Research design and study patients
The study patients and methods were described previously . From 27 May 2015 to 3 August 2016, a total of 71,020 patients were hospitalized in Liaoning Medical University First Affiliated Hospital (LMUFAH). Among them, 1032 consecutive patients were diagnosed with T2D and had complete information on height, weight, and blood pressure. Non-T2D subjects were recruited from hospital’s physical examination center where individuals had regular health examination. One thousand five hundred twenty-two of them had metabolomic profiles measured and were included in this analysis. All subjects were over 18 years old. T2D was diagnosed according to the 1999 World Health Organization’s criteria or use of anti-diabetic drugs . The Ethics Committee for Clinical Research of LMUFAH approved the ethics of the study, and informed consent was waivered by the above ethics committee due to the retrospective nature of the study, which is consistent with the Declaration of Helsinki.
Data collection and definitions
Clinical information including demography, anthropometry, laboratory parameters, medications, and disease status was extracted from electronic medical system (EMS) retrospectively. Age was calculated automatically by EMS as difference between admission year and birth year. Body mass index (BMI) was calculated as weight in kilograms divided by squared height in meters. According to the recommended Chinese criteria , BMI was categorized into four classes: underweight (< 18.5 kg/m2), normal weight (18.5 to 24 kg/m2), overweight (> 24 kg/m2), and obesity (≥ 28 kg/m2). Systolic blood pressure (SBP) over 140 mmHg, high-density lipoprotein cholesterol (HDL-C) less than 1 mmol/L in men and 1.3 mmol/L in women, triglyceride over 1.7 mmol/L, low-density lipoprotein cholesterol (LDL-C) over 2.6 mmol/L were defined as abnormal . Use of oral anti-diabetic drugs and insulin, angiotensin-converting enzyme inhibitors (ACEIs), angiotensin receptor blockers (ARBs), and other anti-hypertensive drugs, statins, and other lipid-lowering drugs in hospital was documented. Complications were extracted from the electronical database, including coronary artery disease, stroke, diabetic retinopathy, and diabetic nephropathy.
Measurements of serum asparagine and aspartate
Details of the amino acid assessment method were published previously . Eight-hour fasting blood was taken and stored as dried blood spot. Eight-hour fasting blood was mostly collected in the morning. As no reliable evidence has suggested circadian rhythm in conversion of asparagine to aspartate, time of sampling was unlikely to affect the levels of amino acids. Amino acids quality control (QC) standards used in our study were provided by Chromsystems (Grafelfing, Germany). LC-MS/MS analysis was performed with AB Sciex 4000 QTrap system (AB Sciex, Framingham, MA, USA). Analyst v1.6.0 software (AB Sciex) was used for system control and data collection. ChemoView 2.0.2 (AB Sciex) was used for data preprocessing.
We first compared difference of clinical and biochemical characteristics of participants between T2D and non-T2D. For continuous variables, Q-Q plot was used to checked normality. Data with normal distribution was expressed as means (standard deviations) while data with skewed distribution was expressed as medians (interquartile ranges). Non-paired Student’s t test (or Mann-Whitney U test when appropriate) was used to measure the difference; for categorical variables, data was presented as frequencies (percentage) and chi-square test (or fisher test if appropriate) was used for difference comparison.
Age was stratified into a binary variable at 50 years, and > 50 years of age in female roughly represented the postmenopausal period. Restricted cubic spline (RCS) nested in the logistic regression was also performed to examine the full-range associations between age with T2D risk and to ascertain the selected cut-off point at 50 years of age . The associations between asparagine or aspartate alone with T2D were examined with RCS too. Next, to visualize potential interaction between asparagine to aspartate ratio with sex or age, we also performed RCS to examine the full-range associations between asparagine to aspartate ratio with T2D in different sex and age groups which derived from cut-off points chosen above. That OR curves turned apart and became unparallel may imply an additive interaction. The point of asparagine to aspartate ratio where curves turned apart and unparallel was selected as the cut-off point to categorize the ratio for further analysis.
Then, we formally tested the additive interaction by calculating relative excess risk due to interaction (RERI), attributable proportion due to interaction (AP), and synergy index (S) . RERI > 0, AP > 0, or S > 1 indicates biological interaction. First, we used univariable and multivariable logistic regression models to obtain odds ratio (OR) and 95% confidence (CI) of asparagine, aspartate, asparagine to aspartate ratio, sex, and age (all as categorical variables) for T2D. Asparagine to aspartate ratio (> 1.5 and ≤ 1.5 μmol/L, sex, age (< 50 and ≥ 50 years, BMI (< 18.5, 18.5 ~ 24.0, 24.0 ~ 28.0 and ≥ 28.0 kg/m2), SBP (< 140 and ≥ 140 mmHg), LDL-C (< 2.60 and ≥ 2.60 mmol/L), HDL-C (< 1.00 mmol/L in male or < 1.30 mmol/L in female as low level and ≥ 1.00 in male or ≥ 1.30 in female as high level) and triglyceride (< 1.70 mmol/L and ≥ 1.70 mmol/L) were included in multivariable analysis. Through steps above, high asparagine, low aspartate, high asparagine to aspartate ratio, female and age over 50 years old were identified as risk factors of T2D (see Table 2); second, to estimate the biological interaction between asparagine to aspartate ratio and age or sex, we created four variables (see Table 2): (1) asparagine to aspartate ratio ≤ 1.5 μmol/L and age < 50 years (or male) (as reference); (2) asparagine to aspartate ratio ≤ 1.5 μmol/L and age ≥ 50 years (or female); (3) asparagine to aspartate ratio > 1.5 μmol/L and age < 50 years (or male); (4) asparagine to aspartate ratio > 1.5 μmol/L and age ≥ 50 years (or female). Confounders listed above were also adjusted in multivariable addictive models.
We further tested the second-order interaction between high asparagine to aspartate ratio and coexistence of female sex and > 50 years of age. The four variables were created as combination of copresence of female sex and > 50 years of age and high asparagine to aspartate ratio: (1) female and > 50 years of age = no plus low asparagine to aspartate ratio (use as reference); (2) female and > 50 years of age = yes plus low asparagine to aspartate ratio; (3) female and > 50 years of age = no plus high asparagine to aspartate ratio; (4) female and > 50 years of age = yes plus high asparagine to aspartate ratio (see Table 3). Likewise, RERI, AP, and S were used to measure biological interactions.
To exclude potential influence of diabetes-related parameters on association between asparagine to aspartate ratio and T2D, especially in the context of gender, we included patients without diabetes complications or use of anti-diabetic medications in the analysis and repeated logistic regression and addictive interaction analysis; to avoid some possible bias from different age distribution for each gender or different gender distribution for each age groups, we also compared male and female gender in different age groups, as well as age over and below 50 years old in each gender.
Partial Pearson correlation (data with normal distribution ) or Spearman correlation (data with skewed distribution) analysis was used to test correlations between asparagine to aspartate ratio and available diabetes traits, i.e., BMI, SBP, triglyceride, LDL-C, HDL-C, HbA1c, and duration of diabetes while adjusted for age and sex.
All analysis was performed using SAS version 9.4 (SAS institute Inc., Cary, NC, USA) and R version 3.6.0. P values of < 0.05 were considered statistically significant.