Physiologically based pharmacokinetic modeling in risk assessment -Development of Bayesian population methods
Toxicology and Risk Assessment, National Institute for Working Life, 112 79 Stockholm, Sweden
In the field of risk assessment, extrapolation of kinetic results, between species or from in vitro to in vivo conditions, are often needed, as pharmacokinetic data from humans are lacking in most cases, due to time, cost and most importantly, the perceived risks associated with experimental exposure of humans. For these applications, PBPK models have been used. Population variability in toxic effect may be estimated by introducing variability in model parameters. Data on these parameters may be found in the scientific literature. However, while the use of PBPK models in risk assessment has become widespread, the issue of PBPK model calibration has received relatively little interest. Since the early seventies, a large number of experimental inhalation studies of the kinetics of several volatile risk chemicals in human volunteers have been performed at the National Institute for Working Life. These data are unique, as the exposures were conducted at high exposure levels, and with simultaneous monitoring in several tissues and body fluids both during and post-exposure. Only very limited analyses of these rich data were performed in conjunction with the original publications, and these data thus provide an excellent opportunity for PBPK model calibration. As the fundamental problem of PBPK model calibration is that of combining information from previous animal and/or in vitro experiments, as manifested in the literature, with information from experimental data, a Bayesian analysis is the most natural choice. When complex models such as population PBPK are considered, it is standard practice to perform the Bayesian analysis by Markov-chain Monte Carlo (MCMC) simulation. The only software available for a convenient implication of population PBPK modeling in a Bayesian framework is the free software MCSim, developed by Dr. Frédéric Bois (available on-line). Examples of results from the application of Bayesian population techniques to several previously published data sets will be given, including predictions of risk distributions in simulated Swedish populations. MCSim is a convenient tool for Bayesian population analyses in general, and shows promise also for applications beyond PBPK modeling.