Longitudinal triglyceride-glucose index and Metal Interactions: A multi-model cohort integrating synergistic and...

Longitudinal dynamics of metal exposures on insulin resistance quantified by the triglyceride-glucose (TyG) index-remain poorly characterized. We aim to explore the combined and interactive effects of metals on TyG index among Chinese elderly. This prospective cohort included 2732 Chinese elders with baseline urinary concentrations of metals quantified via ICP-MS. Analyses employed a multi-model framework beginning with linear mixed-effects models that incorporated random intercepts to assess metal-time interactions, followed by weighted quantile sum regression to evaluate metal mixture effects, and culminating in extended linear mixed models that computed the interaction effects for all pairwise metal-metal-time combinations. Single-metal analysis identified zinc (Zn), molybdenum (Mo), arsenic (As), strontium (Sr), and copper (Cu) as positively associated with the TyG index, while iron (Fe) and cadmium (Cd) were inversely associated. The positive associations for Zn, Mo, As, Sr, and Cu attenuated over time, whereas the inverse association for Fe strengthened. The Weighted Quantile Sum (WQS) model indicated an overall positive mixture effect on the TyG index, primarily driven by Zn and Mo, with Fe acting as an antagonist. Ternary interaction analysis among 78 metal pairs revealed 28 significant metal × metal × time interactions, with vanadium (V) featuring in several key antagonistic combinations. Sex-stratified analyses showed modification of these associations, and findings remained robust in sensitivity analyses. In summary, individual and mixed metal exposures are longitudinally associated with insulin resistance in the elderly, characterized by time-dependent and interactive effects, highlighting the necessity of considering complex exposure dynamics and susceptibility in risk assessment.