The Association of Sex Hormones with Carotid Artery Distensibility in Men and Postmenopausal Women: Multi- Ethnic Study of Atherosclerosis

Abstract

The decline in carotid distensibility with age is steeper in women than in men, however, the correlates of this sex difference are not known.

We examined the association of bioavailable testosterone, estradiol, dehydroepiandrosterone and sex hormone binding globulin, in 2783 postmenopausal women and 2987 men aged 45–84 at the Multi-Ethnic Study of Atherosclerosis baseline examination. Carotid artery lumen diameters by ultrasound and brachial artery blood pressures were measured at systole and diastole. Regression models to determine the association of carotid distensibility coefficient and lumen diameter with sex-specific quartiles of sex hormones were adjusted for age, race, height, weight, diabetes, current smoking, antihypertensive medication use, total and high density lipoprotein cholesterol levels, and hormone replacement therapy in women. A higher DC indicates a more distensible vessel.

In women, higher dehydroepiandrosterone (p = 0.008) and lower sex hormone binding globulin (p = 0.039) were associated with lower distensibility; higher dehydroepiandrosterone and lower estradiol were associated with smaller carotid diameters. In men, higher Bio-T (p = 0.009) and lower estradiol (p = 0.007) were associated with greater distensibility and also with smaller diameters (p = 0.012 and 0.002, respectively).

An androgenic internal milieu is associated with lesser carotid distensibility and diameter remodeling in women, but the opposite is true for men. Higher levels of estradiol are associated with smaller carotid diameters in both sexes. Future longitudinal and experimental studies are needed to reveal the mechanism and clinical consequences of these associations.

Introduction

Older age is more strongly associated with lower carotid distensibility in women as compared to men aged 45–84 years (2.52×10⁻⁵ vs. 2.16×10⁻⁵ /mmHg lower distensibility coefficient (DC) per year of age, p=0.006).¹ One standard deviation lower carotid artery distensibility is associated with a 13–19% higher risk of strokes.² Thus sex differences in arterial stiffness, and thence stiffness related cardiovascular disease^3–6 may explain the observation in US and British national-level cohort analyses that age-related increase in heart disease mortality rates is blunted in men after 45 years of age but this is not seen in women.⁷ The basis of sex differences in carotid artery distensibility in middle aged and older adults are not well understood. Few published reports explore the association of sex hormones with arterial stiffness and distensibility, with inconsistent results in sex-specific studies and special populations for testosterone (T), estrogen and dehydroepiandrosterone (DHEA), as reviewed by Rossi et al.⁸

We aim to find the association of carotid artery distensibility with circulating levels of the sex hormones T, estradiol (E2), DHEA, and sex hormone binding globulin (SHBG) in a large population-based US sample of middle aged and older men and postmenopausal women.

Methods

Study Sample

We analyzed data from the baseline examination of the Multi-Ethnic Study of Atherosclerosis (MESA, 2000–2002) including 5761 adult men and women for the current analyses (enrollment strategy in Supplementary figure S1). The study was approved by the institutional review boards of all participating institutions. Participants gave written informed consent. All procedures complied with international, federal and institutional guidelines.

Clinical examination and risk factor covariates

Participants self-reported their age and race/ethnicity, education, use of medications, smoking, physical activity and menopause and the use of current hormone therapy in women. Seated blood pressure was measured as the average of the second and third readings taken using an automated oscillometric device. Total cholesterol categories (<200, 200–239, 240+) and High Density Lipoprotein (HDL)-cholesterol categories (<40, 40–59, 60+) were defined from fasting lipid profiles. Diabetes was defined as fasting blood glucose ≥ 7 mmol/L or self-reported antidiabetic medication use.

Carotid Artery imaging

Carotid distensibility was estimated from a 20-second ultrasound imaging record of the right common carotid artery and simultaneous brachial blood pressure was measured. Analysis was performed centrally using automated edge detection software. Common carotid intima-media thickness (IMT) was measured as mean of the maximum measurements of the near and far walls of the right common carotid artery.

Sex hormone measurements

Serum from fasting morning samples was immediately stored at −70°C. Assays for Total T, DHEA, estradiol and SHBG were performed at the University of Massachusetts Medical Center in Worcester, MA. Bioavailable testosterone (BioT) was calculated using equilibrium analysis.⁹

Detailed methods for clinical definitions, blood pressure measurement, carotid imaging and sex hormone measurement are provided in the supplementary material.

Statistical analysis

The demographic, cardiovascular risk profile, and sex hormone profile of men and women in the sample was tabulated.

All analyses to test the association of sex hormones with carotid distensibility were performed separately for men and women. Because sex hormone levels are right skewed, we used sex-specific hormone quartiles for these analyses.

If A is the arterial cross sectional area, P is the arterial pressure and Δ represents the change in diameter and pressure from diastole to systole Arterial distensibility coefficient (DC) is defined¹⁰ as:

$$\frac{\mathrm{\Delta }A}{A\mathrm{\Delta }P}\approx \frac{\mathrm{\Delta }(\text{log}A)}{\mathrm{\Delta }P}=(\mathit{\text{slopeof}}\text{log}A\mathit{\text{vs}}.P)$$

Higher DC indicates more distensible vessels. We took advantage of the “slope Δ(logA)/ΔP” definition of DC to calculate the adjusted association of sex hormones with DC. The slope adjusting for both logA and P was calculated in a single step in mixed model linear regression analysis by making the within-person log(diameter) the dependent variable and blood pressure the independent variable of linear regression analyses. The interaction of this slope with other independent variables gives the adjusted association of distensibility with the other independent variables. Illustrative examples of this regression model with small datasets have been previously presented,^{1, 11} and also included in the online appendix. The models adjusted for age, race, height, weight, the mean of systolic and diastolic blood pressure diabetes, current smoking, antihypertensive medication use, total cholesterol and HDL-cholesterol categories and only in women, the use of hormone replacement therapy. Preliminary analyses shows that sex-specific hormone quartiles for Bio-T only were associated with IMT, thus Bio-T analyses were also further adjusted for IMT. Heterogeneity by sex of hormone quartile linear trend with carotid artery properties was assessed using the test of differences using standard errors.

Supplementary data analysis

Association results for quartiles of total testosterone, and analysis stratifying women by current hormone replacement therapy use are presented in the data supplement. We tested if there was any statistically significant interactions by race in the association of any sex hormone with either carotid DC or diameter. We examined if the associations found in the main analysis were robust to further adjustment for (a) pulse rate measured during the imaging (b) educational attainment as a proxy for socioeconomic status (c) habitual moderate to vigorous weekly activity (d) carotid intima media thickness, (e) type of hormone therapy used (estrogen or estrogen and progesterone) for women and (f) medical or surgical menopause in women.

Results

The demographic, cardiovascular risk factor, carotid vascular measurement and sex hormone distributions show the profiles expected for this middle aged and older men and postmenopausal women in a population based sample (Table 1).

Table 1.

Demographic, Cardiometabolic, Vascular and Sex Hormone Characteristics of the Study Sample

Characteristic		Women	Men
Number		2783	2978
Age (years) mean ± SD		64.6 ± 9.1	62.2 ±10.3
Race
White		1044 (37.5%)	1164 (39.1%)
Chinese		336 (12.1%)	377(12.7%)
Black		785 (28.2%)	749 (25.2%)
Hispanic		618 (22.2%)	688 (23.1%)
BP medication use		1055 (37.9%)	906 (30.4%)
Diabetes Mellitus*		336 (12.1%)	409 (13.7%)
Current Smokers		307 (11.0%)	424 (14.2%)
Total cholesterol (mmol/L)		5.2 ± 0.9	4.9 ± 0.9
HDL cholesterol (mmol/L)		1.5 ± 0.4	1.2 ± 0.3
Any Current Hormone Therapy Use		895 (2.9%)
Types of Current Hormone Therapy:
Estrogen and Progesterone		482 (17.3%)
Estrogen only		328 (11.8%)
Type not available		85 (3.1%)
Menopause Type	Natural	2021 (72.6%)
Surgical		762 (27.4%)
Systolic BP (mmHg)		129.2 ± 23.5	126.0 ± 19.4
Diastolic BP (mmHg)		69.2 ± 10.3	75.0 ± 9.4
Carotid Diastolic Diameter (mm)		5.9 ± 0.7	6.4 ± 0.8
Carotid Systolic Diameter (mm)		6.3 ± 0.8	6.8 ± 0.8
Carotid Intima Media Thickness (mm)		0.88 ± 0.20	0.89 ± 0.22
Carotid Distensibility Coefficient (× 10−5 Pa−1)		1.7 [1.3 to .2]	1.9 [1.4 to 2.4]
Total Testosterone (mmol/L)		0.90 [0.58 to 1.32]	14.26 [11.38 to 17.84]
Bioavailable testosterone (mmol/L)		0.21 [0.10 to 0.35]	5.21 [4.23 to 6.49]
Estradiol (mmol/L)		0.073 [0.048 to 0.165]	0.114 [0.088 to 0.139]
Dehydroepiandrosterone (mmol/L)		10.31 [7.01 to 14.61]	12.49 [9.16 to 17.04]
Sex Hormone Binding Globulin (mmol/L)		59.1 [40.6 to 94.6]	40.8 [31.5 to 52.8]

Distributions shown as median [interquartile range], number (%), or mean ± standard deviation.

^*Diabetes was defined as fasting blood glucose ≥ 7 mmol/L or the use of antidiabetic medications,

^†Hypertension was defined as blood pressure ≥ 140/90 or antihypertensive medication use. 1 Pa⁻¹ = 133.32 mmHg⁻¹ (conventional units)

Association of Carotid Artery Properties with Sex Hormone quartiles in Women

Table 2 shows the association of sex hormone quartiles with carotid DC and diameter in women. Higher levels of the androgenic hormones bioavailable testosterone and DHEA were associated with less distensible (stiffer) carotid arteries. However the association of bio-T does not remain significant after adjustment for IMT. Higher levels of SHBG were associated with more distensible carotid arteries. Higher levels of E2 were associated with smaller diameter in women. Non-monotonic relationships of the association of carotid diameter were seen for DHEA in women, the highest quartile being associated with larger diameters.

Table 2.

Association of Sex Hormones with Carotid Artery Distensibility and Diameter in Women

Hormone Quartiles
Vascular Property	1st	2nd	3rd	4th	P(trend)
Bioavailable Testosterone
Δdist	Ref	−0.34 (−1.11 to 0.44)	−0.50 (−1.32 to 0.32)	−0.86 (−1.71 to −0.02)	0.046
Δdiam	Ref	0.04 (−1.14 to 1.21)	0.50 (−.76 to 1.75)	−0.32 (−1.65 to 1.00)	0.82
Bioavailable Testosterone*
Δdist	Ref	−0.15 (−0.53 to 0.24)	−0.21 (−0.62 to 0.20)	−0.34 (−0.76 to 0.09)	0.48
Δdiam	Ref	−0.05 (−1.23 to 1.13)	0.37 (−0.88 to 1.63)	−0.52 (−1.85 to 0.81)	0.82
Estradiol
Δdist	Ref	0.21 (−0.55 to 0.97)	−0.43 (−1.19 to 0.33)	−0.79 (−1.83 to 0.26)	0.11
Δdiam	Ref	−1.27 (−2.45 to −0.09)	−2.08 (−3.29 to −0.88)	−2.62 (−4.19 to −1.06)	<0.001
Dehydroepiandrosterone
Δdist	Ref	−0.18 (−0.95 to 0.58)	−0.93 (−1.71 to −0.14)	−0.92 (−1.72 to −0.11)	0.008
Δdiam	Ref	0.39 (−0.79 to 1.57)	−0.21 (−1.42 to 1.00)	1.56 (0.32 to 2.81)	0.040
Sex Hormone Binding Globulin
Δdist	Ref	0.67 (−0.11 to 1.44)	0.72 (−0.09 to 1.52)	1.00 (0.06 to 1.93)	0.039
Δdiam	Ref	−1.59 (−2.78 to −0.41)	0.06 (−1.18 to 1.31)	−0.67 (−2.13 to 0.79)	0.88

The association of carotid measures with total testosterone quartiles, which combines bioavailable testosterone and SHBG, are presented in the supplementary data table S1, where the association with DC in women is non-significant.

Analysis in women stratified by current HRT use is presented in supplementary data tables S2 and S3. The association of Bio-T with lower DC was stronger in women who were not on HRT, and non-significant in those on HRT. The associations of carotid DC and diameter were qualitatively similar in both strata of women for E2, DHEA and SHBG.

Association of Carotid Artery Properties with Sex Hormone quartiles in Men

In men, higher levels of bioavailable testosterone were associated with more distensible carotid arteries, although this association was significant at the borderline after adjustment for IMT (Table 3). Higher levels of estradiol were associated with less distensible carotid arteries. Higher levels of E2 were associated with smaller carotid diameter. Only the two highest quartiles of Bio-T were associated with smaller carotid diameters.

Table 3.

Association of Sex Hormones with Carotid Artery Distensibility and Diameter in Men

Hormone Quartiles
Vascular Property	1st	2nd	3rd	4th	P(trend)
Bioavailable Testosterone
Δdist	Ref	0.28 (−0.56 to 1.12)	0.57 (−0.30 to 1.44)	1.24 (0.31 to 2.17)	0.009
Δdiam	Ref	0.24 (−0.95 to 1.42)	−1.70 (−2.91 to −0.49)	−1.05 (−2.33 to 0.24)	0.012
Bioavailable Testosterone*
Δdist	Ref	0.13 (−0.29 to 0.56)	0.31 (−0.13 to 0.75)	0.63 (0.16 to 1.09)	0.057
Δdiam	Ref	0.32 (−0.86 to 1.50)	−1.75 (−2.96 to −0.55)	−1.15 (−2.42 to 0.13)	0.002
Estradiol
Δdist	Ref	0.08 (−0.76 to 0.91)	−0.05 (−0.92 to 0.81)	−1.25 (−2.12 to −0.39)	0.007
Δdiam	Ref	−1.53(−2.67 to −0.39)	−1.28 (−2.47 to −0.09)	−2.06 (−3.26 to −0.86)	0.002
Dehydroepiandrosterone
Δdist	Ref	0.27 (−0.57 to 1.12)	0.89 (−0.02 to 1.78)	0.07 (−0.89 to 1.04)	0.54
Δdiam	Ref	−0.67 (−1.86 to 0.52)	−1.52 (−2.76 to −0.28)	−0.26 (−1.59 to 1.06)	0.45
Sex Hormone Binding Globulin
Δdist	Ref	0.17 (−0.71 to 1.05)	−0.03 (−0.94 to 0.87)	0.40 (−0.54 to 1.34)	0.50
Δdiam	Ref	−0.11 (−1.29 to 1.08)	0.01 (−1.22 to 1.24)	−0.29 (−1.59 to 1.01)	0.72

Sex Differences in Linear Trends for Association of Carotid Artery Properties and Sex Hormone Quartiles

There was statistically significant sex difference between the linear trends of the Bio-T quartile association with DC (p = 0.001), DHEA quartile association with DC (p = 0.029), while the sex difference in DHEA quartile association with diameter was borderline significant (p = 0.051).

There were no statistically significant interactions by race for any of the associations. Our results were robust to further adjustment by factors included in sensitivity analyses (detailed in data supplement).

Discussion

Our analysis in a large population based sample shows that a more androgenic circulating milieu in terms of Bio-T and DHEA is associated with less distensible arteries in women, and DHEA is associated with larger arterial diameter. Higher bio-T had the opposite associations in men: higher DC and smaller diameter. Higher SHBG levels were associated with lower DC only in women. The bio-T associations may be mediated by thicker IMT.

Arterial stiffness is associated with cardiovascular events,^3–6 and low carotid distensibility is specifically associated with greater stroke risk.² Consistent with our findings, in prior reports, higher levels of testosterone in men were associated with lower arterial stiffness,^{12, 13} but SHBG or DHEA-sulfate were not,¹² and in a study of 120 postmenopausal women higher levels of estradiol and DHEA-sulfate were associated with greater arterial stiffness.¹⁴ Polycystic ovary syndrome, an androgenic state in women is associated with greater arterial stiffness,¹⁵ while lower testosterone and lower SHBG in men with erectile dysfunction are associated with greater pulse pressure, a measure of arterial stiffness.¹⁶

In MESA, higher levels of E2, testosterone, and DHEA in postmenopausal women were associated with longitudinal increases in blood pressure, but this was attributable to adiposity; the association of low SHBG with increasing blood pressure was independent of adiposity.¹⁷ This finding is consistent with our results in so far as arterial stiffness contributes to higher blood pressure. Cross sectional studies have shown that higher T¹⁸ and DHEA-sulfate¹⁹ levels were associated with high blood pressure in women, in agreement with our findings. Sex differences in cross sectional associations of high blood pressure in low SHBG are inconsistent in different studies:^{18, 20, 21} our findings show sex differences in cross-sectional associations with distensibility, a contributor to BP, show an association in women but not men.

Interventional studies show that exogenous testosterone is associated with lowering of blood pressure^{22, 23} and lowered arterial stiffness in hypogonadal men.²⁴ These findings may suggest that our observational associations may be due to the effect of androgenic sex hormones on arterial stiffness; however, we note that these interventional studies are in special populations, some with small sample sizes. In small studies, T therapy was not associated with any change in blood pressure in hysterectomized women,²⁵ and DHEA therapy was not associated with changes in arterial stiffness in men and women with adrenal deficiency.²⁶ In a small study including men and women, DHEA therapy improved arterial stiffness overall, however, the sample was too small to show sex differences in this effect.²⁷ Estrogen therapy was associated with inconsistent results, either no association with arterial stiffness^{28, 29} or some reduction in arterial stiffness.³⁰ The possible reduction in arterial stiffness in one small study is contrary to the expectation based on our large but cross sectional analysis. In MESA, we have previously shown that higher levels of E2 were associated cross-sectionally with glucose intolerance and diabetes,^{31, 32} and longitudinally with worsening of cardiometabolic profile including central obesity³³ in women and men and diabetes incidence in women.³⁴ We thus believe that higher E2 in postmenopausal women may be more likely to increase rather than decrease vascular risk.

Our findings of the association of carotid arterial size with sex hormones are novel. Larger arterial size is associated with a worse cardiovascular profile,³⁵ active coronary plaque volume³⁶ and cardiovascular events.³⁷ Our findings underline the complex relation of estrogen with arterial health: higher E2 was associated with smaller carotid diameter, but lower carotid DC.

Possible mechanisms of action of sex hormones on the vasculature from animal and cellular studies may include relaxation of vascular smooth muscle cells and increased endothelial nitric oxide production by estrogen (reviewed by Skafar et al.³⁸). Other cellular mechanisms include reduced matrix collagen deposition by estrogen treated human vascular smooth muscle cells as compared to testosterone treated cells, and matrix metalloproteinase expression induced by testosterone.³⁹ However, we cannot currently link these basic biological studies with the specific sex differences found in our study.

The large representative multiethnic sample with standardized clinical and imaging protocols is a strength of our study. Our regression methodology also appropriately adjusted for confounding by blood pressure in a single mathematical step. Our study has certain limitations: we did not measure estrone, the major circulating estrogen in postmenopausal women. However, E2 is the most potent circulating estrogen, thus our findings are likely to be valid. Another limitation is that we used brachial artery blood pressure as a proxy for carotid artery blood pressure. However, this estimate of carotid distensibility has been previously shown to predict clinical events,² and this approximation likely to be useful. Our study only measured distensibility in the right carotid artery. However, we do not believe that there should be a systematic difference in the associations of hormones with left versus right arterial properties.

Perspectives

We have found that a more androgenic circulating mileu is associated with worse carotid artery distensibility and larger diameters in women, but the opposite associations in men. Higher estradiol levels are associated with less distensible, smaller diameter vessels in both men and women. Future cellular and animal studies of the effects of sex steroids on vascular tissue may need to be designed to examine the mechanism of these sex differences. Future longitudinal population studies and larger intervention studies may reveal if these associations are associated with preventable clinical events.

Novelty and Significance.

1) What Is New

We show the association between sex hormone levels with arterial distensibility.

2) What Is Relevant?

An androgenic internal milieu is associated with lesser carotid distensibility and diameter remodeling in women, but the opposite is true for men. Higher levels of estradiol are associated with smaller carotid diameters in both sexes.

3) Summary

We examine a large multi-ethnic US population sample and show that circulating sex hormone levels affect vascular properties, and some associations differ between men and women. Our study provides a possible risk factor for stiff arteries in middle aged and older men and women.