Non-essential metal mixture and depressive symptoms in chinese older adults: cross-sectional associations and exploratory role of interleukin-6.

Exposure to non-essential metals (NEMs) has been linked to depressive symptoms, but evidence in older adults and the potential role of inflammatory pathways remain limited, particularly for metal mixture. This cross-sectional study aimed to evaluate individual and joint associations of six blood NEMs with depressive symptoms and to exploratorily assess whether interleukin-6 (IL-6) statistically accounts for part of the observed associations. Data were derived from 4076 community-dwelling Chinese older adults in a cross-sectional survey. Six NEMs (barium, cadmium, aluminum, gallium, mercury, thallium) were quantified by ICP-MS. Depressive symptoms were defined as a Geriatric Depression Scale score ≥ 5. Logistic regression and restricted cubic splines were used for single-metal analyses. Weighted quantile sum regression, quantile-based g-computation, and Bayesian kernel machine regression were applied to assess the joint mixture effect. Exploratory mediation analyses within a counterfactual framework estimated the proportion of the statistical association explained by IL-6. After full adjustment, thallium (Tl) was significantly associated with higher odds of depressive symptoms (OR = 1.19, 95% CI: 1.08-1.31). Mercury (Hg) showed a borderline positive association after false discovery rate correction (OR = 1.10, 95% CI: 1.01-1.21; FDR = 0.104). All three mixture methods consistently indicated a positive joint effect of the NEM mixture, with Tl and Hg as the main contributors. In exploratory mediation analyses, IL-6 statistically accounted for 17.04% (95% CI: 6.23%-65.46%) of the Hg-depressive symptoms association and 12.34% (95% CI: 3.83%-65.68%) of the NEM mixture-depressive symptoms association. In this study of older adults, exposure to NEMs, particularly Tl and Hg, was associated with higher odds of depressive symptoms. IL-6 statistically explained a modest proportion of this association. These hypothesis-generating findings require validation in longitudinal studies with temporally ordered measurements.