Serum Dioxin, Furan, and PCB Concentrations and Half-Life Study Among the U.S. General Population

ISEE-1205

Background and Objective:

The goal of this study is to quantify the population-based background concentrations of the 29 congeners of dioxins, furans, and PCBs for which World Health Organization (WHO) Toxic Equivalency Factors exist in serum using two datasets: the reference population (Jackson/Calhoun counties) data from the University of Michigan Dioxin Exposure Study (UMDES) 2004-2005, and the non-Hispanic white sub-population data in the National Health and Nutrition Examination Survey (NHANES) 2003-2004. The Jackson/Calhoun data has the advantage of lower limits of detection (LOD), while it is limited by having few participants older than 75 years. However, NHANES has more participants older than 75 years, but it has much higher LOD because of smaller blood samples.

Methods and Results:

Multiple imputation was employed to impute the serum concentrations below the LODs. Correlation coefficient matrix was used to examine the associations of serum concentrations among the 29 congeners. The background mean, median, 75th percentile, and 90th percentile serum concentrations were estimated by age and gender using linear and quantile regressions for complex survey data. By concatenating the UMDES data with the NHANES data, we improved the imputation for the values below the LOD in the NHANES data by incorporating the observed serum dioxin, furan, and PCB measures in the Jackson/Calhoun data. At the same time, inclusion of the NHANES data enhanced the estimates of the upper percentiles of serum dioxin, furan, and PCB concentrations among older people in the Jackson/Calhoun population. Using the regression results, the mean, median, 75th percentile, and 90th percentile serum concentrations of the WHO 29 congeners and their total toxic equivalent (TEQ) value can be estimated for females and males between ages 20-85. In addition, the relative half-lives of these congeners compared to 2,3,7,8-TCDD were estimated, based on the magnitudes of regression coefficients associated with age.