Reassessing models of hepatic extraction.

The aim of this investigation is to compare different mathematical models of the liver in the context of in vitro-in vivo correlation. We reanalyze drugs from the Houston reviews [1, 2], and compare the mathematical models. For the well-stirred model, a particular form of the distributed tubes model, and the dispersion model, fits are done to in vitro and in vivo intrinsic clearance data from microsomal and hepatocyte experiments. The distributed and dispersion models have decreased residuals as compared to the well-stirred model, but neither is to be clearly preferred over theother. It seems likely that drug-specific factors have a major impact on the quality of IVIVC correlations. While new experiments are needed to validate IVIVC models, our results indicate that improved correlation of in vitroand in vivo data is possible for high clearance drugs by using either a dispersion or distributed tube model rather than a well-stirred model.