Pharmacokinetic models in the development of exposure indicators in epidemiology

Physiological pharmacokinetic (PK) models can estimate tissue levels from data on blood solubility, partition coefficients and metabolic rate(s) of a chemical agent. Use of these models in epidemiological studies has been limited. PK models can be useful if four conditions are met: (1) potential causal agents are identified; (2) groups of subjects with substantially different profiles of exposure intensity are identified; (3) a physiological PK model is available to extrapolate target tissue concentrations for potential agents; and (4) mechanistic hypotheses predict substantially different risk for different time profiles of tissue concentration. An example of this approach is presented for cancer risk from mixed hydrocarbon exposures, including the development of three practical dose indices. This use of PK models produces dose-response relationships useful for risk assessment, and it permits epidemiological evaluation of mechanistic hypotheses derived from laboratory studies.