Progress in Antifungal Susceptibility Testing of Candida spp. by Use of Clinical and Laboratory Standards Institute Broth Microdilution Methods, 2010 to 2012

Abstract

Antifungal susceptibility testing of Candida has been standardized and refined and now may play a useful role in managing Candida infections. Important new developments include validation of 24-h reading times for all antifungal agents and the establishment of species-specific epidemiological cutoff values (ECVs) for the systemically active antifungal agents and both common and uncommon species of Candida. The clinical breakpoints (CBPs) for fluconazole, voriconazole, and the echinocandins have been revised to provide species-specific interpretive criteria for the six most common species. The revised CBPs not only are predictive of clinical outcome but also provide a more sensitive means of identifying those strains with acquired or mutational resistance mechanisms. This brief review serves as an update on the new developments in the antifungal susceptibility testing of Candida spp. using Clinical and Laboratory Standards Institute (CLSI) broth microdilution (BMD) methods.

INTRODUCTION

The need for reproducible and clinically relevant antifungal susceptibility testing has been prompted by the increasing number of invasive fungal infections (IFIs), the expanding use of antifungal agents, and the recognition of antifungal resistance as an important clinical problem (5, 12, 17, 26, 35, 51, 55, 57, 86, 88). In vitro antifungal susceptibility testing is now standardized internationally (16) and is becoming essential in patient management and resistance surveillance (57). Although in vitro susceptibility testing is often used to select antimicrobial agents likely to be clinically active for a given infection, perhaps its more important function is the detection of in vitro resistance, i.e., to determine which agents are less likely to be effective (17, 57, 86, 93). Improvements in the ability of antifungal susceptibility testing methods to detect emerging resistance patterns, coupled with molecular characterization of resistance mechanisms, provide useful adjuncts to optimize the effectiveness of antifungal therapy (16, 26, 35, 55, 56, 57, 63, 69, 70, 88).

There are two internationally recognized standard methods for the performance of antifungal susceptibility testing of Candida spp. using broth microdilution (BMD): that of the Clinical and Laboratory Standards Institute (CLSI) (13, 14) and that of the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (23, 24, 87). These methods provide MIC data for azole and echinocandin antifungal agents that are both quantitatively and qualitatively similar (4, 66, 71). As a result, in vitro antifungal susceptibility testing now plays an increasingly important role in guiding therapeutic decision making, as an aid in drug development studies, and as a means of tracking the development of antifungal resistance in epidemiological studies (5, 16, 26, 35, 37, 52, 55, 57, 86, 88). In recent years, studies by the CLSI Subcommittee on Antifungal Susceptibility Testing and collaborating investigators have generated data to support a more rapid time for MIC result reporting (39, 61, 71, 72), established epidemiological cutoff values (ECVs) for Candida and the systemically active antifungal agents (63, 67, 73, 74, 76), and developed new species-specific clinical breakpoints (CBPs) for fluconazole (65), voriconazole (69), and the echinocandins (70). This review summarizes these recent advances in the performance of antifungal susceptibility testing of Candida spp. using CLSI BMD methods and provides commentary on the rationale for, impact of, and recommendations for the clinical application of antifungal susceptibility testing of Candida spp.

VALIDATION OF 24-H MIC DETERMINATIONS FOR AZOLES AND FLUCYTOSINE

The vast majority of Candida species achieve sufficient growth within 24 h of incubation to allow MIC testing using the CLSI BMD method (20–22, 61). Given that a shorter duration of incubation avoids the potentially confounding effect of trailing growth on 48-h MICs (7, 61, 80, 84), is more efficient and practical for use in the clinical laboratory (22, 61), and provides useful information sooner (30, 48, 54), attention has been focused on the validation of 24-h MIC determinations for antifungal agents and Candida. Investigators have used previously gathered Candida MIC data to evaluate the essential agreement (EA; agreement within ±2 log 2 dilutions) and categorical agreement (CA) between MIC values determined at 24-h and 48-h incubation for amphotericin B, fluconazole, flucytosine, itraconazole, posaconazole, and voriconazole (39, 50, 61, 63, 67, 72, 73, 76). It is now clear that MIC determinations at 24 h provide good EA and CA with 48-h values for these antifungal agents and up to 11 different species of Candida, allowing clinical laboratories to read BMD MICs for antifungal susceptibility testing of Candida in the same time frame as other antimicrobial susceptibility tests.

DEVELOPMENT OF EPIDEMIOLOGICAL CUTOFF VALUES (ECVs) FOR CANDIDA SPP. AND THE SYSTEMICALLY ACTIVE ANTIFUNGAL AGENTS

Clinical interpretive breakpoints (CBPs) for in vitro antimicrobial susceptibility testing may be used to indicate those clinical isolates that are likely to respond to treatment with a given antimicrobial agent administered using the approved dosing regimen for that agent (93). Conversely, epidemiological cutoff values (ECVs) can be considered to represent the most sensitive measure of the emergence of strains with decreased susceptibility to a given agent (36, 90). An ECV is an MIC threshold value that allows the discrimination of wild-type (WT) strains (those without mutational or acquired resistance mechanisms) from non-WT strains (those having mutational or acquired resistance mechanisms) (36, 90, 93). The typical MIC distribution for WT organisms covers 3 to 5 doubling dilutions surrounding the modal MIC (3, 36, 92). The ECV for each organism-antimicrobial agent pair is obtained by considering the WT MIC distribution, the modal MIC for each distribution, and the inherent variability of the test (usually within ±1 doubling dilution). For most MIC distributions, the ECV is determined to occur at an MIC that is approximately two dilutions above the modal MIC and encompasses (MIC ≤ ECV) ∼95% of the results in the WT MIC distribution (92). Organisms with mutational or acquired resistance mechanisms may be included among those for which the MIC results are higher than the ECV (3, 36, 74), and ECVs may be used as a means of tracking the emergence of reduced susceptibility to antifungal agents among Candida spp. (63, 73, 74). ECVs may also be used to identify isolates that are less likely to respond to antimicrobial therapy due to acquired resistance mechanisms when limited clinical data preclude the development of CBPs (63, 65, 67, 69–73, 76, 93).

Over the past 3 years, ECVs have been established for amphotericin B, flucytosine, the triazoles (fluconazole, itraconazole, posaconazole, and voriconazole), and the echinocandins (anidulafungin, caspofungin, and micafungin) and 11 species of Candida using CLSI BMD methods (Table 1). The process used to develop the ECVs for each class of antifungal agents and the relationship of the ECVs to the various acquired or mutational resistance mechanisms (when known) are briefly described below.

Table 1.

Epidemiological cutoff values and clinical breakpoints for systemically active antifungal agents and Candida spp. determined by 24-h CLSI broth microdilution methods^a

Organism	Antifungal agent	ECV (μg/ml)		CBP (μg/ml)
		WT	Non-WT	S	SDD	I	R
C. albicans	Amphotericin B	≤2	>2
	Flucytosine	≤0.5	>0.5
	Fluconazole	≤0.5	>0.5	≤2	4		≥8
	Itraconazole	≤0.12	>0.12	≤0.12	0.25–0.5		≥1
	Posaconazole	≤0.06	>0.06
	Voriconazole	≤0.03	>0.03	≤0.12		0.25–0.5	≥1
	Anidulafungin	≤0.12	>0.12	≤0.25		0.5	≥1
	Caspofungin	≤0.12	>0.12	≤0.25		0.5	≥1
	Micafungin	≤0.03	>0.03	≤0.25		0.5	≥1
C. glabrata	Amphotericin B	≤2	>2
	Flucytosine	≤0.5	>0.5
	Fluconazole	≤32	>32		≤32		≥64
	Itraconazole	≤2	>2
	Posaconazole	≤2	>2
	Voriconazole	≤0.5	>0.5
	Anidulafungin	≤0.25	>0.25	≤0.12		0.25	≥0.5
	Caspofungin	≤0.12	>0.12	≤0.12		0.25	≥0.5
	Micafungin	≤0.03	>0.03	≤0.06		0.12	≥0.25
C. parapsilosis	Amphotericin B	≤2	>2
	Flucytosine	≤0.5	>0.5
	Fluconazole	≤2	>2	≤2	4		≥8
	Itraconazole	≤0.5	>0.5
	Posaconazole	≤0.25	>0.25
	Voriconazole	≤0.12	>0.12	≤0.12		0.25–0.5	≥1
	Anidulafungin	≤4	>4	≤2		4	≥8
	Caspofungin	≤1	>1	≤2		4	≥8
	Micafungin	≤4	>4	≤2		4	≥8
C. tropicalis	Amphotericin B	≤2	>2
	Flucytosine	≤0.5	>0.5
	Fluconazole	≤2	>2	≤2	4		≥8
	Itraconazole	≤0.5	>0.5
	Posaconazole	≤0.12	>0.12
	Voriconazole	≤0.06	>0.06	≤0.12		0.25–0.5	≥1
	Anidulafungin	≤0.12	>0.12	≤0.25		0.5	≥1
	Caspofungin	≤0.12	>0.12	≤0.25		0.5	≥1
	Micafungin	≤0.12	>0.12	≤0.25		0.5	≥1
C. krusei	Amphotericin B	≤2	>2
	Flucytosine	≤32	>32
	Fluconazole	≤64	>64
	Itraconazole	≤1	>1
	Posaconazole	≤0.5	>0.5
	Voriconazole	≤0.5	>0.5	≤0.5		1	≥2
	Anidulafungin	≤0.12	>0.12	≤0.25		0.5	≥1
	Caspofungin	≤0.25	>0.25	≤0.25		0.5	≥1
	Micafungin	≤0.12	>0.12	≤0.25		0.5	≥1
C. lusitaniae	Amphotericin B	≤2	>2
	Flucytosine	≤0.5	>0.5
	Fluconazole	≤2	>2
	Itraconazole	≤0.5	>0.5
	Posaconazole	≤0.12	>0.12
	Voriconazole	≤0.03	>0.03
	Anidulafungin	≤2	>2
	Caspofungin	≤1	>1
	Micafungin	≤0.5	>0.5
C. guilliermondii	Amphotericin B	≤2	>2
	Flucytosine	≤1	>1
	Fluconazole	≤8	>8
	Itraconazole	≤1	>1
	Posaconazole	≤0.5	>0.5
	Voriconazole	≤0.25	>0.25
	Anidulafungin	≤4	>4	≤2		4	≥8
	Caspofungin	≤2	>2	≤2		4	≥8
	Micafungin	≤2	>2	≤2		4	≥8
C. dubliniensis	Amphotericin B	≤2	>2
	Flucytosine	≤0.5	>0.5
	Fluconazole	≤0.5	>0.5
	Itraconazole	≤0.25	>0.25
	Posaconazole	≤0.12	>0.12
	Voriconazole	≤0.03	>0.03
	Anidulafungin	≤0.12	>0.12
	Caspofungin	≤0.12	>0.12
	Micafungin	≤0.12	>0.12
C. kefyr	Fluconazole	≤1	>1
	Posaconazole	≤0.25	>0.25
	Voriconazole	≤0.015	>0.015
	Anidulafungin	≤0.25	>0.25
	Caspofungin	≤0.03	>0.03
	Micafungin	≤0.12	>0.12
C. orthopsilosis	Fluconazole	≤2	>2
	Posaconazole	≤0.25	>0.25
	Voriconazole	≤0.06	>0.06
	Anidulafungin	≤2	>2
	Caspofungin	≤0.5	>0.5
	Micafungin	≤1	>1
C. pelliculosa	Fluconazole	≤4	>4
	Posaconazole	≤2	>2
	Voriconazole	≤0.25	>0.25
	Caspofungin	≤0.12	>0.12

^aData compiled from references 65, 67, 69, 73, 74, 76, and 83. ECVs, epidemiological cutoff values; CBPs, clinical breakpoints; WT, wild type; non-WT, non-wild type; S, susceptible; SDD, susceptible, dose dependent; I, intermediate; R, resistant.

Amphotericin B.

The ECVs determined after 24-h incubation for amphotericin B and each species of Candida are shown in Table 1. The ECV was 2 μg/ml for all species and encompasses 97% to 100% of results for the indicated species (76).

In the literature, a default breakpoint for resistance or nonsusceptibility to amphotericin B is variably cited to be an MIC of either >0.5 μg/ml or >1 μg/ml (38, 53, 85). This cutoff is loosely based on the attainment of peak serum concentrations of 2 μg/ml and the pharmacodynamic (PD) correlate of a peak serum concentration-to-MIC ratio of 2 as predictive of near maximal in vivo activity (85). Analyses of both clinical trial data (81, 82) and clinical and microbiological data from population-based surveillance studies (53) have failed to establish any clinical correlation between amphotericin B MICs, as determined by either CLSI BMD or Etest methodology, and clinical outcome. Park et al. (53) specifically addressed the predictive value of a CBP of 1 μg/ml using the CLSI method and found a distinct lack of prediction of clinical outcome; however, the limited data set of 107 cases of candidemia treated with amphotericin B did not contain an episode for which the amphotericin B MIC was greater than 1 μg/ml.

The data thus far suggest that an ECV of 2 μg/ml as determined by 24-h CLSI BMD methods should be used to decide whether an isolate of Candida spp. should be considered WT (MIC ≤ ECV) or non-WT (MIC > ECV) with respect to amphotericin B susceptibility (76). This cutoff would encompass all of the isolates reported by Rex et al. (81, 82) and by Park et al. (53). Notably, these WT strains of Candida were associated with a 50% (53) to 79% (81, 82) clinical success rate when treated with amphotericin B. A similar response rate of 65% was also seen in the amphotericin B arm (115 patients) of a study reported by Mora-Duarte et al. (46) where the MIC range was 0.25 to 2 μg/ml. Thus, an amphotericin B MIC greater than 2 μg/ml should be considered to be distinctively unusual for the vast majority of Candida spp., suggesting that treatment with this agent alone may not be optimal (85, 86).

Flucytosine.

The results of more than 17,000 MIC determinations from 16 different laboratories were used to generate the ECVs for flucytosine and eight different species of Candida (76) (Table 1). The 24-h ECVs were 0.5 to 1 μg/ml for all species, with the exception of C. krusei (ECV, 32 μg/ml). These ECVs approximate the susceptible breakpoint of ≤1 μg/ml established by the British Society for Mycopathology (10) and are well below the CBP for susceptibility of ≤4 μg/ml proposed by the CLSI (13, 14, 85). The CLSI CBP was based on a combination of historical data and in vivo results from animal studies, with little or no consideration of clinical data or mechanisms of resistance (85). The finding that the flucytosine MIC for the vast majority of Candida isolates is ≤0.5 μg/ml (76) raises concern that the CLSI CBPs of ≤4 μg/ml (susceptible [S]), 8 to 16 μg/ml (intermediate [I]), and ≥32 μg/ml (resistant [R]) may be too high and are likely to be insensitive to the development of decreased susceptibility or resistance to flucytosine among the more highly susceptible species of Candida (2).

Resistance mechanisms for flucytosine are well described among various species of Candida and include mutations in the genes FCY2, FCY1, and FUR1 encoding the cytosine permease, cytosine deaminase, and phosphoribosyltransferase enzymes, respectively. Studies of C. albicans (18, 34, 78), C. glabrata (19, 94), C. dubliniensis (45), and C. lusitaniae (25) have elucidated several different patterns of susceptibility to flucytosine, each of which is dependent upon the mutations present. In general, mutations in FCY2 (permease) result in MICs that are somewhat elevated (>0.5 μg/ml but <8 μg/ml) whereas mutations in FCY1 (deaminase) and FUR1 (phosphoribosyltransferase) result in MICs that are 8 to 128 μg/ml, depending upon whether the organism is heterozygous or homozygous for the mutation (18, 19, 25, 34, 78, 94). Dodgson et al. (18) found that isolates of C. albicans representing clade 1 for which FUR1 was WT in both alleles all had MICs for flucytosine that were <0.5 μg/ml, those with a mutation in one allele had MICs of 0.5 to 8 μg/ml, and those with mutations in both alleles all had MICs that were >16 μg/ml. This was confirmed by Hope et al. (34), who also demonstrated that a C. albicans isolate with a mutation in FCY1 exhibited an intermediate level of flucytosine resistance, with an MIC of 4 μg/ml. In the related species C. dubliniensis, McManus et al. (45) reported that isolates with a homozygous mutation in FCY1 demonstrated a high degree of resistance (MIC, ≥128 μg/ml) and that those without a mutation all had flucytosine MICs of 0.25 μg/ml or less. Edlind and Katiyar (19) showed that the haploid yeast C. glabrata exhibited high-level resistance to flucytosine (MIC, ≥32 μg/ml) that was associated with mutations in either FCY1 or FUR1 and that moderately elevated MICs (MIC, 1 μg/ml) were associated with mutations in FCY2. Similar findings with the haploid yeast C. lusitaniae were also reported by Florent et al. (25). Taken together, these findings indicate that normally flucytosine-susceptible (WT) species of Candida exhibit MICs of ≤0.5 μg/ml and do not have mutations in FCY1, FCY2, or FUR1, whereas non-WT strains for which MICs are between 1 μg/ml and 8 μg/ml may have mutations in FCY2 or are heterozygous for mutations in FUR1 or FCY1 and that those strains that are homozygous for mutations in FCY1 or FUR1 are highly resistant, with MICs in excess of 32 μg/ml. These data provide support for the ECVs shown in Table 1 in that WT strains for which the flucytosine MIC is ≤0.5 μg/ml are unlikely to possess a flucytosine resistance mutation, whereas those strains for which the MIC is greater than the ECV (non-WT; MIC, >0.5 μg/ml) are likely to be either homozygous or heterozygous for a flucytosine resistance mutation.

Triazoles.

The systemically active triazoles that are available for the treatment or prevention of invasive candidiasis (IC) include fluconazole, itraconazole, posaconazole, and voriconazole. These agents all share a mechanism of action, inhibition of lanosterol demethylase, as well as several mechanisms of resistance (60). Resistance to the triazole antifungal agents can arise from a modification in the quantity or quality of the target enzyme, reduced access of the drug to the target via either MDR (multidrug resistance) or CDR (Candida drug resistance) efflux pumps, or some combination of those mechanisms (31, 57, 60, 89). The ECVs for each of these triazoles were derived by considering the MIC distributions for several thousand isolates of Candida species and the relationship of resistance mechanisms to the MICs of non-WT strains (63, 67, 72, 73, 76) (Table 1).

The ECVs for fluconazole and 11 different species of Candida were derived from a global surveillance database of 15,839 MIC values, all determined by the 24-h CLSI BMD method (63, 65, 71, 74) (Table 1). The fluconazole ECVs ranged from 0.5 μg/ml to 2 μg/ml for 7 of the 11 species and encompassed 95% to 99% of the results in each MIC distribution. In the cases of C. glabrata (ECV, 32 μg/ml), C. krusei (ECV, 64 μg/ml), C. guilliermondii (ECV, 8 μg/ml), and C. pelliculosa (ECV, 4 μg/ml), the elevated ECVs depict the decreased susceptibility to fluconazole that is intrinsic to those species (68).

The relationship between the fluconazole MIC and the various resistance mechanisms (MDR and CDR efflux pumps, overexpression/mutation of ERG11 [encodes the target enzyme]) has been derived by studying serial isolates of C. albicans from AIDS patients with recurrent orophryngeal candidiasis (OPC) (11, 40, 79, 95–97), as well as from patients with IC (43, 44, 47, 49), and in genetically manipulated strains (41).

One of the first demonstrations of the relationship between the fluconazole MIC for C. albicans and fluconazole resistance mechanisms is found in the work of White and colleagues (95–97), who investigated the resistance mechanisms expressed in an isogenic set of 17 sequential isolates of C. albicans from a single HIV-infected patient with relapsing OPC treated with increasing doses of fluconazole (Table 2). As the fluconazole MIC for successive isolates increased progressively from 0.25 μg/ml (WT) against the pretreatment isolate (isolate 1) to 64 to 128 μg/ml against isolates 16 and 17, the number of identified resistance mechanisms increased from overexpression of the MDR1 efflux pumps (resulting in the MIC increasing from 0.25 μg/ml [WT] to 8 μg/ml [non-WT]) through point mutations in ERG11, loss of allelic variation in ERG11 and overexpression of ERG11 (MIC increased from 8 μg/ml to 16 to 32 μg/ml), and, finally, overexpression of the CDR1 efflux pumps, resulting in high-level resistance to both fluconazole (MIC, 64 to 128 μg/ml) and itraconazole (MIC, 4 to 8 μg/ml). This isolate set demonstrated a stepwise, quantitative increase in fluconazole resistance in which the fluconazole MIC depended on the number and type of resistance mechanisms expressed in each isolate. Isolates for which the fluconazole MIC was less than the ECV of 0.5 μg/ml (Table 1) were found to lack any of the described resistance mechanisms, whereas those for which the MIC was greater than the ECV all possessed one or more fluconazole resistance mechanisms.

Table 2.

Azole resistance mechanisms as they relate to MIC in serial isolates of C. albicans from an HIV-infected patient with recurrent oropharyngeal candidiasis^a

C. albicans isolate(s)	MIC (μg/ml)		Molecular change(s)
	FLC	ITR
1	0.25	0.06	None (WT)
3	8	0.06	Increase in MDR1 mRNA
12–15	16–32	0.12–0.25	Mutation in ERG 11 gene, loss of heterozygosity in ERG 11, increase in ERG11 mRNA
16, 17	64–128	4–8	Increase in CDR mRNA

^aData compiled from references 79, 95, 96, and 97. FLC, fluconazole; ITR, itraconazole; WT, wild type.

The ECVs for itraconazole and eight species of Candida were derived from an MIC database of more than 30,000 values, all determined by the 24-h CLSI BMD method (76) (Table 1). The ECV was lowest for C. albicans (0.12 μg/ml), was 0.25 μg/ml for C. dubliniensis, and was 0.5 μg/ml for all other species, with the exception of C. glabrata (2 μg/ml), C. krusei (1 μg/ml), and C. guilliermondii (1 μg/ml). Aside from C. albicans, these ECVs are all higher than the susceptible CLSI CBP of ≤0.12 μg/ml (83, 85). This CBP was assigned based entirely on MICs and clinical outcomes for isolates of Candida spp. (90% of which were C. albicans) obtained from patients with OPC who were treated with oral itraconazole (capsule and/or solution) and in whom serum concentrations of itraconazole of less than 0.5 μg/ml were common (83).

An example of the relationship between the itraconazole MIC for C. albicans and various azole resistance mechanisms is evident from the data in Table 2, where a shift from a WT MIC phenotype (MIC, ≤0.12 μg/ml) to a non-WT phenotype (MIC, 4 to 8 μg/ml) is influenced primarily by the overexpression of the CDR efflux pumps. Given the ECVs shown in Table 1, it is clear that the CLSI CBPs for itraconazole are not appropriate for any species other than C. albicans. Whereas the existing CLSI CBPs for itraconazole should provide an optimal means for detecting decreased susceptibility among isolates of C. albicans, the ECVs shown in Table 1 should be used for this purpose for all other species.

The ECVs for voriconazole and posaconazole and 11 different species of Candida were derived from a database of 17,010 MIC values, all determined by the 24-h CLSI BMD method (73, 74). The ECV for voriconazole was lowest for C. kefyr (0.015 μg/ml) and was 0.03 to 0.12 μg/ml for all other species, with the exception of C. glabrata (0.5 μg/ml), C. krusei (0.5 μg/ml), C. guilliermondii (0.25 μg/ml), and C. pelliculosa (0.25 μg/ml) (Table 1). Likewise, the ECV for posaconazole was lowest for C. albicans (0.06 μg/ml) and was 0.12 to 0.25 μg/ml for C. tropicalis, C. parapsilosis, C. lusitaniae, C. kefyr, and C. dubliniensis (Table 1). Similar to voriconazole, the species with higher posaconazole ECVs were C. glabrata (2 μg/ml), C. krusei (0.5 μg/ml), C. guilliermondii (0.5 μg/ml), and C. pelliculosa (2 μg/ml), reflecting the decreased susceptibility to the azoles that is intrinsic to these species (68). These ECVs encompass 94% to 100% of the results in each MIC distribution (73, 74).

As with fluconazole, the relationship between the voriconazole and posaconazole MICs for C. albicans and azole resistance mechanisms has been demonstrated using an isogenic series of isolates with one or more resistance mechanisms (41). MacCallum et al. (41) employed sequential genetic manipulations of a single strain of C. albicans to demonstrate the impact of the level of expression of CDR efflux pumps and the presence or absence of mutations in ERG11 on the level of resistance to voriconazole and posaconazole (Table 3). An increase in the expression of CDR genes coupled with a mutation in both of the ERG11 alleles resulted in an increase of the voriconazole MIC from 0.007 μg/ml in the WT parental strain to 2 μg/ml (greater than the ECV of 0.03 μg/ml) in the mutant strain. A similar increase in the posaconazole MIC from 0.03 μg/ml (WT) to 0.25 μg/ml (greater than the ECV of 0.06 μg/ml) was observed. A more modest effect on voriconazole susceptibility (MIC increase from 0.007 μg/ml to either 0.06 μg/ml or 0.12 μg/ml) was seen with mutations in one or both ERG 11 alleles coupled with a basal level of CDR expression. Notably, mutations in ERG11 had no effect on the posaconazole MICs. This is consistent with the understanding that certain mutations near the heme site of the C. albicans lanosterol demethylase result in significant levels of resistance to voriconazole (and fluconazole) but have less effect on the susceptibility of the organisms to posaconazole (and itraconazole). This differential susceptibility to the various azoles is thought to be due to the additional contacts with the target afforded by the long side chains of posaconazole and itraconazole, allowing these agents to retain activity despite decreased target affinity for voriconazole and fluconazole (98). Increased expression of CDR with WT ERG11 resulted in 8-fold and 16-fold increases in the MICs of posaconazole and voriconazole, respectively. Thus, isolates of C. albicans for which the voriconazole and posaconazole MICs are less than their respective ECVs (0.03 μg/ml and 0.06 μg/ml, respectively) lack any of the described azole resistance mechanisms whereas those for which the MIC was greater than the ECV possess one or more azole resistance mechanisms.

Table 3.

Impact of resistance mechanisms on the in vitro susceptibility of C. albicans to voriconazole and posaconazole^a

Strain	Resistance mechanism(s)		MIC (μg/ml)
	CDRb	ERG 11c	VRC	PSC
DSY294	Basal	WT/WT	0.007	0.03
DSY296	Increase	G464S/G464S	2	0.25
DSY3083	Basal	G464S/G464S	0.12	0.03
DSY3604	Basal	G464S/WT	0.06	0.03
DSY3606	Increase	WT/WT	0.12	0.25

^aData compiled from reference 41. VRC, voriconazole; PSC, posaconazole; WT, wild type.

^bData represent the level of expression of CDR1/CDR2 efflux pumps.

^cData represent wild type or a mutation (G464S) in either or both of two ERG 11 alleles.

Echinocandins.

All three of the echinocandins (anidulafungin, caspofungin, and micafungin) are approved for the treatment of candidemia and other forms of IC and are considered to be the agents of first choice for the initial treatment of most episodes of IC (52). These agents share both a common mechanism of action, inhibition of the glucan synthase (GS) enzyme complex, as well as a common mechanism of resistance (56). Echinocandin resistance in C. albicans, C. tropicalis, and C. krusei is associated with point mutations in the fks 1 gene (encodes the target subunit of GS) (56). Likewise, mutations in both fks 1 and fks 2 are responsible for clinical echinocandin resistance in C. glabrata (56, 99). These mutations, which result in elevated MICs (4-to-30-fold MIC increases for caspofungin and 90-to-110-fold increases for anidulafungin and micafungin), reduce the sensitivity of GS to drug inhibition by 30- to 1,000-fold (28, 29). Among the less susceptible species, C. parapsilosis and C. guilliermondii both possess a naturally occurring polymorphism at the edge of hot spot 1 (HS1) in fks 1 that accounts for the MICs of the echinocandins for these species being elevated relative to the WT strains of other species (27).

The ECVs for the echinocandins and 11 different species of Candida were derived from a database of more than 9,000 MIC values all determined by 24-h CLSI BMD methods (67, 74) (Table 1). The ECVs for all three echinocandins were ≤0.25 μg/ml for 7 of the 11 species and encompassed 96% to 100% of the results in each MIC distribution (67, 74). In the case of C. parapsilosis (ECVs of 1 μg/ml to 4 μg/ml), C. lusitaniae (ECVs of 0.5 μg/ml to 2 μg/ml), C. guilliermondii (ECVs of 2 μg/ml to 4 μg/ml), and C. orthopsilosis (ECVs of 0.5 μg/ml to 2 μg/ml), the elevated ECVs clearly show the decreased susceptibility to the echinocandins that is intrinsic to these species.

The ability of the echinocandin ECVs to differentiate WT strains of Candida spp. from those with acquired resistance mutations was demonstrated by applying the ECVs for each species to a collection of 229 WT isolates (no fks mutations) and 50 isolates with fks mutations (70). Using the anidulafungin ECVs for C. albicans (0.12 μg/ml), C. glabrata (0.25 μg/ml), C. tropicalis (0.12 μg/ml), and C. krusei (0.12 μg/ml), the CLSI BMD method correctly classified 90% (45 of 50) of the mutant strains as non-WT and 98% (224 of 229) of those without fks mutations as WT. With caspofungin as the test reagent and ECVs of 0.12 μg/ml (C. albicans, C. glabrata, and C. tropicalis) and 0.25 μg/ml (C. krusei), the CLSI BMD method correctly classified 98% (49 of 50) of mutant strains and 91% (209 of 229) of WT strains. The ECVs for micafungin are 0.03 μg/ml for C. albicans and C. glabrata and 0.12 μg/ml for C. tropicalis and C. krusei (Table 1). Using these ECVs, the CLSI method with micafungin correctly classified all 50 mutant strains and 190 (83%) of 229 WT strains (70). Thus, as seen with flucytosine and the triazoles, the ECVs for the echinocandins provide a sensitive and specific means of differentiating WT from non-WT strains of Candida using the CLSI BMD method.

NEW CLSI SPECIES-SPECIFIC CLINICAL BREAKPOINTS (CBPs) FOR FLUCONAZOLE, VORICONAZOLE, AND THE ECHINOCANDINS AND THE FREQUENTLY ENCOUNTERED SPECIES OF CANDIDA

The CLSI Subcommittee for Antifungal Susceptibility Testing has established CBPs for fluconazole, voriconazole, and the echinocandins versus Candida spp. by taking into account the MIC distributions, pharmacokinetic (PK) and pharmacodynamic (PD) parameters, resistance mechanisms, and clinical outcomes as they relate to MIC values (58, 59, 62). Initially the CLSI Subcommittee did not allow for species-specific CBPs and assigned values for susceptibility (S) of ≤8 μg/ml for fluconazole, ≤1 μg/ml for voriconazole, and ≤2 μg/ml for the echinocandins to be applied to all species of Candida despite clear evidence that the MICs for these agents were significantly lower for some species than others and that clinical outcome data were lacking for all but the six most common species (57). A comparison of the ECVs for the various species-antifungal agent pairs shown in Table 1 with those CBPs shows that the originally proposed CBPs were too high to provide a sensitive means of predicting the emergence of resistance among the highly susceptible species and at the same time bisected the WT MIC distributions of other species (e.g., C. glabrata and fluconazole) (2, 3). As a result, the CLSI Subcommittee reconsidered the MIC distributions for each species and antifungal agent, developed ECVs as shown in Table 1, compiled more data on the relationship of resistance mechanisms to both MICs and outcomes, and related this information to the available PK/PD and outcome data for each species to arrive at species-specific CBPs that both are predictive of clinical outcomes and provide a more sensitive means of detecting the emergence of resistance. The details of these deliberations are well described in the relevant publications (65, 69, 70) and are not repeated here. The revised CBPs for fluconazole, voriconazole, anidulafungin, caspofungin, and micafungin are provided in Table 1. CBPs for these agents are applicable only to the six species shown in Table 1 due to the lack of sufficient clinical outcome data for the less common species. In lieu of CBPs for other species of Candida, the ECVs shown in Table 1 should be used to detect the emergence of strains that have decreased susceptibility to the triazoles and echinocandins. Likewise, given the absence of CBPs for most species and amphotericin B, flucytosine, itraconazole, and posaconazole, the ECVs should be used in efforts to detect the emergence of potential resistance to these agents. Additional clinical outcome data and investigations of resistance mechanisms prevalent among the less common species are required before CBPs can be assigned.

RATIONALE FOR, CLINICAL IMPACT OF, AND RECOMMENDATIONS FOR THE USE OF ANTIFUNGAL SUSCEPTIBILITY TESTING OF CANDIDA SPECIES AGAINST SYSTEMICALLY ACTIVE ANTIFUNGAL AGENTS

The primary objective of all in vitro antimicrobial (e.g., antifungal, antibacterial, and antiviral) susceptibility testing is to predict the likely impact of administration of the tested agent on the outcome of infection caused by the treated organism or similar organisms (86, 93). Antifungal susceptibility testing is performed for the same reasons as antibacterial testing (83, 85, 86): (i) to provide a reliable estimate of the relative activities of two or more antimicrobial agents against the pathogen of interest; (ii) to correlate with in vivo activity and to predict the likely outcome of therapy; (iii) to provide a quantitative means by which to survey the development of resistance among members of a normally susceptible population of organisms; and (iv) to predict the therapeutic potential and spectrum of activity of newly developed investigational agents.

In the clinical microbiology laboratory, the focus of antifungal susceptibility testing is directed toward a specific clinical isolate causing infection in an individual patient. Recent studies examining the clinical use of “real-time” antifungal susceptibility testing have shown that when such testing is available on site, physicians find the results helpful and frequently alter therapy based on the results (8, 32, 33, 37, 42, 52, 54). Collins et al. (15) reported that susceptibility testing of C. glabrata isolates results in lower overall treatment costs, based on de-escalation in therapy from an expensive echinocandin to fluconazole for patients with documented C. glabrata fungemia. Those authors suggest that antifungal susceptibility testing is a necessity in today's world of resistant organisms and expensive agents (15). Likewise, Parkins et al. (54) suggest that accurate and timely antifungal susceptibility testing may be more important than has been recognized previously. In a population-based survey conducted in Canada between July 1999 and June 2004, they found that empirical therapy with an adequate antifungal agent (isolate susceptible in vitro) was associated with a significant reduction in all-cause morality from 46% to 27% (P = 0.02). Notably, empirical fluconazole therapy was more likely to be deemed inadequate and inadequate therapy was an independent predictor of in-hospital death. Thus, it appears that routine antifungal susceptibility testing can serve as an adjunct in the treatment of candidemia in the same way that antibacterial testing aids in the treatment of bacterial infections (6, 17, 26, 86). In considering these findings, one must understand that the prediction of outcome in a complex and dynamic biological system, such as a clinical infection, from results obtained in an artificial and well-defined matrix (in vitro susceptibility test) is an inherently error-prone process in which only modest degrees of correlation can be expected (83, 85, 86).

In order to be useful clinically, in vitro susceptibility testing of antimicrobial agents should reliably predict the in vivo response to therapy in human infections. However, the in vitro susceptibility of an infecting organism to the administered antimicrobial agent is only one of the factors that may influence the likelihood that therapy for an infection will be successful (17, 57, 64, 75, 86, 93). Factors related to the host immune response, severity of underlying disease, drug pharmacokinetics and pharmacodynamics, drug interactions, and proper patient management and factors related to the virulence of the infecting organism and its interaction with both the host and the antimicrobial agent all influence the outcome of treatment of an infectious episode (17, 64, 86, 93). In order to appreciate the clinical value one can expect from antifungal susceptibility testing, it must be understood that after more than 40 years of study, in vitro susceptibility testing can be said to predict the outcome of bacterial infections with an accuracy that has been summarized as the “90–60 rule” (17, 86): infections due to isolates that are susceptible to the agent being given respond to therapy approximately 90% of the time, whereas infections due to isolates that are resistant to the agent being given respond approximately 60% of the time. There is now a considerable body of data indicating that standardized antifungal susceptibility testing (CLSI M27-A3; EUCAST EDef 7.1) for several organism-drug combinations (most notably Candida spp. and azole antifungal agents) provides results that have a predictive utility consistent with the “90–60 rule” (65, 69, 70, 86) (Table 4).

Table 4.

Clinical success for patient-episode-isolate events treated with fluconazole, voriconazole, or itraconazole by their respective CLSI MIC interpretive categories for Candida spp.

Antifungal agent	MIC breakpoint interpretive category (μg/ml)	No. of events	% success
Fluconazolea	Susceptible (≤2 μg/ml)	550	92
	Resistant (≥8 μg/ml)	212	37
Voriconazoleb	Susceptible (≤0.12 μg/ml)	173	76
	Resistant (≥1 μg/ml)	8	38
Itraconazolec	Susceptible (≤0.12 μg/ml)	103	88
	Resistant (≥1 μg/ml)	6	67

^aData compiled from Pfaller et al. (65) for C. albicans, C. tropicalis, and C. parapsilosis.

^bData compiled from Pfaller et al. (69) for C. albicans, C. tropicalis, and C. parapsilosis.

^cData compiled from Rex et al. (83) for C. albicans and patient isolates for which plasma concentrations of itraconazole were >0.5 μg/ml.

Antifungal resistance results in elevated MICs that are associated with poorer outcomes and breakthrough infections during antifungal treatment and prophylaxis. Antifungal resistance and its negative consequences can often be traced to acquisition of a particular resistance mechanism. The most obvious consequence of antifungal resistance may be seen in the results shown in Table 4, where the clinical outcome was significantly poorer for those patients infected with isolates of Candida for which the MIC of fluconazole, voriconazole, or itraconazole was classified as R than for those for which the MIC was classified as S. Similarly, Baddley et al. (9) reported a lower mortality rate among patients with candidemia for which the fluconazole MIC of the infecting isolate was ≤2 μg/ml (S) than among those for which the MIC was ≥8 μg/ml (R). Taken together, these data indicate that isolates with high (R) azole MICs obtained from patients with Candida infections are associated with lower treatment success rates and higher mortality than those with low or susceptible MICs, illustrating the negative impact of antifungal resistance on clinical outcomes (17, 57).

One obstacle to demonstrating clinical relevance of antifungal susceptibility testing within a single Candida species that is usually azole (C. albicans, C. tropicalis, C. parapsilosis) or echinocandin (C. albicans, C. tropicalis, C. glabrata) susceptible is the absence of sufficient numbers of isolates in clinical trials that are resistant to the drug of interest (65, 69, 70, 86). In order to establish a relationship between MIC and clinical outcome, one requires not only sufficient numbers of resistant isolates but also a sufficient number of patients treated with the drug to which the isolate is later shown to be resistant (17, 86). It is often well after a given drug is introduced into clinical practice that sufficient numbers of clinical failures or breakthrough infections are detected to allow the establishment of a resistant susceptibility testing category (70, 91).

Antifungal resistance can lead to breakthrough invasive fungal infections in high-risk patients receiving antifungal prophylaxis. For example, Alexander et al. (1) described eight cases of breakthrough fungemia among 295 adult bone marrow transplant (BMT) recipients receiving fluconazole prophylaxis between October 2002 and June 2004 at Duke University Medical Center. Among the eight cases of breakthrough fungemia, seven were due to C. glabrata, and four of the seven exhibited cross-resistance to all azoles (fluconazole, itraconazole, posaconazole, and voriconazole). Although the resistance mechanism responsible for the pan-azole resistance was not elucidated, it was likely due to elevated CDR gene-encoded efflux pump activity, as this is prevalent in C. glabrata and has been associated with cross-resistance among azole antifungal agents.

Another Duke University Medical Center study examined cases of breakthrough candidiasis among BMT or solid-organ-transplant recipients receiving micafungin prophylaxis (77). Between February 2006 and May 2008, 649 high-risk patients received at least three doses of micafungin and 12 (1.8%) subsequently developed IC involving a total of 19 isolates of Candida (7 of C. parapsilosis, 6 of C. glabrata, 3 of C. tropicalis, and 1 each of C. albicans, C. dubliniensis, and C. krusei). Among the breakthrough isolates, micafungin MICs were elevated for 5 of 7 isolates of C. parapsilosis (MIC range, 4 to 8 μg/ml), for 5 of 6 C. glabrata isolates (MIC range, 4 to 8 μg/ml), and for 2 of 3 C. tropicalis isolates (MICs, 2 μg/ml). All of the C. glabrata and C. tropicalis isolates for which the micafungin MICs were elevated were found to possess fks gene mutations and were cross-resistant to both anidulafungin and caspofungin, establishing the mutational event as important for both an increase in MIC and clinical failure, i.e., breakthrough infection.

Given the data discussed in this review, how then should one use antifungal susceptibility testing results in the care of patients with IC? Guidelines for the use of antifungal susceptibility testing, and other laboratory studies, have been developed (17, 86) and are presented in Table 5. Selective application of antifungal susceptibility testing, coupled with broader identification of Candida to the species level, should prove useful, especially in difficult-to-manage cases of IC (52).

Table 5.

Recommendations for use of antifungal susceptibility testing of Candida spp. in the clinical laboratory

Clinical setting	Recommendation(s)
Routine	Species-level identification of all Candida isolates from deep sites (e.g., blood, normally sterile body fluids, tissues, abscesses)
	Routine antifungal testing of fluconazole and an echinocandin against C. glabrata from deep sites
	Routine testing of fluconazole and an echinocandin against other species of Candida possibly helpful but susceptibility usually predictable by species
	Use CBPs or ECVs to interpret results as appropriate (Table 1)
	Consider cross-resistance between fluconazole and all other azoles to be complete for C. glabrata
	Create an antifungogram
Mucosal candidiasis	Determination of azole susceptibility not routinely necessary
	Susceptibility testing of azoles may be useful for patients unresponsive to therapy
Invasive disease with clinical failure of initial therapy	Consider susceptibility testing as an adjunct—amphotericin B, flucytosine, fluconazole, voriconazole, and an echinocandin
	Consultation with an experienced microbiologist recommended
Infection with species with high rates of intrinsic or acquired resistance	Susceptibility testing not necessary when intrinsic resistance is known
	C. lusitaniae and amphotericin
	C. krusei and fluconazole, flucytosine
	C. guilliermondii and echinocandins
	With high rates of acquired resistance, monitor closely for signs of failure and perform susceptibility testing
	C. glabrata and fluconazole, amphotericin B, and echinocandins
	C. krusei and amphotericin B
	C. guilliermondii and amphotericin B
	C. rugosa and amphotericin B, fluconazole, and echinocandins
New treatment options (e.g., echinocandins, voriconazole, posaconazole) or unusual species	Susceptibility of Candida spp. to echinocandins may be assumed unless initial response is suboptimal
	Susceptibility testing warranted if prior exposure to echinocandins or fluconazole
	Selection of therapy based on published consensus guidelines (52) and review of survey data on the organism-drug combination in question
	Susceptibility testing may be helpful when patient is not responding to what should be effective therapy
Patients who respond to therapy despite being infected with an organism later found to be resistant	Best approach not clear
	Take into account severity of infection, patient immune status, consequences of recurrent infection, etc.
	Consider alternative therapy for infections with isolates that appear to be highly resistant to initial therapy
Selection of susceptibility testing methods	Standardized methods
	CLSI methods
	Broth based, M27-A3
	Agar based, M44-A2
	EUCAST EDef 7.1
	Commercial methods
	Etest
	Sensititre YeastOne
	Vitek 2

SUMMARY

Antifungal susceptibility testing of Candida has benefited greatly from the efforts at standardization conducted by both the CLSI and EUCAST organizations. Progress in refining the CLSI approach to the establishment of CBPs has led to the generation of new species-specific CBPs for the triazoles and echinocandins and the major species of Candida. In lieu of CBPs for those agents and species where clinical data are lacking, ECVs have been established which may serve as sensitive markers for the emergence of decreased susceptibility to the agent of interest. In most instances, the ECVs have been shown to separate those non-WT strains with acquired or mutational resistance mechanisms from WT strains. The role of antifungal susceptibility testing in the management of patients with IC is now coming into focus, and it is clear that such testing can aid in the selection of agents for primary therapy as well as in a de-escalation strategy. Antifungal susceptibility testing continues to progress and to refine both methods and interpretive capabilities in order to aid in optimizing the care of patients with candidemia.