Fredrik Jonsson1, Gunnar Johanson1, 2, and Frédéric Y Bois3
1National Institute for Working Life, Solna, Sweden;2Department of Occupational and Environmental Medicine, Uppsala University Hospital, Sweden; 3B3E - INSERM, Paris, France.
Physiologically based toxicokinetic (PBTK) models are routinely used to assess the kinetics of volatile substances. However, such modeling is often performed without assessing the reliability of the parameters or their variability in the population. This can be addressed by applying a population model using Markov chain Monte Carlo (MCMC) simulations. By using existing data on the physiological parameters, improved estimates of the parameters are produced, merging the prior information on the parameters with the information provided by the experimental data.
This approach was tested on two existing data sets. Åstrand et al. (1975) exposed 14 male volunteers to dichloromethane (DCM) vapor during moderate to heavy exercise. Engström and Bjurström (1977) exposed 12 male volunteers to DCM vapor during moderate exercise. Extensive physiological and kinetic data were collected in both studies. The kinetic data were fit to a PBTK model for DCM, including HbCO, based on Andersen et al. (1991). The analysis was done using the free software MCSim, available at http://rfs63.berkeley.edu/users/fbois/mcsim.html
The results show that already by using prior information on the parameters, these experimental studies are described reasonably well by our model. Furthermore, by using the information supplied by the experimental data it is possible to improve the statistical reliability of the distributions of the physiological parameters. Ultimately, this might have implications for risk assessment of volatile substances, and additional knowledge of their dose-response curves might be gained.
The study was financially supported by the Swedish Council for Work Life Research.
Reference: PAGE 8 (1999) Abstr 155 [www.page-meeting.org/?abstract=155]
Poster: poster