Maternal prenatal mercury exposure from rice and its association with newborn neurobehavioral development

Methylmercury (MeHg) exposure is a global environmental concern. Rice-derived MeHg exposure poses a potential neurodevelopmental risk in inland populations with low fish intake, yet epidemiological evidence remains scarce. This representative mother-infant cohort study was conducted in a mercury mining area in China, involving 420 mother-infant pairs, of which 206 were from mercury-exposed regions and 214 were from control regions. It was first found that the median total mercury and methylmercury levels in rice from mercury-exposed areas (11.19 ng/g and 5.07 ng/g) were significantly higher than those in control areas (3.34 ng/g and 1.09 ng/g). And the exposure levels of total mercury and MeHg in environmental media (rice) and population samples (placenta and umbilical cord blood) were positively correlated with the incidence of neurobehavioral developmental impairment (NBDI) in newborns. Subsequently, using a multistage logistic regression model, a significant association was demonstrated between placental MeHg exposure and NBDI (OR = 1.23, 95 % CI 1.02-1.49). Restricted cubic spline models further supported this positive linear dose-response relationship. Additionally, NBDI risk prediction models related to MeHg exposure were developed. Among them, the LightGBM model exhibited the best predictive performance, with an area under the curve (AUC) of 0.91 (95 % CI 0.86-0.96) and an accuracy of 0.85 (95 % CI 0.78-0.90). Its sensitivity, specificity, positive predictive value, negative predictive value, and F1 score all exceeded 0.82. Moreover, the interpretation of the SHAP model indicated that placental MeHg concentration was one of the most important predictors of NBDI occurrence. These findings reinforced existing epidemiological evidence on the association between MeHg exposure and NBDI, and fill a research gap regarding the health effects of MeHg exposure through rice consumption. They also provide an effective predictive tool for MeHg-related NBDI with clinical applicability.