Estimating xenobiotic half-lives in humans from rat data: influence of log P.

The nature of empirical allometric expressions relating dispositional and kinetic parameters for a given xenobiotic across multiple mammalian species is well known. It has also been demonstrated that a simple allometric relationship may be used to predict kinetic parameters for humans based merely on data for multiple xenobiotics from rats. We decided to explore reasons for the variance in the data arising from the latter method. We were particularly interested in learning whether any physicochemical characteristics of xenobiotics might account for outlying data points (i.e., poor prediction of human half-life from rat half-life). We have explored the influence of lipid solubility as reflected by a xenobiotic's log P value because adipose tissue comprises a significantly larger percentage of total body weight in humans than in rats. We used half-life data from the literature for 127 xenobiotics. A data subset of 102 xenobiotics for which we were able to find estimates of log P values, including several with extremely large log P values, was also analyzed. First and second order models, including and excluding log P, were compared. The simplest of these models can be recast as the familiar allometric relationship having the form Y = a(Xb). The remaining models can be seen as extensions of this relationship. Our results suggest that incorporation of log P into the prediction of xenobiotic half-life in humans from rat half-life data is important only for xenobiotics with extremely large log P values such as dioxins and polychlorinated biphenyls. Moreover, a second order model in logarithm of rat half-life accommodates all data points very well, without specifically accounting for log P values.