Fusarium keratitis: genotyping, in vitro susceptibility and clinical outcomes

Abstract

Purpose

To determine differences in the clinical characteristics and antifungal susceptibility patterns among molecularly characterized ocular Fusarium sp isolates.

Methods

58 Fusarium isolates obtained from 52 eyes of 52 patients were retrieved from the Bascom Palmer Eye Institute’s (BPEI) ocular microbiology laboratory and grown in pure culture. These isolates were characterized based on DNA sequence analysis of the ITS1/2 and rDNA regions. Antifungal susceptibilities were determined for each isolate using broth microdilution methods and the corresponding medical records were reviewed to determine clinical outcomes.

Results

Fusarium (F.) solani isolates had significantly higher voriconazole MIC₉₀ values than F. non-solani organisms (16 and 4ug/ml, respectively). F. solani isolates also exhibited a significantly longer time to cure (65 vs 40.5 days), a worse follow up BCVA (20/118 vs 20/36), and increased need for urgent surgical management (7 vs 0 penetrating keratoplasties) when compared to F. non-solani isolates.

Conclusions

This is the first report to examine the correlation between ocular genotyped Fusarium species and clinical outcomes. It supports the overall worse prognosis for F. solani versus F. non-solani isolates, including higher voriconazole resistance by the former. The clinical implementation of molecular-based diagnostics and antifungal efficacy testing, may yield important prognostic and therapeutic information that could improve the management of fungal ocular infections.

Introduction

Following the recent global outbreak of contact lens associated Fusarium keratitis (2005–2006) there has been a renewed interest and significant focus of research establishing epidemiology, classification, diagnosis and treatment standards. It is now clear that the incidence of Fusarium keratitis has increased dramatically over the past 4 decades, accounting for up to 50% of all microbial keratitis cases in tropical climates. ^1–9 This increasing incidence is multi-factorial and is believed to be due to increased awareness and changes in risk factor profiles throughout the global population, including an increase in the use of topical steroids and antibacterial agents, as well as an increase in surgical procedures, contact lens use, ocular trauma, chronic ocular surface diseases, and immune compromised patients ^{6, 10–13}.

Among the Fusarium species that are pathogenic to the eye, F. solani is the most common, followed by F. oxysporum, F. dimerum, F. incaratum-equiseti, and Gibberella fujikuroi. ^{14, 15} Microbiologic techniques have long served as the diagnostic gold standard in the setting of fungal keratitis. Such techniques are capable of reliably differentiating among the different molds and yeasts causing keratitis. Although highly accurate at genus identification, (Fusarium, Aspergillus, Candida, ect.) morphologic classification of Fusarium isolates to the species level using microbiologic techniques (F. solani, F. oxysporum, etc) is problematic and inconsistent due to the large degree of morphologic variability demonstrated at different growth stages.^{15, 16} In light of the increasing incidence of Fusarium keratitis, this inconsistency in morphologic identification has generated significant interest in finding a more consistent and reliable basis for organism classification. Recent reports have demonstrated genotypic identification systems as a more accurate and reproducible means of properly identifying ocular Fusarium pathogens ^{14, 17} and a clear consensus has emerged ^18–21 that DNA sequence-based methods will be essential for rapid species identification of the Fusarium genus in clinical laboratories²².

The question remains: is it clinically important to accurately differentiate between the different Fusarium species causing keratitis? Recent reports suggest that filamentous fungi harbor unique species-specific in vitro susceptibility profiles to the existing and emerging antifungal agents ^23–25. Within the Fusarium genus and among isolates pathogenic to the eye, however, these species-specific antifungal susceptibility profiles have not been firmly established ^{26, 27}. Furthermore, although the etiology and epidemiology of Fusarium keratitis has been well studied ^{3–5, 28, 29}, very little has been revealed about the differences in clinical characteristics and outcomes associated with infections due to different Fusarium species.

In this study, we investigate and compare the in vitro susceptibility profiles and provide the first report of the clinical characteristics and outcomes among ocular pathogenic Fusarium isolates classified by genotypic analysis. Such information is useful from a prognostic, diagnostic and therapeutic viewpoint to determine the level of pathogen identification that has the potential to directly influence practice patterns and patient outcomes.

Methods

Isolates

Fifty-eight Fusarium sp isolates, representing 52 patients, were retrospectively selected from the Bascom Palmer microbiology isolate library based on the morphologic species classification to include approximately 20 isolates of each: F. solani, F. oxysporum and F. species without further designation. We included consecutive isolates from May 2005 to June 2007. In addition, this group of isolates was supplemented with samples dating back to April 2000 to achieve the desired equal distribution from each morphological designation. The source distribution of the isolates was cornea (41), aqueous humor (4), vitreous (1), contact lens (8), and contact lens case (4). A recent paper from the authors of this study ¹⁵, explains in detail the genotyping procedures, conforming to the most recent Fusarium species complexes classification system^{14, 21}. The quality control (QC) reference strains Candida albicans ATCC 90028, Candida parapsilosis ATCC 22019, Candida krusei ATCC 6258 and Aspergillus flavus ATCC 204304 were included as control isolates for the Clinical and Laboratory Standards Institute (CLSI) (Formerly NCCLS) testing method.

Antifungal Agents

The methodology as described in the CLSI document M38-A reference was followed for broth dilution antifungal susceptibility testing of filamentous fungi. Additive drug dilutions were prepared to yield twice the final strength required for the test. Stock solutions of amphotericin B (AMB) and natamycin (NAT) at their final concentrations were frozen at −80°C until needed. The agents evaluated in this study have well-established microdilution MIC ranges for these QC strains (M38-A). Thirty isolates were also sent to the University of Texas fungal testing laboratory for voriconazole (VOR) susceptibility testing and identity confirmation (following CLSI M38-A methodology).

Medical records review

Following Institutional Review Board (IRB) approval, the medical records from 50 of the 52 affected patients were available for review. All patients included in the chart review had positive corneal culture results for Fusarium sp growth or positive culture from contact lens paraphernalia plus clinical and confocal microscopy appearance of filamentous fungal corneal infection.

The follow up BCVA of patients who underwent penetrating keratoplasty (PK), was the last BCVA measured prior to surgery. For this group of patients, the time to cure was assigned as an estimate, which was the longest time to cure among clinically treated patients.

The follow up BCVA of the patients treated with topical or oral antimicrobial agents, was the VA measured when the patient was considered cured (inactive corneal scar with intact epithelium).

Snellen visual acuities were converted to logMAR by the formula logMAR acuity = minus log (numerator Snellen/denominator Snellen) for the purpose of data analysis. Vision levels classified as count fingers, hand motion, light perception, and no light perception were assigned Snellen acuities of 1/200, .5/200, 20/20,000, and 20/200,000. The corresponding logMAR VAs were 2.3, 2.6, 3.0, 4.0 respectively, similar to a previously published scale ³⁰.

Statistical analyses were performed using SPSS software version 15.0 (SPSS Inc, Chicago, Illinois, USA). Tests of significance were two-tailed with p ≤0.05 for all tests. Univariate comparisons between F. solani and F. non-solani were performed using the two-sided Student t-test for continuous variables. Non-parametric tests (Mann-Whitney Test) were used for analyzing the data which does not follow normal distribution. Chi-square or Fisher’s exact test was performed for categorical variables. Analysis of Variance (ANOVA) was performed for multiple comparisons.

Results

Microbiological and Molecular Identification

Identification of the organisms as belonging to the Fusarium genus took a mean of 3.6 (±2.9) days, with a range of 2 to 12 days, whereas identification to the species level using microbiological techniques required a mean of 8 (±3.7) days.

Molecular identification took up to 24 hours to identify the isolates to the species level. Based on genotype, the 58 isolates were classified into one of five groups: F. solani species complex (FSSC) (75%), F. oxysporum species complex (FOSC) (16%), F. incarnatum-equiseti species complex (FIESC) (5%), F. dimerum species complex (FDSC) (2%), and Fusarium species not otherwise identified (2%). Sequence analysis showed 15 distinct sequences among the 58 isolates. For the purposes of data analysis, and due to isolate characteristics homogeneity, the isolates were divided into 2 groups: F. solani and F. non-solani, which includes all species other than F. solani.

Clinical Characteristics and Epidemiology

The 58 samples were isolated from 52 eyes of 52 patients. The clinical characteristics are shown in table 1.

Table 1.

Clinical characteristics of Fusarium sp keratitis

		Fusarium sp total (N=50)	F. solani (N=38)	F. non-solani (N=12)	p-value
Gender	Male	22 (44%)	17 (45%)	5 (42%)	1.00*
	Female	28 (56%)	21 (55%)	7 (58%)
Age (mean(SD))		39 (19)	39 (20)	40 (18)	0.97**
CL wear		33 (66%)	24 (63%)	9 (75%)	0.51*
Trauma		9 (18%)	7 (18%)	2 (17%)	1.00*
Infiltrate’s location					1.00*
Central 6mm		36 (72%)	27 (71%)	9 (75%)
Peripheral		14 (28%)	11 (29%)	3 (25%)
Smear (+) results
Cornea		14/34 (41%)	11/29 (38%)	3/5 (60%)	0.63*
CL smear		1/12 (8%)	1/5 (20%)	0/7 (0%)	0.42*
A/C		0/3 (0%)	0/3 (0%)	0	N/A
Vitreous		0/1 (0%)	0/1 (0%)	0	N/A
Total		15/50 (30%)	12/38 (32%)	3/12 (25%)	1.00*
Time to (+) culture† (days)		3.6 (2.9) (Range: 2–12)	3.5 (3.1) (range: 2–12)	3.8 (2.1) (range: 2–11)	0.8**
Treatment delay (days)		9.7 (7) (range: 0–35)	10.3 (7.6) (range: 0–35)	7.2 (4.3) (range: 1–12)	0.22**
Time to cure (days)		60 (range: 10–100)	65 (range: 10–100)	40.5 (range: 10–92)	0.035***
Initial BCVA LogMAR (Snellen)		0.83 (20/135) (range: 20/20-LP)	0.89 (20/155) (range: 20/20-LP)	0.65 (20/89) (range: 20/25-HM)	0.38**
Follow up BCVA LogMAR (Snellen)		0.65 (20/89) (range: 20/20-LP)	0.77 (20/118) (range: 20/20-LP)	0.25 (20/36) (range:20/20–20/100)	0.01**
Penetrating Keratoplasty		7 (14%)	7 (18%)	0 (0%)	0.17*
Follow up BCVA<20/200		9 (19%)	9 (25%)	0 (0%)	0.09*

Treatment delay= time from the onset of symptoms to the initiation of antifungal agents.

Time to cure= duration of treatment necessary to eliminate evidence of clinically active infection.

^*Fisher’s exact test

^**t-test

^***Mann-Whitney test

^†Genus level (Fusarium sp)

The follow up time in this study had a mean of 7.3 (±6) months, ranging from 15 days to 24 months. The time from the onset of symptoms until the first evaluation at BPEI ranged from 1 to 33 days, with a mean of 6.6 (±7.1) days. No patients had additional pathology at baseline that accounted for decreased visual acuity at any time during the management.

The infiltrates involved the central 6mm of the cornea in 36 of 50 cases, with no difference observed between F. solani (71%) and F. non-solani isolates (75%). Compared to peripheral infiltrates, central infiltrates were associated with a statistically significant worse initial BCVA (20/235 and 20/33, respectively, p<0.001, t-test) and final BCVA (20/142 and 20/28, respectively, p=0.001, t-test) when compared to peripheral infiltrates.

Risk factors

Risk factors for infectious keratitis were found in 49 patients (Figure 1). Among the 38 patients with F. solani infection, 24 were CL wearers (63%) and among the 12 patients with F. non-solani infection, 9 were CL wearers (75%). The frequency of trauma history between F. solani and F. non-solani infections was similar (20%) in both groups. Three patients had history of previous topical steroid use (2 in the F. solani and 1 in the F. non-solani group).

Figure 1.

Risk factors among Fusarium sp keratitis patients*.

*There were 6 patients that had more than 1 risk factor.

Management

Complete information of the treatment protocols was present in the records of 48 patients. Both topical and systemic antimicrobial agents were used for treatment of these infections. Topical natamycin 5% (Natacyn®, Alcon Labs, Fort Worth, TX) was used in 41 cases, and was used as monotherapy in 26 cases and in combination with other antimicrobial agents in 15 cases. Topical voriconazole 1% (compounded from Vfend®, Pfizer Inc, New York, NY) was prescribed in 7 cases, oral fluconazole 100mg (generic fluconazole, Cipla Pharmaceuticals, Mumbai, India) BID in 6 cases, topical amphotericin B 0.15% (compounded from generic amphotericin B, X-Gen Pharmaceuticals Inc, Big Flats, NY) in 4 cases, moxifloxacin 0.5% (Vigamox®, Alcon Labs, Fort Worth, TX) in 3 cases, combination of fortified cefazolin 5% (compounded from generic cefazolin, Abraxis BioScience, Los Angeles, CA) with tobramycin 1.4% (compounded from Ak-Tob® 0.3%, Akorn, Lake Forest, IL + generic tobramycin, Hospira Pharmaceuticals, Lake Forest, IL) in 3 cases each and oral valacyclovir 500mg (Valtrex®, GlaxoSmithKline, London, UK) 3x/day in 1 case.

Therapeutic penetrating keratoplasty (PK) was necessary in 7 cases, all of which were caused by F. solani isolates. The time from diagnosis to PK ranged from 4 to 57 days, with a mean of 28 (±22) days. The indications for PK were perforation in 4 cases and severe infiltrates non-responsive to clinical therapy in 3 cases. Four of the therapeutic grafts failed (three due to rejection and one to stromal melting) and repeat PK was successful in 2 cases.

Outcomes

Treatment delay (Figure 2) and time to cure (Figure 3) had a significant correlation with the follow up BCVA of F. solani versus F. non-solani isolates.

Figure 2.

Correlation of treatment delay and follow up BCVA for F. solani vs F. non-solani isolates.

Figure 3.

Time to cure versus follow up BCVA in patients with Fusarium sp keratitis

Among the 26 patients that received natamycin as monotherapy, 22 (85%) ended with BCVA ≥ 20/40, and the other 4 patients (15%) had a follow up BCVA between 20/60 and 20/80 (Table 2).

Table 2.

Treatment regimen versus visual outcomes.

	Nata Only (N=26)	Nata+Azoles (N=11)	Nata+Others (N=4)	Others (N=7)	P-value
Mean Initial BCVA	0.66 (0.73) 20/91	1.27 (0.99) 20/372	0.54 (0.38) 20/69	0.92 (0.99) 20/166	0.20*
Mean Follow up BCVA	0.24 (0.46) 20/35	1.74 (1.24) 20/1099	0.11 (0.08) 20/26	0.73 (1.12) 20/107	<0.001*
Mean BCVA improvement	0.43 (0.67)	−0.47† (1.23)	0.43 (0.31)	0.32 (0.42)	0.03*
Initial BCVA Better than or equal to 20/40	11 (42.3%)	1 (9.1%)	2 (50%)	2 (28.6%)	0.22**
Follow up BCVA Better than or equal to 20/40	22 (88%)	3 (27.3%)	4 (100%)	4 (66.7%)	0.001**
BCVA improve 2 or more lines	14 (56%)	2 (18.2%)	4 (100%)	3 (50%)	0.032**

^*Analysis of Variance (ANOVA) was performed.

^**Chi-Square Test

^†Mean follow up BCVA was worse than mean initial BCVA.

Natamycin was used as combined therapy in 15 cases, 11 used natamycin with azoles. Eight of these patients (73%) underwent PK or had final BCVA < 20/400. The other three patients (27%) had a good final outcome, with BCVA ≥ 20/40.

Among the 9 patients with a follow up BCVA <20/200, risk factors were identified in 8 cases (6 CL wearers, 2 used previous topical steroids, 2 immunosupressed patients) and PK was performed in 6 cases. The initial BCVA in this group was LogMAR 1.25 (20/355) and follow up BCVA 1.66 (20/915).

Antifungal Susceptibility Testing

Minimum inhibitory concentrations for 50 and 90% of the isolates (MIC_{50 and} MIC₉₀), medians, and MIC ranges for F. solani and F. non-solani isolates are listed in table 3.

Table 3.

Voriconazole, Natamycin and Amphotericin B MIC₅₀’s, MIC₉₀’s and medians (ug/ml).

	Voriconazole			Amphotericin B			Natamycin
	MIC50 MIC90 (range)	Median	p-value	MIC50 MIC90 (range)	Median	p-value	MIC50 MIC90 (range)	Median	p-value
F. solani (n=44)	16 16 (4- ≥16) n=15	16	<0.001*	2 2 (1-≥16) n=43	2	0.07*	4.8 4.8 (2.4–9.6) n=44	4.8	0.37*
F. non-solani (n=14)	4 4 (2-≥16) n=12	4		2 2 (1–2) n=14	2		2.4 4.8 (2.4–4.8) n=14	3.6

MIC₅₀ and MIC₉₀ values at 48 hours.

^*Mann-Whitney Test

In general, the isolates were equally susceptible to natamycin and amphotericin B. F.solani organisms had significantly higher voriconazole MIC values than F. non-solani organisms.

Discussion

A wide spectrum of Fusarium is known to be pathogenic in the human eye. Our improved understanding of this spectrum begs the questions we set forth to address in this study, namely to determine clinical characteristics and antifungal susceptibilities within the Fusarium genus. Currently, the literature describing these relationships is sparse. Defining these inherent qualities is critical to determining the clinical significance of accurate and rapid species identification.

Clinical Characteristics and Outcomes

In this study, F. solani isolates were associated with significantly longer treatment course (p=0.035, Mann-Whitney test), worse follow up BCVA (p=0.01, t-test), and higher necessity for PK compared to F. non-solani isolates. These observations suggest increased pathogenicity and differences between in vitro and in vivo antifungal susceptibility among F. solani isolates compared to other species within the genus.

Interestingly, F. solani isolates were associated with poorer outcomes compared to other species of Fusarium, despite slight differences between in vitro susceptibilities for the most commonly available and utilized topical antifungal agents (natamycin and amphotericin B). This observation suggests that other factors may be responsible for this discordance between in vitro and in vivo antifungal activities, like inter-species differences in pathogenicity.

Systemically, F. solani organisms are significantly more pathogenic and incur a higher risk of mortality compared to F. non-solani isolates ³¹. The suggestion of enhanced pathogenicity among F. solani isolates (compared to other Fusarium species) observed in our study, as well as other non-ocular studies, may be explained by species-specific differences in virulence strategies, including organism adherence to the ocular surface ³², invasion of the organism into the corneal stroma ³³, or alteration of host and pathogen defense mechanisms ³⁴.

Antifungal susceptibility

The increasing use of DNA-based identification schemes to accurately and rapidly identify fungal organisms to the species level has brought to light the importance of defining specific antifungal susceptibility profiles. This is particularly important for the Fusarium genus, for which minimum inhibitory concentrations (MIC’s) are higher and more variable than for other molds (e.g. Aspergillus sp) ^{23, 26, 27}.

The results of the in vitro susceptibility tests from the isolates in this study (Table 2, 3) demonstrated high levels of in vitro resistance among Fusarium sp to available antifungal agents, particularly when compared to published susceptibility profiles of other filamentous fungi, including Aspergillus sp and Paecilomyces sp ^{26, 35–38}. Across the Fusarium genus, amphotericin B had the lowest MIC values. Finally, different patterns of susceptibility were noted among the different Fusarium species. Specifically, this study demonstrates significantly higher MIC values (p<0.001, Mann-Whitney test) for voriconazole among F. solani isolates compared to F. non-solani isolates.

By the present date, six studies have explored in vitro antifungal susceptibility patterns among genetically characterized Fusarium isolates ^{26, 27, 37–40}. These studies consistently demonstrate variable MICs and high levels of resistance across the spectrum of antifungal agents.

Repeatedly, F. solani isolates have been reported to exhibit greater resistance to antifungal agents than F. non-solani in studies using both morphologic identification ^{23, 25, 41–45} and molecular characterization ^37–39. Among the studies utilizing reliable molecular identification techniques to characterize the isolates ^37–39, F. solani isolates demonstrated higher levels of resistance to voriconazole, posaconazole, and pentamidine when compared to other members of the Fusarium genus.

Management

In the present study, 11 patients received a combined therapy of natamycin and azoles. A statistically significant amount of patients in this group experienced a worse outcome (follow up BCVA < 20/40 and loss of ≥2 lines of BCVA) when compared to the other groups (p<0.001 and 0.032, Chi-square test, respectively). Alternatively, the cases with combination therapy may have had more severe disease which was not reflected by the visual acuity. Of note, it has been demonstrated that combination treatment may produce antagonistic interactions in vitro, particularly when natamycin is used in combination with azoles ⁴⁶.

Limitations of the study

The lower prevalence of F. non-solani when compared with F. solani infections, did not allow for selection of an even distribution of Fusarium species to be genotyped. In future studies, an equal distribution and larger number of isolates from all species would provide more power for further inter-species comparisons. Furthermore, due to the inherent limitations in a retrospective study, the clinical data was incomplete in some cases, specifically risk factor assessment and follow-up information.

Conclusions

This study demonstrates important differences in antifungal susceptibility profiles and clinical characteristics among Fusarium isolates causing keratitis, and suggests inter-species differences in virulence mechanisms yet to be explored. Future studies comparing in vitro and in vivo antifungal activity of the available antifungal agents alone and in combination will allow us to accurately assess drug interactions and optimize specific treatment strategies. Furthermore, this study suggests that accurate species identification, especially if performed with rapid PCR techniques, may yield important prognostic and therapeutic information that can influence management decisions and improve patient outcomes in the setting of Fusarium keratitis.