Concentration-Dependent Effects of Caspofungin on the Metabolic Activity of Aspergillus Species

A total of 29 clinical isolates of Aspergillus spp. was used, including 9 isolates of A. fumigatus, 12 isolates of A. terreus, and 8 isolates of A. flavus. Conidia were harvested after isolates were subcultured on potato dextrose agar at 35°C for 5 to 7 days and were suspended in normal saline containing 0.025% Tween 20. Conidial suspensions were counted with a hemacytometer, and inoculum sizes were verified by quantitative colony counts on Sabouraud dextrose agar plates. Reference strains Candida krusei ATCC 6258 and Candida parapsilosis ATCC 22019 were used as quality controls.

RPMI 1640 medium with l-glutamine but without bicarbonate, buffered to pH 7.0 with 0.165 M 3-N-morpholinopropanesulfonic acid (Cambrex BioScience, Inc., Walkersville, MD), was used as the assay medium.

The tetrazolium salt XTT (Sigma-Aldrich, St. Louis, MO) was dissolved in normal saline at 0.5 mg/ml. Menadione (Sigma-Aldrich) was dissolved in absolute ethanol at 10 mg/ml (58 × 10⁻³ M) and subsequently added to the XTT solution at concentrations of 125, 31.25, or 7.8 μM.

Susceptibility to caspofungin was studied for all Aspergillus isolates using the broth microdilution method based on the CLSI M38-A protocol (19). Caspofungin (Merck & Co., Inc., Whitehouse Station, NJ) was dissolved in sterile water. Twofold serial dilutions in the assay medium were initially prepared in flat-bottom 96-well microtitration plates (Costar 3596; Corning, Inc., Corning, NY) in order to obtain final concentrations ranging from 0.008 to 8 μg/ml at a total volume of 200 μl after inoculation. The range of concentrations studied was chosen based on human pharmacokinetic studies (25, 27). After inoculation of caspofungin trays with a final concentration of 2.5 × 10⁴ conidia/ml and incubation at 37°C for 48 h, the MEC was defined microscopically as the lowest drug concentration that produced short, stubby, and highly branched hyphae (3, 9). In each tray, a row of wells was filled with RPMI 1640 up to a total volume of 200 μl, without fungal inoculum, and served to provide background absorbance values in subsequent spectrophotometric measurements of XTT conversion. The experiments were repeated in triplicate. For reasons of clarity, the use of the term MEC throughout the text implies the microscopically determined MEC, whereas the term X-MEC implies that this value was determined with the XTT assay (see below).

For six Aspergillus isolates (two of each species), immediately after MEC determination 50 μl of the XTT-menadione solutions described above was added to each well, resulting in final concentrations of 100 μg of XTT/ml and 25, 6.25, or 1.56 μM menadione. The plates were incubated at 37°C for 2 h and subsequently shaken for 1 to 2 min (Wallac Plate Shake 1296-004; Wallac OY, Turku, Finland) for further dissolution of the formazan derivatives (2, 14). Color absorbance (A) was then measured at dual wavelengths (450 and 630 nm [reference] in order to correct for the absorbance of hyphae) with a microtitration plate spectrophotometric reader (Elx808; Bio-Tek Instruments, Winooski, VT). The percent metabolic activity for each drug well in relation to the drug-free control was calculated after subtraction of the background absorbance as follows: % metabolic activity = (A_{drug well} −A_{background drug well})/(A_control − A_{background control}) × 100%. For each isolate, the measurements obtained using different menadione concentrations were evaluated with respect to the changes in the percent metabolic activity observed at or one to two wells near the MEC, as well as the degree of experimental variability.

After obtaining the results of these studies (see Results), we additionally investigated whether performing the XTT assay with the higher menadione concentration (25 μM) but shorter incubation period (30 min or 1 h) would yield changes in the percent metabolism at the MEC that would be comparable to those detected using 6.25 μM menadione and a 2-h incubation. For this purpose the six isolates mentioned above were studied, and the XTT assay was performed as described above, with shorter incubation times for the plates to which 25 μM menadione had been added.

Following the above studies of optimization of the XTT assay (see Results), the metabolic activity in the presence of increasing concentrations of caspofungin for the greater collection of Aspergillus isolates was assessed using final concentrations of 100 μg of XTT/ml and 6.25 μM menadione and a 2-h incubation time. The XTT assay was performed as described above, immediately after microscopic MEC determination at 48 h of incubation. For each isolate, the percent metabolic activity obtained at the MEC was recorded. The caspofungin concentration at which the metabolic activity reached its lowest values was also recorded and, for the purpose of the present study, is referred to as the minimum metabolic activity concentration (MMC). In case a paradoxical increase in metabolic activity was observed at higher caspofungin concentrations, it was considered significant if it exceeded the metabolic activity at MMC by at least 40% and was consistent in all three experiments. The cutoff of 40% was used based on previous data demonstrating a coefficient of variation up to 19% for the spectrophotometric measurements of XTT conversion (15). For each species the values of the percent metabolic activity at 8 μg of caspofungin/ml were compared to those at the MMC.

The assessment of the percent metabolic activity corresponding to the microscopic MEC for each Aspergillus isolate allowed the development of a method for the determination of the X-MEC by using the optimized XTT assay (100 μg of XTT/ml and 6.25 μM menadione). Determination of the X-MEC required the use of cutoff values of the percent metabolism (compared to drug-free control) at 48 h that were based on the best agreement obtained between the microscopic MEC and X-MEC values, as described below. Starting from the median value of the percent metabolism at the microscopic MEC obtained for each species, a range of values of metabolic activity in increments of 5% (± from the median) were evaluated as cutoff levels for this purpose. The XTT endpoint (i.e., the X-MEC) was defined as the lowest caspofungin concentration showing a percent metabolic activity equal to or less than the value evaluated as cutoff level. For each species, values of percent metabolic activity for which the best relative agreement (± one dilution) was obtained between the X-MEC and microscopic MEC were defined as cutoff levels for determination of X-MEC with the optimized XTT assay.

The choice of model for description of the concentration-effect relationship between the caspofungin concentration and metabolic activity for Aspergillus spp. depended on the presence or absence of a paradoxical increase of fungal metabolism at higher concentrations.

For isolates for which a paradoxical response was not detected, the E_max model (sigmoid curve with a variable slope) was used. This four-parameter logistic model is described by the equation: Y = Y_min + (Y_max − Y_min)/[1 + 10^{(logEC50-X)*slope}], where X is the log₁₀ of the caspofungin concentration, Y is the corresponding percent metabolic activity, and the four parameters are the values of the maximal (Y_max) and minimal (Y_min) percent metabolic activities, respectively, the EC50, which is the drug concentration showing metabolic activity that is halfway between the Y_max and Y_min, and the slope, which describes the steepness of the curve (18).

For isolates demonstrating a paradoxical increase in metabolic activity at higher caspofungin concentrations, two different models were evaluated and compared. The first model combines two sigmoid concentration-effect relationships to describe what looks like a bell-shaped curve based on the equation: Y = Y_min + (Y_max1 − Y_min)/[1 + 10^{(logEC50_1-X)*nH1}] + (Y_max2 − Y_min)/[1 + 10^{(logEC50_2-X)*nH2}], where X is the log₁₀ of caspofungin concentration and Y is the corresponding percent metabolic activity (17). The biphasic curve begins at Y_max1, turns over at Y_min, and then approaches Y_max2. EC50-1 and EC50-2 are the corresponding EC50 values for each phase of the curve, while the model includes two more parameters, the slope factors n_H1 and n_H2 for each phase of the curve. The problem, however, with the bell-shaped equation is that it is a complicated model, requiring a sufficient number of datum points to adequately define its seven parameters and, consequently, both phases of the response to the drug. For this reason, the possibility of a simpler model that could still describe such a biphasic concentration-effect relationship was also investigated. A relatively simple model for this purpose is based on the following equation that describes the Gaussian distribution: Y = Y_max − range*exp(−1*[(X-midA)/slope]²), where midA is the logEC50 − slope*[−ln(0.5)]. This model includes only four parameters: the maximal response (Y_max), the logEC50, the range (i.e., the difference between the maximal and minimal response), and the slope. Since the Gaussian model requires fewer parameters than the bell-shaped model, it can be used to fit concentration-effect curves with fewer datum points. An important limitation of the Gaussian model, however, is its symmetric nature, implying that both phases of the curve are mirror images (17).

For each of the isolates demonstrating a paradoxical effect, direct comparison of the above two models (bell-shaped and Gaussian) was performed by using the corrected Akaike's information criterion (6). The latter is a valid method for comparing models that is based on information theory and tells which model is more likely to explain the data by balancing the improvement in goodness-of-fit (R²), obtained by using a more complex model, with the increase in the number of parameters. Model building was performed by using GraphPad Prism software (4.0b; San Diego, CA).

Comparisons of metabolic activity at the MEC or MMC among Aspergillus species were performed by nonparametric analysis of variance, using the Kruskal-Wallis test, followed by Dunn's test for multiple comparisons. In order to compare the metabolic activity at 8 μg of caspofungin/ml versus MMC for each species, a Wilcoxon matched-pairs test was used (GraphPad Prism Software 4.0b). A P value of <0.05 was considered statistically significant.

MICs for quality control strains were within reference range, and the MEC ranged from 0.25 to 0.5 μg/ml for all Aspergillus isolates (Table 1) .

In the initial 2-h incubation experiments, when the higher concentration (25 μM) of menadione was used, no significant change in the percent metabolic activity of A. terreus and A. flavus isolates was detected in the presence of increasing caspofungin concentrations, whereas for A. fumigatus the reduction in metabolic activity did not exceed 40% at concentrations greater than or equal to the MEC. With the lower concentration of 6.25 μM menadione, however, a clear reduction in fungal metabolism was detected at caspofungin concentrations greater than or equal to the MEC for all three species. This reduction was on the order of 70 to 80% for the A. fumigatus and A. flavus isolates used in these preliminary studies and 50% for A. terreus. Interestingly, for both A. fumigatus and one of the two A. terreus isolates a paradoxical increase in metabolic activity was observed at higher caspofungin concentrations (Fig. 1). When 1.56 μM menadione was used, the concentration-dependent changes in metabolic activity were similar to those obtained with 6.25 μM. However, low XTT conversion rates (absorbance of 0.2 to 0.3) were frequently observed with the 1.56 μM concentration, which resulted in increased variability in the calculated percent metabolic activity (data not shown).

Using incubation for 1 h or 30 min and 25 μM menadione, we found that the reduction in the percent metabolism detected at the MEC was progressively increased with shorter periods of incubation and at 30 min, for three of the six isolates, it was comparable to that detected using 6.25 μM menadione and a 2-h incubation. However, for the other three isolates very low absorbance values were observed (0.1 to 0.2), resulting again in significant interexperimental variability in the obtained values for the percent metabolic activity.

Consequently, a menadione concentration of 6.25 μM and an incubation time of 2 h were considered to be the optimal conditions for assessment of the caspofungin effects on the metabolic activity of the greater collection of Aspergillus isolates using the XTT assay.

For all Aspergillus isolates, a substantial reduction in metabolic activity was associated with the formation of short, aberrant hyphae in the presence of caspofungin. As demonstrated in Table 1, this reduction was more pronounced (P < 0.01) for A. flavus (median metabolism = 25% of control) compared to A. fumigatus and A. terreus (median metabolic activities of 42 and 53%, respectively). For most isolates, the MMC tended to be one to two dilutions higher than the MEC. The intraspecies range of MMC values appeared to be broader than the MEC values; however, for each isolate the interexperimental variability of MMC rarely exceeded one dilution. The interspecies differences in the metabolic activity recorded at the MMC paralleled those observed for the MEC, with A. flavus exhibiting greater metabolic inhibition than A. fumigatus and A. terreus (P < 0.01). Both at the MEC and at the MMC, significant and reproducible intraspecies variation in the metabolic activity was demonstrated, mainly for A. fumigatus and A. terreus (Table 1).

For approximately half of the A. fumigatus and A. terreus isolates a gradual increase in metabolic activity was observed at caspofungin concentrations higher than the MMC, suggesting a paradoxical effect. This paradoxical increase was infrequent for A. flavus (Table 1). For A. fumigatus and A. terreus isolates, the average metabolic activity at 8 μg of caspofungin/ml was increased by factors of 1.82 and 1.47, respectively, compared to that of MMC (P < 0.001 for both species), whereas for A. flavus it was increased by 1.15, which was not statistically significant.

The significant reduction in fungal metabolism detected at the MEC for Aspergillus spp. could allow determination of X-MEC with the optimized XTT assay. However, because of the interspecies differences in the observed metabolic activity at the microscopic MEC (Table 1), the values of the percent metabolic activity that could serve as cutoff levels for X-MEC determination would have to be different for each Aspergillus species. Indeed, the cutoff levels of the percent metabolism that gave the best relative agreement (± one dilution) with the microscopically determined MEC were different for each species. Using these cutoff levels, the agreement between the X-MEC and microscopic MEC ranged between 93 and 100% (Table 2).

For isolates for which a paradoxical response was not detected (i.e., most of the A. flavus and approximately half of the A. fumigatus and A. terreus isolates, a total 17 of 29 isolates), the data were well described with the E_max model, as demonstrated by R² values ranging between 0.87 and 0.99 (median = 0.97), and no significant deviations with the runs test.

The median (range) of minimal (Y_min) values of the percent metabolism calculated with the E_max model for the above isolates was 35% (21 to 60%) for A. fumigatus, 44% (27 to 77%) for A. terreus, and 20% (12 to 27%) for A. flavus, demonstrating again a greater reduction of metabolic activity for A. flavus compared to other species, in agreement with the findings presented in Table 1. The generated sigmoid curves were steep, as manifested by the high absolute numbers of the (negative) slope factors, which ranged from 4.39 to 35.5 (median = 7.1) without significant interspecies differences. For all isolates, the metabolic activity corresponding to the microscopic MEC was lower than that of EC50, approximating the Y_min. However, because of the steepness of the sigmoid curves, the EC50 and MEC values for every isolate were very close (Fig. 2A). Thus, while the MEC values ranged between 0.25 and 0.5 μg/ml (Table 1), the EC50 values ranged between 0.14 and 0.48 μg/ml. In addition, the ratio MEC/EC50 ranged from 1.03 to 1.93 (median = 1.40), suggesting that the microscopic MEC tended to be slightly higher than the EC50, but their difference would not exceed one twofold dilution. Consequently, the EC50 values generated by the E_max model provided a very good approximation (within one dilution) of the MEC values. These findings suggest that modeling of data and calculation of EC50 values could provide an alternative way for determination of X-MEC with the optimized XTT assay, without the use of the cutoff levels presented in Table 2.

For the remaining 12 isolates for which a paradoxical increase was detected at higher caspofungin concentrations (Table 1), the bell-shaped model was able to fit the data very well, as manifested by high R² values (0.85 to 0.99 [median = 0.97]) (Fig. 2B). The absolute values of slope factors (n_H1) of the first phase of the biphasic curve (associated with the MEC) ranged between 2.28 and 39.92 (median: 6.48) and were comparable to those reported above for the E_max model. However, the second phase of the curve (associated with the paradoxical effect) tended to be less steep, as manifested by lower values of the corresponding slope factors (n_H2), which ranged from 0.73 to 8.32 (median = 1.53). As with the E_max model, the derived EC50-1 values ranged between 0.15 and 0.67 μg/ml and were always within one dilution from the microscopic MEC (Fig. 2B). In the present study, the number of 12 datum points (11 concentrations of caspofungin plus drug-free control) was small relative to the number of seven parameters defined by the bell-shaped equation. Consequently, appropriate initial parameter values needed to be chosen in order to assure that the model converged successfully. Furthermore, wide confidence intervals were sometimes obtained for the best-fit values of some of the parameters (mainly those describing the second phase of the curve).

In contrast to the bell-shaped equation, the relatively simple Gaussian model easily converged for all isolates demonstrating the paradoxical response, and its parameters were well defined. Its symmetric nature, however, proved to be a disadvantage for description of the asymmetric biphasic concentration-effect curve observed with caspofungin, in which the first phase was usually steeper than the second. Consequently, the Gaussian equation did not appear to fit the data as well as the bell-shaped model did (Fig. 3). This problem was also demonstrated by the R² values, which ranged between 0.78 and 0.96 (median = 0.89) (i.e., lower than the corresponding values for the bell-shaped model). Direct comparison of the two models for each of the 12 isolates using the corrected Akaike's information criterion suggested that the bell-shaped model was more adequate to describe the data than the Gaussian equation in 8 of the 12 comparisons.