Prediction of human pharmacokinetics from animal data and molecular structural parameters using multivariate regression analysis: volume of distribution at steady state.

The aim of this study was to develop a regression equation for predicting volume of distribution at steady state (Vd(ss)) in humans to enable application to various types of drugs using animal experimental data for rats and dogs and some molecular structural parameters. The Vd(ss) data for rats, dogs and humans of 64 drugs were obtained from literature. The compounds have various structures, pharmacological activities and pharmacokinetic characteristics. In addition, the molecular weight, calculated partition coefficient (clogP), and the number of hydrogen bond acceptors were used as possible descriptors related to the Vd(ss) in humans. Multivariate regression analyses, multiple linear regression analysis and the partial least squares (PLS) method were used to predict Vd(ss) in humans. Interaction terms were also introduced into the regression analysis to evaluate the non-linear relationship. For the data set used in the present study, PLS with quadratic term descriptors gave the best predictive performance. The PLS model using Vd(ss) data for only two animal species and using easily calculated structural parameters could generally predict Vd(ss) in humans better than an allometric method. In addition, the PLS model with only animal data gave almost the same predictive performance as the PLS model with quadratic term descriptors. This model may be easier to use and be practical in a realistic situation, and could predict Vd(ss) in humans better than the allometric method.