Understanding the kinetics of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) concentrations in humans is an important step for TCDD cancer risk assessment. In this paper longitudinal series of serum TCDD concentration measurements on U.S. veterans of the Vietnam war, who were exposed to dioxin during herbicide-spraying operations, are studied. The overall aim is to use these data to infer the dynamics of TCDD concentrations in humans. This is done by identifying a kinetic model describing the dioxin time course at the individual level. The individual toxicokinetic model is then expanded into a population model within a Bayesian hierarchical framework which allows residual variations across subjects that cannot be explained by observed covariates. Other complications in the data, such as unknown exposure histories, are also resolved implicitly through the hierarchical model. Moreover, the choice of a Bayesian approach enables the accumulation of external source of information in the form of prior distributions. The model is subjected to various diagnostic checks and analyses of sensitivity to distributional assumptions showing a good fit in terms of both the population and the kinetic features.
Population toxicokinetic analysis of 2, 3, 7, 8-tetrachlorodibenzo-p-dioxin using Bayesian techniques
Thomaseth K;Salvan A
2002
Abstract
Understanding the kinetics of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) concentrations in humans is an important step for TCDD cancer risk assessment. In this paper longitudinal series of serum TCDD concentration measurements on U.S. veterans of the Vietnam war, who were exposed to dioxin during herbicide-spraying operations, are studied. The overall aim is to use these data to infer the dynamics of TCDD concentrations in humans. This is done by identifying a kinetic model describing the dioxin time course at the individual level. The individual toxicokinetic model is then expanded into a population model within a Bayesian hierarchical framework which allows residual variations across subjects that cannot be explained by observed covariates. Other complications in the data, such as unknown exposure histories, are also resolved implicitly through the hierarchical model. Moreover, the choice of a Bayesian approach enables the accumulation of external source of information in the form of prior distributions. The model is subjected to various diagnostic checks and analyses of sensitivity to distributional assumptions showing a good fit in terms of both the population and the kinetic features.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
