The importance of protein binding for the in vitro-in vivo extrapolation (IVIVE)-example of ibuprofen, a highly protein-bound substance

A physiologically based human kinetic model (PBHKM) was used to predict the in vivo ibuprofen dose leading to the same concentration-time profile as measured in cultured human hepatic cells (Truisi et al. in Toxicol Lett 233(2):172-186, 2015). We parameterized the PBHKM with data from an in vivo study. Tissue partition coefficients were calculated by an algorithm and also derived from the experimental in vitro data for the liver. The predicted concentration-time profile in plasma was in excellent agreement with human experimental data when the liver partition coefficient was calculated by the algorithm (3.01) demonstrating values in line with findings obtained from human postmortem tissues. The results were less adequate when the liver partition coefficient was based on the experimental in vitro data (11.1). The in vivo doses necessary to reach the in vitro concentrations in the liver cells were 3610 mg using the best fitting model with a liver partition coefficient of 3.01 compared to 2840 mg with the in vitro liver partition coefficient of 11.1. We found that this difference is possibly attributable to the difference between protein binding in vivo (99.9 %) and in vitro (nearly zero) as the partition coefficient is highly dependent on protein binding. Hence, the fraction freely diffusible in the liver tissue is several times higher in vitro than in vivo. In consequence, when extrapolating from in vitro to in vivo liver toxicity, it is important to consider non-intended in vitro/in vivo differences in the tissue concentration which may occur due to a low protein content of the medium.