Materials and Methods

Clinical Studies.

Twenty-two children (8 females) aged between 2 months and 18 years (median 4 years) were investigated. No patient at the time of study, or in the 28 days before the study, had received other drugs known to interfere with P-450 activity. All patients were treated with cyclophosphamide as a constant rate 1-h infusion; the median dose administered was 1050 mg/m² (range 270–2130 mg/m²). Nine children were receiving fluconazole (5 mg/kg once daily, either orally or as a 1-h i.v. infusion) for at least 1 week before the study (Table 1). Children being treated with fluconazole received a higher dose of cyclophosphamide (median 1285 mg/m2 versus 600 mg/m²), and as a result, were more likely to receive mesna during chemotherapy.

For ethical reasons the decision to treat with fluconazole was made by the childrens’ attending physicians and was not done to improve study design; as a result, this was a nonrandomized study. Fluconazole was most often used in children considered to be at high risk of fungal infection, in particular, patients receiving “high-dose” cyclophosphamide therapy. A randomized in vivo study of fluconazole was not considered to be ethical in this patient group. The patients included in the study formed a retrospective case series drawn from a population of over 50 children with a previously established profile of cyclophosphamide metabolism. Within this population, children who were receiving (or had received) treatment within the month before the study with known P-450 enzyme inducers/inhibitors were excluded, leaving 22 children who were analyzed in terms of fluconazole treatment. Of these, nine patients were receiving fluconazole and 13 children could be identified as controls.

Before the infusion and at 0.5, 1, 2, 4, 6, 12, 18, and 24 h after the start of cyclophosphamide administration 3 to 5 ml of blood was collected from an indwelling central venous catheter and anticoagulated with EDTA. Plasma was separated and frozen immediately at −20°C before analysis. The concentrations of cyclophosphamide, CX, and DCCP in plasma were determined using a high performance thin layer chromatography-photographic densitometry assay, which has been described previously (Yule et al., 1995). Unfortunately, this technique does not reliably detect phosphoramide mustard, which requires bedside derivatization to increase its stability. Although several methods for measuring phosphoramide mustard have been described, none have proven to be reproducible (Yule et al., 1993). Cyclophosphamide pharmacokinetics were assumed to follow a one-compartment open model with first order elimination kinetics assuming constant rate drug infusion. Estimates of cyclophosphamide pharmacokinetic parameters were obtained by maximum likelihood using ADAPT II (D’Argenio and Schumitzky, 1992). The variance was weighted in favor of 1/C², where C was the absolute concentration of cyclophosphamide in plasma. To allow for heteroscedasticity, the error variance at time, o²(t), was assumed to take the functional form o²(t) = (g + dm(t))², where m(t) was the expected, or fitted, plasma concentration at time t. The parameters g and d were estimated as part of the fitting process and were not specified a priori. The area under the concentration-time curve (AUC) for individual metabolites was calculated using the trapezoidal rule. Comparisons between children receiving fluconazole and the remainder of the study population were performed using Student’s unpaired t test. Written consent from the child’s parents and, where appropriate, from the subjects themselves, was obtained before participation in the study. The project was approved by the joint ethical committee of the Medical School of the University of Newcastle upon Tyne and the Royal Victoria Infirmary, Newcastle upon Tyne.

Human Liver Microsomal Incubations.

Human liver microsomes were purchased from the Keystone Skin Bank (Exton, PA and The International Institute for the Advancement of Medicine, University of Leicester, UK). Each batch of microsomes had been characterized for the activity of several P-450s by the suppliers; CYP2A6 activity was determined by the rate of coumarin 7-hydroxylation, CYP2C19 activity was determined as the rate of mephenytoin 4-hydroxylation, and CYP3A activity was determined by the rate of production of [¹⁴C]6β-hydroxytestosterone. Several additional studies were performed on these human liver microsomes in our laboratory as part of other ongoing studies. CYP2C9 activity was determined as the rate of diclofenac 4-hydroxylation (Leeman et al., 1993) and the expression of CYP2C8 and CYP3A4 was measured using immunoblotting with enhanced chemiluminescence detection (Amersham International, Amersham, UK) using anti-CYP2C (Dr. P. Beaune, Centre Universitaire des Saints-Pères, Paris, France) and anti-CYP3A antibodies (Dr. J. Hardwick, Northeastern Ohio University, Rootstown, OH).

Cyclophosphamide 4-hydroxylation was ascertained as the fluorescence of the 7-hydroxyquinoline produced by the reaction of 3-aminophenol with the acrolein liberated from 4-hydroxycyclophosphamide (Kurowski et al., 1991). Microsomes were incubated at 0.25 to 0.5 mg/ml protein for 45 min in conditions described previously (Walker et al., 1994). Fluconazole was added to individual microfuge tubes to provide concentrations of 5, 10, 50, 100, 250, and 500 μM and incubated at 37°C for 10 min before the initiation of the reaction by the addition of cyclophosphamide to provide a final substrate concentration of 1 mM. This concentration was selected to reflect in vivo levels of cyclophosphamide in pediatric patients (Juma et al., 1984; Tasso et al., 1992). All measurements were the result of duplicate observations. Experiments were performed on six different microsomes to determine the concentration of fluconazole at which 4-hydroxycyclophosphamide production was reduced to 50% of control values (IC₅₀). Estimates of IC₅₀were made by fitting a Hill equation to data for 4-hydroxycyclophosphamide formation as a function of fluconazole concentration using GraphPad Prism (GraphPad Software, Inc., San Diego, CA). The initial value was fixed at 100% and the Hill coefficient to 1. To investigate whether the presence of mesna altered the extent of cyclophosphamide 4-hydroxylation, a series of microsomal incubations were performed with and without mesna at a concentration of 1 mM.

Results

Clinical Studies.

The disappearance of cyclophosphamide from plasma was monoexponential in all cases. Details of the pharmacokinetic parameters and of individual metabolite production in both groups are presented in Table2. Cyclophosphamide Cl in children receiving fluconazole was lower than that in controls (2.4(0.71) versus 4.2(1.2) l/h/m²; p = .001; Fig.1.). This difference persisted despite using the analysis of covariance to correct for any possible effect of administered dose or patient age on Cl (p = .02 and p = .007, respectively). Children receiving fluconazole also exhibited a longer half-life (6.7(4.9) versus 2.2 (0.49) h; p = .003). There was no accompanying difference in volume of distribution. Fluconazole did not influence the production of either CX or DCCP.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Cyclophosphamide clearance in children receiving fluconazole (5 mg/kg/day) and control patients (p = .02 using the analysis of covariance to adjust for any effect of administered dose).

Human Liver Microsomal Incubations.

The rate of 4-hydroxycyclophosphamide production was linear during the 45-min incubation period and was not affected by the presence of mesna (data not presented). The median IC₅₀ of fluconazole was 38 μM (range 9–80 μM; Fig.2.). No correlation was seen between the extent of inhibition and CYP2A6, CYP2C, CYP2C9, or CYP3A4 activity. Similarly, there was no correlation between the expression of either CYP2C8 or CYP3A4 as measured by immunoblotting and the inhibition of cyclophosphamide 4-hydroxylation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

The inhibition by fluconazole (μM) of the production of 4-hydroxycyclophosphamide by six human liver microsomes. Median IC₅₀ = 38 (range 9–80 μM).

Discussion

The results of this investigation suggest that treatment with fluconazole may inhibit the P-450s involved in cyclophosphamide metabolism. In vivo studies demonstrate a lower plasma Cl of cyclophosphamide in children receiving fluconazole. Although the two groups of patients studied were heterogenous in respect of their clinical condition, prior cytotoxic therapy, and accompanying medication, these variables were unlikely to influence the results in the systematic fashion observed here. Previous studies have demonstrated that the pharmacokinetics of cyclophosphamide remain linear across the doses administered in this investigation; thus, differences between the two groups of children are unlikely to result from saturation of elimination pathway(s) (Wilkinson et al., 1983;Busse et al., 1997). Further confirmation that the observed effect of fluconazole on the Cl of cyclophosphamide was not solely the result of variation in administered dose was provided using the analysis of covariance.

Because of the relationship between cyclophosphamide dose and hemorrhagic cystitis the use of mesna as a uroprotectant occurred almost exclusively in children receiving fluconazole. Children receiving mesna concurrently with cyclophosphamide were treated with equivalent doses of the two drugs (James et al., 1987). Our study found neither evidence of any effect of mesna upon the pharmacokinetics of cyclophosphamide nor interference with 4-hydroxycyclophosphamide production by human liver microsomes upon the addition of mesna. Although mesna forms reversible adducts with circulating 4-hydroxycyclophosphamide, it does not influence the elimination kinetics of cyclophosphamide itself (Shaw and Graham, 1987). Thus it is unlikely that preferential administration of mesna contributed to the observed differences in cyclophosphamide metabolism.

Several patients in both study groups were receiving treatment with oral trimethoprim-sulfamethoxazole (240–480 mg twice daily, administered 3 days per week) as prophylaxis against respiratory infections. Previous reports have demonstrated that trimethoprim-sulfamethoxazole potentiates the clinical effects ofS-warfarin and tolbutamide, probably by reducing the activity of CYP2C9 (O’Reilly, 1980; Wing and Miners, 1985; Kunze et al., 1996). We found no significant effect of trimethoprim-sulfamethoxazole on cyclophosphamide metabolism in this study. This may reflect the relatively small doses of the antibiotic administered in this setting.

At this time it is uncertain whether the 4-hydroxylation andN-dechloroethylation of cyclophosphamide are catalyzed by identical P-450s in vivo. Distinct isoforms are responsible for the production of these metabolites in animals (Yu and Waxman, 1996). DCCP production by human liver microsomes is dependent upon CYP3A4 activity and relatively unaffected by any contribution from the CYP2C family (Ren et al., 1997). The absence of any effect of fluconazole on the production of DCCP in this study also suggests that these reactions are catalyzed by different P-450s. This result should be interpreted with caution because of the difficulty in measuring plasma concentrations of DCCP in vivo. Circulating concentrations of this metabolite were below the lower limit of detection in 44% of children receiving fluconazole. As expected, the production of CX from aldophosphamide by aldehyde dehydrogenase was unaffected by fluconazole (Dockham et al., 1992).

Fluconazole is thought to prevent the replication of C sp by binding to the haem moiety of fungal CYP450 liters1A1 (Lanosterol 14 α-demethylase), thereby reducing the production of ergosterol, an essential component of the fungal cell wall (Hitchcock et al., 1990). Although a less potent inhibitor of CYP450 enzymes than imidazole antifungals such as ketoconazole (Ervine et al., 1996), fluconazole inhibits the metabolism of several CYP2C and CYP3A4 substrates in humans. It is likely that irreversible binding of the triazole ring structure of fluconazole to P-450s from the CYP3A family, and particularly CYP2C9, is responsible for the reduction in cyclophosphamide Cl observed in this study (Kunze et al., 1996).

Two studies have described the pharmacokinetics of fluconazole in children. The first of these reported peak plasma concentrations of 20 to 30 μM after an initial dose of 4 mg/kg. Steady state concentrations, achieved after 4 days of continuous therapy, were approximately 1.6 times greater than the initial peak value (Lee et al., 1992). A second pediatric study reported peak concentrations of 30 to 60 μM after a single dose of 6 mg/kg delivered as a 1-h infusion (Seay et al., 1995). It is likely that only the unbound proportion of circulating fluconazole is available to interact with P-450s in vivo. The plasma-protein binding of fluconazole is low, representing only 11% of total plasma concentrations (Debruyne and Ryckelynck, 1993). Thus, the peak free plasma concentration of fluconazole obtained during the present study lies between 20 and 50 μM. These concentrations are of a similar order to the IC₅₀ values obtained from human liver microsomal studies (9–80 mM). The similarity between these values suggests that significant inhibition of cyclophosphamide metabolism by fluconazole may occur in vivo. Such an effect may be particularly marked in patients with pre-existing renal impairment in whom peak fluconazole concentrations are greater (Lee et al., 1992).

In this study IC_50s were not directly related to the activity of individual P-450s. The absence of a direct correlation may result from the incomplete characterization of the human liver microsomes used or the small number of microsomes tested, but is more likely to reflect the complexity of cyclophosphamide metabolism in humans. Inhibition of a single or a small number of P-450s by fluconazole is likely to have different effects in different microsomes depending upon the level of expression of the affected isoforms and of other unaffected P-450s capable of catalyzing the 4-hydroxylation of cyclophosphamide. Such variation obscured any correlation between the extent of inhibition and the expression of individual P-450s in this study.

We were unable to compare the clinical consequences of the inhibition of cyclophosphamide metabolism by fluconazole in terms of drug toxicity and treatment efficacy because of differences in the underlying malignancies within the study population. Such an analysis is also complicated by the widespread use of multiagent chemotherapy regimens. The conclusions of our investigation are weakened by the fact that they are derived from a retrospective series of patients rather than a randomized study. Further difficulties in the interpretation of our results are provided by the use of a wide range of doses of cyclophosphamide in addition to many other drugs given concurrently. This situation, however, accurately reflects the nature of pediatric oncology patients making pure randomized controlled studies difficult to perform.

Because of the complex nature of cyclophosphamide metabolism the clinical consequences of a reduction in clearance is uncertain. Our results illustrate the need for further studies of the effect of fluconazole on 4-hydroxycyclophosphamide production in vivo to determine whether this interaction reduces the therapeutic effect of cyclophosphamide in clinical practice