Introduction

Arterial stiffness, commonly assessed using pulse wave velocity (PWV), is a feature of vascular aging and closely related to hypertension. Arteries are an organ that is damaged by elevated blood pressure (BP); however, arterial stiffness is an independent predictor of hypertension, future cardiovascular events, and all-cause mortality beyond conventional risk factors, including hypertension [1,2,3,4]. Understanding the factors that influence arterial stiffness is important, as they can be used as therapeutic targets to ameliorate the global burden of cardiovascular disease [5].

The global obesity epidemic is well established, and the prevalence of obesity is increasing worldwide [6]. Obesity, especially visceral obesity, is an important risk factor for cardiovascular diseases and all-cause mortality [6]. Although a strong association between obesity and hypertension has been established, the relationship between obesity and arterial stiffness remains inconsistent [7,8,9,10,11,12,13,14,15,16,17,18]. The paradoxical relationship between obesity index and arterial stiffness might be related to the use of body mass index (BMI), which cannot distinguish between fat mass and muscle mass. Waist circumference (WC) and waist-height ratio (WHtR), as measures of abdominal obesity, correlate with the amount of visceral adipose tissue, which is linked to insulin resistance, dyslipidemia, and systemic chronic low-grade inflammation, all of which play a pivotal role in the pathogenesis of atherosclerosis and increased cardiovascular disease risk [19]. Waist circumference and WHtR are superior to BMI in identifying adults at an increased cardiometabolic risk. However, several studies showed that WC and WHtR, as well as BMI inversely associated with arterial stiffness [11,12,13, 15]. This was believed to be due to the use of traditional obesity indices that do not accurately distinguish net weight, abdominal body fat, and visceral fat [8]. A body shape index (ABSI) and body roundness index (BRI) are newly developed anthropometric indices [20, 21]. A body shape index is based on WC, weight, and height, and was proposed for the estimation of abdominal adiposity. It was found to reflect mortality hazards better than BMI and WC. The BRI predicted overall body fat and visceral fat percentages more accurately than BMI or WC. Recent studies have reported that ABSI and BRI are positively associated with arterial stiffness [8, 22,23,24]. New anthropometric indices may be more strongly associated with visceral obesity than conventional indices, but the relationship between visceral fat area (VFA) assessed using computed tomography (CT), the most accurate measure of visceral adipose tissue [19], and conventional and newer anthropometric indices is not clear in Japanese population. This study aimed to investigate the relationship between VFA and several anthropometric indices in a screened Japanese population, and to examine the association between obesity indices and arterial stiffness.

Methods

Participants

This was an observational cross-sectional study. Participants were recruited from a one-day health checkup initiative for the general population held by the Okinawa Health Promotion Foundation. A total of 29,568 participants underwent optional voluntary brachial-ankle PWV (baPWV) measurements between April 2007 and March 2015. Some individuals participated multiple times during this period; however, only the first data points were analyzed. Data of 2819 participants who voluntarily underwent abdominal CT to evaluate their VFA were analyzed. Visceral fat area was measured at the level of the umbilicus using a 64-slice CT scanner (Somatom Definition; Siemens Medical Solutions, Erlangen, Germany). The VFA was quantified using a AZE Virtual Place (AZE of America, Ltd., Irvine, CA, USA). Participants with persistent atrial fibrillation or severe valvular disease were not eligible for baPWV examination. Participants with missing BMI (n = 6) and WC (n = 4) data were excluded. Because severe atherosclerotic arterial stenosis in a lower limb artery affected the recording and accuracy of baPWV, participants with ankle-brachial index of ≤ 0.9 (n = 19) or of ≥ 1.4 (n = 1) were excluded. Following the exclusion of participants who met the exclusion criteria, 2789 participants were enrolled (Fig. 1). This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for cross-sectional studies [25]. This study was conducted in accordance with the revised Declaration of Helsinki and approved by the Ethics Committee of University of Ryukyus (#1849). All the participants provided written informed consent.

Fig. 1
figure 1

Flow chart of the participants selection process. ABI Ankle-brachial index, baPWV Brachial-ankle pulse wave velocity, BMI Body mass index, CT Computed tomography, WC Waist circumference

Full size image

Anthropometric obesity indices

Body height and weight were measured using light clothing without shoes. Body mass index was calculated as the total weight divided by the square of height (kg/m²). Waist circumference was measured at the umbilical level with the participant in a standing position by a trained staff member. Waist-height ratio was calculated by dividing WC with height (cm). A body shape index and BRI were calculated using the following formulas: ABSI = WC/(BMI2/3 × height1/2), with WC and height in meters [20], and BRI = 364.2 − 365.5 × √1 − (WC/2π)2/(0.5 × height)2, with WC and height in cm [21]. We defined obesity-related indices as follows: general obesity based on BMI of ≥ 25 kg/m², abdominal obesity based on WC of ≥ 85 cm for men and ≥ 90 cm for women [26], and visceral obesity based on VFA of ≥ 100 cm².

Blood pressure and baPWV measurements

Brachial-ankle pulse wave velocity and ankle-brachial index (ABI) were measured using a validated automatic oscillometric device (BP-203RPE, II form PWV/ABI; Omron-Colin Co. Japan), which simultaneously measures pulse volume and BP in the brachial and ankle arteries [27]. The participants were examined at rest in the supine position for at least 5 min. Ankle-brachial index was calculated bilaterally as the ratio of systolic BP (SBP). We used the mean of left and right baPWV for the analysis.

Data collection

Blood samples were drawn after an overnight fasting. Dyslipidemia components of metabolic syndrome (MetS) was defined as high-density lipoprotein cholesterol of < 1.03 mmol/L and triglycerides of ≥ 1.69 mmol/L. High BP components of MetS were defined as BP of ≥ 130/85 mmHg and/or the use of antihypertensive agents. Hyperglycemic components of MetS were defined as fasting blood glucose of ≥ 6.1 mmol/L and/or the use of antidiabetic agents [26]. Estimated glomerular filtration rate (eGFR) was calculated using the Japanese equation [28]. Individual medical histories were collected using self-administered questionnaires and confirmed through physician interviews.

Statistical analysis

Data are expressed as mean (standard deviation) for normally distributed continuous variables, or as the number of participants (percentage) for categorical variables. Data for groups according to sex were compared using the Student’s t-test for continuous variables and the chi-squared test for categorical variables. Correlations among continuous variables were analyzed using the Pearson’s correlation tests. For multivariable analyses, we excluded participants with missing eGFR and fasting blood glucose data (n = 30), and the total sample size was 2759. Multivariate linear regression analyses were used to detect associations between obesity indices and baPWV after adjusting for age, SBP, heart rate, fasting blood glucose, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, eGFR, current smoking, and use of any agents for hypertension, dyslipidemia, and diabetes. A two-sided P value of < 0.05 was considered statistically significant. Analyses were performed using JMP® Pro version 15 (SAS Institute Inc., Cary, North Carolina, USA).

Results

Of the 2789 participants, 50% were women. Baseline characteristics are presented in Table 1. The prevalence of general, abdominal, and visceral obesity were 1314 (47%), 1 483 (53%), and 1289 (46%), respectively. Compared with the women, the men were younger, had a higher BMI, and larger WC and VFA, whereas WHtR, ABSI, and BRI were lower in men. The SBP and baPWV were comparable between women and men. Age-related differences in SBP were modest, whereas baPWV increased significantly with age (Table 2). In women, the anthropometric indices, except BMI, modestly correlated positively with age; however, the correlation between BMI and age was very weak (Fig. 2 and Table 2). In men, the VFA, WHtR, ABSI, and BRI positively correlated with age; however, BMI inversely correlated with age, and WC did not correlate with age.

Table 1 Baseline characteristics of participants by sex
Full size table
Table 2 Relationship to age
Full size table
Fig. 2
figure 2

Age-related differences of obesity indices by sex. A Visceral fat area, B Body mass index, C Waist circumference, D Waist-height ratio, E A body-shaped index, F Body roundness index

Full size image

Visceral fat area showed modestly significant correlation with anthropometric indices in both the sexes. The correlation between ABSI and VFA was modest in men, but very weak in women (Table 3). In women, baPWV showed a modest correlation with VFA and anthropometric indices, but little correlation with BMI. In men, baPWV was modestly correlated with VFA, WHtR, ABSI, and BRI, but inversely correlated with BMI and did not significantly correlate with WC. Visceral fat area and all anthropometric indices moderately correlated with SBP in both the sexes (Table 3). In the age-adjusted model, VFA, WHtR, ABSI, and BRI positively associated with baPWV in women, but BMI and WC did not associate with baPWV (Table 4, model 1). In men, the VFA and ABSI positively associated with baPWV, whereas BMI, WHtR, WC, and BRI were not. In the age- and SBP-adjusted models, VFA was not associated with baPWV in women and men (Table 4, model 2). A multivariable-adjusted model showed that the anthropometric indices, except ABSI, inversely associated with baPWV in women and men (Table 4, model 3). The association between VFA and baPWV was very weak, but statistically significant only for women. In both the sexes, ABSI positively associated with baPWV. Sensitivity analysis after excluding participants using any agents for hypertension, dyslipidemia, and diabetes showed similar results (Supplementary Table 1).

Table 3 Correlation of obesity indices with baPWV and SBP by sex
Full size table
Table 4 Regression analysis for the association between obesity indices and baPWV
Full size table

The VFA and anthropometric indices, except ABSI, were positively associated with SBP in women and men in all the models (Supplementary Table 2). The multivariable-adjusted models showed that ABSI inversely associated with SBP in women, but not in men. We then examined the relationship between obesity indices and MetS components, including high BP, dyslipidemia, and hyperglycemia. The VFA and anthropometric indices, except for ABSI, had significantly higher odds ratio (OR) for the presence of MetS components of high BP, dyslipidemia, and hyperglycemia in women and men (Supplementary Table 3). In women, the OR of ABSI was not significant for all the three MetS components; however, in men, the OR of ABSI was only significant for MetS-Glu.

Discussion

In this study, the VFA positively correlated with baPWV in the univariate analysis, but this association disappeared when adjusted for age and SBP at the time of measurement among women and men. The multivariable-adjusted model showed that VFA and anthropometric obesity indices, except for ABSI, were weakly associated inversely with baPWV only for women. In contrast, the VFA and anthropometric indices, except ABSI, were positively associated with all the three MetS components (high BP, dyslipidemia, and hyperglycemia). These divergent effects of VFA on baPWV and MetS components suggest that VFA may have a protective role in structural arterial wall stiffness. The ABSI had a lower correlation with VFA than the other anthropometric indices, and misclassified individuals at risk for obesity comorbidities.

Changes in body composition with aging result in excess visceral fat mass and decreased muscle mass [29]. The VFA was significantly larger among older men than in younger men, while the BMI and WC were lower or equal in older men than in younger men. These results suggest that an imbalance between skeletal muscle mass and visceral fat mass occurs with age, especially in men. The BMI and WC could not discriminate between skeletal muscle mass, abdominal subcutaneous fat mass, and visceral fat mass. In women, all the anthropometric indices increased with age, similar to the VFA. This is likely because women have less muscle mass and their age-related loss of muscle mass has less impact on BMI and WC than in men. These may be one of the reasons for the different correlations between obesity indices and baPWV. These results are similar to those of a previous health checkup-based study in Japan [30]. A body shape index, based on WC and adjusted for height and weight, was originally proposed to estimate abdominal adiposity in the USA population, mainly white Americans and African Americans, and Mexican [20]. A high ABSI indicates that WC is larger than expected for a given height and weight and corresponds more to the central concentration of body volume. Additionally, ABSI has been shown to be associated with MetS, atherosclerotic cardiovascular diseases, and all-cause mortality, but its effects has been shown to potentially vary by ethnicity. We found that ABSI had a lower correlation with VFA than the other anthropometric indices. In line with our results, the correlation between ABSI and VFA was modest in the Japanese population [22, 23]. The association of ABSI with hypertension, dyslipidemia, and diabetes was found to be weaker than BMI and WC among the Japanese and Indonesian populations [30, 31]. In Asian populations, ABSI may not adequately reflect VFA, which may be influenced by the differences in bone-to-muscle mass ratio and body fluid distribution between Asians and Caucasians [32, 33]. Further studies are needed to determine the usefulness of ABSI for assessing visceral obesity and cardiovascular events in the Japanese population. Body roundness index was developed to quantify individual body shape in a height-independent manner and is a predictor of body fat and visceral adipose tissue percentage [21]. However, the association of BRI with VFA and MetS components was comparable to that of simpler obesity indices, such as BMI, WC, and WHtR.

The paradoxical relationship between obesity indices and arterial stiffness appears to be associated with the use of BMI; however, our study and several previous studies found that not only BMI but also VFA, WC, and WHtR were inversely associated with PWV [11,12,13, 15]. The reason for the inconsistent results can be partially explained by differences in age, sex, race, obesity level, presence of cardiovascular diseases, and the use of any agents for hypertension, dyslipidemia, and diabetes [12, 34]. Some studies did not consider the BP component as a confounding factor in determining arterial stiffness [10]. Arterial stiffness, often assessed using aortic PWV and baPWV, is largely affected by age and BP. In addition to the increase in PWV due to structural arterial wall stiffening caused by prolonged hypertension, the increase in BP at the time of measurement causes passive aortic wall stretching, thereby resulting in an increase in PWV that is reversible by BP control. Persistent hypertension, hyperglycemia, and dyslipidemia associated with visceral obesity were expected to advance structural arterial stiffening. We found that VFA retained a weak positive association with baPWV when adjusted for age, but this association disappeared after adjusting for age and SBP at the time of measurement. The multivariate-adjusted model showed that VFA was weakly associated inversely with baPWV only for women. Cardio-ankle vascular index (CAVI) is an index of arterial rigidity at the central and peripheral levels that is relatively unaffected by BP (but still affected) at the time of measurement, and it provides a better assessment of the intrinsic elastic properties of an artery [35]. A study of middle-aged Japanese men found that CAVI was inversely correlated with BMI and WC, but positively correlated with VFA in univariate analyses; however, a multivariable-adjusted model, including age and SBP showed that CAVI associated inversely with VFA, BMI, and WC [11]. In a study of European outpatients, carotid-femoral PWV was positively associated with WC, but associated inversely with CAVI [36]. The positive relationship between carotid-femoral PWV and WC disappeared after adjusting for several factors, including the BP component. These findings and our results suggest that the positive association of VFA and anthropometric indices, except ABSI, with PWV might be affected by functional stiffening due to elevated BP at the time of measurement rather than by structural stiffening of the aortic wall. In a prospective study of a healthy middle-aged French population, initial BMI, WC, and WHtR were inversely associated with aortic PWV 20 years later in univariate and multivariate-adjusted models, whereas ABSI was positively associated with aortic PWV [34]. A cross-sectional study of healthy young adults showed that aortic PWV was not associated with trunk fat mass assessed by dual X-ray energy absorptiometry, but associated inversely with BMI after adjustment for several risk factors, including SBP [37]. In the same cohort, longitudinal analyses showed that trunk fat mass from childhood to young adulthood was not associated with aortic PWV progression. A lean mass was positively associated with aortic PWV in cross-sectional and longitudinal analyses. Although the mechanisms linking obesity and arterial stiffness are not fully understood, VFA and anthropometric indices, such as BMI, WC, and WHtR are likely to play a protective role in arterial stiffness.

The VFA and anthropometric indices, except ABSI, were significantly associated with adverse metabolic risks, such as hypertension, dyslipidemia, and hyperglycemia, which is consistent with previous studies [38]. Anthropometric obesity indices are not perfect indicators of adiposity, but are well-known risk factors for mortality in cardiovascular disease and all-cause mortality [39]. The link between obesity and cardiovascular disease is largely mediated by hypertension, dyslipidemia, diabetes, and other comorbidities [6, 40]. Despite the fact that obese individuals are more likely to have hypertension and metabolic risk factors, they have paradoxically fewer cardiovascular events than normal-weight individuals, especially those who already have symptomatic cardiovascular disease events [6, 39]. Further studies are needed to determine whether the paradoxical protective effect of obesity on arterial stiffness is related to the “obesity paradox.”

Perspective of Asia

The relationship between body composition and anthropometric indices differs between Asians and Caucasians [19]. Therefore, an obesity index that is useful in Caucasians may not necessarily be useful in Asians. The role of obesity in the pathogenesis of arterial stiffness may also differ between ethnic groups. Further longitudinal studies are needed to examine the effects of visceral obesity and body composition on structural arterial wall stiffness and cardiovascular events in Asian populations.

The strength of this study is the use of CT to measure VFA, which is the gold standard for assessing visceral obesity. Among studies that have examined the association between obesity-related indices and arterial stiffness, several of them have used VFA in their studies; however, most have used indirect measures, such as bioelectrical impedance analysis, and few have used VFA measured directly by CT in a large number of participants [8, 10, 11, 23].

Our study has several limitations. As in all cohort studies, a cause-effect association could not be established. This study was an observational design; therefore, it was subject to potential residual confounding effects despite adjusting for several potential covariates. The study participants were residents of Okinawa prefecture, which has the highest obesity rate in Japan; therefore, it might not be generalizable to the Japanese population. Participants in the present study were self-selected and might have been more concerned about their health than the general population. Although obesity indices were associated with metabolic factors, the pathophysiological mechanisms that were inversely associated with arterial stiffness were not clear in this study. Further studies are required to examine the relationship between body composition and arterial stiffness.

In conclusion, this study demonstrated that VFA and anthropometric indices, except ABSI, were inversely associated with arterial stiffness assessed by baPWV, but positively associated with MetS components, such as hypertension, dyslipidemia, and hyperglycemia.