Effects of Obesity, Body Composition, and Adiponectin on Carotid Intima-Media Thickness in Healthy Women

Abstract

Context

Increased common carotid intima-media thickness (IMT) is predictive of coronary artery disease and stroke.

Objective

In this study, we investigated common carotid IMT by obesity category in a cohort of healthy women without previously known cardiovascular disease.

Design, Setting, Participants, and Main Outcome Measures

One hundred healthy women (aged 24-59 yr) from the general community enrolled in an observational study conducted at an academic medical center participated in the study. B-mode ultrasound imaging of the common carotid arteries was used to measure common carotid IMT in 99 subjects. Fat distribution was determined by computed tomography. Hormonal and inflammatory parameters related to cardiovascular disease and obesity were measured.

Results

IMT was higher in obese [body mass index (BMI) ≥ 30 kg/m²], compared with overweight women (BMI ≥ 25 and < 30 kg/m²) [0.69 mm, interquartile range (IQR) 0.60-0.75 mm] vs. 0.62 mm [IQR 0.56-0.68 mm), P = 0.044] and in comparison with lean women (BMI < 25 kg/m²) [0.69 mm (IQR 0.60-0.75 mm) vs. 0.59 mm (IQR 0.54-0.67 mm), P = 0.016]. In multivariate modeling, age (beta = 0.0050 mm change in IMT per year of age, P = 0.003), smoking (beta = 0.0044 mm change in IMT per pack-year, P = 0.046), and sc abdominal adiposity (beta = 0.00026 mm change in IMT per square centimeter, P = 0.010) were positively associated with IMT, whereas adiponectin (beta =–0.0042 mm change in IMT per milligram per liter, P = 0.045) was negatively associated with IMT. Visceral adiposity (beta = 0.00048 mm change in IMT per square centimeter, P = 0.092) was not significantly associated with IMT after adjusting for age, race, smoking, sc abdominal adiposity, and adiponectin.

Conclusions

Obesity is associated with increased common carotid IMT in young and middle-aged women. Adiponectin and sc abdominal adiposity are associated with carotid IMT in this population.

Carotid intima-media thickness (IMT) is a measure of subclinical atherosclerosis and is predictive of future myocardial infarction and stroke (1-4). In this study, we investigated common carotid IMT among predominantly young, healthy women without previously known cardiovascular disease. Traditional risk factors of cardiovascular disease are known to correlate with carotid IMT (5). Obesity, an increasing epidemic in our nation and globally, is associated with higher risk of coronary heart disease and stroke (6-8). Obesity during childhood has been associated with higher carotid IMT during adulthood (9-11). In this study, we examined differences in IMT between categories of obesity as defined by the World Health Organization. Furthermore, we investigated the effects of adiponectin as well as depot-specific fat content (abdominal visceral fat area and sc abdominal fat area) compared with traditional risk factors for the prediction of increased carotid IMT and early atherosclerosis in this population.

Subjects and Methods

Subjects

Subjects were recruited from October 2000 to May 2004. Recruitment was accomplished through community advertisement and primary care provider referral. Consecutive subjects were enrolled if they met eligibility criteria of age between 18 and 60 yr, female gender, and body mass index (BMI) greater than 20 kg/m² and less than 35 kg/m². The subjects were healthy; without known acute or chronic diseases including any cardiovascular disease or past or current diabetes mellitus; and were not receiving medication for diabetes, hypertension, or dyslipidemia. Exclusion criteria also included use of steroids, GH, oral contraception pills, or any anabolic agents within 3 months of the study. Subjects were excluded if they engaged in substance abuse; were pregnant or had breast-fed in the past year; had a history of oophorectomy; or had had an acute infection within 3 months of the study.

The study was approved by the Massachusetts General Hospital (MGH) and Massachusetts Institute of Technology (MIT) institutional review committees, and subjects provided written informed consent. Eligible subjects were seen at the MIT and MGH General Clinical Research Centers. All testing was performed in the follicular phase when appropriate, after a 12-h overnight fast. Insulin and glucose were measured at 0 and 120 min after a standard 75-g oral glucose challenge.

Biochemical indices

Adiponectin was measured using a RIA (Linco Research, St. Charles, MO). Intraassay and interassay coefficients of variation (CVs) range from 1.8 to 6.2% and 6.9 to 9.3%, respectively. Insulin levels were measured in serum using a RIA (Diagnostic Products Corp., Los Angeles, CA). Intraassay and interassay CVs range from 3.1 to 9.3% and 4.9 to 10.0%, respectively. C-reactive protein (CRP) was measured by ELISA (Diagnostic Systems Laboratories, Webster, TX) with intraassay and interassay CVs ranging from 1.7 to 3.9% and 2.8 to 5.1%, respectively. The sensitivity of the assay is 0.0002 mg/liter. Homocysteine was measured by recombinant enzymatic cycling assay (Carolina Liquid Chemistries, Brea, CA). Intraassay and interassay CVs range from 1.8 to 3.0% and 1.0 to 3.0%, respectively. Low-density lipoprotein (LDL) cholesterol was calculated using the Friedewald formula. Total cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, and glucose were measured using standard techniques. Serum FSH was measured using a solid-phase immunoradiometric assay (Diagnostic Products Corp.) with an intraassay CV of 2.2-3.8%. SHBG level was measured by immunoradiometric assay (Esoterix, Austin, TX) with an intraassay CV of less than 4% and interassay CV of 7.8-10.6%. Homeostasis model assessment of insulin resistance (HOMA-IR) score was calculated by the formula: fasting serum insulin (microunits per milliliter) × fasting plasma glucose (millimoles per liter)/22.5 as described by Matthews et al. (12).

Common carotid IMT

Imaging was conducted using a high-resolution 7.5-MHz phased-array transducer (SONOS 2000/2500; Hewlett-Packard, Andover, MA). Digital images were captured directly to a Windows NT workstation using a high-quality, high-speed frame capture card made by Data Translation (Marlboro, MA). Subjects were positioned with a wedge of approximately 35 degrees such that the subject’s head and torso were at an incline to reduce respiratory variation and subsequent motion in the jugulars. Imaging of the left common carotid artery was performed with the subject turning her head 45 degrees to the right. Imaging was performed in B-mode, and the transducer was swept in cross-section to note the position and orientation of the bifurcation of the carotid artery. The transducer was then applied to the longitudinal view, with images acquired at two angles, 90 and 45 degrees. The 90-degree imaging plane is a frontal plane of the head at the common carotid artery. The 45-degree imaging plane at the common carotid artery is 45 degrees from the 90-degree plane. In each plane, the transducer was manipulated until the best image of the far wall of the distal 1 cm of the common carotid was acquired. Fifty-frame digital video clips of this region were acquired onto the Windows NT imaging workstation. Differences in interadven-titial diameter of the common carotid artery across the 50 frames were used to judge the cardiac cycle and select a frame of minimum diameter (diastole) as the analysis frame. Either the 90- or 45-degree image in diastole was selected as the best view for image quality. Edge detection and mean IMT calculation were accomplished with an in-house computer program. The published reproducibility of the technique is excellent with a sd of 0.007 mm (13). Reproducibility was further tested and determined to be 0.004 among 10 patients. The average IMT over the length of the measured segments on the left is reported.

Body composition

Weight and anthropometric measurements were determined in the morning, before breakfast. Anthropometric measurements were obtained using an inelastic tape measure by the bionutrition staff of the MIT General Clinical Research Center. To assess abdominal visceral and abdominal sc adipose tissue area, a cross-sectional abdominal computed tomography (CT) scan at the level of the L4 pedicle was performed. Scan parameters for each image were standardized (144 table height, 80 kV, 70 mA, 2 sec, 0.25 cm × four-slice thickness, 48 FOV). Fat attenuation coefficients were set at −50 to −250 HU (14).

Statistical methods

Continuous variables were tested for normality of distribution with the use of the Wilk-Shapiro test and examination of the histogram distribution. Because carotid IMT results were not normally distributed in our cohort, carotid IMT and other variables were compared among the three obesity categories by the nonparametric Kruskal-Wallis test. If statistical significance was met in the Kruskal-Wallis test, further comparisons of pairs of groups were performed using the Wilcoxon rank sum test. For dichotomous or categorical variables, comparison between the obesity groups was performed using the χ² test. Race was dichotomized to African-American or not. Forty-five of the 100 study subjects were African-American, 35 were Caucasian, and the other racial groups comprised only a small proportion of the cohort. Statistical significance was assumed when a null hypothesis could be rejected at P < 0.05. Spearman correlation coefficients were assessed in the univariate analysis comparing carotid IMT and other covariates. Multivariate regression analysis was performed to determine the association of covariates to the outcome of carotid IMT. A step-wise forward regression algorithm was used to select variables entering in the final standard least squares model. All predictors that were significant on univariate analysis and biologically plausible to affect IMT, including traditional cardiovascular risk parameters (blood pressure, lipids, inflammatory markers) were chosen for testing in the forward step-wise selection algorithm. Only those variables that entered the model at P < 0.1 were included in the final model. To test whether model assumptions for linear regression were met, we checked for normality of model residuals and equality of variances; both of these assumptions were met. All statistical analyses were performed using SAS JMP statistics software (version 5.1; SAS Institute Inc., Cary, NC).

Results

Characteristics of study subjects

One hundred women enrolled in the study; however, carotid IMT measurements were obtained in 99 participants. Age, demographic, biochemical, and body composition parameters are shown for all subjects and by obesity group in Table 1. Age, smoking, and homocysteine levels did not differ among the categories, but significant trends for increased systolic blood pressure, diastolic blood pressure, visceral fat area, sc abdominal fat area, triglyceride level, CRP, and HOMA-IR were seen across the obesity categories, whereas HDL-cholesterol, adiponectin, and SHBG decreased significantly with increasing weight category. Subjects in the highest obesity category (BMI ≥ 30 kg/m²) were more often African-American than subjects in the other obesity categories. Controlling for differences in race among the groups did not affect the significant relationship between IMT and BMI in our study population. Ten of the subjects were postmenopausal by FSH measurement, but FSH did not differ by obesity category (Table 1). One subject was on estrogen replacement therapy. IMT was higher in obese subjects (BMI ≥ 30 kg/m²), compared with overweight subjects (BMI ≥ 25 and < 30 kg/m²) [0.69 mm, interquartile range (IQR) 0.60-0.75 mm vs. 0.62 mm (IQR 0.56-0.68 mm), P = 0.044] and in comparison with lean women (BMI < 25) [0.69 mm (IQR 0.60-0.75 mm) vs. 0.59 mm (IQR 0.54-0.67 mm), P = 0.016] (Fig. 1).

TABLE 1.

Clinical characteristics of study subjects

	All subjects	Total n	BMI < 25 kg/m2	25 ≤ BMI < 30 kg/m2	BMI ≥ 30 kg/m2	P value
Demographics
Age, yr	40 (24–59)	100	40 (25–59)	39 (24–50)	39 (33–51)	0.72
Race		100				0.013
Caucasian	35 (35%)		16 (43%)	16 (41%)	3 (13%)
African-American	45 (45%)		14 (38%)	14 (36%)	17 (71%)
Hispanic	13 (13%)		2 (5%)	7 (18%)	4 (10%)
Asian	4 (4%)		3 (8%)	1 (3%)	0
Other	3 (3%)		2 (5%)	1 (3%)	0
Current smokers	28 (28%)		8 (22%)	12 (31%)	8 (33%)	0.53
Pack-years among   smokers	5.0 (1.0–24.0)	32	6.9 (0.6–24.0)	2.9 (0.1–18.0)	4.4 (1.4–10.0)	0.38
Body composition
BMI, kg/m2	25.9 (24.0–29.8)	100	23.6 (18.8–24.98)	27.4 (25.0–29.8)	33.0 (30.1–38.5)a,b	<0.0001
Visceral abdominal fat,   cm2	60.8 (9.5–256.1)	100	39.7 (9.5–123.8)	65.5 (26.2–213.7)a	108.1 (34.2–256.1)a,b	<0.0001
Subcutaneous   abdominal fat, cm2	273.9 (77.6–643.5)	100	171.3 (77.6–338.7)	282.0 (173.1–443.0)a	440.7 (272.5–643.5)a,b	<0.0001
Metabolic and endocrine    parameters
Total cholesterol, mg/dl	175 (100–324)	97	174 (128–324)	178 (100–263)	169 (116–226)	0.58
LDL-cholesterol, mg/dl	106 (49–231)	97	99 (57–231)	109 (49–196)	109 (51–173)	0.76
HDL-cholesterol, mg/dl	55 (28–95)	97	64 (26–132)	51 (36–90)a	53 (32–95)a	0.002
Triglycerides, mg/dl	66 (26–280)	97	59 (52–71)	73 (38–280)a	67 (39–163)	0.015
Adiponectin, mg/liter	7.0 (1.4–31.8)	84	8.5 (1.4–31.8)	6.3 (2.2–21.5)	5.7 (1.5–29.6)a	0.045
SHBG, nmol/liter	73.5 (19.0–231.0)	92	92.0 (41.0–231.0)	71.5 (19.0–148.0)a	56.0 (19.0–179.0)a	0.003
FSH, IU/liter	5.6 (2.4–174.0)	95	5.4 (2.4–174.0)	5.6 (2.5–156.0)	5.9 (2.6–29.7)	0.88
HOMA-IR	1.19 (0.34–4.75)	94	0.93 (0.44–3.46)	1.28 (0.34–4.75)a	1.42 (0.71–3.76)a	<0.001
Hemodynamic parameters    and cardiovascular markers
Systolic blood pressure,   mm Hg	112 (72–141)	100	102 (72–141)	115 (92–140)a	118 (96–135)a	<0.001
Diastolic blood   pressure, mm Hg	69 (47–94)	100	62 (47–92)	71 (49–94)a	72 (58–92)a	0.007
Homocysteine, μmol/   liter	6.8 (3.8–12.9)	74	7.0 (5.0–11.6)	6.5 (3.8–11.8)	6.3 (4.6–12.9)	0.38
CRP, mg/liter	1.3 (0.1–25.5)	74	0.8 (0.1–4.3)	1.5 (0.2–8.5)a	3.9 (0.5–25.5)a,b	<0.0001
IMT, mm	0.61 (0.45–1.11)	99	0.59 (0.46–0.96)	0.62 (0.45–1.11)	0.69 (0.49–0.87)a,b	0.035

Results are reported as median with minimum and maximum values in parentheses for continuous variables. For race and smoking, percentages are reported in parentheses. P values reported for Kruskal-Wallis test for continuous variables. For categorical variables, race and smoking status, P values are reported for χ² test. For race, χ² test performed for African-American or not.

^aP < 0.05 vs. BMI < 25 kg/m².

^bP < 0.05 vs. BMI 25–30 kg/m².

For conversion from milligrams per deciliter to millimoles per liter for total cholesterol, HDL-cholesterol, and LDL-cholesterol, multiply by 0.0259, and to convert from milligrams per deciliter to millimoles per liter for triglycerides, multiply by 0.0113.

FIG. 1.

Differences in carotid IMT according to World Health Organization obesity categories. Error bars represent IQR. *, P < 0.05 vs. BMI less than 25 kg/m²; †, P < 0.05 vs. BMI 25–30 kg/m².

Univariate analysis

In univariate regression analysis, carotid IMT was significantly related to age (rho = 0.35, P < 0.001), diastolic blood pressure (rho = 0.29, P = 0.004), BMI (rho = 0.30, P = 0.003), visceral fat area (rho = 0.38, P = 0.0001), sc abdominal fat area (rho = 0.30, P = 0.003), triglyceride level (rho = 0.22, P = 0.033), HOMA-IR score (rho = 0.22, P = 0.034), and SHBG (rho =−0.21, P = 0.047) and tended to relate to adiponectin (rho =–0.20, P = 0.077) (Table 2). Total cholesterol, LDL-cholesterol, HDL-cholesterol, CRP, and homocysteine were not significantly related to carotid IMT on univariate analysis (Table 2).

TABLE 2.

Univariate analysis of correlation with carotid IMT

Variable	Spearman correlation coefficient (rho)	P value for Spearman rho
Age	0.35	<0.001
Smoking pack-years	0.15	0.15
BMI	0.30	0.003
Visceral abdominal fat area	0.38	0.0001
Subcutaneous abdominal  fat area	0.30	0.003
Total cholesterol	0.11	0.27
LDL-cholesterol	0.10	0.31
HDL-cholesterol	−0.13	0.23
Triglycerides	0.22	0.033
Adiponectin	−0.20	0.077
SHBG	−0.21	0.047
FSH	0.11	0.31
HOMA-IR	0.22	0.034
Systolic blood pressure	0.20	0.051
Diastolic blood pressure	0.29	0.004
CRP	0.15	0.21
Homocysteine	−0.16	0.19

Multivariate analysis

Candidate variables that met statistical significance on univariate analysis and additional parameters that were hy-pothesized to have plausible biological relationship with carotid IMT were included in a stepwise regression analysis, with P < 0.10 as a selection cutoff to determine final variables in the model. Variables tested for initial entry into the model for IMT included age, race (African-American or not), smoking pack-years, systolic blood pressure, diastolic blood pressure, BMI, visceral fat area, sc abdominal fat area, total cholesterol, triglycerides, LDL-cholesterol, HDL-cholesterol, adiponectin, CRP, SHBG, FSH, and HOMA-IR. The variables identified as significant by the step-wise forward selection procedure in the model were visceral fat area, smoking pack-years, sc abdominal fat area, age, and adiponectin in that order. We then used these parameters to construct a final standard least-squares model. In addition, we forced race into the final model because of its potential confounding relationship with common carotid IMT. The total R² for the final whole model is 0.39, and the parameter estimates (beta) and P values from the standard least squares model are shown in Table 3. In multivariate modeling, age (beta = 0.0050 mm change in IMT per year of age, P = 0.003), smoking (beta = 0.0044 mm change in IMT per pack-year, P = 0.046), and sc abdominal fat area (beta = 0.00026 mm change in IMT per square centimeter, P = 0.0095) were positivelyassociated with IMT. Adiponectin (beta =–0.0042 mm change in IMT per milligram per liter, P = 0.045) was found to have a significant negative association with IMT in the model controlling for age, race, smoking, sc abdominal adiposity, and visceral fat area. In contrast, the relationship between visceral fat area and carotid IMT (beta = 0.00048 mm change in IMT per square centimeter, P = 0.092) was no longer significant after adjusting for age, race, smoking, sc abdominal adiposity, and adiponectin. When we constructed a model including HOMA-IR, the overall model R² worsened in comparison with the final model selected and the se of the covariate adiponectin rose 15%, indicating collinearity between adiponectin and HOMA-IR. Therefore, HOMA-IR was not forced into our final model.

TABLE 3.

Relationships between adjusted covariates and carotid IMT (mm) in multivariable regression modeling

Parameter	Estimate (beta)	SE	P value
Age, yr	0.0050	0.0016	0.003
Race (African-American or not)	−0.0062	0.011	0.57
Smoking pack-years	0.0044	0.0021	0.046
Subcutaneous abdominal fat  area, cm2	0.00026	0.000099	0.010
Adiponectin, mg/liter	−0.0042	0.0021	0.045
Visceral abdominal fat area, cm2	0.00048	0.00028	0.092

Variables included in the final model were age, race, smoking pack-years, visceral abdominal fat area, and sc abdominal fat area.

Because sc abdominal fat area and visceral fat area could be proxy measures for generalized obesity, we also constructed an alternate model adjusting for BMI. When we included BMI in the model, the overall R² was not improved. Age (beta = 0.0051 mm per year increase in age, P = 0.0027) and adiponectin (beta =–0.0045 mm change in IMT per milligram per liter, P = 0.03) remained significantly associated with IMT. After controlling for BMI, sc abdominal fat tended to be associated with IMT (P = 0.08), but the se increased without change in the beta estimate. Similarly, for visceral fat area, the beta estimate remained unchanged, whereas the se increased, again suggesting collinearity. Therefore, to avoid overadjustment, BMI was not included in our final model investigating the relationship of individual fat depots to IMT.

One woman in the study was taking estrogen replacement therapy. When we repeated the multivariate analysis excluding this subject, the significant relationships among age, adiponectin, sc abdominal fat, and smoking with IMT remained unchanged.

Assumptions for use of linear regression were met because model residuals were found to be normally distributed (Wilk-Shapiro P = 0.32), and equal variances assumption was met.

Discussion

In this study, we demonstrate increased carotid IMT based on degree of obesity in young, healthy women, without known cardiovascular disease. We investigated biochemical and detailed body composition data and identified a number of significant factors associated with carotid IMT. Traditional risk factors of atherosclerosis such as age, diastolic blood pressure, and triglyceride levels were significantly associated with carotid IMT on univariate analysis. In multivariate modeling, however, age, smoking, adiponectin, and sc abdominal fat area were the factors significantly related to IMT in healthy women across a range of weights. In contrast, intraabdominal visceral fat area was no longer significantly associated with IMT after adjustment for age, race, smoking, adiponectin, and sc abdominal fat. These data suggest novel factors that may contribute to the development of atherosclerotic disease in women.

To our knowledge, prior studies have not investigated the relationship between BMI, fat distribution, adipocytokines, and IMT in young, healthy women. Previous studies have shown an association between obesity and carotid IMT including longitudinal studies that demonstrated associations between childhood obesity and carotid IMT in adulthood (10, 11, 15). These analyses included both men and women. When men and women were analyzed separately in the Muscatine study, childhood BMI was found to be a significant predictor of carotid IMT measured in young women but not in men (9).

Studies investigating factors associated with carotid IMT in women are limited. In a study of 86 premenopausal women, De Pergola et al. (16) found carotid IMT to be inversely related to insulin sensitivity after adjusting for age, BMI, waist circumference, mean blood pressure levels, plasma glucose, and lipids. In the Healthy Women Study and the Women’s Healthy Lifestyle Project, premenopausal BMI was found to be significantly correlated with postmenopausal IMT during longitudinal follow-up (17). In another smaller study of 43 obese premenopausal women of similar age to our cohort, baseline common carotid IMT was significantly greater in obese women than 19 age-matched lean controls. After a 3-month weight reduction program, carotid IMT decreased in the women who lost weight (18), suggesting the reversibility of IMT changes with weight reduction. In contrast to the aforementioned studies in women, we further characterized fat distribution by CT imaging instead of waist to hip ratio or BMI alone and examined the potential roles of adiponectin and CRP.

A notable finding in our study is the fact that sc abdominalfat was more strongly associated with increased IMT than visceral adiposity. Prior studies (19, 20) suggested a deleterious effect of abdominal adiposity on atherosclerosis development. Other studies (21, 22) showed positive associations between waist circumference or abdominal sagittal and transverse diameters and carotid IMT; however, these associations were no longer significant after adjustment for BMI. Leptin, a marker of overall fat mass, has been shown to correlate with carotid IMT, independent of age, insulin sensitivity, smoking, systolic blood pressure, fasting glucose, and lipids but not after adjustment for BMI (23). Liu et al. (24) used ultrasound imaging to assess fat compartments, finding mesenteric but not preperitoneal fat thickness as measured by ultrasound was correlated with carotid IMT in both men and women. A weak correlation between carotid IMT and sc fat thickness was shown in women but not in men in their study. Our data assessing specific fat depots using CT scanning further support the notion that the sc abdominal fat depot more than the visceral fat depot is related to IMT among an otherwise healthy cohort of young adult women. These data are in agreement with data from Kortelainen and Sarkioja (25) demonstrating that sc abdominal fat thickness correlated with degree of coronary atherosclerosis seen on autopsy in women between 15 and 50 yr of age. The sc abdominal fat depot may better reflect overall adiposity in otherwise healthy women, and therefore it remains unclear whether it is the sc abdominal fat depot or overall adiposity that contributes to increased IMT in this population.

We examined sc adipose tissue depot in the abdominal region; however, there may be differences between sc adipose tissue in different anatomical regions that we did not assess. Other investigators (26) proposed that accumulation of adipose tissue in the gluteofemoral region may be associated with protective effects, whereas sc fat accumulation in the abdominal region may increase cardiovascular risk. Furthermore, deep abdominal sc adipose tissue but not superficial sc adipose tissue has been associated with insulin resistance (27).

In our cohort, sc abdominal fat maintained a significant relationship with carotid IMT, even after adjusting for cholesterol levels (total cholesterol, triglycerides, LDL-cholesterol, HDL-cholesterol), age, visceral fat area, HOMA-IR, adiponectin, CRP, SHBG, smoking, and blood pressure measurements. On the other hand, after adjustment for these factors, visceral fat area was no longer significantly correlated with IMT. These findings suggest that sc adiposity in the abdominal region may exert an independent effect on atherosclerosis, whereas visceral adiposity might exert its effects through other factors such as adiponectin, triglycerides, or insulin resistance. Previous studies have demonstrated the differential expression of adiponectin in adipose tissue from different fat distributions. Lihn et al. (28) demonstrated reduced adiponectin gene expression in visceral compared with sc adipose tissue from both lean and obese women.

Unlike other adipocytokines, adiponectin is found to be paradoxically low in patients with obesity (29), type 2 diabetes mellitus (30), and coronary artery disease (31). Adiponectin may have antiatherosclerotic and antiinflammatory effects (32, 33). Adiponectin levels are associated inversely with cardiovascular disease in humans. In a prospective nested cohort study of 18,000 men from the Health Professionals Follow-Up Study, a doubling of adiponectin was correlated with a relative risk of cardiovascular events of 0.67, which rose to 0.80 when the investigators corrected for lipids, glycosylated hemoglobin, and CRP (34). In a study of 142 patients with coronary artery disease and 108 control patients, adiponectin correlated negatively with maximum axial IMT (35). In another study of 140 obese juveniles (mean age 13.5 yr), serum levels of adiponectin were negatively correlated with carotid IMT, even after controlling for BMI, HOMA-IR, cholesterol, triglycerides, blood pressure, CRP, gender, and age (36). In comparison, our study also supports the association between adiponectin and the development of early atherosclerotic disease in otherwise healthy young women. In contrast to adiponectin, CRP was not related to IMT among our study subjects.

SHBG had been previously found to be inversely related to carotid atherosclerosis in postmenopausal women in the Atherosclerosis Risk in Communities Cohort (37). The authors hypothesized that decreased SHBG levels may be an indication of insulin resistance because insulin inhibits SHBG production and SHBG may mediate its protective effect on atherosclerosis by affecting levels of bioavailable androgen and estrogen (37). In our study, SHBG was negatively correlated with carotid IMT in univariate but not multivariate analysis.

Cigarette smoking is a well-established cardiovascular risk factor, and in our study, smoking appeared to contribute independently to increased IMT. Our data reinforce the known association of smoking with atherosclerosis but suggest that this relationship exists even among young women without known cardiovascular disease and a shorter smoking duration.

Strengths of our study include the use of CT imaging to measure fat distribution and also the assessment of simultaneous inflammatory and traditional cardiovascular risk factors. In this regard, we are able to show a significant relationship between adiponectin and carotid IMT, controlling for measures of adiposity and other risk factors. One limitation of our study is the cross-sectional design. Although causality cannot be determined from cross-sectional studies, these data suggest the need for further longitudinal and treatment studies targeting low adiponectin, in addition to more traditional risk factors, as a strategy to prevent the development of atherosclerotic disease in women. Because our study included women only, there may be a gender-dependent effect of regional fat distribution and cardiovascular risks.

In summary, our data demonstrate that increased IMT is associated with obesity, even among healthy women without known cardiovascular disease. The accumulation of sc abdominal fat and the decrease of adiponectin levels seem to be important factors mediating the effect of obesity on the development of early atherosclerosis.

Abbreviations

BMI

Body mass index

CRP

C-reactive protein

CT

computed tomography

CV

coefficient of variation

HDL

high-density lipoprotein

HOMA-IR

homeostasis model assessment of insulin resistance

IMT

intima-media thickness

IQR

interquartile range

LDL

low-density lipoprotein