Serum Levels of 18 Trace Elements and Type 2 Diabetes Mellitus: A Propensity Score-Matched Case-Control Study in...

Objective: Current evidence on the association between multiple trace elements and type 2 diabetes mellitus (T2DM) remains limited, particularly regarding combined effects, nonlinear dose-response patterns, and elemental interactions. This study aimed to evaluate the associations between multiple serum trace elements and T2DM using a sequential analytical strategy. Methods: A 1:1 propensity score-matched case-control study based on sex, age, and residence was conducted (637 controls, 637 T2DM cases). Serum concentrations of 18 trace elements were measured using inductively coupled plasma mass spectrometry (ICP-MS). A sequential analytical strategy was applied: LASSO regression for variable selection; logistic regression with single-element (Model 1) and multi-element (Models 2 and 3) approaches for association assessment; weighted quantile sum (WQS) regression and Bayesian kernel machine regression (BKMR) for mixture effects; and restricted cubic spline (RCS) for dose-response patterns. Results: Lithium (Li) and tellurium (Te) showed positive associations with T2DM across multiple analytical methods, whereas boron (B), lead (Pb), and vanadium (V) showed inverse associations. WQS identified V and B as the main contributing elements. BKMR did not identify a statistically significant overall association between the mixture and T2DM. Although bivariate interaction plots suggested possible effect modification between V and Al, B, Li, Pb, and Te, formal interaction tests did not reach statistical significance. RCS showed that serum B was significantly associated with T2DM overall and exhibited a significant non-linear dose-response relationship (P-nonlinear = 0.0028). Serum V showed a linear inverse association with T2DM, and serum Li showed a linear positive association. Conclusions: Our findings suggest that Li, Te, B, and V are associated with T2DM, with the most robust evidence for Li, while the evidence for Te was less consistent. In addition, we observed a trend toward a joint association of the trace element mixture, potential effect modification at the point-estimate level between V and Al, B, Li, Pb, and Te, and distinct dose-response patterns for B, Li, and V. These findings highlight the need for further investigation of trace elements in relation to T2DM, with consideration of both individual and combined associations; future validation of these exploratory findings in larger prospective studies and independent populations is warranted.