The handle http://hdl.handle.net/1887/39795 holds various files of this Leiden University dissertation
Author: Gast, Karin
Title: Insulin resistance and atherosclerosis : the role of visceral fat
Issue Date: 2016-06-01
Chapter 5
Individual contributions of visceral fat and total body fat to subclinical atherosclerosis: the NEO study
K.B. Gast M. den Heijer J.W.A. Smit R.L. Widya H.J. Lamb A. de Roos J.W. Jukema F.R. Rosendaal R. de Mutsert
for the NEO study group
Atherosclerosis 2015; 241: 547-554
aBstraCt
Background
Both overall and abdominal adiposity are established risk factors for cardiovascular disease.
Visceral adipose tissue (VAT) and total body fat (TBF) are strongly correlated and previous stud- ies did not make this distinction.
Objective
We aimed to distinguish individual contributions of TBF, VAT, and the ratio of VAT to abdominal subcutaneous adipose tissue (VAT/SAT) to subclinical atherosclerosis in men and women.
Methods
In this cross-sectional analysis of the Netherlands Epidemiology of Obesity (NEO) study, we assessed VAT and SAT with magnetic resonance imaging, TBF with bio-electrical impendence analysis, and carotid intima-media thickness (cIMT) with ultrasound. We performed linear regression analyses of standardized values of TBF, VAT, VAT/SAT with cIMT. We adjusted the models for confounding factors (age, sex, ethnicity, education, smoking, alcohol, physical activ- ity), and either for VAT or TBF.
Results
This analysis included 2,451 participants, 53% men with mean (SD) cIMT of 615 (91)μm. After adjustment for confounding factors, the difference in cIMT (95% CI) per SD in VAT was 14 (8,21) μm in men and 18 (13,24)μm in women. After adjustment for TBF, this attenuated to 5 (-3,13) μm in men and 13 (5,20)μm in women. In the full model, differences in cIMT (95% CI) per SD of TBF were 14 (6,22)μm in men and 8 (0,16)μm in women, and per SD of VAT/SAT were 7 (-1,15) μm and 9 (3,16)μm respectively.
Conclusions
In this population-based study, VAT contributed beyond overall adiposity to subclinical athero-
sclerosis, particularly in women. This implies a specific role of VAT in the early development of
atherosclerosis.
5
IntroduCtIon
Obesity is a well-established risk factor for cardiovascular disease [8]. Several observational studies have shown that waist circumference, a marker of abdominal obesity, was more strongly associated with cardiovascular mortality [29] and subclinical atherosclerosis [132, 196, 197]
than body mass index (BMI). In an earlier study, we showed that the association between insulin resistance and carotid intima-media thickness (cIMT), a marker of subclinical atherosclerosis, was largely explained by waist circumference, suggesting that abdominal adiposity is respon- sible for both insulin resistance and atherosclerosis [196].
It was hypothesized that the excess risk of abdominal adiposity may be explained by larger amounts of visceral adipose tissue (VAT) [38]. Waist circumference however, does not distin- guish between VAT and abdominal subcutaneous adipose tissue (SAT), which have different biological characteristics and functions. VAT secretes more pro-inflammatory adipokines [70]
and non-esterified fatty acids (NEFAs) [69] per gram adipose tissue than SAT and is therefore considered more atherogenic [38].
Previous studies in the general population that directly assessed VAT by computed tomog- raphy or magnetic resonance imaging (MRI) have shown that more VAT was associated with a larger cIMT [57, 63-68]. It must be noted that VAT is strongly correlated with total body fat [75].
Therefore, when studying specific effects of VAT, results should be adjusted for total body fat [76]. To our knowledge, only two of previous studies adjusted for total body fat and showed inconsistent results [64, 68]. In one study VAT contributed to cIMT above total body fat [64], whereas in the other study VAT did not contribute and total body fat was more important in the association with cIMT [68]. Therefore, the individual contributions of VAT and total body fat to the development of atherosclerosis remain unclear.
Within the abdomen, not only absolute amounts of VAT and SAT may be important in the development of atherosclerosis, but also their proportion. It is hypothesized that excess triglyc- erides are primarily stored in SAT and when this depot cannot sufficiently expand, triglycerides will be stored in the visceral area and eventually in non-adipose tissues such as the liver and skeletal muscle [38, 48]. According to this “lipid overflow” hypothesis [38, 48], a high relative amount of VAT may thereby indicate dysfunctional SAT. The ratio of VAT and abdominal SAT (VAT/SAT) was recently proposed as a marker reflecting the propensity to store triglycerides viscerally [176]. However, when considering this ratio, which can be the same in persons with different amounts of total body fat, it is important to adjust associations for total body fat as well. Furthermore, it is unclear whether the same ratio in persons with different amounts of body fat is associated with similar cardiometabolic risk. To our knowledge, no previous study investigated the association between the VAT/SAT ratio and cIMT in the general population.
Therefore, our aim was to study the individual contributions of total body fat, VAT and the
VAT/SAT ratio to subclinical atherosclerosis, as represented by cIMT, in men and women.
methods
Study design and study population
The Netherlands Epidemiology of Obesity (NEO) study is a population-based prospective co- hort study in individuals aged 45 to 65 years, with an oversampling of persons with a BMI of 27 kg/m
2or higher. The present study is a cross-sectional analysis of the baseline measurements of the participants included in the NEO study. Detailed information about the study design and data collection has been described elsewhere [77]. Between September 2008 and September 2012, men and women aged between 45 and 65 years with a self-reported BMI of 27 kg/m
2or higher living in the greater area of Leiden (in the West of The Netherlands) were eligible to participate in the NEO study. In addition, all inhabitants aged between 45 and 65 years from one municipality (Leiderdorp) were invited irrespective of their BMI, to allow for a reference distribution of BMI.
Prior to the NEO study visit, participants completed a questionnaire about demographic and clinical information. At the study site, participants completed a screening form, asking about anything that might create a health risk or interfere with MRI (most notably metallic devices, or claustrophobia). A body circumference of more than 1.70 m was an additional contraindica- tion for undergoing MRI. Of the participants who were eligible for MRI, approximately 40%
were randomly selected to undergo direct assessment of VAT and abdominal SAT. For the present analysis, we included participants with a measurement of VAT and abdominal SAT, and excluded participants who had missing data. The study was approved by the medical ethics committee of the Leiden University Medical Center (LUMC) and all participants gave written informed consent.
Data collection
On the questionnaire, participants reported ethnicity by self-identification in eight categories which we grouped into white (reference) and other. Tobacco smoking was reported in the three categories: current smoker, former smoker, or never smoker (reference). Highest level of educa- tion was reported in 10 categories according to the Dutch education system and participants with none, primary school or lower vocational education were categorized as poorly educated (reference) and the others as highly educated. Alcohol consumption was reported in 10 catego- ries, based on the average number of alcohol units consumed per day during the last year and this was further categorized in four categories: abstainers (reference), 0-1 unit/day, 2-4 units/
day, ≥5 units/day. Participants reported their physical activity and this was expressed in meta-
bolic equivalents hours per week. Participants reported their medical history of cardiovascular
disease. Pre-existing cardiovascular disease was defined as a prior event of myocardial infarc-
tion, angina, congestive heart failure, stroke, or peripheral vascular disease. At the study site,
height was measured without shoes with a vertically fixed, calibrated tape measure. Brachial
blood pressure was measured in a seated position on the right arm using a validated automatic
5
oscillometric device (OMRON, Model M10-IT, Omron Health Care Inc, IL, USA). Blood pressure was measured three times with five minutes rest between consecutive measurements. The mean systolic and diastolic blood pressure were calculated. Fasting blood samples were drawn from the antecubal vein after five minutes rest of the participant. Fasting plasma glucose concentra- tions, serum total cholesterol and high-density lipoprotein (HDL) cholesterol were determined in the central clinical chemistry laboratory of the LUMC by using standard methods [77].
Assessment of body fat
Body weight and total body fat (%) were measured by the Tanita bio impedance balance (TBF- 310, Tanita International Division, UK) without shoes and one kilogram was subtracted from the body weight. BMI was calculated by dividing the weight in kilograms by the height in meters squared. Waist circumference was measured mid-way between the border of the lower costal margin and the iliac crest.
Imaging was performed on an MR system operating at field strength of 1.5 Tesla (Philips Medi- cal Systems, Best, the Netherlands). VAT and abdominal SAT were quantified by a turbo spin echo imaging protocol. At the level of the fifth lumbar vertebra, three transverse images each with a slice thickness of 10 mm were obtained during a breath-hold [162]. Imaging parameters were:
TR=300 ms; TE=20 ms; flip angle=90°; slice thickness=10 mm, slice gap=2 mm. Mean VAT and abdominal SAT areas for each participant were quantified by converting the number of pixels to square cm for all three slides, and then averaging theses values, using in-house-developed software (MASS, Medis, Leiden, the Netherlands).
Assessment of carotid intima-media thickness
The cIMT was measured in the far wall of the left and right common carotid arteries (CCA’s), along a 15 mm long section, 10 mm proximal of the bifurcation, and in recumbent position. A 7.5-10 MHz linear-array transducer (Art.Lab version 2.1, Esaote, Maastricht, The Netherlands) in B-mode setting was used to visualize the distal CCA and an online wall track system was used to detect the lumen-intima and media-adventitia boundaries. CIMT was measured in three predefined angles per side (180,135 and 90 degrees for the right CCA and 180, 225 and 270 degrees for the left CCA) during six heartbeats. We calculated the mean cIMT for each partici- pant (referred to as cIMT) by averaging all these 36 cIMT measurements within each individual.
Statistical analyses
In the NEO study there is an oversampling of individuals with a BMI of 27 kg/m
2or higher. To
correctly represent associations in the general population [126], adjustments for the oversam-
pling of individuals with a BMI ≥ 27 kg/m
2were made. This was done by weighting individuals
towards the BMI distribution of participants from the Leiderdorp municipality [127], whose BMI
distribution was similar to the BMI distribution of the general Dutch population [128]. All results
were based on weighted analyses. Consequently, the results apply to a population-based study without oversampling of participants with a BMI of 27 kg/m
2or higher.
Baseline characteristics of the weighted study population were expressed as mean (SD), median (IQR) or as proportion (%). Pearson’s correlation coefficients between BMI, total body fat, waist circumference, VAT, SAT, and VAT/SAT ratio were calculated. We calculated Z-scores and standardized the values of total body fat, VAT, abdominal SAT, and VAT/SAT ratio to a mean of zero with a standard deviation of one.
With linear regression analysis we examined the individual associations of total body fat, VAT, abdominal SAT, and VAT/SAT ratio with the cIMT. We calculated regression coefficients with cor- responding 95% confidence intervals, which can be interpreted as the difference in cIMT (μm) per weighted standard deviation of total body fat, VAT, abdominal SAT, or VAT/SAT ratio. We constructed scatter plots between the standardized residuals and the body fat measures and observed that the assumption of linearity was met in our regression models. We calculated vari- ance inflation factors (VIFs) to check for multicollinearity in our regression models. VIF values were below 10 in all models and were considered appropriate. First, we adjusted the crude associations for age, ethnicity, education, tobacco smoking, alcohol consumption, and physical activity. In order to investigate to what extent associations of VAT and VAT/SAT ratio with cIMT could be explained by overall adiposity we additionally adjusted for total body fat. The linear regression analysis with SAT as determinant was additionally adjusted for VAT. The associations of all body fat measures with cIMT were additionally adjusted for diastolic and systolic blood pressure, glucose, total cholesterol, and HDL-cholesterol concentrations, in order to investigate to what extent the associations were mediated by these obesity-related risk factors. We tested for interaction between the body fat measures and sex, in addition to interaction between the body fat measures by including product terms in the models, and performed all analyses for the total study population and for men and women separately.
Second, to further distinguish effects of total body fat and VAT, we stratified the study popu- lation into nine different groups on the basis of tertiles of total body fat and VAT. We calculated differences in cIMT between these nine groups of participants with linear regression, using participants within the lowest tertiles of total body fat and VAT as reference. Adjustments were made for age, ethnicity, education, tobacco smoking, alcohol consumption, and physical activ- ity. Similarly, we examined joint associations of total body fat and VAT/SAT ratio.
All analyses were repeated excluding participants with pre-existing cardiovascular disease
at the baseline study visit. Analyses were performed with STATA Statistical Software (Statacorp,
College Station, Texas, USA), version 12.0.
5
resuLts
Baseline characteristics
In total, 6,673 participants were included in the NEO study, of which 2,581 had undergone MRI of the abdomen and in 2,570 participants complete imaging and quantification of VAT and abdominal SAT was achieved. After consecutive exclusion of participants who had missing data for total body fat (n=4), cIMT (n=23), ethnicity (n=3), education (n=25), tobacco smoking (n=2), alcohol consumption (n=7), or physical activity (n=55), 2,451 participants were included in the present analysis. Characteristics of the 2,451 included participants are presented in table 1.
Men had less total body fat and less abdominal SAT, but a larger waist circumference, more VAT, a higher VAT/SAT ratio, and a larger cIMT as compared with women (table 1).
Correlations between measures of overall and abdominal adiposity
Measures of overall adiposity (BMI, total body fat) and abdominal adiposity (waist circumfer- ence) showed a strong, positive correlation in both men and women. VAT was positively correlated with all other measures of overall and abdominal adiposity, but in men somewhat weaker with SAT. In both men and women, the VAT/SAT ratio was weakly correlated with BMI, total body fat, waist circumference, and SAT, and strongly with VAT (table 2).
Associations of total body fat, VAT, SAT and VAT/SAT ratio with cIMT
The results of the linear regression analyses per standard deviation of total body fat, VAT, abdominal SAT, and VAT/SAT ratio are shown in table 3 for the total study population, and in Figure 1 for men and women seperately. One standard deviation of VAT (men: 58 cm
2, women:
43 cm
2) was associated with 19 µm cIMT (95% CI: 12, 25) in men and 22 µm cIMT (95% CI: 17, 28) in women. These associations attenuated after adjustment for age, ethnicity, tobacco smoking, alcohol consumption, and physical activity (men:14 µm cIMT, 95% CI: 8, 21, women: 18 µm cIMT, 95% CI: 13, 24). After additional adjustment for total body fat, VAT remained stronger associated with cIMT in women (13 µm cIMT, 95% CI: 5, 20) than in men (5 µm cIMT, 95% CI: -3, 13). After adjustment for diastolic and systolic blood pressure, glucose, total cholesterol, and HDL-cholesterol concentrations, visceral fat was no longer associated with cIMT in men (0 µm cIMT, 95% CI: -8, 8), and only weakly in women (7 µm cIMT, 95% CI: -2, 15).
After adjustment for confounding factors, in men one standard deviation of VAT/SAT ratio (0.28) was associated with 9 µm cIMT (95% CI: 1, 16), whereas in women one standard deviation of VAT/SAT ratio (0.13) was associated with 13 µm cIMT (95% CI: 6, 19). After additional adjust- ment for total body fat the differences in cIMT per SD VAT/SAT ratio were 7 µm (95% CI: -1, 15) in men and 9 µm (95% CI: 3, 16) in women. Additional adjustment for diastolic and systolic blood pressure, glucose, total cholesterol, and HDL-cholesterol concentrations, attenuated the association of VAT/SAT ratio with cIMT in both men and women (men: 3 µm cIMT, 95% CI:
-4, 1, women: 3 µm cIMT, 95% CI: -4, 10). After adjustment for VAT, one standard deviation of
total body fat in men (5.5%) was associated with 14 µm cIMT (95% CI: 6, 22). In women, one standard deviation of total body fat (6.5%) was associated with 8 µm cIMT (95% CI: 0, 16). Addi- tional adjustment for diastolic and systolic blood pressure, glucose, total cholesterol, and HDL- cholesterol concentrations did not materially changed associations of total body fat and SAT
table 1. Baseline characteristics of 2,451 participants in the NEO study aged between 45 and 65 years
Characteristics All
(n=2,451)
Men (n=1,289)
Women (n=1,162)
Age (years) 56 (50-61) 57 (50-61) 55 (51-60)
Ethnicity (% whites) 96 96 95
Educational level (% low
a) 19 18 20
Tobacco smoking (%)
Current 14 15 13
Former 46 48 44
Alcohol consumption (%)
Abstainer 9 5 13
0-1 unit per day 61 53 68
2-4 units per day 26 35 18
≥ 5 units per day 4 7 1
Physical activity (MET-hours/week) 38 (33) 40 (37) 37 (29)
Medical history of CVD
b(%) 5 5 5
Systolic blood pressure (mmHg) 131 (17) 135 (15) 127 (18)
Diastolic blood pressure (mmHg) 83 (10) 85 (10) 82 (10)
Fasting concentrations (mmol/L)
Glucose 5.4 (1.0) 5.6 (1.2) 5.3 (0.8)
Total cholesterol 5.72 (1.05) 5.61 (1.00) 5.83 (1.08)
HDL-cholesterol 1.57 (0.46) 1.34 (0.34) 1.78 (0.45)
BMI (kg/m
2) 26.0 (4.0) 26.7 (3.4) 25.3 (4.3)
Total body fat (%) 31 (8) 25 (6) 36 (6)
Waist circumference (cm) 91 (13) 98 (10) 86 (12)
VAT (cm
2) 90 (56) 116 (58) 67 (43)
SAT (cm
2) 236 (97) 210 (80) 259 (104)
VAT/SAT 0.41(0.27) 0.57 (0.28) 0.26 (0.13)
cIMT (μm) 615 (91) 626 (94) 605 (87)
Results were based on weighted analyses. Data are shown as mean (SD), median (IQR) or percentage.
a
Low education: none, primary school or lower vocational education as highest level of education.
b
CVD, cardiovascular disease: defined as myocardial infarction, angina, congestive heart failure, stroke and peripheral vascular disease.
BMI, body mass index; cIMT, carotid intima-media thickness; CVD, cardiovascular disease; HDL, high density
lipoprotein; MET, metabolic equivalent; NEO, Netherlands Epidemiology of Obesity; SAT, abdominal subcu-
taneous adipose tissue; VAT, visceral adipose tissue; VAT/SAT, ratio of visceral adipose tissue and abdominal
subcutaneous adipose tissue
5
with cIMT in the total population and in men and women separately (data not shown). There was a significant interaction between sex and VAT (p=0.024) and between sex and VAT/SAT ratio (p=0.004).The interaction terms of total body fat and SAT with sex were non-significant (p-values > 0.05).
table 2. Pearson correlation coefficients between measures of overall and abdominal adiposity in 1,289 male and 1,162 female participants in the NEO study aged between 45 and 65 years
Pearson correlation coefficients (r)
BMI TBF WC VAT SAT VAT/SAT
BMI 0.86 0.87 0.73 0.88 0.23 W omen
TBF 0.88 0.86 0.70 0.84 0.25
WC 0.88 0.85 0.76 0.83 0.33
VAT 0.62 0.65 0.68 0.62 0.74
SAT 0.81 0.78 0.81 0.46 0.02
VAT/SAT 0.09 0.14 0.15 0.70 -0.20
Men
Results were based on weighted analyses.
The Pearson correlation coefficients in the total study population were: 0.50 between BMI and TBF, 0.84 between BMI and WC, 0.65 between BMI and VAT, 0.77 between BMI and SAT, 0.20 between BMI and VAT/
SAT, 0.22 between TBF and WC, 0.13 between TBF and VAT, 0.74 between TBF and SAT, -0.31 between TBF and VAT/SAT, 0.76 between WC and VAT, 0.58 between WC and SAT, 0.42 between WC and VAT/SAT, 0.35 between VAT and SAT, 0.76 between VAT and VAT/SAT, and -0.23 between SAT and VAT/SAT.
BMI, body mass index; NEO, Netherlands Epidemiology of Obesity; SAT, abdominal subcutaneous adipose tissue; TBF, total body fat; VAT, visceral adipose tissue; VAT/SAT, ratio of visceral adipose tissue and abdomi- nal subcutaneous adipose tissue; WC, waist circumference
table 3. Associations of total body fat, VAT, abdominal SAT, and VAT/SAT ratio with cIMT in 2,451 partici- pants in the NEO study aged between 45 and 65 years
Difference in cIMT (µm)
a(95% CI) per SD of:
TBF (SD=8.3%) VAT (SD=56 cm
2) SAT (SD=97 cm
2) VAT/SAT (SD=0.27)
Crude 7 (2, 12) 23 (18, 27) 10 (5, 14) 19 (13, 24)
Adjusted 23 (17, 30) 18 (12, 23) 14 (10, 18) 12 (5, 18)
Adjusted + VAT 16 (8, 24) 9 (3, 16) 8 (3, 12) 9 (3, 16)
Results were based on weighted analyses. In the adjusted model, results were adjusted for age, sex, ethnic- ity, education, tobacco smoking, alcohol consumption, and physical activity.
a
Beta coefficients (95% CI) from linear regression per weighted SD in total body fat, VAT, abdominal SAT, and VAT/SAT, for all 2,451 participants.
cIMT, carotid intima-media thickness; SAT, abdominal subcutaneous adipose tissue; SD, standard devia-
tion; TBF, total body fat; VAT, visceral adipose tissue; VAT/SAT, ratio of visceral adipose tissue and abdominal
subcutaneous adipose tissue
Joint associations of total body fat, VAT, and VAT/SAT ratio with cIMT
The difference in cIMT between nine groups stratified in tertiles of total body fat and VAT are shown in Figure 2a to 2c, after adjustment for age, ethnicity, education, tobacco smoking, alcohol consumption, and physical activity. Although some groups were small and therefore confidence intervals were wide, VAT also appeared to contribute to cIMT beyond total body fat in the total study population and in men with their total body fat in the second tertile, as the difference in cIMT increased across tertiles of VAT (Figures 2a and 2b). In women, VAT was clearly most important in the association with cIMT, and total body fat influenced cIMT only in the lowest tertile of VAT (Figure 2c). Within the second and third tertile of VAT, total body fat
Figure 1. Associations between total body fat, VAT, abdominal SAT, VAT/SAT ratio and cIMT (μm) in 1,289 men (closed circles) and 1,162 women (open circles) aged between 45 and 65 years
Results were based on weighted analyses. In the adjusted model, results were adjusted for age, ethnicity, education, tobacco smoking, alcohol consumption, and physical activity.
a
Beta coefficients (95% CI) from linear regression per weighted SD in total body fat (men: 5.5%, women:
6.5%), VAT (men: 58 cm
2, women: 43 cm
2), abdominal SAT (men: 80 cm
2, women: 104 cm
2), and VAT/SAT (men: 0.28, women: 0.13), for 1,289 men (closed circles) and 1,162 women (open circles) separately.
cIMT, carotid intima-media thickness; SAT, abdominal subcutaneous adipose tissue; SD, standard devia-
tion; TBF, total body fat; VAT, visceral adipose tissue; VAT/SAT, ratio of visceral adipose tissue and abdominal
subcutaneous adipose tissue
5
did not add to the difference in cIMT. While on average men had more VAT and a higher VAT/
SAT ratio than women, the figures with VAT/SAT ratio were similar in men and women (Figures 2e and 2f). Both total body fat and the VAT/SAT ratio contributed to cIMT, to a similar extent in men and women (Figures 2d, 2e, and 2f). The interaction term of TBF and VAT was signifi- cant in women (p<0.001). Although all other product terms were not statistically different we consistently presented all results for the study population and separately for men and women throughout the manuscript. Exact differences in cIMT with 95% confidence intervals are shown in appendix tables 1 and 2.
After exclusion of participants with pre-existing cardiovascular disease results were similar (data not shown).
dIsCussIon
In this study, we observed that associations between visceral fat and cIMT attenuated after adjustment for total body fat and remained most strongly associated with cIMT in women. In contrast to our expectations, total body fat seemed at first sight more important than VAT in men. However, after stratification in tertiles of VAT and total body fat, VAT also appeared to
A B C
D E F