Adipocytokine and ghrelin levels in relation to cardiovascular disease risk factors in women at midlife: longitudinal associations

Abstract

Background

There are limited data concerning the relationships between changes in adipocytokines and cardiovascular disease (CVD) risk factors.

Objective

To examine the longitudinal associations between leptin, adiponectin, resistin and ghrelin levels and CVD risk factor levels in women at midlife.

Design

Prospective, observational study.

Subjects and measurements

Leptin, adiponectin, resistin, ghrelin levels and CVD risk factors were measured in specimens collected from 40 women at 3 points in time corresponding to the pre-, peri- and postmenopause stages of their natural menopause transition.

Results

In longitudinal analyses adjusted for CVD risk factors and leptin at the previous menopausal stage, aging, education, smoking and physical activity, greater increases in leptin over the menopause transition were associated with greater decreases in high-density lipoprotein cholesterol (HDL-c) and greater increases in diastolic blood pressure, glucose, insulin and insulin resistance (all P<0.05). Larger decreases in adiponectin over the menopause transition were associated with greater increases in systolic blood pressure, insulin and insulin resistance and with greater decreases in HDL-c. Greater increases in ghrelin levels over the menopausal transition were associated with greater low-density lipoprotein cholesterol increases (P = 0.014). Resistin was not associated with CVD risk factor changes.

Conclusion

There were significant adverse associations of adipocytokines and ghrelin with multiple CVD risk factor changes in women across midlife. Given that this time period is dynamic for CVD risk, these data underscore the need for additional prospective studies.

Introduction

Excess weight increases the risk for stroke, incident cardiovascular disease (CVD), cardiovascular mortality and all cause mortality among the middle-aged and elderly.^1–4 This has fueled interest in determining the role of recently characterized regulators of energy metabolism, including the adipocytokines leptin, adiponectin and resistin, and the gut hormone ghrelin, in CVD risk. Higher leptin and resistin levels, and lower adiponectin levels have been described among persons with CVD compared to those without.^5–7 Similar relationships have been reported for leptin, adiponectin and ghrelin with incident CVDevents.^8–10

Leptin regulates body weight by suppressing appetite and stimulating energy expenditure.^11,12 Leptin levels are higher among obese individuals, suggesting leptin resistance.¹¹ Adiponectin is a matrix protein secreted by adipocytes;¹³ in contrast to leptin, lower, not higher, adiponectin levels are associated with obesity.¹¹ Resistin is released from macrophages both in adipose tissue and in the periphery.^14,15 Less is known about resistin in humans, but in mice, resistin is increased in obese states.¹⁶ Ghrelin is secreted primarily by cells in the fundus of the stomach. Acylated ghrelin alters feeding behavior and energy metabolism^17,18 and is down-regulated in human obesity.¹⁹

The mechanism underlying the relationships between adipocytokines/ghrelin and CVD events may be through increasing CVD risk factor levels, but there are limited prospective data examining the associations between adipocytokines/ghrelin and CVD risk factor changes. Because midlife is associated with changes in body composition, alterations in lipid and glucose metabolism and increased prevalence of metabolic syndrome among women,^20–23 the purpose of this study is to report the longitudinal associations between leptin, adiponectin, resistin and ghrelin levels and CVD risk factor levels among midlife women. We hypothesize that increases in leptin, resistin and ghrelin and decreases in adiponectin across midlife will be associated with adverse changes in CVD risk factors.

Methods

Study population and sample selection

Specimens and data are from the longitudinal Michigan Bone Health and Metabolism Study (MBHMS). The MBHMS enrollees are from two sampling frames; The first is the family records of the Tecumseh Community Health Study, a population-based prospective cohort study established in 1959 to study risk factors associated with common chronic and infectious diseases. This was supplemented by information from Kohl’s Directory to provide a listing of Tecumseh women whose families had not been in the Tecumseh Community Health Study. The collective MBHMS cohort is comprised of 664 Caucasian women, aged 24–44 years in 1992. Annual follow-up examinations include physical examinations, urine and blood specimen collection and interviews to characterize menstrual bleeding patterns and behaviors.

The current data are from a substudy nested within the longitudinal MBHMS. Women eligible for the substudy were unrelated and healthy, with endogenous hormone data and serum specimens available at three natural menopause transition stages associated with midlife change. Specimens were selected for assay from each of three menopause transition stages as is described in Figure 1. Women with hysterectomy or a bilateral oophorectomy, diabetes or thyroid disease, or women using hormone therapy were ineligible for consideration. Further, specimen selection was restricted to women in one of two baseline body size groups (body mass index (BMI) 23–27 or 29–36 kg m⁻²) to provide a wide range of adipocytokine levels.

Figure 1.

Selection process for nested, prospective study from population-based cohort (MBHMS) of women at midlife.

Written informed consent was obtained from all participants. This study was approved by the University of Michigan Institutional Review Board, and we certify that all applicable institutional and governmental regulations concerning the ethical use of human volunteers were followed during this research.

Measurements

MBHMS specimens from the pre-, peri- and postmenopausal stages of the menopausal transition were selected for assay. Fasting morning specimens were collected in days 2–7 of the follicular phase of the menstrual cycle. Specimens from postmenopausal women were collected on the anniversary of her study enrollment ±15 days. Biological samples from each annual visit were aliquoted and stored at−80 °C, without thaw, until assay.

Adipocytokines and ghrelin

Serum adipocytokines (leptin, adiponectin and resistin) and acylated ghrelin were determined spectrophotometrically using commercially available colorimetric enzyme immunoassay kits (leptin and acylated ghrelin kits (Cayman Chemical, Ann Arbor, MI, USA); adiponectin and resistin (Linco, St Charles, MO, USA)) according to the manufacturers’ instructions. The percent mean coefficient of variation and standard errors for duplicate samples for each subject and lower limit of detection, respectively, were as follows: adiponectin: 4.64 ± 0.34%, 1.5 ng ml⁻¹; leptin: 4.22 ± 0.36%, 1 ng ml⁻¹; resistin: 3.65 ± 0.24%, 0.16 ng ml⁻¹; and acylated ghrelin: 8.77 ± 1.24%, 2 pg ml⁻¹. Assays were undertaken in the laboratory of Dr Peter Mancuso in batch in 2006.

Lipids and glucose metabolism

Lipid and lipoprotein fractions, glucose and insulin were analyzed from sera; specimens from all years were assayed in batch in 2006. Total cholesterol and triglycerides were analyzed by enzymatic methods and high-density lipoprotein cholesterol (HDL-c) was isolated using heparin-2 M manganese chloride.²⁴ Serum insulin was measured using RIA (DPC Coat-a-count, Los Angeles, CA, USA) procedures and monitored as part of the monthly quality assurance program by the Diabetes Diagnostic Laboratory at the University of Missouri. Glucose was measured using a hexokinase-coupled reaction. Homeostasis model assessment-insulin resistance (HOMA-IR) as an estimate of insulin resistance was calculated (HOMA-IR = fasting serum glucose (mmol l⁻¹) × insulin (μIU ml⁻¹)/22.5).²⁵

Midlife time points

Change during midlife was characterized through the use of menopausal status. Menopausal status designation was based on self-report of menstrual bleeding frequency as well as follicle-stimulating hormone levels. Follicle-stimulating hormone concentrations were measured with a two-site chemiluminometric immunoassay directed to different regions on the beta subunit with CV% of 12.0 and 6.0% and a lower limit of detection of 1.05 mIU ml⁻¹. Women were classified as premenopausal with no increase in menstrual irregularity in the previous year and a follicle-stimulating hormone level >10 mIU ml⁻¹. Perimenopause was defined as an increase in menstrual irregularity and follicle-stimulating hormone level in the range between 10 and 40 mIU ml⁻¹. Postmenopause was characterized as at least 12 consecutive months of amenorrhea and a follicle-stimulating hormone level <40 mIU ml⁻¹.

Anthropometric and body composition measures

Annually, height and weight were measured with a stadiometer and balance beam scale, respectively. BMI was calculated by dividing the weight (in kilograms) by the square of the height (in meters). Annual waist measures were taken with a non-stretching tape at the narrowest location in the midsection of the torso. Bioelectrical impedance resistance and reactance measures were used to estimate total body water and, by extension, fat mass.²⁶ Annually, current physical activity level was measured with a modified self-administered Stanford Five-City instrument, where participants recalled the number of minutes of physical activity during the previous January and the previous July.²⁷ Because the number of questions about different activities was reduced across the 10-year follow-up period, ranks of physical activity levels are used in the current analyses rather than actual METS values.

Statistical analyses

Mean levels of demographic and laboratory characteristics at each time point were calculated using least squares means in the context of linear mixed effect modeling, to account for the correlation in repeated measures over time when comparing time points. The frequency of current smoking at each time point was compared using generalized estimating equations, again to account for the correlation in repeated measures over time. Longitudinal associations between adipocytokines/ghrelin and CVD risk factors across two midlife transitions (pre- to perimenopause and peri- to postmenopause) were assessed by fitting linear mixed models with two repeated measurements, adjusting for age, duration of each transition, education, concurrent smoking and concurrent physical activity. To examine whether body fat was an important attribute of longitudinal relationships between adipocytokines and CVD risk factors, additional adjustment was made for waist circumference and its change across the transitions. Waist circumference was chosen rather than total percent body fat due to the larger increase in abdominal adiposity across the menopause transition compared to total percent body fat. A compound symmetry covariance structure was assumed for all longitudinal models. To explore the possibility of the menopause transition as an effect modifier of relationships between adipocytokines and CVD risk factors, partial Spearman correlations were calculated, stratified by menopausal status. Additionally, multiplicative status-by-adipocytokine interaction terms were tested in the aforementioned longitudinal models. P-values <0.05 were considered statistically significant.

Results

Age increased across the midlife transition, with mean (s.d.) age at premenopause of 39.6 (0.6) years, at perimenopause of 46.3 (0.6) years and at postmenopause of 51.8 (0.6) years (P<0.001). Across time, mean blood pressure, triglyceride, glucose, HDL, BMI and waist circumference increased, while insulin levels decreased (Table 1). The prevalence of current smoking was 20.0% and the mean ranked METS value was 0.15 at premenopause, and neither significantly changed at successive time points.

Table 1.

Mean (s.e.) of laboratory and anthropometric characteristics at each time point

CVD risk factor	Premenopause	Perimenopause	Postmenopause	P-valuea
SBP, mm Hg	117.9 (2.2)	123.1 (2.2)	120.8 (2.2)	0.05
DBP, mm Hg	77.9 (1.5)	77.5 (1.5)	76.2 (1.5)	0.53
HDL-c, mmol l−1	1.2 (0.05)	1.3 (0.05)	1.4 (0.05)	<0.001
LDL-c, mmol l−1	3.6 (0.2)	3.7 (0.2)	3.8 (0.2)	0.64
Triglycerides, mmol l−1 b	1.3 (1.1, 1.6)	1.7 (1.4, 2.0)	1.7 (1.5, 2.1)	<0.001
Glucose, mmol l−1	4.8 (0.1)	5.3 (0.1)	5.5 (0.1)	<0.001
Insulin, pmol l−1 b	100.1 (81.6, 122.9)	59.2 (48.2, 72.8)	69.9 (56.8, 86.9)	<0.001
HOMA-IRb	3.0 (2.4, 3.8)	2.0 (1.6, 2.5)	2.4 (1.9, 3.1)	<0.001
BMI, kg m−2	28.9 (0.9)	30.3 (0.9)	30.9 (0.9)	<0.001
Waist, cm	85.3 (2.0)	91.6 (2.0)	96.2 (2.0)	<0.001
Fat mass, kgb	29.5 (26.4, 32.9)	30.0 (26.9, 33.5)	31.4 (28.2, 35.0)	0.06
Leptin, ng ml−1 b	22.6 (17.9, 28.7)	25.4 (20.1, 32.2)	31.2 (24.7, 39.6)	0.007
Adiponectin, ng ml−1 b	9.5 × 103 (8.5 × 103, 10.72103)	8.1 × 103 (7.3 × 103, 9.1 × 103)	9.3 × 103 (8.3 × 103, 10.4 × 103)	<0.001
Resistin, ng ml−1 b	42.6 (36.8, 49.1)	20.4 (17.7, 23.5)	21.0 (18.2, 24.3)	<0.001
Ghrelin, pg ml−1 b	7.9 (4.5, 13.8)	37.4 (21.4, 65.1)	8.0 (4.6, 14.0)	<0.001

Abbreviations: BMI: body mass index; CVD: cardiovascular disease; DBP: diastolic blood pressure; HDL-c: high-density lipoprotein cholesterol; HOMA-IR: homeostasis model assessment-insulin resistance; LDL-c: low-density lipoprotein cholesterol; SBP: systolic blood pressure.

^aP-value for change in CVD risk factor across time points.

^bValues are geometric mean (95% confidence interval).

Longitudinal associations between adipocytokines/ghrelin and CVD risk factors

In longitudinal analyses, higher leptin levels at the previous time point were associated with subsequent increases in diastolic blood pressure (beta per unit increase in log leptin at the previous time point = 5.4 mm Hg). Greater increases in leptin across the time points were associated with greater increases in diastolic blood pressure, glucose, insulin and HOMA-IR, and with greater decreases in HDL-c, even after adjustment for age, education, smoking status and physical activity level (Table 2).

Table 2.

Adjusted^a beta coefficients as measures of the longitudinal associations between leptin and adiponectin from longitudinal regression models of seven CVD risk factors

	SBP, mm Hg	DBP, mm Hg	HDL-c, mmol l−1	Triglycerides, mmol l−1	Glucose, mmol l−1	Insulin, pmol l−1	HOMA-IR
Leptin models
Log leptin change, ng ml−1 b	−0.4	5.6***	−0.1**	0.3	0.3*	23.5*	0.9*
Adiponectin models
Log adiponectin change, ng ml−1 b	−12.3*	−4.1	0.2**	−0.6	−0.6	−65.6*	−2.7*

Abbreviations: CVD: cardiovascular disease; DBP: diastolic blood pressure; HDL-c: high-density lipoprotein cholesterol; HOMA-IR: homeostasis model assessment-insulin resistance; SBP: systolic blood pressure.

^*Beta estimate P<0.05.

^**Beta estimate P<0.01.

^***Beta estimate P<0.001.

^aAdjusted for duration of menopause stage (time) and corresponding CVD risk factor from the previous menopause stage, age at the previous menopausal stage, education, current smoking, current physical activity and log leptin at the previous menopausal stage for the leptin model and log adiponectin from the previous menopausal stage for the adiponectin model.

^bBeta estimates can be interpreted as the change in the CVD risk factor associated with each log ng ml⁻¹ change in the adipocytokine across the previous menopausal status transition.

Lower adiponectin levels at the previous time point were not associated with subsequent changes in CVD risk factors. However, decreases in adiponectin levels over successive time points were associated with subsequent increases in systolic blood pressure, insulin levels and HOMA-IR, and decreases in HDL-c (Table 2).

It appeared that a portion of the longitudinal relationships were explained by waist circumference and its changes across the midlife. When the aforementioned models were additionally adjusted for waist circumference at the previous time point, and change in waist circumference across time points, statistically significant relationships remained only for leptin changes with glucose changes, and for adiponectin changes with HDL changes. However, although no longer statistically significant, the magnitude of the β coefficients associated with each adipocytokine was only slightly attenuated.

Ghrelin was associated with low-density lipoprotein cholesterol (LDL-c) increases across the midlife, while resistin levels were not associated with CVD risk factor changes. After adjustment, each unit increase in ghrelin across time points (in the log scale) was associated with a 0.1 mmol/L increase in LDL-c (P = 0.014).

Strength of correlations between adipocytokines and CVD risk factors at each menopausal state

Leptin levels increased from pre- to peri- to postmenopause, and resistin levels decreased, while adiponectin and ghrelin levels appeared altered only during the perimenopause (Table 1).

On the basis of Spearman correlations between leptin and adiponectin and the CVD risk factors by menopausal status (Figure 2), relationships between leptin levels and blood pressure may weaken across the menopause transition, while the perimenopause may be of particular importance for relationships between leptin levels and lipids and glucose homeostasis. Adiponectin relationships with CVD risk factors appear to strengthen across the menopause transition. Multiplicative status-by-adipocytokine interaction terms revealed a significant modification of the leptin/systolic blood pressure relationship by menopausal status (P = 0.04). None of the other interaction terms were statistically significant.

Figure 2.

Adjusted^† Spearman correlation coefficients between leptin (a) and adiponectin (b) and CVD risk factors, stratified by menopausal status. CVD: cardiovascular disease; DBP: diastolic blood pressure; HDL-c: high-density lipoprotein cholesterol; HOMA-IR: homeostasis model assessment-insulin resistance; SBP: systolic blood pressure.

Discussion

Levels of adipocytokines and ghrelin and their changes across midlife, indicated by three menopause states, were associated with changes in CVD risk factors. Additionally, these data are suggestive that menopause stage may influence the strength of association of adipocytokines with CVD risk factors and their changes. For leptin, the perimenopause stage, specifically, seemed to evidence a stronger association of leptin with lipids and glucose homeostasis, while for adiponectin, the effects of adiponectin on CVD risk strengthened monotonically across the menopause transition.

This current study showing associations between changes in both leptin and adiponectin and CVD risk factor changes is consistent with the existing limited literature. Franks et al.,²⁸ in a longitudinal study of adipocytokines with the development of the metabolic syndrome in women, also showed significant longitudinal associations between leptin levels and subsequent changes in metabolic parameters and the development of the metabolic syndrome. These longitudinal data are important additions to the literature, since cross-sectional studies are conflicting about leptin and blood pressure and lipid levels; some studies report that higher leptin levels are associated with adverse blood pressure and lipid profiles,^28–31 while some studies report null associations.^32–34 Adiponectin’s association with metabolic parameters has been more consistently reported. Moderately negative associations have been reported between adiponectin levels and blood pressure, lipid fractions and triglycerides in women,^35–37 and strong associations with the development of insulin resistance and diabetes.^38–40 We observed similar associations.

Abdominal adiposity and changes in abdominal adiposity across time points appeared to drive the associations between leptin and adiponectin and CVD risk factors. Given that adiponectin and leptin are released in largest part by adipose tissue, this is not surprising. However, the magnitude of the beta coefficients was not greatly altered with adjustment for abdominal adiposity, perhaps suggesting that with a larger sample size, relationships between leptin and adiponectin may have been found to be independent of changes in adiposity. We used waist circumference as it is the body size measure exhibiting the most change among midlife women, and women were selected for study based on their baseline BMI values. Prospective associations between leptin levels and the development of dyslipidemia and hypertension in women reported by Franks et al.²⁸ were not independent of baseline obesity status (obese/non-obese).

A number of physiological effects have been identified in experimental systems for leptin and adiponectin that may underlie relationships with blood pressure, lipids and glucose homeostatis. Leptin activates the AMPK (adenosine 5′-monophosphate-activated protein kinase) pathway in muscle and liver, decreasing cholesterol, fatty acid and triglyceride synthesis, and increasing glycolysis.^41,42 Leptin may also directly impact insulin secretion from pancreatic β cells, as well as activate the sympathetic nervous system.⁴³ Leptin apparently induces endothelial dysfunction, platelet aggregation, oxidative stress, pro-inflammatory mediator synthesis and vascular smooth muscle proliferation.⁴⁴ Adiponectin likely affects triglyceride metabolism and insulin receptor phosphorylation through regulation of AMPK, acyl CoA oxidase and peroxisome proliferator-activated receptor-γ (PPAR-γ).^45,46 Further, adiponectin knockout mice have reduced levels of endothelial nitric oxide synthase and prostacyclin in aortic tissues and plasma and reduced systolic blood pressure levels.⁴⁷

We identified a longitudinal association between ghrelin and LDL-c, but there is limited information about ghrelin’s role in CVD risk in the current literature. Ghrelin correlated negatively with blood pressure and positively with LDL particle size among middle-aged men.⁴⁸ However, a cross-sectional study among Korean women reported no significant correlations between plasma ghrelin levels and lipids, fasting insulin or blood pressure.³⁸ Continued study into the possible role of ghrelin in CVD risk is needed.

No significant longitudinal associations were identified between resistin levels and cardiovascular risk factor changes. Resistin levels did not change across the study period to the same extent as leptin, adiponectin or ghrelin levels compared to its standard deviation in premenopause, which may have limited the power to see associations with resistin. Previous studies of associations with resistin have been inconsistent not only with lipids and insulin resistance but also with obesity itself.^49–55

Hormone therapy has been shown to alter both leptin and adiponectin levels, and significant associations have been found between endogenous hormones and these adipokines.^56–60 However, to our knowledge, previous studies have not examined the role of menopause in the relationship between adipocytokines and CVD risk. The current data are suggestive, though not conclusive, of possible effect modification of the relationships between adipocytokines and CVD risk factors by the menopausal transition. Given the menopausal transition as a dynamic period for susceptibility of the vasculature to subclinical atherosclerosis and the atherosclerotic benefits of estradiol and lifestyle intervention,^61,62 these data underscore the need for larger future prospective studies powered to examine the role of menopause in modifying the effects of adipose tissue secretions on CVD risk.

This study had several strengths. First, it is among the first to simultaneously measure key adipocytokines and ghrelin, allowing for direct comparison and contrast of the strength and direction of relationships within the same sample. Second, it is also among the first to assess the longitudinal relationships between adipocytokines, adipocytokine changes and subsequent changes in CVD risk factors. Third, the inclusion of actively transitioning women in the current study allowed for an initial examination of the consistency of relationships between adipocytokines and CVD risk factors across the menopause transition, producing intriguing findings that require replication in a larger study.

This study also has limitations. There was insufficient power for testing of adipocytokine-by-menopause status interaction terms, and available power may have also limited the detection of significant risk factor relationships with resistin and ghrelin. Lastly, although these data are longitudinal, they cannot determine whether the associations between adipocytokines and ghrelin and CVD risk factor changes are causal.

This is among the first longitudinal studies to examine the relationships between adipocytokines, their changes and changes in CVD risk factors in women outside of a lifestyle, surgical or pharmacologic intervention. The current study reports significant associations of adipocytokine and ghrelin levels and their changes across midlife with changes in CVD risk factors, including blood pressure, lipid levels and insulin resistance. Given midlife as a period of increasing adiposity and CVD risk, these data underscore the need for additional prospective studies among midlife women.