Population Pharmacokinetic-Pharmacodynamic-Viral Dynamics Modelling of Maraviroc Monotherapy Data Using MONOLIX
P. L. S. Chan (1), P. Jacqmin (2), M. Lavielle (3), L. McFadyen (1), B. Weatherley (1)
(1) Global Pharmacometrics, Pfizer Global Research and Development, Sandwich, UK; (2) Exprimo NV, Mechelen, Belgium; (3) INRIA Saclay, France
Objectives: A 4-differential equation viral dynamics (VD) model was used to describe the kinetics and interaction of target cells, actively infected cells, latently infected cells and viruses in human immunodeficiency virus (HIV) infected patients1. NONMEM has been previously used for fitting pharmacokinetic-pharmacodynamic (PKPD)-VD model to the maraviroc (MVC) monotherapy data2. Not only are computation times very long but there are often convergence problems resulting from numerical difficulties in optimizing the linearized likelihood. Only a few of the parameters can be estimated and it is not feasible to perform simultaneous PKPD-VD modelling. MONOLIX implements a stochastic approximation of the standard expectation maximization (SAEM) algorithm for nonlinear mixed effects models without approximations. The SAEM algorithm replaces the usual estimation step of EM by a stochastic procedure which has been shown to be very efficient with improved convergence toward the maximum likelihood estimates3.
This analysis compares population PKPD-VD modelling of monotherapy MVC data using MONOLIX with NONMEM.
Methods: Plasma concentration (1250 samples) and viral load (1169 observations) arising from 63 asymptomatic HIV infected patients were available. Patients received 10 days MVC monotherapy with doses ranging from 25-300mg QD and 50-300mg BID.
A 2-compartment disposition model with first-order absorption was used to describe the MVC concentrations. An inhibitory Emax model was used to describe the viral inhibition. The need of an effect compartment and/or a lag time was examined to describe the delay in onset of viral inhibition.
Parameter estimation was performed using a 2-stage approach in MONOLIX version 2.4 and NONMEM VI. The predicted PK profile based on the Empirical Bayes Estimates (EBE) obtained from separate PK analysis was used to drive the viral inhibition. A simultaneous approach was also tested with MONOLIX.
Results: With a 2-stage approach, the time taken in MONOLIX to generate population and individual estimates including diagnostics (conditional means and standard errors, log likelihood profile, visual predictive checks and normalized prediction distribution errors) was over 50% less than in NONMEM without diagnostics. Parameter estimates were comparable between MONOLIX and NONMEM.
Conclusions: The SAEM algorithm allows simultaneous estimation of PKPD and viral dynamics parameters. MONOLIX provides an alternative option to NONMEM when facing lengthy computation times or convergence problems.
References:
[1] Funk GA, et al. Quantification of In Vivo Replicative Capacity of HIV-1 in Different Compartments of Infected Cells. J Acquir Immune Defic Syndr. 2001;26(5):397-404
[2] Rosario C, et al. A pharmacokinetic-pharmacodynamic model to optimize the phase IIa development program of maraviroc. J Acquir Immune Defic Syndr. 2006;42:183-91.
[3] Kuhn E, Lavielle M. Maximum likelihood estimation in nonlinear mixed effects models. Comput Statist Data Anal. 2005;49:1020-38.
