Copyright@ IFAC Modelling and Control in Biomedical Systems, KarlsburglGreifswald. Gennany, 2000 INCORPORATING VARlABILI1Y AND UNCERTAINlY INTO INTERSPECIES SCALING OF PHARMACOKINETICS Ivan Nestorov" David CarliIe J , Hannah Jones" Brian Houston 1 ,2, Malcolm Rowland 1 ,2 JCentre for Applied Pharmacokinetic Research, School of Pharmacy and Pharmaceutical Sciences, The University ofManchester, Oxford Road, Manchester M13 9PL, UK 2School ofPharmacy and Pharmaceutical Sciences, The University ofManchester, OxfordRoad, Manchester M13 9PL, UK 3Chiroscience, Cambridge Science Park, Milton Road, Cambridge CB4 4WE, UK Abstract: A Monte Carlo simulation procedure for incorporating measures of the variability and uncertainty· into the prediction of pharmacokinetics is presented. The approach is validated by a Monte-Carlo simulation of the prediction of the in vivo rat intrinsic clearance from in vitro (microsomal and hepatocyte) data for eight benzodiazepine drugs. For most of the drugs a significant overlap of the confidence regions with the line of unity exists. Therefore, the hypothesis, that the scaled intrinsic clearances from microsomes and hepatocytes are identical, can not be rejected, despite the apparently significant difference in the extrapolated mean values observed. An additional benefit of the scheme is that the statistical distributions for the scaling coefficient are also generated. Due to its generality, the proposed approach can be used without any limitations to incorporate the whole interspecies extrapolation scheme. Copyright @ 2000 IFAC Keywords: prediction, uncertainty, variability, Monte Carlo simulation, pharmacokinetic data. 1. INfRODUCTION The ability to predict the pharmacokinetics (PK) of drugs, especially in humans, from physicochemical, in vitro and in vivo preclinical data and other information is of great importance both to academia, in providing a quantitative framework for identification and investigation the key processes involved, and to industry, in facilitating the drug selection and development process. A human PK prediction exercise usually consists of a series of extrapolations, ranging from the prediction of PK parameters based on the physicochemical properties of the compound, through in vitro - in vitro and in 97 vitro - in vivo extrapolation, up to the inter - animal scaling of the PK profiles of the drug. The ultimate goal is to achieve as accurate quantitative prediction of the human pharmacokinetics as possible. It is generally known that the information used for PK predictions and interspecies scaling is characterized by a large degree of inherent variability and uncertainty (Obach, et al., 1997; Houston, 1994; Houston and Carlile, 1997). The latter results in the inevitable presence of a significant variability and uncertainty in the parameter estimates of the models used to carry out the phannacokinetic extrapolations. Yet, during the extrapolation, most researchers use