Pharmacokinetic bias analysis of the epidemiological associations between serum polybrominated diphenyl ether (BDE-47) and timing of menarche.

Abstract Background Associations between serum levels of polybrominated diphenyl ether (PBDE) and timing of pubertal development in adolescent girls ( e.g. , menarche) have been reported in both a cross-sectional and in a longitudinal study. The associations may be biased by growth dilution and pharmacokinetic changes during pubertal development. Objectives To use a physiologically-based pharmacokinetic (PBPK) model to assess how much of the epidemiologic association between PBDE and altered timing of menarche might be attributable to growth dilution and pubertal maturation. Methods We developed a PBPK model of BDE-47, a major congener of PBDE, to perform Monte Carlo (MC) simulation of plasma BDE-47 levels in a hypothetical target population aged 2 to 22 years old. The model used realistic distributions of physiological parameters including timing of growth spurts and menarche. The simulated data were analyzed as if they had come from an epidemiologic study. We compared the results based on the simulated population to those reported. Results The population characteristics, including age and body mass index (BMI) were similar between the simulated and reported groups. In the cross-sectional study design, the association between proportion of subjects with menarche before age 12 years and BDE-47 serum concentration was inverse in our simulated population, whereas the reported association was positive. In the longitudinal study design, simulated data were not suggestive of an association, whereas a delay in pubertal onset with higher concentrations of BDE-47 was observed in the epidemiologic study. Conclusion Results of our simulation suggest that in the previous cross-sectional study there was a small negative bias due to pharmacokinetics in the reported relationship between BDE-47 and age at menarche. However, in the longitudinal study there was little evidence of bias. Our study showed how PBPK modeling can be used to quantify the potential bias in epidemiological studies and also suggested that further studies on the optimal approach to modeling exposure are warranted to better understand and quantify the potential bias in the epidemiological associations with BDE-47 due to pharmacokinetics.