Background
Atherosclerosis is currently the leading cause of death and disability in developing countries and is mainly expressed in the form of coronary artery disease (CAD), which is associated with high morbidity and mortality [
1,
2].
Although systemic cardiovascular risk factors for atherosclerosis (hyperlipidemia, diabetes mellitus, smoking and hypertension) have been identified [
3‐
9], a wide series of researches suggest that the local microenvironment, that comprehends arterial mechanics, matrix remodelling and lipid deposition, plays a key role in regulating the susceptibility to plaque development and progression, regulating the function of endothelial cells. Moreover, these microenvironmental stimuli are capable of modulate other aspects of the microenvironment through collective adaptation [
10]. All these events induce a series of biochemical reactions that generate a wide and complex interplay between metabolites absorbed from or released in blood [
11], thus modifying also the vascular microenvironment.
In this context, the omics approach could represent an innovative method for comprehensively investigating the molecular basis of CAD pathogenesis.
During the past decade, both animal and human studies have enabled the rapid development of metabolomics. By combining targeted and non-targeted approaches, metabolomic analysis has identified small-molecule metabolite profiles of diverse cardiovascular risk factors and diseases [
11‐
13].
On this basis, our hypothesis was that (i) despite similar cardiovascular risk factors, coronary different microenvironments, generated by interaction between inherited and acquired factors, could generate distinct manifestations of CAD and (ii) these microenvironmental differences might be investigated by metabolomics analysis.
Discussion
We conducted a pathophysiological study to identify possible metabolic differences in the microenvironment of coronary arteries, normal or suffering from both stenotic and microvascular disease, using metabolomics, an innovative and highly sensitive method.
For this purpose, we conducted 1H-NMR experiments on blood samples obtained from coronary artery of all enrolled case-group and control patients. We obtained the following results: (1) the OPLS-DA model used here separated the samples into 3 groups, which indicated that, given comparable cardiovascular risk factors, the controls and the Micro and SD patients were characterized by markedly different metabolic profiles. (2) ROC curves, created to evaluate the importance of the relative concentration of detected metabolites, confirmed the identification of the 3 distinct biochemical fingerprints, one belonging to healthy coronary arteries, the other 2 underlying the stenotic and microvascular manifestations of CAD. (3) Pairwise comparisons used to identify the metabolites responsible for the separation among groups revealed that relative to the control group, the Micro group showed higher levels of 2-hydroxybutyrate, alanine, leucine, isoleucine, and N-acetyl groups and lower levels creatine/phosphocreatine, creatinine, and glucose, whereas the SD group showed higher levels of 3-hydroxybutyrate and acetate and lower levels of 2-hydroxybutirate; furthermore, the levels of 2-hydroxybutyrate, alanine, leucine, and N-acetyl groups were higher and those of 3-hydroxybutyrate and acetate lower in the Micro group than in the SD group.
Several studies on the pathophysiological mechanisms involved in atherogenesis have shown that alteration in endothelial function is a key event in the development of this pathology and that it plays a major role in plaque progression and complication [
18‐
21]. By contrast, with regard to the possibility of using metabolomics for detecting the presence of CAD and defining its severity, previous studies have obtained divergent results, first supporting [
22] and then refuting [
23] the likelihood of this approach being employed effectively.
However, the prevalence of an interest in diagnostics diverted the attention from the endothelium as the substrate of atherogenesis.
The investigation of endothelial metabolic responses is critical: the recognized risk factors for atherosclerosis, although typically systemic, necessarily require an alteration of endothelial properties to change from a potential threat to a pathological lesion [
10,
18‐
20].
As stated in the Introduction section, the action of neural, hormonal, paracrine or cytokine signaling directly or indirectly modifies cellular metabolic pathways [
11,
24]. Furthermore, it was shown that even the endothelial mechanoreceptors activity can alter circulating metabolites and may be revealed by metabolomics [
25].
Given all the aforementioned considerations, we designed this study to assess the coronary microenvironment and its relationship with the pattern and degree of arterial anatomical and functional impairment in patients undergoing coronary angiography. Patients with stable angina (confirmed by positive stress test) formed the 2 case groups (Micro and SD), whereas patients with non-ischemic diseases (e.g., valvular heart disease, cardiomyopathies) were enrolled as controls. According to our OPLS-DA model, the coronary artery blood of CAD patients with stenosing disease appeared different, with high specificity, from that of patients with microvascular disease. Furthermore, the metabolomic analysis differentiated both case groups from the control group of non-ischemic patients; this discrimination was further confirmed using ROC curves, which validated the OPLS-DA models in terms of sensibility and sensitivity [
17].
The interaction between “macro” risk factors (e.g. diabetes, hypertension, dyslipidaemia) and intrinsic properties (e.g. genetic, epigenetic) of the local endothelial layer may be critical for atherogenesis [
10,
26]. This interaction may produce a microenvironment suitable for the phenotypic translation of a specific athero-susceptibility, as previously suggested by a wide range of studies [
10].
The metabolic profile provides a “functional” view of a tissue, intended as the ultimate resultant of its genes, RNAs, proteins and environmental factors (e.g. nutrition). In the case of the endothelium, it may provide evidence of a perturbated blood flow (e.g. the presence of a plaque), reflecting the endpoints of the biological processes of that tissue and the involved pathways [
11,
12]. Accordingly, despite the comparable systemic risk factors, all the groups in our study showed different coronary angiography findings and a distinctive metabolic fingerprint, as confirmed by the statistical parameters reported in Table
4. The control group was distinct from both case groups (PC2, Fig.
3), whereas the latter seem to be differentiated along the PC1.
Our results suggest that the normal coronary microenvironment is considerably different from that of coronary arteries affected by both stenotic and microvascular disease. In addition, coronaries with atherosclerosis exhibit different metabolic fingerprints from each other, suggesting the involvement of distinctive pathways in these 2 atherogenic manifestations.
The different expressions or developmental stages of the same disease explained by metabolomics suggest that endothelial dysfunction, though enabled by the classic cardiovascular risk factors, is modulated by other determinants, such as genetic, nutritional and behavioral, and causes various local micro-environments, witnessing an individual sensitivity to atherosclerosis. The cluster of identified metabolites, specific for each group, as confirmed by the ROC curves, suggests the existence of multiple pathways involved in atherosclerotic development and responsible for the differences we encountered.
Coronary blood from the SD group contained higher levels of acetoacetate and 3-hydroxybutyrate but lower levels of 2-hydroxybutyrate than did the blood from both Micro and Control groups. Acetoacetate and 3-hydroxybutyrate, but not 2-hydroxybutyrate, have been associated with atherogenesis in diabetic patients [
27,
28]. Specifically, Kanikarla-Marie and Jain showed that these 2 ketone bodies upregulate NADPH oxidase and thus cause, at the endothelial level, an increase in oxidative stress, ICAM-1 expression, and monocyte adhesion [
28], all of which are involved in atherosclerosis development.
Relative to controls, Micro patients showed elevated levels of alanine, leucine/isoleucine, 2-hydroxybutyrate, and
N-acetyl groups, but lower levels of creatine and creatinine. Creatine exerts anti-inflammatory effects at the endothelial level by inhibiting ICAM-1 and E-selectin expression and thereby reducing neutrophil margination. Furthermore, creatine can reduce endothelial permeability [
29], and Sestili et al. have highlighted the cytoprotective effect of creatine as a radical scavenger against reactive oxygen (in particular hydroxyl radical) and nitrogen species [
30].
Recently, alanine was found to be independently associated with major adverse cardiovascular events, [
31] but reduced alanine levels were also reported in both apolipoprotein E-deficient mice (apolipoprotein E(−/−) mice) [
32] and the plasma of patients with stable carotid atherosclerosis [
33]. This dual behavior of alanine (association with cardiovascular events, but reduction in patients with atherosclerosis) was also observed in our study population: alanine levels were higher in Micro patients than in both controls and SD patients, which suggests a risk profile wherein ischemic events are not associated with atherosclerotic plaques. A similar functional behavior was also observed for leucine/isoleucine, which reduced the serum levels of SOD and GPx in hypercholesterolemic rats [
34], but also positively correlated with acetoacetate and 3-hydroxybutyrate in distinguishing atherosclerotic patients from healthy controls [
33].
Authors’ contributions
MD, work conception and design, data analysis and interpretation, paper drafting and final approval; CP, data analysis and interpretation, paper drafting; CCD, work design, data acquisition, analysis and interpretation, paper drafting; DC, work conception and design, data analysis and interpretation; EL, FA, data acquisition, analysis and interpretation; GdC, data acquisition and interpretation, paper revising; MC, data acquisition and interpretation, paper revising; GL, RP, data acquisition and interpretation, paper revising; LA, data analysis and interpretation, paper drafting, revising, final approval; GM, work conception and design, data analysis and interpretation, paper drafting, revising and final approval. All authors read and approved the final manuscript.