Background
Obesity is a global epidemic. In 2016, 39% of adults worldwide were estimated to be overweight, and 13% obese [
1]. The global prevalence of obesity has almost tripled over the last 30 years and continues to increase [
1]. This trend is of substantial concern, as obesity is associated with an increase in cardiometabolic disorders and cardiovascular disease (CVD), the leading cause of death worldwide. Paradoxically, over this period, the burden of CVDs has been decreasing in industrialised countries [
2,
3]. However, the current increase in obesity could offset or reverse the downward trend of CVD mortality in spite of the decrease in the prevalence of other traditional CVD risk factors, such as smoking and hypertension [
3]. A recent analysis of US data has suggested that if body mass index (BMI) had not increased, life expectancy in 2011 at age 40 would be 0.9 years higher than was actually the case [
4], which in part would be due to a slowdown in the rate of decline of CVD mortality [
5].
Atherosclerosis is a major cause of CVD. Carotid atherosclerosis is easily and non-invasively detected using ultrasound. Carotid intima-media thickness (IMT) consistently predicts future CVD events, but carotid plaque outperforms IMT regarding its predictive ability of future CVD [
6]. Most previous studies on the association between obesity and carotid atherosclerosis in adults have been cross-sectional. Relatively few studies have attempted to look at this relationship prospectively. One population study investigated the association between lifetime BMI and IMT in late adulthood and showed that a reduction in the BMI category was associated with a decrease in IMT [
7]. Several others have investigated determinants of carotid plaque progression, including obesity as one of the potential factors [
8‐
11].
If there is evidence of obesity influencing progression, this is of direct relevance to clinical management in adulthood. Also, compared to general obesity (the exposure in most previous studies), abdominal obesity might play a more important role in progression because of its stronger association with cardiometabolic diseases [
12].
In an earlier investigation of cross-sectional data from the population-based Tromsø Study, we have shown that abdominal obesity was more closely associated with carotid plaque burden assessed by total plaque area (TPA) than general obesity [
13]. We also found that cardiometabolic risk factors such as hyperlipidemia, glucose intolerance, and hypertension mediated much of this association. The aim of the current analysis is to extend these investigations to determine whether the progression of carotid plaque burden over 7 years is related to different measures of obesity in the Tromsø Study, and how far any indication of such an association is mediated by the same set of cardiometabolic risk factors.
Discussion
In this seven-year follow-up of a population-based sample of women and men in late adulthood, we found that abdominal obesity was more strongly associated with the progression of carotid plaque burden than general obesity. New plaque formation among participants without plaque at baseline was, however, not associated with any adiposity measures. Furthermore, all significant associations observed were in part mediated by cardiometabolic risk factors.
Our results suggest that abdominal obesity in late adulthood might contribute to the progression of carotid plaque burden with larger effect estimates than general obesity, imposing an excess risk on the progression of atherosclerosis. This finding concurs with our previous findings from a cross-sectional analysis, although considerable overlap of CIs prevents us from drawing definitive conclusions [
13].
Several previous studies have investigated the effect of general obesity on the progression of carotid atherosclerosis [
8‐
10]. Herder et al. investigated the determinants of the progression of IMT and TPA after a 13-year follow-up; in this study, BMI at baseline did not predict progression of either [
8]. Similarly, van der Meer et al. showed that BMI was not associated with an increase in plaque numbers over an average follow-up of 6.5 years [
9]. Molino-Lova et al. showed that overweight/obesity, according to BMI category, was not associated with the new formation of plaque in 486 elderly participants without plaque at baseline over a three-year follow-up period [
10]. All of these findings agree with the relatively weak association between BMI and the progression of atherosclerosis in our analysis.
On the other hand, not much has been done to clarify the influence of abdominal obesity on the progression of carotid atherosclerosis. One prospective study (
n = 1894) with a four-year follow-up showed that an increase in WC was one of the determinants of the new formation of plaque among 462 participants without plaque at baseline after adjustment for age, sex, and follow-up time, while BMI was not [
11]. However, neither was a determinant of the progression of TPA. In the Rotterdam Study, the determinants of the progression of the number of plaques were analysed in 3409 participants after the 6.5-year follow-up. An increase in WHR was associated with an increase in the number of plaques after adjustment for traditional CVD risk factors. Again, BMI was not associated with increases in the number of plaques [
9]. Although a direct comparison of their findings with ours is difficult due to differences in statistical methods and adjustments, the potentially stronger effect of abdominal obesity compared to general obesity is consistent with our results.
The objectives of the studies mentioned above were to investigate the determinants of the progression of atherosclerosis [
8,
9,
11] or the association between main exposures other than obesity and atherosclerosis [
10]. When restricted to studies directly investigating the association between obesity and atherosclerosis, there is some evidence that IMT in adulthood may partly reflect childhood obesity [
19‐
24]. However, the question of whether or not the development or progression of carotid plaque in adult life is related to obesity in early life has not been investigated.
Most previous studies have used BMI to assess obesity. However, BMI does not provide information about fat distribution. Abdominal obesity, reflecting excess visceral adipose tissue, has a stronger association with inbsulin resistance and dyslipidemia than general obesity [
12]. Regarding whether abdominal obesity is more strongly associated with CVD than general obesity, evidence from observational studies is inconsistent. The Emerging Risk Factors Collaboration analysed 221,934 individuals from 58 cohorts and found that BMI, WC, and WHR were all associated with CVD risk, and the authors concluded that their effect sizes were similar [
25].
On the other hand, some studies suggest a stronger effect of abdominal obesity than general obesity. The INTERHEART Study, a large multi-centre case-control study with 12,461 myocardial infarction cases, suggested that increased WHR was more strongly associated with the occurrence of myocardial infarction than BMI [
26]. Furthermore, the recent INTERSTROKE Study showed that WHR had a stronger association with stroke than BMI had [
27].
Recently, two large Mendelian randomisation studies have shed light on the effect of abdominal obesity on CVD [
28,
29]. Dale et al. analysed data from 14 prospective studies with 66,842 coronary heart disease cases and 12,389 ischemic stroke cases, and compared associations of genetic risk scores for BMI and WHR adjusted for BMI with various cardiometabolic risks. The results showed that WHR might have a stronger effect on coronary heart disease and stroke than BMI. In particular, only WHR was associated with increased risk of ischemic stroke [
29]. Another study, with 111,986 participants from the UK Biobank, showed that a genetic disposition to higher WHR adjusted for BMI was associated with type 2 diabetes and coronary heart disease, supporting causal relationships [
28]. These findings emphasise the important role of abdominal obesity. Using BMI may lead to underestimation of the true risk of obesity for CVD.
In the present study, associations were more statistically significant in the analyses using TPA as an outcome than those using the number of plaques. It is expected that continuous plaque variables such as TPA and total plaque volume (TPV) can capture the small change in plaque over time more easily than a simple categorical plaque variable, requiring smaller sample sizes and potentially shorter follow-up time to detect significant changes. It has also been suggested that TPA is likely to be more sensitive to the progression of atherosclerosis than the more commonly studied outcome of IMT because plaque grows along the axis of the artery 2.4 times faster than it changes in thickness [
30]. One study, with 349 atherosclerotic patients, compared the predictive ability of future CVD events among TPV, TPA, and IMT after a five-year follow-up [
31]. Progression of TPV was significantly associated with CVD events after the adjustment for traditional CVD risk factors. Although the predictive ability of TPA was inferior to TPV, TPA performed better than IMT. While TPV is more sensitive to the progression with its three-dimensional information, TPA would be sensitive enough to detect the progression of plaque burden within a reasonable time frame, which makes TPA an attractive outcome in large population-based studies.
New plaque formation was not associated with any adiposity measures. Considering that our sample was in late adulthood, having no plaque at baseline might mean that participants are to some extent resistant to atherosclerotic changes for genetic or other reasons. This might contribute to the slower progression of carotid atherosclerosis and make it difficult to detect changes in this population. Another potential explanation is lack of power: by restricting the analysis to participants without plaque at baseline, the sample size was almost halved. Besides, like the number of plaques variable, the binary plaque variable provides less statistical information on the progression of plaque burden than a continuous plaque variable capturing size such as TPA.
All effect sizes of observed associations between obesity and the progression of plaque burden in the main analysis were substantially reduced after the adjustment for cardiometabolic risk factors in model 3, and no associations remained significant. This finding is supported by previous studies where strong determinants of the progression of carotid plaque burden included systolic blood pressure and total cholesterol [
8,
9]. Our finding validates that the pathway from obesity to carotid atherosclerosis is at least in part through cardiometabolic risk factors.
Recent studies with CVD mortality as an outcome also showed that the risk of obesity for atherosclerotic CVD is largely or fully mediated by these cardiometabolic risk factors [
32,
33]. These findings suggest that the strict control of metabolic risk factors might in part attenuate the risk of obesity on the progression of atherosclerosis. Interventions to bring long-term and sustained weight loss through lifestyle change have not been uniformly successful [
34,
35]. On the other hand, the effective treatment of cardiometabolic risk factors is established, and with strict control of this might reduce the indirect effect of obesity on atherosclerosis. Nevertheless, the treatment of obesity is vital to block any direct pathway to the progression of atherosclerosis and to control cardiometabolic risk factors better [
36]. Because our main objective was to investigate the association between obesity and plaque burden with and without the adjustment for mediators as a whole, this did not require partitioning of the contribution of a specific mediator to the overall effect.
Our major strength is a large sample size with a relatively long follow-up period, which allowed us a reasonable estimation of the association between the progression of carotid atherosclerosis and obesity. Furthermore, theTromsø Study is one of the few prospective studies with repeated carotid plaque measurements. The use of the quantitative plaque variable TPA is another advantage.
Our main limitation is a loss to follow-up. This might reduce the power of the study to detect associations between adiposity and the progression of carotid plaque burden. In addition, selection bias may be introduced. Thus, the association should be interpreted with caution, as should generalisability to the whole population. Another limitation is the possibility of residual confounding. For example, we did not include statin use in the model because statin use was not common at baseline. With respect to smoking, we only had data on this as a binary variable (current smoker: yes/no), which may have resulted in some residual confounding, as it is known that former smokers have a higher CVD risk than non-smokers [
37].
Furthermore, we did not consider subsequent changes in CVD risk factors and medications. Moreover, the abdominal obesity index that we used is a crude measure of abdominal adipose tissue. However, it is easily available in a real-world clinical setting. Further study using the reliable measurement of visceral adipose tissue and body composition overall would be informative. Several studies have shown that a decrease in lean mass or skeletal muscle mass was associated with CVD and atherosclerosis [
38‐
40]. In terms of WC, we measured WC at the level of the umbilicus, which tends to give higher WC values compared to other WC measurement sites [
41]. However, a systematic literature review concluded that the WC measurement site had little impact on the association of WC with CVD mortality, CVD events, or risk of diabetes [
42]. Finally, only the right carotid artery was assessed.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.