J Aluri, B Zussman, J Sidhu, B Patel and D Boyle
Clinical Pharmacokinetics/Modelling & Simulation, GlaxoSmithKline, King of Prussia, PA, US and Harlow, UK
Background: Cilomilast (ArifloÒ ) is a potent and selective PDE 4 inhibitor currently under development for the oral treatment of chronic obstructive pulmonary disease (COPD). Cilomilast population pharmacokinetics (Pop PK) were characterized in COPD patients in 2 Phase III clinical trials using a sparse sampling paradigm in 681 patients. Cilomilast plasma concentrations (Cp) from a third Phase III study (n=366 patients) were used as an external validation dataset to assess the predictability of the developed Pop PK model .
Objective: The overall objective of this analysis was to develop and validate a Pop PK model for cilomilast using data from 3 identical fixed dose studies in the Phase III program.
Methods: Model development and validation was performed using NONMEM V. Both one and two compartment pharmacokinetic models were examined. The influence of various covariates were examined on cilomilast pharmacokinetics to account for variability in the population. Model validation was performed in 3 three ways: 1) Comparison of observed vs Bayesian estimates of Cp using the prior Pop PK model parameters; 2) Fitting the prior Pop PK model to the validation dataset and comparing parameter estimates with those obtained using the model building data set and 3) By performing a posterior predictive check (PPC). PPC was performed by simulating a total of 1000 Cp values for each time point at which a Cp was determined. The percentage of observed data falling within 5th-95th percentile was determined.
Results: A one-compartment PK model with first order absorption and elimination best described cilomilast pharmacokinetics. Of the covariates examined, body weight was the only significant descriptor of cilomilast CL/F and Vss/F. Results of model validation confirmed predictability of the Pop PK model. PPC analysis indicated that the percent of the observed Cp in the validation dataset lying within the 5th-95th percentile of the simulated Cp was 94.5%.
Conclusions: The population pharmacokinetics of cilomilast were characterized in COPD patients and identified body weight as the only influential descriptor of cilomilast CL/F and Vss/F. External validation demonstrated good predictive performance of the Pop PK model.
Reference: PAGE 12 () Abstr 445 [www.page-meeting.org/?abstract=445]
Poster: poster