Pharmacokinetic modelling
“The ability to integrate in vitro data into PBPK models enhances the value, and aids the interpretation, of many in vitro techniques proposed as alternatives” to animals in toxicological research [1].
That’s one reason why Britain’s Health and Safety Laboratory (HSL) is spending a lot of time and energy developing advanced, easier-to-use physiologically-based pharmacokinetic (PBPK) models. The HSL, an agency of the Health and Safety Executive, is facilitating the wider application of PBPK models in chemical risk assessment, by making them less resource intensive and quicker to develop.
PBPK models are mathematical descriptions of the complex interactions affecting the disposition of substances in the body. In practice, they are computer simulations that can be used to predict drug or chemical absorption, distribution, metabolism and excretion (ADME) and toxicity. They are based on chemical-specific data such as blood and tissue solubility and plasma protein-binding; and on species-specific physiological data such as tissue perfusion rates, fractional blood flow, and the weights of organs and tissues including kidneys, liver, fat and muscle.
Early drug development
PBPK models range from relatively simple to quite complex simulations (e.g. representing separately as many as twelve organs and tissues), depending on the intended application. In drug development, they offer value in the early phase of drug candidate selection, prior to any in vivo study, by predicting pharmacokinetics such as bioavailability or metabolic stability. In particular, PBPK models can avoid known problems of inter-species extrapolations in pharmacokinetics (such as volume of distribution, clearance, and absorption and bioavailability) by using values relevant to the human species.
These simulations for early drug development are informed by physico-chemical information and by in vitro data, such as plasma protein-binding, cell membrane permeability, and microsomal or hepatocyte intrinsic clearance. Software is used to estimate features such as rate and extent of absorption and hepatic clearance, which can then be ‘slotted’ into a whole-body PBPK model to predict the time course of the compound in plasma and different tissues [2].
Applications of PBPK models
A PBPK model of this kind was evaluated at Pfizer for predicting human bioavailability for 16 drug compounds, both commercially available drugs and Pfizer candidates. By integrating readily available drug discovery parameters, the model successfully predicted bioavailability for about 80% of these compounds [3].
In conjunction with other ADME data such as P450 enzyme inhibition, models like this will enable progression of the most promising compounds for further development. By eliminating at an early stage those candidates likely to fail because of inappropriate ADME characteristics, animals are spared and fewer clinical trial volunteers are put at risk.
In the 1990s, the US Food and Drug Administration (FDA) requested a company to undertake a PBPK analysis of all-trans retinoic acid, being introduced at that time as a treatment for sun-damaged skin. The FDA wanted an evaluation of the likely exposure of a fetus when all-trans retinoic acid — a known teratogen — was dermally applied by a pregnant woman. The PBPK analysis was conducted [4] and the drug approved. Although some of the data used in the model derived from animal tests that had already been conducted, the simulation probably pre-empted further animal testing.
Better databases are required of human physiological parameters in health and disease. These should include aspects such as changes in blood flow in cardiovascular disease or diabetes, and alterations in tissue characteristics with age.
Chemical toxicity
In the risk assessment of chemicals for human health effects, data from human volunteer studies are often unavailable and assessment relies on numerous assumptions and estimates. A major challenge is extrapolating laboratory animal data to the exposure of humans in the workplace or wider environment: “For risk assessments based on animal data, the most obvious extrapolation that must be performed is from the tested animal species to humans” [2]. Others include extrapolations from high to low dose, from short- to long-term exposure, and from one exposure route to another.
PBPK modelling offers a powerful approach to increase the reliability of these extrapolations. Perhaps more importantly, by combining in vitro data with human physiological parameters, direct predictions of the human in vivo situation can be made with more confidence. Variations in susceptibility to chemical toxicity in population sub-groups of differing physiology, such as children or pregnant women, can also be modelled by these systems.
The HSL is developing a model equation generator, by which the user is engaged in dialogue about molecule-specific and species-specific details, the output being a document that provides a format for generic storage of models. They have also produced a PBPK parameter database, which contains physico-chemical, biochemical, anatomical and physiological data, vetted for their reliability.
The HSL believes that this approach will reduce the time required to develop a PBPK model from days to minutes, and that for predicting chemical toxicity, “…the greater availability of such a PBPK modelling capacity could potentially lead to a marked reduction in the use of animals” [1].
Human-based PBPK models
PBPK models are progressing in leaps and bounds, and offer several ways of replacing certain animal studies and, at the same time, improving the basis for risk assessment.
The models have many applications, including extrapolating from animal studies to humans; but simulations that incorporate human data would be the gold standard in this area. At present, more and better quality human-based data is needed, some of which can be provided by in vitro studies with human tissues, and some by safe, volunteer microdose studies [5]. This kind of information will also help ensure that PBPK models make valid predictions for human safety in a range of situations.
References and information
1. See HSL’s web pages: www.hsl.gov.uk/capabilities/pbpk.htm
2. Rowland M, Balant L & Peck C (2004). Physiologically based pharmacokinetics in drug development and regulatory science: A workshop report. AAPS PharmSci 6(1) article 6. www.aapspharmsci.org
3. Cai H, Stoner C, Reddy A et al (2006). Evaluation of an integrated in vitro-in silico PBPK (physiologically based pharmacokinetic) model to provide estimates of human bioavailability. Int J Pharm 308:133-139.
4. Clewell HJ, Andersen ME, Wills RJ et al (1997). A physiologically based pharmacokinetic model for retinoic acid and its metabolites. J Am Acad Dermatol 36:S77-85.
5. Dourson ML, Andersen ME, Erdreich LS et al (2001). Using human data to protect the public’s health. Regul Toxicol Pharmacol 33:234-256.
(Originally published in the Dr Hadwen Trust’s Science Review 2006)