Longitudinal metal mixture interactions and dyslipidemia in a prospective nested case control study

Dyslipidemia affects over 40% of adults aged 50 years or older and contributes substantially to cardiovascular mortality, yet prospective evidence on metal mixtures, nonlinear dose–response relationships, and metal–metal interactions in community-dwelling older adults remains limited. In this prospective nested case–control study (2019–2023) within the Yinchuan Community-Dwelling Elderly Cohort, 436 incident dyslipidemia cases and 872 age-, sex-, and BMI-matched controls were included, and 13 urinary metals were quantified by microwave-assisted acid digestion followed by inductively coupled plasma mass spectrometry (ICP-MS), expressed as µg/g creatinine. Single-metal logistic models showed positive dose–response trends for copper and cadmium (P for trend < 0.05), elevated risks for molybdenum (OR = 1.44) and arsenic (OR = 1.43) in the third quartile, and a protective association for zinc in the second quartile (OR = 0.68). Machine learning prioritized zinc, copper, cobalt, strontium, and vanadium, and SHAP revealed nonlinear patterns, including J-shaped associations for copper and vanadium and a biphasic pattern for cobalt. Weighted quantile sum regression indicated that low-concentration metal mixtures were associated with increased dyslipidemia risk, and Bayesian kernel machine regression suggested potential Cu–Sr and V–Zn interactions, which generalized additive interaction models supported on an exploratory basis; these models also suggested that exercise attenuated zinc-associated risk, whereas smoking and alcohol amplified copper and zinc toxicity and meat-based diets exacerbated strontium and cobalt risks. Collectively, metal exposures, individually and as mixtures, are associated with dyslipidemia risk in older adults through nonlinear and interactive mechanisms, supporting their potential relevance to precision public health.