A novel approach to estimate ontogenies for PBPK applications – From literature data to simulations
Hannah Mayer, Rolf Burghaus, Katrin Coböken, Sebastian Frechen, Ibrahim Ince, Jan Schlender
Bayer AG
Objectives: Application of physiologically based pharmacokinetic (PBPK) modelling in pediatric and geriatric populations requires a detailed understanding of age-dependent physiological processes (ontogenies). Important examples include the expression of certain proteins (e.g. plasma proteins, metabolic enzymes or transporters) or functional measures (e.g. glomerular filtration rate). The aim of this study was to develop a novel approach informed by ontogenetic information from literature that allows (1) to cover the complete age range (including maturation and ageing) with one generic ontogeny function and (2) to integrate individual and aggregated (e.g. mean and standard deviation) data. Resulting ontogenies will be populated in PK-Sim® as part of the Open Systems Pharmacology (OSP) suite [1].
Methods: Markov Chain Monte Carlo (MCMC) methods, which allow the usage of aggregated data in combination with individual data, are used to estimate the parameters of a function describing the typical course of the ontogeny as well as the variability around it. The typical course of the ontogeny is described using a piecewise defined function in combination with a flexible estimation of the transition points between the different pieces. The pieces are connected in such a way that the function is continuously differentiable at each point. As default there are three parts: The first one describes an increase during the maturation phase similar to a Hill function, the second one refers to a constant level in (healthy) adults and the third one describes a decrease during ageing similar to a Hill function. The variability may be modelled in an age-dependent manner making use of the continuously differentiable function. Based on the results of the MCMC the relevant quantities can be simulated in future applications taking into account the age of the simulated individual as well as variability and uncertainty for this particular age. As a first feasibility assessment the approach was applied to learn ontogenies for alpha-acid glycoprotein (AAG), Human Serum Albumin (HSA) and Hematocrit (HCT). In cases of differences with respect to gender or ethnicity separate ontogenies are fitted for each subgroup.
Results: A general applicability of the proposed approach with its flexible properties was demonstrated. Goodness of fit plots revealed that the obtained results for AAG, HSA and HCT ontogeny are in good agreement with the data. For example, the following properties shown in the data could be well described: there is no age-dependent decline for AAG, HSA levels in men are elevated compared to those in women and the decline of HCT levels in elderly is more pronounced for men than for women. The ontogenies for the plasma proteins HSA and AAG were qualified by a comparison of measured and predicted fraction unbound values in children and adults.
Conclusions: The newly developed approach can adequately describe ontogenies over the complete age range and reflects the observed variability. It comprises a hill-function-like increase during the maturation phase and a hill-function-like decrease during the ageing phase. This approach could be applied to derive ontogenies for PBPK applications in pediatric and geriatric populations.
References:
[1] http://www.open-systems-pharmacology.org
