Study Population and Biomedical Measurements
A total of 501 participants were recruited from the Department of Cardiology, Zhongshan Hospital Shanghai China, between March 2017 and June 2018, including 254 cases and 247 controls. Cases were clinically diagnosed as ACS with documented ≥ 50% stenosis of at least one epicardial coronary artery during the coronary angiography (11, 12), and controls were non-ACS participants who underwent coronary angiography with a normal coronary artery. Each participant provided written informed consent. The study was approved by the Ethics Committee of the Zhongshan Hospital Fudan University.
Data on demographics (age and gender), anthropometrics (height and weight), lifestyle (smoking), history of disease (i.e., hypertension, diabetes mellitus and hyperlipidemia), kidney function [estimated glomerular filtration rate (eGFR)] were collected from the electronic medical record systems in Department of Cardiology. The severity of coronary atherosclerosis was assessed using the Gensini score, a widely used scoring system to quantify coronary atherosclerosis burden, in which a zero score indicates absence of atherosclerotic disease, and a higher score accounts for a severer proximal lesion by combining the degree of luminal narrowing as well as the location of narrowing(13, 14).
Fasting blood samples were collected using tubes containing EDTA via radial access before heparinization and then immediately stored at -80°C until analysis. Plasma proteins were precipitated with 3 volumes of methanol containing a mixed internal standard of 500ng/ml. After vortex and centrifugation, supernatants were analyzed with an Agilent 1290 Infinity UHPLC instrument (Agilent, USA) on an XBridge BEH Hilic Column (2.5μm,2.1×100mm, Waters, Milford, MA) at a flow rate of 0.35ml/min. LC gradient was starting from 2% 10mM ammonium formate (A) and 98% acetonitrile of (B, PH3.5) over 1min, then increased to 10% A at 6min, holding 1min; then to 15% A at 10min, 30% A at 12min and 40% A at 13min. The supernatants of 20 samples were randomly taken out and mixed as Sample-Quality Control samples to calculate intra-day relative standard deviations and inter-day relative standard deviations, which were all <15% in all the samples for the measured metabolites.
An Agilent 6470 Triple Quadruple (Agilent, USA) equipped with ESI source was used for quantification of TMA, TMAO, choline, and betaine. All the compounds were monitored in positive MRM mode using characteristic precursor-product ion transitions: m/z 60.1-44.2, m/z 76.1-58.1, m/z 104.1-60.1, m/z 118.1-58.1, m/z 162.1-102.9, m/z 114.0-44.1, respectively. The internal standards TMA-d9, TMAO-d9, and Choline-d9 were added to blood samples, and monitored in MRM mode at m/z69.1-49.2, m/z84.9-68.4, and m/z113.1-69.5, respectively. Series concentrations of TMA, TMAO, choline, and betaine standards and a fixed amount of internal standards were spiked into the water to prepare the curves for quantification of blood analytes. L-QC, M-QC and H-QC were inserted into the sequence to evaluate the accuracy of the method.
The standards for TMA, TMAO, choline and betaine were purchased from Sigma-Aldrich (Shanghai, China). The internal standards for TMA-d9 and TMAO-d9 were obtained from Cambridge Isotope Laboratories, Inc (MA, USA).
Statistical analyses
A rank-based inverse normal transformation was applied to approximate the normal distribution of metabolites concentrations(15). Characteristics were presented as mean (SD) for continuous variables and number (frequencies) for categorical variables. Characteristics in cases and controls were compared using the t-test for continuous variables and χ2 -test for categorical variables. Cases were stratified into three groups according to their Gensini scores. We calculated a metabolite score as the weighted sum of levels of four metabolites from the choline pathway: betaine, choline, TMA and TMAO, and modeled the score as the main exposure variable in logistic regressions to estimate the composite association of circulating metabolites in choline pathway with ACS (16). The weight for each metabolite was the regression coefficient for one SD increment in the blood concentration estimated from the adjusted multivariable logistic regression model. A ratio of betaine to choline was also calculated and modeled as an exposure variable in the regression models. This ratio can be considered as a better predictor of metabolic stress as it combines the predictive power of betaine and choline to metabolic stress together, and it was able to capture the composite associations of betaine and choline with metabolic disturbances(17).
Multivariable (adjusted) logistic regression models were used to evaluate the odds ratios (ORs) and corresponding 95% confidence intervals (CIs) to estimate the association of circulating metabolites and the metabolite score with the odds of ACS. Circulating metabolites were analyzed as both quartiles (using cut-points defined among controls) and continuous variables (per 1-SD increment). To test the linear trend across quartiles, the median of each quartile was assigned and analyzed as a continuous variable. Logistic regression models were adjusted for age, sex, smoking index (pack-years) and body mass index (BMI) in model 1, model 2 was additionally adjusted for history of the disease (ie., hypertension, diabetes mellitus and hyperlipidemia), and model 3 was further adjusted for kidney function. We used participants assigned to the first quartile of the levels of each metabolite, the metabolite score and betaine to choline ratio, as the reference group in each model. The correlations between circulating metabolites, metabolite score and betaine-to-choline ratio were tested by Spearman correlation. In ACS cases, the relations between circulating metabolites and the coronary atherosclerosis burden measured by Gensini score were also measured.
All statistical analyses and data visualizations were performed using R 3.6.2 (https://www.r-project.org/) and a two-sided p value<0.05 was considered statistically significant.