Pharmacokinetically based dosing of weekly paclitaxel to reduce drug-related neurotoxicity based on a single sample strategy.

PURPOSE: The present simulation study was initiated to develop a limited sampling strategy and pharmacokinetically based dosing algorithm of weekly paclitaxel based on pharmacokinetic (PK) and chemotherapy-induced peripheral neuropathy (CIPN) data from a large database.
METHODS: We used paclitaxel plasma concentrations from 200 patients with solid tumors receiving weekly paclitaxel infusions to build a population PK model and a proportional odds model on CIPN. Different limited sampling strategies were tested on their accuracy to estimate the individual paclitaxel time-above-threshold-concentration of 0.05 µmol/L (T c>0.05µM), which is a common threshold for paclitaxel. A dosing algorithm was developed based on the population distribution of paclitaxel T c>0.05µM and the correlation between paclitaxel T c>0.05µM and CIPN. A trial simulation based on paclitaxel PK and CIPN was performed using empirical Bayes estimations, applying the proposed dosing algorithm and a single 24-h paclitaxel PK sample.
RESULTS: A single paclitaxel plasma concentration taken 18-30 h after the start of chemotherapy infusion adequately predicted T c>0.05µM. By using an empirical dosing algorithm to target an average paclitaxel T c>0.05µM between 10 and 14 h, Bayesian simulations of repetitive (adapted) dosing suggested a potential reduction of grade 2 CIPN from 9.6 to 4.4 %.
CONCLUSIONS: This simulation study proposes a pharmacokinetically based dosing algorithm for weekly paclitaxel and shows potential improvement of the benefit/risk ratio by using empirical Bayesian models.