Introduction

Invasive fungal disease (IFD) caused by Candida and Cryptococcus species represents significant morbidity with mortality rates as high as 70–75% [1, 2]. Rapid diagnosis including detection of drug resistance is essential for early initiation of appropriate antifungal treatment [3]. Candida comprise the commonest yeasts encountered in the clinical mycology laboratory where Candida albicans remains the most frequent; however, the prevalence of species with reduced susceptibility or resistance to antifungal drugs, most notably Candida glabrata, is increasing due to selective pressures from increased antifungal drug use [4]. C. glabrata and Candida parapsilosis are the second and third most common causes of candidaemia in Australia [5], North America [6] and Northern Europe [7•] whilst in India [8] and Latin America [9], Candida tropicalis is the more, if not most, frequent species, although in some studies from Latin America, C. parapsilosis is the second most common species after C. albicans. Of concern, the frequency of azole resistance and of echinocandin resistance within C. glabrata has reached up to 30 and 23%, respectively, in some countries, and may be centre-specific. Currently, rates of azole resistance amongst C. tropicalis are estimated at 8–16.7% [5, 8].

Other than Candida, the two main species of the Cryptococcus neoformans-Cryptococcus gattii complex (C. neoformans and C. gattii) are typically the next most common yeasts encountered. Each of C. neoformans and C. gattii, incorporating their hybrid species, comprises four major molecular types (VNI-VNIV and VGI-VGIV, respectively) [10]. As these molecular types may be associated with differences in virulence and antifungal susceptibility [11•, 12], elucidating their identification and drug susceptibility profiles is important. C. gattii isolates are reported to be less susceptible to some antifungal drugs than isolates of C. neoformans, and in some studies, there have been statistically significant differences in fluconazole minimum inhibitory concentration (MIC) values amongst molecular types of C. gattii, with type VGII having the highest fluconazole geometric mean (GM) MICs [13,14,15].

This article summarises the methods used in diagnostic mycology laboratories to detect drug resistance and the challenges thereof in Candida and Cryptococcus species. We also briefly describe the major molecular mechanisms of drug resistance and whether phenotypic susceptibility testing methods correlate with the molecular basis for resistance. For this review, the term ‘Candida’ is used to encompass the main ascomycetous yeasts, which have historically been pooled into the artificial genus Candida (including Pichia, Clavispora and Meyerozyma) [3, 4].

Antifungal Susceptibility Testing Methods

In vitro antifungal susceptibility testing is performed to guide selection of antifungal drug therapy but is also important to detect drug resistance, or probability that treatment will fail [16]. Although species identification can help predict resistance in intrinsically resistant species, e.g. Candida krusei and fluconazole, it cannot infer presence of acquired resistance.

There are two reference standards for susceptibility testing of yeasts: the Clinical and Laboratory Standards Institute (CLSI) methods [17] and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) methods [18]. Both are based on broth microdilution (BMD) and, despite technical differences, e.g. inoculum size and determination of MIC endpoint, are able to reliably discriminate between susceptible ‘wild-type’ strains (no acquired resistance mechanism) and resistant strains with intrinsic or acquired resistance [16]. Both methodologies have developed clinical breakpoints (CBPs) [19] and epidemiological cut-off values (ECVs) for Candida spp. [20]. The CBP is the concentration (mg/L) of drug which defines whether the organism is susceptible, susceptible-dose dependent (S-DD) or resistant and is based on the in vitro response of the organism to the antifungal at levels corresponding to blood or tissue levels attainable with usually prescribed doses of that agent [17]. The ECV is the upper limit of the wild-type population and usually incorporates 90–95% of a given population for a specific agent [21•], i.e. it is the MIC concentration that separates a population into those with and without acquired and/or mutational resistance based on their phenotype [22]. CBPs and/or ECVs have been established for the more common species of Candida for most antifungal drugs [7•, 18,19,20, 23] and ECVs also developed for Cryptococcus species (see section on ‘Cryptococcus’). Importantly, these CBPs or ECVs are species-specific and the reader is referred to the appropriate CLSI or EUCAST documents.

Because these ‘gold standard’ reference techniques require 48–72 h to perform [24], many clinical laboratories utilise commercial tests such as the Sensititre® YeastOne® (TREK Diagnostics Systems, Cleveland, Ohio) and Etest (bioMérieux, Marcy d’Etoile, France) for susceptibility testing for convenience, but standardisation of these methods against the reference BMD tests is not absolute for all drug-bug combinations and may result in the misclassification of both ‘susceptible’ and ‘resistant’ Candida strains when reference CBPs are applied [25]. Guinea et al. [26] have demonstrated that the Etest performed directly on positive blood samples performed well with the azoles (fluconazole, voriconazole and isavuconazole) compared with the CLSI standard [17], offering a rapid (24 h) susceptibility result in patients with candidaemia. The Sensititre® YeastOne® has also been used to test susceptibility of cryptococcal isolates (see section on ‘Cryptococcus’).

The Vitek 2 (bioMérieux) automated yeast identification and susceptibility system, which reduces result variability by using spectrophotometric reading of results [27], provides MIC results more rapidly within 14–18 h [28]. Comparative studies between CLSI BMD [17, 19] and Vitek 2 MICs in Candida species have generally shown excellent essential agreement (EA) (within 2 dilutions) as well as categorical agreements (CA) for fluconazole (97.9 and 96.8%), voriconazole (96.7 and 96.5%), caspofungin (99.5 and 99.8%), micafungin (98.6 and 98.2%) and posaconazole (95.6 and 98.1%) [27, 29]. However, intra-species discrepancies in MICs were observed particularly for C. glabrata and C. krusei; the Vitek 2 is also unable to detect the susceptibility of C. neoformans isolates to voriconazole [30]. Further, when Candida auris isolates were tested for amphotericin B susceptibility, highly elevated MICs were noted but with only 10% EA between Vitek and the CLSI reference method which may lead to the selection of inappropriate therapy [31].

Candida Species

Azole Resistance in Candida

The widespread use of triazoles has resulted in both primary resistance, observed as a shift towards inherently less susceptible species (e.g. C. glabrata and C. krusei), and secondary resistance, which selects for resistant sub-populations in normally susceptible strains (e.g. C. albicans) [32]. Secondary or acquired resistance is less predictable and may be under-reported [7•]. The prevalence of azole resistance varies with species and even within species. C. glabrata exhibits intrinsic decreased susceptibility to azoles and may develop high-level resistance following azole exposure [33•]. Amongst C. albicans, bloodstream isolates have been associated with the lowest incidence (0–5%) of azole resistance [34, 35] with those from the oropharynx after fluconazole treatment exhibiting higher resistance rates [33•]. Surveys of invasive candidiasis have noted overall fluconazole resistance rates of 10–25% with many isolates, particularly C. glabrata, cross-resistant to other triazoles [24, 33•]. Isavuconazole, the most recent marketed azole, has broad in vitro activity against Candida species similar to that of voriconazole; up to 15% of C. glabrata and C tropicalis may have non-WT MICs depending on the centre [36•].

Azole resistance implicating a range of candidate genes inked to resistance has been most extensively studied in C. albicans. Studies have examined orthologous genes in C. parapsilosis and the recently described Candida auris with mixed results whilst comparatively little is known about azole resistance in C. tropicalis [33•, 37•]. There are four main mechanisms of azole resistance in Candida species (summarised in Table 1): (i) induction of drug efflux pumps encoded by the Multi-Drug Resistance (MDR) and Candida Drug Resistance (CDR) genes, which decrease drug concentrations at the site of the target enzyme, 14-α-sterol demethylase; (ii) acquisition of point mutations in the ERG11 gene which encodes the target enzyme resulting in a target with reduced affinity for azoles; (iii) overexpression or up-regulation of ERG11 which helps overcome azole activity and (iv) development of bypass pathways, linked to mutations of the ERG3 gene, which negate the fungal membrane-disruptive effects of azoles [16]. A resistant strain may have more than one mechanism active with additive effects. Induction of CDR gene-encoded efflux pumps usually affects all azole drugs and is sufficient for resistance. However, efflux pumps encoded by MDR genes are usually selective for fluconazole resistance only [16].

Table 1 Genetic mechanisms leading to resistance by antifungal compound in Candida species and molecular detection tools (modified from Perlin [32] and Cuence-Estrella [24])
Full size table

Detecting Azole Resistance in Candida

In vitro susceptibility testing is useful to guide classification of an isolate as susceptible or resistant, although it cannot determine if an isolate harbours a mechanism of acquired antifungal resistance [38]. Nonetheless, for the azoles, with the exceptions of isavuconazole where data is relatively lacking and C. glabrata, species-specific interpretative CBPs have in general been shown to have utility in predicting outcomes [39]. Hence, for the clinical laboratory, phenotypic MIC testing remains a core function.

Conversely, molecular assays have the ability to determine the underlying genetic basis of triazole resistance [32]. However, they have the drawback in that they are not standardised and that in many circumstances, correlation of presence of a genetic change with phenotypic resistance is uncertain (Table 1) [24].

Mutations in the ERG11 gene are clustered into three ‘regional hotspots’ within the genome corresponding to amino acid positions 105 to 165, 266 to 287 and 405 to 488 [40]. Direct sequence analysis of the ERG11 gene is the most direct method of detecting these point mutations [41] but other methods include a simple and rapid (2 h) padlock probe and rolling circle amplification (RCA)-based method for their detection [42] with good specificity. Whether such gene mutations confer phenotypic resistance to the azoles in diploid Candida species such as C. albicans is complicated by the need to demonstrate mutations in one or more alleles, and to demonstrate a causal effect by experiments involving matched pairs of susceptible and resistant isolates from the same patient or by gene expression studies [43].

Quantitative reverse transcriptase-PCR (RT-PCR) has now replaced semi-quantitative Northern blot analysis to assess transcript levels of CDR1, CDR2 and MDR1 genes [44, 45]. Unlike specific mutation detection, expression profiling of these genes requires cell cultures grown in the presence/absence of drug and threshold levels need to be associated with the resistance phenotype [32]. One study used RT-PCR to quantify mRNA levels in CDR1, CDR2, ERG11 and MDR1 genes in the absence of azole exposure in C. albicans isolates with variable fluconazole susceptibilities [44]. ERG11 expression levels were significantly higher in the fluconazole resistant isolates compared with S-DD and susceptible isolates, as well as in S-DD isolates compared with the susceptible isolates. Additionally, CDR1 and CDR2 overexpression, but not MDR1, was noted amongst resistant but not susceptible isolates. The assay had improved sensitivity, specificity and speed of analysis over Northern hybridization and was able to quantify very low and high mRNA levels [44].

A recent study by Xisto et al. [9] found 50% of C. tropicalis isolates (n = 22) were phenotypically resistant to one or more azoles and six were pan-azole resistant. By conventional PCR and DNA sequencing to amplify hot spot region 1 (HS1) (amino acid positions 641–649) and hot spot region 2 (HS2) (amino acid positions 1357–1364) in ERG11, a novel amino acid substitution at location K143R, near the active binding site of ERG11, together with an up-regulation of the ABC transporters, was associated with phenotypically pan-azole resistant isolates. K143R has been associated with fluconazole resistance in C. albicans [9].

Most recently, next generation sequencing (NGS) has been used to analyse genes commonly linked to azole resistance (ERG11, ERG3, TAC1 and CgPDR1) in C. albicans, C. glabrata and C. parapsilosis [46]. NGS provides a genome-wide view of gene mutations and is able to detect new combinations of mutations that would probably be missed by targeted DNA sequencing [46]. Lockhart et al. [37•] compared ERG11 amino acid sequences for C. auris obtained by NGS to C. albicans ERG11 and identified nine amino acid substitutions associated with azole resistance as well as three additional hotspot amino acid substitutions (F126T, Y132F and K143R) that are either proposed or proved to confer fluconazole resistance in C. albicans. However, NGS is less accessible to clinical laboratories than Sanger sequencing and although costly (AUD 80/sample), these costs are likely to decrease with technological advancements.

Mechanisms of Echinocandin Resistance in Candida

Echinocandin resistance in Candida species (including C. albicans, C. glabrata, C. krusei, C. tropicalis and C. dubliniensis) has been linked to point mutations in two ‘hotspot’ regions (HS1: region 640–650 and HS2: 1345–1365) of the FKS1 gene, which encodes the major subunit of 1,3-β-d-glucan synthase complex. Additional mutations in the FKS2 gene of C. glabrata are also linked to resistance [3, 16, 47, 48]. The most frequent amino acid changes in C. albicans is at positions Ser641 and Ser645, whereas in C. glabrata, amino acid substitutions at positions Ser629 in FKS1 and Ser663 and Phe659 in FKS2 are the most prominent [49]. These unique mutations result not only in higher MICs to one or more echinocandins but also clinical failure [50]. In C. glabrata, the presence of FKS mutations are an independent risk for treatment failure in patients with invasive candidiasis [51•].

For most Candida species, echinocandin resistance remains relatively low, for example, at <3% in C. albicans [52]; however for C. glabrata, echinocandin resistance rates is rising (from 4.9 to 12.3% in 10 years in one US hospital) [50], and many isolates also show co-resistance to azoles [49, 53, 54]. Naturally occurring polymorphisms in the FKS genes (P660A at the 3′-end of HS1) [21•] of the C. parapsilosis complex (C. parapsilosis sensu stricto, C. orthopsilosis and C. metapsilosis) and C. guilliermondii result in higher echinocandin MICs, relative to other Candida spp., but these have not been associated with treatment failure [49].

Detecting Echinocandin Resistance in Candida Species

Both CLSI [55] and EUCAST [7•] have established species-specific CBPs for the echinocandins, although EUCAST has not set CBPs for caspofungin. In addition, ECVs, which define the upper limit of the wild-type MIC population in the absence of a known FKS resistance mechanism, have been defined for anidulafungin and micafungin against common Candida species [20].

For the echinocandins, in vitro MIC susceptibility testing is not able to completely distinguish WT isolates from echinocandin-resistant (FKS mutant) isolates, especially for caspofungin [7•] where there is substantial inter-laboratory variability of results even with reference methodologies; this has led to isolates falsely classed as ‘resistant’. This limitation of MIC testing may be overcome by molecular testing to detect FKS mutations by either DNA sequencing, real-time PCR or NGS (Table 1) [46, 48, 56, 57]. Molecular testing is preferred because (i) an FKS mutation is a primary clinical indicator for reduced therapeutic response; (ii) the number of Candida strains exhibiting resistance in vitro is low and (iii) there is a limited spectrum of mutations conferring resistance [24, 51•]. A multiplex PCR assay detected resistance mutations in C. glabrata, with 98% concordance between mutation detection and echinocandin MICs [48]. The assay was inexpensive, simple and quick (<4 h), but was unable to detect a three nucleotide mutation encoding the F659 at Fks2p. DNA sequencing is accurate for identifying FKS mutations but requires multiple PCR assays, and is costly. Pham et al. [58] developed a rapid, high-throughput Luminex microsphere-based assay to identify C. glabrata FKS mutants, but this approach is also expensive.

NGS has also been used to analyse FKS gene mutations in clinical isolates of C. glabrata, demonstrating high concordance between drug MICs and genomic variation [46, 57]. Biswas et al. [57] recently used NGS to retrospectively study three strain pairs of C. glabrata from three patients where antifungal resistance developed during treatment. For two of three isolate pairs, there was a >60-fold increase in MICs to all echinocandins and NGS detected mutations in either the FKS1 (S629P) or FKS2 (S663P) genes of the resistant isolate.

Amphotericin B Resistance in Candida

Despite more than 50 years of clinical use, resistance to amphotericin B is uncommon in Candida species (1–3%), possibly due to its fungicidal activity limiting the selection of mutants [3]. There have been reports of increasing MICs to amphotericin B amongst C. krusei and C. glabrata isolates and intrinsic polyene resistance is reported for Candida lusitaniae, but resistant isolates are difficult to reproduce in vitro [59]. There are no reliable molecular tools to detect amphotericin B resistance (Table 1) and phenotypic susceptibility testing to determine the MIC remains the more clinically relevant [24] despite there being no CBPs for polyenes. Most clinicians use a MIC ≥1.0 μg/mL to flag ‘possible drug resistance’ [59].

Cryptococcus neoformans

The widespread practice of fluconazole use for long-term suppressive therapy of cryptococcosis, especially in patients with HIV/AIDS, has led to concern for development of fluconazole (and other azole) resistance [60, 61]. The guidelines of the Infectious Diseases Society of America (IDSA), however, consider primary resistance of C. neoformans to fluconazole not to be a significant clinical problem [60] and that susceptibility testing be performed as guided by clinical need, e.g. clinical failure, rather than as routine. This recommendation is based on findings of several studies from disparate geographic regions where the prevalence of fluconazole ‘resistance’ was <5% (based on a MIC of >16 mg/L—see below) [62,63,64].

Yet emergence of azole resistance or non-susceptibility may be growing. A global study from 1997 to 2007 documented an increase in resistance to fluconazole when results from time periods 1997–2000 (7.3%), 2002–2004 (10.9%) and 2005–2007 (11.7%) were compared, being most evident for isolates from the Asia Pacific, Africa and Latin America [65]. The rate of fluconazole-non-susceptible C. neoformans isolates also increased significantly from 2001 to 2012 in Taiwan [66] as in Uganda where between 2010 and 2014, there was an upward shift in fluconazole MICs (MIC90 32 mg/L vs. 8 mg/L in 1998–99) [67]. However, these results are confounded by the use of different methodologies across studies and different criteria to define ‘resistance’ or ‘non-susceptibility’. Similar variation is noted for studies reporting different (low) rates of resistance to amphotericin B and 5-flucytosine but employing cut-off MIC values [68].

Clinical Breakpoints and Epidemiological Cut-off Values

Unlike for Candida spp., the availability of methodologies to enable recognition of drug-resistant isolates by interpretative CBPs by both CLSI and EUCAST to infer susceptibility/resistance is not well defined. It is also unclear whether the current CBPs for Candida spp. are appropriate for Cryptococcus; hence, there are no standard breakpoints for C. neoformans or C. gattii for any antifungal drug. CBPs are based not only on parameters such as MIC distributions, pharmacokinetic (PK) and pharmacodynamic (PD) data but also on clinical outcomes of therapy. In cryptococcosis, the impact of drug MICs on clinical outcomes is uncertain with relatively few systematic data. An earlier review found that by using CLSI M27-A3 BMD methodology [17], high fluconazole MICs (>8–16 mg/L) for C. neoformans were predictive of clinical failure [69] as did the findings of Aller et al. [70] where reduced treatment efficacy was noted when the fluconazole MIC was >16 mg/L. However, others report that MICs do not predict clinical outcomes as also noted for amphotericin B and 5-flucytosine [67, 71].

As CBPs are not validated for the azoles, amphotericin B or 5-flucytosine, ECVs (based mostly on MIC distributions) have been proposed to help identify presence of resistance mechanisms that could lead to reduced susceptibility [72]. In a multicentre study, using CLSI M27-A3 methodology [17], ECVs of amphotericin B (0.5 to 1.0 mg/L) and 5-flucytosine (4 to 16 mg/L) were proposed where these values are both species- and molecular type-specific (Table 2). Similarly, the same consortium established ECVs for the four major azoles in clinical use, based on CLSI WT MIC distributions (Table 2): the species- and molecular type-specific ECVs of fluconazole were 8–32 mg/L, of itraconazole, 0.12–0.25 mg/L, and of voriconazole and posaconazole, 0.12–0.25 mg/L and 0.25–0.5 mg/L, respectively [73]. A similar study by the CDC, USA, using the Sensititre® YeastOne® system (TREK Diagnostics) established ECVs for each azole and the molecular types of C. gattii [11•]. In general, ECVs were identical or similar or to those of Espinell-Ingroff et al. [72, 73] and are shown as numerical values within brackets in Table 2.

Table 2 Epidemiologic cut-off values for the major species of C. neoformans and C. gattii (modified from references [11•] and [72]) to antifungal drugs
Full size table

There are a number of important issues around the interpretation of ECVs for Crypotococcus. Because clinical outcome data following treatment of cryptococcosis and PK/PD data are relatively lacking for the antifungal drugs used, the ECVs are established as a surrogate for CBPs to define WT MIC distributions and identify isolates that may not respond to therapy. For Cryptococcus, it is evident from Table 2 that the development of a single ‘one size fits all’ ECV for azoles or indeed for other agents that covers all C. neoformans and C. gattii molecular types is impractical and may be even incorrect.

Nonetheless, all azoles and amphotericin B have good in vitro activity against Cryptococcus. For fluconazole, higher GM MICs for C. gattii molecular-type VGII may be evident, and for all Cryptococcus, the GM MICs of itraconazole, posaconazole and voriconazole are lower than that for fluconazole. Selection of one azole over another purely on the basis of MICs is flawed since there are no CBPs to recommend one over another. Fluconazole (and to lesser extent itraconazole) has been time-tested as the preferred azole for maintenance therapy of cryptococcosis, although the newer azoles also have a therapeutic role as second line or alternate agents where fluconazole is not able to be used [60].

Of note also is that the overall ECV for fluconazole for Cryptococcus would be in the range that most Candida species would be considered resistant (MIC > 4 mg/L). The same is true for the other azoles. So whilst an ECV of 32 mg/L, for example, may seem high compared with Candida ECVs or CBPs, it may be that these values are in the susceptible range for C. neoformans and C. gattii, especially given the high dose of fluconazole used to treat cyptococcosis [60].

Isavuconazole

Isavuconazole has good activity against C. neoformans and C gattii. In one study, the GM MIC for 90 C. gattii isolates was 0.057 mg/L with no significant increase observed for the molecular type VGII. In absolute terms, isavuconazole had about tenfold greater activity against C. gattii compared with fluconazole [74]. Another study focussing on C. gattii also determined that isavucoanzole had the lowest MIC90 (0.125 mg/L) for all isolates (GM MIC 0.051 mg/L) [15]. Both these studies employed CLSI M27-A3 methodology [17]. Thompson et al. [75] compared the isavuconazole Etest (A Biodisk, Solna, Sweden) with CLSI broth microdilution and found a 97.8% agreement between these methods, without major discrepancies (>2-well dilution difference). The GM MICs for C. gattii and C. neoformans were 0.03 and 0.024 mg/L, respectively by Etest and 0.027 and 0.023, respectively, by broth microdilution [75]. However, clinical correlation with these results is lacking.

Methods for Susceptibility Testing of Cryptococcus

Ideally, the CLSI and EUCAST reference BMD methodology is the preferred method for performing susceptibility testing [17, 18]. The good agreement by Etest for isavuconazole susceptibility testing is noted above [75]. Cuenca-Estrella et al. [28] compared the Vitek 2 Antifungal susceptibility system (AST-YS01; bioMerieux Inc., Hazelwood, MO) with the CLSI and EUCAST reference procedures to determine susceptibility to amphotericin B, 5-flucytosine, fluconazole and voriconazole for 16 C. neoformans isolates; the average EA was 92%, and the Vitek 2 system (bioMerieux) provided results more rapidly and was an easier alternative for clinical laboratories.

Mechanisms of Antifungal Drug Resistance

C. neoformans strains that manifest resistance to every class of antifungal drug have been described [69]. For 5-flucytosine, resistance occurs by mutations within the pyrimidine pathway specifically linked to a single mutation within the FCY1 or FYC2 gene [76]. Whether their resistance can be predicted by non-WT MIC values for 5-flucytosine in a given population is uncertain. However, the clinical utility of 5-flucytosine lies in its synergistic activity with amphotericin B so despite drug resistance, it may still be useful [60]. Resistance to amphotericin B may occur through defects in the target enzyme sterol delta 8→7 isomerase, other mutations in sterol synthesis genes or an increase in drug efflux [77]. Amphotericin B MICs of >2 mg/L have been associated with presence of more than one of these defects with clinical failure.

The molecular basis of azole resistance in C. neoformans is also not well studied. One resistance mechanism is through the duplication of chromosome 1 and consequently of two of its resident genes—ERG11 and AFR1 which encodes for an ABC transporter [78, 79] where up-regulation of AFR1 is involved in resistance to fluconazole. Unlike Candida species, however, only a limited number of ERG11 mutations have been reported to confer azole resistance; these include those leading to the amino acid substitutions G484S and Y145F [80, 81]. That leading to G484S was present in isolates with fluconazole MICs of 16–32 mg/L whilst the substitution Y145F conferred resistance to both fluconazole (MIC >256 mg/L) and voriconazole (≥2 mg/L), although interestingly increased susceptibility to posacoanzole and itraconazole. Of note, Gast et al. [82] examined the roles in azole resistance of ERG11 overexpression or mutations in the ERG11 gene in 25 C. gattii isolates with high MICs (GM MIC 20 mg/L) and found no correlation between either molecular-based results [82].

For all drug classes, much remains to be examined regarding the relationship between MICs and resistance mechanisms, in particular in determining the value of current ECVs for different bug-drug combinations in identifying isolates which harbour resistance mechanisms. For the present, MIC determination remains important for monitoring trends in susceptibility.

Conclusions

Determination of susceptibility or resistance to antifungal drugs remains important for Candida and Cryptococcus spp. In vitro MIC testing, particularly for Candida spp., is well standardised by CLSI and EUCAST methodologies. Overall, species-specific CBPs and ECVs for Candida spp.-, and species-, as well as molecular type-specific ECVs for Cryptococcus spp. are helpful in identifying isolates that are more/less likely to respond to an antifungal or are more/less likely to harbour a mechanism of drug resistance. However, for Candida spp. and the echinocandins, detection of FKS gene mutations linked to resistance, rather than MIC determination, predicts clinical outcomes. In the future, testing for molecular markers of resistance including the use of NGS technologies may become the preferred method. Further work on correlating MICs and ECVs with clinical outcomes and with underlying drug resistance mechanisms is required.