Population pharmacokinetic modelling of unbound and total plasma concentrations of paclitaxel in cancer patients.

The aim of this study was to validate and further develop a mechanism-based population pharmacokinetic model for paclitaxel (Taxol; Bristol-Myers Squibb Co, Princeton, NJ, USA) based on the knowledge of Cremophor EL (CrEL) micelle entrapment and to evaluate the exposure/toxicity relationships. Paclitaxel (total and unbound) and CrEL concentrations were obtained according to a sparse sampling scheme with on average only 3.5 samples per course from 45 patients with solid tumours who received 3-hour infusions of paclitaxel (final dose range 112-233 mg/m(2)). The present data were predicted well by the mechanism-based model. In addition, bilirubin and body size were found to be significant as covariates. A change in body surface area (BSA) of 0.1 m(2) typically caused a change in clearance (CL) of 22.3 l/h and an increase in bilirubin of 10 microM typically caused a decrease in CL of 41 l/h. Toxicity was best described by a threshold model. In conclusion, even with a sparse sampling scheme, the same mechanism-based binding components as in the previously developed model could be identified. Once the CrEL and total paclitaxel plasma concentrations are known, the unbound concentrations, which are more closely related to the haematological toxicity, can be predicted.