba keratitis: risk factors and outcome. Br J Ophthalmol 1995;79:1078–82. Tabin G, Taylor H, Snibson G, et al. Atypical presentation of Acanthamoeba keratitis. Cornea 2001;20:757–9. Shukla Kent S, Robert MC, Tokarewicz AC, Mather R. Painless Acanthamoeba keratitis. Can J Ophthalmol 2012;47: 383–4. Kabra A, Lalitha P, Mahadevan K, et al. Herpes simplex keratitis and visual impairment: a case series. Indian J Ophthalmol 2006;54:23–7. Inoue Y, Shimomura Y, Fukuda M, et al. Multicentre clinical study of the herpes simplex virus immunochromatographic
24. 25.
26.
27. 28. 29. 30.
assay kit for the diagnosis of herpetic epithelial keratitis. Br J Ophthalmol 2013;97:1108–12. Bacon AS, Frazer DG, Dart JK, et al. A review of 72 consecutive cases of Acanthamoeba keratitis, 1984-1992. Eye (Lond) 1993;7(Pt 6):719–25. Lehmann OJ, Green SM, Morlet N, et al. Polymerase chain reaction analysis of corneal epithelial and tear samples in the diagnosis of Acanthamoeba keratitis. Invest Ophthalmol Vis Sci 1998;39:1261–5. Boggild AK, Martin DS, Lee TY, et al. Laboratory diagnosis of amoebic keratitis: comparison of four diagnostic methods for different types of clinical specimens. J Clin Microbiol 2009;47:1314–8. Cariello AJ, Passos RM, Yu MC, Hofling-Lima AL. Microbial keratitis at a referral center in Brazil. Int Ophthalmol 2011;31: 197–204. Allan BD, Dart JK. Strategies for the management of microbial keratitis. Br J Ophthalmol 1995;79:777–86. Butler TK, Males JJ, Robinson LP, et al. Six-year review of Acanthamoeba keratitis in New South Wales, Australia: 19972002. Clin Experiment Ophthalmol 2005;33:41–6. McClellan K, Howard K, Niederkorn JY, Alizadeh H. Effect of steroids on Acanthamoeba cysts and trophozoites. Invest Ophthalmol Vis Sci 2001;42:2885–93.
Footnotes and Financial Disclosures Originally received: November 22, 2013. Final revision: January 27, 2014. Accepted: January 28, 2014. Available online: ---.
Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Manuscript no. 2013-1948.
1
National Institute of Health Research Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, London, United Kingdom.
2
Moorfields Eye Hospital NHS Foundation Trust, London, United Kingdom.
3
EpiVision Ophthalmic Epidemiology Consultants, Bucks, United Kingdom.
6
Abbreviations and Acronyms: AK ¼ Acanthamoeba keratitis; CI ¼ confidence interval; HSK ¼ herpes simplex keratitis; OR ¼ odds ratio; PCR ¼ polymerase chain reaction; VA ¼ visual acuity. Correspondence: John K. G. Dart, MD, FRCOphth, Moorfields Eye Hospital NHS Foundation Trust, 162 City Rd., London, EC1V 2PD, UK. E-mail: j.dart@ucl. ac.uk.