Introduction

Top Abstract Introduction Patients and Methods Results Discussion References

Ifosfamide (Holoxan, Ifex) is an alkylating agent, which has been proven to be active against a number of solid tumors and hematological malignancies in adults and children. Currently, ifosfamide is mostly used in combination with other anticancer drugs, but also as single agent. Many different infusion schedules are in use, e.g., continuous infusion over either 24, 72, or 240 h with doses of 5, 9, or 7.5-15 g/m² ifosfamide, respectively (Kaijser et al., 1996; Nielsen et al., 2000).

Ifosfamide is a prodrug, which needs activation by cytochrome P450-3A4 (CYP3A4¹) to 4-hydroxyifosfamide, as depicted in Fig. 1. Spontaneous decomposition of 4-hydroxyifosfamide yields the ultimate alkylating metabolite ifosforamide mustard and acrolein (Kerbusch et al., 2001). The alkylating activity of ifosforamide mustard is responsible for both the antitumor activity and hematological toxicity. Acrolein causes hemorrhagic cystitis, which can generally be prevented by mesna coadministration. Ifosfamide is deactivated to the noncytotoxic metabolites 2- and 3-dechloroethylifosfamide. Each dechloroethylation reaction yields an equimolar amount of chloracetaldehyde, which is held responsible for neurotoxicity observed in about 10% of all patients receiving conventional single-agent dosing of ifosfamide (Cerny and Küpfer, 1992). In addition, observed renal tubular abnormalities may also be correlated with the formation of chloracetaldehyde (Kerbusch et al., 2001). Chloracetaldehyde is very unstable. Therefore, assessment of 2- and 3-dechloroethylifosfamide exposure is preferred when quantifying the relationship between the pharmacokinetics and neurotoxicity after ifosfamide infusion.

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Metabolism of ifosfamide.

Ifosfamide is subject to autoinduction, which leads to an increase in metabolism of ifosfamide over time (Kerbusch et al., 2001). Several comparative studies observed that fractionation of the ifosfamide dose leads to increased autoinduction (Lewis et al., 1990; Kerbusch et al., 2000b). Pharmacokinetic studies can provide further insight into the effect of the infusion schedule on the autoinduction of ifosfamide metabolism. Aim of this study is to quantify and to compare the pharmacokinetics of ifosfamide, 2- and 3-dechloroethylifosfamide, 4-hydroxyifosfamide, and ifosforamide mustard in short (1-4 h), medium (24-72 h), and long infusion durations (96-240 h) of ifosfamide. Assessment of pharmacokinetics will indicate whether the development of autoinduction is dependent on the dose and/or infusion duration, and may further aid in the design of ifosfamide infusion schedules.

	Patients and Methods

Top Abstract Introduction Patients and Methods Results Discussion References

Eligibility Criteria. Thirty-three patients received ifosfamide (Holoxan, ASTA Medica, Diemen, The Netherlands) as i.v. infusion as part of the regular therapy with ifosfamide as single agent or as combination chemotherapy with cisplatin, carboplatin, etoposide, Adriamycin, vincristine, or methotrexate, during the period of March 1997 and April 2000. None of these anticancer agents are known to induce CYP-mediated metabolism or are known to pharmacokinetically interact with ifosfamide. Typical infusion schedules used for ifosfamide were 1.2 g/m² once daily over 1, 3, or 4 h for three consecutive days, 1.5 g/m² once daily in 22.5 h for two consecutive days, 5 g/m² in 24 h once, and 9 g/m² by means of a 72-h continuous infusion. The study protocol was approved by the Ethics Board of the Antoni van Leeuwenhoek Hospital/The Netherlands Cancer Institute. Another group of 23 patients received a 10-day continuous infusion of single-agent ifosfamide using an ambulatory pump. This group was treated at the Leiden University Medical Center, in a study that was approved by the Ethics Board of the hospital (Kaijser et al., 1996). Supportive care consisted of mesna and extensive hydration to prevent hemorrhagic cystitis and bicarbonate to prevent acidosis. Standard 5-hydroxytryptamine₃ blockers as antiemetics and methylene blue (a neurotoxicity antidote) were given when indicated (Cerny and Küpfer, 1992). Pharmacokinetic sampling was performed after written informed consent was obtained from the patients.

Pharmacokinetic Sampling. Blood samples were drawn before the start of, during, and after the end of the infusion at selected time points, depending on the infusion schedule. Ten milliliters of whole blood was collected in a lithium heparin-coated Vacutainer (Becton Dickinson, Plymouth, UK) and placed in ice water. The plasma was immediately separated by centrifugation at 1000g for 5 min at 4°C. The plasma was aliquoted in 4 volumes of which three were precisely 1-ml volumes. To two 1-ml volumes, 100 µl of 2 M semicarbazide solution, pH 7.4, was added to stabilize 4-hydroxyifosfamide. To another 1-ml volume, 100 µl of 1 M sodium chloride, 2 M semicarbazide solution, pH 8.0, was added to prevent 4-hydroxyifosfamide degradation to ifosforamide mustard and to stabilize the latter. The remaining plasma was used for ifosfamide, 2- and 3-dechloroethylifosfamide analysis. Urine was collected from the start of the (first) infusion until 24 h after the end of the (last) ifosfamide infusion and was analyzed for ifosfamide, 2- and 3-dechloroethylifosfamide. Both plasma and urine samples were stored at 70°C, pending analysis.

Bioanalysis. Gas chromatography with selective nitrogen-phosphorus detection was used for the determination of ifosfamide, 2- and 3-dechloroethylifosfamide in urine and plasma (Kerbusch et al., 2000c). Sample pretreatment consisted of alkalinized liquid-liquid extraction with ethyl acetate, transfer of the organic extract to another tube, evaporation to dryness, and subsequent reconstitution in ethyl acetate. This method was validated and proved to be specific, sensitive, accurate (93.3-104.5%), and precise (within and between day <5.5%) within the concentration range of 0.192 to 383 µM, with a lower limit of quantification (LLQ) of 0.192 µM for ifosfamide, 2- and 3-dechloroethylifosfamide.

Data Evaluation. Pharmacokinetic models were fitted to the plasma concentration data from the individuals using the population pharmacokinetic program NONMEM (Nonlinear Mixed Effects Modeling, version V 1.1, double precision, first order estimation) (Beal et al., 1992; Boeckman et al., 1994). The population pharmacokinetic models resulted in model predictions and Bayesian individual predictions for the concentrations of ifosfamide and metabolites (Sheiner and Grasela, 1991). The nonlinear pharmacokinetics of ifosfamide were described using a recently developed model, which incorporated the development of autoinduction (Kerbusch et al., 2000b). According to this recent study the time-dependent pharmacokinetics of ifosfamide can be described by a one-compartment model (Fig. 2). The change of the amount of ifosfamide (A_ifo, µmol) in the central compartment over time is described by eq. 1:

(1)

in which T_inf (h) is the infusion duration, R (µmol · h¹) is the infusion rate of ifosfamide, CL (liters · h¹) the ifosfamide clearance, which changes over time, and V_ifo (liters) the volume of distribution. The predicted concentration of ifosfamide C_ifo (µM) is the ratio of A_ifo and V_ifo. The development of autoinduction is described using a hypothetical enzyme compartment. CL is expressed as initial clearance (CL_ini,ifo, liters · h¹) multiplied by the relative amount of enzyme (A_enz) in the hypothetical enzyme compartment, as given in eq. 2:

(2)

The change of A_enz (a relative measure with no dimension) over time in the enzyme compartment is dependent on C_ifo as follows:

(3)

in which K_enz,out (h¹) is the first order rate constant for enzyme degradation/inactivation and IC₅₀ (µM) is the ifosfamide concentration at 50% of the maximum inhibition of enzyme degradation. At t = 0 and A_enz = 1, the enzyme formation rate (K_enz,in) is equal to K_enz,out (Fig. 2). The induction half-life of the enzyme (t_1/2enz, h) was calculated by the ratio of ln(2) and K_enz,out.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Pharmacokinetic model for ifosfamide metabolism describing the autoinducible pharmacokinetics of ifosfamide (IFO), 2-dechloroethylifosfamide (2DCE), 3-dechloroethylifosfamide (3DCE), 4-hydroxyifosfamide (4OHIF), and ifosforamide mustard (IFM). Autoinduction is modeled with a hypothetical enzyme compartment described by an enzyme formation rate (Kenz,in) and elimination rate (Kenz,out). The amount of enzyme increases the clearance of IFO (CLIFO). The amount of IFO, described by the IFO concentration and IFO volume of distribution (VIFO), inhibits Kenz,out. Metabolite compartments are described by metabolite formation (F) and elimination rates (K).

(4)

1

1

(5)

1

(6)

(7)

	Results

Top Abstract Introduction Patients and Methods Results Discussion References

A total of 56 patients was divided into three subpopulations with 12 patients receiving short (1-4 h), 21 patients receiving medium (24-72 h), and 23 patients receiving long (96-240 h) infusion durations of ifosfamide (Table 1). The number of patients in which 4-hydroxyifosfamide and ifosforamide mustard were analyzed was insufficient for the separate estimation of the pharmacokinetics of these metabolites for each infusion schedule. The pharmacokinetic population parameters with their standard error and interindividual variability and the residual variability of ifosfamide and 2- and 3-dechloroethylifosfamide are presented in Table 2. The corresponding estimates of 4-hydroxyifosfamide and ifosforamide mustard are listed in Table 3.

The estimates of most pharmacokinetic parameters were not influenced by the duration of infusion. Only minor differences were found for K_enz,out, F* of 2-dechloroethylifosfamide (F*_2dce), and 3-dechloroethylifosfamide (F*_3dce). K_enz,out during long infusion durations was 52% (^long = 0.48) lower in comparison with short and medium infusion durations. F*_2dce and F*_3dce decreased with 51% (^medium = 0.487) and 33% (^medium = 0.666), respectively, after medium infusion durations when compared with short and long infusion durations. No outliers or trends could be observed in the plots of the population predictions and individual Bayesian predictions of the concentrations versus the observed concentrations (plots not shown). The observed and individual Bayesian estimated pharmacokinetic profiles of ifosfamide of all patients are presented in Fig. 3.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Bayesian individual predicted plasma concentration-time profiles of ifosfamide (IFO) for all patients during short (1-4 h), medium (24-72 h), and long (96-240 h) infusion durations. The observations are represented by the dots.

The urinary excretion of unchanged ifosfamide and its dechloroethylated metabolites are presented in Table 4. Increase of infusion duration resulted in a significant decrease of urinary recovery of unchanged ifosfamide and the sum of ifosfamide and metabolites. For unchanged ifosfamide the urinary recovery after short and long infusion durations was 21.2 ± 6.6 and 8.6 ± 2.6%, respectively. No clear trend was observed for 2- and 3-dechloroethylifosfamide, although urine recovery of 2-dechloroethylifosfamide during long infusions was slightly lower than during medium infusion duration.

The individual profiles of the ifosfamide clearances over time of each infusion schedule are depicted in Fig. 4. Modest interindividual variability was observed. The clearance of ifosfamide increased during the infusion and decreased after cessation. Notably, during the first 24 h for all patients the model predicted an average doubling of the clearance.

View larger version (25K): [in this window] [in a new window]   Fig. 4.   Individual time profiles of the clearance of ifosfamide (IFO) for all patients. The profiles were obtained on the basis of the individual Bayesian parameter estimates based on the total population.

Figure 5 depicts the relationships between the AUCs of ifosfamide and its metabolites and the ifosfamide dose for the total population. The AUC of ifosfamide increased less than proportional with increase in absolute ifosfamide dose. In contrast, the AUCs of the metabolites increased more than proportional with increase in absolute ifosfamide dose.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   From top left to bottom right: relationships (with Loess-smooth) between the ifosfamide dose (mmol) and ifosfamide (IFO), 4-hydroxyifosfamide (4OHIF), ifosforamide mustard (IFM), 2-dechloroethylifosfamide (2DCE), and 3-dechloroethylifosfamide (3DCE) exposures as calculated by the AUC of each patient.

The dose-corrected AUCs are depicted in Fig. 6. The dose-corrected exposure to ifosfamide during long infusions was decreased 26% (p < 0.001) and 25% (p < 0.05) compared with short and medium infusion duration, respectively. In comparison with medium infusion durations dose-corrected exposure to 3-dechloroethylifosfamide during long infusion durations was increased with 45% (p < 0.01).

View larger version (19K): [in this window] [in a new window]   Fig. 6.   From top left to bottom right: relationships between the ifosfamide infusion duration (h) and ifosfamide (IFO), 4-hydroxyifosfamide (4OHIF), ifosforamide mustard (IFM), 2-dechloroethylifosfamide (2DCE), and 3-dechloroethylifosfamide (3DCE) dose-standardized exposures as calculated by the ratio of the AUC and the ifosfamide dose of each patient. The bar indicates the mean. *, significantly decreased 26% (p < 0.001) and 25% (p < 0.05), respectively, compared with 1-4- and 24-72-h infusions. **, significantly increased 45% (p < 0.01) compared with 24-72-h infusions (Kolmogorov-Smirnov).

	Discussion

Top Abstract Introduction Patients and Methods Results Discussion References

This study investigated the effect of the duration of the ifosfamide infusion on the pharmacokinetics of ifosfamide and its metabolites. In a review on the clinical pharmacokinetics of ifosfamide in 1994 it was proposed that fractionation of the ifosfamide dose increased the metabolism of ifosfamide (Wagner, 1994). However, more recent studies have indicated that the ifosfamide infusion schedule does not influence the degree of metabolism. In a comparative study in pediatric patients receiving 3 g/m² ifosfamide administered as a 1-h infusion for three consecutive days or 9 g/m² ifosfamide administered as a 72-h continuous infusion, no major differences in exposure were observed, except a minor decrease in exposure to deactivated metabolites with the bolus infusion (Boddy et al., 1995). In addition, no differences in metabolic exposures were found in a similar study in adults (Singer et al., 1998). Singer and coworkers concluded that there is no identifiable pharmacokinetic basis for insistence on either bolus or infusional methods of ifosfamide administration.

In our study a group of 56 patients was divided into three subpopulations receiving either short (1-4 h), medium (24-72 h), or long (96-240 h) infusion durations with ifosfamide. All patients studied received ifosfamide as part of standard therapy for the treatment of various malignancies. A previously developed autoinduction model was applied to assess the population pharmacokinetics of ifosfamide and metabolites (Kerbusch et al., 2000b). This model describes the development of autoinduction with a relationship between the ifosfamide plasma concentration and a hypothetical, dynamic enzyme compartment, which is responsible for ifosfamide metabolism. Thus, clearance of ifosfamide over time changes with ifosfamide concentrations. From some of the patients included in the study only a limited number of blood samples were available for pharmacokinetic analysis. In such cases conventional noncompartmental analysis of the concentration-time profiles does not allow accurate estimation of the pharmacokinetic parameters, due to insufficient data. The applied population approach with Bayesian estimation, on the other hand, does not have these drawbacks.

The influence of the ifosfamide concentration on the enzyme amount was modeled as an inhibiting influence on the degradation of the enzyme. Ifosfamide has been shown to reduce protein synthesis and could have a similar effect on the enzymes responsible for CYP3A4 degradation/inactivation, thus increasing ifosfamide clearance (Lewis et al., 1990). A similar mechanism of inhibition was proposed for troleandomycin, which did not increase protein synthesis but inhibited catalytic enzymes responsible for CYP degradation (Watkins et al., 1986). However, increase of enzyme synthesis through stabilization of mRNA was suggested for cyclophosphamide (Hassan et al. 1999) and ifosfamide (Chang et al., 1997). Whatever the mechanism, all modeling approaches will result in an increased amount of enzyme (Hassan et al. 1999). We also included the interaction term on the synthesis of the enzyme and an identical fit to the concentration-time data of ifosfamide was achieved. Thus, modeling autoinduction as increased enzyme translation/transcription or as decreased enzyme degradation will yield similar conclusions.

Statistically significant differences were found for K_enz,out between long and short infusion durations and for F*_2dce and F*_3dce between medium and short infusion durations. These differences in parameter estimates were considered to be minor (33-52%) and not clinically relevant, since they were in the same order of magnitude as the interindividual variability of the respective parameters (22-43%). The observed differences may be explained by different amounts of data (information) within the concentration-time profiles of the various infusion schedules, which are necessary for the estimation of these pharmacokinetic parameters. Estimation of K_enz,out in long infusion durations is more feasible than in short infusion durations, because the effect of autoinduction can be estimated over a longer period of time (Fig. 4). Nevertheless, the derived half-life of the enzyme (t_1/2enz) increased from 8.5 h with short duration infusions to 17.7 with long duration infusions. Thus, in the latter group maximum induction would be reached after approximately 88 h (5 × t_1/2enz,long), whereas maximum clearance in the short duration infusions would already be reached after 42 h. This difference in enzymatic behavior can also be observed in Fig. 4. The relatively lower number of samples (on average three per patient) drawn during the first 24 h of medium infusion durations, compared with short infusion durations (on average seven per patient), may have resulted in different estimations of the formation rates of the deactivated metabolites. Sampling frequency during the first 24 h with long infusions was more similar to that of short infusion durations (on average five samples were collected per patient). Hence, no difference was observed in metabolite formation with long infusion durations compared with short.

All other estimates of the population parameters were not significantly influenced by infusion duration. This indicates that no clinically relevant differences were observed in the pharmacokinetics between the different infusion durations.

The population pharmacokinetic parameters were in accordance with a recent study on the pharmacokinetics of ifosfamide after administration of 9 to 12 g/m² ifosfamide in a 72-h continuous intravenous infusion (Kerbusch et al., 2000b). In another study in 13 cancer patients receiving 1.5 g/m² for 5 days ifosfamide clearance and autoinduction were described using a population pharmacokinetic model for the enantiomers of ifosfamide and its metabolites (Di Marco et al., 2000). In that study clearance increased from 4 to 7 and 5 to 10 l/h from day 1 to 5 for R-ifosfamide and S-ifosfamide, respectively (Di Marco et al., 2000). This is in accordance with our findings as presented in Fig. 4. Individual ifosfamide clearances increase from 4 to 7 l/h (range 5-15).

Initially, K values of the metabolites (K_2DCE = 0.385, K_3DCE = 0.124, K_4OHIF = 73.8) were greater than the value calculated for ifosfamide (CL_{ini, ifo}/V_ifo = 0.0656 h¹). This indicates that the observed decay in plasma of the metabolites was formation rate limited. Maximum autoinduction was reached after 42, 42, and 88 h in short, medium, and long duration infusions, respectively. The K value of ifosfamide ranged then approximately between 0.109 and 0.327 h¹. In other words, after autoinduction ifosfamide elimination half-life decreased from 10.5 to 2.1 to 6.4 h and is now slower than the elimination half-life of 2- and 3-dechloroethylifosfamide (2 and 6 h, respectively). Thus, no formation rate limited elimination was observed with the dechloroethylated metabolites after development of autoinduction. Visual inspection of the terminal slopes confirmed this (data not shown). The observed lack of formation rate-limited elimination warranted robust and uncorrelated estimates for these metabolite parameters. In contrast, 4-hydroxyifosfamide still exhibited formation rate-limited elimination. Unfortunately, no long infusion duration data were available for this metabolite, preventing robust parameter estimates.

These findings are in accordance with previous findings. Di Marco et al. (2000) observed the same "flip-flop" phenomenon with formation rate-limited dechloroethylated metabolite elimination at day 1, and elimination rate-limited elimination after the development of autoinduction on day 5.

Previously, Kaijser et al. (1996) calculated metabolite half-lives of 47 and 19 h for 2- and 3-dechloroethylifosfamide, respectively. This corresponds with lower K values of 0.015 and 0.036 h¹, respectively. However, these values were based on the assumption that these metabolites are not metabolized further. This assumption may not be valid because a didechloroethylifosfamide metabolite has been detected in human plasma (Gilard et al., 1993).

Table 4 indicates that considerable amounts of unchanged ifosfamide are excreted in the urine. Separate estimation of renal clearance, however, did not improve the goodness-of-fit of the model. The urinary excretion data found in this study were similar to previous observations. Gilard et al. (1993) reported urinary excretion of 18, 16, and 7% for ifosfamide, 2- and 3-dechloroethylifosfamide, respectively, in patients receiving 3 g/m²/3 h ifosfamide for 3 days. The decrease in excretion of ifosfamide with increasing infusion duration can be explained by the development of the autoinduction. The decrease in excretion of ifosfamide with increasing infusion durations can be explained by decreasing AUCs due to autoinduction. For 2- and 3-dechloroethylifosfamide increased AUCs would theoretically produce increased urine recovery. This was however not observed. The mechanism of this discrepancy remains to be elucidated. Nevertheless, the latter observation is in accordance with the previously reported finding that the urinary excretions of 2- and 3-dechloroethylifosfamide were superimposable for the different infusion schedules in pediatric patients (Boos et al. 1995).

The AUC of ifosfamide increased less than proportional with the absolute dose and the AUCs of the metabolites increased more than proportional with the absolute ifosfamide dose. This can be explained by a greater extent of autoinduction due to higher ifosfamide plasma concentrations (eqs. 2-4). These findings are in accordance with a previous report in which the pharmacokinetic model for ifosfamide was developed (Kerbusch et al., 2000b). The dose-corrected exposures to ifosfamide and its metabolites changed with increasing infusion duration. In comparison with short infusions, a decrease of 25% in dose-corrected ifosfamide exposure and an increase of 45% in 3-dechloroethylifosfamide exposure were observed with long infusion durations. The decrease in ifosfamide exposure can be explained by an increased average clearance during the long infusions (Fig. 4).

As a result of the increased average clearance of ifosfamide during long infusions dose-corrected exposure to the dechloroethylated metabolites was expected to increase. Both isoenzymes CYP2B6 and 3A4 have been indicated in the dechloroethylation of ifosfamide (Roy et al., 1999). Equal changes for both deactivated metabolites were expected. However, the increased average clearance observed with 3-dechloroethylifosfamide was not observed with 2-dechloroethylifosfamide. This contrasting finding may be explained by the observed large interindividual variability that could obscure the expected effect of long infusion durations on the dose-corrected exposure of 2-dechloroethylifosfamide (Fig. 6). Unfortunately, the effect of infusion duration on the dose-corrected exposure to 4-hydroxyifosfamide and ifosforamide mustard could not be evaluated since no samples were available for the long infusions. Nevertheless, the observed increased exposure to the dechloroethylated metabolites during long infusion durations demonstrates the importance of the assessment of the pharmacokinetics when comparing the pharmacodynamics during different infusion durations.

In conclusion, a population pharmacokinetic model for ifosfamide and its metabolites was used to investigate any differences in ifosfamide metabolism due to the duration of the ifosfamide infusion. Estimates of K_enz,out, F*_2dce, and F*_3dce were found to depend on the infusion schedule. The clinical relevancy of the observed differences was, however, considered to be of minor importance. Concentration-dependent development of autoinduction produced a less than proportional increase in ifosfamide exposure and more than proportional increase in metabolite exposure with increasing ifosfamide dose. Schedule dependence could be demonstrated for the dose-corrected exposures to ifosfamide and 3-dechloroethylifosfamide between short and long infusion durations, which could be explained by the increased average clearance during infusion. No change in dose-corrected exposure to ifosfamide and metabolites was observed between short and medium infusion durations in this population. The observed dependence on dose and infusion duration should be taken into account when the pharmacodynamics of different infusion schedules of ifosfamide is evaluated.