Development of a pharmacokinetic limited sampling model for temozolomide and its active metabolite MTIC

Purpose: To develop a pharmacokinetic limited sampling model (LSM) for temozolomide and its metabolite MTIC in infants and children. Methods: LSMs consisting of either two or four samples were determined using a modification of the D-optimality algorithm. This accounted for prior distribution of temozolomide and MTIC pharmacokinetic parameters based on full pharmacokinetic sampling from 38 patients with 120 pharmacokinetic studies (dosage range 145-200 mg/m² per day orally). Accuracy and bias of each LSM were determined relative to the full sampling method. We also assessed the predictive performance of the LSMs using Monte-Carlo simulations. Results: The four strategies generated from the D-optimality algorithm were as follows: LSM 1=0.25, 1.25, and 3 h; LSM 2=0.25, 1.25, and 6 h; LSM 3=0.25, 0.5, 1.25, and 3 h; LSM 4=0.25, 0.5, 1.25, and 6 h. LSM 2 demonstrated the best combination of low bias [0.1% (-8.9%, 11%) and 11% (4.3%, 15%)] and high accuracy [-1.0% (-12%, 24%) and 14% (7.9%, 37%)] for temozolomide clearance and MTIC AUC, respectively. Furthermore, adding a fourth sample (e.g., LSM 4) did not substantially decrease the bias or increase the accuracy for temozolomide clearance or MTIC AUC. Results from Monte-Carlo simulations also revealed that LSM 2 had the best combination of lowest bias (0.1±6.1% and -0.8±6.5%), and the highest accuracy (4.5±4.1% and 5.0±4.3%) for temozolomide clearance and MTIC apparent clearance, respectively. Conclusions: Using data derived from our population analysis, the sampling times for a limited sample pharmacokinetic model for temozolomide and MTIC in children are prior to the temozolomide dose, and 15 min, 1.25 h and 6 h after the dose.