Abstract Body (Do not enter title and authors here): Introduction
Heavy metals are harmful environmental pollutants found in air, water, soil, and food. Epidemiological and clinical studies have shown a strong association between heavy metal exposure and cardiovascular disease (CVD) and CVD-related mortality. While some studies have identified synergic or antagonistic interactions between heavy metals in relation to CVD risk, there is limited evidence on how these interactions affect cholesterol levels. In this study, we examine both linear and non-linear interactions between serum heavy metals and their association with total cholesterol level in a US-based population.

Method
We included adult participants from the 2021–2023 NHANES cycles with available serum cholesterol and heavy metal data (N=4,916), excluding those with missing smoking status. Laboratory measures were log-transformed and z-scored. Serum selenium, cadmium, manganese, mercury, and lead were included. Each individual metal and total heavy metal level were analyzed for its association with total serum cholesterol using survey-weighted logistic regression, adjusting for age, sex, race/ethnicity, BMI, socioeconomic status, and smoking status. Interactions between metals were assessed using both Bayesian kernel machine regression (BKMR) and linear regression with the same adjustment. A significance threshold was set at 0.05.

Result
Lead, mercury, selenium, and total heavy metals were positively associated with total serum cholesterol, while manganese showed a negative association. (Figure 1) Cadmium showed no significant relationship. BKMR identified the strongest interaction between selenium-manganese (interaction score = 0.26) (Figure 2), followed by selenium-lead (interaction score = 0.23) (Figure 3), both indicating synergistically additive effects. In contrast, linear regression detected a significant interaction only between lead-selenium, but not between manganese-selenium. These findings suggest that selenium–manganese interaction is likely non-linear, whereas the selenium–lead interaction includes a linear component.

Conclusion
This study reveals novel evidence of synergistically additive effects of serum heavy metals on serum total cholesterol, particularly selenium–manganese and selenium–lead interactions. The use of nonlinear modeling methods such as BKMR may uncover relationships that are not apparent in traditional linear models.