Early-life exposure to multiple persistent organic pollutants and metals and birth weight: Pooled analysis in four Flemish birth cohorts

BACKGROUND AND AIMS: Prenatal chemical exposure has frequently been associated with reduced fetal growth although results have been inconsistent. Most studies associate single pollutant exposure to this health outcome, even though this does not reflect real life situations as humans are exposed to many pollutants during their life time. The objective of this study is to investigate the association between prenatal exposure to a mixture of persistent environmental chemicals and birth weight using multipollutant models. METHODS: We combined exposure biomarker data measured in cord blood samples of 1579 women from four Flemish birth cohorts collected over a 10 years' time period. The common set of available and detectable exposure measures in these cohorts are three polychlorinated biphenyls (PCB) congeners (138, 153 and 180), hexachlorobenzene (HCB), dichlorodiphenyldichloroethylene (p,p'-DDE) and the metals cadmium and lead. Multiple linear regression (MLR), Bayesian Information Criterion (BIC), penalized regression using minimax concave penalty (MCP) and Bayesian Adaptive Sampling (BAS) were applied to assess the influence of multiple pollutants in a single analysis on birth weight, adjusted for a priori selected covariates. RESULTS: In the pooled dataset, a median (P25-P75) birth weight and gestational age of 3420 (3140-3700) grams and 39 (39-40) weeks was observed respectively. The median contaminant levels in cord blood were: 15.8, 26.5, 18.0, 16.9 and 91.5 ng/g lipid for PCB 138, PCB 153, PCB 180, HCB and p,p'-DDE, respectively, 0.075 µg/L for cadmium and 9.7 µg/L for lead. According to the applied statistical methods for multipollutant assessment, p,p'-DDE and PCB 180 were most consistently associated with birth weight. In addition, PCB 153 was selected when applying MCP and BAS. An inverse association with birth weight was found for the PCB congeners, while an increased birth weight was observed for elevated levels of p,p'-DDE. CONCLUSIONS: Assessing the health risk of combinations of exposure biomarkers reflects better real-world situations and thereby allows more effective risk assessment. Our results add to the existing evidence based on detrimental effects of PCBs on birth weight and indicate a possible increase in birth weight due to p,p'-DDE (while correcting for PCBs).