Relationships between minerals' intake and blood homocysteine levels based on three machine learning methods: a large cross-sectional study

Abstract

Background: Blood homocysteine (Hcy) level has become a sensitive indicator in predicting the development of cardiovascular disease. Studies have shown an association between individual mineral intake and blood Hcy levels. The effect of mixed minerals' intake on blood Hcy levels is unknown.

Methods: Data were obtained from the baseline survey data of the Shanghai Suburban Adult Cohort and Biobank(SSACB) in 2016. A total of 38273 participants aged 20-74 years met our inclusion and exclusion criteria. Food frequency questionnaire (FFQ) was used to calculate the intake of 10 minerals (calcium, potassium, magnesium, sodium, iron, zinc, selenium, phosphorus, copper and manganese). Measuring the concentration of Hcy in the morning fasting blood sample. Traditional regression models were used to assess the relationship between individual minerals' intake and blood Hcy levels. Three machine learning models (WQS, Qg-comp, and BKMR) were used to the relationship between mixed minerals' intake and blood Hcy levels, distinguishing the individual effects of each mineral and determining their respective weights in the joint effect.

Results: Traditional regression model showed that higher intake of calcium, phosphorus, potassium, magnesium, iron, zinc, copper, and manganese was associated with lower blood Hcy levels. Both Qg-comp and BKMR results consistently indicate that higher intake of mixed minerals is associated with lower blood Hcy levels. Calcium exhibits the highest weight in the joint effect in the WQS model. In Qg-comp, iron has the highest positive weight, while manganese has the highest negative weight. The BKMR results of the subsample after 10,000 iterations showed that except for sodium, all nine minerals had the high weights in the joint effect on the effect of blood Hcy levels.

Conclusion: Overall, higher mixed mineral's intake was associated with lower blood Hcy levels, and each mineral contributed differently to the joint effect. Future studies are available to further explore the mechanisms underlying this association, and the potential impact of mixed minerals' intake on other health indicators needs to be further investigated. These efforts will help provide additional insights to deepen our understanding of mixed minerals and their potential role in health maintenance.

Fig. 1

Study flowchart.

Fig. 2. WQS model regression index weights for homocysteine (figure left) and hHcy (figure right).

The WQS models were adjusted by age, sex, education levels, marriage, cigarette smoking, alcohol drinking, tea drinking, energy intake, physical activity, BMI, hypertension, coronary heart disease, diabetes, and the intake of Vb₆ and Vb₁₂. WQS weighted quantile sum, hHcy Hyperhomocysteinemia.

Fig. 3. Qg-comp model regression index weights and joint effect (95% CI) for Hcy (figure left) and hHcy (figure right).

The Qg-comp models were adjusted by age, sex, education levels, marriage, cigarette smoking, alcohol drinking, tea drinking, energy intake, physical activity, BMI, hypertension, coronary heart disease, diabetes, and the intake of Vb₆ and Vb₁₂. Qg-comp Quantile g-computation, hHcy Hyperhomocysteinemia.

Fig. 4. Qg-comp regression analysis of the relationship between minerals mixture and blood Hcy level (left) and hHcy risk(right).

The Qg-comp models were adjusted by age, sex, education levels, marriage, cigarette smoking, alcohol drinking, tea drinking, energy intake, physical activity, BMI, hypertension, coronary heart disease, diabetes, and the intake of Vb₆ and Vb₁₂. Qg-comp Quantile g-computation, hHcy Hyperhomocysteinemia.

Fig. 5. Associations between mixed minerals and Hcy (figure left) and hHcy (figure right) obtained by BKMR model.

The BKMR models were adjusted by age, sex, education levels, marriage, cigarette smoking, alcohol drinking, tea drinking, energy intake, physical activity, BMI, hypertension, coronary heart disease, diabetes, and the intake of Vb₆ and Vb₁₂. BKMR Bayesian kernel machine regression, Hcy homocysteine, hHcy Hyperhomocysteinemia.

Fig. 6. Exposure-response relationship of minerals and blood Hcy levels based on BKMR.

Univariate exposure–response functions and 95% confidence interval for each mineral on the effect of Hcy (a) and hHcy (b), with other minerals fixed at the median. The BKMR models were adjusted by age, sex, education levels, marriage, cigarette smoking, alcohol drinking, tea drinking, energy intake, physical activity, BMI, hypertension, coronary heart disease, diabetes, and the intake of Vb6 and Vb12. BKMR Bayesian kernel machine regression, Hcy homocysteine, hHcy hyperhomocysteinemia.