AFRICAN AMERICAN ETHNICITY AND CARDIOVASCULAR RISK FACTORS ARE RELATED TO AORTIC PULSE WAVE VELOCITY PROGRESSION

Abstract

Background

Accelerated central arterial stiffening as represented by progression of aortic pulse-wave velocity (PWV) may be influenced by cardiovascular disease (CVD) risk factors. Little is known about the relationships between CVD risk factors and PWV progression among women transitioning through the menopause, or whether these relationships vary by ethnicity. To address this knowledge gap, we conducted a subgroup analysis of 303 African American and Caucasian participants in the Study of Women's Health Across the Nation (SWAN) Heart Study received PWV scans at baseline examination and at a follow-up examination an average of 2.3 years later. CVD risk factors were also assessed at baseline.

Methods and Results

Systolic blood pressure (SBP) and waist circumference were the strongest predictors of PWV progression, after adjustment for age, low density lipoprotein cholesterol (LDL-C), high density lipoprotein cholesterol (HDL-C), diastolic blood pressure (DBP), glucose, and triglyceride levels. The magnitude of the influence of SBP, DBP, LDL-C, and glucose on PWV progression varied by ethnicity (difference in slopes: p=0.02 for SBP, p=0.0009 for DBP, p=0.005 for LDL-C, and p=0.02 for glucose). The positive relationship between SBP and PWV progression was significant among women of both ethnicities. LDL-C, DBP, and, to a lesser extent, glucose levels were positively associated with PWV progression only among African Americans.

Conclusions

Blood pressure, LDL-C, glucose, and excess body size may be important targets for improving vascular health and preventing clinical outcomes related to arterial stiffening, particularly among African American women.

INTRODUCTION

Arterial stiffening, as measured by aortic pulse-wave velocity (PWV), is an age-related process associated with clinical cardiovascular disease (CVD) events and mortality ^1-4 . However, premature arterial stiffening may occur as a response to CVD risk factors such as excess body weight, hypertension, type 2 diabetes mellitus, and the metabolic syndrome ⁵. These risk factors may also predict the progression of aortic PWV over time ^6-9.

African American women have greater incidence of and mortality from clinical cardiovascular events ¹⁰, and appear to have higher PWV than do Caucasians ^9,11. In addition, African American women have a greater prevalence of obesity, hypertension, and type 2 diabetes mellitus ¹². This disproportionate CVD risk factor burden may contribute to higher PWV progression among African American women.

The effects of certain CVD risk factors on arterial stiffness may also vary by ethnicity. Hypertensive African Americans have a cardiovascular mortality rate 3 to 5 times higher than that of hypertensive Caucasians,¹³ and hypertension is an especially robust predictor of coronary heart disease (CHD) among African American women¹². African Americans also present with higher systolic blood pressures (SBP) than do Caucasians beginning in early adulthood ¹⁴; greater arterial stiffness among African Americans may reflect a greater cumulative lifetime exposure to hypertension. Several studies also suggest that African Americans may be particularly susceptible to vascular remodeling after exposure to hypertension, high mean arterial pressures, or to cytokines and growth factors related to atherosclerosis ^15-17. Therefore, it is conceivable that African American women with hypertension or atherogenic risk factors at baseline may develop more accelerated arterial stiffening than do Caucasian women with similar baseline risk factors. This question has not been assessed in relation to PWV progression, and may identify a potential pathway by which African American women develop more prevalent clinical disease.

Identification and modification of risk factors that contribute to early arterial stiffening among women transitioning through the menopause may help to advance preventative care and improve clinical outcomes. In this study of CHD-free, middle-aged women, we assessed: 1) the relationships of baseline levels of SBP, DBP, waist circumference, LDL-C, HDL-C, glucose, and triglycerides with PWV progression, and 2) whether associations between PWV progression and these individual CVD risk factors varied by ethnicity. We hypothesized that risk factors related to blood pressure would be most strongly associated with accelerated PWV progression among African American women.

METHODS

Study Population

The SWAN study examines the changing biological and psychological health of women across the menopause transition. Descriptions of the study design and methods have been reported elsewhere ¹⁸. Between 1996 and 1997, 3302 pre-menopausal or peri-menopausal women were recruited to one of seven research centers: Detroit, MI; Boston, MA; Chicago, IL; Oakland, CA; Los Angeles, CA; Newark, NJ; and Pittsburgh, PA. Eligibility criteria for the SWAN were as follows: aged 42-52 at enrollment, intact uterus and at least one ovary, menstrual bleeding within the prior three months, no current pregnancy or breast-feeding, no usage of reproductive hormones within the prior three months, and self-identification as a member of 1 of 5 ethnic groups depending on the research site: Caucasian (all sites), African American (Boston, Chicago, Detroit, Pittsburgh), Chinese/Chinese American (Oakland), Hispanic (Newark) or Japanese/Japanese American (Los Angeles).

SWAN Heart is an ancillary study designed to assess subclinical CVD in mid-life African American and Caucasian women. SWAN participants from the Chicago and Pittsburgh sites (n=608) were recruited to the SWAN Heart cohort between 2001 and 2003 if they met the following criteria: no history of CHD, stroke, hysterectomy/postmenopausal status, and no usage of diabetes medications or hormone therapy.

Participants who met study criteria received their initial PWV measurements at SWAN Heart baseline (n=554). Of the original cohort, 316 participants returned for a follow-up PWV measurement an average of 2.3 years later (range: 1.1 - 4.3 years). Time between scans did not vary by ethnicity (p=0.55). Women who did not return were more likely to be African American, enrolled at the Chicago site, and had marginally higher SBP, glucose levels, and waist circumferences. HDL-C, DBP, LDL-C, age, triglycerides, educational status, pulse rate, or smoking status did not differ between women who returned for follow-up and those who did not. Additional participants were excluded from analysis: implausible PWV values or unreadable waveforms at either visit (n=7), missing plasma measurements (n=1), usage of anti-hypertensive medications at the screening exam (n=5). The final sample of 303 women included 204 Caucasians and 99 African Americans.

Measurement of Aortic Pulse Wave Velocity

Arterial stiffness was assessed by the gold standard, carotid-femoral PWV, using the protocol described previously¹⁹. In brief, arterial flow waves were simultaneously and non-invasively recorded at the right carotid and femoral arteries of supine participants, using unidirectional transcutaneous Doppler flow probes (model 810-a, 8-10 MHz, Parks Medical Electronics, Aloha, OR). Aortic PWV is calculated as distance/transit time (cm/seconds). The transit time of the pulse wave was calculated as the delay between the averaged carotid and femoral waveforms. The distance traveled by the pulse waveform was estimated by measurement over the participant's torso, above the plane of the body. The distance from the carotid to aortic site (at the suprasternal notch) was subtracted from the sum of the aortic to umbilicus and umbilicus to femoral site to adjust for the opposite direction of the blood flow in that arterial branch. For each participant, three data collection runs were performed and results were averaged. A higher PWV indicates a stiffer vessel. This measure has been demonstrated to have good reproducibility, with an overall laboratory intraclass correlation of 0.77¹⁹. PWV values at follow-up were normally distributed.

The following plasma, physical, and covariate measurements were taken at the SWAN visit coincident with or closest in time to each participant's baseline SWAN Heart visit (SWAN years 4-7).

Plasma Measures

Blood was drawn once, usually during days 2 to 5 of the follicular phase of the menstrual cycle and after a twelve hour fast. Plasma samples were maintained at 4°C until separated, and then were frozen at –80°C and shipped on dry ice to the Medical Research Laboratories (Lexington, KY), which is certified by the National Heart Lung and Blood Institute, Centers for Disease Control Part III program. Low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) were analyzed on EDTA-treated plasma. HDL-C was isolated using heparin-2M Mn(II)Cl. Triglycerides were analyzed by enzymatic methods via a Hitachi 747 analyzer (Boehringer Mannheim Diagnostics, Indianapolis, IN, USA). Glucose was quantified via a hexokinase-coupled reaction (Boehringer Mannheim Diagnostics).

Physical Measures

Resting blood pressure at baseline was measured three times after each participant had been seated for five minutes, and the average of the final two readings was recorded via a standard mercury sphygmomanometer. Hypertension was defined as SBP ≥140 mm Hg or DBP ≥90 mm Hg during the baseline visit, or current use of antihypertensive medications; this definition is generally in accordance with criteria developed by the Joint National Committee on Prevention, Detection, Treatment and Control of High Blood Pressure (JNC 7).²⁰ Participants wore light clothing for body size assessments. Height and weight were used to calculate BMI (kg/m²). Waist circumference was measured at the narrowest aspect of the torso as seen from the anterior aspect or at the smallest horizontal circumference between the lowest ribs and the iliac crest.

Covariate Measures

Following a 5-minute rest, pulse rate (beats per minute) was calculated by multiplying each participant's pulse at 30 seconds by a factor of two.

Ethnicity, age, educational attainment, smoking status, menopausal status, and study site (Chicago or Pittsburgh) were assessed via self- or interview-administered questionnaires. Antihypertensive medication use was self-reported and verified by medication review at baseline and follow-up examinations. Education was originally coded as one of five options, but was collapsed to the following three categories due to low frequencies: high school education or less, some college (including associate's and college degree), or postgraduate education. Current smoking status (yes/no/unknown) was defined as intake of at least one cigarette per day between the previous SWAN visit and the SWAN Heart baseline year. Women with missing smoking data (n=26) were assigned as unknown smoking status.

Menopause status was determined via self-reported menstrual patterns. Pre-menopause was defined as menstrual bleeding within the prior 3 months without changes in regularity, early peri-menopause was defined as menstrual bleeding within the prior 3 months with change in regularity, late peri-menopause was defined as no menstrual bleeding within the prior 3 months but bleeding within the prior 12 months, and natural post-menopause was defined as no menstrual bleeding within at least the prior 12 months. Hysterectomy was defined as removal of the uterus and/or double oophorectomy. Women who did not report menstrual bleeding patterns or who used exogenous estrogen within the past year were designated as unknown menopausal status. Due to small counts, menopausal status was categorized as follows: premenopause or early peri-menopause (n=175), late peri- or postmenopause (n=103), and hysterectomy/unknown/hormone use (n=25).

Data Analysis

T-tests and Chi-Square tests were used to assess ethnic differences in CVD risk factors and covariates. PWV progression was analyzed as follow-up PWV values adjusted simultaneously for both the continuous baseline PWV values and the time between SWAN Heart visits. To these minimally-adjusted models, baseline LDL-C, HDL-C, glucose, SBP, DBP, waist circumference, and triglycerides were added separately or in combination to investigate the relationships of CVD risk factors with PWV progression. Fully-adjusted models predicting follow-up PWV included the following covariates: baseline PWV measurements, time between SWAN Heart visits, participant age, menopausal status, education, smoking status, and pulse rate. BMI and waist circumference were highly correlated (r=0.90, p<0.0001). We retained waist circumference in the regression models as it explained more of the variance in PWV than did BMI, and was a predictor of PWV at follow-up in the models.

To formally test whether the effects of CVD risk factors on PWV progression were affected by ethnicity, interaction terms between ethnicity and each CVD risk factor were included in each minimally-adjusted linear regression model; those models with significant interactions were then stratified by ethnicity. Models were two-tailed with alpha=0.05. SAS version 9.1 software (SAS Institute, Cary, NC) was used for analyses.

RESULTS

Participants were on average fifty years old, most had at least some college education and most were pre- or early peri-menopausal. Participants had somewhat elevated BMI, LDL-C, SBP and waist circumferences. Mean pulse rate, DBP, HDL-C, fasting glucose, and triglyceride levels were within normal ranges. Baseline PWV ranged from 377.7 to 1531.3 cm/s, and follow-up PWV ranged from 334.6 to 1887.2 cm/s.

African American women had significantly higher SBP, DBP, BMI, and waist circumferences than did Caucasian women; Caucasian women had higher triglycerides and marginally higher HDL-C levels. At baseline, African American women reported significantly more anti-hypertensive medication usage than Caucasian women. Over the follow-up period, a similar number of African Americans and Caucasians commenced usage of anti-hypertensive medications (n=13 and n=10, respectively). Mean baseline PWV among African Americans was 35.5 cm/s higher than among Caucasians, but this difference was not statistically significant. Follow-up PWV was significantly higher among African American women (Table 1).

Table 1.

Characteristics of the Total Sample and by Ethnicity

	Total (n=303)	Caucasian (n=204)	African American (n=99)	P value (Ethnicity)
Age, years	50.1 (2.6)	50.3 (2.7)	49.7 (2.5)	0.07
BMI, kg/m2	28.6 (6.3)	27.8 (5.7)	30.4 (7.0)	0.002
SBP, mm Hg	117.5 (15.1)	114.0 (14.0)	124.8 (14.9)	<0.0001
DBP, mm Hg	74.8 (9.7)	72.6 (9.0)	79.2 (9.6)	<0.0001
Anti-hypertensive medication usage at Baseline %(N)	17.0 (51)	14.4 (29)	22.4 (22)	0.003
Waist Circumference, cm	87.6 (13.7)	86.2 (13.3)	90.5 (14.1)	0.012
LDL-C, mg/dL	119.6 (34.7)	119.3 (34.3)	120.3 (35.6)	0.82
HDL-C, mg/dL	57.7 (14.1)	58.7 (14.3)	55.6 (13.4)	0.08
Triglycerides, mg/dL	117.4 (88.4)	124.7 (100.0)	102.2 (55.8)	0.013
Glucose mg/dL	90.8 (15.3)	90.3 (16.8)	91.8 (11.5)	0.35
Baseline PWV, cm/s	800.1 (193.9)	788.6 (190.0)	824.1 (200.6)	0.14
Follow-up PWV, cm/s	868.5 (195.5)	839.5 (179.1)	928.4 (214.5)	0.0005
Pulse rate, beats/minute	70.8 (9.3)	70.4 (8.7)	71.7 (10.4)	0.31
Menopausal Status % (N)
Late Peri/Post-menopause	34.2 (104)	34.1 (70)	34.3 (34)	0.28
Pre-/Early Peri-menopause	57.6 (175)	55.6 (114)	61.6 (61)
Other or missing	8.2 (25)	10.2 (21)	4.0 (4)
Educational Level % (N)
High School or less	17.8 (52)	17.6 (35)	18.1 (17)	0.99
Any College	51.5 (151)	51.2 (102)	52.1 (49)
Post-Graduate Education	30.7 (90)	31.2 (62)	29.8 (28)

Mean (S.D.) unless indicated

Hypertension = SBP ≥ 140 mm Hg or DBP ≥ 90 mm Hg

Missings: DBP (n=5), Waist Circumference (n=9), LDL-C (n=1), Educational Level (n=11), anti-hypertensive medications (n=1), diagnosis of hypertension (n=1)\

Standardized regression coefficients of follow-up PWV on each CVD risk factor or ethnicity are presented in Table 2. African American ethnicity, SBP, DBP, waist circumference were each positively associated with follow-up PWV in minimally-and fully-adjusted models. HDL-C was negatively and significantly related to higher follow-up PWV in minimally-adjusted models, but this association was not significant in fully-adjusted models. LDL-C was marginally associated with higher follow-up PWV both in minimally- and fully-adjusted models. Triglyceride and glucose levels appeared to be unrelated to follow-up PWV.

Table 2.

Individual Regression Coefficients of Follow-up PWV Scores on Each CVD Risk Factor or Ethnicity

	Model 1 Adjustments			Model 2 Adjustments
	β (S.E.)	P value	Change in Model R2	β (S.E.)	P value	Change in Model R2
Ethnicity
African American	79.9 (22.7)	0.0005	0.04	86.8 (24.2)	0.0004	0.004
Caucasian	Referent			Referent
SBP	4.56 (0.68)	<0.0001	0.12	5.00 (0.76)	<0.0001	0.12
DBP	5.84 (1.08)	<0.0001	0.088	6.42 (1.20)	<0.0001	0.08
Waist Circumference	4.58 (0.79)	<0.0001	0.10	4.52 (0.83)	<0.0001	0.08
LDL-C	0.61 (0.32)	0.053	0.012	0.55 (0.33)	0.10	0.01
HDL-C	-1.70 (0.76)	0.03	0.02	-1.37 (0.81)	0.09	0.01
Triglycerides	0.08 (0.12)	0.51	0.002	0.06 (0.13)	0.66	0
Glucose	0.84 (0.72)	0.24	0.005	0.64 (0.76)	0.39	0

Model 1 Adjustments: Baseline PWV and time between scans, Model R² = 0.082; Model 2 Adjustments: Model 1 adjustments, age, pulse rate, study site, menopausal status, educational level, smoking status, Model R²= 0.11; Each Risk Factor or ethnicity was assessed individually with either Model 1 or Model 2 adjustments.

CVD risk factors and ethnicity were entered simultaneously into a fully-adjusted model predicting follow-up PWV (Table 3). SBP and waist circumference remained significantly associated with greater PWV progression. African American ethnicity was not independently related to higher PWV after adjustment for SBP: (β (S.E.): 35.4 (22.9), p=0.12). LDL-C, DBP, HDL-C, triglycerides, glucose, and the covariates were not related to follow-up PWV in this model.

Table 3.

Associations of CVD Risk Factors and Ethnicity on Follow-Up PWV Scores: Multivariate Linear Regression Models

	β (S.E.)	P value
Ethnicity
African American	34.6 (24.9)	0.17
Caucasian	Referent
SBP	3.46 (1.25)	0.006
DBP	0.34 (1.87)	0.86
Waist Circumference	2.92 (0.95)	0.0023
LDL-C	0.32 (0.32)	0.32
HDL-C	-0.29 (0.88)	0.74
Triglycerides	-0.15 (0.20)	0.47
Glucose	-0.18 (0.76)	0.81
Age	6.63 (4.57)	0.15
Pulse Rate	1.63 (1.18)	0.17
Study Site
Pittsburgh	-35.3 (24.8)	0.16
Chicago	Referent
Menopausal Status
Late	-14.6 (26.3)	0.58
Missing	3.48 (42.0)	0.93
Early	Referent
Educational Level
High School or Less	16.7 (32.5)	0.61
Any College	10.0 (24.4)	0.68
Post-graduate	Referent
Smoking Status
Smoker	20.2 (32.4)	0.53
Missing	-19.2 (55.7)	0.73
Non-Smoker	Referent

Model also adjusted for baseline PWV and time between scans; Model R²= 0.27

To determine whether associations between each risk factor and PWV progression varied by ethnicity, interaction terms between ethnicity and individual risk factors were examined using minimally-adjusted models. No effect modification was observed between ethnicity and waist circumference, HDL-C, or triglycerides. However, significant effect modification was observed between ethnicity and the following covariates: SBP (p=0.02), DBP (p=0.001), glucose (p=0.02), and LDL-C (p=0.005).

In ethnicity-stratified, minimally-adjusted models, SBP was significantly associated with higher follow-up PWV in both ethnicities (Figure 1). Among African-Americans only, DBP and LDLC were associated with higher follow-up PWV values: β (S.E.) for DBP: 10.1 (2.0), p<0.0001 (Figure 2); β (S.E.) for LDL-C: 1.74 (0.58), p=0.004 (Figure 3). DBP and LDL-C were not related to higher follow-up PWV values among Caucasians: β (S.E.) for DBP: 2.35 (1.36), p=0.09; β (S.E.) for LDL-C: 0.020 (0.36), p=0.95. Glucose levels were marginally associated with higher follow-up PWV among African Americans: β (S.E.): 3.51 (1.86), p=0.06 (Figure 4), but not among Caucasians: β (S.E.): 0.08(0.73), p=0.91.

Figure 1.

Relationship of SBP and PWV Progression by Ethnicity

Figure 2.

Relationship of DBP and PWV Progression by Ethnicity

Figure 3.

Relationship of LDL-C and PWV Progression by Ethnicity

Figure 4.

Relationship of Glucose and PWV Progression by Ethnicity

DISCUSSION

In the current study of middle-aged women, baseline SBP and waist circumference measures were most strongly associated with higher two-year progression of aortic PWV after adjustments for age, cardiovascular, biological, and psychosocial risk factors. DBP, LDL-C, and to a lesser extent, glucose levels were associated with greater PWV progression only among African American women. These relationships were not attributable to menopausal status, educational attainment, pulse rate, waist circumference, or smoking status. To our knowledge, our findings are novel in a population of healthy women. Furthermore, these findings suggest that the most robust CVD risk factors for women over a relatively brief follow-up period include blood pressure, LDL, and waist circumference; these risk factors may be particularly important to modulate in this population.

Our findings are consistent with existing literature. A recent review noted that blood pressure variables were most strongly associated with cross-sectional PWV across multiple studies; diabetes was also was consistently but more modestly associated with PWV²⁰. No associations between PWV and triglycerides or HDL were observed. In addition, Ferreira and colleagues observed in a study of men that Afro-Brazilians presented with a greater SBP-dependent increase in cross-sectional PWV measurements than did Caucasians; substantially higher PWV scores were observed among Afro-Brazilians with hypertension as compared to Caucasians with hypertension ¹⁶. Similarly, our findings indicate a robust association between blood pressure and PWV progression; HDL and triglycerides were also unrelated to PWV progression in our sample.

Reduction of elasticity in the aorta normally occurs with age, and is accompanied by increases in collagen content and hypertrophy of vascular smooth muscle cells ²¹. A stiffer artery is more susceptible to endothelial injury¹⁹ and may contribute to higher blood pressure; a reinforcing feedback cycle of pressure and vessel damage has been linked to clinical outcomes such as cerebrovascular disease ²². Because higher PWV is associated with greater risk of clinical cardiovascular disease events^1-4,10, the excess burden of clinical disease among African American women may be partly related to accelerated arterial stiffening. Results from the current study suggest that even relatively healthy African American women have a higher risk of arterial stiffening than do Caucasians. Given the excess prevalence of CVD risk factors such as obesity, hypertension, and type 2 diabetes mellitus among African American women ²³—and therefore, the higher potential for vascular damage-- our findings underscore the importance of early clinical management of these risk factors to prevent cardiovascular outcomes in this group²⁵.

It is unclear why the vasculature of African Americans may be more susceptible to adverse remodeling upon exposure to specific CVD risk factors. Given that African Americans disproportionately generate keloids in response to injury, investigators have hypothesized that a similar vascular process may occur in response to hypertension-related injuries or cytokine damage secondary to atherosclerosis ¹⁵. Increased intravascular collagen content could contribute to greater central arterial stiffness among African Americans. Alternatively, recent work by the Bogalusa Study indicated that luminal enlargement associated with atherosclerosis was more strongly related to mean arterial pressures and vascular wall thickness among young African American adults than among Caucasians ¹⁷. This finding suggests that in a milieu of atherosclerotic or pressure-related risk factors, the vasculature of African Americans is more inclined to undergo compensatory remodeling. This mechanism may initially preserve perfusion, but once the reserve capacity of the vessels is exhausted, the effects of CVD risk factors may be detrimental on the vasculature. Lastly, endothelial injury may enhance PWV progression by limiting the endogenous nitric oxide supply needed for normal vasodilation. African Americans generally have less of a vasodilatory response to nitric oxide than do Caucasians,²⁴ and damage to the endothelium from hypertension or through early plaque formation may disproportionately lead to accelerated PWV progression in this group.

Our study has several limitations. It is possible that CVD risk factors measured at baseline changed during the follow-up period for some women. Wildman and colleagues found that baseline measures of body size were most strongly associated with two-year progression of PWV, although several measures of change in body size were associated with PWV progression⁹. Similarly, the effect of changes in CVD risk factors over time should also be assessed to provide additional insight regarding how changes in CVD risk factors may exert positive or negative effects on arterial stiffening. For example, future work should examine the effects of hypertension on pulse wave velocity scores over time; it is also important to assess is whether usage of anti-hypertensive medications significantly modulates arterial stiffening among women with hypertension.

In conclusion, in this sample of healthy, middle-aged African American and Caucasian women, SBP and waist circumference were associated with PWV progression. Among African American women, SBP, DBP, and LDL-C levels were robust predictors of PWV progression, and glucose levels were marginally related to PWV progression. SBP was also associated with greater PWV progression among Caucasian women. Future research is needed to identify CVD risk factors that contribute to PWV progression among Caucasian women; clinical CVD is also the leading cause of mortality in this group. Future studies should also examine whether the current clinical recommendations for acceptable risk factor levels—particularly with regard to LDL-C, glucose, SBP and DBP-- are adequate to prevent adverse clinical outcomes among African American women.