Stefanie Kraff (1), Andreas Lindauer (1), Markus Joerger (2), Salvatore Salamone (3), Ulrich Jaehde (1)
(1) Institute of Pharmacy, Clinical Pharmacy, University of Bonn, Germany, (2) Department of Oncology & Hematology, Cantonal Hospital, St. Gallen, Switzerland, (3) Saladax Biomedical Inc., Bethlehem, PA, USA
Objectives: In prospective clinical trials pharmacokinetically guided dose adjustment was performed for paclitaxel and docetaxel [1, 2]. As target parameters for dose adaptation the time above a paclitaxel threshold concentration of 0.05 μmol/L (Tc>0.05) is used for paclitaxel and the area under the curve (AUC) for docetaxel. Individual Tc>0.05 and AUC values are estimated based on previously published pharmacokinetic (PK) models of paclitaxel and docetaxel [3, 4] by using the software NONMEM®. Since many clinicians are not familiar with the use of NONMEM® we developed Excel®-based dosing tools performing comparable parameter estimations like NONMEM®.
Methods: Typical population PK parameters and interindividual variability were taken from published models [3, 4]. An Alglib VBA code was implemented in Excel® to compute differential equations for the paclitaxel PK model [5]. Bayesian estimates of the PK parameters (EBE) were determined by the Excel® solver using individual drug concentrations and in addition for the paclitaxel PK model bilirubin concentrations, BSA, gender and age as covariates. For paclitaxel, concentrations from 50 patients were simulated receiving 6 cycles of chemotherapy. The paclitaxel dose was adapted according to a previously published algorithm [6]. For docetaxel, concentrations from 300 patients were simulated receiving one cycle of chemotherapy. Predictions of the Excel® tool were compared with those of NONMEM®, where EBEs were obtained using the POSTHOC function.
Results: The results suggested that there was no major difference in the predictive performance between Excel® and NONMEM® regarding drug concentrations and Tc>0.05 or AUC values. Bias (median prediction error) for Tc>0.05 of paclitaxel was 0.07% with NONMEM® and 2.81% with Excel®, for the AUC of docetaxel 0.84% and 2.91%, respectively. Precision (median error) for Tc>0.05 was 10.14% and 11.44%, for the AUC 14.27% and 14.75%, respectively. The mean deviation of the estimated paclitaxel Tc>0.05 values between both programs was about 8 minutes. In 11% of the dose calculations, diverging Tc>0.05 values between the Excel® and NONMEM® Bayesian estimations resulted in different doses. The mean deviation of the estimated docetaxel AUC values was 0.158 mg·h/L.
Conclusions: The PK models of paclitaxel and docetaxel could be adequately implemented in Excel® with a predictive performance comparable to that of NONMEM. The developed dosing tools are self-explanatory and easy-to-use with acceptable computation times.
References:
[1] Central European Society for Anticancer Research-EWIV (CESAR) Study of Paclitaxel Therapeutic Drug Monitoring (CEPAC-TDM study), www.clinicaltrials.gov
[2] Engels et al. Clin Cancer Res 2011, 17: 353-362
[3] Joerger et al. Clin Cancer Res 2007; 13: 6410-6418
[4] Bruno et al. J Pharmacokinet Biopharm 1996; 24: 153-172
[5] http://www.alglib.net
Reference: PAGE 22 () Abstr 2735 [www.page-meeting.org/?abstract=2735]
Poster: Oncology