ORIGINAL ABSTRACT

Fetal growth is a critical health outcome influenced by prenatal exposure to environmental chemicals, particularly endocrine-disrupting chemicals (EDCs). Studies have shown that pregnant women are simultaneously exposed to multiple chemicals, illustrating the need for methods that can examine the health effects of cumulative exposures. We compared four statistical methods-Principal Component Analysis (PCA), k-means clustering, Weighted Quantile Sum Regression (WQSR), and Bayesian Kernel Machine Regression (BKMR), -to identify associations between the mixture of chemicals and birth weight z-scores, including phthalate metabolites, bisphenol A, bisphenol A alternatives, triclosan, organophosphate pesticides, arsenic species, glyphosate and its breakdown product aminomethylphosphonic acid (AMPA), solvent metabolites, organophosphate flame retardants, fluoride, polybrominated diphenyl ethers, polychlorinated biphenyls (PCBs), organochlorine pesticides (OCs), cotinine, per- and polyfluoroalkyl substances (PFAS), and five metals. Our complete case analysis investigated the potential effects of a mixture of 46 chemicals on birth weight z-score, using 1127 mother-infant pairs from the Maternal-Infant Research on Environmental Chemicals (MIREC) study. PCA showed a significant inverse association between birth weight z-score and a component with loadings highest for PCBs (-0.035, 95%CI: (-0.068, -0.002)) using multivariable linear regression. The k-means analysis revealed distinct clusters of chemical profiles associated with lower birth weight z-score (-0.17, 95% CI: (-0.34, -0.01)) using multivariable linear regression, and primarily composed of arsenic (As), mercury (Hg), lead(Pb), ΣOC Chlordane and the PCBs, WQSR showed an inverse association with the overall mixture index (-0.065; CI: (-0.171, 0.04) driven mostly by Aroclor 1260, ΣOC Chlordane, glyphosate, and PCB180. BKMR highlighted that the birth weight z-score was 0.054 (-0.100, 0.209)when all chemicals in the 25thpercentile were compared to their medians, which decreased to -0.04 (-0.219, 0.14)when all chemicals in the 75th percentile were compared to their median values. After stratification by infant sex, associations tended to be larger in magnitude in females. We observed, according to all four approaches, that ΣOC Chlordane, ΣOC Insecticides, Aroclor 1260, dimethylarsinic acid, Pb, PCB170 and PCB180 were most often associated with decreased birth weight. These findings underscore the potential adverse effects of chemical mixtures on birth weight, the usefulness of using multiple methods, and suggest the need for continued research for evaluating cumulative environmental exposures in prenatal health outcomes.