Use of Epidemiological Cutoff Values To Examine 9-Year Trends in Susceptibility of Candida Species to Anidulafungin, Caspofungin, and Micafungin

RESULTS AND DISCUSSION

Table 1 demonstrates the species distribution of Candida isolates from the bloodstream and normally sterile sites according to each year of the survey, 2001 to 2009. In general, the rank order remained the same from year to year: C. albicans > C. glabrata > C. parapsilosis > C. tropicalis > C. krusei. The only exceptions to this rank order came in the years 2003 and 2004, when the frequency of C. parapsilosis exceeded that of C. glabrata. Overall, there was no consistent trend indicating an increase or decrease in the frequency of any given species over time.

The ECVs for each echinocandin and the five Candida species studied are shown in Table 2. The ECVs for these five species were determined in a previous study of 8,023 isolates of Candida spp. representing the years 2003 to 2007 and were tested against all three echinocandins by use of the CLSI BMD method (29). We have demonstrated the ability of these echinocandin ECVs to discriminate WT strains (with MICs less than or equal to the ECV) from those with acquired resistance mechanisms (5, 28, 30). In an earlier study of 136 isolates of Candida (105 WT strains and 31 non-WT strains with fks mutations), the ECVs for anidulafungin, caspofungin, and micafungin identified 90% (anidulafungin) to 100% (caspofungin and micafungin) of the 31 mutant strains as non-WT strains (28). Furthermore, the ECVs correctly classified 74% (micafungin) to 96% (anidulafungin) of the 105 WT strains. In a subsequent study of C. glabrata (30), we also demonstrated that 6 of 12 non-WT strains for each echinocandin were fks mutants.

Although ECVs may be considered an early step in the development of CBPs (15, 16), the most important role for these cutoffs is to detect the emergence of reduced susceptibility to the agent of interest in the context of a resistance surveillance program. This is especially noteworthy in the case of the echinocandins and Candida, where the existing CBP for NS of >2 μg/ml is clearly not sensitive enough to detect the emergence of non-WT strains (Tables 3 and 4). The insensitivity of the CBP to detect the emergence of potential R to caspofungin is shown in Table 3, where both the CBP for NS and the ECV for each species are applied to the collection of Candida isolates spanning the 9-year period from 2001 through 2009. Application of the ECVs shows that the mean proportion of non-WT isolates per year was 0.1% for C. albicans, 1.3% for C. glabrata, 1.5% for C. parapsilosis, 0.7% for C. tropicalis, and 6.5% for C. krusei, whereas the CBP criteria for NS show only 0.0% to 0.3% of each species to be NS (Table 3). Notably, the application of the ECV for C. parapsilosis of 1 μg/ml documents a steady emergence of strains with decreased susceptibility to caspofungin from 2001 (0.5%) through 2009 (3.1%) (P = 0.04). No such trend was detected for any species by use of the CBP to detect the emergence of NS isolates.

A similar analysis is seen in Table 4, where the ECVs for anidulafungin and micafungin are applied to isolates from the years 2004 through 2009. Using the ECV for each echinocandin and species of Candida, the mean proportions of non-WT isolates per year for anidulafungin and micafungin, respectively, were as follows: 0.3% and 2.1% for C. albicans, 0.8% and 1.6% for C. glabrata, 0.0% and 0.5% for C. parapsilosis, 0.9% and 0.9% for C. tropicalis, and 0.5% and 3.5% for C. krusei (Table 4). Conversely, most years showed a complete lack of NS isolates detected by the CBP. Temporal trends of interest uncovered by the ECVs included increases in the proportion of non-WT isolates, from 0.4% (2004) to 1.8% (2009) for anidulafungin and C. glabrata (P = 0.09), from 2.4% (2004) to 5.7% (2009) for micafungin and C. krusei (P = 0.01), and from 0.0% (2004 to 2008) to 3.1% (2009) for micafungin and C. parapsilosis (P = 0.001). Although the CBPs for anidulafungin and micafungin identified more isolates of C. parapsilosis as NS than those detected by the ECVs, the CBPs for this species may need to be adjusted to avoid bisecting the WT population (1).

In light of the data presented in Tables 3 and 4, one may ask (i) what is the importance of detecting non-WT strains of Candida species for each echinocandin and (ii) given the slight differences in the percentage of non-WT strains for each species and echinocandin, what is the level of cross-resistance among the three echinocandins? Regarding the first question, application of the ECV criteria for each species and each echinocandin is simply a way to identify those strains that are phenotypically different from WT strains in their in vitro response to a given agent. The ECVs were generated earlier from a subset of the total collection described in the present study and were determined using the statistical methods of Turnidge et al. (35), which take into account the variation inherent in determining MICs due to the time of testing and reader variation. The non-WT strains detected by application of the ECVs to a population of MIC values represent strains that may have an acquired resistance mechanism and which may require additional testing to confirm the species identification, MIC value, and resistance mechanism. One of the main goals of the present work is to demonstrate that the use of ECVs provides data that may be “hidden” by CBPs. Such an approach thus highlights strains that should be characterized further. Although the non-WT strains detected in this study were not characterized further for the presence of fks mutations, we have demonstrated the sensitivity and specificity of these ECVs in distinguishing those isolates with a WT MIC for each echinocandin from non-WT strains containing fks mutations in previous studies (28, 30). Thus, the data herein should represent a reasonable approximation of the frequency of strains with decreased susceptibility to each of the echinocandins.

The issue of cross-resistance among the echinocandins has certainly been examined with respect to the molecular and biochemical characteristics of strains containing different fks mutations (11–13, 23, 24, 31, 34, 37). Whereas a given mutation may have a class effect in reducing the sensitivity of the modified enzyme to inhibition by drug, it does not necessarily manifest itself phenotypically with the same magnitude of change in various MIC values. Despite such differences, we demonstrated previously that ECVs do provide a phenotypic marker that indicates those strains most likely to contain an fks mutation and that they do so for each echinocandin, despite differences in the absolute values of the MICs (28, 30, 31). In assessing cross-resistance in the present study, we examined a subset of 3,400 MIC results where the isolates were tested against all three echinocandins (data not shown). We compared the levels of categorical agreement (CA) of the different agents by using the respective ECVs to separate the populations tested into WT and non-WT MIC groups. The overall CA between anidulafungin and caspofungin was 98.6% (0.8% very major discrepancies [VMD]), that between micafungin and caspofungin was 96.6% (0.7% VMD), and that between anidulafungin and micafungin was 98.0% (1% VMD) (data not shown). Thus, although there were some differences in the percentages of non-WT strains detected by the ECVs for the three echinocandins, the overall concordance was excellent and demonstrated a high degree of cross-resistance and -susceptibility across the three agents.

In summary, we have demonstrated the importance of ECVs in detecting the emergence of decreased susceptibility to the echinocandins in surveys of antifungal resistance. Whereas CBPs may serve the same purpose for C. parapsilosis, they appear to be too insensitive to be of epidemiological value in monitoring the more susceptible Candida species. Although uncommon, decreased S among Candida spp. was observed for each of the echinocandins and varied according to species over the 9-year study period. Application of in vitro susceptibility testing and the use of ECVs are important in determining R trends among the echinocandins and Candida spp. and will also serve as important steps in developing improved species-specific CBPs.