With the intensification of population aging, the incidence rate of atherosclerotic cardiovascular diseases (ASCVD) will significantly rise[33, 34]. Early detection and intervention are effective measures for reducing the incidence and mortality of cardiovascular diseases. While traditional clinical risk factors account for a significant portion of the attributable risk at the population level, their predictive ability for future ASCVD events in individuals is relatively limited[11, 35]. Therefore, more research focus on high-dimensional molecular signatures to better assess the cardiovascular disease risk and develop precise diagnostic and prognostic approaches[36]. We globally analyzed the dysregulated metabolome of CHD and elucidated the metabolic signature of the progression of this disease. These metabolic features can be developed further for CHD early diagnosis and acute myocardial infarction (AMI) prevention. Besides, diabetes-CHD exhibited specific metabolic dysregulations in comparison to any of single disease, which provide new hints and potential targets for concurrent diabetes and CHD.
The mostly reported differential metabolites in CHD include amino acids, short chain acylcarnitines, bile acids, TCA cycle-related metabolites, lyso-lipids, phospholipids, and sphingolipids[37, 38, 39, 40, 41, 42, 43]. What we identified the differentiated metabolites in CHD mainly consist of amino acids, xanthines, TAGs, and phospholipids (Fig. 1B, Fig. 2D). We compared the details with the work of Fan et al.[43], found that the most contribute lipids for group classification were lysoPC. For example, in the binary classification of normal and nonobstructive coronary atherosclerosis, they selected 10 biomarkers for diagnosis and eight of them were short chain lysoPC. Whereas the significantly changed metabolites in our classification of NCHD_0 and NCHD_0–50 contained five phospholipids (i.e., PC 38:2, PC 34:3, PE 40:5, PC 38:3, and PE 40:7) and two fatty acids (Suppl. Table 7). In their group classification of AMI and unstable angina, the metabolic biomarkers were lysoPC, amino acids, and fatty acids. In another cohort study of AMI, it was found that lysoPC could predict STEMI and NSTEMI[44]. In our result, the significantly changed metabolites in AMI were long chain TAGs, fatty acids, and β-oxidation related metabolites compared to NAMI (Suppl. Table 8). It is possible that the disparities between our two studies arose due to variations in study design, methodologies, and technical platforms. For example, the plasma metabolites were extracted with acetonitrile, detected by LC–quadrupole TOF MS in the previous study[43]. We performed this process with different ways. As described in method part, the plasma metabolites were extracted with methanol/methyl tert-butyl ether (MTBE)/water buffer, which can separate the hydrophilic polar compounds and hydrophobic lipids in one step. Further, both phases were detected with polar- and lipid-LC-MS platforms, separately. According to the octanol/water partition coefficient (X log P) range of common metabolites in plasma and the polarity of organic solvents used for metabolite extraction, acetonitrile is more suitable for relative polar compound extraction, such as part of amino acids and acylcarnitines, bile acids, fatty acids, lyso-/ and phospholipids, whereas methanol/MTBE have a wider coverage on TAGs but not on amino acids, acylcarnitines, and lyso-phospholipids[45]. This explains why they detected a higher amount of lysoPC and carnitines, while we observed a higher abundance of TAGs. Despite the vast number of studies conducted in the field of metabolomics, the diversity of metabolites, including variations in chemical structures and wide ranges of abundance, continues to pose significant challenges in achieving comprehensive coverage of the metabolome and ensuring reproducibility across different studies.
Atherosclerosis is initiated by the deposition of atherogenic lipoproteins[46], which are further aggravated by reactive oxygen species and inflammatory factors, leading to the formation and growth of atheroma (plaque). The blockage of arteries is a slow and intricate process that primarily leads to complications of ASCVD. Consequently, the blockage degree represents the disease progression and severity in some degree. As previous research stated that CHD was defined as ≥ 50% stenosis of the left main coronary artery or ≥ 70% of the left anterior descending coronary artery[47]. In our study, NAMI and NCHD-MAX0 cannot be classified well based on metabolic pattern although they have obvious difference on stenosis (Fig. S1). While AMI shown satisfied classification with NCHD_MAX0, NCHD_0–50, and even NAMI. This suggests that the stenosis of coronary artery visualized by the “gold standard” approach[48], coronary angiography, can be used for the diagnosis of suspected CHD but have a week association with the molecular biology of the disease. The explanations may result from the complex of the cause and effect of CHD and the complications. Matthew Nayor et al. summarized the molecular signatures of cardiovascular disease from multi-omics[36]. Circulating metabolome contains the information from biological system as well as external exposures[49], thereby enabling provide multi-dimensional clues for CHD risk assessment. This means that metabolic signatures process the potential to be developed as new technique for CHD precise diagnosis compared to traditional angiography approaches. Another possibility is the complications of CHD. As we exhibited in Fig. 6A-B, AMI and NCHD_MAX0 have better classification when complicated with diabetes. Xia et al. also found that the glucose utilization related metabolites, lysoPC and N-lactoyl-phenylalanin, skewed the metabolic pattern of AMI-diabetes relative to AMI-nondiabetes[32]. This information indicate that the molecular microenvironment of AMI-diabetes contributes to compromised cardiac function and worse outcomes[50, 51]. From the demographic analysis, diabetes mellitus (DM) history shown no statistic difference between four sub-groups of suspected CHD (Table 1), indicating there were no potential biases from this comorbidity. Hence, we further deciphered the metabolic interference and cross-regulations of DM and CHD, revealing that TAGs were the most prominent differentiated metabolites in this comorbid condition (Fig. 6). This observation aligns with previous studies that have demonstrated the association between specific TAG species and an increased risk of CVD and DM. For example, Stegemann et al. found that TAG of low carbon number and double-bond content have a strong and consistent associations with CVD from a population-based plasma lipidomic study[52]. Rhee et al. identified that a combination of TAGs was a metabolic signature of insulin resistance and improved the DM prediction[53]. For the CHD-DM complications, Dieren et al. investigated the associations of lipid concentrations and CVD risk in patients with diabetes and found that TAGs were associated with CVD independent of fasting[54]. All information indicate that TAGs could be promising molecules for CHD-DM prediction and surveillance. Further research should be undertaken to identify and validate specific TAG molecules across different studies, instead of focusing solely on bulk TAG analysis.
Although there is a discrepancy existed between angiographic stenosis severity and the presence of myocardial ischemia[55, 56], stenosis severity had significant association with the risk of clinical cardiovascular events[57]. In the previous studies, plaque characteristics and physiological stenosis severity are associated, and both were the predictors of per-vessel fractional flow reserve[58, 59]. Monitoring and assessing the severity of coronary artery stenosis is important in the management of CHD. For example, patients with 50–69% stenosis would be offered surgery to prevent severe cardiovascular events[22]. The developed approaches for determination the degree of stenosis mainly include invasive and less-invasive imaging tests, including intra-arterial angiography[60], ultrasound[61], magnetic resonance angiography[62], computed tomographic angiography[63], and contrast-enhanced MRA[64]. These techniques are recommended for patients with symptomatic or rapidly progressing severe aortic stenosis, and for asymptomatic patients with significant decline of the left ventricular ejection fraction < 50%[65]. However, there are increasing concerns that the irreversible myocardial damage has already happened before these clinical events presence[66]. Thus, this complex, multifaceted and systemic disease requires a more comprehensive assessment and monitoring approach[67]. In addition to echocardiography, measurement of blood biomarkers is one of the simple and most important methods for risk stratification of patients with or suspected of having vascular stenosis[68]. The available blood biomarkers for early reflecting the disease process and prognosis mainly include brain natriuretic peptide (BNP), N-terminal pro B-type natriuretic peptide (NT-proBNP), Troponin, Galectin-3, and other reported ones, such as cardiac myosin-binding protein (cMyC), lipoprotein (a) OxPL-apoB, CK-MB, CA125, hs-CRP, human epididymis protein (HE4)[68]. In this study, we identified 59 small molecular metabolites significantly correlated with the blockage progression (Suppl. Table 4). Some of these have been shown by us to be associated with clinical markers of CHD (Fig. 5). Among these metabolic markers, N-epsilon,N-epsilon,N-epsilon-trimethyllysine, the precursor of the atherogenic-related metabolite trimethylamine N-oxide (TMAO), was associated not only with clinical marker but also with CHD progression and stenosis degree (Fig. 2–3)[69]. This confirmed that N-epsilon,N-epsilon,N-epsilon-trimethyllysine could be a novel metabolic marker for ASCVD in conjunction with TMAO. These metabolic biomarkers can be further developed and validated for stenosis assessment and surveillance in a noninvasive and efficient way. Moreover, metabolic biomarkers demonstrated greater responsiveness to different stages of disease progression in comparison to conventional blood biomarkers. For instance, the clinical biomarkers associated with CHD showed a strong correlation primarily with AMI rather than NAMI, as depicted in Figure S4. This suggests that the clinical biomarkers of CHD may possess low sensitivity for diagnostic purposes. However, at the metabolic level (Fig. 2D), there were more accurate distinguishing patterns for both AMI and NAMI, particularly with the involvement of specific TAG species.
There is no single method or platform can resolve the entire complexity of plasma metabolome. In the conducted study, we employed separate extraction methods for hydrophilic and hydrophobic metabolites, which were subsequently analyzed using LC-MS polar, LC-MS lipid, and GC-MS platform, respectively. This approach ensured comprehensive coverage across the chemical and physical range of metabolites. However, there are still specific species of metabolites identified by other related studies that not yet annotated in our feature list. The chemical library and analytical methods should be improved furtherly. Furthermore, it is important to note that the cohort study was conducted on a limited group size, which may have certain limitations in terms of generalizability. Therefore, the potential biomarkers identified in this study should be subjected to validation in a larger population-based cohort study with well-established group settings involving multiple centers to assess the robustness and reliability of the identified biomarkers across a more diverse and representative population.