PAGE. Abstracts of the Annual Meeting of the Population Approach Group in Europe.
PAGE 23 (2014) Abstr 3103 [www.page-meeting.org/?abstract=3103]
Click to open
Poster: Methodology - Model Evaluation
Chayan Acharya, Andrew C. Hooker, Siv J÷nsson, Mats O. Karlsson
Department of Pharmaceutical Biosciences, Uppsala University, P.O. Box 591, SE-751 24 Uppsala, Sweden
Objectives: Non-compartmental analysis (NCA) calculates pharmacokinetic (PK) measures related to the systemic exposure to a drug following administration, e.g. area under the concentration-time curve and peak concentration. We developed a new functionality in R, nca_ppc, to (i) perform diagnostic checks for a population model (PopPK diagnosis) and (ii) perform a traditional NCA.
Methods: The nca_ppc functionality estimates the PK measures by traditional NCA procedures  for the observed data set. For the PopPK diagnosis, a set of concentration-time profiles is simulated using the associated population model. The nca_ppc functionality estimates the PK measures for each simulated data set and compares them with those estimated from the observed data. The analysis is performed at the population as well as the individual level. The 95% nonparametric prediction interval of the distribution of the simulated population means of each PK measure is compared with the corresponding observed population mean. The individual level comparison is performed based on the deviation of the mean of any simulated PK measure for an individual from the corresponding observed value. The nca_ppc functionality also reports the normalized prediction distribution error (NPDE) of the simulated PK measures for each individual .
Results: The nca_ppc functionality produces two default outputs depending on the type of analysis performed, i.e., PopPK diagnosis and regular NCA. The PopPK diagnosis feature of nca_ppc functionality produces 7 sets of graphical outputs to assess the ability of a population model to simulate the concentration-time profile of a drug and thereby identify model misspecification. In addition, tabular outputs are generated showing the values of the PK measures estimated from the observed and the simulated data, along with the deviation, NPDE, regression parameters used to estimate the elimination rate constant and the related population statistics. The tabular output for a regular NCA is similar to the output obtained in commercial software.
Conclusions: The nca_ppc functionality is a versatile and flexible tool-set written in R that successfully estimates PK properties from observed and simulated concentration-time data. It produces a comprehensive set of graphical and tabular output to summarize the statistical results including the model specific outliers. The output is easy to interpret and to use in evaluation of a population model.
Acknowledgement: This work was supported by the DDMoRe (www.ddmore.eu) project.