Body Mass Index or Waist-Hip Ratio, Which Correlated With Arterial Stiffness Based on Brachial-Ankle Pulse Wave Velocity in Chinese Rural Adults With Hypertension?

Background: The aim of this study was to investigate the association betweenbody mass index (BMI), waist-hip ratio (WHR) and arterial stiffness (AS) based on brachial-ankle pulse wave velocity (baPWV) in Chinese rural adults with hypertension. Methods: In this analysis, selected 5,049 Chinese rural adults with hypertension were divided into three groups according to BMI ( <24 kg/m 2 , control; 24 ‐ 28 kg/m 2 , overweight; and ≥ 28 kg/m 2 , obesity), WHR ≥ 0.9 for male and ≥ 0.85 for female was dened as central obesity, while baPWV ≥ 18.0 m/s was considered as increased AS. Multivariate analysis was used to examine the association between BMI, WHR (central obesity) and AS based on baPWV in different models. Furthermore, the generalized additive model and smooth curve tting was used to visually show the relationship between BMI or WHR with baPWV. Finally, to ensure the robustness between BMI group or central obesity with increased AS, we also did the subgroup analyses that were performed using stratied multivariate regression and interaction analyses and presented in tabulated form or forest plot. Results: The prevalence of overweight, general obesity, central obesity and increased AS were 32.62%, 8.58%, 63.85% and 44.01%, respectively. In comparsion with control group, there are a statistically signicant lower prevalence of increased AS in population with overweight or general obesity (adjusted-OR: 0.78, 95% CI 0.65 to 0.92, P <0.001; adjusted-OR: 0.54, 95% CI 0.40 to 0.72, P <0.001, respectively; P for trend <0.001). Whereby in comparsion with non-central obesity group, there are an statistically signicant higher prevalence of increased AS in population with central obesity (adjusted-OR: 1.54, 95% CI 1.30 to 1.83, P <0.001). The multivariate analyses indicated that BMI was negatively associated with baPWV (adjusted-β per SD increase: -0.49, 95% CI -0.60

Atherosclerosis would be the main mechanism linking obesity to CVD [6,7]. As part of the atherosclerotic process, there is an increase in arterial stiffness (AS), which has been associated with an increase in pulse wave velocity (PWV) [8][9][10]. In recent decades, PWV as a measure of arterial stiffness [11] has been increasingly used for CVD risk assessment [12,13]. Brachial-ankle pulse wave velocity (baPWV), considered as an unique indicator in assessing the degree of arteriosclerosis, has been gradually adopted as an epidemiological investigation method because of its simplicity and reproducibility [11].
Body mass index (BMI) is one of the main measures for general obesity, and is commonly used in clinical assessment and screenings. However, researches on the association between BMI and PWV in various population including the hypertensive patients [14][15][16][17][18] which often accompanied with overweight or obesity [19], remian controversial: studies have reported an insigni cant relationship [14,18,[20][21][22][23][24], positive relationship [15,17,[25][26][27][28], or negative relationship [16,29,[30][31][32][33]. Many causes could interpret these apparent discrepancies including diverse sample size, different origins of the study population, ethnic and regional disparity. On the other hand, BMI is not able to differentiate fat-free muscle from fat mass, or account for sex and racial differences in fat content and distribution of visceral and subcutaneous fat [34]. Thus, the impact that BMI and obesity have on hypertensive patients needs further study.
Waist-hip ratio (WHR), a measure of central obesity and abdominal fat, may be better indicate risk for atherosclerosis or CVD associated with obesity than other anthropometric measures, including BMI [35][36], as WHR is potentially less in uenced by muscle and bone mass. The higher WHR is often accompanied with increased arterial stiffening measured by PWV [25,27]. Moreover, China bears the biggest hypertension burden in the world and hypertension is often accompanied with overweight or obesity [19]. Furthermore, because rural adults engage in stronger physical labor in the long run, this crowd have more muscular tissue content and better physical tness than civil servants, thus we consider that BMI might not be suitable for the evaluation of obesity and to indicate risk for atherosclerosis in Chinese rural adults with hypertension. We therefore analysed the association between BMI, WHR and arterial stiffness based on baPWV in Chinese rural adults with hypertension.

Study design and participants
Data analyzed in this study was the baseline of the ongoing China H-type Hypertension Registry Study (Registration number: ChiCTR1800017274). The method of data collection and the exclusion criteria have been described previously [37]. Brie y, the study is a real-world, multicenter, observational study, conducted from Wuyuan, Jiangxi province of China, which conducted in March 2018. Eligible participants were adults aged 18 years and older who had hypertension, de ned as seated, resting systolic blood pressure (SBP) ≥ 140 mmHg or diastolic blood pressure (DBP) ≥ 90 mmHg at the screening, or who were on antihypertensive medications. The exclusion criteria included neurological abnormalities, unable to be followed-up according to the study protocol, or plans to relocate shortly, and the patients, who are not suitable for inclusion or for long-term follow-up as assessed by study physicians. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of Institute of Biomedicine, Anhui Medical University. All participants provided written informed consent.
In this study, a total of 5, 233 participants completed the measurement of baPWV. In this analysis, we excluded 78 individuals with ABI < 0.9 (11,24) (in cases of peripheral arterial disease, the reliability of baPWV measurement is attenuated) and 104 patients with atrial brillation as well as 2 participants without BMI measurements, nally 5, 049 participants were included in our analysis ( Supplementary   Fig. 1).

Data collection and indexes determination
Participants' demographics characteristics (age, gender), lifestyle (smoking and drinking status), past medical history [such as hypertension, diabetes mellitus (DM), dyslipidemia and atrial brillation (AF) ] and medication usage (such as antihypertensive agents, hypoglycemic agents, lipid-lowering agents and antiplatelet agents) through questionnaires were collected by trained research staff. AF is diagnosed by medical history and resting supine standard 12-lead surface electrocardiograms (25 mm/s, 10 mm/mV).
They also collected participants' anthropometric measurement indicators included weight, height, waistline, hipline, systolic blood pressure (SBP), diastolic blood pressure (DBP) and heart rate (HR). Due to the health screening protocols, single point measurements of anthropometric indicators were obtained. Waistline and hipline were measured using an inelastic measuring tape, with 0.1 cm resolution and length of 1.5 m. Individuals were at upright position, with abdomen relaxed at the end of gentle expiration. Four consecutive o ce BP measurements (each time interval was 1-2 min) were taken for each individual with a validated non-invasive electronic oscillometric device using the appropriate cuff size for the upper right arm. The mean values of the last three readings were used for analysis in order to reduce the impact of reactivity on the blood pressure (higher rst reading). BMI was calculated as the body weight in kilograms divided by the square of the height in meters (kg/m 2 ). WHR was calculated as the waistline in centimeters divided by the hipline in centimeters. In our study, Subjects were divided into three groups according to BMI (< 24 kg/m 2 , control; 24-28 kg/m 2 , overweight; and ≥ 28 kg/m 2 , obesity) based on published values for the Chinese population [29,38]. Furthermore, a WHR of < 0.9 for male and < 0.85 for female was de ned as normal, while ≥ 0.9 for male and ≥ 0.85 for female was de ned as central obesity [38].
Blood samples were collected utilizing venipuncture after an overnight fast of at least 12 hours. Blood biochemicaltests for plasma total homocysteine, fasting blood glucose (FBG), total cholesterol (TC), total triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), serum uric acid and creatinine, blood urea nitrogen (BUN), total and direct bilirubin, aspartate aminotransferase (AST) and alanine aminotransferase (ALT) measured using automatic clinical analyzers (Beckman Coulter) at the core laboratory of the National Clinical Research Center for Kidney Disease, Guangzhou, China. All laboratory measurements met a standardization and certi cation program. In our study, DM was de ned as the FBG levels above 7.0 mmol/L or on treatment for diagnosed DM formerly. Dyslipidemia was de ned as having one of the following features: elevated TG (≥ 2.3 mmol/L), TC (≥ 6.2 mmol/L), LDL-C (≥ 4.1 mmol/L), reduced HDL-C (< 1.0 mmol/L) or on appropriate lipid-lowering agents treatment formerly. The formula for estimated glomerular ltration rate (eGFR) used the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation which more accurate than Modi cation of Diet in Renal Disease (MDRD) equation for subjects at higher eGFR levels.

Measurement of PWV and de nition of arterial stiffness
For baPWV values we used a BP-203 RPE III networked arteriosclerosis detection device (Omron Health Care, Kyoto, Japan). Participants underwent baPWV measurement after at least 5 minutes of rest in the supine position in a quiet room. Coffee, tea, cigarette use, or alcohol use were not allowed for 30 minutes before the test. Blood pressure cuffs were wrapped on both arms and ankles and connected to a plethysmographic sensor to determine the volume pulse waveform. The lower edge of the arm cuff was positioned 2-3 cm above the cubital fossa transverse striation, while the lower edge of the ankle cuff was positioned 1-2 cm above the medial malleolus. The heart beat monitor was placed on the left edge of the sternum, and electrocardiogram electrodes value directly. The methodology for baPWV measurement remained the same for all participants. The average of baPWV measured on bilateral limbs of each patient was used for analysis. While baPWV value has been shown to have prognostic signi cance, with a threshold of increased risk that exists around 18 m/s [11]. Therefore, baPWV ≥ 18 m/s was considered as increased AS in our study.

Statistical Analysis
Clinical characteristics of participants were grouped by BMI group or central obesity. Continuous variables were expressed as means ± standard deviation (SD) or median (Q1-Q3). For the variables in both groups that exhibit normal distribution and homoscedasticity, independent t-test or one-way analysis of variance were used. For the variables that do not show normal distribution, nonparametric test was used. Categorical variables were expressed as count (percentage), differences between groups were measured by chi-square test or Fisher's exact probability test. Secondly, we used three different linear regression models to examine the associations of BMI or WHR with increased AS or baPWV. The crude model was not adjusted for any confounder. The model was adjusted for age, gender, SBP, DBP, HR, WHR or BMI except the independent variable itself. The model was confounder model. The confounder model screened covariates including age, gender, SBP, DBP, heart rate, BMI, WHR, DM, smoking and drinking status, homocysteine, TC, TG, HDL-C, LDL-C, AST, ALT, serum uric acid, eGFR, total and direct bilirubin, lipid-lowering agents and antiplatelet agents except the independent variable itself. We selected these confounders on the basis that, when added to this model, its changed the matched odds ratio by at least 10 percent. Supplementary tables 1 show the associations of each confounder with the outcomes of interest. We considered the confounder model to be the main model. In addition, we performed tests for linear trend by entering the median value of each category of BMI levels or categorical variables as a continuous variable in the models.
Furthermore, the generalized additive model and smooth curve tting (penalized spline method) was used to visually show the relationship between BMI or WHR with baPWV (the selection principle of confounders was equal to confounder model). Finally, to ensure the robustness between BMI group or central obesity with increased AS, we also did the subgroup analyses that were performed using strati ed multivariate regression and interaction analyses and presented in tabulated form or forest plot.
All statistical analyses were performed using the statistical package R (http://www.R-project.org, The R Foundation) and the Empower (R; www.empowerstats.com; X&Y Solutions, Inc, Boston, MA, USA). All Pvalues are two-tailed, and P < 0.05 was considered statistically signi cant.

Clinical characteristics of the study population
The present study included 5, 049 Chinese adult hypertensive individuals (age: 64.46 ± 9.45 years, range 29-93 years; male, 49.89%), and the prevalence of overweight, general obesity, central obesity and increased AS were 32.62%, 8.58%, 63.85% and 44.01%, respectively. The clinical characteristics of the study participants were presented in Table 1 grouped by BMI group or central obesity, Supplement Table 2 grouped by baPWV quartiles.  On the one hand, the participants with overweight or general obesity were more likely to be central obesity, had lower extents of baPWV and lower prevalence of increased AS. On the other hand, the subjects with central obesity were more likely to be overweight or general obesity, had elevated extents of baPWV and higher prevalence of increased AS (Table 1). There were statistically signi cant differences in the above clinical characteristics among different groups (P < 0.05). Furthermore, Compared to baPWV ≤ 15.36 m/s, there were a statistically signi cant elevated extents of SBP and DBP, lower prevalence of overweight or general obesity as well as higher prevalence of central obesity in the third and highest baPWV quartiles (Supplement Table 2).

Associations between BMI and increased AS
The multivariate analyses indicated that BMI was negatively associated with the risk of increased AS (adjusted-OR per SD increase: 0.75, 95% CI 0.68 to 0.82, P < 0.001;

Associations between WHR and increased AS
The multivariate analyses indicated that WHR was positively associated with the risk of increased AS (adjusted-OR per SD increase: 1.24, 95% CI 1.14 to 1.35, P < 0.001;

Subgroup analyses by potential effect modi ers
To explore whether the inverse association between BMI group and increased AS were still stable in different subgroups, we conducted the strati ed and interaction analyses. The subgroup analyses showed that there were not statistically signi cant interactions in any of the subgroups, including age (< 60 vs.  Table 5).

Discussion
In the current study, we presented an interesting nding that there was a inversely relationship between BMI and baPWV or increased AS in Chinese rural adults with hypertension, whereas WHR or central obesity is positively associated with baPWV and increased AS. To our knowledge, these ndings have not been previously described.
Hypertension is often accompanied with overweight or obesity [19]. In our study, We found that the prevalence of overweight, general obesity, central obesity and were 32.62%, 8.58% and 63.85%, respectively. Moreover, PWV was increased in hypertensive patients, and the degree of PWV increase was associated with baseline blood pressure [39]. In our study, baPWV was also associated with most clinical characteristics including SBP and DBP, nearly half (44.01%) of the hypertensive participants have increased AS (baPWV ≥ 18 m/s) (Supplement Table 1 and 2), which has been considered to be an independent factor in indicating the relapse of CVD as well as an irreversible progress of atherosclerosis [11].
Obesity is an accredited dangerous factor for CVD [23]. However, some studies have shown that obesity as de ned by BMI has a survival bene ts on some speci c populations [40]. The phenomenon of the obesity paradox may be related to genetics, cardiorespiratory tness, bene cial adipose tissue and weaker sympathetic activation [34,40].
In our study, we found that there was a inversely relationship between BMI and baPWV or increased AS in our crowd. Similarly, Huang et al. [29] and Liu et al. [18] found that there was a negative relationship between BMI and baPWV among male hypertension participants. On the contrary, previous studies showed that a higher BMI was associated with increased PWV in Grade I essential hypertension [15] or obese and non-obese hypertensive patients [17]. Different origins of the study participants might contribute to the study differences. The Liu et al. [18] study was based on 699 male hypertensive patients who were hospitalization or had other complications. The Samir et al. [15] study was based on 114 patients with Grade I essential hypertension who were civil servants. The Huang et al. [29] study enrolled 10 1510 participants, a coal occupation group in labor-intensive enterprise, from 11 hospitals in the Kailuan community, which is most similar to our rural hypertensive crowd. In consideration of that BMI fails to account for body composition [34] and physical habits, whereby rural adults are associated with increased muscle mass (muscle mass is greater than fat mass) and better physical tness, both of which may account for the inversely relationship between BMI and baPWV or increased AS in our crowd.
WHR, a more sensitive marker for central obesity, may be better indicate risk for atherosclerosis or CVD associated with obesity than BMI [35,36], as high WHR can re ect both increased visceral fat as well as low gluteal muscle mass (and/or low peripheral fat mass), both of which have been found to be independently associated with cardiovascular disease risk [41,42]. We demonstrated that WHR or central obesity is positively associated with baPWV and increased AS in our crowd. The result was consistent with similar ndings in the general population [25,27]. For example, the Whitehall II study, a prospective study of 10 308 civil servants, showed that standardized effects of central adiposity on aortic PWV increase was obvious and previous adiposity was associated with aortic stiffening independent of change in adiposity, glycaemia, and lipid levels across PWV assessments [25].
Seidell et al. [43] demonstrated that WHR is highly correlated with visceral fat but not with subcutaneous fat (the latter which in fact may be protective), while waist circumference (WC) was highly correlated with both that. The incorporation of a reference body size may explain why WHR is a better predictor of outcomes than WC, a primitive measurement for abdominal obesity [35].
The lack of consistency between WHR and BMI may re ect that these measures identify different characteristics of obesity (central obesity in case of WHR vs. subcutaneous/total fat in case of BMI).
Bouchi et al. [44], in a cross-sectional study with patients with diabetes and non-obese (normal BMI), noticed that increased visceral fat seems to be associated with increased AS based on baPWV. Notably, BMI fails to account for body composition, however, skeletal muscle mass [45] and physical tness [46] counteract risk with obesity. As Chinses rural adults are associated with increased muscle mass and better physical tness, thus BMI might not be suitable for the evaluation of obesity and the degree of atherosclerosis may be underestimated by relying on BMI in this crowd, while WHR (central obesity) may be better indicate risk for atherosclerosis.
There are some strengths in our study, which should be considered in the interpretation of results. Firstly, the study was conducted in Wuyuan, a typical county in China, which offered us an exceptional opportunity to examine the association between obesity indicators and PWV in this crowd under a natural state. In addition, although the real-world, multicenter, observational study was susceptible to potential confounding, we used rigorous statistical adjustments to minimize residual confounding.

Limitations
Our study also has several limitations that should be considered. Firstly, the study was performed by cross-sectional design and so provides the association of relationships instead of predictive values of WHR on the progression of arterial stiffness. Secondly, results are limited by the absence of gold standard for measurement of visceral fat, such as computerized tomography (CT) and magnetic resonance imaging (MRI) for quanti cation of body fat composition. Furthermore, WHR measurements were performed by different individuals, although they were provided with clear instructions. Finally, this study was conducted in Chinese hypertension populations, the generalizability of the ndings to other populations remains to be determined.

Conclusions
In conclusion, hypertension was often accompanied with central obesity and increased AS in Chinese rural adults with hypertension. We presented an interesting nding that there was a inversely relationship between BMI and baPWV or increased AS in this crowd, whereas WHR or central obesity is positively associated with baPWV and increased AS. Given that rural adults are associated with increased muscle mass and better physical tness, BMI might not be suitable for the evaluation of obesity and the degree of atherosclerosis may be underestimated by relying on BMI in this crowd, whereas WHR (central obesity) may be a helpful tool for risk strati cation in Chinese rural adults with hypertension. Additional studies should also evaluate whether intentional reduction in WHR, whether through exercise, medication or surgical intervention, is associated with a decrease in CVD events. MRI:magnetic resonance imaging; SD:standard deviation; OR:odds ratio; CI:con dence interval. Figure 3 Effect size of central obesity on increased AS in prespeci ed and exploratory subgroups. Abbreviations: AS, arterial stiffness; SBP, systolic blood pressure; BMI, body mass index; DM, diabetes mellitus; HDL-C, high-density lipoprotein cholesterol; eGFR, estimated glomerular ltration rate; ALT, alanine aminotransferase; OR, odds ratio; CI, con dence interval. Each strati cation adjusted for age, gender, SBP, HR, BMI, smoking status, DM, HDL-C, eGFR and ALT except the subgroup variable.