2011 - Athens - Greece

PAGE 2011: Estimation methods
Georgios Vlasakakis

The impact of modified-release formulations on bridging of pharmacokinetic data from adults to children

Georgios Vlasakakis1, Massimo Cella1 and Oscar Della Pasqua1,2

1.GlaxoSmithKline, Stockley Park West, Uxbridge, UK, 2.Division of Pharmacology, Leiden/Amsterdam Centre for Drug Research, Leiden, The Netherlands

Objectives: Besides the known effects of age and weight on drug disposition, formulation-specific differences create additional complexities in terms of data fitting and extrapolation from adults to children. Of particular interest in paediatric development is the role of transit time and gastric emptying on lag time, rate and extent of the absorption processes. Yet, such processes may be mathematically intractable by conventional estimation procedures. The aim of this study was therefore to assess the suitability of Bayesian methods in NONMEM and WinBUGS to deal with data sparseness and problematic absorption using small datasets. Diclofenac was selected as a paradigm compound.

Methods: Pharmacokinetic data from enteric-coated diclofenac tablets (30 adults) derived from a previously analysed large population (141 adults and children) was used in this evaluation. The original dataset consisted of different formulations, but the enteric coated data was discarded due to long lag time observed in the trial [1]. Experimental data was modelled using a 1-compartment model with first-order absorption (WinBUGS) and a 1-compartment model with 2 sites of first-order absorption (NONMEM7) using Bayesian methods. Goodness-of-fit, statistical and graphical diagnostic measures were used to assess model performance. Model parameters were subsequently used for extrapolation purposes.

Results: WinBUGS was required to characterise the lag time, whereas absorption rate and drug disposition parameters were estimated in NONMEM7. Different model parameterisations based on algebraic and analytical solutions were explored, but our results show that the description of the lag phase cannot be obtained by conventional methods.

Conclusion: The use of pharmacokinetic bridging is highly desirable in paediatric drug development. However, whilst data sparseness has been the focus of modelling methodologies, pharmaceutical factors have important effects on drug disposition, which cannot be overlooked and may not easily addressed by maximum likelihood methods. The incorporation of priors ensures stabilisation of parameter estimation, preventing bias and deterioration of precision. 

[1] Standing, J.F., Tibboel, D.,Korpela, R.,Olkkola, K.T. Diclofenac pharmacokinetic meta-analysis and dose recommendations for surgical pain in children aged 1-12 years. Paediatr Anaesth. 21, 316-324 (2011).

Reference: PAGE 20 (2011) Abstr 2164 [www.page-meeting.org/?abstract=2164]
Poster: Estimation methods
Click to open PDF poster/presentation (click to open)