This study used data from the Beijing Xiaotangshan Examination Center [17], which is a large-scale longitudinal cohort study investigating the risk factors and biomarkers of cardiometabolic disorders. Beijing Xiaotangshan Examination Center included participants undergoing physical examinations from 2008, which is still ongoing and forms dynamic open cohort data with annual resurveys. In this study, a total of 12,505 participants aged 18 years or older who underwent the first comprehensive health examination without diabetes between 2008 and 2013 were primarily included as baseline. Then, we excluded 207 participants due to being underweight (n=126), being 18 years of age or younger (n=22), or lacking fasting glucose data (n=59) at baseline. Finally, 12,298 participants were selected and annually followed to incident diabetes or the end of 2019, including 8184 participants with available data on waist circumference as shown in Figure 1.

The data on demographic characteristics, lifestyle, disease history, and medication use were collected through a standard questionnaire and face-to-face interview. Educational level was categorized as illiteracy or primary school (primary), middle school or high school (secondary), and bachelor’s degree or above (third). Smoking and drinking status was divided into current or not. Physical activity was defined as having moderate or intense activity at work or during leisure time more than 4 times and 80 minutes each week. The disease history of dyslipidemia and hypertension was self-reported.

The physical examination and biochemical data were measured by the trained staff and acquired from the electronic record system of Beijing Xiaotangshan Hospital, including height, weight, waist circumference, and blood pressure. BMI was calculated as weight in kilograms divided by height in meters squared. Obesity was defined as BMI ≥28 kg/m², and abdominal obesity was defined as waist circumference ≥90 cm for male individuals and ≥85 cm for female individuals, using the standard for the Chinese population [18,19]. The concentrations of fasting glucose and glycated hemoglobin A_1c (HbA_1c) were tested using the automatic biochemical analyzers Roche Cobas C 701 and SYSMEX HLC-723G8. Diabetes was defined as fasting glucose ≥7.0 mmol/L, HbA_1c ≥6.5%, or use of any glucose-lowering medication or self-reported diagnosis history of diabetes.

Arterial stiffness level was measured by brachial-ankle pulse wave velocity (baPWV) using the Omron Colin BP-203RPE III device (Omron Health Care). Elevated arterial stiffness was defined as baPWV ≥1400 cm/s, as previously described [18,19]. The ankle-brachial index (ABI) was calculated as the pressure ratio between posterior tibial artery and brachial artery.

Arterial stiffness obesity phenotype was divided into four groups: (1) normal arterial stiffness and no obesity (NASNO), (2) normal arterial stiffness and obesity (NASO), (3) elevated arterial stiffness and no obesity (ASNO), and (4) elevated arterial stiffness and obesity (ASO). On the other hand, waist circumference is another index to inflect the accumulation of abdominal fat, which is more closely associated with the risk of diabetes. Thus, arterial stiffness abdominal obesity phenotype was divided into four groups: (1) normal arterial stiffness and no abdominal obesity (NASNAO), (2) normal arterial stiffness and abdominal obesity (NASAO), (3) elevated arterial stiffness and no abdominal obesity (ASNAO), and (4) elevated arterial stiffness and abdominal obesity (ASAO).

Baseline characteristics were described using mean (SD) and frequency (proportion) according to arterial stiffness abdominal obesity phenotype. The incidence rate and cumulative incidence of diabetes were calculated.

We used Kaplan-Meier curves to present the cumulative hazard of diabetes onset stratified by arterial stiffness obesity and arterial stiffness abdominal obesity phenotype. The effect of arterial stiffness (or obesity) on incident diabetes was evaluated stratified by obesity (arterial stiffness) status. Then, the Cox proportional hazard model was used to explore the longitudinal association of arterial stiffness abdominal obesity phenotype with incident diabetes. Hazard ratios (HRs) and 95% CIs were calculated in the following models: model 1 was primarily adjusted for age and sex, and model 2 was further adjusted for education level, physical activity, current smoking, current drinking, dyslipidemia or not, hypertension or not, mean arterial pressure (MAP), and fasting glucose concentration. We adjusted the level of MAP in the regression analysis as MAP was more dependent on baPWV level compared to systolic blood pressure and diastolic blood pressure. The interaction effect between arterial stiffness and abdominal obesity was tested as a multiplicative term. We performed multiple sensitivity analyses, including additionally adjusting HbA_1c levels, excluding participants using antihypertensive medication, or repeating analyses among participants with an ABI >0.9. To address the dynamic changes of arterial stiffness abdominal obesity phenotype between baseline and follow-up, we repeated the analyses among participants of stable arterial stiffness abdominal obesity status. In addition, the effect of arterial stiffness abdominal obesity phenotype on diabetes onset was explored in subgroups of sex, age, and baseline fasting glucose level.

All statistical analyses were performed using R software (version 4.1.0; R Foundation for Statistical Computing), and a 2-sided P value <.05 was considered statistically significant.

This study was conducted in accordance with the principles of the Declaration of Helsinki and approved by the Ethics Committee of the Beijing Xiaotangshan Center (XTS021431). The study data were anonymous. All participants provided written informed consent before taking part in this study.

In this study, we proposed the concept of arterial stiffness obesity phenotype using noninvasive and simple measurements. We found that arterial stiffness obesity phenotype could improve the risk stratification of diabetes onset. People with ASO had the highest risk of incident diabetes, independent of their baseline glucose level. The results remained consistent among multiple sensitivity analyses and subgroups of sex, age, and fasting glucose level.

Many studies have suggested a relationship between arterial stiffness and diabetes, indicating that arterial stiffness could predict the development of diabetes [5,6,8]. A Kailuan study of 14,159 participants found that arterial stiffness appeared to precede the increase in fasting glucose [7]. Notably, the independent association between arterial stiffness and diabetes was not consistent among all studies. A large prospective observational study of Japanese company employees found that atherosclerosis was observed only in those with diabetes combined with hypertension, and the increased arterial stiffness may not be associated with the onset of diabetes or prediabetes alone [20]. On the other hand, obesity is an important risk factor involved in the etiopathogenesis of diabetes and is the most important culprit of insulin resistance [21,22]. In a meta-analysis in the United States and Europe, men with obesity had a 7-fold higher risk of diabetes, and women with obesity had a 12-fold higher risk [23]. A longitudinal study from the United Kingdom investigated 369,362 participants aged between 2 and 15 years and found that most of the patients with diabetes were obese (47.1%) [24]. Similar results have been found in other countries [25]. Studies have shown that the distribution of adipose tissue is a key factor in the development of insulin resistance, independent of the stage of obesity [26]. A growing body of evidence suggests that obesity cannot be assessed by BMI. Individuals of normal weight but with excess visceral adipose tissue are at high risk of diabetes, while individuals with obesity who can expand their subcutaneous adipose tissue mass, especially in the hip and femoral regions, may be at much lower risk than expected [11]. Thus, both excess body weight (BMI) and ectopic fat (such as abdominal obesity) determine the risk of diabetes onset. This study supplemented the evidence about the combined effect of arterial stiffness and obesity or abdominal obesity on the incident diabetes using a larger cohort. Compared with the ideal vascular function and nonobese group, the highest risk of diabetes was observed in the elevated arterial stiffness with obesity or abdominal obesity group. Additionally, there exists a potential interaction effect between general obesity and arterial stiffness on diabetes onset.

Metabolic health obesity phenotype has been proposed as a health index [27]. In general, people with different metabolic health obesity phenotypes have a differentiated risk of type 2 diabetes, cardiovascular diseases, and all-cause mortality [28-32]. The number and severity of metabolic abnormalities could further identify the risk of adverse outcomes [33,34]. In terms of diabetes, data suggest that the risk of developing diabetes is 5-20 times higher in metabolically unhealthy people with obesity than in the metabolically healthy nonobese group, and the risk is 4 times higher for the metabolically healthy obese group [35]. However, the definition of metabolic health status needs physical examinations (blood pressure), blood test (glucose and lipid), and a questionnaire survey (disease history and medication use). In this study, we proposed the concept of arterial stiffness abdominal obesity phenotype to stratify people at different risks of diabetes onset. The measurements of arterial stiffness and obesity status are simple, fast, and noninvasive, which enhance the applicability of arterial stiffness abdominal obesity phenotype in the risk stratification and management of diabetes.

Potential mechanisms linking arterial stiffness and diabetes include endothelial dysfunction, chronic inflammation, oxidative stress, microvascular dysfunction, and shared genetic background [7,8,36]. First, endothelial dysfunction is associated with arterial stiffness [37]. Arterial stiffness may lead to increased arterial pulse pressure and pulsation shear, resulting in endothelial dysfunction and metabolic dysregulation [38]. It has been suggested that endothelial dysfunction can cause the development of diabetes, and there is a common pathway that may link arterial stiffness and endothelial dysfunction to the development of diabetes, or possibly that these 2 factors reinforce each other [8,39]. Second, arterial stiffness may lead to microvascular dysfunction, which in turn leads to damage to low-resistance organs (eg, the pancreas), resulting in reduced tissue perfusion, including insulin-mediated muscle perfusion. This will lead to impaired glucose metabolism, insulin resistance, and an elevated fasting glucose level [7,40]. In this process, endothelial dysfunction and impaired endothelium-dependent vasodilation may exacerbate insulin resistance by limiting glucose delivery to key target tissues [41]. Third, increased oxidative stress and chronic low-grade inflammation may be common risk factors for atherosclerosis and diabetes [42,43].

The mechanisms underpinning the relationship of obesity with diabetes are only partially understood. Most of the hypotheses about obesity causing diabetes in recent years have been based on the coexistence of insulin resistance. Randle et al [44], for the first time, explained the relationship between obesity and diabetes by the “glucose-fatty acid cycle,” proposing a theory that obesity inhibits the glycolytic enzymes pyruvate dehydrogenase, phosphofructokinase, and hexokinase, thus causing an imbalance in glucose metabolism [45]. Another hypothesis is that adipose tissue is a secretory organ that produces and releases a variety of factors that may lead to insulin resistance. Most of the data suggested that tumor necrosis factor alpha (TNF-α) plays a mediating role [46]. Upregulated TNF-α induces multiple adverse effects, such as impaired insulin signaling and inhibition of glucose transporter type 4 expression, which inhibits glucose uptake [47]. TNF-α could reduce the expression of lipocalin, a protein that is abundantly expressed in adipocytes and has direct antidiabetic and antiatherosclerotic effects [48]. In addition, there are several hypotheses about signaling pathways of adipose tissue inflammation [49], endoplasmic reticulum stress [50], oxidative stress [51], and accumulation of immune cells [52], which are related to insulin resistance and insulin secretion.

This cohort study proposed the concept of arterial stiffness abdominal obesity phenotype to stratify the risk of incident diabetes. However, the results should be interpreted in the context of limitations. First, baPWV measures the stiffness of both the elastic aorta and muscular artery, and other index of arterial stiffness such as carotid-femoral pulse wave velocity was not collected in this study. Second, this was an observational study design, and we were unable to claim the causal effect of arterial stiffness obesity groups on diabetes onset. Third, although we adjusted for the important confounding factors, including fasting glucose level, there was still a possibility of residual confounding bias, such as dietary factors. The observed results require further validation in other populations.

The findings indicated the combined effect of arterial stiffness and obesity status on diabetes onset, independent of fasting glucose level. This study proposed the concept of arterial stiffness abdominal obesity phenotype, providing a noninvasive and simple panel for the risk stratification and potential management of diabetes.