Bayesian Analysis of a Physiologically Based Pharmacokinetic Model for Perchloroethylene in Humans

Abstract

Perchloroethylene (PCE) is a widely distributed pollutant in the environment, and is the primary chemical used in dry cleaning. PCE-induced liver cancer was observed in mice, and central nervous system (CNS) effects were reported in dry-cleaning workers. To support reconstruction of human PCE exposures, including the potential for CNS effects, an existing physiologically based pharmacokinetic (PBPK) model for PCE in the human (Covington et al., 2007) was modified by adding a brain compartment. A Bayesian approach, using Markov chain Monte Carlo (MCMC) analysis, was employed to re-estimate the parameters in the modified model by combining information from prior distributions for the model parameters and experimental data. Experimental data were obtained from five different human pharmacokinetic studies of PCE inhalation exposures ranging from 150 ppm to as low as 0.495 ppm. The data include alveolar or exhaled breath concentrations of PCE, blood concentrations of PCE and trichloroacetic acid (TCA), and urinary excretion of TCA. The PBPK model was used to predict target tissue dosimetry of PCE and its key metabolite, TCA, during and after the inhalation exposures. Posterior analysis was performed to see whether convergence criteria for each parameter were satisfied and whether the model with posterior distributions may be used to make accurate predictions of human kinetic data. With posteriors, the trend of percent of PCE metabolized in the liver at low concentrations was predicted under different exposure conditions. The 95th percentile for the fraction PCE metabolized at a concentration of 1 ppb was estimated to be 1.89%.