Correlation of Pseudomonas aeruginosa Genotype With Antibiotic Susceptibility and Clinical Features of Induced Central Keratitis

Abstract

Purpose.: To determine the association of type III secretion system (T3SS) genotype with antibiotic susceptibility, serotypes, and clinical manifestation of centrally located Pseudomonas aeruginosa keratitis.

Methods.: Clinical characteristics of P. aeruginosa keratitis cases from 2001 to 2011 were analyzed. Each strain was serotyped and genotyped for T3SS exotoxin genes. Antibiotic sensitivity and minimum inhibitory concentrations were determined with agar dilution method. Prognostic factors affecting final visual outcomes and time to re-epithelialization were analyzed using linear and Cox regression models.

Results.: Forty-four invasive and 28 cytotoxic strains were identified. Invasive strains occurred mostly in males (P = 0.03) with older age (P = 0.009) and ocular surgery or trauma history (P = 0.006), poor presenting (P = 0.04) and final (P < 0.0001) best spectacle-corrected visual acuity (BSCVA), and larger infiltration size (P = 0.03). Cytotoxic strains were associated with contact lens wear (P < 0.0001). Poor prognostic factors for final BCVA after adjusting for sex, age, and contact lens wear included invasive strains (P = 0.02), larger infiltration area (P = 0.02), deeper infiltrations (P = 0.02), and presence of hypopyon (P = 0.09). Factors affecting complete re-epithelialization time included invasive strains (P = 0.02), infiltration area (P = 0.01), depth (P = 0.02), and hypopyon (P = 0.03) after adjusting for age and sex. Serotypes 2, 6, and 11 were more commonly found among all isolates. A trend to increased fluroquinolone resistance is noted for invasive strains, although not reaching statistically significant difference.

Conclusions.: Cytotoxic strains are associated with contact lens wear while invasive strains are related to poor prognosis and increased resistance to fluoroquinolones. Clinicians should be aware of the two different genotypes of P. aeruginosa affecting prognosis.

Introduction

Pseudomonas aeruginosa is the most commonly isolated Gram negative pathogen causing bacterial keratitis, particularly among contact lens wearers.^1–3 P. aeruginosa keratitis may progress rapidly and often results in permanent loss of vision.⁴ Previously, type III secretion system (T3SS) was identified as a significant virulence factor in the pathogenesis of pseudomonal keratitis.^5,6 Depending upon the type of T3SS exotoxin secreted, P. aeruginosa strains are categorized as invasive or cytotoxic genotypes, which affect corneal epithelial cells differently. Invasive strains carrying the exoS gene are capable of corneal epithelial cell invasion, while cytotoxic strains with the exoU gene encoding a phospholipase cause rapid necrotic death of host cells within 1 to 2 hours.^7,8 Analysis of genotype distribution found that isolates from patients with contact lens-associated microbial keratitis (CLMK) were mostly of the cytotoxic genotype.^9,10 Although both in vitro and animal models of pseudomonal keratitis have shown different manifestations between these two types of P. aeruginosa strains, only a few reports have indicated a difference in the clinical characteristics between the strains in humans.¹¹ Recently, strains with invasive genotypes were found to be associated with better initial visual acuity but poor visual outcomes at 3 months of follow-up.¹¹ This suggests the possibility of more drug resistance or difficulty in eradicating these organisms as invasive strains tend to “hide” within the host cells. This study compared the clinical manifestation of centrally located pseudomonal keratitis between these two genotypes and sought to determine if an association of antibiotic sensitivity and serotypes exists. 

Methods

Centrally located, that is, involving the pupillary axis, and laboratory-proven P. aeruginosa keratitis cases from 2001 to 2011 seen at the Department of Ophthalmology, National Taiwan University Hospital, Taipei, Taiwan, were included in this observational case series study. This retrospective study was approved by our institutional review board. All patients had a minimal follow-up period of at least 3 months. Presenting and final best spectacle-corrected visual acuities (BSCVA) were recorded and converted to logMAR units for analysis. Clinical manifestations such as infiltration size, depth, and presence of hypopyon were recorded using slit-lamp biomicroscopy. The widest diameters of the infiltrate at two meridians with axes perpendicular to each other were measured and multiplied to obtain the area of infiltration. The depth of the infiltration was recorded as percentage of total corneal thickness involved. The time for complete re-epithelialization was recorded and used for regression analysis. Patients requiring surgical interventions such as therapeutic penetrating keratoplasty or evisceration were regarded as medical treatment failures. 

Bacterial Strains and Culture Conditions

All strains had been confirmed by the central microbiological laboratory (National Taiwan University Hospital, Department of Laboratory Medicine, Bacteriology Section) as P. aeruginosa by green pigment production and a positive cytochrome oxidase test (BD Oxidase Reagent Droppers; BD Bioscience, Mexico City, Mexico). The isolates were kept as frozen stocks and were defrosted when needed for experimentation. Bacteria were grown to the stationary phase in 15 mL trypticase soy broth (TSB; BD Biosciences, Baltimore, MD, USA) at 37°C and then harvested by centrifugation at 9600g for 15 minutes. The pellet was washed twice with 5 mL normal saline and resuspended to various concentrations as required for experimentation. 

Genotyping

The presence of exoU and exoS genes was determined using multiplex PCR on all clinical P. aeruginosa strains. Using primer pairs reported by Ajayi et al.,¹² the following gene fragments were amplified: exoU (134 bp) and exoS (118 bp). Bacteria were grown overnight at 37°C in TSB, and DNA was isolated by using a DNA purification kit according to the manufacturer's protocol (Viogene, Sunnyvale, CA, USA). The PCR was set up as follows: 1 μL DNA template (500 ng), 1 μL total PCR primers (MDBio, Inc., Taipei, Taiwan, Republic of China) (a final 40 mM concentration of each primer), 12.5 μL 2× GoTaq Green Master Mix (Promega Corporation, Sunnyvale, CA, USA), and 10.5 μL sterile water. The negative control contained GoTaq Green Master Mix, no DNA, and 11.5 μL sterile water. The positive controls for the exoU and exoS genes were from the standard strains 6206 and 6294, respectively.^8,13 The standard reaction included 1 μL each of bacterial DNA, GoTaq Green Master Mix, and 10.5 μL sterile water. Polymerase chain reaction amplification was carried out as follows: initial denaturation at 94°C for 10 minutes; 35 cycles of 94°C for 45 seconds, 55°C for 45 seconds, and 72°C for 45 seconds; and a final extension step at 72°C for 7 minutes. The reaction was run in a 3% agarose gel (Sea Kem LE agarose; BMA, Rockland, MD, USA) with 1:10000 SYBR Safe DNA Gel Stain (Invitrogen, Carlsbad, CA, USA). Strains that were exoU+/exoS− were considered cytotoxic. ExoS+/exoU− strains were classified as invasive. Atypical strains that were either positive or negative for both genes were excluded from this study. 

Serotyping

Serotyping was performed on all clinical isolates using antiserum from Denka Seiken Co., Ltd. (Tokyo, Japan). Live P. aeruginosa bacterial cells were used. Agglutination reactions to specific antiserum indicated the O-antigen serotype. The Lanyi and Bergan¹⁴ digital coding corresponding to the Japan letter coding system (as specified by the instruction manual) was used. 

Antibiotic Susceptibility Testing and Minimum Inhibitory Concentration (MIC) Determination

P. aeruginosa susceptibility to aminoglycosides (gentamicin and amikacin), β-lactams (ceftazidime and piperacillin), and fluoroquinolones (ciprofloxacin, levofloxacin, gatifloxacin, and moxifloxacin) was determined in vitro by the Kirby-Bauer disc diffusion method. Since breakpoints for topical antimicrobials are not available, the National Committee for Clinical Laboratory Standards serum standards were used as the disc susceptibility breakpoint for each antimicrobial agent. Isolates of intermediate sensitivity were categorized as nonsusceptible organisms. 

The MIC (μg/mL) is the lowest concentration of an agent that will prevent or inhibit microbial growth. The MIC for each strain was determined in triplicate using the agar dilution method according to standards from the Clinical and Laboratory Standards Institute (CLSI, 2010) in Mueller-Hinton agar (BD Biosciences) containing gentamicin, amikacin, ceftazidime, piperacillin, ciprofloxacin, levofloxacin, gatifloxacin, and moxifloxacin in final concentrations of 128, 64, 32, 16, 8, 4, 2, 1, 0.5, 0.25, 0.12, 0.06, and 0 μg/mL. Inoculums of approximately 10⁴ colony-forming units (CFU) were transferred to the agar surface with a 96-pin applicator from a microtitration plate containing suspensions of the individual strains. The MIC was defined as the lowest concentration inhibiting visible bacterial growth as evaluated after 18-hour incubation at 37°C. Multidrug-resistant strains are strains resistant to two or more classes of antibiotics. 

Statistical Analysis

To compare the clinical features between different genotypes (invasive versus cytotoxic) of P. aeruginosa, Fisher exact tests were used when comparing binary variables. T-tests were used for comparing continuous variables. Linear regression models were used to evaluate the factors affecting visual outcome. To evaluate the factors affecting the time to re-epithelialization, Cox regression models were used. Complete re-epithelialization was considered to be the primary endpoint, and treatment failures requiring penetrating keratoplasty or evisceration were treated as censored events for Cox regression analysis. The candidate factors used for regression analyses included age, sex, genotypes (invasive versus cytotoxic), serotypes (serotype 11 versus others), contact lens wear, infiltration size, infiltration depth, and presence of hypopyon. A P value less than 0.05 was considered statistically significant. SAS 9.1 (SAS Institute, Inc., Cary, NC, USA) was used for all statistical analyses. 

Results

From 2001 to 2011, 72 eyes of centrally located P. aeruginosa keratitis from 72 patients were included in this study. Genotypic analyses found 44 invasive (exoS+exoU−) strains and 28 cytotoxic (exoS−exoU+) strains. All strains had mutually exclusive genotypes. Clinical characteristics at presentation and outcome measures of the two different strains are shown in Table 1. Comparatively, pseudomonal keratitis caused by invasive strains occurred more commonly in older males (P = 0.0009 and P = 0.03, respectively) with a history of ocular trauma or surgery (P = 0.006) and poor presenting and final visual acuity (P = 0.04 and P < 0.0001, respectively). The size of infiltration was significantly larger for invasive strains (P = 0.03). Hypopyon was also more frequently seen for invasive strains (P = 0.03). Cytotoxic strains, in contrast, occurred more often in younger females and were strongly associated with contact lens wear (P < 0.0001). Keratitis caused by invasive P. aeruginosa had significantly more cases with treatment failure indicating requirement for surgical interventions such as therapeutic penetrating keratoplasty or evisceration (P = 0.005). 

Table 1

 

Clinical Manifestation of Microbial Keratitis Caused by Different P. aeruginosa T3SS Genotypes

	T3SS Genotype		P Value
	Invasive, n = 44	Cytotoxic, n = 28
Age, y	51.5 ± 23.9	32.9 ± 19.7	0.0009, t-test
Sex, female	16, 36.3%	18, 64.3%	0.03, Fisher exact test
Contact lens wear	17, 38.6%	24, 85.7%	<0.0001, Fisher exact test
History of ocular trauma or previous surgery	14, 31.8%	1, 3.6%	0.006, Fisher exact test
Presenting logMAR of BSCVA	1.69 ± 1.08	1.15 ± 0.99	0.04, t-test
Final logMAR of BSCVA	1.62 ± 1.07	0.66 ± 0.67	<0.0001, t-test
Infiltration size, mm2	14.2 ± 11.8	8.1 ± 7.8	0.03, t-test
Maximal depth of infiltrate, % of cornea thickness	52.4 ± 24.5	43.5 ± 13.7	0.07, t-test
Presence of hypopyon	25, 56.8%	8, 28.6%	0.03, Fisher exact test
Treatment failure	11, 25.0%	0, 0%	0.005, Fisher exact test

 All continuous variables are expressed as mean ± standard deviation. All binary variables are expressed as case number (percentage of total cases).

Table 2 shows the serotype distribution among clinical isolates. The most commonly encountered serotypes among all clinical isolates were serotypes 11 (38.9%), 2 (26.4%), and 6 (15.3%). Further analysis of the isolates showed that invasive strains from non-CLMK cases were mostly of serotype 2. Cytotoxic strains from CLMK cases were predominantly of serotype 11. Seventy-five percent of invasive strains with serotype 2 from CLMK patients were sensitive to fluoroquinolones (Table 3). Overall, 71.4% and 75% of the invasive strains with serotype 11 from non-CLMK and CLMK patients were sensitive to aminoglycosides and fourth-generation fluoroquinolones (gatifloxacin and moxifloxacin), respectively (Tables 3, 4). 

Table 2

 

Distribution of P. aeruginosa Serotypes Among Clinical Isolates

Serotype	All Isolates	T3SS Genotype
	No. of Isolates	Invasive (exoS+)		Cytotoxic (exoU+)
		CL Wear +	CL Wear −	CL Wear +	CL Wear −
1	5	1	2	2	0
2	19	3	12	2	2
3	1	0	1	0	0
4	2	0	1	1	0
6	11	6	5	0	0
7	4	0	1	3	0
9	0	0	0	0	0
10	1	0	0	1	0
11	28	7	4	15	2
12	0	0	0	0	0
13	0	0	0	0	0
14	0	0	0	0	0
15	1	0	1	0	0
Total	72	17	27	24	4

Table 3

 

Percentage of CLMK Isolates Susceptible to Antibiotics Tested

Serotype	2		6		11
Antibiotics	Invasive	Cytotoxic	Invasive	Cytotoxic	Invasive	Cytotoxic
Gentamicin	100	100	100	NA	71.4	100
Amikacin	100	100	100	NA	71.4	100
Ceftazidime	100	100	83.3	NA	100	100
Piperacillin	100	100	100	NA	100	100
Ciprofloxacin	75	100	100	NA	100	100
Levofloxacin	75	100	100	NA	100	100
Gatifloxacin	75	100	100	NA	100	100
Moxifloxacin	75	100	83.3	NA	100	93.3

 NA, not applicable.

Table 4

 

Percentage of Non-CLMK Isolates Susceptible to Antibiotics Tested

Serotype	2		6		11
Antibiotics	Invasive	Cytotoxic	Invasive	Cytotoxic	Invasive	Cytotoxic
Gentamicin	91.7	100	60	NA	100	50
Amikacin	100	100	60	NA	100	100
Ceftazidime	100	100	80	NA	100	100
Piperacillin	100	100	80	NA	100	100
Ciprofloxacin	100	100	100	NA	100	100
Levofloxacin	100	100	100	NA	100	100
Gatifloxacin	100	100	100	NA	75	100
Moxifloxacin	91.7	100	100	NA	75	50

The antibiotic sensitivities with MIC values are listed in Tables 5 and 6 for CLMK and non-CLMK isolates, respectively. For CLMK isolates, all isolates were susceptible to aminoglycosides (gentamicin and amikacin). Susceptibility to moxifloxacin was 100% for cytotoxic strains and 94.1% for invasive strains. For non-CLMK isolates, all cytotoxic strains showed 100% susceptibility to most antibiotics tested, except for 75% to gentamicin and 75% to moxifloxacin. For invasive strains from non-CLMK isolates, susceptibility to gentamicin, amikacin, and moxifloxacin was 85.1%, 88.8%, and 81.5%, respectively. In addition, some highly resistant invasive strains from non-CLMK isolates were found as indicated by the high maximum MIC values. There were seven multidrug-resistant strains, six of which were of invasive genotype. Two of the six multidrug-resistant invasive strains required therapeutic penetrating keratoplasty or evisceration. 

Table 5

 

Antibiotic Susceptibility of CLMK Isolates With Different T3SS Genotypes

	Invasive Strains of CLMK Isolates, 17 Strains					Cytotoxic Strains of CLMK Isolates, 24 Strains
Antibiotics	MIC50	MIC90	Min MIC	Max MIC	Susceptible, %	MIC50	MIC90	Min MIC	Max MIC	Susceptible, %
Gentamicin	4	4	1	8	100	4	4	2	4	100
Amikacin	4	8	2	8	100	4	8	2	8	100
Ceftazidime	2	4	2	4	100	4	4	2	8	100
Piperacillin	4	8	1	8	100	4	8	2	16	100
Ciprofloxacin	0.25	1	0.125	2	100	0.25	0.25	0.125	1	100
Levofloxacin	0.5	2	0.25	8	94.1	0.5	2	0.5	4	100
Gatifloxacin	1	2	0.25	8	94.1	1	2	0.5	2	100
Moxifloxacin	2	4	0.25	8	94.1	2	2	1	4	100

 MIC₅₀, the minimum inhibitory concentration inhibiting growth in 50% of all isolates; MIC₉₀, the minimum inhibitory concentration inhibiting growth in 90% of all isolates; Min, minimal; Max, maximal.

Table 6

 

Antibiotic Susceptibility of Non-CLMK Isolates With Different T3SS Genotypes

	Invasive Strains of Non-CLMK Isolates, 27 Strains					Cytotoxic Strains of Non-CLMK Isolates, 4 Strains
Antibiotics	MIC50	MIC90	Min MIC	Max MIC	Susceptible, %	MIC50	MIC90	Min MIC	Max MIC	Susceptible, %
Gentamicin	2	32	0.5	128	85.1	4	8	2	8	75.0
Amikacin	4	8	1	128	88.8	4	16	2	16	100
Ceftazidime	2	8	2	32	92.5	2	4	2	4	100
Piperacillin	4	16	2	128	92.5	4	8	2	8	100
Ciprofloxacin	0.25	0.5	0.125	16	92.5	0.25	0.5	0.06	0.5	100
Levofloxacin	0.5	2	0.25	32	92.5	0.5	2	0.5	2	100
Gatifloxacin	1	2	0.25	16	92.5	1	2	0.5	2	100
Moxifloxacin	2	4	0.5	64	81.5	2	8	1	8	75.0

Prognostic Factors for Visual Outcome

Table 7 shows the results of linear regression analysis for the logMAR of the final BSCVAs. Without adjustment, older patients (P < 0.0001) infected with invasive strains (P < 0.0001) presenting with hypopyon P = 0.007), large infiltration area (P = 0.002), and deep infiltration depth (P = 0.008) had poor final visual acuity. After adjustment for age and sex, invasive strains presenting with hypopyon, larger infiltration areas, and deeper infiltration depths were still significantly associated with poor visual outcomes (P = 0.006, P = 0.04, P = 0.01, P = 0.02, respectively). Contact lens wearers had better visual outcomes than noncontact lens wearers (P = 0.03 after adjustment for age and sex). After adjusting for age, sex, and contact lens wear, patients infected with serotype 11 tended to have better visual outcomes (P = 0.02). After adjusting for age, sex, and contact lens wear, invasive strains, larger infiltration areas, and deeper infiltration depths at presentation were all correlated with poorer final BSCVAs (P = 0.02 for each factor). 

Table 7

 

Linear Regression Analysis for Logarithm of the Minimal Angle of Resolution of Final Best Spectacle-Corrected Visual Acuity

	Unadjusted		Adjusted for Age and Sex		Adjusted for Age, Sex, and Contact Lens Wear
	Coefficient	P Value	Coefficient	P Value	Coefficient	P Value
Age, y	0.021	<0.0001	-	-	-	-
Sex, F = 0, M = 1	0.143	0.57	-	-	-	-
Contact lens wear	−0.987	<0.0001	−0.592	0.03	-	-
Strain, invasive = 0, cytotoxic = 1	−0.962	<0.0001	−0.731	0.006	−0.600	0.02
Serotype, serotype 11 = 0, others = 1	0.951	0.0003	0.681	0.008	0.600	0.02
Infiltration area at presentation, mm2	0.037	0.002	0.027	0.01	0.026	0.02
Infiltration depth at presentation, % of cornea thickness	0.016	0.008	0.014	0.02	0.013	0.02
Hypopyon	0.669	0.007	0.479	0.04	0.410	0.09

Prognostic Factors for Re-Epithelialization

Among the 72 eyes in this study, 61 (84.7%) had complete re-epithelialization and 11 (15.3%) received penetrating keratoplasty or evisceration due to treatment failures. The mean healing time was 43.5 ± 20.2 days for the 61 eyes with complete re-epithelialization. Table 8 shows the results of Cox regression analysis for complete re-epithelialization. Younger-aged (P = 0.03), female (P = 0.01), contact lens–wearing patients (P = 0.007 after adjusting for age and sex) tended to exhibit faster re-epithelialization. Patients infected with invasive strains experienced poor re-epithelialization compared with those infected with cytotoxic strains (P = 0.02 after adjusting for age and sex). Larger infiltration areas and deeper infiltration depths at presentation were both correlated with delayed re-epithelialization (P = 0.02 for each after adjusting for age, sex, and contact lens wear). 

Table 8

 

Cox Regression Analysis for the Time to Complete Re-epithelialization

	Unadjusted			Adjusted for Age and Sex			Adjusted for Age, Sex, and Contact Lens Wear
	Hazard Ratio	95% CI	P Value	Hazard Ratio	9% CI	P Value	Hazard Ratio	% CI	P Value
Age, y	0.987	0.976–0.999	0.03	-	-	-	-	-	-
Sex, F = 0, M = 1	0.493	0.283–0.860	0.01	-	-	-	-	-	-
Contact lens wear	2.957	1.522–5.204	0.0007	2.607	1.264–4.937	0.007	-	-	-
Strain, invasive = 0, cytotoxic = 1	2.784	1.558–4.975	0.0005	2.137	1.138–4.011	0.02	1.691	0.870–3.286	0.12
Serotype, serotype 11 = 0, others = 1	0.490	0.269–0.892	0.02	0.621	0.330–1.167	0.14	0.823	0.428–1.585	0.56
Infiltration area at presentation, mm2	0.943	0.904–0.984	0.006	0.952	0.917–0.989	0.01	0.955	0.919–0.993	0.02
Infiltration depth at presentation, % of cornea thickness	0.980	0.966–0.994	0.006	0.982	0.967–0.997	0.02	0.981	0.966–0.997	0.02
Hypopyon	0.520	0.303–0.892	0.02	0.527	0.299–0.926	0.03	0.616	0.349–1.089	0.10

Discussion

In this study, the clinical manifestations of centrally located P. aeruginosa keratitis and antibiotic susceptibility were compared between the T3SS genotypes. Presently, only one major study, using isolates collected from the Steroids for Corneal Ulcer Trial (SCUT), has compared the clinical characteristics, specifically visual outcomes, between invasive and cytotoxic P. aeruginosa strains causing bacterial keratitis.¹¹ The SCUT study concluded that determining the T3SS genotype may be beneficial in assisting clinical management decisions. However, antibiotic susceptibility and clinical outcomes besides visual acuity were not included. Since both peripherally and centrally located corneal ulcers were included in that study, visual outcomes had to be adjusted for ulcer location, a major confounding factor. Furthermore, the number of cytotoxic strains included was limited (18 out of 74 in SCUT study versus 28 out of 72 in the present study), and analysis of differences in antibiotic sensitivity between the genotypes was not done. To the best of our knowledge, our study may be the only clinical study to date that concurrently analyzed antibiotic susceptibility, strain serotypes, and the clinical outcome of all centrally located P. aeruginosa ulcers between different T3SS genotypes. 

Our results show that pseudomonal keratitis caused by invasive strains occurred more frequently in elderly males with average age of approximately 51 years. In a small case series of P. aeruginosa keratitis isolates, invasive strains were found also predominately in males over the age of 50.¹⁵ Ocular trauma involving males had previously been identified as a significant risk factor for microbial keratitis development for all types of pathogens.¹⁶ Pseudomonal keratitis due to invasive strains, in particular, was found to be significantly related to previous eye trauma or surgery in this study. Invasive strains often presented with poor initial visual acuity and caused more fulminate disease as lesions tended to be large and deep with the presence of severe anterior chamber reactions. The final visual outcome of invasive strains is thus significantly worse than those caused by cytotoxic strains. Previously, an analysis of P. aeruginosa strains collected in the SCUT yielded similar results, showing that invasive strains presented with larger sizes of infiltration.¹¹ However, patients infected with invasive strains had better presenting visual acuity but worse final visual outcomes at 3 months as compared to those patients infected with cytotoxic strains.¹¹ This may be indicative of poor therapeutic response for invasive strains. Greater invasiveness was also found to be associated with worse final visual outcomes.¹¹ Thus, results from both SCUT and our study indicate greater disease severity with poor clinical outcomes for invasive strains. 

In contrast to invasive strains, cytotoxic strains were significantly associated with contact lens wear, which explains the younger age and female preponderance. The association of contact lens wear and cytotoxic strains has been previously demonstrated, as isolates from CLMK cases were mostly cytotoxic.^9,10 Reasons for the association of cytotoxic strains with contact lens wear need further investigation. Although the quantity of bacteria adhering to various contact lens materials was not significantly different between the two genotypes,⁹ resistance of cytotoxic strains to contact lens disinfection solutions has been reported and may be one explanation for its predominance among CLMK cases.¹⁷ With the increasing popularity of contact lens wear, even for emmetropic individuals desiring cosmetic effects, it is worthwhile to point out that the commonly found cytotoxic strains among contact lens wearers generally caused less severe clinical disease with better final visual outcome even after adjustment for age and sex. These results are also supported by Borkar et al.,¹¹ who noted a three-line improvement in final visual acuity with highly cytotoxic strains. 

Among our total isolates, serotypes 11, 2, and 6 were most commonly identified. Although the finding was not statistically significant, serotypes 2 and 6 were predominately seen in invasive strains, while serotype 11 was more common in cytotoxic strains. Zhu et al.¹⁸ found serotypes 11 and 6 in 28% and 20% of P. aeruginosa isolates from microbial keratitis cases, respectively. From the same study, serotype 11 was significantly correlated with cytotoxic strains.¹⁸ Strains from noncontact lens–related microbial keratitis were found to be mostly invasive strains that were highly resistant to fluroquinolones.¹⁰ We found that approximately one-fourth of our invasive strains with the serotype 11 from non-CLMK isolates were resistant to fourth-generation fluoroquinolones. We also noted that 25% of the invasive strains with serotype 2 from CLMK cases were resistant to all fluoroquinolones. Therefore, the invasive strains of serotypes 2 and 11 tended to be more resistant to fluoroquinolones. Thus, a relatively quick serotype determination along with genotype determination may be useful clinically as a prognostic indicator for suspected pseudomonal keratitis. 

Analyses of the prognostic factors for final BSCVA and epithelial healing showed that invasive genotypes, greater infiltration size and depth, and the presence of hypopyon were significantly associated with poor final BSCVA and delayed epithelial healing. After adjustments for age and sex, these same factors were again significant in influencing visual acuity outcomes and healing times. Signs of more aggressive disease such as larger infiltration size and depth with hypopyon certainly contributed to the poor visual outcomes and delayed re-epithelialization. The SCUT study by Borkar et al.¹¹ did not specify the incidence of treatment failures, but our study showed that patients infected with invasive strains are at high risk of treatment failure indicating requirement for corneal transplants or even evisceration. Clinical prognosis is thus significantly correlated to T3SS genotypes. 

Although the T3SS genotype may be a significant prognostic factor clinically, clinical and in vitro outcomes may not be similarly influenced by the two genotypes. Previous in vitro cell culture models have shown that cytotoxic strains were more virulent in causing rapid corneal epithelial cell death.⁸ In vivo mouse models found that corneas infected with cytotoxic strains showed ring infiltrates with edema and ectasia.¹⁶ Mice infected with invasive strains have mostly focal infiltrations.¹⁹ Infection by both strains causes severe anterior chamber reaction.¹⁹ Other virulence factors for P. aeruginosa such as biofilm formation, production of las B, elastase, P. aeruginosa small protease (PASP), and alkaline protease have also been shown to affect the antibiotic sensitivity or clinical outcome of not just ocular disease but also other systemic diseases such as lung and urinary tract infections.^10,20–25 Invasive strains with strong biofilm formation are more likely to be resistant to fluroquinolones.¹⁰ Overall, experimental models of P. aeruginosa infections have shown variable results. At present, results from our study and that of Borkar et al.¹¹ concur in demonstrating a generally worse prognosis for invasive strains. Thus, clinicians should be aware of the existence of two different genotypes of P. aeruginosa strains at presentation and should follow up with appropriate treatment. 

The treatment of pseudomonal keratitis generally involves aminoglycosides or fluoroquinolones. However, strains from our non-CLMK cases were more commonly found to be resistant to aminoglycosides or fluroquinolones. Monotherapy with fluoroquinolones is generally favored over fortified aminoglycosides due to lower epithelial toxicity and commercial availability.^26–28 Nonetheless, emerging resistance to fluoroquinolones has been noted.^29–31 The invasive strains from our non-CLMK cases had relatively high maximum MIC values for all tested antibiotics. Nearly all cytotoxic strains were sensitive to most fluroquinolones except for moxifloxacin, while only approximately 81% to 94% of invasive strains were sensitive to fluroquinolones. This is a very disturbing sign, as moxifloxacin is one of the newer potent fourth-generation fluoroquinolones. Murine P. aeruginosa keratitis models comparing the effectiveness of fluroquinolones versus aminoglycosides between invasive and cytotoxic strains demonstrated that aminoglycosides such as tobramycin were less effective in eradicating invasive strains.³² Although the resistance mechanisms to aminoglycosides were not within the scope of this study, the resistance mechanism of P. aeruginosa to aminoglycosides may be quite complicated and may involve inactivating enzymes encoded by plasmids or transposons, impermeability of the antibiotics, and also other adaptive resistance mechanisms related to virulence of the bacteria.^23–25,33 Since invasive strains are capable of intracellular invasion and replication, fluoroquinolones may be the preferred choice of treatment due to their ease of cellular membrane penetration.³² Aminoglycosides are less cell permeable and thus less effective than fluoroquinolones in treating invasive strains.³⁴ Due to the tendency of invasive strains to be multidrug resistant, we emphasize the importance of early medical treatment and possible aggressive surgical debridement for patients infected with invasive strains to prevent uncontrollable intraocular infection and eventual evisceration. 

In conclusion, we have demonstrated that cytotoxic strains are associated with contact lens wear while invasive strains occur more frequently in older males with a history of ocular trauma or surgery. Invasive strains also tend to result in treatment failures and thus poor visual outcomes. Increasing resistance to fluoroquinolones, which may be the treatment of choice for invasive strains, is alarming. Clinicians are advised to raise awareness of the two genotypes of P. aeruginosa causing keratitis of variable prognoses. Continued monitoring of antibiotic resistance and close follow-up of patients suspected of having P. aeruginosa infections, especially of the invasive genotype, are required to prevent rapid progression and permanent vision loss. 

Acknowledgments

We thank Suzanne M. J. Fleiszig, OD, PhD, University of California-Berkeley, School of Optometry and Programs in Vision Sciences, Infectious Diseases and Immunity, and Microbiology for her generosity in providing the standard strains 6206 and 6294. 

Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Seattle, Washington, United States, May 5–9, 2013. 

Supported in part by the National Science Council, the Executive Yuan, Taiwan, NSC102-2628-B-002-051-MY3; the National Center of Excellence for Clinical Trial and Research Grant DOH101-TD-B111-001; and a grant from the Taipei Tzuchi Hospital, The Buddhist Tzuchi Medical Foundation, TCRD-TPE-101-34. The authors have no proprietary or commercial interest in the research presented herein. 

Disclosure: E.P. Shen, None; Y.-T. Hsieh, None; H.-S. Chu, None; S.-C. Chang, None; F.-R. Hu, None