Comprehensive exploration of interactions between persistent organic pollutants and nutritional biomarkers on birth weight

•Background and Aim Pregnant women are exposed to complex chemical mixtures, either nutrients or environmental stressors, whose joint impacts on fetal growth has been scarcely explored. The main objective of this study was screening interactions between chemicals and nutrients to better characterize their joint associations with birth weight. •Methods We analysed data from the population-based INMA birth cohort, relying on blood biomarkers from the first trimester and weight at birth. The most representative biomarkers of organochlorine compounds (OCs) and perfluoroalkyl substances (PFAS) were included, as well as nutritional biomarkers for vitamins and carotenoids. A dataset with minimal imputation was established with complete data of OCs and PFAS, carotenoids and vitamins (n=810). The statistical framework included a screening phase to identify interactions using Gradient Boosting Regression and Glinternet. Identified interactions were further characterized using regression models. •Results The screening step revealed interactions between perfluorooctanesulfonic acid (PFOS) and carotenoids, but also with other chemicals like the pesticide dichlorodiphenyldichloroethylene (p,p’-DDE). The regression methods showed that at low levels of β-cryptoxanthin (tertile 1 and 2) the associations between PFOS and birth weight were clearly strengthened compared the flat slope of the highest tertile, yet the interaction was not statistically significant. Moreover, statistically significant interaction between PFOS and p,p’-DDE was confirmed: a log increase of PFOS, increased 131.0 g (95% confidence interval 18.4-243.6) of birth weight in the Tertile 1 of p,p’-DDE whereas a reduction of -59.6 g (95% confidence interval -155.9, 36.7) was found in the Tertile 3 of p,p’-DDE (p interaction= 0.01). •Conclusions The results of this study supports the presence of complex interactions between chemicals and nutrients during pregnancy and birth weight. These findings may help to better characterize the heterogeneous findings in the overall body of epidemiological evidence, yet further research is needed to confirm those findings.