Department of Medicine, University of Manchester, Department of Biochemistry, Hope Hospital
Department of Biochemistry, Wythenshawe Hospital, Manchester, United Kingdom
Ordway Research Institute, Albany, New York
The strategy of combining antifungal drugs in a treatment regimen may improve the outcome of invasive candidiasis. Using a well-validated pharmacodynamic murine model of invasive candidiasis, we defined the effect of the combination of amphotericin B deoxycholate (AmB) and 5-fluorocytosine (5FC) by use of the Greco model of drug interaction. The combination was additive, meaning that the experimental effect did not deviate in a statistically significant manner from the null reference model (or additive surface) of the combined effect. From a clinical perspective, the addition of 5FC to a regimen of AmB may enable the near-maximum effect to be reached in circumstances in which the administration of a given dose of AmB alone produces a submaximum effect but an increase in the dose is not possible, because of dose-related toxicity. Our methods provide a way in which some of the complex issues surrounding antifungal combination treatment can be addressed.
Invasive candidiasis (INVC) is a life-threatening illness. The attributable mortality is 38%49% [1, 2], and the rate of therapeutic failure is 20%50% [35]; these data provide the major impetus to continually develop and refine treatment strategies. The combination of antifungal drugs is increasingly touted as a useful approach [6, 7]; the potential benefits include an attainment of effect that is greater than the effect of either drug administered alone, the use of smaller doses than is otherwise possible with monotherapy, and the maximization of the spectrum of antifungal coverage [7].
Amphotericin B deoxycholate (AmB), despite its well-documented toxicity, remains the reference standard for the treatment of INVC and in clinical trials examining new antifungal drugs [3, 5]. 5-Fluorocytosine (5FC) is a fluorinated pyrimidine analogue that acts as a subversive substrate within the pyrimidine salvage pathway and disrupts both DNA and RNA synthesis [8]. 5FC is active against most Candida species, but, because of the rapid induction of resistance, it is generally prescribed in combination with other antifungal drugs.
The combination of AmB and 5FC is considered to be the standard of care in cryptococcal meningitis [9]. The addition of 5FC to a regimen of AmB may also be useful in cases of INVC that are refractory to AmB alone or in the treatment of infection with Candida species at sites where drug penetration may be compromised. In the present study, we sought to examine, using the Greco model of drug interaction in a well-validated pharmacodynamic murine model of INVC, the nature and the magnitude of the interaction between AmB and 5FC.
MATERIALS AND METHODS
Drugs, drug measurement, and the MICs of AmB and 5FC.
AmB powder (Fungizone) and 5FC solution (10 mg/L) were obtained from Bristol-Myers Squibb Pharmaceuticals and Valeant Pharmaceuticals, respectively. Serum AmB concentrations were measured by high-performance liquid chromatography (HPLC) using a modification of the method of Groll et al. [10]. Briefly, 500 L of HPLC-grade methanol was added to 100 L of mouse serum, vortexed, and centrifuged, and the methanolic supernatant was then evaporated to dryness. The residual was reconstituted in 200 L of mobile phase, which consisted of methanol : acetonitrile : 0.0025 mol/L NaEDTA in a ratio of 500 : 350 : 240 (vol : vol : vol), and 100 L of this solution was used for detection. HPLC analysis was performed using an Alltech Altima C18 column (250 × 4.6 mm; Knauer). The flow rate of the mobile phase (composition as above) was 1.2 mL/min. AmB was eluted after 6.8 min of flow, and its concentration at 382 nm was measured with a UV detector (Kratos). The coefficients of variation (CVs) were 7% and 13% for 0.5 and 0.05 mg/L, respectively. The limit of quantification was 0.05 mg/L.
Samples for the measurement of serum 5FC concentrations were prepared by adding 100 L of 5% trichloroacetic acid to 20 L of mouse serum. HPLC analysis was performed using a Hichrom SCX 10 cm × 4.6 mm column (Thermo Electron) and 10 L of sample. The mobile phase consisted of 10 mmol/L ammonium acetate buffer (pH 4.1), and the flow rate was 1.0 mL/min. 5FC was eluted after 2 min of flow, and its concentration at 270 nm was measured with a UV detector (Thermo Electron). The CV was <7% between 0.2 and 13.0 mg/L. The limit of detection was 0.1 mg/L. The MICs for AmB and 5FC were determined using a microtiter modification of the National Committee for Clinical Laboratory Standards M27-A method [11] and the method recommended by the European Committee on Antimicrobial Susceptibility Testing [12], respectively.
Model of invasive candidiasis.
All in vivo models were approved by the institutional ethics committees. A well-characterized clinical isolate of Candida albicans, F/6862, was used. The isolate was retrieved 24 h before use from beads stored at -70°C, was placed in Sabouraud liquid medium (Oxoid), and was incubated at 35°C on a shaker. The final inoculum was determined by progressive dilution in PBS and was verified by quantitative culture. Male CD1 mice (Charles River Laboratories), 2426 g of body weight, were immunosuppressed with 200 mg/kg cyclophosphamide (Pharmacia) in 0.2 mL of 0.9% saline, which was administered intravenously (iv) via the lateral tail vein 3 days before infection (day -3). On day 0, mice were injected iv with 2 × 104 C. albicans organisms in 0.2 mL of PBS via the lateral tail vein. At 5 h after infection (time 0), AmB in 0.2 mL of 5% dextrose, 5FC in 0.2 mL of 0.9% saline, and the combination of AmB and 5FC in their respective diluents were administered intraperitoneally (ip) in 2 separate and immediately sequential injections. An early control group (4 mice) was killed immediately after the administration of the relevant agent (time 0); early control data were subsequently compared with late control data, which were obtained at 24 h after the initiation of therapy, to ensure that logarithmic growth had been established. Treated mice were killed at 24 h after the initiation of therapy. Both kidneys were dissected, weighed, and processed together by homogenization in 2 mL of PBS. The homogenates were serially diluted to a concentration of 1 : 10 in PBS, and quantitative cultures were determined.
Pharmacokinetics of AmB, 5FC, and the combination of AmB and 5FC.
All pharmacokinetic relationships were determined in infected mice. Three mice were used for each data point. For AmB, 0.625, 1, 2.5, and 5 mg/kg were administered ip once, and blood samples were collected at 0.5, 1, 3, 6, 10, 17, and 24 h after the initiation of therapy (i.e., 84 mice were used to study 7 time points for each of the 4 dose regimens). For 5FC, 6.25, 25, 100, and 200 mg/kg were administered ip once, and blood samples were collected at 0.5, 1, 1.5, 2, 3, 4, and 6 h after the initiation of therapy (i.e., 84 mice were used to study 7 time points for each of the 4 dose regimens). Blood was collected by terminal cardiac puncture and was allowed to clot on ice. Samples were stored at -70°C until analysis.
To investigate the possibility of a pharmacokinetic interaction between AmB and 5FC, both drugs were administered alone and in combination. Six cohorts, which were composed of groups of 3 mice, received AmB and/or 5FC in the following combinations: 0 : 3, 0 : 6, 2 : 0, 0.3 : 6, 2 : 3, and 2 : 6 mg/kg. AmB was administered once at time 0, and 5FC was coadministered at time 0 and then alone at 8 and 16 h after the initiation of therapy. The doses were chosen on the basis of the dose-response relationships defined for each drug when administered alone (see below). Blood samples were collected at 0.5, 3, 16.5, 17, 17.5, 18, and 24 h after the initiation of therapy. Blood samples were collected intensively during the third dosing interval (i.e., between 16 and 24 h after the initiation of therapy) to ensure that there was no evidence of progressive drug accumulation.
Pharmacokinetic data analysis.
The concentrations of AmB and 5FC were modeled using a population pharmacokinetic analysis. The nonparametric adaptive grid (NPAG) with adaptive program of Leary et al. was used [13]. Data were weighted by the inverse of the observed variance of the measured drug concentrations for each group of mice. For both drugs, an open 2-compartment model was used. This model was composed of a central compartment (c), with volume Vc (in L), and a peripheral compartment (p) that were connected by the first-order transfer rate constants Kcp (h-1) and Kpc (h-1) and had bolus input into the peritoneal cavity, first-order absorption from the peritoneal cavity into the central compartment (Ka [h-1]), and first-order clearance (CL) from the central compartment (in L/h). Bayesian parameter estimates were obtained using the "population of one" utility within NPAG.
Determination of the dose-response relationships of 5FC and AmB administered alone.
The dose-response relationships for AmB and 5FC administered alone were defined before the combination matrix was designed. Groups of 3 mice were used for each drug dose. For AmB, the effect induced by 0, 0.01, 0.0625, 0.1, 0.3, 1, and 2 mg/kg administered once at time 0 was determined. For 5FC, the effect induced by 0, 0.125, 0.1, 0.4, 0.6, 0.9, 1.56, 2, 3, and 6.25 mg/kg administered at time 0 and at 8 and 16 h after the initiation of therapy was determined. For both drugs, the observed effect, in terms of the reduction of fungal burden (in log10 cfu/g of kidney), was assessed by fitting to the data the following inhibitory sigmoid Emax model:
where E is the effect of drug exposure (in log10 cfu/g of kidney), Econ is the fungal burden (in log10 cfu/g of kidney) in the absence of treatment, Emax is the maximum reduction in fungal burden (in log10cfu/g of kidney) induced by drug exposure, "exposure" is the dose or other pharmacodynamic variable, and H is the Hill (or slope) constant. Observed data were weighted by the inverse of the observed variance. Pharmacokinetic and MIC data were used to convert the exposure term from dose (mg/kg) to the relevant pharmacodynamic variable, which was taken to be the area under the curve (AUC) : MIC ratio, for AmB, and the fraction of the dosing interval (T) that serum concentrations were greater than the MIC (T>MIC), for 5FC. The model was implemented and fitted using the identification module of the ADAPT II software package [14].
Drug interaction modeling.
To model drug interaction, the interaction model of Greco [15] (equation [1]) was used and implemented in the identification module of the ADAPT II software package [14]. The Greco model provides parameter estimates and their associated 95% confidence intervals (CIs), thus enabling the quantification of the combined effect as well as the level of statistical significance to be determined. The Greco equation, when applied to the present study, takes the form
where DAmB and D5FC are the concentrations of AmB and 5FC, respectively, that produce effect E; mAmB and m5FC are the respective slope parameters for the 2 drugs; IC50,AmB is the AUC : MIC ratio for AmB that produces 50% of the maximum effect; IC50,5FC is the fraction of the dosing interval that the serum concentration of 5FC is above the MIC that produces 50% of the maximum effect; and is the interaction parameter.
The first 2 terms on the right side of equation (1) define the additive effect (Loewe additivity); the third is the interaction term and contains the interaction parameter, . If the 95% CI near the point estimate of crosses 0, the combined effect is additive. If is positive and the lower bound of its 95% CI does not cross 0, synergy is present. If is negative and the upper bound of its 95% CI does not cross 0, antagonism is present.
RESULTS
MICs.
The MICs of AmB and 5FC for C. albicans were 0.03 mg/L and 0.125 mg/L, respectively, in 2 experiments.
Model of INVC.
IV injection of immunosuppressed mice with C. albicans F/6862 generated a reproducible and sublethal infection. The mean ± SD fungal burden of C. albicans at 5 h after infection (time 0) and at 24 h after the initiation of therapy in control mice was 2.68 ± 0.27 and 5.87 ± 0.32 log10 cfu/g of kidney, respectively.
Pharmacokinetics of 5FC and AmB alone and in combination.
A total of 28 sampling points were available for use in the pharmacokinetic analysis for both AmB and 5FC. The estimates of the mean ± SD values for each parameter derived from the NPAG analysis of the 2 compartment models are summarized in table 1. Concentrations of both AmB and 5FC when given in combination were adequately accounted for by the model for either drug alone, and this indicates that there was no significant pharmacokinetic interaction, at least at the levels proposed and subsequently used in the combination matrix (data not shown).
DISCUSSION
The combination of AmB and 5FC in the treatment of INVC is attractive, in theory, because the membrane damage induced by AmB may facilitate the intracellular accumulation of 5FC and thereby cause an effect beyond that observed with either drug alone [17, 18]. In the present study, we have demonstrated, using a rigorous definition of drug interaction and a well-validated pharmacodynamic murine model of INVC, that the combination of AmB and 5FC has an additive effect. On the basis of previous work with 5FC, we employed T>MIC as the pharmacodynamic variable linked to outcome [16]. In the case of AmB, Andes et al. suggested that the peak concentration : MIC ratio is the relevant pharmacodynamic parameter [19]. In the present study, only a single dose of AmB was administered; in this circumstance, there is complete colinearity between peak concentration : MIC and the AUC : MIC ratio, and, for the purposes of computational tractability, we chose to employ the AUC : MIC ratio as the relevant linked pharmacodynamic variable.
The combination of AmB and 5FC against Candida species has been studied in vitro, in vivo, and in a limited number of clinical contexts. In vitro, different definitions and modeling techniques, including the fractional inhibitory concentration and surface response methods, have been used to assess the combined effect [2024]. The majority of studies have concluded that the combination of AmB and 5FC is potentially beneficial, although disparities in methodological methods, analysis, and interpretation prevent any definitive conclusions regarding the overall nature and magnitude of the combined effect. In vivo studies examining the combination of AmB and 5FC have used survival [25, 26] and fungal tissue burden [27] as therapeutic end points; in this context, the combination has been shown to improve survival and reduce fungal tissue burden to a greater extent than is observed with monotherapy. Finally, limited clinical data that describe the safety and efficacy of the combination of AmB and 5FC in a small number of patients in a variety of clinical contexts are available [18, 28], although, at present, no clinical data support the routine addition of 5FC to a regimen of AmB for the treatment of INVC. Thus, the potential benefits of combining AmB and 5FC for INVC remain unclear. The present study enables some of the potential advantages of the combination to be further explored.
A variety of definitions have been employed to define the interaction between 2 drugs; these have been extensively reviewed elsewhere [15]. The Greco model, which is based on the concept of Loewe additivity, was used in this study. The interaction parameters in equation (1) are estimated from the entire data set, at once, using a weighted nonlinear least-squares approach [29]. In the Greco model, the null reference model (or additive surface) is defined when = 0; consequently, the third term in equation (1) is also 0, and the Greco model defaults to an equation of Loewe additivity [29]. If is positive, a greater-than-expected effect (synergy) is present; conversely, if is negative, a less-than-expected effect (antagonism) is present [29]. In the present study, the combination of AmB and 5FC has an additive effect, because the model estimate of , the interaction term, is essentially 0, and the lower bound of the 95% CI crosses 0; thus, the observed experimental effect does not deviate in a statistically significant manner from the null reference model (or additive surface) of the combined effect.
The present study has many limitations. First, only a single isolate was studied; ideally, multiple C. albicans isolates would be investigated to examine whether there are significant strain-to-strain differences in terms of the combined effect of AmB and 5FC. Second, the results are applicable only to isolates with the same susceptibility profile we examined; isolates with higher AmB and 5FC MICs may need higher doses of drugs to produce a near-maximum effect. Third, the results are not directly applicable to the treatment of infections with Candida species at sanctuary sites, where drug penetration may be compromised. Fourth, the results cannot be extrapolated to other organisms, such as Cryptococcus species, although the same methods could be used to answer similar clinically relevant questions. Fifth, the potential differences in outcome induced by prolonged and sequential dosing used in human infections, in comparison with the short duration of treatment often used in pharmacodynamic models, remains poorly elucidated and is a matter for further study. Finally, unrecognized differences in protein binding between experimental animals and humans may limit the extent to which the results obtained in the murine model can be confidently extrapolated to a clinical context.
References
1. Wey SB, Mori M, Pfaller MA, Woolson RF, Wenzel RP. Hospital-acquired candidemia: the attributable mortality and excess length of stay. Arch Intern Med 1988; 148:26425. First citation in article
2. Gudlaugsson O, Gillespie S, Lee K, et al. Attributable mortality of nosocomial candidemia, revisited. Clin Infect Dis 2003; 37:11727. First citation in article
3. Rex JH, Bennett JE, Sugar AM, et al. A randomized trial comparing fluconazole with amphotericin B for the treatment of candidemia in patients without neutropenia. N Engl J Med 1994; 331:132530. First citation in article
4. Rex JH, Pappas PG, Karchmer AW, et al. A randomized and blinded multicenter trial of high-dose fluconazole plus placebo versus fluconazole plus amphotericin B as therapy for candidemia and its consequences in nonneutropenic subjects. Clin Infect Dis 2003; 36:12218. First citation in article
5. Mora-Duarte J, Betts R, Rotstein C, et al. Comparison of caspofungin and amphotericin B for invasive candidiasis. N Engl J Med 2002; 347:20209. First citation in article
6. Marr KA, Boeckh M, Carter RA, Kim HW, Corey L. Combination antifungal therapy for invasive aspergillosis. Clin Infect Dis 2004; 39:797802. First citation in article
7. Sobel JD. Combination therapy for invasive mycoses: evaluation of past clinical trial designs. Clin Infect Dis 2004; 39(Suppl 4):S2247. First citation in article
8. Hope WW, Tabernero L, Denning DW, Anderson MJ. Molecular mechanisms of primary resistance to flucytosine in Candida albicans. Antimicrob Agents Chemother 2004; 48:437786. First citation in article
9. Saag MS, Graybill RJ, Larsen RA, et al. Practice guidelines for the management of cryptococcal disease. Infectious Diseases Society of America. Clin Infect Dis 2000; 30:7108. First citation in article
10. Groll AH, Mickiene D, Petraitis V, et al. Comparative drug disposition, urinary pharmacokinetics and renal effects of multilamellar liposomal nystatin and amphotericin B deoxycholate in rabbits. Antimicrob Agents Chemother 2003; 47:391725. First citation in article
11. National Committee for Clinical Laboratory Standards. Reference method for broth dilution antifungal susceptibility testing of yeasts [approved standard M27-A]. Wayne, PA: National Committee for Clinical Laboratory Standards, 1997. First citation in article
12. Rodriguez-Tudela JL, Barchiesi F, Bille J, et al. Method for the determination of minimum inhibitory concentration (MIC) by broth dilution of fermentative yeasts. Clin Microbiol Infect 2003; 9:18. First citation in article
13. Leary R, Jelliffe R, Schumitzky A, van Guilder M. An adaptive grid, non-parametric approach to pharmacokinetic and dynamic (PK/PD) models. In: Proceedings of the 14th Institute of Electrical and Electronics Engineers (IEEE) Symposium on Computer-Based Medical Systems. Bethesda, MD: IEEE Computer Society, 2001:38994. First citation in article
14. D'Argenio DZ, Schumitzky A. ADAPT II: a program for simulation, identification, and optimal experimental design. User manual. Los Angeles: Biomedical Simulations Resource, University of Southern California, 1997. Available at: http://bmsr.usc.edu/. Accessed 5 July 2005. First citation in article
15. Greco WR, Bravo G, Parsons JC. The search for synergy: a critical review from a response surface perspective. Pharmacol Rev 1995; 47:33185. First citation in article
16. Andes D, van Ogtrop M. In vivo characterization of the pharmacodynamics of flucytosine in a neutropenic murine disseminated candidiasis model. Antimicrob Agents Chemother 2000; 44:93842. First citation in article
17. Medoff G. Controversial areas in antifungal chemotherapy: short-course and combination therapy with amphotericin B. Rev Infect Dis 1987; 9:4037. First citation in article
18. Eilard T, Beskow D, Norrby R, Wahlen P, Alestig K. Combined treatment with amphotericin B and flucytosine in severe fungal infections. J Antimicrob Chemother 1976; 2:23946. First citation in article
19. Andes D, Stamsted T, Conklin R. Pharmacodynamics of amphotericin B in a neutropenic-mouse disseminated-candidiasis model. Antimicrob Agents Chemother 2001; 45:9226. First citation in article
20. Te Dorsthorst DT, Verweij PE, Meletiadis J, et al. In vitro interaction of flucytosine combined with amphotericin B or fluconazole against thirty-five yeast isolates determined by both the fractional inhibitory concentration index and the response surface approach. Antimicrob Agents Chemother 2002; 46:29829. First citation in article
21. Keele DJ, DeLallo VC, Lewis RE, Ernst EJ, Klepser ME. Evaluation of amphotericin B and flucytosine in combination against Candida albicans and Cryptococcus neoformans using time-kill methodology. Diagn Microbiol Infect Dis 2001; 41:1216. First citation in article
22. Odds FC. Interactions among amphotericin B, 5-fluorocytosine, ketoconazole, and miconazole against pathogenic fungi in vitro. Antimicrob Agents Chemother 1982; 22:76370. First citation in article
23. Montgomerie JZ, Edwards JE Jr, Guze LB. Synergism of amphotericin B and 5-fluorocytosine for Candida species. J Infect Dis 1975; 132:826. First citation in article
24. Medoff G, Comfort M, Kobayashi GS. Synergistic action of amphotericin B and 5-fluorocytosine against yeast-like organisms. Proc Soc Exp Biol Med 1971; 138:5714. First citation in article
25. Rabinovich S, Shaw BD, Bryant T, Donta ST. Effect of 5-fluorocytosine and amphotericin B on Candida albicans infection in mice. J Infect Dis 1974; 130:2831. First citation in article
26. Polak A, Scholer HJ, Wall M. Combination therapy of experimental candidiasis, cryptococcosis and aspergillosis in mice. Chemotherapy 1982; 28:46179. First citation in article
27. Thaler M, Bacher J, O'Leary T, Pizzo PA. Evaluation of single-drug and combination antifungal therapy in an experimental model of candidiasis in rabbits with prolonged neutropenia. J Infect Dis 1988; 158:808. First citation in article
28. Abele-Horn M, Kopp A, Sternberg U, et al. A randomized study comparing fluconazole with amphotericin B/5-flucytosine for the treatment of systemic Candida infections in intensive care patients. Infection 1996; 24:42632. First citation in article
29. Drusano GL, D'Argenio DZ, Symonds W, et al. Nucleoside analog 1592U89 and human immunodeficiency virus protease inhibitor 141W94 are synergistic in vitro. Antimicrob Agents Chemother 1998; 42:21539. First citation in article
Protective Immunity Associated with a Strong Serological Response to a Cryptosporidium-Specific Antigen Group, in HIV-Infected Individuals
Interactive Functional Specificity of the Stress and Immune Responses: The Ying, the Yang, and the Defense against 2 Major Classes of Bacteria
Human Cervicovaginal Lavage Fluid Contains an Inhibitor of HIV Binding to Dendritic CellSpecific Intercellular Adhesion Molecule 3Grabbing Nonintegrin
Cytomegalovirus (CMV) Blood DNA Load, CMV Retinitis Progression, and Occurrence of Resistant CMV in Patients with CMV Retinitis
Highly Sensitive Detection and Localization of Maternally Acquired Human Cytomegalovirus in Placental Tissue by In Situ Polymerase Chain Reaction
Frequent Detection of Polyomaviruses in Stool Samples from Hospitalized Children
Virulence May Determine the Necessary Duration and Dosage of Oseltamivir Treatment for Highly Pathogenic A/Vietnam/1203/04 Influenza Virus in Mice
Baseline Correlation and Comparative Kinetics of Cerebrospinal Fluid Colony-Forming Unit Counts and Antigen Titers in Cryptococcal Meningitis
Fluorogenic Assay for Molecular Typing of the Leishmania donovani Complex: Taxonomic and Clinical Applications
Fms-Like Tyrosine Kinase 3Based Immunoprophylaxis against Infection Is Improved by Adjuvant Treatment with AntiInterleukin-10 Antibody