Physiologically based pharmacokinetic (PBPK) modeling and simulation in neonatal drug development: how clinicians can contribute

Introduction: Legal initiatives to stimulate neonatal drug development should be accompanied by development of valid research tools. Physiologically based (PB)-pharmacokinetic (PK) modeling and simulation are established tools, accepted by regulatory authorities. Consequently, PBPK holds promise to be a strong research tool to support neonatal drug development.

Area covered: The currently available PBPK models still have poor predictive performance in neonates. Using an illustrative approach on distinct PK processes of absorption, distribution, metabolism, excretion, and real-world data in neonates, we provide evidence on the need to further refine available PBPK system parameters through generation and integration of new knowledge. This necessitates cross talk between clinicians and modelers to integrate knowledge (PK datasets, system knowledge, maturational physiology) or test and refine PBPK models.

Expert opinion: Besides refining these models for ‘small molecules’, PBPK model development should also be more widely applied for therapeutic proteins and to determine exposure through breastfeeding. Researchers should also be aware that PBPK modeling in combination with clinical observations can also be used to elucidate age-related changes that are almost impossible to study based on in vivo or in vitro data. This approach has been explored for hepatic biliary excretion, renal tubular activity, and central nervous system exposure.     

A Physiologically based pharmacokinetic and pharmacodynamic (PBPK/PD) model for the organophosphate insecticide chlorpyrifos in rats and humans.

A PBPK/PD model was developed for the organophosphate insecticide chlorpyrifos (CPF) (O,O-diethyl-O-[3,5,6-trichloro-2-pyridyl]-phosphorothioate), and the major metabolites CPF-oxon and 3,5,6-trichloro-2-pyridinol (TCP) in rats and humans. This model integrates target tissue dosimetry and dynamic response (i.e., esterase inhibition) describing uptake, metabolism, and disposition of CPF, CPF-oxon, and TCP and the associated cholinesterase (ChE) inhibition kinetics in blood and tissues following acute and chronic oral and dermal exposure. To facilitate model development, single oral-dose pharmacokinetic studies were conducted in rats (0.5-100 mg/kg) and humans (0.5-2 mg/kg), and the kinetics of CPF, CPF-oxon, and TCP were determined, as well as the extent of blood (plasma/RBC) and brain (rats only) ChE inhibition. In blood, the concentration of analytes followed the order TCP > CPF > CPF-oxon; in humans CPF-oxon was not quantifiable. Simulations were compared against experimental data and previously published studies in rats and humans. The model was utilized to quantitatively compare dosimetry and dynamic response between rats and humans over a range of CPF doses. The time course of CPF and TCP in both species was linear over the dose range evaluated, and the model reasonably simulated the dose-dependent inhibition of plasma ChE, RBC acetylcholinesterase (AChE), and brain (rat only) AChE. Model simulations suggest that rats exhibit greater metabolism of CPF to CPF-oxon than humans do, and that the depletion of nontarget B-esterase is associated with a nonlinear, dose-dependent increase in CPF-oxon blood and brain concentration. This CPF PBPK/PD model quantitatively estimates target tissue dosimetry and AChE inhibition and is a strong framework for further organophosphate (OP) model development and for refining a biologically based risk assessment for exposure to CPF under a variety of scenarios.     

Analysis of methylmercury disposition in humans utilizing a PBPK model and animal pharmacokinetic data

Physiologically based pharmacokinetic (PBPK) models are excellent tools to aid in the extrapolation of animal data to humans. When the fate of the chemical is the same among species being compared, animal data can appropriately be considered as a model for human exposure. For methylmercury exposure, sufficient data exist to allow comparison of numerous mammalian species to humans. PBPK model validation entails obtaining blood and tissue concentrations of the parent chemical and metabolite(s) at various times following a known exposure. From ethical and practical considerations, human tissue concentrations following a known exposure to an environmental toxicant are scarce. While animal-to-human extrapolation demands that sufficient human data exist to validate the model, the validation requirements are less stringent if multiple animal models are utilized within a single model template. A versatile PBPK model was used to analyze the distribution and elimination of methylmercury and its metabolite, inorganic mercury. Uniquely, the model is formed in a generic way from a single basic template during the initial program compilation. Basic parameters are defined for diffferent PBPK models for mammalian species that span a relatively large range of sizes. In this article, the analyses include 12 species (mouse, hamster, rat, guinea pig, cat, rabbit, monkey, sheep, pig, goat, cow, and human). Allometric (weight-based) correlations of tissue binding coefficients, metabolism rate constants, and elimination parameters for both methylmercury and inorganic mercury are presented for species for which sufficient data are available. The resulting human model, in accord with the animal models, predicts relatively high inorganic mercury levels in the kidneys long after the disappearance of methylmercury from the blood.     

Physiologically based pharmacokinetic (PBPK) modeling of caffeine and theophylline in neonates and adults: implications for assessing children's risks from environmental agents

Children's risks can differ from those in adults for numerous reasons, one being differences in the pharmacokinetic handling of chemicals. Immature metabolism and a variety of other factors in neonates can affect chemical disposition and clearance. These factors can be incorporated into physiologically based pharmacokinetic (PBPK) models that simulate the fate of environmental toxicants in both children and adults. PBPK models are most informative when supported by empirical data, but typically pediatric pharmacokinetic data for toxicants are not available. In contrast, pharmacokinetic data in children are readily available for therapeutic drugs. The current analysis utilizes data for caffeine and theophylline, closely related xanthines that are both cytochrome P-450 (CYP) 1A2 substrates, in developing PBPK models for neonates and adults. Model development involved scale-up of in vitro metabolic parameters to whole liver and adjusting metabolic function for the ontological pattern of CYP1A2 and other CYPs. Model runs were able to simulate the large differences in half-life and clearance between neonates and adults. Further, the models were able to reproduce the faster metabolic clearance of theophylline relative to caffeine in neonates. This differential between xanthines was found to be due primarily to an extra metabolic pathway available to theophylline, back-methylation to caffeine, that is not available to caffeine itself. This pathway is not observed in adults exemplifying the importance of secondary or novel routes of metabolism in the immature liver. Greater CYP2E1 metabolism of theophylline relative to caffeine in neonates also occurs. Neonatal PBPK models developed for these drugs may be adapted to other CYP1A2 substrates (e.g., arylamine toxicants). A stepwise approach for modeling environmental toxicants in children is proposed.     

Physiologically based pharmacokinetic modeling of 1,4-Dioxane in rats, mice, and humans.

1,4-Dioxane (CAS No. 123-91-1) is used primarily as a solvent or as a solvent stabilizer. It can cause lung, liver, and kidney damage at sufficiently high exposure levels. Two physiologically based pharmacokinetic (PBPK) models of 1,4-dioxane and its major metabolite, hydroxyethoxyacetic acid (HEAA), were published in 1990. These models have uncertainties and deficiencies that could be addressed and the model strengthened for use in a contemporary cancer risk assessment for 1,4-dioxane. Studies were performed to fill data gaps and reduce uncertainties pertaining to the pharmacokinetics of 1,4-dioxane and HEAA in rats, mice, and humans. Three types of studies were performed: partition coefficient measurements, blood time course in mice, and in vitro pharmacokinetics using rat, mouse, and human hepatocytes. Updated PBPK models were developed based on these new data and previously available data. The optimized rate of metabolism for the mouse was significantly higher than the value previously estimated. The optimized rat kinetic parameters were similar to those in the 1990 models. Only two human studies were identified. Model predictions were consistent with one study, but did not fit the second as well. In addition, a rat nasal exposure was completed. The results confirmed water directly contacts rat nasal tissues during drinking water under bioassay conditions. Consistent with previous PBPK models, nasal tissues were not specifically included in the model. Use of these models will reduce the uncertainty in future 1,4-dioxane risk assessments.     

Physiologically based pharmacokinetic modeling of cyclotrimethylenetrinitramine in male rats

A physiologically based pharmacokinetic (PBPK) model for simulating the kinetics of cyclotrimethylene trinitramine (RDX) in male rats was developed. The model consisted of five compartments interconnected by systemic circulation. The tissue uptake of RDX was described as a perfusion‐limited process whereas hepatic clearance and gastrointestinal absorption were described as first‐order processes. The physiological parameters for the rat were obtained from the literature whereas the tissue : blood partition coefficients were estimated on the basis of the tissue and blood composition as well as the lipophilicity characteristics of RDX (logP = 0.87). The tissue : blood partition coefficients (brain, 1.4; muscle, 1; fat, 7.55; liver, 1.2) obtained with this algorithmic approach were used without any adjustment, since a focused in vitro study indicated that the relative concentration of RDX in whole blood and plasma is about 1 : 1. An initial estimate of metabolic clearance of RDX (2.2 h^?1 kg^?1) was obtained by fitting PBPK model simulations to the data on plasma kinetics in rats administered 5.5 mg kg^?1 i.v. The rat PBPK model without any further change in parameter values adequately simulated the blood kinetic data for RDX at much lower doses (0.77 and 1.04 mg ^?1 i.v.), collected in this study. The same model, with the incorporation of a first order oral absorption rate constant (K_a 0.75 h^?1), reproduced the blood kinetics of RDX in rats receiving a single gavage dose of 1.53 or 2.02 mg kg^?1. Additionally, the model simulated the plasma and blood kinetics of orally administered RDX at a higher dose (100 mg kg^?1) or lower doses (0.2 or 1.24 mg kg^?1) in male rats. Overall, the rat PBPK model for RDX with its parameters adequately simulates the blood and plasma kinetic data, obtained following i.v. doses ranging from 0.77 to 5.5 mg kg^?1 as well as oral doses ranging from 0.2 to 100 mg kg^?1. Published in 2009 by John Wiley & Sons, Ltd.     

Physiologically based pharmacokinetic model parameter estimation and sensitivity and variability analyses for acrylonitrile disposition in humans.

A physiologically based pharmacokinetic (PBPK) model of acrylonitrile (ACN) and cyanoethylene oxide (CEO) disposition in humans was developed and is based on human in vitro data and scaling from a rat model (G. L. Kedderis et al., 1996, TOXICOL: Appl. Pharmacol.140, 422-435) for application to risk assessment. All of the major biotransformation and reactivity pathways, including metabolism of ACN to glutathione conjugates and CEO, reaction rates of ACN and CEO with glutathione and tissues, and the metabolism of CEO by hydrolysis and glutathione conjugation, were described in the human PBPK model. Model simulations indicated that predicted blood and brain ACN and CEO concentrations were similar in rats and humans exposed to ACN by inhalation. In contrast, rats consuming ACN in drinking water had higher predicted blood concentrations of ACN than humans exposed to the same concentration in water. Sensitivity and variability analyses were conducted on the model. While many parameters contributed to the estimated variability of the model predictions, the reaction rate of CEO with glutathione, hydrolysis rate for CEO, and blood:brain partition coefficient of CEO were the parameters predicted to make the greatest contributions to variability of blood and brain CEO concentrations in humans. The main contributor to predicted variance in human blood ACN concentrations in people exposed through drinking water was the Vmax for conversion of ACN to CEO. In contrast, the main contributors for variance in people exposed by inhalation were expected to be the rate of blood flow to the liver and alveolar ventilation rate, with the brain:blood partition coefficient also contributing to variability in predicted concentrations of ACN in the brain. Expected variability in blood CEO concentrations (peak or average) in humans exposed by inhalation or drinking water was modest, with a 95th-percentile individual expected to have blood concentrations 1.8-times higher than an average individual.     

Physiologically based pharmacokinetic modeling of arsenic in the mouse

A remarkable feature of the carcinogenicity of inorganic arsenic is that while human exposures to high concentrations of inorganic arsenic in drinking water are associated with increases in skin, lung, and bladder cancer, inorganic arsenic has not typically caused tumors in standard laboratory animal test protocols. Inorganic arsenic administered for periods of up to 2 yr to various strains of laboratory mice, including the Swiss CD-1, Swiss CR:NIH(S), C57Bl/6p53(+/-), and C57Bl/6p53(+/+), has not resulted in significant increases in tumor incidence. However, Ng et al. (1999) have reported a 40% tumor incidence in C57Bl/6J mice exposed to arsenic in their drinking water throughout their lifetime, with no tumors reported in controls. In order to investigate the potential role of tissue dosimetry in differential susceptibility to arsenic carcinogenicity, a physiologically based pharmacokinetic (PBPK) model for inorganic arsenic in the rat, hamster, monkey, and human (Mann et al., 1996a, 1996b) was extended to describe the kinetics in the mouse. The PBPK model was parameterized in the mouse using published data from acute exposures of B6C3F1 mice to arsenate, arsenite, monomethylarsonic acid (MMA), and dimethylarsinic acid (DMA) and validated using data from acute exposures of C57Black mice. Predictions of the acute model were then compared with data from chronic exposures. There was no evidence of changes in the apparent volume of distribution or in the tissue-plasma concentration ratios between acute and chronic exposure that might support the possibility of inducible arsenite efflux. The PBPK model was also used to project tissue dosimetry in the C57Bl/6J study, in comparison with tissue levels in studies having shorter duration but higher arsenic treatment concentrations. The model evaluation indicates that pharmacokinetic factors do not provide an explanation for the difference in outcomes across the various mouse bioassays. Other possible explanations may relate to strain-specific differences, or to the different durations of dosing in each of the mouse studies, given the evidence that inorganic arsenic is likely to be active in the later stages of the carcinogenic process.     

Physiologically based pharmacokinetic modeling of inhaled 2-butoxyethanol in man

The arterial blood concentration of 2-butoxyethanol (ethylene glycol monobutyl ether) was simulated in a physiologically based pharmacokinetic model developed for a 70-kg man. Elimination data (Vmax and Km) were extrapolated from the perfused rat liver, while flows and volumes were from the literature. Simulated inhalation exposure to 2-butoxyethanol at 20 ppm (0.8 mmol/m3) and physical exercise at 50 W agrees well with the results from experimental exposure of human volunteers under identical conditions. In further simulations, the marked effects of physical exercise and co-exposure to ethanol are illustrated. The relatively rapid decay of 2-butoxyethanol in all compartments indicates that the parent compound is not likely to accumulate in the body. Further, linear kinetics may be expected at occupational inhalation exposure to 2-butoxyethanol. The study serves as an example of how a physiologically based pharmacokinetic model may be used to illustrate some aspects of occupational solvent exposure.     

Influence of body composition on disposition of the highly fat distributed compound as analysed by physiologically based pharmacokinetic (PBPK) modeling and simulation

A recent study suggested that the pharmacokinetics (PK) of highly fat distributed compounds can be affected by acute changes in the volume of adipose tissue. The present study investigates possible influences of body composition on the disposition of the highly lipophilic compound TAK‐357 in two rat strains. Physiologically based PK (PBPK) modeling and simulation was applied on single and multiple dose PK data of TAK‐357 in obese Wistar fatty rats and Wistar lean rats having approximately 45% and 13% body fat, respectively. The observed effects of an elevated fat mass in Wistar fatty rats on the plasma concentrations appeared to be partly compensated for by other differences between the two rat strains. A decrease in the tissue to blood partition coefficients under high body fat conditions was identified as another factor contributing to the difference in PK. A higher lipid content in the plasma in high body fat animals may result in relatively lower tissue to blood partition coefficients. PBPK‐based simulations indicate that the plasma concentrations of lipophilic compounds in high body fat conditions can differ by up to two‐times at steady‐state. This confirms that there is only a small impact of body composition change on the plasma concentration of highly lipophilic drugs and that the need for therapeutic dose adjustments may be limited.     

Physiologically based pharmacokinetic prediction of p -phenylbenzoic acid disposition in the pregnant rat

Disposition of p -phenylbenzoic acid (PPBA) in the pregnant Wistar rat (for both mother and fetuses) was predicted by using a physiologically based pharmacokinetic model. This model was constructed from ten organs for the mother and eight organs for fetuses, with fetal blood flow based on anatomical circulation in uteri and skin–amniotic fluid drug exchange. Plasma total clearance, and renal and nonrenal clearances were measured, and transplacental clearance, skin–amniotic fluid clearances and fetal metabolic clearance were taken from previously reported compartment analysis. Tissue-to-plasma partition coefficients (K_p) for the mother were almost the same as that of the interstitial fluid space (0.055–0.28), except for the kidney and liver. In contrast, K_p values for fetuses were small when membrane restricted and diffusion-limited uptakes were assumed in brain, gut, spleen, muscle, fat, and skin for the mother. The physiological model successfully predicted the PPBA concentration–time profiles for both mother and fetuses after intravenous injection into the mother. Further, the model could be applied to predict the results obtained via two other routes of administration. Fetal plasma PPBA concentrations were well predicted after PPBA injection into umbilical vein and fetal muscle. © 1998 John Wiley & Sons, Ltd.     

Physiologically based pharmacokinetic modeling of dibromoacetic acid in F344 rats

A novel physiologically based pharmacokinetic (PBPK) model structure, which includes submodels for the common metabolites (glyoxylate (GXA) and oxalate (OXA)) that may be involved in the toxicity or carcinogenicity of dibromoacetic acid (DBA), has been developed. Particular attention is paid to the representation of hepatic metabolism, which is the primary elimination mechanism. DBA-induced suicide inhibition is modeled by irreversible covalent binding of the intermediate metabolite α-halocarboxymethylglutathione (αH1) to the glutathione-S-transferase zeta (GSTzeta) enzyme. We also present data illustrating the presence of a secondary non-GSTzeta metabolic pathway for DBA, but not dichloroacetic acid (DCA), that produces GXA. The model is calibrated with plasma and urine concentration data from DBA exposures in female F344 rats through intravenous (IV), oral gavage, and drinking water routes. Sensitivity analysis is performed to confirm identifiability of estimated parameters. Finally, model validation is performed with data sets not used during calibration. Given the structural similarity of dihaloacetates (DHAs), we hypothesize that the PBPK model presented here has the capacity to describe the kinetics of any member or mixture of members of this class in any species with the alteration of chemical-and species-specific parameters.     

Physiologically based pharmacokinetic modeling of inhalation exposure of humans to dichloromethane during moderate to heavy exercise.

Dichloromethane (methylene chloride, DCM) is metabolized via two pathways in humans: mixed-function oxidases (MFO) and glutathione-S:-transferase (GST). Most likely, the carcinogenicity for DCM is related to metabolic activation of DCM via the GST pathway. However, as the two pathways are competing, the metabolic capacity for the MFO pathway in vivo is also of interest in risk assessment for DCM. Past estimates of MFO metabolism are based on the in vitro activity of tissue samples. The aim of the present study was to develop a population model for DCM in order to gain more knowledge on the variability of DCM inhalation toxicokinetics in humans, with main emphasis on the MFO metabolic pathway. This was done by merging published in vitro data on DCM metabolism and partitioning with inhalation toxicokinetic data (Astrand et al., 1975, Scand. J. Work.Environ. Health 1, 78-94) from five human volunteers, using the MCMC technique within a population PBPK model. Our results indicate that the metabolic capacity for the MFO pathway in humans is slightly larger than previously estimated from four human liver samples. Furthermore, the interindividual variability of the MFO pathway in vivo is smaller among our five subjects than indicated by the in vitro samples. We also derive a Bayesian estimate of the population distribution of the MFO metabolism (median maximum metabolic rate 28, 95% confidence interval 12-66 micromol/min) that is a compromise between the information from the in vitro data and the toxicokinetic information present in the experimental data.     

Physiologically based pharmacokinetic modeling as a tool for drug development

Since the pioneering work of Haggard and Teorell in the first half of the 20th century, and of Bischoff and Dedrick in the late 1960s, physiologically based pharmacokinetic (PBPK) modeling has gone through cycles of general acceptance, and of healthy skepticism. Recently, however, the trend in the pharmaceuticals industry has been away from PBPK models. This is understandable when one considers the time and effort necessary to develop, test, and implement a typical PBPK model, and the fact that in the present-day environment for drug development, efficacy and safety must be demonstrated and drugs brought to market more rapidly. Although there are many modeling tools available to the pharmacokineticist today, many of which are preferable to PBPK modeling in most circumstances, there are several situations in which PBPK modeling provides distinct benefits that outweigh the drawbacks of increased time and effort for implementation. In this Commentary, we draw on our experience with this modeling technique in an industry setting to provide guidelines on when PBPK modeling techniques could be applied in an industrial setting to satisfy the needs of regulatory customers. We hope these guidelines will assist researchers in deciding when to apply PBPK modeling techniques. It is our contention that PBPK modeling should be viewed as one of many modeling tools for drug development.     

Physiologically Based Pharmacokinetic (PBPK) Modeling of Pharmaceutical Nanoparticles

With the great interests in the discovery and development of drug products containing nanoparticles, there is a great demand of quantitative tools for assessing quality, safety, and efficacy of these products. Physiologically based pharmacokinetic (PBPK) modeling and simulation approaches provide excellent tools for describing and predicting in vivo absorption, distribution, metabolism, and excretion (ADME) of nanoparticles administered through various routes. PBPK modeling of nanoparticles is an emerging field, and more than 20 PBPK models of nanoparticles used in pharmaceutical products have been published in the past decade. This review provides an overview of the ADME characteristics of nanoparticles and how these ADME processes are described in PBPK models. Recent advances in PBPK modeling of pharmaceutical nanoparticles are summarized. The major challenges in model development and validation and possible solutions are also discussed.