Energetics and Metabolism in the Failing Heart: Important But Poorly Understood

Abstract

Purpose of review

Profound abnormalities in myocardial energy metabolism occur in heart failure and correlate with clinical symptoms and survival. Available comprehensive human metabolic data comes from small studies, enrolling patients across heart failure etiologies, at different disease stages, and using different methodologies, and is often contradictory. Remaining fundamental gaps in knowledge include whether observed shifts in cardiac substrate utilization are adaptive or maladaptive, causal or an epiphenomenon of heart failure.

Recent findings

Recent studies have characterized the temporal changes in myocardial substrate metabolism involved in progression of heart failure, the role of insulin resistance, and the mechanisms of mitochondrial dysfunction in heart failure. The concept of metabolic inflexibility has been proposed to explain the lack of energetic and mechanical reserve in the failing heart.

Summary

Despite current therapies, which provide substantial benefits to patients, heart failure remains a progressive disease, and new approaches to treatment are necessary. Developing metabolic interventions would be facilitated by systems-level integration of current knowledge on myocardial metabolic control. Although preliminary evidence suggests that metabolic modulators inducing a shift towards carbohydrate utilization seem generally beneficial in the failing heart, such interventions should be matched to the stage of metabolic deregulation in the progression of heart failure.

Introduction

Heart failure (HF) affects nearly six million people in the United States. The current magnitude of this problem from a public health standpoint cannot be overstated, and the future burden on the healthcare system is likely to be increasingly significant in view of the continued high incidence of the most common risk factors such as coronary atherosclerosis, diabetes mellitus and hypertension. The estimated direct and indirect cost of HF management in the United States is almost 40 billion dollars per year.

The clinical hallmark of HF is the inability of the heart to provide adequate perfusion to the body. To serve its function as a pump, the myocardium must turn over vast amounts of ATP to fuel sarcomeric contraction and relaxation, ion pumps, and macromolecular synthesis. With such high energetic requirements and an inextricable link between substrate metabolism and contractile function, it is not surprising that any compromise in the myocardium’s ability to generate high-energy phosphates would be associated with HF.

Despite the central role of dysregulated metabolism in HF, there is a surprising lack of clarity with regard to the precise metabolic changes that occur in humans with this disease. Only a relatively small number of HF patients have been characterized in terms of myocardial substrate oxidation and uptake, each from various stages of disease, with heterogeneous etiologies of left ventricular (LV) dysfunction, and with different methods of study. Not surprisingly, divergent conclusions have been drawn. The purpose of this review is to highlight current areas of controversy and agreement in substrate metabolism of the failing heart, with a particular emphasis on alterations in glucose utilization.

The failing myocardium is energy starved

There is clear evidence that the myocardium in HF is energy-depleted. Myocardial ATP levels, particularly in the context of advanced HF, are progressively reduced by approximately 25–35%. [1, 2] Furthermore, there are substantial decreases in creatine kinase activity and total creatine pool (sum of free and phosphocreatine content) by as much as 50–70% in the failing heart.[3–5] Several lines of evidence also suggest an impaired transfer of phosphoryl groups between spatially distinct intracellular sites of ATP production and utilization in HF, normally occurring by means of metabolic relays (via creatine kinase, adenylate kinase, and glycolysis) that compartmentalize the ATP supply inside the cardiomyocyte without exchange with bulk cytosolic pools for improved efficiency.[6] Thus disruption of excitation-contractile function in HF may be more substantial than is simply suggested by measurements of bulk myocardial ATP concentrations. The altered energetic state in HF is complex, involving limitations in all three major ATP-synthesis pathways, namely reduced mitochondrial ATP synthesis, reduced capacity for phosphoryl transfer, and impaired glycolytic reserve.

Reversion to the fetal metabolic phenotype - still a hypothesis

The biological mechanisms leading to the abnormal myocardial energetic state in HF remain poorly understood. Under normal aerobic and hemodynamic load conditions, cardiac function is not limited by availability of the main energy-providing oxidizable substrates, long-chain fatty acids (FA) and carbohydrates. However, since the profile of myocardial substrate oxidation in most models of HF shifts away from FA, multiple authors have drawn analogies to the metabolic behavior of the heart immediately after birth where major changes in substrate utilization also occur. In utero, where insulin levels are relatively high and oxygen tension low, glucose and lactate metabolism are the predominant source of myocardial fuel. After birth, oxygen tension and myocardial work increase, requiring a switch towards oxidation of FA. Ultimately, about 70% of the energy requirement of the adult heart under resting conditions is provided by oxidation of long-chain FA. Myocardial glucose and FA metabolism remain tightly coupled, with increased FA metabolism inhibiting glucose metabolism and vice versa, an effect referred to as the glucose-fatty acid, or Randle cycle.

Several lines of evidence, including ultrastructural and gene expression studies, indirectly support the paradigm that in HF there is reversion to the fetal metabolic phenotype of preferential carbohydrate utilization at the expense of FA. The number and size of mitochondria are decreased, and their function is impaired.[7–10] Myocardial expression of genes responsible for FA oxidation is suppressed[11, 12] in concert with downregulation of key effectors of lipid metabolism and mitochondrial biogenesis, such as PPAR-α and PGC-1α,[13–15] suggesting the switch is regulated at both transcriptional and post-translational level.

A considerable body of in vivo experimental evidence also directly supports this hypothesis of fetal phenotype switching, based on assessment of cardiac metabolic parameters using various methodologies. An early study using aortic-banded guinea pigs as a model of HF noted impaired palmitate oxidation coupled with significantly lower carnitine concentrations in failing myocardium compared with controls.[16] The site of this ‘block’ appeared to be the rate of conversion of the FA to its cognate carnitine. Notably, however, these investigators did not report any “compensatory” increase in the rate of glucose oxidation.

Three decades later, using a canine model of pacing-induced LV dysfunction, investigators confirmed the previously observed decreased rates of FA oxidation in the failing heart.[17, 18] This drop was associated with lower activities of carnitine palmitoyl transferase-I (CPT-1) and medium-chain acyl-CoA dehydrogenase (MCAD), both key enzymes in FA metabolism. Importantly, rates of glucose uptake and oxidation were increased with HF, again supporting the previously proposed concept of reversion to a fetal metabolic phenotype.

Since then, these results have been borne out in humans. In a small study involving seven patients with idiopathic dilated cardiomyopathy (DCM), rates of myocardial FA uptake and oxidation measured by positron emission tomography (PET) were significantly lower compared with controls, while glucose uptake was increased (glucose oxidation was not measured).[19] This shift from FA to glucose utilization was also found by directly measuring arterio-venous substrate gradients in an elegant experiment by Neglia et al.[20] Here, ten patients with IDCM underwent cardiac catheterization with coronary sinus sampling. Baseline (and importantly pacing-induced stress) measurements of substrate extraction and oxidation were made following infusions of radiolabeled substrates. Rates of FA uptake and oxidation were convincingly lower and glucose uptake increased in the HF group compared with controls (although again, glucose oxidation was not directly measured).

The concept that HF promotes a fetal pattern of substrate utilization is widely discussed.[21, 22•] Given the very small clinical studies to date and the limitations in measuring substrate oxidation in humans, little is known about the extent to which different metabolic pathways are deregulated in patients with various forms of HF. Moreover, it remains unclear whether this shift away from FA oxidation is a late consequence of HF, an adaptive response associated with the overall fetal molecular phenotype, or even causal in the progression of HF. Addressing these issues is of fundamental importance for the design of stage-specific metabolic interventions aimed to restore myocardial substrate oxidation in HF patients.

Alternative substrate-utilization paradigms

It is important to recognize that depressed FA oxidation has not been observed in every study of HF. In fact, the first study to examine substrate preference of the failing heart in humans by indirect calorimetry documented significant increases in FA uptake and utilization in HF patients compared with controls.[23] Conversely, glucose metabolism was found to be suppressed. Of note, the study was small and enrolled HF patients across non-ischemic etiologies.

Several subsequent PET studies supported these observations. Using [¹⁸F]-FDG and the FA tracer [¹⁸F]-fluoro-6-thia-heptadecanoic acid, Taylor et al reported lower glucose and higher FA uptake among 12 patients with New York Heart Association (NYHA) Class III HF compared to historical controls (the study lacked matched controls).[24] This group then went on to show that following three months of treatment with the mixed α- and β-blocker carvedilol, FA use by PET was significantly decreased while glucose remained unchanged (but trended up).[25] The authors noted a significant increase in ejection fraction (26±2% to 37±4%) associated with carvedilol therapy and the fall in FA utilization, a finding that would seemingly argue against the idea of impaired FA oxidation in HF. Indeed, previous studies have also found that β-blockers increase myocardial energetics and function by enhancing glucose oxidation at the expense of FAs.[26–28] Myocardial glucose uptake was similarly not different (but slightly lower) in clinically stable HF patients with DCM compared to age-matched healthy controls when studied by PET in euglycemic hyperinsulinemia, even though interestingly oxidative metabolism was increased in the right but not the left ventricle suggesting a metabolic imbalance between ventricles in the failing heart.[29]

Consistent with the above PET studies, direct quantification of myocardial metabolic substrate uptake by the arterio-venous gradient technique combined with a systemic infusion of ²H₂-palmitate revealed no differences between patients with compensated HF, type-2 diabetes, and matched controls.[30•] Although limited by inclusion of patients with both idiopathic and ischemic HF etiologies, lack of coronary blood flow and substrate oxidation measurements, this study suggested that the human moderately failing heart is not clearly prone to increased glucose utilization.

Most recently, we reported transcoronary gradients of substrate extraction among 37 patients undergoing cardiac surgery.[31•] Neither glucose nor FA extraction were significantly different between patients with or without LV dysfunction at baseline. Although rates of substrate oxidation were not measured directly, transcoronary elution of acylcarnitines - metabolic intermediates reflective of FA, amino acid and glycolytic metabolism - were not different based on the presence or absence of LV dysfunction. Once again, there was no clear signal in humans to support the hypothesis of reversion to a fetal pattern of metabolism in HF.

The roles of hyperadrenergic state and insulin resistance in metabolic abnormalities in HF

Opie [32] proposed an intriguing concept that the hyperadrenergic state of HF initiates an adverse metabolic vicious cycle, whereby aberrant metabolism and in some cases insulin resistance (IR) further promote the progression of HF. Briefly, the compensatory sympathetic activation invoked by HF results in the following parallel chains of events: a) increased circulating FA - reduced glycolysis and glucose uptake (by heart and skeletal muscle) - increased plasma glucose – increased reactive oxygen species –reduced glycolysis (glyceraldehyde-3-phosphate dehydrogenase) – multifactorial insulin resistance; b) increased circulating FA – increased FA uptake – FA activation and mitochondrial transport (CPT-1) – uncoupled mitochondrial respiration and oxygen waste.[33]

Increasing epidemiological evidence suggests a bidirectional relationship between IR and HF. A robust association exists between type-2 diabetes, IR, and increased incidence of HF, while diabetic cardiomyopathy has become a diagnostic entity with recognized myocardial metabolic alterations supporting a possible etiologic link. Conversely, preliminary evidence also suggests that HF may predispose to IR and type-2 diabetes. Early hyperinsulinemic-euglycemic clamp studies reported that, irrespective of etiology, HF is associated with systemic IR that independently predicts increased mortality.[34] However, it has since been proposed that although whole-body and myocardial IR do not invariably coexist and may have different pathophysiologic mechanisms, HF is associated with the development of myocardial IR. This concept is based on several lines of evidence. First, a number of PET studies suggest that failing human myocardium has reduced glucose uptake in favor of FA uptake,[23, 24] irrespective of diabetic status or ischemic versus non-ischemic etiology of HF.[35] Two recent elegant experimental studies in rat models of pressure overload [36•] and postinfarct induced HF [37•] have reported the development of myocardial IR associated with substantial reductions in both FA and glucose oxidation. In the case of pressure overload HF, cardiac IR preceded the onset of mitochondrial and contractile dysfunction,[36•] suggesting a potential causal link. This is consistent with data in genetically engineered mice with impaired myocardial insulin signaling by cardiomyocyte-selective ablation of the insulin receptor, which led to mitochondrial dysfunction through coordinated reductions in tricarboxylic acid and FA oxidative capacity, as well as oxidative stress and mitochondrial uncoupling.[38•] Finally, indirect evidence is provided by preliminary studies of the incretin hormone glucagon-like peptide 1, known to increase plasma insulin levels and mitigate IR, which was shown to ameliorate HF.[39]

The complex interaction between myocardial substrate metabolism, degree of contractile dysfunction, insulin resistance, and β-blocker medication has been recently addressed in both clinical and experimental studies. Tuunanen et al [40] made the intriguing observation that, although myocardial substrate metabolism (by PET) in DCM patients is generally shifted from FA utilization towards alternative substrates, this shift is modulated and counteracted by further deterioration of LV function and development of IR.

The metabolic effects of β-blockers are to decrease myocardial FA uptake and improve work efficiency. Most studies have shown that β-blockers increase whole-body insulin sensitivity in patients with HF. In the heart, a shift towards increased glucose metabolism in patients with HF has been reported with nonselective β-blockers but not with β1-specific blockers.[25, 41]

Myocardial utilization of the main metabolic substrates glucose and long-chain FA is regulated not only by substrate availability and competition at the level of the mitochondria, but also at the site of cellular entry. Both experimental and human data have shown that shifts in myocardial fuel balance associated with HF can be linked to altered regulation of sarcolemmal substrate transporters, which we briefly review below. In the case of glucose metabolism, deregulation of both facilitated-diffusion glucose transporters (GLUT1 and GLUT4) and sodium-dependent glucose transporter (SGLT1) has been demonstrated in HF. Overall downregulation of myocardial GLUT expression is accompanied by an increase in GLUT4/GLUT1 ratio in HF,[12] without significant differences between diabetic and non-diabetic patients.[42•] Conversely, myocardial SGLT1 is upregulated in end-stage diabetic and ischemic cardiomyopathy, but not changed in DCM. Furthermore, SGLT1 expression was increased following implantation of a left ventricular assist device, and correlated with functional recovery, suggesting that upregulation of SGLT1 may be an adaptive response to injury.[43•] One functional role of glucose transporters in human myocardium is to mediate the substrate-dependent positive inotropic effects of insulin, which has been demonstrated to occur in both normal and failing (ICM>DCM) human hearts via the glucose transporters GLUT4 and SGLT1.[44•] Additional insight is provided from genetically engineered GLUT4 knockout mice, which surprisingly displayed cardiac hypertrophy [45] associated with a compensatory upregulation of GLUT1 and downregulation of proteins required in FA utilization.[46] Increasing insulin-independent glucose uptake and glycolysis in adult hearts by cardiac-specific overexpression of GLUT1 in mice not only did not compromise cardiac function but protected against progression to HF and improved survival after chronic pressure overload, seemingly because of the higher efficiency of glucose compared with FA utilization with respect to oxygen consumption.[47] However, a follow-up study using the same transgenic mouse model of cardiac-specific GLUT1 overexpression showed that, while complete adaptation to chronic increases in glucose uptake and oxidation occurred, failure to upregulate myocardial FA oxidation during high fat diet-induced obesity resulted in increased oxidative stress, FA overload and contractile dysfunction. The authors conclude that chronic changes in substrate availability induce adjustments in cardiac metabolic networks at the expense of network flexibility, thus rendering it vulnerable to accelerated deterioration when other stressors superimpose.[48•]

On the other hand, the concept that both glucose and FA oxidation are required for optimal function of the failing heart is supported by clinical studies of acute FA deprivation by acipimox, an inhibitor of lipolysis, that unexpectedly resulted in further depression of cardiac work, uncoupling of cardiac contractile function from oxidative metabolism, and reduced cardiac efficiency in DCM HF patients in contrast to healthy controls.[49] The importance of these results is two-fold. Mechanistically, it suggests that myocardial FA oxidation enzymes can be acutely upregulated in patients with moderately severe HF when FA levels are suppressed. Therapeutically, it argues against the prevalent current paradigm that acutely switching myocardial substrate metabolism from FA to glucose may be beneficial for the failing heart, at least when secondary to DCM; however, these conclusions cannot be generalized to long-term metabolic modulation.

The complexity of myocardial substrate utilization in HF is also indirectly suggested by several studies demonstrating a “reverse epidemiology” of obesity and hyperglycemia in the prognosis of patients with chronic, established HF. Higher BMI, high HbA_1c, [50, 51•] and higher levels of fasting glycemia [52•] have been unexpectedly associated with improved rather than impaired outcomes in chronic HF, irrespective of diabetic status. Though not directly tested, availability of glucose as the preferred fuel for the failing heart with compromised energetic state has been speculated as a possible explanation for these paradoxical findings.

As a corollary to the studies presented above, Taegtmeyer [53] proposed the concept that metabolic extremes, either as fuel overabundance or absence, can cause lipotoxicity or glucotoxicity, both of which are adverse for the failing heart and lead to contractile dysfunction.

Reconciling the discrepancies

The currently available data on substrate preferences in HF stem from small studies enrolling patients across HF etiologies, with different clinical staging and metabolic backgrounds, and importantly using different technologies. This makes reconciliation of these disparate findings difficult. However, insights provided by recent studies in animal models provide some useful clues.

Chandler et al,[54] created an experimental model of permanent LV dysfunction – in contrast to reversible HF models induced by rapid pacing - by embolizing microparticles to the coronary arteries in dogs. These hearts demonstrated significant functional impairment as measured by rate-pressure product, dP/dt, and LV ejection fraction (EF) (28±1%), but the animals were otherwise considered to be well-compensated and their HF characterized as “moderate.” Under these conditions, FA uptake and oxidation was higher and glucose utilization was lower than in control animals, again seemingly at odds with the hypothesis of reversion to a fetal metabolic phenotype. The authors speculated that myocardial glucose metabolism may only be enhanced in late stages of HF.

A subsequent study of ischemic cardiomyopathy in mice demonstrated that while genes involved in the metabolism of both glucose and FA were initially downregulated for the first month after myocardial infarction, by three months levels of these transcripts had essentially normalized in the surviving myocardium.[55] These changes were accompanied by activation of other markers of the fetal gene program (e.g. ANF and β- MHC induction). These results suggest that although dysregulated in the acute phase, during the more stable, chronic phase of HF, there may be gradual normalization of metabolic control.

In follow-up to their group’s previous reports of tachy-pacing induced HF,[17, 18] Qanud et al recently reported that substrate switching towards enhanced glucose metabolism during severe HF was reversible in the recovery phase following cessation of pacing.[56] Although largely normalized towards baseline, the recovering ventricles still demonstrated some functional impairment, including depressed LV EF and dP/dt_max. The authors concluded that cardiac substrate selection returns to normal during post-HF recovery, but an additional interpretation, consistent with Chandler et al, is that milder forms of compensated LV dysfunction are not associated with significant changes in myocardial substrate preferences.

Most recently, using a rat model of chronic pressure overload-induced hypertrophy and transition to HF, Doenst et al [57••] sequentially quantified myocardial FA and glucose oxidation ex vivo during the development of HF and contractile dysfunction. They found an early impairment of FA oxidation (at 2 weeks) preceding the onset of HF in vivo, which was associated with a progressive decrease in glucose oxidation. In contrast, mitochondrial respiratory capacity was maintained in the early stages of HF (with preserved LV EF), and declined only later (after 10 weeks) in hearts with contractile dysfunction. Interestingly, although the observed increased ratios of glucose to FA oxidation in HF could be interpreted as a shift towards greater reliance on glucose utilization, in fact both oxidation rates decreased and the ‘shift’ only reflected a greater relative reduction in FA oxidation. Using the same rat model of chronic pressure overload this group has further reported that development of mitochondrial dysfunction, which appears specific to the interfibrillar mitochondria, may outweigh the benefits of preferential glucose oxidation by limiting overall oxidative capacity in the heart, and contribute to cardiac decompensation in late stage HF.[58•]

Hence, a potential unifying notion to explain the disparate data of cardiac substrate selectivity in humans with HF would be that in the early phases of the disease, there is an increase in FA metabolism (due to increased availability), which inhibits myocardial glucose metabolism but ultimately overloads the system and results in accumulation of FA intermediates and negative feedback inhibition of FA oxidation via malonyl CoA and CPT-1. This perpetual metabolic inefficiency leads first to loss of energy reserve supported by creatine kinase, with a reduction in phosphocreatine levels. In advanced HF, FA metabolism is significantly reduced, driven in principal by downregulation of PPAR-α as the heart suppresses adult metabolic gene transcripts and reverts to a fetal phenotype, in parallel with an increase in glycolysis uncoupled from glucose oxidation. This metabolic remodeling, mirroring that seen in hibernating myocardium,[59, 60] then becomes a pro-survival adaptation for the cardiomyocyte in response to stress.[21] However, further depression of LV function and development of profound insulin resistance induce a shift back to greater FA uptake and oxidation, which appears to be independent of β-blocker medication.[40] Ultimately, impaired mitochondrial oxidative capacity and metabolic inflexibility result in catastrophic energy failure with reductions in both ATP and phosphocreatine. These alterations in mitochondrial function and biogenesis in end-stage HF are caused by reduced expression of the metabolic master regulator PGC-1α, and are at least in part mediated via endothelin-1 and angiotensin II.[61•]

Although no study to date has enrolled sufficient patients in detailed metabolic analyses to adequately stratify by disease duration or NYHA class, reasonable evidence exists that different pathways for ATP synthesis are compromised at different times and to different extents in the progression of HF, and therefore interventions designed to alter metabolic pathways must be matched to the stage of metabolic deregulation, analogous to NYHA classes.

The dysfunctional heart lacks metabolic flexibility

In recent years, a striking picture has begun to emerge that the failing heart, which clearly lacks mechanical reserve, also lacks the metabolic and energetic reserve to deal with hemodynamic stress. Qanud et al,[56] demonstrated that in response to rapid pacing, the failing canine heart was unable to significantly increase its oxidation of FA or glucose, as was the normal myocardium. Importantly, even in those hearts which had been allowed to recover to near-normal function following tachycardia-induced pacing and which appeared metabolically identical to control hearts at baseline, the initiation of rapid pacing led to significant drops in FA oxidation and the inability to increase myocardial oxygen consumption.

The contention that the failing heart lacks metabolic flexibility to deal with stress has been shown in humans, as well. In their study of patients with non-ischemic cardiomyopathy, Neglia et al,[20] found the dysfunctional ventricles unable to augment their substrate uptake or oxidation in response to pacing stress. The result was mechanical inefficiency and an inability to increase the indexed rate-pressure product compared with normal controls.

Although no significant differences in baseline substrate utilization were seen between patients with and without HF, in the study by Turer et al marked changes in substrate uptake were noted among those with LV dysfunction following global myocardial ischemia-reperfusion associated with surgical cardioplegic arrest.[31] Under these conditions, fuel uptake was massively deranged, and the myocardium appeared to favor anaerobic metabolism. “Unmasking” of this underlying metabolic fragility by the ischemic stress was associated with increased duration of inotropic support, again highlighting the relationship between hemodynamic decompensation and perturbed cardiac metabolism.

Overall, it appears that the dysfunctional ventricle, which undergoes chronic modulation to alter its substrate selection, nonetheless remains at a precarious equilibrium. Underlying metabolic derangements become manifest under stress conditions (e.g. ischemia or high workloads). Whether a direct cause of the resulting hemodynamic effects or just an association is unclear, but it is intriguing to speculate that acute metabolic shifts may be a key inciting mechanism of HF flares. Indeed, recent transgenic murine models of PDH-inhibition[62] and phosphocreatine deficiency,[63] which at baseline display no obvious myocardial phenotype, undergo cardiac decompensation with stress.

Conclusions

Excitation-contraction-relaxation coupling in the heart is critically dependent on continuous supply of high-energy phosphates. Since myocardial energetic state is clearly altered in the setting of HF, it is reasonable to suggest that it may also play a role in the progression of disease. However, the processes involved in disruption of ATP synthesis, the timing and consequences of switches in substrate utilization, and the potential reversibility of energetic defects in HF remain incompletely characterized. Discrepancies in the published clinical and experimental data likely reflect the heterogeneity of study design and patient characteristics, as well as methodological difficulties of measuring metabolic fluxes in intact tissues. These prominent gaps in our knowledge present hurdles to our ability to modulate metabolism of the failing heart.

HF is a public health problem with enormous financial and human tolls. With risk factors continuing to increase in prevalence world-wide, HF prevalence will also continue to rise. Our current level of understanding of alterations in cardiac metabolism remains inadequate given the scope of this disease. Our goals should be to more clearly understand how perturbations in cardiac metabolism may be related to the pathophysiology of HF and importantly how metabolism may be modulated to maximize the efficiency of existing myocardium. This will require larger studies enrolling patients with different etiologies and stages of disease progression.