Pulse wave velocity is decreased with obesity in an elderly Chinese population

Abstract

Obesity is generally considered an undesirable risk factor for cardiovascular disease; however, obese subjects with heart failure paradoxically can have better outcomes than their lean counterparts. This study aimed to investigate this characteristic in an elderly Chinese population. Elderly participants (N = 414, age 77 ± 11 years, 211 males) were recruited from a Chinese community‐dwelling elderly population. Subjects were divided into 3 groups according to body mass index (BMI ≤ 25, normal; 25‐28, overweight; and ≥28, obese). Arterial stiffness was assessed by brachial‐ankle pulse wave velocity (baPWV), and the atherosclerosis status was evaluated with the ankle brachial index (ABI). Brachial systolic blood pressure (BSBP) was significantly higher as BMI increased (135 ± 18.4, 138 ± 18.3, 147 ± 17.6 mm Hg; P = .003) adjusted for age, sex, and heart rate. However, baPWV was significantly lower as BMI increased (baPWV 1830 ± 18, 1793 ± 25, 1704 ± 36 cm/s; P = .008) in the three groups, even with additional adjustment for BSBP. BMI showed a significant negative correlation with baPWV (r = −.170, P = .001) after adjusting for confounding factors. Using multiple linear regression, BMI was negatively and independently associated with baPWV (β = −.190, P < .001) especially for age <80 years. Arterial stiffness, as measured by baPWV, is lower in overweight subjects in a Chinese elderly population compared to those with normal body weight. ABI showed no relationship with BMI. These findings suggest that reduced arterial stiffness in the overweight population, independent of confounding factors, may contribute to the explanation of the “obesity paradox.”

1. INTRODUCTION

Obesity, generally considered undesirable on account of its association with metabolic syndrome and with other risk factors for cardiovascular disease, presents a global health problem.1 The incidence of obesity in China ranks second in the world,1 affecting nearly one‐third of the population. Obesity remains a major risk factor for development of heart failure.2 However, paradoxically, overweight and other obese individuals with established disease can often have better long‐term outcome for revascularization and other therapies than do their lean counterparts.3, 4, 5 The obesity paradox has been shown to be present in populations with heart failure, hypertension, and even in the general population.6, 7, 8, 9, 10 The ARIC study 10 suggested that the 5‐year mortality was lower in overweight (HR 0.68‐0.79) and obesity (HR 0.50 ~ 0.63) compared to normal weight in heart failure patients.

The mechanisms of the obesity paradox may be related to genetics,11 cardiopulmonary health,10, 12, 13 beneficial adipose tissue producing various hormones and cytokines,7 weaker sympathetic activation,10, 14 and endothelial function.15 The function of large arteries may also be an important factor. Obesity is associated with impaired function of large arteries, which might be affected by metabolic dysregulation and inflammatory pathways.16 A study found an association between excess body weight and increased vascular stiffness in adults as young as 20‐30 years of age, highlighting the effect of obesity since a very early stage of vascular aging.17 A measure of arterial stiffness, brachial‐ankle pulse wave velocity (baPWV), has been found to be in close association with other surrogate cardiovascular indices.18 Many indicators of vascular function are associated with body fat.19 Although obesity may be a factor in arterial remodeling, which is accompanied by other hemodynamic and arterial changes,20 vascular endothelial function in the obese population may still be normal,15 even those with obesity may have improved arterial elasticity.21 The effects of obesity alone, without comorbidities in the elderly, have not been established. Therefore, our hypothesis is that if elderly patients with high body mass and blood pressure (BP) are found to have “youthful” arterial elasticity and show normal or even lower pulse wave velocity (PWV), this may be a vascular mechanism that could contribute to the “obesity paradox.”

2. METHODS

2.1. Study population

A total of 628 participants who received a routine physical examination in the Community Department of Health Assessment Center between 2016 and 2017 were recruited from the Xuhui Community in Shanghai. Main exclusion criteria were those participants with presence of a malignant disease, with any cardiovascular symptoms and who had taken antihypertensive medication. A total of 214 subjects were excluded from the analysis because BMI of 35 subjects, including 19 subjects with missing baPWV, of age >60 years was not measured, and those of age <60 years (179 subjects) were also excluded. The number of participants included in the present analysis was 414 (211 males and 203 females, age 60‐99 years) including 233 normotensive and 181 treatment‐naïve hypertensive elderly subjects. There were 163 subjects of >80 years of age. Body height and weight were measured without shoes using a portable stadiometer. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared (kg/m²). Subjects were divided into three groups according to BMI ( ≤25 kg/m², normal; 25‐28 kg/m², overweight; and ≥28 kg/m², obese) based on published values for the Chinese population.22, 23 All subjects provided written informed consent, and the Ethics Committee of Ruijin Hospital North, Shanghai, approved the study protocol.

2.2. Indices of arterial stiffness and atherosclerosis

Noninvasive measurements of peripheral BP and PWV were performed in the supine position using the Omron device (BP‐203RPEIII VP‐1000). The subjects were required not to smoke nor take any vasoactive medication or caffein for 8 hours before the study, which may influence BP and baPWV. After the subject had rested at least for10 minutes in a quiet room in the supine position at a temperature of 22°C, the appropriate four pressure cuffs were wrapped on bilateral upper arms and ankles to measure blood pressure and record the pulse wave form.

The device simultaneously and automatically measured BP in both ankles and upper arms and distance according to the height of the subject. baPWV was calculated using the equation:

$$\text{baPWV}\left[\text{cm/s}\right]=\frac{\text{La}-\text{Lb}}{\text{Tba}}$$

where La is path length from the heart to the ankle, Lb is the path length from the heart to the brachium, and Tba is the delay time from the brachial waveform to the ankle waveform.24 Ankle‐brachial index (ABI) was shown by the device as the ratio of the systolic blood pressures at the ankle to the systolic blood pressures in the upper arm (brachial). An ABI value of 0.9‐1.2 was considered normal, with ABI < 0.9 indicating arterial disease.25 The average value of right and left baPWV or ABI was calculated and analyzed in this study.

2.3. Statistical analysis

All analyses were performed using SPSS 24.0 for Windows (SPSS Inc). A 2‐sided P < .05 was considered statistically significant. Continuous variables are expressed as mean ± SD. One‐way ANOVA analyses was used for comparisons across the three BMI groups. The Pearson coefficient correlation and partial correlation (adjusted for age, sex, and heart rate (HR) or systolic BP were used to assess the relations between BMI and baPWV and ABI, respectively. The association of BMI with baPWV was assessed by univariate and linear regression.

3. RESULTS

Characteristics of the three BMI groups are shown in Table 1. A total of 414 subjects (mean age, 77 ± 11 years; 51% male) were studied. Compared with normal weight, brachial systolic (BSBP) and diastolic (BDBP) pressures were significantly higher in the overweight/obesity group (both P < .05). ABI and HR did not differ significantly. There were no differences in baPWV between normal and overweight/obese groups (P > .05), although a slightly nonsignificant trend for reduction was observed between normal weight and overweight groups. When adjusted for age, sex, and HR, BSBP was significantly higher as BMI increased (135 ± 18.4, 138 ± 18.3, 147 ± 17.6 mm Hg; P = .003), and baPWV was significantly lower as BMI increased (baPWV 1830 ± 272, 1793 ± 279, 1704 ± 279cm/s; P = .008) in the three groups described by mean ± SD, even with additional adjustment for BSBP by means of univariate analysis.

Table 1.

Subjects grouped by their body mass index, n = 414

	Total	Normal weight BMI＜25 kg/m2	Overweight BMI 25 ~ 28 kg/m2	Obese BMI ≥ 28 kg/m2	P value
N	414	229	125	60
Age (y)	76.6 ± 11.5	78.3 ± 12.0	73 ± 10**	77.5 ± 10.7	<.001
BSBP (mm Hg)	137 ± 18.5	135.0 ± 18.4	137.0 ± 17.5	144.8 ± 19.0**	.001
BDBP (mm Hg)	77.3 ± 10.4	76.3 ± 10.6	77.9 ± 9.5	79.8 ± 11.3*	.047
HR (beat/min)	7.6 ± 11.6	71.7 ± 11.4	7.8 ± 12.9	73.2 ± 9.5	.218
baPWV (cm/s)	1800 ± 391	1825 ± 401	1749 ± 353	1812 ± 422	.204
ABI	1.09 ± 0.12	1.09 ± 0.12	1.10 ± 0.13	1.08 ± 0.13	.430
BMI (kg/m2)	24.7 ± 3.5	22.3 ± 2.2	26.4 ± 0.9**	30.2 ± 2.1**	<.001
Height (cm)	164 ± 8.5	164 ± 8.6	165.4 ± 8.7	161.2 ± 6.9*	.007

Values are mean ± SD.

Abbreviations: ABI: ankle brachial index; baPWV: brachial‐ankle pulse wave velocity; BDBP: brachial diastolic blood pressure; BMI: body mass index; BSBP: brachial systolic blood pressure; HR: heart rate.

^* P < 0.05 with normal weight group.

^** P < 0.001 with normal weight group.

Data are shown separately for males and females (Table 2). For the whole population, age and height were higher in males compared to females; however, there was no significant difference in baPWV between males and females in both overweight and obese groups between BMI classes. Pearson's and partial correlations are shown in Table 3. BMI significantly correlated positively with BSBP and BDBP (r = .170, r = .157, P < .001, respectively) but not with baPWV or ABI (P > 0.05). Following adjustment for age, sex, height, and BSBP, however, BMI showed a significant negative correlation with baPWV (r = −.170, P < .001).

Table 2.

Subjects grouped by their body mass index separately for males and females

	Total	Normal weight BMI＜25 kg/m2	Overweight BMI 25 ~ 28 kg/m2	Obese BMI ≥ 28 kg/m2
N (male/female）	211/203	119/110	69/56	23/37
Age (y)	78/75**	80/77*	74/72	82/75*
BSBP (mm Hg)	135/139**	132/139**	136/138	147/144
BDBP (mm Hg)	77/77	76/77	78/78	81/79
HR (beat/min)	70/75**	69/75**	72/75	71/74
baPWV (cm/s)	1764/1839	1779/1875	1735/1765	1768/1841
ABI	1.10/1.10	1.11/1.07*	1.08/1.13*	1.07/1.08
BMI (kg/m2)	24/25	22/22	26/27	30/30
Height (cm)	170/158**	170/158**	170/159**	167/157**

Values are mean.

Abbreviations: ABI: ankle brachial index; baPWV: brachial‐ankle pulse wave velocity; BDBP: brachial diastolic blood pressure; BMI: body mass index; BSBP: brachial systolic blood pressure.

^** P < 0.01 with female.

^* P < 0.05 with female.

Table 3.

Pearson's and partial correlation coefficients adjusted by age, sex, heart rate, height, and BSBP between variables and BMI

Variables	BMI (kg/m2) unadjusted		BMI (kg/m2) adjusted
	r	P	r	P
BSBP (mm Hg)	.170	.001	—	—
BDBP (mm Hg)	.157	.001	−.009	.860
baPWV (cm/s)	−.082	.094	−.170	.001
ABI	−.013	.800	−.045	.365

Abbreviations: ABI: ankle brachial index; baPWV: brachial‐ankle pulse wave velocity; BDBP: brachial diastolic blood pressure; BMI: body mass index; BSBP: brachial systolic blood pressure.

Using baPWV as the independent continuous variable in multiple linear regression, age, BSBP, and HR were all positively associated with baPWV, whereas BMI was negatively associated with baPWV (β = −.126, P < .001) after adjusting for confounding factors such as age, sex, height, BSBP, BMI, and HR (Table 4). With age as a continuous variable and BMI as categorical variable, BMI still showed a significant negative association with baPWV (β = −.088, P = .013). In addition, with BMI and age as categorical variables, the study showed that age older than 80 years was positively associated with baPWV, whereas BMI ≥ 25 kg/m² was negatively associated with baPWV (β =−0.100, P = 0.006) when adjusted for age (≥80 years vs.<80 years), sex, BMI (≥25 kg/m² vs. <25), height, HR, and mean BP. As BMI is also related to height, collinearity between BMI and height was tested using variance inflation factor, which resulted in a value of <2.4. This indicated that there was no problem with collinearity and height remained in the regression model.

Table 4.

Linear regression analysis determinants of baPWV

Variable	B	SE	baPWV (cm/s)		R 2
			β	P value
Model 1
Age (y)	12.2	1.28	.359	<.001	.527
Sex (female vs male）	−27.9	41.11	−.036	.0498
BMI(kg/m2)	−14.2	4.03	−.126	<.001
BSBP (mm Hg)	11.2	0.75	.531	<.001
HR (beat/min)	5.5	1.18	.163	<.001
Height (cm)	−5.9	2.41	−.129	.014
Model 2
Age (y)	12.7	1.28	.372	<.001	.520
Sex (female vs male)	−27.2	41.45	−.035	.511
BMI (≥25 kg/m2vs. <25 kg/m2)	−69.0	27.74	−.088	.013
BSBP (mm Hg)	10.9	0.75	.519	<.001
HR (beat/min)	5.4	1.19	.161	<.001
Height (cm)	−5.5	2.42	−.120	.023
Model 3
Age (≥80y vs.<80y)	262.3	30.35	.328	<.001	.497
Sex (female vs male)	−54.4	41.95	−.07	.195
BMI (≥25kg/m2vs. <25kg/m2)	−78.2	28.32	−.100	.006
BSBP (mm Hg)	11.6	0.76	.551	<.001
HR (beat/min)	5.99	1.23	.178	<.001
Height (cm)	−8.1	2.42	−.176	.001

Adjusted by age, sex, BMI, height, and BSBP.

Abbreviations: baPWV: brachial‐ankle pulse wave velocity; BMI: body mass index; BSBP: brachial systolic blood pressure; HR: heart rate.

In multivariate regression analysis of individuals of age <80 years, after adjusting for confounding factors such as age, sex, height, and BSBP, we found that BMI was negatively associated with baPWV (β = −.190, P < .001). However, there was no significant relationship between BMI and baPWV (β = −.094, P = .162) in age ≥80 years (Table 5). HR (β = .162, β = .206; P < .01, respectively) and BSBP (β = .635, β = .555; P < .001, respectively) showed a positive association with baPWV for both age <80 years and age ≥80 years.

Table 5.

Linear regression analysis of associations between BMI and baPWV stratified by age level

Variable	B	SE	baPWV (cm/s)		R 2
			β	P value
Age < 80 y
Height (cm)	−10.7	2.86	−.254	<.001	.527
Sex (female vs male)	−75.3	48.89	−.105	.125
BMI (kg/m2)	−21.1	4.97	−.190	<.001
BSBP (mm Hg)	13.2	0.94	.635	<.001
HR (beat/min)	4.8	1.34	.162	<.001
Age ≥ 80 y
Height (cm)	−5.2	4.15	−.116	.210	.345
Sex (female vs male)	−10.3	73.72	−.013	.889
BMI (kg/m2)	−9.8	6.99	−.094	.162
BSBP (mm Hg)	10.4	1.26	.555	<.001
HR (beat/min)	7.8	2.47	.206	.002

Abbreviations: baPWV: brachial‐ankle pulse wave velocity; BMI, body mass index; BSBP: brachial systolic blood pressure; HR: heart rate.

When the data were divided into tertiles of SBP (<128, 128‐144, >144 mm Hg) and pooled for the overweight and obese BMI groups (Figure 1), we found that the trend of the reduced baPWV with increase in BMI reached statistical significance in the second tertile (P = .012). For this tertile, a 24.7% increase in BMI was significantly associated with 7% decrease in baPWV (P = .012). baPWV increased with BSBP for both the normal weight group and overweight/obesity group. Compared to normal weight group (r = .614), baPWV was lower in the overweight and obesity group at the same BSBP (r = .558; P < .001; Figure 2).

Figure 1.

The relationship between BSBP divided into tertiles (≤128, 128‐144, ≥144 mm Hg) and baPWV in normal, overweight, and obese BMI. baPWV, brachial‐ankle pulse wave velocity; BMI, body mass index; BSBP, brachial systolic blood pressure. *P = .012 compared to the normal weight

Figure 2.

Regression lines between brachial SPB and baPWV in overweight, obesity, and normal weight elderly subjects. Overweight/obesity, continuous line and square; normal weight, dotted line and circle; baPWV, brachial‐ankle pulse wave velocity; SBP, systolic blood pressure

4. DISCUSSION

The present study performed in a total of 414 elderly subjects (51% male) showed that BSBP was significantly higher as BMI increased after adjusting for confounding factors, whereas baPWV was significantly lower as BMI increased in the three BMI groups. BMI showed a negative correlation with baPWV after adjusting for confounding factors. baPWV was significantly lower in the overweight/obesity group in the SBP range of 128‐144 mm Hg. Thus, baPWV increased at a slower rate with BSBP in the overweight/obesity group compared to the normal weight group. baPWV was significantly lower in the overweight/obesity group in the SBP range of 128‐144 mm Hg. baPWV showed significant negative association with BMI especially in the elderly of age less than 80 years, whereas it was positively associated with age, BSBP, and HR in the whole population.

Several longitudinal studies7, 9 elucidated an obesity paradox in the elderly with different comorbidities and body composition. A recent study26 suggested that high BMI was positively associated with a lower mortality in elderly patients (≥65 years) with atrial fibrillation, whereas no benefit was observed in younger (<65 years) patients. For patients with ST‐segment elevation myocardial infarction (STEMI), another recent study27 demonstrated that mild obesity was associated with lower long‐term risk of outcome, whereas normal weight and extreme obesity were associated with worse outcomes. A study involving more than 120,000 patients with acute myocardial infarction (AMI)16showed that overweight and obesity were associated with lifespan longevity at all ages, but morbid obesity was only associated with better survival in patients ≥75 years of age at the time of acute MI. The obesity paradox has also been shown to exist in the general elderly population.7, 27 A population‐based study6 suggested that for the general population, the mortality (AMI, chronic heart failure, stroke, heart disease) was 18%~36% lower for obese class I relative to normal weight. A community‐based cohort study in Japan28 also reported that the obese population had a lower risk of all‐cause mortality, compared with normal weight population, with the multivariate hazard ratios (95% confidence interval) being 0.86 (0.62‐1.19) for obesity and 0.83 (0.73‐0.94) for overweight, respectively. However, underweight populations had the highest mortality (hazard ratio = 1.60, 1.40‐1.82). There are several potential speculations for the obesity paradox in the elderly. Firstly, for the frail elderly, the potential protective effect of adipose tissue is greater than that of adverse cardiovascular effects by obesity.26 For example, elevation of triacylglycerol level caused by obesity independently extends the lifespan through energy exhaustion.29 Additionally, lean cells without triacylglycerol biosynthetic capabilities can be subjected to apoptosis during the process of aging.29 Secondly, those with cardiovascular disease caused by overweight and obesity might be given more aggressive and earlier drug therapy, such as ACEI/ARB and beta blockers,26 according to relevant hypertension guidelines. Thirdly, the elderly obese have been shown to have better cognitive function than normal weight elderly, not affected by age or ethnicity, and so enabling elderly patients to be beneficial from executive function processes with high compliance.30

By contrast, normal or underweight populations are more likely to have a poor nutritional status, such as anemia or low albumin levels, which may affect the prognosis of patients with higher risk of chronic heart failure, chronic obstructive pulmonary disease, chronic kidney disease, dementia, and cancer.16 Finally, obesity, as a condition of nutrient energy reserves, can respond to acute stress events and meet the increased metabolic requirements, leading to survival benefits, which have been proven by several studies in subjects with STEMI and AMI.16, 27 A study also showed7 that those with higher BMI often have youthful arteries with lower PWV and greater elasticity than those with normal BMI, hence providing a potential explanation for the obesity paradox.

PWV is an important biomarker to predict cardiovascular risk.31 Approximately 40% of our participants had a baPWV > 2000 cm/s, which is similar or higher than those values reported in studies of patients with CVD32, 33 (where increased baPWV was defined as >1426 cm/s). These studies found that metabolically abnormal normal weight subjects had increased baPWV compared to metabolically healthy normal weight metabolically healthy obese. PWV is associated with BP, age, gender, height, heart rate, and other factors. However, BP and age are the two major factors influencing PWV. Therefore, higher baPWV is attributed to old age and higher BP.

Our study showed there were no significant differences between normal and obese groups and the statistical significance was reached because of the multiple adjustments. BMI was independently and negatively correlated with baPWV, which suggests that there is still an obesity paradox in Chinese elderly. The present study suggests that those with overweight and obesity have better vascular function in relation to stiffness of large arteries and so might be associated with lower risk of cardiovascular events. BMI has also been shown to be associated with severity of carotid siphon calcification34 with carotid siphon calcifications also being associated with abnormal ankle‐brachial index. Our study also suggested that BMI insignificantly correlated with ABI. ABI is closely related to arterial wall calcification and atherosclerosis, suggesting that obesity has protective effects on vascular function, and the specific mechanisms need further study. In addition, our study showed that age, BP, gender, and HR were positively correlated with baPWV and negatively correlated with ABI. These findings are in agreement with those in previous studies.35, 36

There are several limitations to be acknowledged in the current study. First, the study was performed by cross‐sectional design and so provides the association of relationships instead of predictive values of BMI on the progression of arterial stiffness. Second, there is lack of data on history of medication and other previous disease status. Due to the health screening protocols, single point measurements of blood pressure and baPWV were obtained, although measurement was made with the participant in a stable state. Third, the age of the participants is over 70 years old, which limits the applicable findings only to the Chinese elderly population with smaller samples. Finally, there was no measurement of fat mass or distribution in this study. BMI does not directly reflect body composition (fat mass and lean mass) and physical habits; however, skeletal muscle mass37 and physical fitness38 counteract risk with obesity. In addition, fat people have fat upper arms and much thinner arms below the elbow, and the full pressure in the cuff is not evenly transmitted to the artery above the elbow especially in the very obese, which can lead to an overestimation of BP when using a cylindrical cuff. Troncoconical cuffs should be used in obese individuals.39

5. CONCLUSION

For the elderly in China, BMI shows an independent negative correlation with baPWV especially for age <80 years. It is speculated that the improvement in vascular function may be one of the possible causes of the obesity paradox. These findings can explain at least some features of the “obesity paradox.”

CONFLICT OF INTEREST

The authors declare there is no conflict of interest in the current study.

Yang H, Zhao J, Deng X, et al. Pulse wave velocity is decreased with obesity in an elderly Chinese population. J Clin Hypertens. 2019;21:1379–1385. 10.1111/jch.13659