PAGE. Abstracts of the Annual Meeting of the Population Approach Group in Europe.
PAGE 20 (2011) Abstr 2122 [www.page-meeting.org/?abstract=2122]
Click to open
Poster: Absorption and physiology-based PK
K. Zhudenkov (1), S. Vinogradova (2), T. Karelina (1,2)
(1) Institute for Systems Biology SPb, Moscow, Russia; (2) Moscow State University, Moscow, Russia
Objectives: To investigate PK of PEG-IFN Alpha in HCV infected patients using population modeling and to compare two software packages - NONMEM 7 and Monolix 3.2. Previously the PK of PEG-IFN was studied in  without using a population modeling approach. However, data available from  suggested application of such methods with the aim of getting more details on the PK of PEG-IFN.
Methods: We modeled the plasma concentrations of PEG-IFN collected in 24 patients on two occasions (after 0'th and 7'th day of treatment). Also a variety of covariate data for each patient (race, level of inflammation, level of CD4+ cells, level of alanine aminotransferase (ALT), level of response - sustained virological responders (SVRs) and nonresponders (NRs))  was available. All ODE models built were tested with NONMEM 7 running FOCE with INTERACTION algorithm and with Monolix 3.2 running SAEM algorithm.
The first step of building a model was to choose a basic structural model - estimate number of compartments, model of absorption, check for saturation in absorption and elimination. Further model building implied estimation of inter-individual variability (IIV) for structure model parameters. After IIV estimation we selected the best error model. Further model verification included covariate analysis - building transformations of parameters with IIV for continuous covariates or analysis of different groups of parameters using categorical covariates or even analysis of categorical variability of parameters with IIV using categorical covariates.
Results: The model selection was similar using either NONMEM or MONOLIX. The final PK structural model contained one compartment with first order absorption and linear elimination. IIV analysis showed the necessity of application of IIV to volume V and absorption Ka. Analysis using categorical and continuous covariates showed no significant correlations (all correlation coefficients showed to be between -0.5 and 0.5) between means of parameters with IIV and covariates. In contrast, categorical variabilities for V from level of response and for Ka from race were apparent. Further analysis showed that such variabilities did not lead to significant lowering of OFV in NONMEM and -2xLL in Monolix.
Conclusions: Final structural model gave the values for V, Ka and Ke parameters (1.08 L/kg, 1.32 1/day and 0.41 1/day respectively) close to values obtained by Andrew H. Talal et al . IIV for V and Ka parameters (0.298 and 0.89 respectively) allowed the data to be more accurately described. No difference between pharmacokinetic parameters for Responders and Non-responders was revealed by population analysis.