Biomarker-Based Characterization of Chemical Exposures and Physiological Responses

Abstract

The chemisome is the chemical components of the exposome, defined as the totality of all exposures and their impact on health. Most current approaches, however, are limited in addressing this “totality” by only studying one chemical or one chemical family at a time in one exposed population. In addition, studying the links between chemical exposures and health is challenging due to an incomplete understanding of how physiological responses are associated with adverse health outcomes. This challenge is further complicated due to how chemical exposures change with demographics such as age, sex, race, and occupation. Thus, this dissertation aims to address these challenges by applying an unbiased approach to datasets of chemical biomarker levels and physiological measurements to systematically identify susceptible populations using the National Health and Nutrition Examination Survey. In the first project, I use quadratic regression models to characterize non-linear, age-based trends of chemical exposure in a sample comprised of 74,942 participants. I screen across 141 chemicals to identify those of higher concentrations in children relative to the older population. Children exhibit higher exposures to chemicals in consumer products such as phthalates, brominated flame retardants, lead, and tungsten. In contrast, restricted and highly persistent chemicals such as polychlorinated biphenyls and dioxins are higher in the older population. In the second project, I apply generalized linear models to evaluate exposure disparities by race/ethnicity for 143 chemicals in a representative sample of 38,080 US women. Compared to non-Hispanic White women, significant disparities are observed for non-Hispanic Black, Mexican American, Other Hispanic, and Other Race/Multi-Racial women. These women have higher levels of pesticides, including 2,5-dichlorophenol and 2,4-dichlorophenol, compounds in personal care products, including parabens and mono-ethyl phthalate, and heavy metals, such as mercury and arsenic. These findings are being coupled with toxicological data to prioritize chemicals to evaluate their role in health disparities. In the third project, I develop a framework using hierarchical clustering to characterize occupational exposures and physiological responses among 26,186 blue- and white-collar workers across 20 employment sectors for 108 chemicals and 27 physiological indicators. Blue-collar workers have higher levels of toxicants such as lead, cadmium, volatile organic chemicals, and polycyclic aromatic hydrocarbons compared to white-collar workers. Moreover, blue-collar workers exhibit higher levels of alkaline phosphatase (indicative of liver disease) and C-reactive proteins (indicative of inflammation). Together, these results suggest that blue-collar workers are exposed to higher levels of toxicants, which may induce physiological dysfunction. In the final project, I implement 10-fold cross-validated regression models to characterize the linear and non-linear associations between all-cause mortality and 27 physiological indicators to identify directionalities indicative of increased mortality risk in a sample of 45,032 participants. Twenty-four out of 27 indicators show non-linear associations, while height, triglycerides, and 60-second pulse show linear associations. Cholesterol-related indicators and glomerular filtration rate unexpectedly show parabolic associations, implying that higher mortality risk is associated with measurements in either extreme of the distribution instead of in one extreme. These findings highlight a need to study associations between these indicators and other health endpoints to gain insights into the physiological profiles associated with adverse health outcomes. Together, this thesis contributes to a better understanding of how chemical exposures can impact human health across multiple subpopulations. It also enables further exploration of how chemical exposures can perturb physiologic function conducive to increasing the risk for adverse health outcomes.