Association between metal mixture exposure and abnormal glucose metabolism in multiple mixture exposure models: Evidence from NHANES 2015–2016

Abstract

Previous studies primarily focused on the single metal exposure and one-sided glucose metabolism disordered states, leading to conflicting results. Herein, we combined diabetes and prediabetes as abnormal glucose metabolism (AGM) to describe the effect of metal mixture exposure on it. Eligible data were obtained from the National Health and Nutrition Examination Survey (NHANES) 2015–2016. In the generalized linear model (GLM), Cd (OR: 1.060, 95 %CI: 1.032–1.089, P value < 0.001) and Tl (OR: 1.039, 95 %CI: 1.004–1.075, P value = 0.031) exposure were positively associated with AGM. In the weighted quantile sum (WQS) regression model, the positive index was obviously associated with AGM (OR: 1.358, 95 %CI: 1.007–1.832, P value = 0.045). In the least absolute shrinkage and selection operator (LASSO) regression model, Cd and Tl were selected as the most contributors. In the Bayesian kernel machine regression (BKMR) model, the effect of co-exposure to metal mixture was associated with AGM, and Cd exposure showed a significantly positive trend. In conclusion, Cd and Tl exposure exhibited independent positive effects on AGM among metal mixture exposure, consistent with their effects on prediabetes.

Keywords

• Metals
• Co-exposure
• Abnormal glucose metabolism
• Diabetes
• Prediabetes
• Machine learning