Risk of arterial stiffness according to metabolically healthy obese phenotype: a combined cross-sectional and longitudinal study in kailuan cohort

We aim to investigate the risk of incident arterial stiffness according to metabolically healthy obese (MHO) phenotype in Chinese population. 37,180 participants with at least one-time measurement of branchial-ankle pulse wave velocity (baPWV) were included in the cross-sectional analysis, and 16,236 participants with repeated measurement of baPWV during the follow-ups were included in the longitudinal study. Cross-classification of body mass index (BMI) categories and metabolic health status created six groups. Linear and logistic regression analyses were used. The results of cross-sectional and longitudinal investigation were essentially the same, as the abnormality of baPWV increased with BMI categories in metabolically healthy participants, while the increasing tendency disappeared in metabolically unhealthy participants. A 1.4-fold, 2.2-fold increased risk for the new occurrence of arterial stiffness were documented in MHO and metabolically unhealthy obese participants compared to metabolically healthy normal-weight controls in the fully adjusted model. Further stratified analysis showed that metabolic health status was an interaction factor between BMI and arterial stiffness in all study populations (P=0.0001 for cross-sectional study and P=0.0238 for longitudinal study). In conclusion, metabolic health status and BMI categories contribute to the progression of arterial stiffness, while BMI is positively associated with arterial stiffness only in metabolically healthy participants.


INTRODUCTION
Branchial-ankle pulse wave velocity (baPWV), a promising indicator of both central and peripheral arterial stiffness [1], has been proven to be strongly associated with cardio-cerebrovascular morbidity and mortality in a recent meta-analysis of 8 studies [2]. The subclinical state of atherosclerosis could be improved AGING with baPWV-guided lifestyle modification and therapeutic intervention [3].
Metabolic syndrome (MetS) is recognized as a cluster of risk factors for atherosclerotic cardiovascular disease (CVD), including hypertension, hyperlipidemia, hyperglycemia, and broadened waist circumference (WC) [4,5]. It is known that baPWV increases with MetS, as well as the number of MetS components [6][7][8]. Obesity, which has reached an epidemic level owing to economic development in China, often coexists with MetS [9]. While no consensus has been reached on the correlation between obesity and baPWV [10,11]. Currently, a subset of obese individuals without MetS, identified as metabolically healthy obese (MHO), have attracted extensive attention due to the controversial results regarding cardiovascular risk. Some studies demonstrated that obesity status exerted no extra influence on CVD [12,13]. Others indicated MHO was a transient condition from metabolically healthy to unhealthy phenotypes, and obesity carried an increased risk of CVD regardless of metabolic health status [14][15][16]. The development of clinical CVD events usually requires a long period of time, subclinical atherosclerosis as arterial stiffness might better estimate the impact of obesity or MetS on CVD within a short period of time. Considering the abovementioned research, the role of obesity on the association between MetS and baPWV is worth further exploration. Therefore, we aimed to investigate the risk of incident arterial stiffness according to MHO phenotype in Chinese population using the Kailuan cohort study.

Patients characteristics
In the cross-sectional analysis, out of the 37,180 enrolled participants, 10,295 (27.69%) were metabolically healthy normal weight (MH-NW), 7,171 (19.29%) were metabolically healthy overweight (MH-OW), 2,337 (6.29%) were MHO, 4,253 (11.44%) were metabolically unhealthy normal weight (MUH-NW), 8,314 (22.36%) were metabolically unhealthy overweight (MUH-OW), and 4,810 (12.94%) were metabolically unhealthy obese (MUO). The baseline characteristics of participants among the body mass index (BMI)-MetS categories were presented in Table 1. In addition to risk factors referred to MetS, the individuals in the metabolically unhealthy group were more likely to be older, male, less educated, a current smoker or alcoholic, and having slightly higher average income and salt intake. People who were overweight or obese had a larger WC than those with normal weight and this tendency was more pronounced in the metabolically unhealthy group.

Cross-sectional investigation
8,182 cases of arterial stiffness were documented based on the first measurement of baPWV. Values are (%) for categorical variables and mean ± SD or median (IQR) for continuous variables; BMI, body mass index; WC, waist circumstance; LDL, low-density lipoprotein; HDL, high-density lipoprotein; CRP, C-reactive protein.
of CVD remained statistically significant. Additional information was given in Table 5.

Stratified analysis
Metabolic health status was an interaction factor between BMI and arterial stiffness in all study participants (P=0.0001 for cross-sectional study and P=0.0238 for longitudinal study). In metabolically health participants, BMI demonstrated a dose-dependent increase in the risk of abnormal baPWV, with adjusted ORs of 1.07 (95%CI 0.97-1.18), 1.22 (95%CI 1.06-1.41) in the overweight and obese group. By contrast, no relation was found in participants with MetS in the cross-sectional study. These results were further validated in the cohort study (Table 6).

DISCUSSION
This study assessed the cross-sectional and longitudinal associations between BMI-MetS phenotypes and baPWV (either as a continuous or dichotomous variable). We found that metabolic health status and BMI categories contributed to the AGING    increased risk of abnormal baPWV [18]. Several studies reported that obese individuals had youthful arteries with lower PWV [19,20]. By contrast, some studies found an irrelevant association between them [21,22]. Obesity often coexists with other risk factors and accelerates arterial stiffness through its associated metabolic abnormalities. Although MetS is served as a risk enhancer, it is difficult to predict CVD risk quantitatively due to the mediating role of obesity. To take the contribution of obesity and other cardio-metabolic risk factors separately, we thus used a modified harmonized IDF-MetS definition and subdivided it by the degree of obesity. The present study indicated that obesity did interact with metabolic status and BMI was positively associated with baPWV only in metabolically healthy participants.
It is noteworthy that BMI cannot fully reflect body composition and adiposity distribution. Those with excess visceral fat exhibited a greater risk of CVD than AGING those with subcutaneous fat. WC is a more reliable index capable of differentiating between overall adiposity and abdominal adiposity among the same BMI range [23]. In our study, a significant positive correlation between WC and BMI categories was observed, which could partially offset the inadequacies of BMI. Beyond that, the issue of obesity paradox has aroused great concern. Although obesity contributes to the development of CVD, the long-term prognosis of obese individuals is often better due to their superior cardiorespiratory fitness against acute stress [24]. In terms of pathophysiological mechanisms, there was no paradoxical association between obesity and subclinical CVD as adipose tissue could impair vascular function through specific hormones and proinflammatory cytokines [25].
The strength of this study is its combined crosssectional and longitudinal aspects. Nonetheless, there are still some limitations. First, the harmless feature of metabolically healthy phenotype was hung in doubt given the definition of MetS. There were 75.9% of MHO participants having one metabolic risk factor in our study. which could exert an additional effect on baPWV apart from obesity. Secondly, accumulated evidence indicated that MHO was not only an intermediate-stage, but also a transit condition from metabolically healthy to unhealthy status [14,15]. Because of the relatively short follow-up time, the conversion of MHO status was not included in our statistics. Further study is needed to provide insight into the dynamic relationship between metabolically healthy obese phenotype and arterial stiffness.
In conclusion, both metabolic health status and BMI categories contribute to the progression of arterial stiffness, while BMI is positively associated with arterial stiffness only in metabolically healthy participants due to its fully mediating role through associated metabolic risk factors. Moreover, MHO is an intermediate stage between metabolically healthy and unhealthy status rather than a benign status, which highlights the need for active weight reduction and risk factor management.

Ethics
The study was conducted in accordance with guidelines from the Helsinki Declaration and was approved by the Ethics Committees of both Kailuan General Hospital and Beijing Tiantan Hospital. All participants or their legal representatives provided written informed consent (Trial registration: ChiCTR-TNRC-11001489).

Study population
The Kailuan study is an ongoing prospective cohort study, details about the design and methods of this study have been published in detail previously [26].  [27]. MetS was defined as having 2 or more abnormalities of the following components based on the modified harmonized International Diabetes Federation (IDF) criteria [5], (1) systolic blood pressure (BP) ≥ 130 mmHg, diastolic BP ≥ 85 mmHg, use of antihypertension medication, or self-reported history of hypertension; (2) fasting blood glucose ≥ 5.6 mmol/L (100 mg/dL), current use of anti-diabetic medication, or self-reported history of diabetes; (3) triglycerides ≥ 1.7 mmol/L (150 mg/dL) or current use of lipid-lowering medication; (4) high-density lipoprotein cholesterol < 1.0 mmol/L (40 mg/dL) for men and < 1.3 mmol/L (50 mg/dL) for women. WC was not included in the definition of MetS, due to its collinearity with BMI [28].
Combing the BMI categories and metabolic health status together, participants were then divided into six

Measurement of baPWV
Bilateral baPWV was evaluated by utilizing an automatic arteriosclerosis detection device (BP-203RPE III; Omron Healthcare Co., Kyoto, Japan). Information of the participants was recorded prior to the measurement, including age, sex, height, and weight. Before the examination, participants should stay away from cigarettes, caffeinated or alcoholic beverages for at least 3 h and have a minimum resting time of 5 min in a supine position. Cuffs were attached to both the upper arms and ankles with certain strain. The lower border of the branchial cuff was tied 2-3 cm above the cubital fossa transverse, and the lower border of the ankle cuff was tied 1-2 cm above the medial malleolus. The cardiechema collecting device was placed at the left border of the sternum, with electrodes clipping to both waists for electrocardiography acquisition. The measurement of baPWV was repeated twice by trained nurses, and the second value was recorded. The maximum value of left-and right-side baPWV was used in further analysis. BaPWV ≥1800 cm/s was considered as arterial stiffness [29]. Moreover, the second measurement of baPWV was performed during the two-year interval follow-ups. The change of baPWV was calculated as re-examined baPWV subtracting baseline baPWV, and the new occurrence of baPWV abnormality was defined as normal baPWV at baseline but abnormal baPWV at follow-up.

Other baseline measurements
Data on demographic characteristics as age, sex, education level, average income, smoking status, drinking status, physical activity, salt intake, past medical history (including hypertension, diabetes, dyslipidemia, myocardial infarction, stroke), and current medication were self-reported on a questionnaire at baseline. WC and BP were measured on admission. Fasting glucose, total cholesterol, triglycerides, lowdensity lipoprotein, high-density lipoprotein, and Creactive protein were analyzed by an auto-analyzer (Hitachi 747; Hitachi, Tokyo, Japan) at the central laboratory of the Kailuan hospital.

Statistical analysis
Continuous variables were expressed as mean ± standard deviation (SD) or median (interquartile range, IQR), categorical variables were presented as count (percentage). The ANOVA or nonparametric Kruskal-Wallis test was used to compare group differences for continuous variables, and χ 2 test was used for categories variables.
Linear and logistic regression analyses were used to assess the association between BMI-metabolic status phenotypes and baseline baPWV in mono-factor and multi-factor models. To verify the causality of obesity status or metabolic syndrome on baPWV, indicated as the change of baPWV or the new occurrence of arterial stiffness, we further performed linear and logistic regression models in participants of the longitudinal study with β coefficients and ORs calculated. Multiple regression models were run as follows. Model 1 was adjusted for age and sex. Model 2 was adjusted for variates in model 1 plus educational level, average income, smoking, drinking, physical activity, sodium intake, history of myocardial infarction, and history of stroke. Model 3 was further adjusted for C-reactive protein. Inverse probability weighting (IPW) was used to minimize selection bias. Weights were based on results from a model of follow-up status, estimated using logistic regression with being followed up or not as the dependent variables and atherogenic risk factors as independent variables. We used the multivariate logistic regression analyses, before and after IPW, to assess whether BMI-MetS phenotypes were associated with higher odds of the change or new occurrence of baPWV abnormality. Sensitivity analyses excluding participants with a history of CVD in both crosssectional and longitudinal analyses were conducted. Additionally, stratified analysis was performed to assess the cross-sectional as well as the longitudinal association between BMI and metabolic health status. All statistical analyses were performed using SAS software, version 9.4 (SAS Institute, Cary, NC, USA), and a 2-sided value of P<0.05 was considered statistically significant.

AUTHOR CONTRIBUTIONS
Anxin Wang and Yu Wang performed the experiments, interpreted the results of statistical analysis, and drafted the manuscript. Yingting Zuo, Xue Tian, Shuohua Chen, Yihan Ma, and Xu Han conducted the statistical analysis and interpreted the data. Shouling Wu revising the manuscript for intellectual content. Shouling Wu and Xingquan Zhao had full access to all of the data and take responsibility for the integrity of the data and the accuracy of the data analysis.