ABSTRACT
We tested 16,191 strains of Candida against posaconazole and voriconazole, using the CLSI M27-A3 broth microdilution (BMD) method (24-h incubation), in order to define wild-type (WT) populations and epidemiological cutoff values (ECVs). From 2001 to 2009, 8,619 isolates of Candida albicans, 2,415 isolates of C. glabrata, 2,278 isolates of C. parapsilosis, 1,895 isolates of C. tropicalis, 508 isolates of C. krusei, 205 isolates of C. lusitaniae, 177 isolates of C. guilliermondii, and 93 isolates of C. kefyr were obtained from over 100 centers worldwide. The modal MICs (μg/ml) for posaconazole and voriconazole, respectively, were as follows: for C. albicans, 0.016 and 0.007; for C. glabrata, 0.5 and 0.06; for C. parapsilosis, 0.06 and 0.007; for C. tropicalis, 0.03 and 0.015; for C. krusei, 0.25 and 0.12; for C. lusitaniae, 0.03 and 0.007; for C. guilliermondii, 0.12 and 0.03; and for C. kefyr, 0.06 and 0.007. The ECVs (μg/ml [% of isolates that had MICs equal to or less than the ECV]) for posaconazole and voriconazole, respectively, were as follows: 0.06 (98.5) and 0.03 (98.9) for C. albicans, 2 (96.2) and 0.5 (90.4%) for C. glabrata, 0.25 (99.3) and 0.12 (97.9) for C. parapsilosis, 0.12 (97.6) and 0.06 (97.2) for C. tropicalis, 0.5 (99.8) and 0.5 (99.4) for C. krusei, 0.12 (95.6) and 0.03 (96.6) for C. lusitaniae, 0.5 (98.9) and 0.25 (98.3) for C. guilliermondii, and 0.25 (100.0) and 0.015 (100.0) for C. kefyr. In the absence of clinical breakpoints (CBPs) for posaconazole, these WT distributions and ECVs will be useful in surveillance for emergence of reduced susceptibility to posaconazole among Candida spp. Whereas a CBP for susceptibility of ≤1 μg/ml has been established for voriconazole and all species of Candida, it is notable that ECVs for this agent range from 10- to >100-fold lower than the CBP, depending on the species of Candida. The CBP is inadequate in detecting the emergence of voriconazole resistance among most Candida species encountered clinically. The CBPs for voriconazole should be reassessed, with consideration for development of species-specific CBPs.
INTRODUCTION
As experience with the Clinical and Laboratory Standards Institute (CLSI) broth microdilution (BMD) method for in vitro susceptibility testing of Candida versus the azoles has grown, it has become apparent that the majority of Candida spp. achieve suitable growth for MIC testing by 24 h of incubation (14–16, 36). Given the facts that a shorter incubation period is more efficient and practical for use in the clinical laboratory (16, 36) and that research has shown that early/correct antifungal treatment has a positive effect on outcomes for patients with candidemia (5, 6, 12, 19, 20, 23, 27, 29, 33, 53), the CLSI has determined that fluconazole MICs may be used after a 24-h incubation, providing earlier clinically useful information to clinicians caring for patients with invasive candidiasis (IC) (31, 36). Currently, the CLSI recommends that MICs for the newer triazoles, posaconazole and voriconazole, be read after 48 h of incubation for testing of Candida spp. (9).
Voriconazole and posaconazole are extended-spectrum triazole antifungal agents with potent activity against Candida spp., Cryptococcus spp., dimorphic fungi, and filamentous fungi (7, 30). Voriconazole was introduced in 2001 and is approved for the treatment of mucosal and systemic candidiasis (candidemia and other forms of IC), invasive aspergillosis, and refractory fungal infections caused by Scedosporium and Fusarium spp. (2, 22, 26, 34, 51). Posaconazole was approved in 2006 for prophylaxis of invasive fungal infection (IFI) in hematopoietic stem cell transplant (HSCT) recipients and patients with hematologic malignancies (11). It is also approved for the treatment of oropharyngeal candidiasis (49, 50). In Europe, posaconazole is also approved for refractory IFIs, including aspergillosis, fusariosis, chromoblastomycosis, mycetoma, and coccidioidomycosis (30, 40, 44, 45, 48, 52).
The CLSI Subcommittee for Antifungal Testing established clinical breakpoints (CBPs) for voriconazole and Candida by taking into account the MIC distributions, pharmacokinetic (PK) and pharmacodynamic (PD) parameters, and clinical outcomes as they relate to the MIC values (9, 10, 35). This approach did not allow for species-specific breakpoints and used the “90-60 rule” (41) to arrive at the following CBPs: susceptible (S), MICs of ≤1 μg/ml; susceptible but dose dependent (SDD), MIC = 2 μg/ml; and resistant (R), MICs of ≥4 μg/ml. These CBPs require 48 h of incubation and are to be applied to all species of Candida. Subsequently, Arendrup and Denning (3) raised the question of whether a non-species-dependent CBP for voriconazole susceptibility of ≤1 μg/ml is appropriate given the fact that the vast majority of cases of IC caused by isolates defined as S were cases of Candida albicans, C. tropicalis, and C. parapsilosis infections, with MICs that were 4 to 5 log below the suggested CBP (35). These authors recommended the use of microbiological cutoff values for C. albicans and other species for which the voriconazole MICs for wild-type (WT) isolates determined by CLSI BMD are very low (3). The European Committee for Antimicrobial Susceptibility Testing (EUCAST) Subcommittee on Antifungal Susceptibility Testing analyzed the 24-h MIC values for WT Candida spp. and developed both epidemiological cutoff values (ECVs or ECOFFs) and CBPs for voriconazole and selected species of Candida (18, 42): the CBPs for S and R are MICs of ≤0.125 and >0.125 μg/ml, respectively, to be applied to C. albicans, C. tropicalis, and C. parapsilosis. Although ECVs were assigned to both C. glabrata and C. krusei, there was insufficient evidence for any correlation between MIC and clinical outcome for these species. In the EUCAST approach, the WT MIC distribution for a species is defined as the MIC distribution for isolates that exhibit no acquired or mutational resistance to the drug in question, whereas non-WT isolates may possess acquired or mutational resistance mechanisms (4, 24, 25, 46, 47). The upper limit for the WT population is defined as the ECV. In general, the ECV encompasses at least 95% of isolates in the WT distribution (46, 47).
Although both agar-based and BMD antifungal susceptibility testing methods have been validated for testing posaconazole against Candida (13–15), CBPs have not yet been developed for this agent, resulting in the use of either fluconazole or voriconazole as a surrogate marker to predict the susceptibility of Candida spp. to posaconazole (37). These findings suggest that in addition to a prolonged (48-h) incubation time, the CBPs for voriconazole may lack sensitivity in detecting the emergence of resistance to voriconazole and posaconazole among the more susceptible species, such as C. albicans, C. tropicalis, and C. parapsilosis (3). For these reasons, and in the absence of CBPs for posaconazole, we considered that the determination of 24- and 48-h WT MIC distributions and ECVs would be useful in surveillance for emergence of reduced susceptibility for both of these agents among Candida spp. Furthermore, this may be considered a necessary first step toward the development of useful, species-specific 24-h CBPs (17, 18, 38, 39, 42).
MATERIALS AND METHODS
Organisms.A total of 16,191 clinical isolates obtained from more than 100 medical centers worldwide from 2001 through 2009 were tested. The collection included 8,619 isolates of C. albicans, 2,415 isolates of C. glabrata, 2,279 isolates of C. parapsilosis, 1,895 isolates of C. tropicalis, 508 isolates of C. krusei, 205 isolates of C. lusitaniae, 177 isolates of C. guilliermondii, and 93 isolates of C. kefyr (Table 1). All isolates were obtained from blood or other normally sterile sites and represented the incident isolates from individual infectious episodes. The isolates were collected at individual study sites and were sent to the University of Iowa (Iowa City, IA) for identification and susceptibility testing as described previously (35, 36, 38). The isolates were identified by standard methods (21) and stored as water suspensions until used in the study. Prior to being tested, each isolate was passaged at least twice onto potato dextrose agar (Remel) and CHROMagar Candida medium (Becton Dickinson and Company, Sparks, MD) to ensure purity and viability.
WT MIC distributions of posaconazole and voriconazole for eight species of Candida obtained using CLSI BMD methods
Antifungal agents.Reference powders of posaconazole and voriconazole were obtained from their respective manufacturers. Stock solutions were prepared in dimethyl sulfoxide, and serial 2-fold dilutions were made in RPMI 1640 medium (Sigma, St. Louis, MO) buffered to pH 7.0 with 0.165 M MOPS (morpholinepropanesulfonic acid) buffer (Sigma).
Antifungal susceptibility testing.BMD testing was performed in accordance with the guidelines in CLSI document M27-A3 (9), using RPMI 1640 medium, an inoculum of 0.5 × 103 to 2.5 × 103 cells/ml, and incubation at 35°C. MICs were determined visually, after 24 and 48 h of incubation, as the lowest concentrations of drug that caused a significant diminution (≥50% inhibition) of growth relative to that of the growth control (9, 35, 36).
Quality control.Quality control was performed by testing CLSI-recommended strains C. krusei ATCC 6258 and C. parapsilosis ATCC 22019 (9, 10).
Definitions.The definitions of WT organisms and ECVs were those outlined previously (24, 38, 39, 46, 47). A WT organism is defined as a strain which does not harbor any acquired resistance to the particular antimicrobial agent being examined. The typical MIC distribution for WT organisms covers three to five 2-fold dilution steps surrounding the modal MIC (4, 25). Inclusion of WT strains in the present study was ensured by testing only the incident isolate for each infectious episode.
The ECVs for posaconazole and voriconazole for each species of Candida were obtained as described by EUCAST (24), by considering the WT MIC distribution, the modal MIC for each distribution, and the inherent variability of the test (usually within 1 log2 dilution). In general, the ECV should encompass at least 95% of isolates in the WT distribution (39, 46, 47). Organisms with acquired resistance mechanisms may be included among those for which the MICs are higher than the ECVs (4, 24, 25, 38, 39).
The voriconazole CBPs for S (MICs of ≤1 μg/ml), SDD (MIC = 2 μg/ml), and R (MICs of ≥4 μg/ml) used in this study were those defined by Pfaller et al. (35) and the CLSI (10).
RESULTS AND DISCUSSION
Wild-type MIC distributions and ECVs.The WT MIC distributions for posaconazole and voriconazole and each of eight species of Candida are shown in Table 1. These distributions clearly show the very low MICs typical of WT strains of C. albicans, C. tropicalis, C. parapsilosis, C. lusitaniae, and C. kefyr and the higher MICs typical of C. glabrata, C. krusei, and C. guilliermondii for both of the triazoles.
The modal MICs (percentages of isolates with a MIC equal to the mode) at 24/48 h of incubation for posaconazole and voriconazole, respectively, were as follows (Table 2): for C. albicans, 0.015/0.015 μg/ml (47.5%/43.4%) and 0.007/0.007 μg/ml (94.0%/85.1%); for C. glabrata, 0.5/1 μg/ml (36.2%/35.4%) and 0.06/0.25 μg/ml (42.1%/30.6%); for C. parapsilosis, 0.06/0.06 μg/ml (35.8%/40.9%) and 0.007/0.015 μg/ml (55.2%/42.7%); for C. tropicalis, 0.03/0.06 μg/ml (34.7%/35.7%) and 0.015/0.03 μg/ml (36.0%/35.5%); for C. krusei, 0.25/0.25 μg/ml (50.8%/47.8%) and 0.12/0.25 μg/ml (55.6%/46.7%); for C. lusitaniae, 0.03/0.06 μg/ml (35.1%/38.0%) and 0.007/0.007 μg/ml (81.0%/72.7%); for C. guilliermondii, 0.12/0.25 μg/ml (42.4%/43.5%) and 0.03/0.06 μg/ml (40.1%/48.0%); and for C. kefyr, 0.06/0.12 μg/ml (31.2%/32.3%) and 0.007/0.007 μg/ml (82.8%/69.9%). The MIC distributions in this study were all determined in a single reference laboratory (University of Iowa) by CLSI-recommended BMD methods and thus may be less broad, with lower modal MICs, than distributions generated by multiple laboratories. This is recognized as a potential limitation in the assignment of the ECVs. These concerns may be mitigated by the fact that the data were generated over a 9-year period and employed multiple lots of BMD trays and antifungal agents as well as multiple readers of the MICs.
ECVs for posaconazole and voriconazole and eight species of Candida
The 24/48-h ECVs (percentages of isolates with MICs that were less than or equal to the ECVs) were determined as described previously (38, 39) and were as follows for posaconazole and voriconazole, respectively (Table 2): 0.06/0.06 μg/ml (98.4%/97.8%) and 0.03/0.03 μg/ml (99.0%/98.4%) for C. albicans, 2/4 μg/ml (96.1%/94.9%) and 0.5/1 μg/ml (90.4%/91.1%) for C. glabrata, 0.25/0.25 μg/ml (99.3%/98.6%) and 0.12/0.12 μg/ml (97.8%/94.9%) for C. parapsilosis, 0.12/0.25 μg/ml (97.8%/99.2%) and 0.06/0.12 μg/ml (97.3%/98.0%) for C. tropicalis, 0.5/1 μg/ml (99.0%/99.4%) and 0.5/1 μg/ml (99.4%/99.6%) for C. krusei, 0.12/0.25 μg/ml (95.6%/96.6%) and 0.03/0.06 μg/ml (96.6%/95.6%) for C. lusitaniae, 0.5/0.5 μg/ml (98.9%/96.0%) and 0.25/0.25 μg/ml (98.3%/97.2%) for C. guilliermondii, and 0.25/0.5 μg/ml (100.0%/100.0%) and 0.015/0.03 μg/ml (100.0%/97.8%) for C. kefyr.
Compared to the voriconazole CBP value for S of ≤1 μg/ml (surrogate CBP for posaconazole), the ECVs are between 4- and 33-fold lower for the two triazoles and C. albicans, C. tropicalis, C. parapsilosis, C. lusitaniae, and C. kefyr. Whereas the CBP encompasses 99.9% to 100.0% of the isolates of these five species (data not shown), the ECVs of each agent encompass 94% to 100.0% of the isolates (Table 2), highlighting the small number of isolates of each species that fall outside the WT distribution yet remain susceptible to each agent according to the CBP. In contrast, the ECVs for the three less-susceptible species, C. glabrata, C. krusei, and C. guilliermondii, are similar to or higher than the CBPs for voriconazole.
As noted previously (38, 39), CBPs are used to indicate those isolates that are likely to respond to treatment with a given antimicrobial agent administered using the approved dosing regimen for that agent, whereas the ECV can be used as the most sensitive measure of the emergence of strains with decreased susceptibility to a given agent. Although organisms whose MICs exceed the ECV show reduced susceptibility compared with the WT population and may exhibit one or more acquired resistance mechanisms, they may yet respond to clinical treatment, as their MIC may lie below the CBP (35, 43, 47).
Comparison of WT distributions and ECVs for voriconazole determined by CLSI and EUCAST BMD methods.The EUCAST Subcommittee on Antifungal Susceptibility Testing determined the voriconazole MICs (determined with 24-h incubation) for 25,284 isolates of Candida spp. and published ECVs of 0.12 μg/ml for C. albicans, C. parapsilosis, and C. tropicalis and 1 μg/ml for C. glabrata and C. krusei (Table 3) (18). Both the WT MIC distributions and the ECVs determined by EUCAST are very similar to those determined by the CLSI BMD method read after 24 h of incubation, confirming the comparability of the two methods for voriconazole susceptibility testing (8, 14). Further harmonization of these two approaches, including the establishment of comparable CBPs, should be possible.
ECVs for voriconazole and five species of Candida obtained using the 24-h CLSI and EUCAST BMD methodsb
Application of ECVs in resistance surveillance.Arguably, the most important role of in vitro susceptibility testing is in detecting resistance, that is, determining which agents will not work (3, 41, 47). Thus, it would be desirable to use the most sensitive means available to detect emerging resistance. The existing CLSI CBP for resistance to voriconazole of ≥4 μg/ml is clearly insensitive in detecting isolates of C. albicans, C. tropicalis, and C. parapsilosis with decreased susceptibility compared to WT strains (Table 2). The insensitivity of the CBP to detect emergence of potential resistance to voriconazole is demonstrated further in Table 4, where both the CBP for voriconazole resistance (MICs of ≥4 μg/ml) and the ECV for each species are applied to a collection of Candida isolates spanning a 9-year period, from 2001 through 2009. Application of the ECVs shows that the mean proportion of non-WT isolates per year was 0.9% for C. albicans, 9.7% for C. glabrata, 2.3% for C. parapsilosis, 2.8% for C. tropicalis, and 0.5% for C. krusei, whereas the CBP for R (MICs of ≥4 μg/ml) shows only 0.0 to 3.4% of each species to be resistant. Notably, the application of the ECV for C. parapsilosis of 0.12 μg/ml documents a steady emergence of strains with decreased susceptibility to voriconazole from 2001 through 2009, whereas no such trend is evident using the CBP for R (Table 4). Although application of the CBP to the collection of C. glabrata isolates shows a trend toward increasing resistance over time, the lower ECV of 0.5 μg/ml suggests that the rate of decreased susceptibility may be 2- to 4-fold higher than that indicated by the CBP. The latter observation is consistent with several case series where C. glabrata isolates from cases breaking through voriconazole treatment were shown to have voriconazole MICs of 1 to 4 μg/ml (1, 28, 32).
Variation in susceptibility of Candida species to voriconazole over a 9-year period (2001 to 2009)a
Although CBPs are not available for posaconazole, application of the ECVs to the different species of Candida tested over a 9-year period demonstrated the emergence of decreased susceptibility among isolates of C. albicans and C. glabrata (Table 5). Using the ECVs for posaconazole and each species of Candida, the mean proportion of non-WT isolates per year was 1.8% for C. albicans, 3.9% for C. glabrata, 0.6% for C. parapsilosis, 2.2% for C. tropicalis, and 1.1% for C. krusei.
Variation in susceptibility of Candida species to posaconazole over a 9-year period (2001 to 2009)a
Conclusions.In this study, we established both 24- and 48-h ECVs for posaconazole and voriconazole, using an extensive geographically diverse collection of Candida species. Importantly, we have shown little difference between the 24- and 48-h values for eight different species of Candida, suggesting that the routine determination of MICs after only 24 h of incubation is feasible for these agents. Furthermore, we have shown that 24-h ECVs for voriconazole determined by the CLSI BMD method are essentially the same as those determined using the EUCAST method, setting the stage for further harmonization of these two reference standards.
The ECVs determined for C. albicans, C. glabrata, C. tropicalis, C. parapsilosis, C. krusei, C. guilliermondii, C. lusitaniae, and C. kefyr will be important in detecting the emergence of decreased susceptibility to voriconazole in ongoing surveillance efforts. The previous CBPs for voriconazole established by the CLSI appear to be too insensitive to be of epidemiological value in monitoring the emergence of decreased susceptibility to this agent, especially among the more susceptible species (e.g., C. albicans, C. tropicalis, and C. parapsilosis). Given the absence of CBPs for posaconazole, the ECVs established herein will prove useful in detecting the emergence of potential resistance as this agent is employed as prophylaxis in high-risk patient populations.
Future studies must include molecular analysis of resistance mechanisms for the strains that fall outside the ECV to better understand the frequency and clinical importance of such strains and mechanisms. The establishment of the WT MIC distributions and ECVs for posaconazole and voriconazole and each species of Candida will be useful in resistance surveillance and may prove to be an important step in the development of species-specific 24-h CBPs for these important antifungal agents.
ACKNOWLEDGMENTS
Caitlin Howard provided excellent support in the preparation of the manuscript.
This work was supported in part by grants from Pfizer and Schering-Plough.
FOOTNOTES
- Received 25 October 2010.
- Returned for modification 1 December 2010.
- Accepted 8 December 2010.
- Accepted manuscript posted online 15 December 2010.
- Copyright © 2011, American Society for Microbiology. All Rights Reserved.