Nathalie H Gosselin (1), Mohamad-Samer Mouksassi (1), Nastya Kassir (1), Leng Hong Pheng (1), JF Marier (1)
(1) Pharsight - A Certara Company, Montreal, Quebec, Canada
Objectives: A study design of a drug with an active metabolite must assess the pharmacokinetics (PK) of the parent drug (PG) and its active metabolite (MET) to understand its contribution to the therapeutic effects. The purpose of this study was to determine a trial design aimed of optimizing the precision on PK parameters of a PG-MET model in a pediatric population from age 2-18 years.
Methods: Adult PK data were used to develop a population PK model that links the parent drug (PG) and its active metabolite (MET) with NONMEM. The model included an allometric function with body weight (WT) and a biotransformation rate with the corresponding variability. The apparent clearance (CL/F) of PG and its MET were markedly different. To account of the complexities of the PG/MET model and the covariate distributions, a simulation/re-estimation approach (SIM-RE) was deemed the most appropriate. To reduce the number of the scenarios and iterations, the initial sampling schedule for PG was determined based on the optimization of population Fisher information matrix in WINPOPT using the mean WT values for each age group. The precisions of the PK parameters of PG and its MET, derived from this optimal design were re-assessed using SIM-RE in NONMEM using 50 replicates for each scenario. Realistic weight-age distribution were incorporated into the simulated data for patients 2 to 18 years using a generalized additive model for location scale and shape (GAMLSS)1 built on available data from CDC2. The asymptotic RSE derived from NONMEM covariance step of the 50 replicates were summarized. This procedure was tested for several possible N to determine the minimum number of subjects that would meet the desired precision.
Results: Optimal sampling schedule for the PG from WinPOPT gave RSE on CL/F and central volume of distribution (V/F) of approximately 12% and 23%, respectively (with N=25). The same trial design tested with the simulation and re-estimation approach using NONMEM resulted in similar values for RSE of the PG (i.e., mean [95%CI] 10% [9.4-10.6] for CL/F and of 24% [22.3-25.6] for V/F). The precision obtained with N=36 resulted in RSE (lower than 20%) for CL/F and V/F of the PG and its MET that met FDA recommendation.
Conclusions: Using optimal design software reduces the number of possible time points and accelerates applying the SIM-RE analysis which is the best method that account for the covariate distributions in the target population.
References:
[1] Mouksassi et al. Clin Pharmacol Ther. 2009;86(6):667-71. 2) 2000 CDC Growth Charts for the United States: Methods and Development. Centers for Disease Control and Prevention, Department of Health and Human Services, May 2002.
Reference: PAGE 21 (2012) Abstr 2545 [www.page-meeting.org/?abstract=2545]
Poster: Study Design