This study was part of the Tongji Maternal and Child Health Cohort (TMCHC) study, which was a prospective cohort study designed to investigate the association between maternal diet during pregnancy and the outcomes of mother-offspring pairs in Wuhan, China. Pregnant women attending their antenatal care appointment before 16 weeks of gestation in one of three public hospitals in Wuhan, between January 2013 and May 2016 were recruited to join the cohort . All procedures and study protocols were approved by the Ethics Review Committee of Tongji Medical College of Huazhong University of Science and Technology (NO. 201302) and written informed consent was obtained at enrollment.
In the current analysis, we included TMCHC participants met the following criteria: (1) with a singleton pregnancy, (2) completed an oral glucose tolerance test (OGTT), (3) had complete information on anthropometric measurements, blood glucose and lipid profiles (TG, total cholesterol [TC], HDL-C and low-density lipoprotein cholesterol [LDL-C] ) during 15-19 gestational weeks. Women without information on blood pressure at enrollment (n=6) were excluded. Women with known diseases prior to pregnancy (e.g., chronic viral hepatitis, hypertension, diabetes) (n=47) or fasting plasma glucose ≥7.0 mmol/L at first prenatal visit (n=38) were excluded. Those with the outliers of fasting plasma glucose concentrations and/or lipid profiles (n=63) were excluded. Finally, a total of 1747 women were involved (Fig. 1).
Data collection and measurement
Basic information was obtained by questionnaire in an in-person interview at enrollment. The content of the standardized self-reported questionnaire included age, education level, average personal income, lifestyle (including smoking, alcohol habits and physical activity), medical history, and family history of diseases including diabetes, hypertension and hyperlipidemia. The educational level was classified according to the number of completed academic years. Maternal height was measured without shoes at enrollment. Their weight before pregnancy was self-reported, and current weight was measured by our trained staffs in each regular antenatal examination. Pre-pregnancy BMI was calculated as pre-pregnancy weight in kilograms divided by the square of the height in meters, and was classified as underweight (pre-pregnancy BMI＜18.5), normal (18.5≤pre-pregnancy BMI＜24.0) and overweight/obese (pre-pregnancy BMI≥24.0) . Alcohol consumers were defined as those consuming alcohol before pregnancy or currently. Participants were defined as smokers if they smoked before or during pregnancy. Physical activity during pregnancy was assessed and classified according to whether they take moderate or vigorous intensity exercise more than 3 times per week for more than 30 min once.
Blood samples were obtained after an overnight fast. Blood glucose was measured through the glucose oxidase method, TG was measured with GPO-PAP method, TC was measured with CHOD-PAP method, HDL-C was measured with catalase scavenging method and LDL-C was measured with surfactant scavenging method using commercially available kits (Biosino, Bio-technology and Science Inc. Beijing, China), by Mindray BS-200 automatic biochemical analyzer (Shenzhen, China). METS-IR was calculated by applying the modified formula, as Ln(2×FPG[mg/dl]+TG[mg/dl])×BMI[kg/m2]/Ln(HDL-C[mg/dl]) . Considering progressive weight gain during pregnancy, we used “pre-pregnancy BMI×current gestational weight/ideal gestational weight” instead of the BMI in the original formula to take in account the influence of both pre-pregnancy BMI and gestational weight gain. The ideal gestational weight was calculated using pre-pregnancy weight plus the ideal weight gain which was determined by recommendation of Institute of Medicine (IOM), based on the gestational age of blood testing and pre-pregnancy BMI of each participant .
A 75-g 2-h OGTT was performed during 24 and 28 gestational weeks. According to the criteria recommended by IADPSG, GDM was diagnosed when fasting plasma glucose ≥5.1 mmol/L, or 1-h plasma glucose ≥10.0 mmol/L, or 2-h plasma glucose ≥8.5 mmol/L .
All continuous variables were described as median (interquartile range) because of skewed distribution, and categorical variables were presented as number (percentage). The Kruskal-Wallis H tests and chi-square test were used to compare continuous and categorical variables between groups, respectively.
Participants were classified into 4 groups based on quartiles of METS-IR in pre-pregnancy normal-weight women. A modified Poisson regression model  was used to estimate risk ratio (RR) and 95% confidence interval (CI) for the association of METS-IR during early midpregnancy with incident GDM. In adjusted models, we controlled potential confounders including age, parity, family history of diabetes, family history of hypertension, family history of hyperlipidemia, systolic blood pressure (SBP), diastolic blood pressure (DBP), gestational week at blood sampling, TC and LDL-C. Stratified and interaction analyses were also conducted. Statistical analyses were done using SAS 9.4 (SAS Inst., Cary, NC, USA). Two-sided P＜0.05 was considered statistically significant.