The Emerging Role of Metabolomics in the Diagnosis and Prognosis of Cardiovascular Disease

Overview of Cellular Metabolism

Substrate metabolism is an essential component of cellular heath and survival. Both anabolic and catabolic processes are necessary to support the numerous cellular events that contribute to cell, organ, and organism survival. From a simplistic perspective, when cells in the body generate energy in the form of adenosine triphosphate (ATP), they catabolize the various energy sources available, either from endogenous energy stores (primarily glycogen or triacylglycerols), or from exogenous substrates circulating in the blood (e.g., carbohydrates, fatty acids, amino acids, and ketone bodies, among others). By contrast, when cells need to build materials (e.g., phospholipids for membranes, proteins for growth, fatty acids for de novo lipogenesis, and so on) or perform many cellular processes (e.g., growth, ionic homeostasis, signal transduction, contraction, among others), they consume ATP, releasing the energy needed to support these functions. For the vast majority of cells in our body, carbohydrates (primarily in the form of glucose and lactate) and fatty acids represent the most common energy substrates metabolized by our cells to produce ATP.

Of the many substrates that can be used to generate energy, glucose and fatty acids are the major contributors to overall ATP production. For the former, following transport from the circulation into the cell, glucose can undergo anaerobic glycolysis to produce pyruvate and small amounts of ATP (2 ATP per glucose molecule) (Figure 1). In some cell types, if the energy need is low, the majority of this pyruvate is converted into lactate, which can exit the cell into the circulation (Figure 1). However, if the cellular energy demands are high and oxygen is present, pyruvate derived from glucose and/or lactate can enter the mitochondria, where it is converted to acetyl coenzyme A (acetyl-CoA) by pyruvate dehydrogenase (PDH). Acetyl-CoA is the common intermediate that links oxidative metabolism of all nutrient energy sources, and this acetyl-CoA is utilized by the tricarboxylic acid (TCA) cycle (also known as the Krebs cycle) to produce reducing equivalents (e.g., nicotinamide adenine dinucleotide and flavin adenine dinucleotide), which act as electron donors to drive the proton motive force that fuels ATP synthesis (Figure 1). This latter process of glucose breakdown from pyruvate to eventual ATP production is termed glucose oxidation, and ultimately generates a significantly greater amount of ATP than glycolysis (31 ATP from glucose oxidation vs. 2 ATP from glycolysis). During cellular fatty acid catabolism, fatty acids are either transported into and/or passively enter the cell, converted into fatty acyl-CoA esters, and then converted into a fatty acylcarnitine, which permits the fatty acid to traverse the mitochondrial membrane. There, it is reconverted back into a fatty acyl-CoA ester for subsequent mitochondrial β-oxidation (Figures 1 and 2) (1). The mitochondrial β-oxidation enzymatic machinery proceeds to repeatedly remove acetyl-CoA from the fatty acyl-CoA ester until it has been completely oxidized. Hence, complete oxidation of a fatty acid molecule, such as oleate or palmitate, produces more acetyl-CoA and reducing equivalents, and thereby a much larger amount of ATP than the complete oxidation of a glucose molecule (104 ATP from palmitate vs. 31 ATP from glucose oxidation).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Cellular Metabolism and Implications of Metabolomic Profiling

The primary substrates used by our cells to produce energy (ATP) include carbohydrates, fatty acids, ketone bodies, and amino acids. In the absence of oxygen, ATP can be generated from glycolysis (step 1) as glucose is converted into pyruvate and then lactate. In the presence of oxygen, pyruvate can enter the mitochondria and be oxidized via PDH to produce acetyl-CoA for the TCA cycle. This process is referred to as glucose oxidation (step 2). Similarly, fatty acids are transported into the cell, converted to fatty acyl-CoA, and then enter the mitochondria via the CPT1/CPT2 system. Following this, the fatty acyl-CoA ester enters the β-oxidation pathway for further breakdown to produce acetyl-CoA (step 3). BCAAs can also be oxidized to produce acetyl-CoA, but must first be converted into their respective BCKA derivatives, which are subsequently oxidized by BCKD as they enter the mitochondria, undergoing a series of reactions that lead to the formation of acetyl-CoA (step 4). Ketone bodies such as β-hydroxybutyrate can also be oxidized (via BDH) in the mitochondria and result in the formation of acetyl-CoA (step 5). Thus, acetyl-CoA represents the common metabolite linking oxidative energy metabolism of the main substrates identified herein, and this acetyl-CoA feeds into the TCA cycle and results in the formation of reducing equivalents NADH and FADH₂, which fuel ATP production. During cardiometabolic disease progression, metabolic pathways are often perturbed and lead to the accumulation or loss of various metabolites that can be detected in the circulation via metabolomic profiling. These key metabolites capable of spilling over into the circulation, and which can reflect a “snapshot” of cellular metabolism, are indicated in blue text. ATP = adenosine triphosphate; BCAA = branched-chain amino acid; BCKA = branched-chain α-keto-acid; BDH = β-hydroxybutyrate dehydrogenase; CoA = coenzyme A; CPT = carnitine palmitoyltransferase; FADH₂ = flavin adenine dinucleotide; L-C = long-chain; NADH = nicotinamide adenine dinucleotide; PDH = pyruvate dehydrogenase; TCA = tricarboxylic acid.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Use of Acylcarnitine Profiling to Infer Changes in Fatty Acid Metabolism

During fatty acid oxidation, fatty acyl-CoA esters are converted into fatty acylcarnitine derivatives (via CPT-1), which allows fatty acids to enter the mitochondria where they are reconverted back into fatty acyl-CoA (via CPT-2) for subsequent mitochondrial β-oxidation. General defects in the enzymes of the β-oxidative machinery that inhibit the oxidation of L-C fatty acids will increase the accumulation of mitochondrial L-C acyl-CoAs and in turn lead to increases in L-C acylcarnitines via CPT-2. However, impaired L-C fatty acid β-oxidation will produce decreases in S-C and M-C acyl-CoAs, as L-C fatty acids are not broken down via repeated cycles of β-oxidation, leading to corresponding reductions in the respective S-C and M-C acylcarnitine derivatives. Therefore, reductions in S-C/L-C or M-C/L-C acylcarnitine ratios in the circulation often reflect impairments in mitochondrial L-C fatty acid oxidation rates. In contrast, general increases in numerous S-C, M-C, and L-C acylcarnitines often infer elevations in fatty acid oxidation. If fatty acid oxidation is impaired due to inhibition of mitochondrial fatty acid uptake, decreases in S-C, M-C, and L-C acylcarnitines are observed. CrAT = carnitine acetyltransferase; M-C = medium-chain; S-C = short-chain; other abbreviations as in Figure 1.

Although glucose and fatty acids represent the most common fuels used by our cells to produce energy, the majority of cells in our bodies are also capable of metabolizing amino acids (e.g., hepatocytes, skeletal muscle myocytes) and ketone bodies (e.g., neurons) to produce acetyl-CoA for oxidative energy metabolism (Figure 1). In particular, cellular metabolism of the branched-chain amino acids (BCAAs) leucine, isoleucine, and valine, has been extensively studied, as BCAAs are potent regulators of systemic metabolism, energy expenditure, and muscle protein synthesis (2). In order for BCAAs to enter the mitochondria for oxidative metabolism, they must first be converted into their respective branched-chain α-keto-acids (BCKAs), which are subsequently oxidized via BCKA dehydrogenase (Figure 3). BCAA/BCKA metabolism is of particular relevance during fasting/starvation, where skeletal muscle catabolism of BCAAs produces alanine to support hepatic gluconeogenesis, and hepatic catabolism of BCAAs leads to the formation of precursors for hepatic ketogenesis (Figure 3). Of interest, the increase in cellular metabolism of BCAAs in muscle has been strongly correlated with reduced insulin sensitivity and impaired glucose homeostasis (see the section Additional Considerations for the Use of Metabolomics in the Diagnosis and Prognosis of CVD) (2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Cellular Fates of BCAA Catabolism

Although Figure 1 depicts that BCAA metabolism can be used to generate acetyl-CoA for oxidative energy metabolism, BCAA catabolism serves a number of other purposes. In particular, during fasting/starvation, the first step in BCAA catabolism involves transfer of amino groups to α-ketoglutarate, followed by subsequent amino group transfer to pyruvate, producing alanine. This is of particular importance in skeletal muscle, where this alanine is exported to the liver to support hepatic gluconeogenesis (step 1). Conversely, enhanced BCAA (leucine and isoleucine) catabolism in the liver during fasting/starvation can lead to the formation of precursors for the biosynthesis of ketone bodies (step 2). Finally, BCAA catabolism can be used to refuel TCA cycle intermediates through anaplerosis, either at the level of α-ketoglutarate, or through formation of succinyl-CoA produced via subsequent breakdown of valine or isoleucine (step 3). Abbreviations as in Figure 1.

Myocardial Metabolism

In the healthy adult heart, virtually all ATP production is derived from mitochondrial oxidative metabolism, with the remainder primarily arising from glycolysis (3). Because myocardial ATP stores are relatively low (∼300 mg of total ATP for an average heart) compared with the high rate of ATP utilization, there is a nearly complete turnover of the myocardial ATP pool every 10 s (3). In order to meet this enormous energy demand, the healthy heart metabolizes more than 30 g of fat and 20 g of carbohydrate per day, and uses the equivalent of nearly 35 liters of oxygen while doing so (3). Furthermore, the metabolic flexibility of the heart is extremely dynamic, as demonstrated by its ability to rapidly alter its pattern of fuel utilization in order to adapt to its substrate and hormonal environment, whereby virtually all energy substrates, including fatty acids, glucose, lactate, ketone bodies, and amino acids, can be consumed by the heart to generate ATP.

Interestingly, the normal patterns of cardiac substrate metabolism are often perturbed in a variety of CVDs, and it is widely accepted that these metabolic changes directly contribute to disease pathophysiology. Indeed, alterations in cardiac metabolism assessed via coronary sinus catheterization combined with infusion of radiolabeled tracers, positron emission tomography (PET) imaging, or cardiac magnetic resonance imaging have been observed in patients with obesity/diabetes, ischemic heart disease (IHD), and/or heart failure (HF) (3). However, these methods can be invasive and costly, present potential health hazards due to use of radioisotopes with long half-lives (e.g., carbon-14 [¹⁴C]), and are often deployed in patients who already have established or advanced CVD, thus limiting the diagnostic potential of these assessments. Perturbations in myocardial metabolism frequently lead to the accumulation or loss of specific metabolites from the various metabolic pathways described previously, and changes in these metabolites are often reflected within the circulation (Figure 1). These metabolites represent the terminal product of the environment’s interaction with the genome-transcriptome-proteome (Central Illustration) in response to a disease process (4,5). Hence, the ability to accurately identify and quantify changes in these metabolites, and to use this information to detect perturbations in a specific metabolic process, adds to the arsenal of diagnostic information available to clinicians. Fortunately, significant advancements in metabolomics-based technology platforms now enable us to detect these various metabolites with high sensitivity and selectivity. Furthermore, a number of these metabolites may have clinical utility with regard to predicting (screening biomarkers) and detecting (diagnostic biomarkers) disease, evaluating disease progression/remission (prognostic biomarkers), and assessing therapeutic effectiveness.

Metabolomics

Limitations with the use of metabolomics to assess intermediary metabolism

Metabolomic profiling performed on blood or urine samples can provide significant insight into metabolic pathways that may be altered during the progression of numerous pathologies (Figures 4, 5, and 6, Table 1). Many of these diseases, such as obesity and/or type 2 diabetes (T2D) are complex syndromes that involve pathological changes in multiple organ systems, each of which may significantly contribute to the blood- and urine-based metabolomic profile. As such, using blood- and urine-based metabolomics as a diagnostic or prognostic tool is valuable, but cannot be considered definitive in informing us about new molecular pathophysiological processes that are occurring within the heart or any specific organ. Indeed, assessment of metabolism at the organ level necessitates the measurement of changes in metabolites across the organ. For example, the delta in metabolite levels between arterial blood and the coronary sinus for the heart or the renal artery and vein for the kidney provides more precise information as to the metabolic changes that are occurring in those organs. In addition, because metabolomics only provides a snapshot of metabolic flux through a particular pathway, the reason for increased or decreased metabolite levels may not be readily apparent. For instance, an aberrant metabolite concentration may be due to a defective metabolic pathway, changes in substrate supply, or abnormal metabolic rate of the primary product. Thus, predictions of defective metabolic processes on the basis of blood- or urine-based samples alone must be interpreted with caution. Regardless of these limitations, identification of numerous altered metabolites related to a common metabolic pathway in the blood or urine can yield significant insight into potential perturbations in metabolism that take place during disease progression. As such, metabolomic signatures in blood or urine can often yield hypothesis-generating findings. These findings ultimately can lead to intricate molecular explorations in a diseased organ, cell, or animal model, which in turn will enhance clinical utility and our understanding of disease pathobiology.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Metabolomics in IHD and Atherosclerosis

During myocardial ischemia, a reduction in blood flow and oxidative metabolism decreases substrate extraction by the heart, which contributes to the elevation in circulating FFAs and BCAAs, whereas anaerobic glycolysis increases to produce ATP in the absence of oxygen. Hence, increases in circulating lactate levels are frequently observed during ischemia. CAD resulting from atherosclerosis is often a major contributor to ischemic heart disease or myocardial infarction. It has been demonstrated that gut microbiota are capable of converting phosphatidylcholine/choline and L-carnitine into TMA, which the liver converts into TMAO. TMAO increases the risk for atherosclerosis potentially via interfering with reverse cholesterol transport and promoting plaque formation, which promotes IHD. Together, all of these changes in the metabolomic profile can provide significant diagnostic insight into the risk for atherosclerosis or presence of CAD. CAD = coronary artery disease; FFA = free fatty acid; IHD = ischemic heart disease; TMA = trimethylamine; TMAO = trimethylamine-N-oxide; other abbreviations as in Figure 1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Metabolomics in HF

The failing heart undergoes a well-characterized metabolic remodeling that consists of general reductions in mitochondrial function and oxidative energy metabolism, which is compensated via an increase in anaerobic glucose metabolism (glycolysis). Hence, circulating metabolomic profiles often yield increases in L-C acylcarnitines and lactate in HF patients. Circulating BCAAs are also often elevated in HF patients, although reasons for this increase are unknown. As recent studies in HF patients demonstrate impairments in the heart’s ability to break down BCKAs (Figure 3), this may lead to a buildup of myocardial BCAAs that spill over into the circulation. Furthermore, recent studies have identified that the failing heart may increase its reliance on ketone bodies as an oxidative energy source to compensate for the reduction in mitochondrial fatty acid oxidation, and circulating ketone bodies (e.g., β-hydroxybutyrate and acetoacetate) are often decreased in HF patients. Of interest, decreased circulating ketone bodies and increased circulating L-C acylcarnitines may also have diagnostic capability in differentiating HFrEF from HFpEF patients, in that serum concentrations of ketone bodies and L-C acylcarnitines are decreased and increased, respectively, in HFrEF patients compared with HFpEF patients. HF = heart failure; HFpEF = heart failure with preserved ejection fraction; HFrEF = heart failure with reduced ejection fraction; other abbreviations as in Figures 1 and 2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Metabolomics in Obesity and Diabetes

Obesity and type 2 diabetes contribute to a number of deleterious changes in multiple organs in the body. These changes include adipose tissue inflammation, skeletal muscle insulin resistance, nonalcoholic fatty liver disease, β-cell dysfunction, vascular dysfunction, and cardiomyopathy. Associated with these changes are metabolic derangements in these organs that can contribute to an increased risk for CVD in these individuals. Blood-based metabolomic profiles of individuals with obesity and type 2 diabetes often yield increased acylcarnitines, lactate, and BCAAs. CVD = cardiovascular disease; other abbreviations as in Figure 1.

• Download figure
• Open in new tab
• Download powerpoint

Central Illustration

Integration of Various Omics Platforms for Patient Phenotyping

The metabolome represents one of the downstream end products of the environment’s interaction with the genome-transcriptome-proteome.

Metabolomics and IHD

IHD is caused by inadequate perfusion of the myocardium and encompasses both stable and unstable angina, as well as myocardial infarction (MI). IHD represents the most common form of CVD, and although survival rates following an acute MI have greatly improved in recent years, mortality due to IHD still remains the leading cause of cardiovascular death globally (10). Because oxygen and nutrient supply/delivery to the myocardium are markedly reduced during ischemic periods, there are a number of key changes that manifest with regard to myocardial intermediary energy metabolism (Figure 4), often marked by a reduction in overall oxidative metabolism. In response to this decline in oxidative metabolism, glycogen breakdown and glycolysis rates are increased, as glycolysis can produce ATP anaerobically at a rapid rate (3). The extent of the increase in myocardial glycolysis is dependent on the severity of ischemia and its duration. Whether these myocardial metabolic alterations during myocardial ischemia can be detected via use of blood-based metabolomics is a question of great interest due to its diagnostic potential.

Metabolomics and Atherosclerosis

Because the vast majority of IHD is due to the presence of atherosclerotic plaques in the coronary vessels, there has been great interest in determining whether metabolomic profiling can identify individuals at increased risk for IHD due to underlying atherosclerosis. This is especially important considering a large number of patients with clinically diagnosed CAD exhibit derangements in myocardial metabolism, and the use of techniques such as PET imaging to confirm these derangements is costly. Hence, the use of blood-based metabolomic profiling to identify subclinical atherosclerosis and early CAD may be of clinical value. Indeed, a number of metabolomic studies in patient cohorts have identified a variety of potential biomarkers that can predict risk for CAD and subsequent cardiovascular events, including MI and death. Most notably, sophisticated studies have demonstrated that circulating trimethylamine-N-oxide (TMAO) is a significant predictor for atherosclerosis and incident risk for MI and stroke (26,27), with the gut microbiome being implicated as a critical factor regulating this process (26). Specifically, dietary phosphatidylcholine, choline, and carnitine are substrates for the generation of trimethylamine (TMA) via the host gut microbiome. This TMA is subsequently released into the circulation and is then oxidized by the liver to TMAO. TMAO, in turn, promotes the progression of atherosclerosis, potentially via interfering with reverse cholesterol transport, which subsequently increases the risk of cardiovascular events (Figure 4) (26). Recent studies indicate that TMAO may also increase platelet hyper-reactivity, as application of exogenous TMAO at physiological levels to human platelet-rich plasma enhanced platelet aggregation, while increasing the adherence of fluorescently labeled platelets within human whole blood to collagen (27). Furthermore, studies have shown that interfering with the gut microbiome’s ability to convert dietary choline or carnitine into TMA, via inhibiting bacterial TMA lyase with 3,3-dimethyl-1-butanol in a nonlethal manner, reduces circulating TMAO levels and the subsequent progression of atherosclerotic lesions in a mouse model of atherosclerosis (28). Interestingly, these studies support that increased circulating levels of phosphotidylcholine/choline and/or carnitine are associated with increased cardiovascular risk, yet only appear to provide true prognostic value in those individuals in which TMAO is also elevated (26). Together, these findings support TMAO as a novel biomarker for the diagnosis of atherosclerosis/CAD, and exemplify the power of metabolomics to produce hypothesis-generating findings that can lead to intricate molecular explorations, which ultimately improve our understanding of disease pathology.

Other identified biomarkers from blood-based metabolomic profiling include observations that increased levels of 18:2 monoglyceride or decreased levels of 18:2 lysophosphatidylcholine and 28:1 sphingomyelin are associated with increased risk for incident CAD events in pooled data from 3 case-cohort studies (29). In patients with angina or MI versus healthy subjects, an increased prevalence for CAD was associated with increased circulating levels of lysophosphatidylcholines containing unsaturated fatty acids, and decreased circulating levels of lysophosphatidylcholines containing saturated fatty acids (30). In addition, this study also identified that circulating phosphatidylcholines containing ceramide, sphingomyelin, diacylglycerol, or palmitic acid were associated with increased prevalence for MI. The diagnostic information these circulating metabolites harbor is currently unknown, but once again illustrates the ability of blood-based metabolomics to potentially identify novel metabolic processes that may contribute to disease pathology. In obese and T2D patients, numerous tissues often display marked accumulation of lipid intermediates, including sphingomyelins, ceramides, and DAGs (31). These lipids often contribute to a phenomenon referred to as lipotoxicity to reflect lipid-induced cellular toxicity and dysfunction. If increases in these circulating metabolites also reflect accumulation of these metabolites in organs such as the liver and skeletal muscle, perhaps these metabolites may have clinical utility in the diagnosis and prognosis of insulin resistance and T2D (see the section Additional Considerations for the Use of Metabolomics in the Diagnosis and Prognosis of CVD). As alluded to previously, circulating BCAAs predict risk for CAD, often even after correcting for T2D (15,24,25), and LC/MS-based metabolic profiling in a cross-sectional study of 472 Chinese subjects demonstrated that serum BCAAs are increased and positively correlated with carotid intima-media thickness, a well-established index of subclinical atherosclerosis (32). Importantly, if changes in these circulating metabolites are truly indicative of atherosclerotic risk, the detection of these biomarkers may allow for early intervention with the appropriate therapies.

Metabolomics and HF

HF affects more than 5 million people in North America, with more than one-half-million new cases in the United States every year (33,34). Despite the advent of several new medications and devices used to manage HF, there remains a high mortality rate of approximately 30% to 50% within 5 years (35). Although a vast array of pathophysiological and molecular events contributes to the development and progression of HF (36–39), a long-standing concept in HF is that the failing heart has defects in metabolic processes that normally permit proper ATP production necessary to maintain contractile function (39) (Figure 5). Within the failing heart, oxidative phosphorylation is impaired, oxygen consumption is depressed, and ATP production is compromised (39), with failing human hearts demonstrating approximately 30% lower ATP levels than nonfailing hearts (40).

Metabolomic profiling, when used alone or in combination with standard HF biomarkers (e.g., B-type natriuretic peptide [BNP]), has significant diagnostic and prognostic value (40). Although the majority of reports utilizing metabolomic analyses in HF are comparing non-HF controls with HF patients (41–43), other studies have shown that blood-based metabolomics can also differentiate between the degree of HF severity (44), or between HF with preserved ejection fraction (HFpEF) and HF with reduced ejection fraction (HFrEF) (42,45). As a result of the profound changes that occur in energy metabolism during HF, the identification of blood- or urine-based metabolic signatures for these patients may uniquely assist clinicians with the clinical management of HF through detection, classification, and/or markers of effective treatment (46,47). Even though changes in the levels of numerous metabolites have been shown to occur in HF, there appear to be some key metabolite classes that consistently change in metabolomic profiles of HF patients (Table 2), which will be discussed herein.

Metabolomics and CVD in Obesity and Diabetes

The prevalence of T2D is increasing at an alarming rate, due in large part to the obesity epidemic and our aging population (64–66). The disease usually develops later in life, and is often detected incidentally or after sequelae of the associated vascular disease or CVD has occurred (65,67). The application of high-throughput metabolomics to populations at risk of T2D has generated considerable excitement, given the potential to identify biomarkers predictive of incident T2D and its associated sequelae, and the possibility of gaining insight into novel modifiable disease pathways and therapeutic strategies for reversing the disease and its associated comorbidities (Figure 6) (68). Furthermore, identification of circulating metabolites that correlate with risk for insulin resistance (e.g., BCAAs) via metabolomics may aid in the treatment and overall prognosis of CVD. Hence, the use of metabolomics to screen for numerous metabolites linked to insulin resistance could theoretically aid in the diagnosis, prognosis, and management of CVD in obese and/or T2D patients. In the last decade, metabolomic profiling in obese and/or T2D subjects has identified a number of potential metabolites that may be mechanistically involved in the pathogenesis of these conditions and their associated comorbidities. We will herein discuss some of these perturbed metabolic pathways and their associated metabolites that may reflect perturbed myocardial metabolism, as well as their potential diagnostic implications.

PET imaging with 1-¹¹C-palmitate demonstrates increases in both myocardial fatty acid uptake and oxidation in obese and insulin-resistant women, as well as in men with uncomplicated T2D (1,69). Furthermore, numerous studies directly assessing fatty acid oxidation rates in hearts from obese and/or T2D animals have observed similar results (1,69). Though no assessment of the acylcarnitine profile has been made in myocardial biopsies in obese and/or T2D subjects, myocardial extracts in obese versus lean mice demonstrate elevations in S-C, M-C, and L-C acylcarnitines (70), reflecting observations seen in circulating and skeletal muscle acylcarnitine profiles of obese and/or T2D humans and animals (see the section Additional Considerations for the Use of Metabolomics in the Diagnosis and Prognosis of CVD) (50,52). Although the heart’s contribution to the circulating acylcarnitine pool remains unknown, the expression of carnitine acetyltransferase, a key enzyme involved in mediating carnitine transfer to acetyl-CoA for mitochondrial export (Figure 2), is greater in the heart than the skeletal muscle (71). This observation supports the possibility that the heart may contribute to the circulating acylcarnitine profile, and that increases in S-C, M-C, and L-C acylcarnitines in the circulation of obese and/or T2D patients do reflect increases in myocardial fatty acid oxidation rates. With regard to myocardial glucose metabolism in obese and/or T2D patients, the most frequently observed defect appears to take place at the level of glucose oxidation (1,70). Nevertheless, the vast majority of studies supporting this concept have been performed in animals, due to the lack of available methods to accurately assess myocardial glucose oxidation rates in humans. Currently, studies of myocardial ketone body and amino acid metabolism in humans are limited, and even more so in obese and/or T2D patients. Nevertheless, with recent metabolomic findings illustrating that the failing heart relies on ketone bodies as an adaptive mechanism to meet its oxidative energy needs, whereas myocardial BCAA/BCKA metabolism is impaired (see the section Metabolomics and HF) (59,63), it is likely that investigations on myocardial ketone body and BCAA metabolism in obesity and/or T2D will increase in order to elucidate their link with HF (72,73).

Additional Considerations for the Use of Metabolomics in the Diagnosis and Prognosis of CVD

Although the majority of this review has focused on how blood-based metabolomics can aid in the diagnosis and prognosis of various CVDs, patients with CVD often also have associated comorbidities, such as kidney disease and hepatic steatosis, as well as obesity and/or T2D (8,17). In fact, obesity and/or T2D may add extreme complexity to the understanding of the metabolomic profile, because these conditions often result in significant pathology and perturbed metabolism in other peripheral tissues (e.g., skeletal muscle, liver, adipose tissue, among others), and these peripheral tissues also contribute to the circulating metabolomic profile measured in patients with CVD. Indeed, systemic insulin resistance is a hallmark feature of obesity and T2D, and metabolomic profiling in obese and/or diabetic subjects often reveals increases in circulating lactate levels (74,75), which on the surface suggest increases in anaerobic glycolysis (Figure 1). However, increases in anaerobic glycolysis are not consistent with systemic insulin resistance, which impairs glucose metabolism in numerous tissues, including skeletal muscle. Conversely, proteomic studies have observed increases in the expression of glycolytic enzymes in vastus lateralis biopsies from obese and T2D patients (76). Furthermore, the source of elevated lactate in the circulation is difficult to ascertain, and can arise from other sources in addition to skeletal muscle. Because adipose tissue dysfunction in obesity is often associated with tissue hypoxia (77), increased anaerobic glycolysis in the expanded adipose depots of obese individuals may also contribute to this elevation in circulating lactate.

As circulating free fatty acids and triacylglycerols are markedly increased during obesity/T2D, fatty acid supply, uptake, and metabolism are frequently increased in peripheral tissues, such as the skeletal muscle and heart. Metabolomic profiling to assess alterations in fatty acid metabolism in obese and/or diabetic subjects has revealed increases in numerous circulating acylcarnitine species (52), whereas other studies have reported no differences (78). It has been suggested that a very large portion of the circulating acylcarnitine pool arises from skeletal muscle-mediated export, and muscle biopsies in obese humans also demonstrate increases in M-C and L-C acylcarnitines compared with lean controls (79). Thus, assessment of acylcarnitine profiles via blood-based metabolomics supports increases in both skeletal muscle and myocardial fatty acid oxidation rates (see the section Metabolomics and CVD in Obesity and Diabetes), although additional investigation is required to determine the contributions individual organs make to the circulating acylcarnitine profile.

Circulating ketone bodies (e.g., β-hydroxybutyrate and acetoacetate) are frequently elevated in obese and/or T2D patients (80), and elevations in circulating ketone bodies predict worsening of hyperglycemia in men with or without newly diagnosed T2D (81). Changes in circulating β-hydroxybutyrylcarnitine may reflect changes in peripheral tissue ketone body oxidation because β-hydroxybutyrylcarnitine is a mitochondrial intermediate derived directly from β-hydroxybutyryl-CoA (30). Of interest, recent findings from the EMPA-REG OUTCOME trial have observed that treatment with the sodium-glucose cotransporter 2 inhibitor, empagliflozin, markedly lowers cardiovascular risk in T2D patients, including a 35% reduction in hospitalization for HF (82). As these exciting findings were associated with a significant increase in circulating β-hydroxybutyrate levels, it has been suggested that empagliflozin reduces cardiovascular risk by increasing ketone body uptake and oxidation in the heart (83). This is consistent with increased ketone body oxidation being a potential beneficial and adaptive mechanism in the failing heart (see the section Metabolomics and HF), and highlights the need to understand the significance of blood-based metabolomic profiles revealing perturbations in ketone body metabolism.

Numerous studies have illustrated a tight correlation between circulating BCAA levels and risk for insulin resistance/T2D (2,75). Since these initial observations, several investigators have utilized novel metabolomic profiling techniques to implicate BCAAs in obesity and insulin resistance (2,17), with BCAAs and aromatic amino acids emerging as predictors of future T2D. Intriguingly, generation of an amino acid score on the basis of 3 of these amino acids (isoleucine, phenylalanine, and tyrosine) was found to significantly predict incident T2D (84). In light of findings suggesting impaired myocardial BCAA/BCKA metabolism in HF (63), it will be important to understand how the circulating BCAA profile in obesity reflects deranged systemic and myocardial metabolism.

Taken together, although blood-based metabolomics is proving to be a powerful tool that can assist in the diagnosis and prognosis of CVD, in the context of obesity and/or T2D, caution must be applied in how we infer the associated peripheral metabolic perturbations. In such instances, assessment of changes in metabolites across the diseased organ would prove more valuable, such as sampling arterial blood and the coronary sinus for the heart (see limitations of metabolomics described in the section Metabolomics), or having a tissue biopsy to compare against the serum/plasma sample. Nonetheless, due to the ease of obtaining a serum/plasma sample versus collecting a tissue biopsy or invasive catheterization, we must rely on molecular explorations in animal models of disease to confirm and/or understand novel metabolic alterations identified from human blood-based metabolomic signatures.

Final Summary and Implications

Metabolomic profiling technologies have significantly evolved to a point where high-resolution and high-throughput assessment of thousands of metabolites within a biological sample can be performed with relative ease. Due to the expansive biochemical diversity of metabolites within the human metabolome, several complementary platforms and techniques are necessary for comprehensive metabolic characterization. These techniques, MS and NMR coupled with LC or gas chromatography, are exquisitely sensitive, generating the metabolic fingerprint of a biological sample that not only reflects the condition of interest, but also diet, drug effects, sex, and other comorbidities and exposures. Thus, human metabolomic studies are consequently prone to clinical confounding factors that may lead to spurious diagnoses/conclusions. Yet the application of metabolomic profiling to several large population-based epidemiological cohorts has allowed for robust statistical adjustment for potential confounders, and has resulted in externally reproducible findings. In addition, metabolomics has been applied in smaller studies of unique clinical scenarios that allow for serial sampling before and after a controlled biological perturbation (e.g., exercise testing, planned MI, drug administration, and so on). Such paired experiments allow subjects to serve as their own controls, mitigating the risk of confounding factors. In addition, emerging investigational techniques are integrating metabolomics with other “omics” platforms in order to gain insight into pathophysiological interactions of metabolites, proteins, genes, and disease states. Furthermore, the state-of-the-art metabolomic technologies now available provide us with a snapshot of the metabolic fingerprints of individual patients, which can serve as diagnostic and/or prognostic tools that can be used to identify impairments in systemic or myocardial metabolism occurring during the development and worsening of CVD, as well as identifying the types and timing of specific interventions/therapies. Thus, metabolomics is a powerful technology that is transforming our ability to predict, detect, and understand a myriad of cardiometabolic disease states, and to monitor the effectiveness of therapeutic interventions. In doing so, metabolomics continues to advance our societal objective of personalizing the practice of medicine.