Associations of circulating choline and its related metabolites with cardiometabolic biomarkers: an international pooled analysis

Abstract

Background: Choline is an essential nutrient; however, the associations of choline and its related metabolites with cardiometabolic risk remain unclear.

Objective: We examined the associations of circulating choline, betaine, carnitine, and dimethylglycine (DMG) with cardiometabolic biomarkers and their potential dietary and nondietary determinants.

Methods: The cross-sectional analyses included 32,853 participants from 17 studies, who were free of cancer, cardiovascular diseases, chronic kidney diseases, and inflammatory bowel disease. In each study, metabolites and biomarkers were log-transformed and standardized by means and SDs, and linear regression coefficients (β) and 95% CIs were estimated with adjustments for potential confounders. Study-specific results were combined by random-effects meta-analyses. A false discovery rate <0.05 was considered significant.

Results: We observed moderate positive associations of circulating choline, carnitine, and DMG with creatinine [β (95% CI): 0.136 (0.084, 0.188), 0.106 (0.045, 0.168), and 0.128 (0.087, 0.169), respectively, for each SD increase in biomarkers on the log scale], carnitine with triglycerides (β = 0.076; 95% CI: 0.042, 0.109), homocysteine (β = 0.064; 95% CI: 0.033, 0.095), and LDL cholesterol (β = 0.055; 95% CI: 0.013, 0.096), DMG with homocysteine (β = 0.068; 95% CI: 0.023, 0.114), insulin (β = 0.068; 95% CI: 0.043, 0.093), and IL-6 (β = 0.060; 95% CI: 0.027, 0.094), but moderate inverse associations of betaine with triglycerides (β = -0.146; 95% CI: -0.188, -0.104), insulin (β = -0.106; 95% CI: -0.130, -0.082), homocysteine (β = -0.097; 95% CI: -0.149, -0.045), and total cholesterol (β = -0.074; 95% CI: -0.102, -0.047). In the whole pooled population, no dietary factor was associated with circulating choline; red meat intake was associated with circulating carnitine [β = 0.092 (0.042, 0.142) for a 1 serving/d increase], whereas plant protein was associated with circulating betaine [β = 0.249 (0.110, 0.388) for a 5% energy increase]. Demographics, lifestyle, and metabolic disease history showed differential associations with these metabolites.

Conclusions: Circulating choline, carnitine, and DMG were associated with unfavorable cardiometabolic risk profiles, whereas circulating betaine was associated with a favorable cardiometabolic risk profile. Future prospective studies are needed to examine the associations of these metabolites with incident cardiovascular events.

Keywords: betaine; biomarkers; cardiometabolic disease; carnitine; choline; dimethylglycine.

FIGURE 1

Heat map for the associations of choline and related metabolites with cardiometabolic biomarkers. Linear regression coefficients (β) were adjusted for age; sex; ethnicity; fasting time; education; obesity; central obesity; smoking status; alcohol drinking; physical activity level; use of multivitamins; menopausal status and hormone therapy in women; intakes of red meat, eggs, and fish; and history of diabetes, hypertension, dyslipidemia, and nonalcoholic fatty liver disease. βs indicate the increase or decrease in SD units of choline, betaine, carnitine, and DMG on the log scale. Statistical significance is indicated by false discovery rate–adjusted P values (Q values) <0.05. CRP, C-reactive protein; DBP, diastolic blood pressure; DMG, dimethylglycine; HbA1c, glycated hemoglobin; SBP, systolic blood pressure.

FIGURE 2

Circulating choline (A), betaine (B), carnitine (C), and DMG (D) in relation to cardiometabolic biomarkers. Regression coefficients (β) and 95% CIs were adjusted for age; sex; ethnicity; fasting time; education; obesity; central obesity; smoking status; alcohol drinking; physical activity level; use of multivitamins; menopausal status and hormone therapy in women; intakes of red meat, eggs, and fish; and history of diabetes, hypertension, dyslipidemia, and nonalcoholic fatty liver disease. βs indicate the increase or decrease in SD units of choline, betaine, carnitine, and DMG on the log scale. Q values represent corrected P values for each group of analyses by controlling the false discovery rate. CRP, C-reactive protein; DBP, diastolic blood pressure; DMG, dimethylglycine; HbA1c, glycated hemoglobin; SBP, systolic blood pressure.

FIGURE 3

Circulating choline in relation to dietary (A) and nondietary (B) factors. Regression coefficients (β) and 95% CIs were adjusted for age; sex; ethnicity; fasting time; education; obesity; central obesity; smoking status; alcohol drinking; physical activity level; use of multivitamins; menopausal status and hormone therapy in women; intakes of red meat, eggs, and fish; history of diabetes, hypertension, dyslipidemia, and NAFLD; and total energy. Dietary covariates were mutually adjusted for other foods and included in the model as the cohort- and sex-specific quintiles. Total energy intake was additionally adjusted for in diet-related analyses. βs indicate the increase or decrease in SD units of choline on the log scale. Q values represent corrected P values for each group of analyses by controlling the false discovery rate. NAFLD, nonalcoholic fatty liver disease; WHR, waist-to-height ratio.

FIGURE 4

Circulating betaine in relation to dietary (A) and nondietary (B) factors. Regression coefficients (β) and 95% CIs were adjusted for age; sex; ethnicity; fasting time; education; obesity; central obesity; smoking status; alcohol drinking; physical activity level; use of multivitamins; menopausal status and hormone therapy in women; intakes of red meat, eggs, and fish; history of diabetes, hypertension, dyslipidemia, and NAFLD; and total energy. Dietary covariates were mutually adjusted for other foods and included in the model as the cohort- and sex-specific quintiles. Total energy intake was additionally adjusted for in diet-related analyses. βs indicate the increase or decrease in SD units of betaine on the log scale. Q values represent corrected P values for each group of analyses by controlling the false discovery rate. NAFLD, nonalcoholic fatty liver disease; WHR, waist-to-height ratio.

FIGURE 5

Circulating carnitine in relation to dietary (A) and nondietary (B) factors. Regression coefficients (β) and 95% CIs were adjusted for age; sex; ethnicity; fasting time; education; obesity; central obesity; smoking status; alcohol drinking; physical activity level; use of multivitamins; menopausal status and hormone therapy in women; intakes of red meat, eggs, and fish; history of diabetes, hypertension, dyslipidemia, and NAFLD; and total energy. Dietary covariates were mutually adjusted for other foods and included in the model as the cohort- and sex-specific quintiles. Total energy intake was additionally adjusted for in diet-related analyses. βs indicate the increase or decrease in SD units of carnitine on the log scale. Q values represent corrected P values for each group of analyses by controlling the false discovery rate. NAFLD, nonalcoholic fatty liver disease; WHR, waist-to-height ratio.

FIGURE 6

Circulating DMG in relation to dietary (A) and nondietary (B) factors. Regression coefficients (β) and 95% CIs were adjusted for age; sex; ethnicity; fasting time; education; obesity; central obesity; smoking status; alcohol drinking; physical activity level; use of multivitamins; menopausal status and hormone therapy in women; intakes of red meat, eggs, and fish; history of diabetes, hypertension, dyslipidemia, and NAFLD; and total energy. Dietary covariates were mutually adjusted for other foods and included in the model as the cohort- and sex-specific quintiles. Total energy intake was additionally adjusted for in diet-related analyses. βs indicate the increase or decrease in SD units of DMG on the log scale. Q values represent corrected P values for each groups of analyses by controlling the false discovery rate. DMG, dimethylglycine; NAFLD, nonalcoholic fatty liver disease; WHR, waist-to-height ratio.