Background
The distribution of body fat appears to be important in cardiovascular disease (CVD) risk [
1]. With recent advances in imaging techniques, attention has focussed on ectopic visceral fat distribution including the fat around the heart, known as epicardial adipose tissue (EAT). This interest is underpinned by population cohort studies that reported the association between EAT and adverse CV outcome [
2,
3]. EAT is a metabolically active visceral and perivascular fat depot, surrounding the myocardium and coronary arteries with no separating fascia. Under pathologic circumstances, EAT can act both locally or by paracrine secretion of mediators including adipokines, inflammatory cytokines, or reactive oxidative species that can potentially adversely affect the adjacent coronary vessels and myocardium [
4]. Local secretion of bioactive molecules by the EAT has been implicated in the formation of atherosclerotic plaques on the adjacent coronary arteries [
5‐
8]. The local effects of these bioactive molecules on the neighbouring myocardium have also been linked to atrial remodelling in atrial fibrillation (AF) [
9]. It is noteworthy that the effects of EAT on left ventricular remodelling such as left ventricular hypertrophy is less well studied.
Recent studies have shown that EAT is related to atherosclerotic disease in other vascular beds. In the population based Rotterdam Study, EAT was associated with multiple vessel beds and was independent of CV risk factors [
10]. This observation may imply endocrine systemic effects of EAT [
4]. EAT is metabolically active, secreting numerous substances associated with CVD such as TNFα, IL-6 and IL-1β [
11] that may suggest a systemic effect of EAT mediated via metabolic and inflammatory pathways. Exploring this link might help to provide a better understanding of the mechanism by which EAT is associated with CVD. Tentative supportive evidence of a systemic effect of EAT is provided by the association between EAT and arterial stiffness, a known predictor of adverse CV outcome [
12]. Previous studies have demonstrated an independent association between EAT and arterial stiffness in various groups of patients with CVD risk [
13,
14]. However, this relationship has not been studied in patients with cardio-metabolic disease and as yet the mechanism by which EAT and arterial stiffness are linked is poorly understood. The relationship between EAT and left ventricular hypertrophy in patients with cardio-metabolic disease has also not been studied.
The aim of this study was to assess the association between EAT, metabolic and inflammatory biomarkers, peripheral arterial stiffness (as measured by PWV) and left ventricular mass (LVM) in a cohort of patients with a spectrum of cardio-metabolic disease compared to controls to provide an understanding of the underlying local and systemic pathophysiological process by which EAT might lead to adverse CVD outcome.
Discussion
In this study we identified several novel findings. First, we have shown that patients with overt cardio-metabolic disease have significantly more EAT than those without. Second, we have shown that EAT is significantly associated with PWV, a marker of arterial stiffness that is linked to adverse CVD outcome. We did not, however, find any significant association between EAT and LVM. Finally, we have shown, to the best of our knowledge, for the first time, a clear association between EAT and IL-6, a biomarker associated with vascular inflammation, and that both EAT and IL-6 are significantly independently associated with PWV even when baseline clinical characteristics are accounted for and have incremental association PWV when added to clinical variables.
Arterial stiffness measured by PWV has been associated with an increased risk of CVD in a wide range of patients and has been found to be an independent predictor of major adverse CV events [
12]. Furthermore, it has also been shown to have incremental additional prognostic value when used in combination with traditional risk predictors such as blood pressure and cholesterol [
22]. Our finding that EAT is a significantly independently associated with arterial stiffness as measured by PWV is consistent with previous studies that showed that EAT, measured by echocardiography, was independently associated with endothelial function measured by PWV [
23]. In a study of 100 patients with biopsy-proven non-alcoholic fatty liver disease (NAFLD), a condition associated with inflammation and atherosclerosis, the authors found that epicardial adipose thickness was significantly higher NAFLD patients compared to controls and was a significant independent predictor of PWV [
24]. Our study extends the results of this study to patients with cardio-metabolic disease by assessing patients with CMR (the gold-standard). In our study, the availability of a comprehensive panel of biomarkers may help provide a mechanistic understanding for these observations.
Prior studies have demonstrated that inflammation, oxidative stress, and visceral fat volume predispose to CVD [
1,
25]. There have been several studies examining the link between pericardial adipose and CVD. In a study of over 6000 patients from the MESA cohort the authors found that pericardial adipose volume had a significant positive association with IL-6, and also had a borderline significance for CVD outcomes after adjustment for traditional risk factors [
26]. More specifically, pericardial adipose also predicted incident coronary heart disease in a subset of these patients [
27]. Finally, in a large cohort of over 1000 patients from the Framingham study, Tadros et al. reported an association between pericardial fat and inflammatory biomarkers including IL-6 [
28].
More recent work has suggested that EAT is associated with systemic inflammation. A meta-analysis of 40 gene expression studies on EAT suggested that EAT might well mediate its effects on CVD through IL6 [
29]. These observations lend support to the increasing recognition that EAT is a metabolically active endocrine organ and a source of pro-inflammatory adipokines that have significant impact on remote vascular tissues [
30]. The findings of our study lend support to this. In our study IL-6, an inflammation mediator, was strongly associated with both EAT and with PWV. It is noteworthy that IL-6 is capable of inducing vascular smooth muscle proliferation that can lead to endothelial dysfunction and arterial stiffness [
31]. In our study, we observed an incremental independent association of both EAT and IL-6 with PWV shown, for the first time, in our study. Indeed, further supporting our inflammatory theory is the association with CD40L, although this did not remain significant when included in the multivariable model. CD40L has also been associated with cardio-metabolic disease and is known to be released by adipose tissue [
32,
33]. Obviously, any indication that the observed relationship between EAT and PWV are due to the effects of IL-6 remains purely speculative and cannot be inferred directly from this observational study. Additionally, the fact that EAT was still independently associated with PWV even when IL-6 was added to the model might suggest that there may be other pathways which need to be further elucidated by which EAT is associated with increased PWV.
In our study we found that although EAT was associated with LVM in a univariate analysis, it was not significant in multivariable analysis. Prior studies using echocardiography have reported differing results regarding the relationship between EAT and LVM. Fosshaug et al. evaluated 60 heart failure patients and found that EAT was related to LVM [
34]. These results were replicated in a more recent study using PET-CT by Bakkum et al. in over 200 obese patients without significant CAD [
35]. Both of these studies contrast the study by Gates et al. in 119 healthy males suggesting that overall body fat was linked to LVM rather than EAT [
36]. Our study differs from all of these studies by using CMR, the gold-standard for assessment of LVMI. Additionally, our study is also, to the best of our knowledge, the first to include patients with CVD and T2D disease rather than just healthy volunteers. Because of this, there may however have been confounding of LVM based on medication use such as ACE inhibitors. It may however be the case that LVM is not influenced by local adipose but rather by overall adiposity and BMI. A large study from the Framingham cohort found that while pericardial adipose was associated with LVM in univariable analysis, this association was no longer significant once adjusted for body weight or visceral adipose [
37].
Limitations
As our findings come from a single-center study, these findings require confirmation within a larger multi-center setting. Due to the nature of recruitment of these patients in which the full biomarker panel was only available in a single centre, the number of patients for analysis is relatively small, although this is one of the largest papers to examine EAT, PWV and such a comprehensive panel of biomarkers using CMR. Additionally as a cross-sectional study causality cannot be proven and only associations can be made. Although we feel that our selection of 51 biomarkers is both pragmatic and broadly covers most pathophysiologcally important biomarkers, we cannot completely exclude the possibility that there may be other biomarkers which could be important, for example adiponectin. Finally, we only performed measurement of EAT in the 4-chamber view. Theoretically 3-dimensional measurement of EAT may have provided different results, however we wished to perform a protocol that was simple to utilise and not too time-consuming for practicing clinicians. However, measurement of adipose in the 4-chamber view has shown excellent correlation with 3-dimensional quantification in several other studies [
17‐
19]. Despite these limitations, we believe our findings are of potential importance and can be seen as hypothesis-generating.