Study population
It was a prospective cohort study of participants aged over 40 years from Jiading District in Shanghai, China. The recruitment and baseline evaluation were completed in 2010 and the follow-up examination was conducted in 2014. A detailed description of the study design, eligibility criteria and sampling has been published [14]. Among 10,375 individuals examined at baseline, we firstly excluded those who had diagnosed diabetes (n = 2000), had CVD (n = 206), did not complete baseline glycemic measurements or subclinical atherosclerosis detection (n = 381), leaving 7788 individuals for analysis. Then, we excluded those who did not complete follow-up glycemic measurements or subclinical atherosclerosis detection (n = 3049). Thus, a total of 4739 subjects were eventually included in the analysis with a median follow-up duration for 4.3 years. All of them did not take medication that could influence plasma glucose. In the additional analysis of CIMT, we included 4644 individuals who completed CIMT measurements at baseline and follow-up.
The study protocol was approved by the Institutional Review Board of Ruijin Hospital affiliated to the Shanghai Jiaotong University School of Medicine. Written informed consent was obtained from each participant.
Baseline data collection
Face-to-face interviews were performed at baseline and in the follow-up period by trained personnel using a standard questionnaire to collect information on socioeconomic characteristics, lifestyle factors, medical history and current use of medications. Current smokers or drinkers were defined as those who had smoked cigarettes or consumed alcohol regularly in the past 6 months.
Measurements of body weight, height and waist circumstance were performed by trained nurses according to standard protocols. Body mass index (BMI) was calculated by dividing weight (kg) by height (m) squared. The blood pressure measurement was performed on the non-dominant arm, using an automated electronic device (OMRON Model HEM-752 FUZZY). Three blood pressure measurements were taken with participants in a seated position after 5 min of quiet rest and the mean value of three measurements was used in analysis.
All the participants were required to fast for ≥ 10 h before their visits and underwent a 75 g OGTT and blood samples were collected at 0 and 2 h during the test. FPG and 2 h- PPG were measured within 2 h after blood sample collection. HbA1c level was determined through High Performance Liquid Chromatography (HPLC) (BIO-RAD, Hercules, CA, USA). Total cholesterol, low density lipoprotein-cholesterol (LDL-C), high density lipoprotein-cholesterol (HDL-C), triglycerides (TG) and serum uric acid were measured using the chemiluminescence method with an auto-analyzer (Modular E170; Roche, Basel, Switzerland). Estimated glomerular filtration rate (eGFR) was assessed based on the Chronic Kidney Disease Epidemiology Collaboration formula for White/other (except Black) expressed in milliliters per minute per 1.73 m2: (1) females: Cr ≤ 0.7 mg/dL, eGFR = 144 × (Cr/0.7)−0.329 × (0.993) age; Cr > 0.7 mg/dL, eGFR = 144 × (Cr/0.7)−1.209 × (0.993)age; and (2) males: Cr ≤ 0.9 mg/dL, eGFR = 141 × (Cr/0.9)−0.411 × (0.993)age; Cr > 0.9 mg/dL, eGFR = 141 × (Cr/0.9)−1.209 × (0.993)age. The Triglyceride–glucose index (TyG index) was calculated as ln [TG (mg/dL) × FPG (mg/dL)/2] [15].
All the participants underwent branchial-ankle pulse wave velocity (baPWV) measurements, with baPWV values determined by Colin VP-1000 (model BP203RPEII, form PWV/ABI; Omron Colin Medical Instruments, Tokyo, Japan). Briefly, participants attached cuffs around both arms and ankles after having rested for 10–15 min at room temperature (25 °C). Measurements from the brachial and tibial arteries were obtained simultaneously. Transit time, the time interval between the initial increase in brachial and tibial waveforms, and transit distance between the arm and ankle were measured. The value of baPWV was calculated as the transit distance divided by the transit time. We adopted the mean value of the right and left common baPWV for analysis. CIMT measurements were performed by an experienced sonographer using a high-resolution B-mode tomographic ultrasound system (Esaote Biomedica SpA, Italy) with a linear 7.5-MHz transducer. The CIMT was measured on the far wall of the right and left common carotid arteries, 1.5 cm proximal to the bifurcation. The distance between the leading edge of the first echogenic line and that of the second echogenic line at the end of diastole was calculated as the CIMT of either side. The larger value of the right and left CIMT was used for analysis.
Definitions of prediabetes at baseline
In this population without diagnosed diabetes at baseline, we categorized participants according to ADA 2021 definitions for prediabetes based on FPG levels (100–125 mg/dL) and/or 2 h-PPG levels (140–199 mg/dL) and/or HbA1c levels (5.7–6.4%) [16]. We also defined prediabetes as having elevated HbA1c or impaired fasting glucose (IFG) or impaired glucose tolerance (IGT). Those whose FPG < 100 mg/dl and 2 h-PPG < 140 mg/dL and HbA1c < 5.7% were categorized to normal glucose regulation (NGR).
Data collection at follow-up visit
In the follow-up examination, glucose parameters, baPWV and CIMT measurements were performed using the same protocol used during the baseline examination. The development of subclinical atherosclerosis was defined as new-onset increased baPWV or increased CIMT. We calculated the delta-baPWV as follow-up baPWV minus baseline baPWV, and new-onset increased baPWV was defined as having delta-baPWV over the 90th percentile. The increased CIMT was defined using the same method.
Statistical analysis
Baseline characteristics were demonstrated according to the age and glycemic status. Continuous variables were presented as medians (interquartile ranges) and categorical variables were presented as numbers and proportions.
Odds ratios (ORs) and their corresponding 95% confidence intervals (CIs) for prediabetes status in relation to incident increased baPWV or incident increased CIMT in age groups were evaluated through multivariate logistic regression analyses. The ORs of prediabetes were calculated compared to normal glucose regulation. The ORs of IFG were calculated compared to normal FPG, and those of IGT were estimated compared to normal 2 h-PPG, and those of elevated HbA1c were calculated compared to normal HbA1c. Model 1 adjusted for age and sex, and Model 2 adjusted for age, sex, systolic blood pressure (SBP), diastolic blood pressure (DBP), BMI, TG, LDL-C, HDL-C, uric acid, eGFR, current smoking, drinking and anti-hypertensive drug. The interaction term between prediabetes status and age was used to obtain the P for interaction and the young-to-old ratio of odds ratios (RORs). To further evaluate the influence of sex, blood pressure, BMI, and TyG index on the age-specific associations, we performed logistic regression analyses in subgroups. The four subgroups were divided respectively by men or women, with hypertension or without hypertension, with BMI < 24 kg/m2 or with BMI ≥ 24 kg/m2, with TyG index < 4.64 (defined by the median of TyG index) or with TyG index ≥ 4.64.
Potential nonlinear associations between the levels of FPG, 2 h-PPG and HbA1c and the ORs of increased baPWV in different age groups were examined with restricted cubic splines [17]. Analyses adjusted for variables in Model 2, and 4 knots were located for each of the three glycemic measurements. Tests for nonlinearity, which compared a model containing only the linear term with a model containing the linear and restricted cubic spline terms were conducted by using likelihood ratio tests.
Repeated baPWV and glycemic measurements constituted a typical cross-lagged panel design [18, 19]. This design measured the effect size of baseline glycemic measurements on subsequent baPWV and the effect size of baseline baPWV on subsequent glycemic measurements simultaneously, adjusting for the auto-regressive effects. This analysis was performed in the total cohort and in age subgroups. Baseline and follow-up baPWV and glycemic measurements were adjusted for potential confounders before the cross-lagged analysis, including age, sex, SBP, DBP, BMI, TG, LDL-C, HDL-C, eGFR, uric acid, current smoking, drinking, and anti-hypertensive drug. We standardized both baPWV and glycemic measurements to means as 0 and SD as 1, because the ranges of the variates were different. Statistical difference in the effect size was examined using t test. Considering that those with abnormal glucose were excluded in the analysis, we performed a sensitivity analysis and further excluded those with baseline baPWV higher than the 95th percentile in the cross-lagged panel model.