Comparison of the Vitek 2 Antifungal Susceptibility System with the Clinical and Laboratory Standards Institute (CLSI) and European Committee on Antimicrobial Susceptibility Testing (EUCAST) Broth Microdilution Reference Methods and with the Sensititre YeastOne and Etest Techniques for In Vitro Detection of Antifungal Resistance in Yeast Isolates

Abstract

The commercial technique Vitek 2 system for antifungal susceptibility testing of yeast species was evaluated. A collection of 154 clinical yeast isolates, including amphotericin B- and azole-resistant organisms, was tested. Results were compared with those obtained by the reference procedures of both the CLSI and the European Committee on Antimicrobial Susceptibility Testing (EUCAST). Two other commercial techniques approved for clinical use, the Etest and the Sensititre YeastOne, were included in the comparative exercise as well. The average essential agreement (EA) between the Vitek 2 system and the reference procedures was >95%, comparable with the average EAs observed between the reference procedures and the Sensititre YeastOne and Etest. The EA values were >97% for Candida spp. and stood at 92% for Cryptococcus neoformans. Intraclass correlation coefficients (ICC) between the commercial techniques and the reference procedures were statistically significant (P < 0.01). Percentages of very major errors were 2.6% between Vitek 2 and the EUCAST technique and 1.6% between Vitek 2 and the CLSI technique. The Vitek 2 MIC results were available after 14 to 18 h of incubation for all Candida spp. (average time to reading, 15.5 h). The Vitek 2 system was shown to be a reliable technique to determine antifungal susceptibility testing of yeast species and a more rapid and easier alternative for clinical laboratories than the procedures developed by either the CLSI or EUCAST.

The standardization of reference procedures for antifungal-susceptibility testing (AST) of yeasts and molds by the Clinical and Laboratory Standards Institute (CLSI) (2, 3) and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (17, 18) has led to the development of several automated or semiautomated commercial systems to provide simple, flexible, and affordable alternative susceptibility testing methods for use in the clinical laboratory (5). Many microbiologists prefer to use commercial systems which have particular advantages, such as ease of performance, economy, or more rapid results.

The Vitek 2 antifungal susceptibility system (bioMérieux, Inc., Hazelwood, MO) is a fully automated commercial method that determines yeast growth spectrophotometrically and that allows both fungal identification and MIC determination simultaneously. The Vitek 2 system includes the Vitek 2 cards that allow species identification by comparison of the biochemical profile with an extensive database (1, 11). The system also incorporates the AST-YS01 card, which is designed for susceptibility testing of amphotericin B (AMB), fluconazole (FLC), flucytosine (5FC), and voriconazole (VRC). The AST-YS01 card is essentially a miniaturized version of the doubling-dilution technique used for determining the MIC by microdilution methodology in μg/ml. It consists of 64 wells containing aliquots of a specific antifungal agent. Once the card is placed in the Vitek 2 system with the appropriate organism suspension, no further handling is required. Dilutions are made, and the card is inoculated using a vacuum-filling process and then sealed and placed into the reader/incubator automatically. The system integrates a software program which validates and interprets susceptibility test results, and the computer determines when a well demonstrates growth based on the attenuation of light measured by an optical scanner. These data are used to determine the MIC values for the antifungal agents (14).

In published evaluations, the Vitek 2 system has demonstrated a high level of reproducibility and an excellent categorical agreement with the CLSI microdilution reference procedure (>95%) for fluconazole and, more recently, for amphotericin B, flucytosine, and voriconazole (13, 15). It must be noted that the Vitek 2 system was able to determine the MIC endpoints after 9.1 to 27.1 h of incubation (mean, 12 to 14 h) (14). In view of these data, the U.S. FDA approved in 2006 the clinical use of this system to detect resistance to fluconazole (13).

The purpose of the present study was the evaluation of the commercial technique Vitek 2 system for antifungal susceptibility testing of yeast species and its comparison with results obtained by the CLSI and EUCAST reference procedures. In the case of the CLSI method, agreement was calculated for two reading times, 24 h and 48 h of incubation. Two other commercial techniques approved for clinical use, the Etest and the Sensititre YeastOne methods, were included in the comparative exercise as well.

MATERIALS AND METHODS

Fungi.

A collection of 154 clinical isolates was tested. The majority of isolates (n = 75) were obtained from blood cultures and the remainder from deep-site specimens (n = 36) or oropharyngeal exudates (n = 33). Species distribution was as follows: 41 Candida albicans isolates, 18 Candida tropicalis isolates, 17 Candida parapsilosis isolates, 16 Candida glabrata isolates, 22 Candida krusei isolates, 8 Candida lusitaniae isolates, 5 Candida guilliermondii isolates, 3 Candida famata isolates, 1 Candida dubliniensis isolate, 1 Candida kefyr isolate, 16 Cryptococcus neoformans isolates, 3 Rhodotorula mucilaginosa isolates, and 3 Dipodascus capitatus isolates. They were selected to represent ranges of susceptibilities in vitro that were as broad as possible, including for organisms resistant in vitro to antifungal agents (4, 5). A total of 4 isolates had an AMB MIC above or equal to 1 μg/ml, 18 had a 5FC MIC value of >4 μg/ml, 85 exhibited an FLC MIC value of >4 μg/ml (10 at 16 μg/ml, 25 at 32 μg/ml, and 26 at >64 μg/ml), and 57 had a VRC MIC of >0.12 μg/ml (3 at 2 μg/ml, 3 at 4 μg/ml, and 10 at >8 μg/ml) by both reference AST procedures. Isolates were identified by morphological and physiological techniques and by sequencing of DNA targets (7, 10). Candida parapsilosis (ATCC 22019) and Candida krusei (ATCC 6258) were incorporated as quality control strains in each set of experiments.

Reference susceptibility testing procedures.

The MICs for all isolates were determined by strictly following the reference microdilution M27-A3 procedure of the CLSI (3). The antifungal agents used in the study were AMB (Sigma-Aldrich Química S.A., Madrid, Spain), 5FC (Sigma-Aldrich), FLC (Pfizer S.A., Madrid, Spain), and VRC (Pfizer Ltd., Sandwich, United Kingdom).

In addition, AST was performed using the EUCAST standard for testing fermentative yeasts (document 7.1) (17). For Cryptococcus neoformans and other species of nonfermentative fastidious yeasts, susceptibility testing was also performed by the EUCAST standard, but a minor modification was done to improve their growth: microdilution plates were sealed to limit evaporation, attached to an electrically driven wheel inside the incubator, and agitated at 350 rpm at 30°C for 48 h. MIC endpoints were determined spectrophotometrically at 24 and 48 h. The minor modification has been shown as having no significant influence on susceptibility test results since growth of organisms is increased without overestimating MIC values (4, 9, 12, 16). For testing the susceptibility of yeast to amphotericin B by the EUCAST method, the MIC endpoints were defined as the lowest drug concentration that resulted in a reduction in growth of 90% or more compared with the growth in a drug-free control well. For flucytosine and azoles, the MIC endpoint was defined as a 50% reduction in optical density.

Commercial techniques.

Apart from the Vitek 2 system, two other commercial methods were analyzed, the Sensititre YeastOne panel (Trek Diagnostic Systems, Ltd., East Grinstead, United Kingdom) and Etest strips on RPMI 1640-2% glucose agar (bioMérieux). Susceptibility testing, reading, and interpretations of the results were performed in accordance with the manufacturer's instructions. A standardized inoculum suspension was placed into a Vitek 2 antifungal susceptibility test card for each organism. Cards were filled, incubated, and read automatically when the rate of growth permitted the endpoint determination, according to the manufacturer's instructions. The results are expressed as MIC values in μg/ml.

Analysis of results.

Both on-scale and off-scale results obtained by the AST procedures were included in the analysis. The low off-scale MICs were left unchanged, and the high off-scale MICs were converted to the next-highest concentration.

The reproducibility of the results obtained by the reference techniques and by the commercial systems was calculated to determine the percentages of essential agreement (EA) between MIC values (8, 13). Agreement was defined as discrepancies in MIC results of no more than ±2 2-fold dilutions. In addition, the correlation between the results was evaluated using the intraclass correlation coefficient (ICC) (21, 22, 23), which was expressed to a maximum value of 1 and had a confidence interval (CI) of 95% (5, 6). In order to approximate a normal distribution, the MICs were transformed to log₂ values. A P of <0.01 was considered statistically significant due to the multiple comparisons/correlations attempted in the study. The ICC is a reverse measurement of the variability of the counting values. The ICC was calculated using the formula ICC = (group mean square − error mean square)/(group mean square + error mean square) and thus has a maximum value of 1 if there is a perfect correlation and a minimum value of −1 if there is a complete absence of a correlation. The ICC evaluates the correlations between values, offering a statistical significance since it takes into account the number of cases and the absolute value of the counting. The ICC is the analysis which exhibits the highest statistical power for correlation studies.

Categorical agreements and percentages of discrepancies were also evaluated. Categories were defined depending on the existence of interpretative breakpoints. The CLSI has defined breakpoints for 5FC, FLC, and VRC (3). The EUCAST has defined interpretive breakpoints for FLC and VRC only (19, 20). For antifungal agents whose breakpoints have been set, categorical agreement was defined as the percentage of isolates classified in the same category by the reference procedures and the commercial systems, and discrepancies between methods were considered very major errors (VME) if an isolate classified as showing resistance in vitro by the reference method was categorized as susceptible by the commercial method. Discrepancies were considered major errors (ME) if an isolate classified as susceptible by the reference method was classified as resistant by the commercial technique. Minor errors (MiE) were considered to have occurred when a susceptible isolate was classified as intermediate or susceptible-dose dependent (S-DD), when a resistant organism was grouped with intermediate or S-DD isolates, when intermediate or S-DD strains were considered susceptible, or when intermediate or S-DD isolates were classified as resistant organisms.

If breakpoints were not defined, discrepancies were classified as nonsubstantial and substantial differences. Nonsubstantial differences were defined as discrepancies in MIC results of 3 or 4 2-fold dilutions, and substantial differences were defined as discrepancies of more than 4 2-fold dilutions. All statistical analyses were done with the Statistical Package for the Social Sciences (version 17.0; SPSS S.L., Madrid, Spain).

RESULTS

A total of 149 out of 154 strains (96.8%) grew when the Vitek 2 system was used, and MIC values could be determined. Five isolates belonging to slow-growing species (two C. neoformans strains, one Rhodotorula mucilaginosa strain, one Dipodascus capitatus strain, and one Candia kefyr strain) did not grow, and MICs could not be read. The Vitek 2 MIC results were available after 14 to 18 h of incubation for all Candida spp. (average time to reading, 15.5 h). The MIC determination by Vitek 2 needed an incubation period of 34 h for C. neoformans, R. mucilaginosa, and D. capitatus.

The essential agreement (EA) between the results obtained by the Vitek 2 system and the reference procedures was very high (>95%), comparable to those observed between the reference procedures and the other two commercial techniques, the Sensititre YeastOne and the Etest. Table 1 shows EA rates between procedures by antifungal agent. The rates of EA were also very high when results were analyzed per species. The EA value surpassed 95% for all species except C. neoformans, whose EA rate stood at 92% between results obtained by Vitek 2 and by the reference procedures.

TABLE 1.

Essential agreement rates between results given by the Vitek 2 system, Sensititre YeastOne, and Etest and the reference procedures, by antifungal agent

Antifungal agenta	% EA betweenb:
	Vitek and EUCAST	Vitek and CLSI24H	Vitek and CLSI48H	Etest and EUCAST	Etest and CLSI24H	Etest and CLSI48H	SYOne and EUCAST	SYOne and CLSI24H	SYOne and CLSI48H
AMB	98.7	99.3	100	98.4	97.4	96.4	97.9	97.4	96.0
5FC	98.0	98.6	96.0	96.4	95.2	95.2	96.0	95.2	95.2
FLC	97.5	96.6	96.2	97.2	96.4	95.2	97.2	96.0	95.6
VRC	97.5	96.8	96.6	95.2	95.2	95.2	95.5	95.6	95.2
Total	97.9	97.8	97.3	96.8	96.1	95.5	96.6	96.1	95.5

^aAMB, amphotericin B; 5FC, flucytosine; FLC, fluconazole; VRC, voriconazole.

^bVitek, Vitek 2 system; EUCAST, EUCAST reference procedure; CLSI24H, CLSI reference procedure after 24 h of incubation; CLSI48H, CLSI reference procedure after 48 h of incubation; Etest, Etest technique; SYOne, Sensititre YeastOne technique.

Table 2 includes the ICCs between methods. Correlation between each combination of techniques was significant (P < 0.01). ICCs for the Vitek 2 system and the reference procedures were significant as well but somewhat lower than those observed for the other two commercial methods and the reference techniques; this may be because the ranges of antifungal agents in the Vitek 2 system do not match exactly the ranges of the other techniques (i.e., the range of fluconazole in the reference procedures was 0.12 to 64 mg/liter, and the range of fluconazole in Vitek 2 was 1 to 64 mg/liter). The ICCs when data were analyzed by antifungal agent and by species were significant as well.

TABLE 2.

Intraclass correlation coefficient indexes between techniques^a

Method	ICC index
	EUCAST	CLSI24H	CLSI48H	SYOne	Etest	Vitek
EUCAST	1.0	0.987	0.984	0.945	0.933	0.871
CLSI24H	0.987	1.0	9.987	0.941	0.934	0.862
CLSI48H	0.984	0.987	1.0	0.938	0.925	0.851
SYOne	0.945	0.941	0.938	1.0	0.943	0.897
Etest	0.933	0.934	0.925	0.943	1.0	0.889
Vitek	0.911	0.915	0.909	0.912	0.923	1.0

^aICC indexes were expressed to a maximum value of 1. All ICC indexes were statistically significant (P < 0.01). See Table 1 for explanations of abbreviations.

Regarding the categorical agreement, percentages of very major errors (VME), major errors (ME), minor errors (MiE), substantial discrepancies (SD), and nonsubstantial discrepancies (NSD) by antifungal agent are displayed in Tables 3, 4, and 5. It should be noted that VME and SD were uncommon, and most errors detected were MiE and NSD.

TABLE 3.

Numbers of errors and discrepancies between MICs obtained by the commercial techniques and the EUCAST reference procedure, by antifungal agent

Antifungal agenta	No. of errors or discrepancies between EUCAST andb:
	Vitek					Etest					SYOne
	VME	ME	MiE	SD	NSD	VME	ME	MiE	SD	NSD	VME	ME	MiE	SD	NSD
AMB				0	2				0	5				0	5
5FC				0	3				1	4				0	3
FLZ	8	0	23			7	0	24			7	0	20
VOR	0	0	10			0	0	6			0	0	8
Total	8	0	33	0	5	7	0	30	1	9	7	0	28	0	8

^aAMB, amphotericin B; 5FC, flucytosine; FLC, fluconazole; VRC, voriconazole.

^bVitek, Vitek 2 system technique; EUCAST, EUCAST reference procedure; Etest, Etest method; SYOne, Sensititre YeastOne technique; VME, very major errors; ME, major errors; MiE, minor errors; SD, substantial discrepancies; NSD, nonsubstantial discrepancies.

TABLE 4.

Numbers of errors and discrepancies between MICs obtained by commercial techniques and the CLSI reference procedure after 24 h of incubation, by antifungal agent

Antifungal agenta	No. of errors or discrepancies between CLSI24H andb:
	Vitek					Etest					SYOne
	VME	ME	MiE	SD	NSD	VME	ME	MiE	SD	NSD	VME	ME	MiE	SD	NSD
AMB				0	1				0	7				0	6
5FC	1	0	8			0	1	18			0	0	12
FLZ	3	0	22			1	0	22			1	0	24
VOR	2	2	3			0	1	5			0	2	4
Total	6	2	33	0	1	1	2	45	0	7	1	2	40	0	6

^aAMB, amphotericin B; 5FC, flucytosine; FLC, fluconazole; VRC, voriconazole.

^bVitek, Vitek 2 system technique; CLSI24H, CLSI reference procedure after 24 h of incubation; Etest, Etest method; SYOne, Sensititre YeastOne technique; VME, very major errors; ME, major errors; MiE, minor errors; SD, substantial discrepancies; NSD, nonsubstantial discrepancies.

TABLE 5.

Numbers of errors and discrepancies between MICs obtained by commercial techniques and the CLSI reference procedure after 48 h of incubation, by antifungal agent

Antifungal agenta	No. of errors or discrepancies between CLSI48H andb:
	Vitek					Etest					SYOne
	VME	ME	MiE	SD	NSD	VME	ME	MiE	SD	NSD	VME	ME	MiE	SD	NSD
AMB				0	0				0	8				0	10
5FC	1	0	14			0	1	23			0	0	16
FLZ	3	0	25			1	0	33			1	0	30
VOR	2	2	2			0	1	5			0	1	5
Total	6	2	41	0	0	1	2	61	0	8	1	1	51	0	10

^aAMB, amphotericin B; 5FC, flucytosine; FLC, fluconazole; VRC, voriconazole.

^bVitek, Vitek 2 system technique; CLSI48H, CLSI reference procedure after 48 h of incubation; Etest, Etest method; SYOne, Sensititre YeastOne technique; VME, very major errors; ME, major errors; MiE, minor errors; SD, substantial discrepancies; NSD, nonsubstantial discrepancies.

The percentage of VME between Vitek 2 and the EUCAST technique was 2.7% (8/298 determinations), and substantial discrepancies were not observed. VME were detected in eight cases and for FLC MIC determination only. Those eight strains were FLC-resistant organisms (four C. albicans strains, one C. guilliermondii strain, and three C. neoformans strains) which exhibited characteristics of slow-growing strains. The VME rate between results of the Etest and the EUCAST reference procedure was 2.3%, and the VME rate between Sensititre YeastOne results and those of the EUCAST method was also 2.3%.

The rate of VME between Vitek 2 and the CLSI technique was 1.3% (6/447 determinations; one for 5FC, three for FLC, and two for VRC). VME were also observed in strains exhibiting slower growth. Rates of VME between the other two commercial methods and the CLSI methodology were both 0.2%.

DISCUSSION

The treatment of systemic fungal infections has undergone changes in the past few years as several new antifungal agents have come on the market. Because of the existence of these therapeutic alternatives, it is clear that not all fungal infections should be treated in the same manner; thus, the identification of fungal species and susceptibility testing are increasingly important. The emergence of strains resistant to antifungal agents has led to variations in the treatment guidelines between different geographical areas. In addition, knowledge of the MIC values of various antifungals for every isolate can be significant in the management of a particular case of fungal infection.

The Vitek 2 system has the advantage of being a fully automated methodology that determines yeast growth spectrophotometrically and that allows both fungal identification and MIC determination simultaneously. This method is easy to perform, and no complex handling is required. Dilutions are made, and cards are inoculated using a vacuum-filling process and then sealed and placed into the reader/incubator automatically. The system uses an integrated software program which validates and interprets the results. One more advantage is that the Vitek 2 system is able to determine the MIC value more rapidly than other techniques. Pfaller et al. (11) reported that this system was able to determine the MIC endpoints of fluconazole after 9.1 to 27.1 h of incubation (mean, 12 to 14 h). In our study, the MIC values for Candida isolates were obtained after 14 to 18 h of incubation, with an average time to reading of 15.5 h.

The essential agreement and the ICC indexes between the results obtained by the Vitek 2 system and the reference procedures were very high and statistically significant (P < 0.01). The agreement was >95%, similar to those previously reported (13-15). The rates of agreement between Vitek 2 and the reference procedures were comparable to those observed between the reference procedures and the other two commercial techniques, the Sensititre YeastOne and the Etest, which are used to detect resistance to FLC in vitro and are considered by the FDA to be research-use-only methods for AST of certain antifungal agents, such as AMB.

Regarding the categorical agreement between the Vitek 2 system and the reference procedures, it should be noted that very major errors and substantial discrepancies were uncommon, and most errors detected were minor errors and nonsubstantial discrepancies. The percentage of strains classified as resistant in vitro by the EUCAST procedure and as susceptible in vitro by the Vitek 2 system was 2.6%, and the percentage classified as resistant by the CLSI method and as susceptible by the Vitek 2 system was 1.6%.

It can be concluded that the Vitek 2 system is a reliable technique to determine antifungal susceptibility testing of yeast species, and that it can be clinically useful for determining the susceptibility of both Candida spp. and other yeast species, such as C. neoformans and emerging species. It is also a reliable technique for identification of azole and amphotericin B resistance in vitro. It has the advantage of being more rapid than and an easier alternative to the procedures developed by either the CLSI or the EUCAST. From a clinical-laboratory point of view, echinocandins and posaconazole should be included in the system to improve its clinical usefulness.