Using structural information to create physiologically based pharmacokinetic models for all polychlorinated biphenyls

Abstract

Physiologically based pharmacokinetic (PBPK) models are useful in describing the distribution, metabolism, and fate of xenobiotics across multiple species. The eventual goal of the present research is to create PBPK models for all 209 polychlorinated biphenyls (PCBs). Key parameters in any PBPK model are the tissue-to-blood partition coefficients. Tissue:blood partition coefficients relate the compound's concentration in a target tissue to its concentration in blood under equilibrium conditions. Data on the adipose:plasma partition coefficients of 24 PCBs were used in a regression analysis to find an expression for the adipose:plasma partition coefficient as a function of molecular structure. Using stepwise regression, it was found that three simple structural descriptors were sufficient to predict adipose:plasma partition coefficients for all 209 PCB congeners. Data on the distribution of PCBs among blood components were used to derive the adipose:blood partition coefficient from the adipose:plasma partition coefficient. The lipid contents of liver, muscle, and skin were used to derive the tissue:blood partition coefficient for those tissues from the adipose:blood partition coefficient. These results allow for the calculation of tissue:blood partition coefficients for liver, skin, muscles, and fat for all 209 PCB congeners.