Endothelial dysfunction and increased intima–media thickness (IMT) have been found in obese patients. Both regional fat distribution and liver steatosis may influence these markers of subclinical atherosclerosis. We sought to determine the interrelationships of endothelial function, carotid IMT, visceral and subcutaneous adipose tissue accumulation, and liver steatosis in severely obese subjects. In 64 severely obese patients (BMI 42.3 ± 4.3 kg/m²), we determined (i) endothelial function as flow-mediated dilation (FMD) of the brachial artery, (ii) carotid IMT, (iii) visceral fat diameter, and (iv) degree of liver steatosis using ultrasound. FMD was associated inversely with visceral fat diameter and degree of steatosis (r = −0.577, P < 0.0001 and r = −0.523, P < 0.0001, respectively). Carotid IMT correlated with visceral fat mass (r = 0.343, P = 0.007) but not with liver steatosis. After adjustment for conventional cardiovascular risk factors, FMD was predicted independently by the visceral fat diameter, age, and sex (r2 = 0.48, P < 0.0001), but not by the degree of liver steatosis or plasma adiponectin levels. In contrast, age and sex were the only predictors of IMT (r2 = 0.33, P < 0.001). In obese patients, visceral fat diameter is a major determinant of endothelial dysfunction, independent of traditional risk factors or the degree of liver steatosis and plasma adiponectin. Measurement of visceral fat diameter by ultrasound is a novel and simple method to identify subjects with an increased risk for atherosclerosis within an obese population.
Obesity is a major risk factor for the development of metabolic disorders and cardiovascular disease (1). Adulthood obesity is associated with an increased morbidity and mortality, partly due to cardiovascular events (2). Visceral fat has been shown to be correlated closer to cardiovascular disease than total body fat (3,4,5). Visceral obesity plays a central role in the development of insulin resistance, metabolic syndrome, type 2 diabetes, hypertension, and dyslipidemia, which are all well-documented risk factors for cardiovascular disease (5,6).
Nonalcoholic fatty liver disease is frequently found in obese subjects and in the metabolic syndrome (7). Recently, liver steatosis has been found to be associated with carotid atherosclerosis, as measured by ultrasound in the carotid arteries. Liver steatosis has been suggested to be a risk factor for carotid atherosclerosis beyond its association with the metabolic syndrome (8).
Endothelial function—measured by flow-mediated dilation (FMD)—as well as intima–media thickness (IMT) are established early markers of the atherosclerotic process and can be assessed noninvasively with high-resolution ultrasound. A growing body of evidence suggests that measurement of endothelial dysfunction and carotid IMT are predictive of cardiovascular events (9,10).
Previous reports described an association of visceral obesity with endothelial function. These studies usually used waist circumference and/or waist-to-hip ratio as an estimate of visceral obesity (5). However, measurement of these circumferences are only poorly standardized and strongly operator dependent, especially in severely obese patients. Moreover, waist circumference cannot distinguish between visceral and subcutaneous adipose tissue. Several studies used computed tomography (CT) scans to distinguish between visceral and subcutaneous fat mass confirming the association of visceral fat with endothelial function (11). However, CT scanning involves radiation and is more expensive compared to an ultrasound examination.
The interrelationships of endothelial function, carotid IMT, and sonographically determined visceral obesity and liver steatosis in severely obese patients have not been reported. Even more, biochemical pathways of such interactions have not been elucidated yet. Therefore, we hypothesized that endothelial dysfunction, increased IMT might be dependent on visceral fat diameter. We performed a cross-sectional evaluation of these relationships in 64 severely obese patients.
Methods and Procedures
Sixty-four obese subjects (BMI >35 kg/m2) were recruited consecutively at our metabolism outpatient department (12). Exclusion criteria were overt diabetes, blood pressure >160/90 mm Hg, history of cardiovascular disease, medical treatment with antihypertensive or lipid-lowering drugs, and alcohol consumption of >20 g per day. Acute infectious and inflammatory disease was excluded by taking a medical history and performing physical and laboratory examinations. All subjects gave written informed consent. The study protocol was approved by the local ethics committee at the Innsbruck Medical University.
Anthropometric measurements, vascular and abdominal ultrasound, and laboratory parameters were performed on the same day in all patients.
After a 12-h fasting period, blood was obtained for measurement of plasma lipids, glucose, insulin, and adiponectin. Glucose and fasting lipid levels were determined by standard laboratory methods. Serum free insulin concentrations were quantified by radioimmunoassay (Pharmacia, Uppsala, Sweden). Insulin resistance was estimated using the homeostasis model assessment for insulin resistance, as described (13). Adiponectin concentrations were determined using a commercially available enzyme-linked immunosorbent assay (Linco Research, St. Charles, MO). Body composition was determined by impedance analysis using InBody 3.0 Body Composition Analyzer from Biospace Europe (Dietzenbach, Germany) with an integrated scale in fasted patients.
All ultrasound examinations were performed by a single investigator (W.S.). Ultrasound examination was performed between 1 and 2 pm. All subjects had eaten a typical continental breakfast and refrained from caffeine, alcohol, food intake, and smoking thereafter. Images were stored and interpreted using calipers offline by the same reader (W.S.) on two separate days. The investigator was blinded to anthropometric and laboratory parameters.
Brachial artery study. Endothelial dependent dilation (FMD) was determined as described previously (14,15,16,17,18). The brachial artery was scanned in a longitudinal section 2–15 cm above the elbow with the use of a 14.0-MHz linear array transducer and a standard Acuson Sequoia 512 system (Acuson, Mountain View, CA). After recording a rest scan, a pneumatic cuff was placed around the forearm and inflated to a pressure of 250 mm Hg for 4.5 min, followed by pressure release. Five measurements of the brachial artery diameter in the end-diastolic phase of the cardiac cycle using electrocardiogram triggering were obtained 45–60 s after deflation. FMD was determined as percent diameter change relative to baseline measurements. Intraobserver coefficient of variation of brachial artery diameter was 0.5% corresponding to 0.13 s.e.m. or 0.013 s.d.
Carotid artery study. Longitudinal B-mode scans of the common carotid artery were obtained immediately after the brachial artery studies, using a 14.0-MHz linear array transducer. The far wall of the common carotid artery was assessed just proximal to the carotid bulb (last 2 cm) to identify the maximal IMT, defined as the distance between the junction of the lumen and the intima and that of the media and adventitia measured in the end-diastolic phase of the cardiac cycle using electrocardiogram triggering (19). Three measurements of the right and left carotid artery were averaged to determine the IMT. Intraobserver coefficient of variation of IMT was 5.3% corresponding to 1.6 s.e.m. or 2.0 s.d.
Abdominal ultrasound studies. Each patient underwent abdominal ultrasonography using a 3.0-MHz curved array transducer. Grading of liver steatosis was performed as described previously (20). Level 0 was defined as a normal hepatic echo pattern, level 1 as a slight increase in echo pattern with normal visualization of vessels and diaphragm, level 2 as a moderate increase in echogenicity with reduced visibility of portal veins and diaphragm, and level 3 as a pronounced increase in hepatic echo pattern with poor visibility of intrahepatic vessels and posterior right lobe of the liver. Subcutaneous and visceral fat diameter were determined as described by Pontiroli et al. A 3.0-MHz curved array transducer was placed along the xypho-umbilical line next to the umbilicus, and visceral and subcutaneous fat were measured after smooth exspiration. Visceral fat was measured from the internal surface of the Musculus rectus abdominalis to the near wall of the aorta. Subcutaneous fat was measured at the same position as the distance between the external surface of the muscle and the skin. The thickness of the muscle and skin were excluded. Measurements were performed in triplicates, mean values were calculated and used for analyses (21). Intraobserver coefficients of variation were 2.8 and 2.6%, for subcutaneous fat diameter and visceral fat diameter, respectively.
Histological liver scoring
Fifteen subjects agreed to the performance of a liver biopsy for research reasons. Histological scoring of steatosis of all biopsy material was performed in a standard manner according to histological criteria used by Dixon et al. Liver steatosis was graded from 0 (<5% of parenchyma involved) to 4 (>75% of lobular parenchyma involved) (22).
Values are presented as means ± s.d. Values that were not normally distributed were naturally log-transformed. Univariate correlation coefficients were calculated using the Pearson's method, except for degree of steatosis, which was calculated using the Spearman-Rho rank test. To reduce the number of variables within the multiple stepwise linear regression models, we included only parameters with significant univariate correlation coefficients. Testing for collinearity resulted in variance inflation factors <3 for all variables tested, hence, all variables with significant univariate correlation were used for regression analyses. The multiple stepwise linear regression model using FMD as dependent variable included age, sex, waist circumference, waist-to-hip ratio, visceral fat diameter, subcutaneous fat diameter, systolic blood pressure, and degree of steatosis as independent variables. The multiple stepwise linear regression model using IMT as dependent variable included age, sex, waist circumference, waist-to-hip ratio, visceral fat diameter, fasting glucose, and degree of steatosis as independent variables. First, a model containing only a constant is created. Then, the coefficient with the highest correlation coefficient is entered as the first variable. If this variable improves the model, it remains in the model and the next variable with the highest correlation is entered and stays in the model, if it again improves the ability to predict the outcome. Now a removal test is made of the least useful predictor to reassess whether any redundant predictor can be removed. All statistical analyses were performed using SPSS/PC statistical program (version 11.5 for Windows; SPSS, Chicago, IL). A value of P < 0.05 was considered statistically significant.
Clinical and laboratory characteristics
Participants of the study were severely obese (BMI: 35.7–53.9 kg/m2, fat mass: 33.5–81.7 kg, waist: 90–150 cm, hip: 105–177 cm, waist-to-hip ratio: 0.68–1.19). Mean values of total cholesterol, high-density lipoprotein–cholesterol, low-density lipoprotein–cholesterol, triglycerides, plasma glucose, and blood pressure were within the normal range. Mean adiponectin plasma levels were 10.1 ± 4.3 µg/ml, and the mean homeostasis model assessment for insulin resistance was 5.36 ± 4.27 (Table 1).
Table 1. Characteristics of the study population
Univariate correlation of FMD, IMT, and clinical parameters
Neither FMD nor carotid IMT showed a significant correlation with BMI and absolute fat mass. Blood pressure was correlated negatively with FMD, but showed no correlation with carotid IMT (Table 2). The strongest correlations could be observed between visceral fat diameter and FMD on the one hand (r = −0.577, P < 0.0001; Figure 1a) and between visceral fat diameter and IMT on the other (r = 0.303, P < 0.05; Figure 1b). In contrast to visceral fat, subcutaneous fat diameter showed a weak positive correlation with FMD (r = 0.279, P = 0.026). The degree of liver steatosis showed a strong negative correlation with the functional vascular parameter FMD (r = −0.523, P < 0.0001) (Figure 2a), but not with the structural vascular parameter carotid IMT (r = 0.22, not significant) (Figure 2b). Histological scoring of liver steatosis showed a positive correlation with IMT (r = 0.687, P = 0.04) and a negative correlation (r = −0.299, P = not significant) with FMD. We observed a significant association of the degree of liver steatosis with visceral fat diameter (r = 0.729, P < 0.0001) but not with subcutaneous fat diameter (r = −0.162, P = 0.20). Within the 15 subjects who underwent liver biopsy, sonographic grading of liver steatosis showed a strong correlation with the histological scoring of steatosis (r = 0.711, P = 0.003).
Table 2. Univariate correlation coefficients of FMD and IMT
Univariate correlation of FMD and IMT with laboratory parameters
Plasma adiponectin levels correlated negatively with visceral fat (r = −0.342, P = 0.008), degree of steatosis (r = −0.347, P = 0.007), insulin sensitivity (r = −0.368, P = 0.006), and blood pressure (r = −0.280, P = 0.03) and positively with high-density lipoprotein–cholesterol (r = 0.399, P = 0.002). However, no correlation of structural or functional vascular parameters with adiponectin or blood lipids could be detected (Table 2).
Independent predictors of FMD and IMT
In a multiple stepwise regression, we used FMD as the dependent variable and parameters with significant univariate correlation coefficients, namely age, sex, waist circumference, waist-to-hip ratio, visceral fat diameter, subcutaneous fat diameter, systolic blood pressure, and degree of steatosis as independent variables. Visceral fat diameter entered the model as the first independent parameter (standardized β −0.621; P < 0.0001), explaining 37% of the variance in FMD. In step 2, sex entered the model as independent predictor, increasing the predictive power to 43% (Table 3). In a similar calculation using IMT as the dependent variable, only sex and age were independent predictors of carotid IMT, explaining 35% of the variance in IMT (Table 3). The significant univariate correlation of the degree of liver steatosis with endothelial function lost significance after adjustment for visceral fat diameter. After inclusion of other clinical relevant cardiovascular risk factors, which displayed no univariate correlation with the vascular parameters (smoking history, low-density lipoprotein– and high-density lipoprotein–cholesterol, homeostasis model assessment for insulin resistance, and BMI) as independent variables in the regression model, age emerged as an independent predictor of FMD increasing the predictive power slightly (r2 = 0.48, Table 3). For IMT as the dependent variable, the addition of those independent variables decreased the predictive power of the model slightly (r2 = 0.33, data not shown).
Table 3. Stepwise linear regression for FMD and IMT
In this study, we investigated the influence of visceral fat accumulation, the degree of liver steatosis, and established risk factors on structural (carotid IMT) and functional (brachial artery FMD) vascular parameters in severely obese subjects.
In univariate analysis, FMD was correlated negatively with parameters of visceral obesity, but showed a weak positive correlation with subcutaneous fat diameter. No correlation with BMI, or absolute fat mass could be detected. Because all subjects were severely obese (BMI ranging from 35.7 to 53.9 kg/m2), it was not the absolute fat mass but the regional fat distribution that influenced the endothelial function in our population. To dissect independent risk factors, we performed multiple stepwise regression analyses. Visceral fat diameter entered the model as the first independent variable exclusively explaining 36% of the variation in endothelial function. In the next steps, only sex and age entered the model as further independent predictors, increasing the model's predictive power up to 43%. Inclusion of all clinical relevant risk factors resulted in an independent association of age with FMD at a third step increasing the predictive power of the model to 48%. Although the degree of liver steatosis was correlated with endothelial function in univariate analysis and after adjustment for age, sex, and blood pressure, no independent correlation could be detected after additional introduction of visceral fat diameter into the statistical model. Thus, multiple stepwise regression analysis supports the theory that the relationship between liver steatosis and vascular parameters reflects the adverse impact of visceral fat accumulation on steatosis and atherosclerosis, rather than a direct influence of liver steatosis on endothelial function.
We observed a strong correlation between visceral fat diameter and IMT. Consistent with our findings, several cross-sectional studies reported a positive correlation of IMT to visceral fat accumulation (23,24,25). However, sex and age were the only independent predictors of carotid IMT in multiple stepwise regression analysis. We explain this discrepancy between the functional parameter FMD and the structural parameter IMT by essentially normal, low mean IMT values (0.58 ± 0.14 mm) in our young study population.
Several previous publications described an independent association of visceral obesity with atherosclerosis and cardiovascular risk (26,27). Experimental studies in small populations with overweight or obese patients without cardiovascular disease corroborate our findings that visceral fat diameter is a principal determinant of vascular damage (4,5,11). A recent population-based study supports the notion that central obesity—measured by abdominal height—is a powerful predictor of elevated alanine aminotransferase and γ-glutamyl transferase levels, probably representing unrecognized cases of nonalcoholic fatty liver disease (28). Moreover, Villanova et al. demonstrated an inverse relationship of FMD to nonalcoholic fatty liver disease in 52 consecutive patients (29). The major new finding here is that in severely obese but otherwise healthy subjects, visceral fat diameter determined using ultrasound is the most powerful predictor of endothelial dysfunction, independent of traditional cardiovascular risk factors and liver steatosis. Visceral fat diameter was also correlated with increased IMT and the degree of liver steatosis. These findings suggest that visceral fat accumulation is a key mediator in the pathogenesis of the atherosclerotic process and the development of liver steatosis, most likely the effects mediated by multiple circulating adipocytokines produced by visceral fat cells (30,31,32). High-sensitivity C-reactive protein as an indicator of subclinical inflammation has been implicated as a marker or mediator of atherosclerosis. Although high-sensitivity C-reactive protein significantly correlated with liver steatosis and parameters of obesity, no significant correlation with IMT or FMD could be detected in this study population (Table 2). Furthermore, in particular adiponectin has been suggested to play a key role in metabolic syndrome, inflammation, endothelial function, cardiovascular disease, and nonalcoholic fatty liver disease (31,33,34,35,36,37,38). However, in our study, adiponectin did not contribute to the vascular impairment in these otherwise healthy subjects, although plasma adiponectin was correlated negatively with visceral obesity, blood pressure, insulin resistance, and liver steatosis. Previous data linking hypoadiponectinemia with endothelial dysfunction were generated either in patients with type 2 diabetes (35)—whereas we investigated healthy, young severely obese subjects—or endothelial function was determined using different methods (36), possibly explaining these discrepant findings. Moreover, our observation is in accordance with a recent report, failing to detect a significant correlation between plasma adiponectin levels with endothelial function (39). In another report, adiponectin could be associated with endothelial independent function only, but not with endothelial dependent function (40).
A negative correlation between visceral obesity and FMD has already been described previously. Previous studies usually used waist circumference and/or waist-to-hip ratio as an estimate of visceral obesity. However, measurement of these circumferences are only poorly standardized and strongly operator dependent, especially in severely obese patients. Moreover, waist circumference cannot distinguish between visceral and subcutaneous adipose tissue. Several studies used CT scans to distinguish between visceral and subcutaneous fat mass. CT scanning involves radiation and is more expensive compared to an ultrasound examination. We can show that a standard ultrasound equipment allows to distinguish between visceral and subcutaneous fat in obese subjects and that the visceral fat diameter as determined by ultrasound has the highest univariate correlation with FMD and IMT. Although bowel distention could interfere with measurement of visceral fat diameter, previous studies demonstrated a high correlation of visceral fat area determined by CT scan and sonographically determined visceral fat diameter (21). We conclude that ultrasound measurement is a quick and reliable method of low costs and without radiation exposure to identify patients with increased vascular risk, especially in obese populations.
This study has some limitations. This is an uncontrolled, cross-sectional study of only severely obese subjects, thus, its results cannot be extrapolated to the general population. Therefore, further studies are warranted to validate this new method before it can be used to determine visceral obesity in large-scale studies.
In summary, the absolute visceral fat diameter is a key mediator of early markers of the atherosclerotic process and liver steatosis, independent of traditional risk factors, BMI, adiponectin levels, and markers for insulin resistance. These findings, therefore, emphasize the role of avoiding or reducing visceral obesity to reduce the burden of cardiovascular disease and liver disease. Measurement of visceral fat using ultrasound is an easily feasible method to identify subjects with increased risk of atherosclerosis and liver disease.
Histological evaluation and scoring of the degree of steatosis was performed by Sylvia Stadlmann, MD from the Department of Pathology, University of Innsbruck. This work was supported by a grant from the OeNB 11172 and MFF Tirol to C. Ebenbichler and the Austrian Science Fund to H. Tilg (P17447).