PFIM

Since the 90's, the importance of evaluating and optimising population designs has been highlighted. Indeed, the use of population analysis in the evaluation of the pharmacokinetics (PK) or the pharmacodynamics (PD) of drugs has been widely demonstrated as useful. Population analysis rely on nonlinear mixed effects modells, for which several parameter estimation methods have been proposed and implemented. Several software such as NONMEM or more recently MONOLIX exist, and these methods are also available in functions of more general statistical software such as the NLMIXED function of SAS or the nlme function of R. However, prior to performing the estimation, pharmacologists are confronted with the choice of the population design; this choice is crucial for an unbiased and efficient estimation of the model parameters. Population designs consist of a set of designs to be performed in groups of patients. Simulation studies confirmed that the balance between the number of groups, the number of subjects per group and the number and the allocation of the sampling times in each group strongly influence the precision of the parameter estimates, leading sometimes to unreliable results. A special section in the US FDA guidelines (1999) is dedicated to the importance of having informative samples in population analysis, especially in sparse datasets.

The classical approach to evaluate a design is based on simulation; however, this approach is very time consuming and does not allow testing several designs in a reasonable delay. PFIM is a R tool proposed to circumvent this problem [1-3]. It allows evaluation and optimisation of population designs and is based on the expression of the Fisher information matrix in nonlinear mixed effects models [4-10].

Since 2008, two main versions are implemented in parallel: a graphical user interface package using the R software (PFIM Interface) and an R script version (PFIM): PFIM Interface 2.1 allows both evaluation and optimisation for single response models, while PFIM 3.0 was extended to accommodate multiple responses models [7, 8] . The latter was based on extensions of the R functions PFIM 1.2 and PFIMOPT 1.0, released in 2003 for design evaluation and optimisation respectively. Conversely to the previous versions, we only have one function for both evaluation and optimisation of population designs.

Concerning the R script version, a new version PFIM 3.2 is now available as an extension of the version PFIM 3.0. This new version includes improvements in terms of model specification and expression of the Fisher information matrix. It is now possible in PFIM 3.2 to use models with inter-occasion variability (IOV) [10] and models including fixed effects for the influence of discrete covariates on the parameters [6].
More recently, PFIM Interface 3.1, a new version of the graphical user interface PFIM Interface 2.1, was made available. PFIM Interface 3.1 implements most features of the R script version PFIM 3.0 dedicated to design evaluation and optimisation for multiple response models. In addition to these features, options have been added for model specification and development of the expression of the Fisher information matrix.

More details on additional options available in PFIM Interface 3.1 and PFIM 3.2 are given in the news section. The documentations for both these new versions are available in the relevant section and include detailed explanations and examples as to how to use these versions of PFIM.

[1] Retout S, Duffull S, Mentré F. Development and implementation of the population Fisher information matrix for evaluation of population pharmacokinetic designs, Computer Methods and Programs in Biomedicine, 2001, 65: 141-151.

[2] Retout S, Mentré F. Optimisation of individual and population designs using Splus, Journal of Pharmacokinetic and Pharmacodynamics, 2003, 30: 417-443.

[3] Bazzoli C, Retout S, Mentré F. Design evaluation and optimisation in multiple response nonlinear mixed effect models: PFIM 3.0, Computer Methods and Programs in Biomedicine, 2010, 98 : 55-65.

[4] Mentré F, Mallet A, Baccar D. Optimal design in random-effects regression models, Biometrika, 1997, 84: 429-442.

[5] Retout S, Mentré F, Bruno R. Fisher information matrix for nonlinear mixed-effects models: evaluation and application for optimal design of enoxaparin population, Statistics in Medicine, 2002, 21: 2623-2639.

[6] Retout S, Mentré F. Further developments of the Fisher information matrix in nonlinear mixed-effects models with evaluation in population pharmacokinetics, Journal of Biopharmaceutical Statistics, 2003, 13: 209-227.

[7] Retout S, Comets E, Samson A, Mentré F . Design in nonlinear mixed effects models: optimization using the Fedorov-Wynn algorithm and power of the Wald test for binary covariates, Statistics in Medicine, 2007, 26: 5162-5179.

[8] Bazzoli C, Retout S, Mentré F. Fisher information matrix for nonlinear mixed effects multiple response models: evaluation of the appropriateness of the first order linearization using a pharmacokinetic/pharmacodynamic model, Statistics in Medicine, 2009, 28 :1940-1956.

[9] Retout S, Comets E, Bazzoli C, Mentré F. Design optimisation in nonlinear mixed effects models using cost functions:application to a joint model of infliximab and methotrexate pharmacokinetics, Communication in Statistics: Theory and Methods, 2009, 38 : 3351–3368.

[10] Nguyen TT, Bazzoli C, Mentré F. Design evaluation and optimisation in crossover pharmacokinetic studies analysed by nonlinear mixed effects models, Statistics in Medecine, 2012, 31: 1043-1058.