Clinical pharmacology of ifosfamide and metabolites

Abstract

Introduction Ifosfamide is an alkylating agent, which is used in the treatment of various types of malignant diseases in adults and childeren. Its use, however, can be accompanied by severe haematological, neuro- and nephrotoxicities. Since its development in the middle of the 1960’s, most of its extensive metabolism has been elucidated. Ifosfamide is a prodrug, which needs to undergo activation by cytochrome P450-3A4 (CYP3A4) to 4-hydroxyifosfamide. Intracellular spontaneous decomposition of 4-hydroxyifosfamide yields the ultimate alkylating metabolite ifosforamide mustard. Ifosforamide mustard binds to DNA causing cross-links followed by apoptosis. Ifosfamide is deactivated to the non-cytotoxic metabolites 2- and 3-dechloroethylifosfamide, yielding an equimolar amount of neurotoxic chloroacetaldehyde. Ifosfamide metabolism is autoinducible. In chapter 1 a review of the literature is given addressing key issues in the clinical pharmacology of ifosfamide and its metabolites. Understanding the relationship between plasma concentrations (pharmacokinetics) and effect (pharmacodynamics) is important for the rational development and the safe and effective use of every therapeutic agent. Since the efficacy and specific toxicities of the treatment with ifosfamide may be linked to its extensive metabolism, pharmacokinetic assessment of ifosfamide and its metabolites is of special interest in helping to explain the unpredictable chances of success and failure in ifosfamide treatment. To characterize the pharmacokinetic-pharmacodynamic relationships and their variability adequately, studies in a representative population using a relatively large number of patients, are needed. However, for practical and ethical reasons extensive pharmacokinetic and pharmacodynamic studies in a large number of often critically ill patients are not possible. Therefore, there was a need for an approach with the help of which description of the relevant pharmacokinetic and pharmacodynamic relationships and of their variability is possible on the basis of sparse data (few data-points per patient) collected under unbalanced designs. Such a population approach will also allow therapeutic drug monitoring of ifosfamide, which can help in further tailoring ifosfamide treatment to the specific needs of the individual patient. The aims of the research described in this thesis were to develop bioanalytical methods for determination of ifosfamide and its metabolites, to build population pharmacokinetic models for these compounds and to apply thee techniques in various clinical studies on ifosfamide. Bioanalysis of ifosfamide and metabolites Assessment of the pharmacokinetics of ifosfamide and metabolites has long been impaired by the lack of reliable bioanalytical assays. The availability of high through-put bioanalytical methods for all relevant metabolites of ifosfamide is a prerequisite for the pharmacological evaluation of the drug in cancer patients. In chapter 2 several new analytical methods are described for the determination of ifosfamide and its main metabolites. Gas chromatography was used to determine ifosfamide and its deactivated metabolites (chapter 2.1). Knowledge of the pharmacokinetics of the deactivated metabolites 2- and 3-dechloroethylifosfamide is crucial for insight into the risk of central neurotoxicity during ifosfamide treatment. In a comparison between nitrogen-phosphorous detection and positive ion electron-impact ion-trap mass spectrometry, the former proved to be superior in sensitivity, accuracy and precision. The main activated metabolites 4-hydroxyifosfamide (chapter 2.2) and ifosforamide mustard (chapter 2.3) were determined using two high-performance liquid chromatography assays. The limitation of the highly unstable character of these metabolites was addressed by derivatization creating stabile, UV-absorbing derivatives. All analytical methods proved to be specific, sensitive, accurate and precise, and could be employed in the analysis of patient samples in a hospital setting. The choice of the matrix for the pharmacokinetic assessment might be of importance, as literature previously indicated possible differences in metabolite distribution between plasma and erythrocytes. Consequently, drug concentration-time profiles in whole blood and plasma could differ, thereby yielding different values for the pharmacokinetic parameters. These data led us to investigate the in vitro and in vivo distribution of ifosfamide and its metabolites. These studies indicated that ifosfamide and its metabolites rapidly reach distribution equilibrium between erythrocytes and plasma, with ifosforamide mustard being the slowest (chapter 2.4). A strong parallelism in the erythrocyte and plasma concentration profiles was observed for all compounds, which indicates that no differences will arise in the assessment of pharmacokinetic parameters using either matrix. Thus, pharmacokinetic assessment using only plasma sampling can yield direct, accurate and relevant relationships with efficacy and toxicity of ifosfamide treated patients. Development of population pharmacokinetic models The development of improved bioanalytical assays allowed extensive pharmacokinetic assessment of ifosfamide and its metabolites. Pharmacokinetic assessment can identify key issues like population differences in pharmacokinetic parameters and the influence of dose and schedule of administration. A review of clinical pharmacokinetic studies on ifosfamide demonstrates that in most studies autoinduction has been observed (chapter 1). Although the mechanism of autoinduction is currently not completely understood, this phenomenon needed to be taken into account when establishing a pharmacokinetic model for ifosfamide and its metabolites. Development of these models was accelerated by using rich data populations. In a population of 15 patients with soft tissue sarcoma who received a 72-hour continuous infusion of ifosfamide, a non-linear population pharmacokinetic model was developed using non-linear mixed effect modelling (NONMEM). This model allowed quantification of the effect of autoinduction on the concentration-time profiles of ifosfamide with correlations between the ifosfamide plasma concentrations and the extent of the autoinduction (chapter 3.1). In order to determine whether this population pharmacokinetic model was adequate in describing the pharmacokinetics of ifosfamide, a comparison was made with two other structural models: one without autoinduction and one with a time-dependent development of autoinduction of ifosfamide (chapter 3.2). The comparison was made in a population of 14 patients with small cell lung cancer who received a 1-hour intravenous infusion of ifosfamide daily for one or two days in combination with paclitaxel and carboplatin. Again, the model presented in chapter 3.1 described the concentration-time profiles of ifosfamide best. The Bayesian estimates of the pharmacokinetic parameters were used to describe the pharmacokinetics of 2- and 3-dechloroethylifosfamide and 4-hydroxyifosfamide. Dose-fractionation over two days compared to one day resulted in increased metabolite formation, especially for 2-dechloroethylifosfamide, probably due to increased autoinduction. This effect of schedule was investigated further in a population pharmacokinetic study of all quantifiable analytes (chapter 2) in 56 patients, who were divided in three groups according to the length of ifosfamide infusion (chapter 3.3). The rate by which the autoinduction developed and the fractions metabolized to 2- and 3-dechloroethyl-ifosfamide were found to be significantly dependent on the infusion schedule. The observed differences in the parameters were, however, comparable to their interindividual variability and were, therefore, considered to be of minor clinical importance. Autoinduction caused a less than proportional increase in the area under the ifosfamide plasma concentration-time curve (measurement of exposure) and more than proportional increase in metabolite exposure with increasing ifosfamide dose. This study demonstrated that the duration of ifosfamide infusion influences the exposure to the parent and its metabolite 3-dechloroethylifosfamide. The observed dose and infusion duration dependency should be taken into account when the pharmacodynamics of different infusion schedules are evaluated. Clinical applications The developed pharmacokinetic population models (chapter 3) were valuable tools in the determination of the pharmacokinetic profiles of ifosfamide and its metabolites in various clinical studies on ifosfamide (chapter 4). Responses to ifosfamide treatment and toxicities vary to a great degree among patients. This variability might be explained by differences in pharmacokinetics. Characterising covariates which contribute to the variation in the pharmacokinetics is therefore of paramount importance. Therefore, the pharmacokinetics, relations between the pharmacokinetics and covariates and pharmacodynamics of ifosfamide and 2- and 3-dechloroethylifosfamide and 4-hydroxyifosfamide were assessed in a population of 20 patients with soft tissue sarcoma, who received ifosfamide administered as a 72-hour continuous intravenous infusion (chapter 4.1). The population pharmacokinetic model (according to chapter 3.1) was built in a sequential manner, starting with a covariate-free model and progressing to a covariate model with the aid of generalized additive modelling. The addition of the covariates weight, body surface area, albumin, serum creatinine, serum urea, alkaline phosphatase and lactate dehydrogenase improved the prediction errors of the model. Significant pharmacokinetic-pharmacodynamic relationships were observed between the exposure to 2- and 3-dechloroethylifosfamide and orientational disorder, a neurotoxic side-effect. No pharmacokinetic-pharmacodynamic relationships between exposure to 4-hydroxyifosfamide and haematological toxicities could be observed in this population. Children are a special population within the field of oncology. Only limited information is available on the pharmacokinetics of ifosfamide and its metabolites in paediatric patients, due to the ethical problems involved in these studies. We assessed the feasibility of a sparse data approach for the determination of the population pharmacokinetics of ifosfamide, 2- and 3-dechloroethylifosfamide and 4-hydroxyifosfamide in 32 children treated with various schedules of single agent ifosfamide therapy against various malignant tumours (chapter 4.2). A non-linear population pharmacokinetic model with linear development of autoinduction was implemented. In chapter 3.2 this model proved less accurate in the description of the ifosfamide concentration-time profiles than the model presented in chapters 3.1 and 3.3. However, the more elaborate ifosfamide concentration dependent autoinduction model could not be applied in this sparse data population. Cross-validation by bootstrapping the data indicated accurate description of the population pharmacokinetic parameter without bias. Specific isoenzymes are responsible for ifosfamide metabolism; activation is mediated by CYP3A4 and deactivation by CYP3A4 and CYP2B6. Therefore, modulation of the metabolism of ifosfamide may lead to an improved efficacy/toxicity ratio. Modulation was investigated by co-administrating ketoconazole and rifampicin, a potent inhibitor of CYP3A4 and inducer of CYP3A4/2B6, respectively (chapter 4.3). In a double randomized, two-way cross-over study a total of 16 patients received ifosfamide either alone or in combination with ketoconazole or rifampicin. Pharmacokinetics were assessed using the models developed in chapter 3. Rifampicin increased metabolism of ifosfamide without specifically favouring the activation or deactivation route of ifosfamide. Ketoconazole decreased activation to 4-hydroxyifosfamide. Thus, disappointingly this pharmacokinetic study indicated that no therapeutic benefit may be gained by modulation of ifosfamide therapy with rifampicin or ketoconazole in humans. Commonly, chemotherapeutic agents are applied in combination-schedules in order to maximize the efficacy. The novel combination of ifosfamide and the topoisomerase I inhibitor topotecan was investigated in a phase I trial (chapter 4.4). Preliminary results indicate that the combination of topotecan administered as a 30-minute infusion daily times 3 with ifosfamide administered as a 1-hour infusion daily times 3 every three weeks to patients with advanced malignancies was feasible. Haematological toxicities were dose limiting. The maximum tolerated dose of topotecan has not yet been reached. The pharmacokinetics of topotecan and ifosfamide and its metabolites were similar to those observed after single agent administration suggesting that the drugs did not interact pharmacokinetically. Sigmoidal Emax models could be fit to the relationship between the areas under the plasma concentration-time curve of topotecan lactone and total topotecan, and the decrease in absolute neutrophil count. Partial responses were documented in three patients with ovarian cancer. Possible clinical benefit of this combination needs to be evaluated in phase II/III studies. Perspectives and final remarks Although ifosfamide has successfully been used for over 30 years in the treatment of various malignant diseases, there is still a need for understanding its variability in success and failure of treatment. Now that satisfactory bioanalytical methods and population pharmacokinetic models have been developed, the underlying mechanisms of this varibility may be better understood and a superior ifosfamide therapy may be developed, tailored to dosing-requirements of the individual patient. Furthermore, as individuals differ from each other in their concentration-effect relationships, the implication of schedule dependence is that varying the administration pattern, for example the infusion duration, between individuals may prove to be as important as individualising the dose-size of ifosfamide. One of the remaining issues in the clinical pharmacology of ifosfamide is the firm establishment of pharmacokinetic-pharmacodynamic relationships that provide enough information for precise predictions of the clinical outcome of therapy based on the pharmacokinetics of ifosfamide. Only if this requirement is met the further development of therapeutic drug monitoring and dose individualization of ifosfamide treatment will be possible. This thesis is the first step in that direction, but many more pieces of the puzzle need to be put together to achieve complete understanding of the extent and limitations of the clinical activity and toxicity of ifosfamide therapy.