In Vitro Pharmacodynamic Characteristics of Amphotericin B, Caspofungin, Fluconazole, and Voriconazole against Bloodstream Isolates of Infrequent Candida Species from Patients with Hematologic Malignancies

Abstract

Time-kill and postantifungal effect (PAFE) of amphotericin B, caspofungin, fluconazole, and voriconazole were determined against clinical isolates of Candida guilliermondii, Candida kefyr, and Candida lusitaniae. Azoles displayed fungistatic activity and no measurable PAFE, regardless of the concentration tested. Amphotericin B and caspofungin demonstrated concentration-dependent fungicidal activity, although amphotericin B only produced a significant dose-dependent PAFE against all isolates tested.

Invasive fungal infections are important causes of morbidity and mortality in immunosuppressed patients (10). Although C. albicans, C. glabrata, C. parapsilosis, and C. tropicalis account for the majority of Candida bloodstream infections, recent epidemiologic trends indicate a shift toward infections by the less frequently isolated non-albicans Candida (NAC) species (12). Among NAC species, C. kefyr, C. guilliermondii, and C. lusitaniae are rare causes of invasive infections but are increasingly encountered among severely immunosuppressed patients occurring in nosocomial clusters and/or exhibiting innate or acquired resistance to one or more established antifungal agents, often related to intravascular catheters and breaks in infection control precautions (3-5, 11, 12, 14).

Currently, knowledge of the in vitro pharmacodynamic characteristics of C. kefyr, C. guilliermondii, and C. lusitaniae is poor and limited to amphotericin B (AMB) and voriconazole (VRC) only (7, 15). Therefore, we conducted time-kill and postantifungal effect (PAFE) studies with AMB, caspofungin, fluconazole (FLC), and VRC against bloodstream isolates of C. guilliermondii, C. kefyr, and C. lusitaniae from neutropenic patients.

Antifungal agents.

Stock solutions of AMB (Sigma-Aldrich SRL, Milan, Italy), caspofungin (Merck Sharp & Dohme Italia SpA, Rome, Italy), FLC (Pfizer Inc., New York, N.Y.), and VRC (Pfizer) were prepared in RPMI 1640 medium (Sigma) buffered to a pH of 7.0 with 0.165 M morpholinepropanesulfonic acid (MOPS) buffer (Sigma) and stored at −80°C until use. Antifungals were solubilized in sterile water, except AMB in dimethyl sulfoxide (Sigma).

Test isolates.

Six Candida isolates were obtained from the Clinical Microbiology Service, Department of Hematology and Oncology, “Spirito Santo” Hospital, Pescara, Italy, for use in this study: two strains each of C. guilliermondii (337 and 555), C. kefyr (240 and 270), and C. lusitaniae (325 and 447) were selected for testing.

Antifungal susceptibility testing.

The MIC for each isolate was determined, in triplicate, by broth microdilution techniques as outlined by the National Committee for Clinical Laboratory Standards (17). The endpoint was defined as 50% inhibition of visible growth for azoles and complete inhibition of visible growth for AMB and caspofungin.

Time-kill.

Before the time-kill studies were initiated, antifungal carryover effects were examined. Briefly, 100 μl of a standardized suspension (1 × 10³ CFU/ml) of each isolate were added to either 900 μl of sterile water with (sample) or without (control) antifungal at 4 and 8 times the MIC. Immediately following the addition of the antifungal, 100 μl was removed and streaked across Sabouraud dextrose agar (SDA) for colony count determination. Antifungal carryover was defined as a reduction in colony counts in samples by >25% compared to control.

Time-kill experiments, conducted in duplicate, began by suspending colonies from a 24-h SDA growth in RPMI 1640 with MOPS (RPMI 1640-MOPS) to approximately 10⁶ CFU/ml. One milliliter of this fungal suspension was then added to 9 ml of RPMI 1640-MOPS without (growth control) or with (test) antifungal at concentrations ranging from 0.125 to 8 times the MIC. The culture was then incubated at 37°C with shaking. At predetermined time points (0, 1, 2, 4, 8, 24, and 48 h), 100 μl samples were removed and serially diluted in cold sterile distilled water, and 100 μl was plated onto SDA for colony counting.

PAFE.

Growth control and test vials were prepared as for time-kill experiments. Following an incubation period of 1 h, antifungal was removed by three sequential centrifugations (3,000 × g, 10 min). The fungal pellet was resuspended in normal saline after the first two centrifugation periods and in 9 ml of warm RPMI 1640-MOPS after the final centrifugation. The resuspended samples were then incubated at 35°C with agitation, and colony counts were performed at 0, 2, 4, 6, 8, and 24 h after the final wash. PAFE experiments were conducted in duplicate.

Analysis.

Fungicidal activity was defined as a ≥3 log₁₀ (99.9%) reduction in CFU per milliliter from the starting inoculum concentration. To quantify the extent of antifungal activity, maximal effect (E_max), the concentration producing 50% of E_max (EC₅₀), the EC₉₀, and the range of the net change (log₁₀ CFU per milliliter) in fungal density were determined for each time point by a sigmoidal Hill three-parameter model with GraphPad Prism (version 4; GraphPad Software Inc., San Diego, Calif.). PAFE was calculated by taking the difference in time required for control and test isolates to grow 1 log₁₀ following drug removal.

Antifungal susceptibility results.

Median MICs ranged from 0.5 to 4.0 μg/ml for AMB, 0.25 to 128 μg/ml for caspofungin, 0.12 to 2.0 μg/ml for FLC, and 0.12 to 1.0 μg/ml for VRC (Table 1).

TABLE 1.

Microdilution broth results for study isolates

Test isolate	Median MIC (μg/ml) (n = 6)
	AMB	Caspofungin	FLC	VRC
C. guilliermondii 337	2	128	2	1
C. guilliermondii 555	0.5	128	2	1
C. kefyr 240	4	0.25	0.12	0.12
C. kefyr 270	4	0.25	0.25	0.25
C. lusitaniae 325	4	0.5	2	0.5
C. lusitaniae 447	4	0.5	2	1

Time-kill results.

No antifungal carryover has been observed with any of the isolates at the concentrations tested. Time-kill plots representative of those noted in this study are showed in Fig. 1. AMB yielded fungicidal activity at concentrations ≥4 times the MIC against C. guilliermondii isolates, reaching fungicidal activity between 1.1 and 3.1 h. AMB exhibited fungicidal activity beginning at the MIC against both C. kefyr isolates (fungicidal endpoint range, 0.5 to 3.9 h). However, regrowth was observed for strain 240 after 8 h at 1 and 4 times the MIC. AMB showed fungicidal activity against both C. lusitaniae isolates beginning at 0.25 times the MIC (fungicidal endpoint range, 0.6 to 6.2 h). Caspofungin exhibited fungicidal activity against both C. guilliermondii isolates tested at ≥4 times the MIC (fungicidal endpoint range, 0.5 to 13 h). A comparable trend was observed for C. kefyr isolates (fungicidal endpoint range, 2.5 to 11 h), although regrowth was observed for both C. kefyr isolates after 8-h exposition at 0.125, 0.25,r and 1 times the MIC. Caspofungin tested against C. lusitaniae produced fungistatic activity only, except at 8 times the MIC against C. lusitaniae 447 reaching the fungicidal endpoint after 15.6 h. FLC and VRC showed fungistatic activity against all test isolates, regardless of the concentrations tested. Composite EC_max model parameters are summarized in Table 2. The EC₉₀ for both AMB and caspofungin decreased to about the MIC at 8 h and then increased to about 2 times the MIC at 24 h. EC_max increased over time up to 24 h for both antifungals, although AMB yielded fungicidal activity more rapidly than caspofungin did (2 versus 8 h, respectively). The EC₉₀ for both FLC and VRC was approximately 2 times the MIC up to 24 h.

FIG. 1.

Representative time-kill plots for the following: AMB against C. guilliermondii 337 (A) and C. lusitaniae 447 (B); caspofungin against C. kefyr 270 (C); and VRC against C. lusitaniae 325 (D). Antifungals were tested at the following concentrations: 0.125 times the MIC (▴), 0.25 times the MIC (▾), the MIC (⧫), 4 times the MIC (•), and 8 times the MIC (□). ▪, control.

TABLE 2.

Composite E_max model parameters

Antifungal and time point (h)	EC50 (multiple of MIC)	EC90 (multiple of MIC)	ECmax (log10 CFU/ ml)	Net inoculum change (log10 CFU/ml)
AMB
1	0.99	1.97	−2.96	−5.28 to 0.25
2	0.80	1.68	−3.65	−5.56 to 0.17
4	0.64	1.78	−4.42	−5.56 to 0.36
8	0.37	1.18	−4.63	−5.56 to 1.00
24	0.70	1.94	−4.55	−5.56 to 2.37
Caspofungin
1	2.02	2.97	−1.45	−5.47 to −0.14
2	1.40	2.34	−1.91	−5.47 to 0.13
4	1.39	2.32	−2.44	−5.47 to −0.50
8	0.48	1.20	−3.06	−5.47 to −0.33
24	1.05	1.94	−3.62	−5.47 to 0.60
FLC
4	1.34	2.28	0.33	−0.04 to 1.02
8	1.18	2.10	0.41	0.03 to 2.08
24	1.26	2.18	0.59	0.06 to 2.50
VRC
4	0.99	1.99	0.31	−0.20 to 1.03
8	0.74	1.60	0.30	−0.34 to 2.09
24	0.88	1.77	0.47	−0.09 to 2.89

PAFE results.

Yeasts exposed to FLC and VRC did not produce any measurable PAFE regardless of antifungal concentrations. Caspofungin induced measurable PAFEs for C. lusitaniae 325 only, regardless of concentration tested. In contrast, significant PAFEs were induced by AMB against each of the test isolates. PAFE was generally influenced by concentration in a dose-dependent manner ranging from 1.3 to 9.4 h (mean, 5.5 h), 3.6 to 10 h (mean, 7.8 h), and 9.2 to 14.9 h (mean, 11.2 h) at 0.125, 0.25, and 1 times the MIC, respectively. At 4 and 8 times the MIC, AMB yielded a PAFE of about 13 h for C. lusitaniae 325 only.

Discussion.

According to previous findings (7, 15), AMB was fungicidal against all isolates tested, showing the greatest activity against C. lusitaniae isolates despite the fact that C. lusitaniae is known for rapid development of AMB resistance (4). Caspofungin displayed fungicidal activity against all isolates but C. lusitaniae 325, suggesting that its fungicidal activity is isolate and species specific. Composite EC_max model parameters suggested that the rate of killing of caspofungin and AMB was influenced by increases in the drug concentration. In fact, by comparing the relative EC₉₀s over time, we observed decreased values until 8 h, hence increased up to 24 h, suggesting that higher concentrations result in more-rapid expression of activity. AMB proved to be more potent than caspofungin, as suggested by EC₅₀ data (mean EC₅₀, 0.8 versus 1.3, respectively) against all isolates tested up to 24 h after antifungal exposure. Further, the EC_max of caspofungin and AMB did increase over time, suggesting that maximal activity occurs at the latest time point observed. However, AMB reached the fungicidal endpoint more rapidly than caspofungin did (2 versus 8 h, respectively). The azoles produced fungistatic effects regardless of concentration, supporting previously published data for C. albicans and other NAC species (7, 13). EC₅₀s and EC₉₀s calculated for FLC and VRC suggested that fungistatic activity did not improve with increasing concentrations of the azoles. The increase observed in the EC₅₀s and EC₉₀s of antifungals tested at 24 h may represent an amount of regrowth seen at fungistatic concentrations, too small to be appreciated in time-kill plots but enough to modify the EC₅₀ and EC₉₀.

The information obtained in this study could be further applied to optimize future dosing regimens for this agent. MIC₉₀s of AMB, caspofungin, FLC, and VRC for the NAC species we tested have been previously reported to range from 0.5 to 8.0, 1 to 16, 0.5 to 16, and 0.06 to 0.5 μg/ml, respectively (1, 8, 9, 16, 18-22). Considering mean EC₉₀s calculated from our E_max model (1.4 times the MIC for AMB and 2 times the MIC for caspofungin and both azoles), we predict that close-to-maximal effect (EC₉₀) would be observed if in vivo AMB, caspofungin, FLC, and VRC concentrations of 1 to 16, 1.4 to 22.4, 1 to 32, and 0.12 to 1 μg/ml were achieved, respectively. These values exceed the peak plasma FLC and VRC concentrations (40 and 4 μg/ml, respectively) achieved in human serum following clinically utilized doses (23, 24), suggesting that concentration-independent fungistatic activity is likely to predominate at clinically achieved FLC and VRC concentrations. On the contrary, peak plasma AMB and caspofungin concentrations (2 and 1 μg/ml, respectively) (24) suggest that current dosing strategies for these antifungals should be reevaluated in order to produce higher peak serum levels.

This is the first published study on the PAFE produced by antifungal agents against C. guilliermondii, C. kefyr, and C. lusitaniae. Although no PAFE was induced by the azoles in any of the strains tested, the method we used for removing drug gets the azole concentrations below the sub-MIC levels that have been previously reported by Turnidge et al. (25) to result in a clinically relevant slower growth of the yeasts. AMB produced a significant dose-dependent PAFE for all isolates tested, as has been previously described for other Candida spp. (2, 25). Caspofungin, although it exhibited fungicidal activity as well as AMB, did not induce a measurable PAFE, in contrast with findings of Ernst et al. (6) concerning C. albicans and Cryptococcus neoformans.

In conclusion, our results demonstrate that AMB may be useful for the treatment of infections caused by C. kefyr, C. guilliermondii, and C. lusitaniae. Animal and clinical studies are warranted to define the clinical relevance of our data.