INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Data collected from time-kill studies have provided critical information regarding the rate and extent of bactericidal activity, pharmacodynamic characteristics (i.e., relationship between concentration and effect and the postantibiotic effect), and potential antagonism or synergy between antibacterial agents administered concomitantly. These data have significantly enhanced our understanding regarding the dynamic relationships which exist between antimicrobial agents and their effects on bacteria. In fact, time-kill testing has become an indispensable tool for assessing the activity of antimicrobials against bacteria. Standardized methods providing instruction on the implementation of time-kill methods have been proposed by the National Committee for Clinical Laboratory Standards to ensure the reproducibility and accuracy of test results (11).

Despite widespread recognition of the value of data generated with time-kill studies against bacteria, similar data for fungi are virtually nonexistent. This, however, is not entirely surprising if one considers that it was only recently that guidelines for conducting and interpreting antifungal susceptibility test results were established and approved (13, 16). With the framework for in vitro testing of antifungal agents in place, it is only a matter of time before increased interest in antifungal time-kill testing is generated. Guidelines regarding procedures for antifungal time-kill testing have yet to be established. Significant interlaboratory variability, nonreproducible results, conflicting data, and unstandardized interpretation of results are only a few of the potential pitfalls that may be encountered if time-kill methods are not standardized. Similar problems were encountered and slowed the development and acceptance of in vitro antifungal susceptibility testing procedures.

There are currently relatively few published reports of antifungal time-kill studies. Our group has conducted several studies utilizing time-kill procedures for the study of antifungal activity (1-10, 17, 18). The methods used to conducted these studies were developed in our laboratory. During the course of these studies, we evaluated several variables for their effect on time-kill curve results. This report details our findings regarding the influence of various test conditions on antifungal time-kill results.

(A preliminary report of this work has been presented previously [7].)

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Antifungal agents. Fluconazole (Pfizer Inc., New York, N.Y.), amphotericin B (Sigma Chemical Company, St. Louis, Mo.), and LY303366 (Eli Lilly and Co., Indianapolis, Ind.) were utilized for susceptibility determinations and time-kill studies. Stock solutions of each agent were prepared utilizing RPMI 1640 (Sigma) buffered to a pH of 7.0 with 0.165 M morpholinepropanesulfonic acid (MOPS) buffer (Sigma) as the solvent. Dimethyl sulfoxide (DMSO) was used to aid the solubilization of each of the drugs. The final concentration of DMSO in the time-kill test solutions was 1% (vol/vol) of the solution composition. To establish that exposure to DMSO did not affect the growth of the test isolates, fungi were grown in the presence of 1% (vol/vol) DMSO and compared with growth of fungi naive to DMSO. Stock solution were separated into unit-of-use aliquots and stored at 70°C until used.

Test isolates. Two clinical isolates of Candida glabrata (strains 582 and 350) and of Candida tropicalis (strains 2697 and 3829) were selected for testing. Additionally, one American Type Culture Collection strain (ATCC 90028) and one clinical isolate (OY31.5) of Candida albicans were utilized. Isolates were obtained from the Department of Pathology, The University of Iowa College of Medicine.

Antifungal susceptibility testing. The MIC of each antifungal was determined against test isolates by using broth microdilution techniques as described by the National Committee for Clinical Laboratory Standards (13). MICs were determined in RPMI 1640 buffered to a pH of 7.0 with MOPS. The starting inoculum was approximately 1 × 10³ to 5 × 10³ CFU/ml. Microtiter trays were incubated at 35°C in a moist, dark chamber, and MICs were recorded after 48 h of incubation. The susceptibility endpoints for fluconazole and LY303366 were defined as the lowest concentration of antifungal which resulted in an 80% reduction in visual growth compared with growth of the control (7, 13). In contrast, the MIC of amphotericin B was defined as the lowest concentration of drug which resulted in total inhibition of visual growth.

Limit of quantitation. The lower limit of accurately detectable CFU/milliliter or the limit of quantitation was determined for each of the six isolates. A fungal suspension was made in sterile water with each isolate and adjusted to a 0.5 McFarland turbidity standard (approximately 1 × 10⁶ to 5 × 10⁶ CFU/ml). A series of dilutions, using sterile water, were made with the standardized suspensions, resulting in three suspensions with fungal concentrations of approximately 100, 50, and 30 CFU/ml for each isolate. Thirty microliters was removed from each suspension and plated on potato dextrose agar (PDA) plates (Remel, Lexena, Kans.) for colony count determination. Plates were incubated at 35°C, and viable-colony counts were determined after 24 to 48 h. Experiments were conducted in quintuplicate.

Antifungal carryover. Antifungal carryover determinations were conducted as previously described (9). Briefly, fungi were obtained from stored samples and subcultured twice on PDA plates (Remel) prior to testing. Fungal suspensions were prepared in sterile water by touching one or two colonies from a 24- to 48-h-old culture plate and adjusting the resulting suspension to a 0.5 McFarland turbidity standard (approximately 1 × 10⁶ to 5 × 10⁶ CFU/ml) by spectrophotometric methods. The resulting suspension was then diluted via sequential dilutions of 1:100 and 1:2 with sterile water to yield a fungal suspension of approximately 5 × 10³ CFU/ml. One hundred microliters of the diluted fungal suspension was then added to 900 µl of sterile water or sterile water containing fluconazole, amphotericin B, or LY303366, resulting in a starting inoculum of approximately 5 × 10² CFU/ml. Antifungal carryover was evaluated over a range of antifungal concentrations from 1× to 16× MIC. Immediately following the addition of the fungal suspension to the aqueous solutions, an aliquot was removed from each tube and streaked across PDA plates (Remel) or RPMI 1640 agar plates (135-mm diameter; Remel) (RPMI agar was used only for the plating of 100-µl samples) for colony count determinations. Because of problems in the acquisition of 135-mm-diameter PDA plates, RPMI agar was used. Three methods were initially evaluated for colony count determinations: direct plating of 10, 30, or 100 µl (amphotericin B and LY303366 only) of test solutions. If antifungal carryover was noted with any of these methods, the study was repeated adding a fourth plating method: dilution of 30 µl of the test sample in 10 ml of sterile water followed by vacuum filtration through a 0.45-µm-pore-size filter and subsequent placement of the filter onto a PDA plate. Colony counts were determined following incubation at 35°C for 24 to 48 h. Tests were conducted in quintuplicate. Reproducibility of results were evaluated by determining the coefficients of variation associated with results obtained with control samples by each of the sampling methods.

Agitation. Time-kill procedures were conducted as previously described (5, 9). Fungal suspensions, adjusted to a 0.5 McFarland turbidity standard, were prepared as described above. A 1:10 dilution of this suspension was made by adding 1 ml of fungal suspension to 9 ml of RPMI 1640 with or without (control) the desired amount of antifungal. This dilution yielded a starting inoculum of approximately 1 × 10⁵ to 5 × 10⁵ CFU/ml. Antifungals were tested at a concentration equal to 2× MIC for test isolates. Two identical sets of solutions were prepared for each isolate: control (drug-free) (tube 1), fluconazole (tube 2), amphotericin B (tube 3), and LY303366 (tube 4). All solutions were incubated at 35°C; however, one set of tubes was placed on an orbital shaker and incubated with agitation, whereas the other set of tubes was incubated without agitation. At predetermined time points (0, 1, 2, 3, 4, 6, 8, 12, and 24 h), a 100-µl sample was removed from each tube and serially diluted 10-fold with sterile water, and a 30-µl aliquot was plated onto a PDA plate for colony count determination. When colony counts were expected to be less than 1,000 CFU/ml, a 30-µl sample was taken directly from the test solution and plated onto a PDA plate without dilution. Plates were then incubated for 24 to 48 h at 35°C prior to examination. All experiments were conducted in duplicate.

Inoculum. Time-kill methods as detailed above were utilized with the following modifications. Three ranges of starting inocula for each isolate were studied against fluconazole, amphotericin B, and LY303366: 5 × 10² to 1 × 10⁴, >1 × 10⁴ to 1 × 10⁶, and >1 × 10⁶ to 1 × 10⁸ CFU/ml. Antifungals were tested at concentrations equal to 2× MIC for each isolate. Test samples were incubated at 35°C with agitation. Aliquots were removed from each test solution for colony count determination at 0, 1, 2, 3, 4, 5, 6, 7, 8, 12, and 24 h following inoculation. The plating procedures described above were followed for high- and medium-inoculum samples; however, the limit of fungal quantitation was lowered to approximately 30 CFU/ml for each isolate in the low-inoculum group. This was accomplished by plating 100 µl of the test sample directly onto an RPMI 1640 agar plate without dilution. Plates were incubated at 35°C for 24 to 48 h prior to determination of colony counts. All experiments were conducted in duplicate.

Analysis. For the quantitation limit and antifungal carryover studies, intraspecies results were combined. Antifungal carryover colony count results for all three drugs, at each multiple of the MIC, were compared to the control value for each of the sampling methods. Significant antifungal carryover was defined as >25% reduction in CFU/milliliter compared to the control value (14).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Antifungal susceptibility. Susceptibility data for each isolate are presented in Table 1. C. glabrata 350 and C. tropicalis 2697 both exhibit resistance to fluconazole, with MICs of >128 µg/ml (16).

Limit of quantitation. Using a 30-µl sampling volume, the lower limit of accurate and reproducible quantitation was 50 CFU/ml for each of the isolates. According to these sampling methods, the cumulative (all species) percent coefficient of variation (% CV) of colony counts resulting from control samples was 22.7. Cumulative % CVs calculated for the 10-µl, 100-µl, and filtered sampling methods were 37.2, 16.6, and 24.4, respectively. Sampling variability appeared to be slightly greater with C. tropicalis (% CV for 30-µl sampling = 34.2) compared with the other species (% CV = 17). The presence of antifungal in solution did not affect the reproducibility of sampling results.

Antifungal carryover. Antifungal carryover data are summarized in Table 2. No antifungal carryover was observed with LY303366 against test isolates with any of the three sampling volumes. For fluconazole, significant carryover was observed against C. tropicalis only. The effect began at concentrations of fluconazole equal to 16× and 8× MIC for the 10- and 30-µl sampling methods, respectively. Direct plating of 100 µl was not evaluated with fluconazole. In contrast, significant carryover was noted for amphotericin B against all of the species tested. At sampling volumes of 10 and 30 µl, carryover was observed only against C. albicans and only at concentrations equal to 16× and 8× MIC, respectively. Carryover was noted among all three species at a sampling volume of 100 µl, even with concentrations as low as 2× MIC (C. glabrata). Filtration was effective in eliminating the carryover noted with amphotericin B. 

Agitation. The rate and extent of growth for each of the control samples appeared to be independent of agitation (Fig. 1). Likewise, agitation or the lack thereof did not affect the activity observed with any of the test agents. The difference between agitated and respective nonagitated samples did not exceed 0.7 log₁₀ CFU/ml at any of the time points. Additionally, by the end of the 24-h study period, differences between samples were generally <0.3 log₁₀ CFU/ml. Only for C. albicans OY31.5 was a difference of >0.3 log₁₀ CFU/ml observed at 24 h. For this isolate, a difference of 0.7 log₁₀ CFU/ml was detected between shaken and nonshaken samples of fluconazole. Against this isolate, slightly fewer CFU/milliliter were noted with the agitated sample than with the nonshaken sample.

View larger version (19K): [in this window] [in a new window]   FIG. 1.   Time-kill plot of the effect of agitation on C. tropicalis 3829.

Inoculum. The maximum change in log₁₀ CFU/milliliter observed over the study period for each isolate is presented in Table 3. The control growth curves for the low and medium starting inocula paralleled each other until the 12-h time point for each of the isolates (Fig. 2). By 24 h, the curves had converged to a common maximal fungal concentration. Control curves for the highest inocula exhibited minimal increases in CFU/milliliter, generally less than 1 log₁₀ unit. Rather, colony counts remained relatively constant at levels near the concentration of maximal sustainable growth throughout the study period. The level of maximal sustainable growth varied among the test isolates and ranged from approximately 10⁶ CFU/ml for C. albicans 90028 to 10⁸ CFU/ml for C. glabrata 582. 

View larger version (17K): [in this window] [in a new window]   View larger version (16K): [in this window] [in a new window]   FIG. 2.   Time-kill plots demonstrating the effects of starting inoculum on the activities of fluconazole (A), amphotericin B (B), and LY303366 (C) against C. albicans 90028.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

As clinical interest in fungi and antifungal therapies continue to grow, there is a pressing need to enhance our understanding of the fungicidal properties and pharmacodynamic characteristics of these agents. Despite the relative lack of data describing the use of time-kill methods for the study of fungi, we have found these techniques to be valuable in examination of antifungal dynamics. The methodology we employ for the study of antifungal agents is based upon adaptation of the procedures proposed for the time-kill evaluation of antibacterial agents (11). However, several methodological modifications to these procedures were required to facilitate the study of fungi and were based primarily upon the established guidelines for in vitro susceptibility testing of antifungals (13, 16). As a result, selection of test variables such as choice of growth medium and incubation temperature are common to both procedures. However, several test conditions specific to time-kill testing were identified as having the potential of significantly affecting test results. Therefore, the impact of variables, such as antifungal carryover, agitation, and starting inoculum, on test results are highlighted in this report.

For bacteria, the starting inoculum used for time-kill tests is similar to that recommended for in vitro susceptibility determinations, approximately 5 × 10⁵ CFU/ml (12). Use of the same starting inoculum facilitates comparisons between MIC and time-kill data by precluding potential discrepancies resulting from an inoculum effect. However, a starting inoculum of 5 × 10² to 2.5 × 10³ CFU/ml has been recommended for in vitro susceptibility testing of antifungals. Since the starting inoculum recommended for in vitro antifungal susceptibility testing is much lower than the inoculum recommendations for antibacterial time-kill testing, we felt that the following three questions needed to be answered. (i) What is the limit of quantitation for fungi? (ii) Does antifungal carryover occur with the test agents we have selected for evaluation and if so to what degree? (iii) Does an inoculum effect exist with fungi which would preclude comparisons between time-kill results and MICs if a higher starting inoculum were selected for time-kill studies?

We quickly realized that use of a starting inoculum, similar to that used for antifungal susceptibility determinations, was not viable for time-kill testing. Because of limitations imposed by the limit of quantitation, the use of a low starting inoculum affected our ability to detect a 99.9% reduction in CFU/milliliter with fungicidal agents. Additionally, with amphotericin B, if methods to lower the limit of quantitation are employed, antifungal carryover would be a concern unless samples were filtered.

Since we deemed the use of a low starting inoculum to be unacceptable, we evaluated the possibility of using two higher starting inocula. Using a starting inoculum between 10⁴ and 10⁶ CFU/ml, we were better able to characterize the kill curves of fungicidal agents. At the highest inoculum, however, we noted that the level of maximal sustainable growth, the plateau of the growth curve, was approximately 1 log₁₀ CFU/ml higher than the starting inoculum. As a result, detection of fungistatic activity with fluconazole was hindered because of the minimal separation observed between control and fluconazole curves. As a result of the limitations encountered with the highest and lowest inocula regarding the characterization of fungistatic and fungicidal activities, respectively, we recommend that a starting inoculum of approximately 10⁵ CFU/ml be employed for antifungal time-kill studies. Additionally, since the activity observed did not differ between the low and medium test inocula, MIC and time-kill results should be reflective of each other. In previous studies, we have been able to correlate MICs with observed time-kill activities of several agents including fluconazole, amphotericin B, and LY303366 (4, 7-9, 15).

In this study, we also attempted to examine the impact of sample agitation on time-kill results. Although we did not observe an appreciable difference among test samples, this may have been an artifact created by our sampling methods. Prior to the removal of each sample for colony count determination, all time-kill tubes were vortexed. In this study, colony count samples were obtained at 2, 4, 6, 8, 12, and 24 h. As a result, vortexing may have had a greater impact on results than did sample agitation. Three isolates were subsequently selected for reevaluation of the effect of agitation on results using a much less aggressive sampling schedule (data not shown). These data again failed to demonstrate an appreciable effect of agitation on results of any of the antifungal agents tested. Against some strains, however, agitation may have resulted in an increased rate of growth of control samples. Agitation did not affect the level of maximally sustainable fungal growth. Therefore, even though we did not detect a difference in results between agitated and nonagitated samples, we do recommend that time-kill samples be incubated with agitation.

Time-kill studies conducted so far have been conducted with Candida species and nonmucoid strains of Cryptococcus neoformans. We have selected these species for evaluation because of their clinical importance and because in vitro susceptibility tests were standardized primarily with these species. Prior to use in time-kill studies, we evaluated all isolates to confirm favorable growth characteristics. Strains were selected only if control cultures exhibited rapid and sustained growth and produced relatively large well-defined colonies on PDA plates. Therefore, discretion should be exercised in extrapolating time-kill methodology to mucoid strains of C. neoformans, filamentous fungi, and molds until methods are carefully evaluated with these organisms.

We have evaluated several time-kill variables and assessed their impact on test results. Factors such as starting inocula and sampling method can significantly influence time-kill results and/or the interpretation of results. Therefore, in an effort to minimize interstudy variation, we propose that the following procedures be adhered to when conducting antifungal time-kill studies. (i) A starting inoculum of 10⁴ to 10⁶ CFU/ml should be used. (ii) RPMI 1640 buffered to a pH of 7.0 with MOPS (or the medium used for susceptibility testing) should serve as the growth medium. (iii) Time-kill samples should be incubated at 35°C with agitation. (iv) Sampling methods should be evaluated for effect on antifungal carryover prior to implementation. (v) Sampling should continue for at least 24 h. Additionally, criteria used to describe bacterial time-kill data such as cidal (99.9% reduction in CFU/milliliter from the starting inoculum) and static (<99.9% reduction in CFU/milliliter from the starting inoculum) activity or synergy (reduction of 2 log₁₀ CFU/ml by the combination over the most active agent alone), should employed when describing antifungal time-kill results. In our laboratory, we have found that results generated following these procedures are reproducible; however, multicenter validation of these methods should be conducted.