A new paradigm this website is that toxicity is determined by the critical concentration and time of exposure to the critical compound (or metabolite) at the critical site of action of the compound. Biokinetics is an important part of this paradigm. PBBK models take into account the fact that organs are linked together. Knowledge of in vitro kinetics can be combined with in vitro toxicodynamic data and incorporated into a model to predict in vivo systemic toxicity. An example of this is acrylamide for which in vitro data on neuronal

toxicity was known ( DeJongh et al., 1999). To date, ADME software packages, although showing promising predictive capacities, especially for absorption and distribution, have not yet been sufficiently validated and still require improvements. A report of an expert meeting organized by COST B15 that reviewed the use of QSAR in drug screening (Boobis et al., 2002) suggested that predictions using QSAR are no worse than those made using invitro tests, and have the added advantage that they need significantly less investment

in technology, resources and time. The report went onto describe a lack of confidence in these approaches and that more effort should be made by the software producers towards more transparency, in order to improve the Thiazovivin confidence of their consumers. It was also felt that controlled access to data from pharmaceutical companies would help to validate the models. If QSAR is used as the first step in risk assessment, then compounds that are flagged up as toxic can Phosphatidylethanolamine N-methyltransferase be de-selected, thus providing a 3Rs and cost-effective screening process. The workshop recommended that the basic parameters of the chemical should be considered (e.g. physicochemical properties) as

well as its partitioning into the tissues (indicated by the octanol:water partition coefficient versus the fat:blood partition coefficient) and the physiology of the organ (e.g. structure, blood flow, metabolic capacity, etc.). In addition, there should be more data generated to add to the predictive power of models. Further developments should combine in vitro and in silico data to feed PBBK models. To this end, increased efforts are needed to develop medium throughput systems to establish absorption (e.g. Caco-2), partitioning coefficients and metabolic parameters for the most important metabolizing organs, i.e. liver and skin. The use of publicly available tools such as the Model Equation GENerator (MEGen, http://xnet.hsl.gov.uk/megen, see Table 2) should be encouraged. Resulting PBBK models can be used to prioritize in vitro development projects. In order for a prediction model to be built, the extrapolation between the concentration of a compound in the incubation medium in vitro and the equipotent plasma concentration is a crucial step, involving predictive TK modelling.