A physiologically-based pharmacokinetic (PBPK) and pharmacodynamic (PD) model of docetaxel (Doc) and neutropenia in humans

Abstract

2567 Background: We have scaled our previously described murine Doc PBPK model (Florian et. al, Proc. AACR, 2006) to humans to determine if it predicts human plasma Doc concentration. The resulting human plasma Doc predictions were then coupled to a low- order neutrophil model from the literature (Karlsson et. al, Clin. Cancer Res., 2006), and individual patient absolute neutrophil count (ANC) predictions were compared with actual ANC data. Methods: Plasma Doc concentration vs. time data were obtained from 75 patients given 1-h Doc infusions and sampled out to 48 h. 67 patients received 50–75 mg/m2 q3wk, and 8 patients received 35 mg/m2 weekly. Plasma Doc concentrations were determined by LC-MS, & ANC were measured weekly during cycle 1. Human organ volumes & blood flow rates for humans were obtained from the literature and scaled based on individual patient weight. Intra-tissue exchange rates & liver clearance rate from the mouse PBPK model were used unchanged as parameters in the scaled model. The PBPK model was used to drive the ANC model, which had been developed with data from patients treated on a q3wk schedule. Simulations were performed using MATLAB, and model predictions were compared with actual data based on a weighted sum squared error metric. Results: The initial, scaled PBPK model performed well in describing actual plasma Doc concentrations during the infusion & at times after 5 h, but not at the intermediate time points. Consequently, transition rates for the “other” compartment were estimated for the human population. The updated model was more accurate in predicting human plasma Doc concentrations (mean improvement 17%; range -3 to 70%). When the scaled PBPK model was coupled with the neutrophil model, the resulting predicted neutrophil profiles agreed with those from patients dosed on a q3w regimen, but not with those dosed weekly. Conclusions: This unique PBPK/PD model allows the prediction of human plasma Doc concentrations provided the patient weight & dose are known. Further validation of the coupled PBPK/PD model, including predicted tumor concentrations, is ongoing. No significant financial relationships to disclose.