Predictive toxicodynamics: Empirical/mechanistic approaches

A major objective of the toxicological sciences is to predict the in vivo toxicological consequences of human exposure to pure chemicals, complex mixtures and commercial formulations. Historically, the experimental approach to this goal has been to investigate toxicological processes in whole animal models and extrapolate the results obtained to predict human risk using various extrapolation procedures (high-dose/low-dose extrapolation, interspecies extrapolation and route-to-route extrapolation). Can in vitro methods be more widely employed in quantitative risk assessment? One major limitation to the broader application of in vitro toxicity testing methods is the lack of validated techniques for the extrapolation of in vitro-derived toxicodynamic data to the in vitro situation. The objective of this paper is to describe some approaches to the development of techniques to extrapolate in vitro toxicity testing data to predict in vivo toxicological responses. An empirical approach within the context of a mechanistic framework is explored. The basic hypothesis is that the in vivo response can be constructed from a cellular toxicity factor that accounts for the cellular response and a toxicodynamic factor that relates toxicological events at the cellular level to the observable in vivo responses. A predictive paradigm to describe the in vivo acute target organ toxicity (hepatotoxicity) of a model chemical (cadmium) is discussed. The cellular toxicity factor is derived from in vitro toxicity testing studies using isolated rat hepatocytes. The toxicodynamic factor is derived through Biologically-Based Response (BBR) modelling techniques to predict target organ toxicity markers (i.e. plasma hepatic enzyme levels as markers for acute hepatotoxicity). The ultimate goal is to develop validated extrapolation procedures that can be applied to predicting target organ toxicity quantitatively in human populations based on in vitro toxicity studies using human cellular models. (C) 1997 Published by Elsevier Science Ltd.