PFIM 4.0: new features for optimal design in nonlinear mixed effects models using R
France Mentré (1), Thu Thuy Nguyen (1), Giulia Lestini (1), Cyrielle Dumont (1) and the PFIM group
(1) IAME, UMR 1137, INSERM and University Paris Diderot, 75018 Paris, France
Objectives: To extend PFIM, the only tool for population optimal design in R, with several new features.
Methods: Model based optimal design approaches are increasingly used in population pharmacokinetic/ pharmacodynamics (PKPD) [1]. These approaches rely on the Fisher information matrix (FIM) for nonlinear mixed effect models and are a good alternative to clinical trial simulation. Several software tools are available and were recently compared [2]. They all incorporate a PKPD library of models and model defined by differential equations. PFIM (www.pfim.biostat.fr) is available since 2001 and was extended in version 3 to multi-response models, inter-occasion variability, discrete covariates with prediction of power of Wald test [3].
We released in spring 2014 the version 4 of PFIM with several new features that we applied on several PKPD examples.
Results: For population designs, optimization can be done with fixed parameters or fixed sampling times. Previous information already obtained can be assumed and loaded through a predicted or an observed FIM. This is crucial to performed adaptive designs which are a strong requirement in drug industry and one of the task of the DDMoRe project [4].
Additional features for design in Bayesian estimation of individual parameters were added. The Bayesian information matrix was implemented. Design for MAP can be evaluated or optimized [5]. The predicted shrinkage is also reported [5]. We show with various examples the influence of design on shrinkage. This is a useful feature to select informative sampling times in therapeutic drug monitoring.
Conclusions: This new version of PFIM fulfilled some of the needs expressed in industry [1]. The examples again showed the importance of model based optimal design to predict good studies and anticipate ‘fatal’ ones.
This work was supported partly by the DDMoRe project (www.ddmore.eu).
References:
[1] Mentré F, Chenel M, Comets E, Grevel J, Hooker A, Karlsson MO, Lavielle M, Gueorguieva I. Current use and developments needed for optimal design in pharmacometrics: a study performed amongst DDMoRe's EFPIA members. CPT Pharmacometrics Syst Pharmacol., 2013;2:e46.
[2] Nyberg J, Bazzoli C, Ogungbenro K, Aliev A, Leonov S, Dufull S, Hooker A, Mentré F. Methods and software tools for design evaluation in population pharmacokinetics-pharmacodynamics. Br J Clin Pharmacol, 2014; in press.
[3] Bazzoli C, Retout S, Mentré F. Design evaluation and optimization in multiple response nonlinear mixed effect models: PFIM 3.0. Comput Methods Programs Biomed, 2010; 98:55-65.
[4] Harnisch L, Matthews I, Chard J, Karlsson MO. Drug and disease model resources: a consortium to create standards and tools to enhance model-based drug development. CPT Pharmacometrics Syst Pharmacol, 2013;2:e34.
[5] Combes F, Retout S, Frey N, Mentré F. Prediction of shrinkage of individual parameters using the Bayesian information matrix in nonlinear mixed-effect models with application in pharmacokinetics. Pharm Res, 2013; 30:2355-67.
