Animal-Alternative Approaches to Understand Pharmacokinetics and Metabolism
Published:December 6, 2007
A member of the editorial boards of Toxicological Sciences, International Journal of Toxicology, Journal of Applied Toxicology and Journal of Child Health, Dr. Krishnan has authored a text book and co-authored over 100 full-length publications. He was a key contributor to the US EPA document on “Approaches for the application of physiologically-based pharmacokinetic (PBPK) data and models in risk assessment” (Federal Register 71: 55469-55470). The PBPK models developed by his group have been used by US EPA, US National Academy of Sciences, and Health Canada to derive guideline values for air pollutants and drinking water contaminants. Listed in Canadian WHO’s WHO (2006-), Dr. Krishnan is currently Fellow of the Academy of Toxicological Sciences and has received the prestigious Veylian Henderson Award (2000) from the Society of Toxicology of Canada and the Best Paper Award in Toxicological Sciences (2003) from the Society of Toxicology (US).
Dr. Kannan Krishnan
University de Montreal
2375 Cote Ste. Catherine
Montreal, PQ, H3C 3J7
Numerous in vitro and in silico tools are available to predict pathways of metabolism and quantitative estimates for some pharmacokinetic determinants (e.g., dermal permeability coefficient, oral absorption constant). Their use in toxicology and risk assessment is limited, because these parameters by themselves do not adequately reflect what goes on in the whole animal. Several scientists, following the works of Melvin Andersen and Yuichi Sugiyama, have continued to explore and make use of physiology-based pharmacokinetic (PBPK) models as a means of integrating in vitro data with other critical determinants to forecast ADME in whole animals and humans.
The identification of in vitro systems that give values for pharmacokinetic determinants and metabolic rates (phase I and phase II) consistent with those operative in vivo is crucial. There has been some success with the use of data from freshly isolated hepatocytes and post-mitochrondrial fractions within PBPK models to simulate in vivo kinetics of chemicals. Focussed efforts to predict metabolic rates from one in vitro system to another as well as to the in vivo situation would be a significant step forward. In this regard, high throughput assays need to be linked with QSAR modeling approaches to develop a more efficient way of predicting metabolic constants for developing PBPK models. This is an area where progress is clearly lacking. A logical starting point would be to develop QSAR/PBPK models based on in vitro data for specific categories of chemicals or those metabolized by specific isozymes (e.g., CYP2E1, CYP1A1/2, GSTµ).
Global QSAR approaches applied to chemicals without regard to the specific isozymes involved are likely to be of limited use. Existing in silico tools such as pharmacophore models and molecular lego may be used to classify or group the chemicals before submitting them to quantitative structure-property analyses of, for example, maximal velocity (Vmax) and Michaelis constant (Km). Given the substantial inter-individual variability in the metabolism rates, a distributional approach should be adopted in analyzing and describing the data. In metabolism studies, when the isozyme-specific protein content is known, a Bayesian approach can be implemented to make full use of the in vitro dataset. Accordingly, for example, the available prior knowledge of the distributions of enzyme content and rates can be used along with the subject-specific value of intrinsic clearance (CLint) to generate posterior probability distributions. Such an approach would facilitate the effective use of in vitro data and QSAR results in the context of population health risk assessment.
The development of in silico tools for predicting volume of distribution based on in vitro data on solubility and protein binding has been pursued by several groups, including our own lab. This is happening both in the pharmaceutical and toxicology arenas. Such efforts should also be extended to transporters, which are critical determinants of tissue uptake and clearance for specific groups of compounds and their metabolites.
Depending upon the in vitro or in silico approach used, it is likely that we may end up getting only approximate values for certain pharmacokinetic determinants. Before additional effort and resources are spent on refining such estimates or experimental systems, we need to evaluate the extent of precision needed for the intended purpose. For example, at any given exposure level, the internal dose is somewhere between zero (theoretical minimum) and potential dose (theoretical maximum). This large uncertainty is due to the fact that there is a lack of precise knowledge regarding the key pharmacokinetic determinants (e.g., tissue uptake, metabolic and pulmonary clearance, volume of distribution), each of which might range from a near zero value to infinity or physiological limit. In silico and in vitro methods clearly can help reduce the uncertainty but the extent to which precision on a certain parameter is needed depends upon the sensitivity of that particular parameter within the pharmacokinetic space defined by the combination of input parameters (see Figure). In other terms, a screening level QSAR method or semi-quantitative characterization of certain parameters (e.g., high, medium or low hepatic extraction) may be sufficient depending upon the extent to which uncertainty related to the particular parameter contributes to the uncertainty in the output. Such systematic uncertainty and sensitivity analyses using PBPK models, in the context of QSAR modeling and in vitro experimentation, would help optimal and focused use of limited resources.
A PBPK modeling framework depicting the evaluation of the key input parameters as a function of their impact on the model output. The physiological limit of certain processes are depicted using Qp, Q1, R1 and Kh which refer to alveolar ventilation rate, hepatic perfusion rate, tissue:blood ration of lipid content and Henry’s law constant.
©2007 Kannan Krishnan