Multilaboratory Testing of Antifungal Drug Combinations against Candida Species and Aspergillus fumigatus: Utility of 100 Percent Inhibition as the Endpoint

Abstract

Four laboratories tested three isolates of Candida species and two isolates of Aspergillus fumigatus using 96-well plates containing combinations of amphotericin B, anidulafungin, caspofungin, micafungin, fluconazole, itraconazole, posaconazole, and voriconazole. The majority of summation fractional inhibitory concentration indices (ΣFICI) based on the Lowe additivity formula suggested indifferent drug interactions (ΣFICI > 0.5 and ≤4.0) and no instance of drug antagonism (ΣFICI > 4.0). The intra- and interlaboratory agreement rates were superior when MIC₁₀₀ readings were used as endpoints (at a 99% confidence interval [CI]).

TEXT

The incidence of systemic fungal infections continues to rise and to cause high morbidity and mortality (1, 2, 3). In many instances, hospitalized patients still receive empirical antifungal therapy, which prolongs hospitalization and increases mortality risk (4). Antifungal susceptibility test results are crucial for making informed clinical decisions (5). Currently, a number of standardized methods and devices are available for antifungal susceptibility testing in clinical laboratories (6–11). However, there is no standard method for laboratory testing of antifungal combinations, despite emerging evidence for the efficacy of combination therapy for fungal infection (12, 13). The present study details results of multilaboratory testing of two antifungal drug combinations with the ultimate goal of standardizing a method suitable for routine use in the clinical laboratories.

Multidrug-resistant fungal isolates of Candida albicans 20464.007 (AMB^R, FLC^R, ITC^R, VRC^R), Aspergillus fumigatus 20684.002 (AMB^R, AND^R, CSP^R, FLC^R, ITC^R, MCF^R) and 20684.007 (AMB^R, AND^R, CSP^R, FLC^R, ITC^R, MCF^R) were obtained from the Department of Pathology, University of Iowa. Candida glabrata M1409 (AMB^R, FLC^R, ITC^R, PSC^R, VRC^R) was selected from the Wadsworth Center Mycology Laboratory Culture Collection. The multidrug-resistant isolates were included because they are especially useful for evaluation of the efficacy of two-drug combinations due to their known resistance to at least one drug in the combination. Candida krusei ATCC 6258 was the quality control (QC) strain from participating sites. The 96-well plates in broth microdilution format were commercially prepared (Trek Diagnostics Systems, Cleveland, OH). One hundred microliters of amphotericin B (AMB; Sigma Chemical Co., St. Louis, MO), anidulafungin (AND; Pfizer, Inc., Groton, CT), caspofungin (CSP; Merck & Co., Rahway, NJ), micafungin (MCF; Astellas Pharma Inc., Tokyo, Japan), fluconazole (FLC; Sigma Chemical Co., St. Louis, MO), itraconazole (ITC; Sigma Chemical Co., St. Louis, MO), posaconazole (PSC; Schering-Plough Corp., Kenilworth, NJ), and voriconazole (VRC; Pfizer, Inc., Groton, CT) was dispensed singly or in pairwise combinations as detailed elsewhere (14, 15). One test set included 17 plates with 282 possible drug combinations. RPMI 1640 broth was used as a culture medium and for drug dilutions. The plates were shipped frozen by the manufacturer and were stored at −70°C. The test inoculum, setup, and reading followed CLSI guidelines M27-A3 for yeasts and M38-A2 for molds, with modifications (14, 15). All plates were read at 48 h for 50% and 100% growth inhibitions compared to growth control wells (MIC₅₀ and MIC₁₀₀). Drug combination interactions were determined by the summation of fractional inhibitory concentration indices (ΣFICI), which were calculated based on the Lowe additivity formula (16). The interpretations were as follows: ≤0.5, synergistic; >0.5 to ≤4.0, indifferent (no antagonism); >4.0, antagonistic (15, 17, 18). The interpretations of ΣFICI were categorized as A, B, and C, corresponding to synergy, indifference, and antagonism, respectively. If there was no ΣFICI value for a specific drug combination, there was no interpretation, which was categorized as D. Agreements and intra-/interclass correlation coefficients (ICCs) were used to examine the reliability of the results. Agreement was calculated as the percentage of the number of the most frequent categorized ΣFICI values divided by the number of all ΣFICI values. The agreement between all of the duplex experiment events for each laboratory was defined as the intralaboratory agreement rate. The agreement between four laboratories for each tested fungal isolate was defined as interlaboratory agreement rate. Intra- and interlaboratory ICCs (IaCC and IeCC, respectively) were calculated based on the ordered categorized ΣFICI values (19, 20).

The four participating laboratories collected about three thousand MIC readings for eight single drugs and 28 two-drug combinations. The single-drug MIC results for Candida species were in close agreement, while A. fumigatus isolates were reported to have a more diverse MIC range (data not shown). ΣFICI values were calculated antifungal drug combinations obtained from MIC₅₀ and MIC₁₀₀ readings (Table 1). The majority of the ΣFICI values were between 0.5 and 4.0, indicating indifference. A few ΣFICI values were greater than 4.0, which indicated antagonism. However, the antagonism was not seen consistently for any drug combination or isolate tested (Table 1). Mean ΣFICI₅₀ values were in the range of 0.06 to 3.63 and mean ΣFICI₁₀₀ values were in the range of 0.18 to 2.31 for all five isolates against all 28 drug combinations tested (Table 1). Interestingly, synergistic interactions of antifungals assessed by ΣFICI (≤0.5) at both 50% and 100% inhibition levels included CSP-PSC synergism against C. albicans 20464.007, AMB-PSC synergism against C. glabrata M1409, PSC-VRC synergism against A. fumigatus 20684.002, and AND-PSC synergism against A. fumigatus 20684.007 (see Table S1 in the supplemental material). Some synergistic drug interactions had a high percentage of agreement, and some had a low percentage of agreement.

TABLE 1.

ΣFICI values for 28 possible antifungal drug combinations obtained from MIC₅₀ and MIC₁₀₀ readings^a

^aShading indicates synergistic interactions (ΣFICI ≤ 0.5); boldface and underlining indicate antagonistic interactions. NA, not available; AMB, amphotericin B; AND, anidulafungin; CSP, caspofungin; FLC, fluconazole; ITC, itraconazole; MCF, micafungin; PSC, posaconazole; VRC, voriconazole.

The performance of each laboratory on two separate testing occasions was compared at different MIC reading levels against all 28 drug combinations using intralaboratory agreement rates and their correlation coefficients (IaCCs) (Table 2). IaCC was defined as the conformity between all the duplex events under the same experimental conditions. All four laboratories had over 80% intralaboratory agreement rates when MIC₁₀₀ was used as the endpoint (Table 2). A comparison of intralaboratory agreement rates between MIC₅₀ and MIC₁₀₀ readings showed significant difference (P < 0.05). Furthermore, IaCCs were greater than 0.9 for all the laboratories at the MIC₁₀₀ reading level, but one of the IaCCs at the MIC₅₀ reading level was as low as 0.677. The performance between different laboratories for the yeasts or molds tested was compared at both the MIC₅₀ and the MIC₁₀₀ reading levels against 28 drug combinations by interlaboratory agreement rates and their correlation coefficients (IeCCs) (Table 3). IeCC aimed to examine the conformity between different laboratories for the experiment results on the same isolate. Again, the interlaboratory agreement rates were higher at the MIC₁₀₀ reading level (86% ∼ 90%) than at the MIC₅₀ reading level (73% ∼ 76%). The IeCCs at the MIC₁₀₀ reading level were greater than 0.96 with a 99% confidence interval (CI) for both Candida species and A. fumigatus. The IeCCs were around 0.86 with 99% CI for different isolates at the MIC₅₀ reading level.

TABLE 2.

Intralaboratory agreement rates and correlation coefficients (IaCCs) for each participating laboratory

Laboratory	Value for MIC50/value for MIC100
	Agreement of ΣFICI interpretation (%)a,c	IaCCb
A	81/87	0.927/0.925
B	79/95	0.893/0.983
C	76/89	0.876/0.955
D	68/82	0.677/0.921

^aAgreement was calculated as the percentage of the number of the most frequent categorized ΣFICI values divided by the number of all ΣFICI values.

^bCalculated at a 99% confidence interval.

^cThe agreement rates at the 100% inhibition level were statistically significantly higher than those at the 50% inhibition level (P < 0.05).

TABLE 3.

Interlaboratory agreement rates and correlation coefficients (IeCCs) for each yeast or mold tested

Fungus	Value for MIC50/value for MIC100
	Agreement of ΣFICI interpretation (%)a,c	IeCCb
Candida species	76/86	0.868/0.962
Aspergillus fumigatus	73/90	0.861/0.978

^aAgreement was calculated as the percentage of the number of the most frequent categorized ΣFICI values divided by the number of all ΣFICI values.

^bCalculated at a 99% confidence interval.

^cThe agreement rates at the 100% inhibition level were statistically significantly higher than those at the 50% inhibition level (P < 0.05).

This study marks another milestone in our quest to define a standard method for antifungal combination testing in clinical laboratories (14, 15). The superior intra- and interlaboratory agreements obtained using MIC₁₀₀ endpoints were the highlight of this study. However, we also found that different drug combinations yielded variable interpretations for the same isolate, even when MIC₁₀₀ endpoints were used. These results affirm earlier suggestions that the FIC index model is subjective and sensitive to experimental error (17). The checkerboard dilution method used for the calculation of ΣFICI assumes that all antifungal drugs interact with each other in a linear model, providing an all-or-none view, thus artificially creating ΣFICI values (21, 22). Notably, fixed concentrations or fixed ratios of antimicrobial combinations were found to yield more useful results, especially for antibacterial combination testing (20, 23, 24). A single study found fixed ratios to be a better approach for antifungal combination testing (25). Thus, our future multilaboratory studies will assess fixed-concentration or fixed-ratio antifungal combination testing by the microbroth method using MIC₁₀₀ endpoints.