Enzyme autoinduction by mitotane supported by population pharmacokinetic modelling in a large cohort of adrenocortical carcinoma patients U Arshad1*, M Taubert1*, M Kurlbaum2, S Frechen1, S Herterich3, F Megerle2, S Hamacher4, M Fassnacht2,3,5, U Fuhr1, M Kroiss2,3,5.
1 Department I of Pharmacology, University Hospital Cologne, Germany
2 Division of Endocrinology and Diabetology, Department of Internal Medicine I, University Hospital, University of Würzburg, Germany
3 Clinical Chemistry and Laboratory Medicine, University Hospital Würzburg, Germany
4 Institute of Medical Statistics and Computational Biology, University of Cologne, Germany
5 Comprehensive Cancer Center Mainfranken, University of Würzburg, Würzburg, Germany
Correspondence: Usman Arshad, Department I of Pharmacology University Hospital of Cologne Gleueler Str. 24 50931 Cologne Germany Phone: +49 221 478 32187, Fax: +49 221 478 7011 EMail: usman.arshad@uk-koeln.de
Running title: Enzyme autoinduction model of mitotane
Key words: mitotane, enzyme autoinduction, adrenocortical carcinoma, pharmacokinetics
*Both authors contributed equally.
Abstract
Objective: Mitotane is used for the treatment of adrenocortical carcinoma. High oral daily doses of typically 1- 6 g are required to attain therapeutic concentrations. The drug has a narrow therapeutic index and patient management is difficult because of a high volume of distribution, very long elimination half-life, and drug interaction through induction of metabolizing enzymes. The present evaluation aimed at the development of a population pharmacokinetic model of mitotane to facilitate therapeutic drug monitoring.
Methods: Appropriate dosing information, plasma concentrations (1137 data points) and covariates were available from therapeutic drug monitoring (TDM) of 76 adrenocortical carcinoma patients treated with mitotane. Using nonlinear mixed effects modeling, a simple structural model was first developed, with subsequent introduction of metabolic autoinduction. Covariate data were analyzed to improve overall model predictability. Simulations were performed to assess the attainment of therapeutic concentrations with clinical dosing schedules.
Results: A one-compartment pharmacokinetic model with first order absorption was found suitable to describe the data, with an estimated central volume of distribution of 6086 L related to a high interindividual variability of 81.5%. Increase in clearance of mitotane during treatment could be modeled by a linear enzyme autoinduction process. Body mass index was found to have an influence upon disposition kinetics of mitotane. Model simulations favor a high dose regimen to rapidly attain therapeutic concentrations, with the first TDM suggested on day 16 of treatment to avoid systemic toxicity.
Conclusion: The proposed model describes mitotane pharmacokinetics and can be used to facilitate therapy by predicting plasma concentrations.
Enzyme autoinduction model of mitotane
1. Introduction
The adrenolytic drug mitotane (1-chloro-2-[2,2-dichloro-1-(4-chlorophenyl) ethyl] benzene) is the only approved treatment of the orphan malignant disease adrenocortical carcinoma (ACC). ACC has a high rate of recurrence after complete tumor resection and a dismal prognosis in advanced stages (1, 2). Mitotane is used both as an adjuvant treatment after complete tumor resection (3, 4) and for palliative treatment of advanced disease (5). Clinically used drug effects of mitotane include reduction of tumor related steroid hormone excess and a direct cytotoxic effect leading to objective treatment response in ~20% of cases (6) which appears to be relatively specific to cells of the adrenal cortex. Several molecular mechanisms for mitotane action appear to contribute to mitotane efficacy (7, 8, 9). We recently found sterol-O-acyl transferase 1 to be inhibited by mitotane which leads to impaired steroidogenesis and lipid induced endoplasmic reticulum stress (10). Published data on the pharmacokinetics of mitotane are scarce and have been conducted in small patient series only. It has been shown that mitotane has a low oral bioavailability (F) of 35-40% (11) and a high volume of distribution which is likely due to its lipophilic nature and extensive accumulation in adipose tissue (12). The majority of mitotane has been found to be bound to lipoprotein particles in circulation with pharmacological activity limited to the unbound fraction (13, 14, 15).
Efficacy of mitotane treatment is associated with plasma concentrations >14 mg/l which could be demonstrated in several retrospective series both in adjuvant and palliative treatment with mitotane monotherapy (6, 16, 17, 18, 19, 20) but also in combination with cytotoxic drugs in advanced disease (21, 22). Adverse effects, including CNS toxicity are associated with plasma concentrations exceeding 20 mg/L (23). However, the time interval to achieve therapeutic plasma concentrations of mitotane limits the clinical utility
Enzyme autoinduction model of mitotane
of the drug regardless of the dosing regimen applied (24, 25, 26). Accordingly, therapeutic drug monitoring (TDM) is needed for continuous treatment evaluation and decision making. There is a poor correlation of mitotane dose with plasma concentrations, which suggests the involvement of other factors influencing the attainment of therapeutic concentrations (27).
Main metabolites of mitotane are o,p’-dichlorodiphenyl-ethene (o,p’-DDE) and -acetate (o,p’-DDA) (28, 29). o,p’-DDA can be detected at ten-fold higher concentration in blood than mitotane itself whereas o,p’-DDE is barely detectable in most cases (19, 30). Small amounts of these derivatives apparently undergo aromatic hydroxylation and glycine conjugation (28). The compound is a strong inducer of hepatic CYP3A4 in vitro and in vivo, which causes interactions with co-administered drugs such as sunitinib (31, 32, 33, 34). Orally administered mitotane is excreted in urine and bile and has a long elimination half-life ranging from 18-159 days (35).
In vitro, drug metabolizing enzymes and transporters beyond CYP3A4 were induced by mitotane, probably via the pregnane X receptor (PXR) (32). PXR ligands transcriptionally induce the activity of a broad range of processes in drug metabolism, which in turn often also accelerate the metabolism of PXR ligands, a phenomenon called autoinduction (36, 37). It is therefore conceivable that mitotane metabolism may be affected by autoinduction, and variability in autoinduction may contribute to differences in clinical toxicity and efficacy among patients. Thus, it might be helpful to account for enzyme induction in order to appropriately describe the pharmacokinetics of mitotane during a long-term treatment. A quantitative description of enzyme induction by mitotane in patients may also be useful to predict drug interactions that may limit the exposure to co- administered chemotherapeutic or targeted agents (3). Previous modeling efforts have not considered autoinduction (38).
Enzyme autoinduction model of mitotane
The objective of the present evaluation is thus to develop a model describing mitotane pharmacokinetics incorporating enzyme autoinduction, which should contribute to optimizing mitotane dosing schedules.
2. Subjects and methods
2.1 Patients characteristics and data preparation
Clinical and demographical data were retrieved from records of patients participating in the German ACC Registry and the European Network for the Study of Adrenal Tumors (ENSAT) at a single reference center. Both registries have been approved by the ethics committee of the University of Würzburg (approval number 86/03 and 88/11) and all patients provided written informed consent to participate in the study. The following parameters were collected: age, sex, weight, height, body mass index, ENSAT tumor stage, treatment intention, concomitant systemic therapy, albumin, triglyceride, high and low-density lipoprotein, cholesterol, creatinine, and y-glutamyltransferase (y-GT) plasma concentrations. Mitotane plasma concentrations were measured within the Lysosafe® TDM provided on behalf of the manufacturer, HRA-Pharma (Paris, France) using HPLC. A total number of 103 patients with adrenocortical carcinoma were treated with oral mitotane doses (0.5-10g per day, with interruptions).
R (version 3.2.3) with ‘dyplyr’, ‘tidyr’ ‘lubridate’ and ‘ggplot2’ packages was used for data manipulation, cleaning and visualization (39, 40, 41, 42). All patients treated with mitotane over the age of 18 were eligible. Only patients with missing dosing at the initiation of therapy were excluded from analysis. Data from patients with missing dosing information during the treatment course was excluded partly by evaluating only data
Enzyme autoinduction model of mitotane
gathered during the period prior up to the missing information. Exploratory data analysis was performed to judge general trends in the data.
2.2 Data analysis and pharmacokinetic model development
Nonlinear mixed effect modeling was performed for data analysis. Estimation of pharmacokinetic parameters was performed by first order conditional estimation with interaction (FOCE-I) using NONMEM 7.4.1 (ICON, Development Solutions, Elliot City, MD, USA) (43). Model development was aided by Pearl-speaks-NONMEM toolkit (Version 4.7.0) (44). Graphical user interface Pirana (Version 2.9.6) was used for model management and execution, output generation and interpretation of results (45). Xpose4 package with R was used for visualizing output data, post processing and analyzing NONMEM output (46).
A compartmental approach was adapted in a stepwise manner to develop a pharmacokinetic model. In the first step, pharmacokinetic parameters representing a typical individual of the population were estimated using the basic structural model, followed by estimation of interindividual variability (IIV). Subsequently, a hypothetical mitotane metabolizing enzyme compartment was introduced.
Change in amount of drug in the central compartment (AD.cent) was described by equation 1.
dt
dAD,cent / + = ka . AD,gut - Cp . CLbaseline . Aenz (1)
Where,
ka is the absorption rate constant,
AD,gut is the amount of drug in gut compartment,
Cp is the mitotane plasma concentration,
Enzyme autoinduction model of mitotane
CLbaseline is the baseline mitotane clearance,
Aenz is the relative amount of enzyme in hypothetical enzyme compartment.
Mitotane plasma concentrations influenced enzyme turnover rate and enhanced enzyme synthesis as described in equation 2.
dA enz = Kin . (1 + Slope . Cp) - Kout . Aenz (2) dt
The enzyme induction model assumes that the rate of enzyme synthesis (Kin) follows zero-order kinetics, while the rate of degradation (Kout ) follows first-order kinetics dependent upon relative amount of enzyme.
At steady state enzyme concentrations,
Kin = Kout (3)
Parameters which could not be estimated from the data because of insufficient information were fixed according to published values or to arbitrary and/or physiologically plausible values which were subsequently evaluated by sensitivity analysis.
Other models tested included: (i) a two-compartment model; (ii) models with mitotane being eliminated by two distinct inducible and uninducible pathways; and (iii) models incorporating both gut wall and hepatic enzyme induction. A more complex physiologically based approach was also tested including (iv) a minimal physiologically based pharmacokinetic model (47); and (v) a semiphysiological well stirred liver model to incorporate the first pass effect (48). Both linear and nonlinear relationships (Emax and sigmoidal Emax models) were tested to describe the effect of mitotane on enzyme formation.
Enzyme autoinduction model of mitotane
IIV was introduced to volumes of distribution and slope (Equation 4) assuming a normal distribution of n with mean zero and variance @2.
Øj=0 ;· exp(nij) (4)
Where Puj is the jth individual pharmacokinetic parameter of the ith subject, 0j the population estimate of the respective pharmacokinetic parameter and nij the deviation of the subject’s individual parameter from the population point estimate.
Additive, proportional and combined error models were scrutinized to obtain estimates for residual unexplained variability (RUV). For nested (hierarchical) models, the likelihood ratio test was used which assumes that the difference in objective function values (OFV) (representating an overall prediction error) between two models is chi- squared distributed. Decisions regarding model preference were based on a preselected level of significance (p=0.05), degrees of freedom (difference in total number of parameters), and a critical chi-square value (for the chosen level of significance and degrees of freedom). Nested models with fewer parameters and an OFV lower by an amount larger than the critical chi-square value was finally given a preference. For non- nested (non-hierarchical) models, the Akaike Information Criterion (AIC: OFV plus two times the number of parameters) was used and the model with a lower AIC value was preferred. Goodness of fit (GOF) plots were evaluated to assess the discrepancy between the observed and predicted data and included individual/population predicted concentrations (IPRED/PRED) vs. the observed concentrations and conditional weighted residuals (CWRES) vs. observed concentrations and vs. time range. Physiological plausibility and precision of parameter estimates assessed via bootstrap statistics with 1,000 samples were further criteria of model selection.
Enzyme autoinduction model of mitotane
2.3 Covariate analysis
After successful development of a basic structural model, covariates were analyzed to provide an explanation for IIV and to improve overall model performance. Covariate pre- selection was based primarily on physiological plausibility. Graphical screening for potential covariates was performed including CWRES and individual pharmacokinetic parameters estimates versus covariates. Correlated covariates were avoided to be tested together and preference among those was given to the covariate with greater scientific plausibility if they provided a similar improvement of the model. Mitotane is reported to alter y-GT (49) and triglyceride levels in patients and it is suggested to closely monitor the lipid profile during mitotane treatment (50). As the drug is known to be accumulated in the adipose tissue, greater body fat proportion in women might have an impact influence upon its volume of distribution (51). Considering these facts, parameter covariate relationships were tested on volume of distribution (BMI and sex) and Slope (CLCR, plasma y-GT and triglyceride levels).
Categorical covariate relationship (sex) was tested as a fractional change (0cov) from the typical value of a parameter estimate (01)
TVP (typical value of parameter) = 01 . [1 + 0cov . (COV)] (5)
Whereas, continuous covariates (BMI, CLCR, plasma y-GT and triglyceride levels) were analyzed as linear relationships,
TVP = 01 . [1 + 0cov .(COV - COVmedian)] (6)
Final covariate inclusion in the model was mainly based upon the decrease in OFV and IIV.
Enzyme autoinduction model of mitotane
2.4 Simulation study
Relative change in clearance over time was evaluated graphically by designing stochastic simulations with model estimates. Both high dose (day1: 1.5 g, day2: 3g, day3: 4.5g, day4-onwards: 6g daily) and low dose (day1-2: 1g, day3-5: 1.5g, day6-8: 2g, day 9-11: 2.5g, day12-onwards: 3g daily) regimens used in clinical practice (24, 25) were tested in simulated population over a period of 3 months. Statistical and graphical evaluation was performed with the aim to attain therapeutic mitotane concentrations (14-20mg/L), avoiding toxic concentrations and to define appropriate timing of first TDM.
3. Results
Data regarding 76 patients out of 103, 45 females and 31 males, could finally be included in the model development process. These patients were aged between 17 and 75 years, the body weight was between 44 and 129 kg. 1137 observations of concentration data were part of the analysis. Descriptive statistics of patient, disease and treatment characteristics are presented in Table 1.
Absorption from gut compartment was modeled as a first order process. A one- compartment model was given preference over a two-compartment model, apparently because there was not sufficient information available in the data to precisely estimate peripheral volume of distribution and intercompartmental clearance. Although an empirical two-compartment model provided a lower OFV, this was at the expense of highly imprecise parameter estimates for volume of distribution and intercompartmental clearance. Attempt to fix these parameters according to published fat to plasma concentration ratios (12) resulted in even a higher IIV with regard to central volume of
Enzyme autoinduction model of mitotane
distribution and therefore conflicted with the model selection criteria. Fig. 1 provides a schematic representation of the model.
The parameter value for F was fixed to 0.35 (11) and value for ka (49.9 day-1) was also taken from the literature because of insufficient information available during the absorption phase of mitotane in the present data (52). Model simulations demonstrated that parameter estimate for Kout did not have any substantial impact upon concentration time profile. Therefore, the parameter value for Kout was fixed to 0.23 day-1 according to literature (33). CLbaseline was a priori assumed to be not estimable and sensitivity analysis using different CLbaseline values of up to 60 L/day did not exhibit any changes in OFV, hence a value of 1 was used to describe the relative change over time. The drug was estimated to have a high central volume of distribution of 6086 L (4743-7676L; bootstrap 95% CI). Clearance of the drug was found to increase over time due to the enzyme induction process. A relative linear increase of 3.97 L/day (“Slope”, 3.22-4.80; bootstrap 95% CI) per day per mg/L mitotane plasma concentration was observed. A simple linear model was preferred over Emax and sigmoidal Emax models because of implausible estimates for EC50 and Emax-
High IIV was associated with the volume of distribution (81.5%; point estimate). Extent of induction (Slope) was also found to be highly variable among the population (78.8%; point estimate). Estimates for RUV were adequately obtained with a combined error model. BMI was found to be a significant covariate for the volume of distribution, with an objective function value (OFV) decrease by 11 points. IIV was marginally reduced by 4.1% upon volume of distribution and 3.5% upon Slope. The comparison of the basic model and the covariate model in terms of decrease in IIV and OFV is represented in table 2. Table 3 presents bootstrap parameter estimates.
Enzyme autoinduction model of mitotane
Fig. 2 displays the basic goodness-of-fit (GOF) plots for the developed model. The IPRED were symmetrically distributed along the line of unity without any major outlier trends and PRED were adequate. CWRES were evenly distributed around zero depicting an adequate model performance over the concentration and time range. Individual plots (Fig. 3) exhibit the concentration time profiles for the observed and model predicted concentrations for four individuals exposed to mitotane treatment over different time periods ranging from 150 to 600 days. These plots illustrate that the model appropriately describes the diverse pharmacokinetic profiles across the studied population.
Monte-Carlo simulations
Temporal changes in clearance over time in a simulated population are depicted in Fig. 4. Median simulated plasma mitotane concentrations and respective percentiles (5th and 99th) with the high and the low dose regimen, respectively, are shown in Fig. 5. Persistent increase in plasma concentrations was observed. The 99th percentile reached the upper range of the therapeutic window at around day 16 for the high dose regimen and on day 55 for the low dose regimen.
4. Discussion
In this by far largest and well characterized series of ACC patients on mitotane treatment, we investigate the pharmacokinetics of mitotane by implementing a nonlinear mixed effect modeling approach. The approach not only considers the fixed effects (descriptors of a process e.g., pharmacokinetic parameters and respective covariates) but also estimates the random variability (reflected by IIV and RUV) across the population by making use of nonlinear regression techniques. Mitotane showed a large and highly variable volume of distribution, partly explained by interindividual differences in BMI.
Enzyme autoinduction model of mitotane
Pronounced variability was also found for concentration dependent mitotane clearance attributable to autoinduction of mitotane metabolism.
Sparsity of data points as well as the complex pharmacokinetics were well handled by the population pharmacokinetic model as indicated by the GOF plots. A high volume of distribution is compatible with the extensive distribution of mitotane drug to adipose tissue (53). Moreover, high IIV associated with volume of distribution suggests a significant influence of individual patient characteristics. Individual patient demographics and lipid profiles were analyzed as possible sources of variability. While BMI was found to be a significant covariate for the volume of distribution, the reduction in IIV was marginal and individualization of therapy based on BMI seems not useful.
Our study has the particular strength of a uniform mode of measurement in regular time intervals and a large individual number of data points and corresponding clinical and laboratory data available. In comparison with a previous modeling effort (38) we come to a similar conclusion regarding high volume of distribution. Nevertheless, the published study has some important limitations although the data were informative enough to develop a three-compartmental model. Thus, the estimate for absorption rate constant (0.005 hr-1) corresponds to a physiologically implausible absorption half-life of 138 hours in that study. Importantly, this model does not take enzyme autoinduction into account. Empirical modeling approaches of metabolic autoinduction are frequently based on either of the two following approaches. The drug may be assumed to increase the rate of enzyme synthesis as demonstrated in case of rifampin, or it may decrease the rate of enzyme degradation as modeled in the case of ifosfamide (55, 56). Studies elucidating the mechanism of enzyme induction responsible for drug interactions concluded that mitotane increases gene expression of a number of transporters and enzymes, including CYP3A4 (57). Therefore, we assumed an increase in enzyme synthesis as an approach for
Enzyme autoinduction model of mitotane
modeling enzyme induction. An attempt was made to estimate a baseline value for mitotane clearance prior to any enzyme induction process but limited information in the initial phase of dosing precluded a reliable estimate. The sparsity of initial TDM data reflects the current practice of monitoring mitotane 3-4 weeks after the first dose. Therefore, the enzyme compartment which represented a time changing clearance of the drug was initialized to a baseline value of 1 (100%). With regard to the magnitude of mitotane induction of drug metabolizing enzymes, a limited parallel group comparison study revealed 18.3 fold and 5.0 fold decrease in midazolam (a CYP3A probe drug) and sunitinib AUCs (both administered orally) by mitotane, respectively (31). As a comparison, the very potent known CYP3A inducer rifampin caused a 4-fold reduction in oral sunitinib plasma exposure (58), while it decreased AUC of orally and intravenously administered CYP3A probe substrates by 10-20 fold and 1.9-3.5 fold respectively (33). The effect on intravenously administered CYP3A probe substrates reflects hepatic CYP3A induction. A reasonable estimate for maximal induction of hepatic CYP3A activity by mitotane may therefore be about 5-fold. The larger changes in mitotane clearance according to our empirical model (Fig. 4) suggest an additional induction of gut wall enzymes.
In order to quickly establish antitumor efficacy, it is imperative to attain target concentrations as early as possible. Previously, efforts have been made to develop an appropriate dosing regimen with mitotane considering ≥14 mg/L as the target concentration. Appropriate plasma levels can be ultimately achieved with a chronic dose from the beginning of treatment but at the expense of a lag time to achieve therapeutic concentrations (24). This period may be shortened with high loading dose regimens but at the cost of a higher risk to attain potentially toxic concentrations (25). Studies indicating high variation in plasma concentration buildup suggest the involvement of factors other
Enzyme autoinduction model of mitotane
than the dosage regimen such as enzyme induction and differences in intestinal absorption due to dietary variation (26). Our model was not able to identify major sources of interindividual variability; therefore, an exact prediction of the individual required dose is not possible. Hence TDM remains essential for mitotane treatment. Simulations based on our model however were supportive of using the high dose regimen for the rapid attainment of therapeutic concentrations (Fig 5).
The current model is expected to be helpful for decision making in clinical management of ACC with mitotane. It can be used for a posteriori dose adjustments in patients using Maximum a posteriori (MAP) Bayesian methods (59) based on limited TDM measurements (two or three) to estimate individual pharmacokinetic parameters. MAP Bayesian approaches have proven their benefit in TDM of a number of anticancer drugs including methotrexate and carboplatin (60, 61). Another important aspect regarding decision making is the addition of cytotoxic chemotherapy in patients who are not predicted to attain therapeutic concentrations within a clinically useful time frame (e.g. 90 days) when using their individual maximum tolerated dose. Also, in a palliative setting where response to mitotane monotherapy is limited (6), optimal dosing may be supported when drug exposure related parameters are taken into account in addition to tissue based markers of response (10). From a research perspective, the model may be used to link mitotane pharmacokinetics to pharmacodynamic endpoints such as tumor growth inhibition.
The limitations of the evaluation related to retrospective nature of the present study suggest to design a prospective study for a more physiological and a more detailed description of mitotane pharmacokinetics. It would be desirable to (i) take more samples (e.g., 1 per day) during the initial build-up of plasma concentration; (ii) occasionally apply a dense sampling scheme including several samples within the first hour after dose
Enzyme autoinduction model of mitotane
during one dosing interval to describe mitotane absorption kinetics; (iii) evaluate the effect of additional covariates such as food intake, disease state and co-medications on mitotane pharmacokinetics, and (iv) to quantify enzyme induction during therapy by separate CYP3A probe drugs such as midazolam.
5. Conclusion
The proposed model appropriately describes plasma concentrations during chronic treatment with mitotane. It includes concentration dependent induction of metabolizing enzymes that considerably accelerates mitotane elimination. If tolerated, using the high dose regimen with a first TDM on day 16 of treatment might be a good treatment strategy. The model is a next important step to use pharmacokinetic modeling to improve personalized dose selection as well as establishing the timing of TDM, while more data are urgently needed.
6. Declaration of interest
We declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
7. Funding
The study was supported by a scholarship grant from Higher Education Commission, Pakistan in collaboration with German Academic Exchange Program (DAAD), Germany (to U.A.), a fellowship by the Comprehensive Cancer Center Mainfranken (to M.K.) and an individual research grant by the German Research Council (FA 466/4-1, FA 466/4-2, KR 4371/1-1, KR 4371/1-2) to M.K. and M.F. and within the CRC/Transregio 205/1 “The Adrenal: Central Relay in Health and Disease” to MK and MF.
Enzyme autoinduction model of mitotane
8. Acknowledgments
We thank all patients and their families for contributing their data, Michaela Haaf and Henriette Herbst for data curation as part of the ENSAT registry.
9. References
1. Fassnacht M, Libé R, Kroiss M & Allolio B. Adrenocortical carcinoma: a clinician’s update. Nature Reviews Endocrinology 2011 7 323-335.
2. Else T, Kim AC, Sabolch A, Raymond VM, Kandathil A, Caoili EM, Jolly S, Miller BS, Giordano TJ & Hammer GD. Adrenocortical carcinoma. Endocrine Reviews 2014 35 282-326.
3. Terzolo M, Angeli A, Fassnacht M, Daffara F, Tauchmanova L, Conton PA, Rossetto R, Buci L, Sperone P, Grossrubatscher E et al. Adjuvant mitotane treatment for adrenocortical carcinoma. New England Journal of Medicine 2007 356 2372-2380.
4. Berruti A, Grisanti S, Pulzer A, Claps M, Daffara F, Loli P, Mannelli M, Boscaro M, Arvat E, Tiberio G et al. Long-term outcomes of adjuvant mitotane therapy in patients with radically resected adrenocortical carcinoma. The Journal of Clinical Endocrinology & Metabolism 2017 102 1358-1365.
5. Temple TE, Jones DJ, Liddle GW & Dexter RN. Treatment of Cushing’s disease. New England Journal of Medicine 1969 281 801-805.
6. Megerle F, Herrmann W, Schloetelburg W, Ronchi CL, Pulzer A, Quinkler M, Beuschlein F, Hahner S, Kroiss M & Fassnacht M. Mitotane monotherapy in patients with advanced adrenocortical carcinoma. The Journal of Clinical Endocrinology & Metabolism 2018 103 1686-1695.
7. Hescot S, Slama A, Lombes A, Paci A, Remy H, Leboulleux S, Chadarevian R,
Enzyme autoinduction model of mitotane
Trabado S, Amazit L, Young J et al. Mitotane alters mitochondrial respiratory chain activity by inducing cytochrome c oxidase defect in human adrenocortical cells. Endocrine Related Cancer 2013 20 371-381.
8. Cai W, Counsell RE, Djanegara T, Schteingart DE, Sinsheimer JE & Wotring LL. Metabolic activation and binding of mitotane in adrenal cortex homogenates. Journal of Pharmaceutical Sciences 1995 84 134-138.
9. Scheidt HA, Haralampiev I, Theisgen S, Schirbel A, Sbiera S, Huster D, Kroiss M & Müller P. The adrenal specific toxicant mitotane directly interacts with lipid membranes and alters membrane properties depending on lipid composition. Molecular and Cellular Endocrinology 2016 428 68-81.
10. Sbiera S, Leich E, Liebisch G, Sbiera I, Schirbel A, Wiemer L, Matysik S, Eckhardt C, Gardill F, Gehl A et al. Mitotane inhibits sterol-O-acyl transferase 1 triggering lipid- mediated endoplasmic reticulum stress and apoptosis in adrenocortical carcinoma cells. Endocrinology 2015 156 3895-3908.
11. MOY RH. Studies of the pharmacology of o,p’DDD in man. The Journal of laboratory and clinical medicine 1961 58 296-304.
12. van Slooten H, van Seters AP, Smeenk D & Moolenaar AJ. O,p’-DDD (Mitotane) levels in plasma and tissues during chemotherapy and at autopsy. Cancer Chemotherapy and Pharmacology 1982 9 85-88.
13. Gebhardt DOE, Moolenaar AJ, van Seters AP, van der Velde EA & Gevers Leuven JA. The distribution ofo,p’-DDD (Mitotane) among serum lipoproteins in normo- and hypertriglyceridemia. Cancer Chemotherapy and Pharmacology 1992 29 331-334.
14. Hescot S, Seck A, Guerin M, Cockenpot F, Huby T, Broutin S, Young J, Paci A, Baudin E & Lombès M. Lipoprotein-free mitotane exerts high cytotoxic activity in
Enzyme autoinduction model of mitotane
adrenocortical carcinoma. The Journal of Clinical Endocrinology & Metabolism 2015 100 2890-2898.
15. Kroiss M, Plonné D, Kendl S, Schirmer D, Ronchi CL, Schirbel A, Zink M, Lapa C, Klinker H, Fassnacht M et al. Association of mitotane with chylomicrons and serum lipoproteins: practical implications for treatment of adrenocortical carcinoma. European Journal of Endocrinology 2016 174 343-353.
16. van Slooten H, Moolenaar AJ, van Seters AP & Smeenk D. The treatment of adrenocortical carcinoma with o,p’-DDD: prognostic implications of serum level monitoring. European journal of cancer & clinical oncology 1984 20 47-53.
17. Haak H, Hermans J, Velde C van de, Lentjes E, Goslings B, Fleuren GJ & Krans H. Optimal treatment of adrenocortical carcinoma with mitotane: results in a consecutive series of 96 patients. British Journal of Cancer 1994 69 947-951.
18. Baudin E, Pellegriti G, Bonnay M, Penfornis A, Laplanche A, Vassal G & Schlumberger M. Impact of monitoring plasma 1,1-dichlorodiphenildichloroethane (o,p?DDD) levels on the treatment of patients with adrenocortical carcinoma. Cancer 2001 92 1385-1392.
19. Hermsen IG, Fassnacht M, Terzolo M, Houterman S, Hartigh J den, Leboulleux S, Daffara F, Berruti A, Chadarevian R, Schlumberger M et al. Plasma Concentrations of o,p’DDD, o,p’DDA, and o,p’DDE as Predictors of Tumor Response to Mitotane in Adrenocortical Carcinoma: Results of a Retrospective ENS@T Multicenter Study. The Journal of Clinical Endocrinology & Metabolism 2011 96 1844-1851.
20. Terzolo M, Baudin AE, Ardito A, Kroiss M, Leboulleux S, Daffara F, Perotti P, Feelders RA, DeVries JH, Zaggia B et al. Mitotane levels predict the outcome of patients with adrenocortical carcinoma treated adjuvantly following radical resection.
European Journal of Endocrinology 2013 169 263-270.
21. Fassnacht M, Terzolo M, Allolio B, Baudin E, Haak H, Berruti A, Welin S, Schade- Brittinger C, Lacroix A, Jarzab B et al. Combination chemotherapy in advanced adrenocortical carcinoma. New England Journal of Medicine 2012 366 2189-2197.
22. Henning JEK, Deutschbein T, Altieri B, Steinhauer S, Kircher S, Sbiera S, Wild V, Schlötelburg W, Kroiss M, Perotti P et al. Gemcitabine-based chemotherapy in adrenocortical carcinoma: A Multicenter Study of Efficacy and Predictive Factors. The Journal of Clinical Endocrinology & Metabolism 2017 102 4323-4332.
23. Mauclère-Denost S, Leboulleux S, Borget I, Paci A, Young J, Al Ghuzlan A, Deandreis D, Drouard L, Tabarin A, Chanson P et al. High-dose mitotane strategy in adrenocortical carcinoma: Prospective analysis of plasma mitotane measurement during the first 3 months of follow-up. European Journal of Endocrinology 2012 166 261-268.
24. Terzolo M, Pia A, Berruti A, Osella G, Alì A, Carbone V, Testa E, Dogliotti L & Angeli A. Low-dose monitored mitotane treatment achieves the therapeutic range with manageable side effects in patients with adrenocortical cancer 1. The Journal of Clinical Endocrinology & Metabolism 2000 85 2234-2238.
25. Faggiano A, Leboulleux S, Young J, Schlumberger M & Baudin E. Rapidly progressing high o,p’DDD doses shorten the time required to reach the therapeutic threshold with an acceptable tolerance: preliminary results. Clinical Endocrinology 2006 64 110-113.
26. Kerkhofs TM, Baudin E, Terzolo M, Allolio B, Chadarevian R, Mueller HH, Skogseid B, Leboulleux S, Mantero F, Haak HR et al. Comparison of two mitotane starting dose regimens in patients with advanced adrenocortical carcinoma. The Journal of Clinical
Enzyme autoinduction model of mitotane
Endocrinology & Metabolism 2013 98 4759-4767.
27. Daffara F, De Francia S, Reimondo G, Zaggia B, Aroasio E, Porpiglia F, Volante M, Termine A, Di Carlo F, Dogliotti L et al. Prospective evaluation of mitotane toxicity in adrenocortical cancer patients treated adjuvantly. Endocrine Related Cancer 2008 15 1043-1053.
28. Reif VD, Sinsheimer JE, Ward JC & Schteingart DE. Aromatic hydroxylation and alkyl oxidation in metabolism of mitotane (o,p’-DDD) in humans. Journal of pharmaceutical sciences 1974 63 1730-1736.
29. Sinsheimer JE, Guilford J, Bobrin LJ & Schteingart DE. Identification of o,p’- dichlorodiphenyl acetic acid as a urinary metabolite of 1-(o-chlorophenyl)-1-(p- chlorophenyl)-2,2-dichloroethane. Journal of Pharmaceutical Sciences 1972 61 314- 316.
30. Mornar A, Sertić M, Turk N, Nigović B & Koršić M. Simultaneous analysis of mitotane and its main metabolites in human blood and urine samples by SPE-HPLC technique. Biomedical Chromatography 2012 26 1308-1314.
31. van Erp NP, Guchelaar HJ, Ploeger BA, Romijn JA, Hartigh JD & Gelderblom H. Mitotane has a strong and a durable inducing effect on CYP3A4 activity. European Journal of Endocrinology 2011 164 621-626.
32. Theile D, Haefeli WE & Weiss J. Effects of adrenolytic mitotane on drug elimination pathways assessed in vitro. Endocrine 2015 49 842-853.
33. Baneyx G, Parrott N, Meille C, Iliadis A & Lavé T. Physiologically based pharmacokinetic modeling of CYP3A4 induction by rifampicin in human: Influence of time between substrate and inducer administration. European Journal of Pharmaceutical Sciences 2014 56 1-15.
Enzyme autoinduction model of mitotane
34. Chortis V, Taylor AE, Schneider P, Tomlinson JW, Hughes BA, O’Neil DM, Libé R, Allolio B, Bertagna X, Bertherat J et al. Mitotane therapy in adrenocortical cancer induces CYP3A4 and inhibits 5a-reductase, explaining the need for personalized glucocorticoid and androgen replacement. The Journal of Clinical Endocrinology & Metabolism 2013 98 161-171.
35. Moolenaar AJ, van Slooten H, van Seters AP & Smeenk D. Blood levels of o,p’-DDD following administration in various vehicles after a single dose and during long-term treatment. Cancer chemotherapy and pharmacology 1981 7 51-54.
36. Ihunnah CA, Jiang M & Xie W. Nuclear receptor PXR, transcriptional circuits and metabolic relevance. Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease 2011 1812 956-963.
37. Sinz MW. Evaluation of pregnane X receptor (PXR)-mediated CYP3A4 drug-drug interactions in drug development. Drug Metabolism Reviews 2013 45 3-14.
38. Kerkhofs TMA, Derijks LJJ, Ettaieb H, den Hartigh J, Neef K, Gelderblom H, Guchelaar HJ & Haak HR. Development of a Pharmacokinetic model of mitotane. Therapeutic Drug Monitoring 2015 37 58-65.
39. Wickham H, Francois R, Henry L & Müller K. dplyr: A Grammar of Data Manipulatio n 2017. [R package version 0.7.4]. Available from: https://CRAN.R-project.org/packag e=dplyr
40. Wikham H. Easily Tidy Data with ‘spread()’ and ‘gather()’ Functions 2017. [R packag e tidyr version 0.8.0]. Available from: https://CRAN.R-project.org/package=tidyr
41. Grolemund G & Wickham H. Dates and Times Made Easy with lubridate. Journal of Statistical Software 2011 40 1-25.
42. Wickham H. ggplot2: Elegant Graphics for Data Analysis. Springer International
Enzyme autoinduction model of mitotane
Publishing 2016.
43. Zhang L, Beal SL & Sheiner LB. Simultaneous vs. sequential analysis for population PK/PD data I: best-case performance. Journal of pharmacokinetics and pharmacodynamics 2003 30 387-404.
44. Lindbom L, Ribbing J & Jonsson EN. Perl-speaks-NONMEM (PsN)-a Perl module for NONMEM related programming. Computer Methods and Programs in Biomedicine 2004 75 85-94.
45. Keizer RJ, van Benten M, Beijnen JH, Schellens JHM & Huitema ADR. Piraña and PCluster: A modeling environment and cluster infrastructure for NONMEM. Computer Methods and Programs in Biomedicine 2011 101 72-79.
46. Jonsson EN & Karlsson MO. Xpose — an S-PLUS based population pharmacokinetic/pharmacodynamic model building aid for NONMEM. Computer methods and programs in biomedicine 1999 58 51-64.
47. Cao Y & Jusko WJ. Applications of minimal physiologically-based pharmacokinetic models. J Pharmacokinet Pharmacodyn 2012 39 711-723.
48. Brill MJE, Valitalo PAJ, Darwich AS, van Ramshorst B, van Dongen HP, Rostami- Hodjegan A, Danhof M & Knibbe CAJ. Semiphysiologically based pharmacokinetic model for midazolam and CYP3A mediated metabolite 1-OH-midazolam in morbidly obese and weight loss surgery patients. CPT: Pharmacometrics and Systems Pharmacology 2016 5 20-30.
49. Fassnacht M, Kroiss M & Allolio B. Update in adrenocortical carcinoma. The Journal of Clinical Endocrinology & Metabolism 2013 98 4551-4564.
50. Larumbe T, Soto C, Goyogana A, De Linares UL, Sesmero A, Vina V, Olmedo C, Garcia M, Roussel O & Munoz A. Severe hypertriglyceridemia in relation to toxic
Enzyme autoinduction model of mitotane
levels of mitotane in a patient with stage IV adrenocortical carcinoma (ACC). Endocrine Abstracts 2017 .
51. Schwartz JB. The Influence of sex on pharmacokinetics. Clinical Pharmacokinetics 2003 42 107-121.
52. Watson AD, Rijnberk A & Moolenaar AJ. Systemic availability of o,p’-DDD in normal dogs, fasted and fed, and in dogs with hyperadrenocorticism. Research in veterinary science 1987 43 160-165.
53. Paci A, Hescot S, Seck A, Jublanc C, Mercier L, Vezzosi D, Drui D, Quinkler M, Fassnacht M, Bruckert E et al. Dyslipidemia causes overestimation of plasma mitotane measurements. Endocrinology, Diabetes & Metabolism Case Reports 2016 .
55. Smythe W, Khandelwal A, Merle C, Rustomjee R, Gninafon M, Bocar Lo M, Sow OB, Olliaro PL, Lienhardt C, Horton J et al. A semimechanistic pharmacokinetic- enzyme turnover model for rifampin autoinduction in adult tuberculosis patients. Antimicrobial agents and chemotherapy 2012 56 2091-2098.
56. Kerbusch T, Huitema AD, Ouwerkerk J, Keizer HJ, Mathôt RA, Schellens JH & Beijnen JH. Evaluation of the autoinduction of ifosfamide metabolism by a population pharmacokinetic approach using NONMEM. British journal of clinical pharmacology 2000 49 555-561.
57. Takeshita A, Igarashi-Migitaka J, Koibuchi N & Takeuchi Y. Mitotane induces CYP3A4 expression via activation of the steroid and xenobiotic receptor. Journal of Endocrinology 2013 216 297-305.
58. Bello C, Houk B, Sherman L, Misbah S, Sarapa N, Smeraglia J & Haung X. Effect of rifampin on the pharmacokinetics of SU11248 in healthy volunteers. Journal of Clinical Oncology 2005 23 3078-3078.
Enzyme autoinduction model of mitotane
59. Rousseau A & Marquet P. Application of pharmacokinetic modelling to the routine therapeutic drug monitoring of anticancer drugs. Fundamental & Clinical Pharmacology 2002 16 253-262.
60. Pignon T, Lacarelle B, Duffaud F, Guillet P, Catalin J, Durand A & Favre R. Dosage adjustment of high dose methotrexate using Bayesian estimation: a comparative study of two different concentrations at the end of 8-h infusions. Therapeutic Drug Monitoring 1995 17 164-169.
61. Guillet P, Monjanel S, Nicoara A, Duffaud F, Lacarelle B, Baqarry-Liegey D, Durand A, Catalin J & Favre R. A Bayesian dosing method for carboplatin given by continuous infusion for 120h. Cancer Chemotherapy and Pharmacology 1997 40 143- 149.
10. Figure legends
Fig 1. A schematic representation of one compartment pharmacokinetic model linked to the enzyme induction model. Mitotane plasma concentrations increasing the enzyme formation rate and enzyme amount is in turn affecting the mitotane clearance.
Fig 2. Goodness fo fit plots; observed vs individual predicted concentration (A) observed vs population predicted concentrations (B) conditional weighted residuals vs population predicted concentrations (C) conditional weighted residuals (CWRES) vs time after first dose (D).
Fig 3. Individual plots of 4 patients treated with mean mitotane doses of 4.12g (A), 4.57g (B), 3.91g (C) and 4.77g over different periods of time (TAFD, time after first dose; -·- observed concentrations; - A-predicted concentrations). Doses (rectangular boxes with associated lines) represent actual daily doses.
Fig 4. Change in clearance over time in a simulated population of 500 virtual subjects (median with 5th and 95th percentiles).
Fig 5. Concentration vs time plot for a simulated population of 500 virtual subjects administered with the clinically used highy dose (A) and low dose (B) mitotane regimens (median, 5th and 99th percentiles). The region between the dashed lines represents the therapeutic window. The vertical lines identify the time points proposed for the first TDM sampling for the two dosing regimens.
Enzyme autoinduction model of mitotane
11. Table legends
Table 1. Patient, disease and treatment characteristics
Table 2. Comparison of base and covariate model
Table 3. Model parameter estimates
Gut ☒ compartment
ka
Kin’(1+Slope.Cp)
Central compartment (V) ☒ mediates
induces
Enzyme compartment
Kout
CL
60
OBS
40
20
0
0
20
40
60
80
IPRED
B
80
60
OBS
40
20
0
0
20
40
60
80
PRED
C
6
3
CWRES
0
-3
-6
0
10
20
30
PRED
D 6
3
CWRES
0
-3
-6
0
1000
2000
3000
4000
TAFD (days)
A
25
Plasma concentration [mg/L]
20
15
10
5
6
4.5
4.5
3
3.5
1.5
2
0
0
50
100
150
200
250
TAFD (days)
B
15
Plasma concentration [mg/L]
10
5.5
5
5
4
3
2
1
0
0
100
200
300
400
500
TAFD (days)
C
25
Plasma concentration [mg/L]
20
15
10
6
5
4.5
4.5
3
2.5
3
1.5
0
0
50
100
150
200
TAFD (days)
D
25
Plasma concentration [mg/L]
20
15
10
6
5
4.5
4.5
3
3
1.5
0
0
0
200
400
600
TAFD (days)
0
30
T 60
T 90
| Page 31 150 | of | 35 | ||||||
| 100 50 Clearance [L/day] 0 | ||||||||
| 1 | T | |||||||
Time after first dose [days]
Plasma concentration [mg/L] >>
30
20
10
0
0
30
60
90
Time after first dose [days]
Plasma concentration [mg/L] 00
20
10
0
0
30
60
90
Time after first dose [days]
| Patient characteristic | total (n = 76) |
|---|---|
| Age, years [mean ± sd] | 48.6± 11.8 |
| Sex, female [n (%)] | 45 (59.2%) |
| Body height, m [mean ± sd] | 1.71 ± 0.92 |
| Body weight, kg [mean ± sd] | 75.4 ±14.5 |
| BMI, kg/m2 [mean ± sd] | 25.6±4.2 |
| Plasma cholesterol, mg/dL [mean ± sd] | 267.6 ±83 |
| Plasma creatinine, mg/dL [mean ± sd] | 0.82 ± 0.26 |
| Plasma y-glutamyltransferase, U/L [mean ± sd] | 187.3± 158.8 |
| Disease characteristics | patient n (%) |
| Stage at treatment initiation | |
| ENSAT I | 1 (1) |
| ENSAT II | 23 (30) |
| ENSAT III | 11 (14) |
| ENSAT IV | 37 (49) |
| local recurrence | 3 (4) |
| unknown | 1 (1) |
| Treatment characteristics | patient n (%) / Median (range) |
| treatment intention at start of mitotane | |
| adjuvant | 12 (16) |
| adjuvant and palliative* | 21 (28) |
| palliative | 37 (49) |
| unknown | 1 (1) |
| Concomitant systemic therapy any time during observation period | 44 (58) |
| yes | |
| unknown | 2 (3) |
| Chemotherapy regimens | |
| Number of regimens | 5 (1-5) |
| EDP | 34 (45) |
| Streptozotocin | 29 (38) |
| Gemcitabine/Capecitabine | 13 (17) |
| other | 16 (21) |
| unknown | 2 (3) |
*Treatment initiation in adjuvant intention, later continued as palliative treatment
BMI: Body Mass Index, ENSAT: European Network for the Study of Adrenal Tumors, EDP: Etoposide, Doxorubicin, Cisplatin regimen.
| Model | Objective function value | Interindividual variability (%CV) |
|---|---|---|
| Base model | 4764.4 | 81.5 (V), 78.8 (Slope) ☒ |
| Covariate model | 4753.3 | 77.3 (V), 75.2 (Slope) ☒ |
CV: coefficient of variation, V: volume of distribution
Enzyme autoinduction pharmacokinetic model of mitotane
| Pharmacokinetic parameters | ||
|---|---|---|
| Parameter | median estimate | 95% CI |
| ka (day-1) | 49.9 | Fixed |
| F (%) | 35 | Fixed |
| V/F (L) | 6086 | 4743-7673 |
| Kin (day1) | 0.23 | Fixed |
| Kout (day-1) | 0.23 | Fixed |
| slope (L day1 day1) | 3.97 | 3.22-4.80 |
| BMI covariate effect upon V (fractional change | 0.055 | 0.01-0.11 |
| from typical value of V per unit of BMI) | ||
| Interindividual variability | ||
|---|---|---|
| Parameter | median estimate | 95% CI |
| niv(@2) | 0.54 | 0.28-0.87 |
| n; slope (co2) | 0.56 | 0.35-0.84 |
| Residual variability (combined error model) | ||
|---|---|---|
| Type | median estimate | 95% CI |
| Additive (o2) | 0.24 | 0.18-0.29 |
| Proportional (o2) | 2.28 | 1.50-3.13 |
ka: absorption rate constant, F: bioavailability, V: volume of distribution, Kin: rate of enzyme synthesis, Kout- rate of enzyme degradation, ni”: interindividual variability in V, n, slope: interindividual variability in slope, @2 and o2: variance, BMI: body mass index as kg m-2, CI: confidence interval.