Author + information
- Received June 8, 2015
- Revision received September 7, 2015
- Accepted October 22, 2015
- Published online January 26, 2016.
- Tariq Ahmad, MD, MPH∗,†∗ (, )
- Jacob P. Kelly, MD†,‡,
- Robert W. McGarrah, MD†,§,
- Anne S. Hellkamp, PhD‡,
- Mona Fiuzat, PharmD‡,
- Jeffrey M. Testani, MD, MTR∗,
- Teresa S. Wang, MD‖,
- Amanda Verma, MD†,
- Marc D. Samsky, MD†,
- Mark P. Donahue, MD†,
- Olga R. Ilkayeva, PhD§,
- Dawn E. Bowles, PhD¶,
- Chetan B. Patel, MD†,‡,
- Carmelo A. Milano, MD¶,
- Joseph G. Rogers, MD†,‡,
- G. Michael Felker, MD, MHS†,‡,
- Christopher M. O’Connor, MD‡,#,
- Svati H. Shah, MD, MPH†,§ and
- William E. Kraus, MD†,§
- ∗Section of Cardiovascular Medicine, Yale School of Medicine, New Haven, Connecticut
- †Department of Internal Medicine, Division of Cardiology, Duke University Medical Center, Durham, North Carolina
- ‡Duke Clinical Research Institute, Duke University, Durham, North Carolina
- §Duke Molecular Physiology Institute, Duke University, Durham, North Carolina
- ‖Department of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania
- ¶Division of Cardiac Surgery, Duke University Medical Center, Durham, North Carolina
- #Inova Heart and Vascular Institute, Falls Church, Virginia
- ↵∗Reprint requests and correspondence:
Dr. Tariq Ahmad, Section of Cardiovascular Medicine, Yale University School of Medicine, 330 Cedar Street, New Haven, Connecticut 06510.
Background Heart failure (HF) is characterized by perturbations in energy homeostasis and metabolism. The reversibility and prognostic value of circulating markers associated with these changes remain unclear.
Objectives This study sought to describe the metabolomic profiles of patients along the spectrum of systolic HF, determine their association with adverse outcomes in a clinical trial of HF, and evaluate whether identified metabolites change with treatment for end-stage systolic HF.
Methods To assess association of metabolites with clinical outcomes, we evaluated a population of 453 chronic systolic HF patients who had been randomized to exercise training versus usual care. To assess change in metabolites with mechanical circulatory support, 41 patients with end-stage HF who underwent left ventricular assist device (LVAD) placement were studied. Targeted, quantitative profiling of 60 metabolites using tandem flow injection mass spectrometry was performed on frozen plasma samples obtained prior to randomization, as well as prior to and ≥90 days post-placement in the LVAD group. Principal components analysis was used for data reduction.
Results Five principal components analysis–derived factors were significantly associated with peak Vo2 levels at baseline in fully adjusted models. Of these, factor 5 (composed of long-chain acylcarnitines) was associated with increased risk of all 3 pre-specified clinical trial outcomes: all-cause mortality/all-cause hospitalization, all cause-hospitalization, and cardiovascular death or cardiovascular hospitalization. Individual components of factor 5 were significantly higher in patients with end-stage HF prior to LVAD placement and decreased significantly post-implantation.
Conclusions In chronic HF patients, circulating long-chain acylcarnitine metabolite levels were independently associated with adverse clinical outcomes and decreased after long-term mechanical circulatory support. These metabolites may serve as potential targets for new diagnostics or therapeutic interventions. (Exercise Training Program to Improve Clinical Outcomes in Individuals With Congestive Heart Failure; NCT00047437)
Heart failure (HF) is a global health problem with an estimated prevalence of 38 million patients worldwide, a number that is increasing with the aging of the population. The most common diagnosis in patients 65 years or older admitted to a hospital in high-income nations (1), HF possesses a prognosis worse than that of most cancers. Pharmaceutical treatments have primarily focused on neurohormonal blockade with β-blockers, angiotensin-converting enzyme inhibitors, angiotensin-II receptor blockers, and aldosterone antagonists. Because HF is a complex syndrome, however, identification of novel molecular mechanisms might lead to new therapies.
Patients with failing hearts are characterized by structural, functional, inflammatory, and metabolic derangements that develop and worsen during disease progression (2). Despite the multifactorial causes for HF, it has been suggested that as hearts begin to fail, altered energetics play an increasingly important role in pathogenesis, that is, the heart becomes “an engine out of fuel” (3). The heart is among the most metabolically active organs in the body, utilizing an entire supply of adenosine triphosphate every 13 s; to accomplish this, it primarily uses free fatty acids (FAs) as energy substrates, and switches to favor glucose metabolism during states of stress. During the progression of HF, glycolysis rises as an adaptation to the reduced oxidative metabolism that is uncoupled from glucose oxidation. This is further exacerbated by an increase in circulating FAs that occurs via several mechanisms, including reduced update and mitochondrial oxidative metabolism (4).
Metabolomics, the study of small-molecule metabolites, aims to uncover the underlying pathophysiological processes of the body with regard to energy homeostasis and metabolism; however, the prognostic or therapeutic implications of metabolic profile derangements in HF remain unclear (5–7). With the availability of novel compounds that stabilize or reverse mitochondrial dysfunction, human studies to delineate the circulating profile of this derangement, and establishing potential for reversibility of metabolic derangements in HF, could lead to a better understanding of whether they might be efficacious in this disease state (8,9).
We therefore sought to characterize circulating metabolites associated with poor outcomes in chronic systolic HF patients and assess whether these prognostic profiles are modifiable with mechanical circulatory support for end-stage HF with long-term left ventricular assist device (LVAD) support.
To assess the prognostic significance of metabolites, we analyzed a subgroup of the HF-ACTION (Heart Failure: A Controlled Trial Investigating Outcomes of Exercise Training) trial of chronic systolic HF patients. Details of the participants have been previously discussed (10). Specifically, 452 of 2,331 patients enrolled in the HF-ACTION study had agreed to participate in a biomarker substudy that required collection and banking of peripheral blood samples for purpose of research, and on whom metabolomics profiling was performed. Patients were 18 years or older, with left ventricular systolic dysfunction (left ventricular ejection fraction [LVEF] <35%) and ambulatory HF; they were randomized to exercise training versus usual HF care. Outcomes of interest included changes in exercise capacity measured by distance walked during a 6-min walk test and peak oxygen consumption (Vo2) measured from a cardiopulmonary exercise test; clinical endpoints included all-cause mortality, cardiovascular mortality, cardiovascular hospitalization, and HF hospitalization.
To assess modifiability of the prognostic metabolites with mechanical treatment for HF, we studied 41 consecutive patients, ages 18 years and older, who were deemed to have end-stage HF and required mechanical circulatory support with continuous-flow LVAD as a bridge to transplantation or destination therapy at Duke University Medical Center between January 1, 2011, and October 30, 2012. Details of the cohort and methodology are described elsewhere (2). These patients had agreed to collection and banking of peripheral blood samples for the purpose of research. Patients had blood samples collected prior to LVAD and had paired long-term samples available post-LVAD placement (median time: >136 days [range: 94 to 180 days]).
Approved by the Duke University Medical Center Institutional Review Board, these studies were performed in accordance with the ethical guidelines of the Declaration of Helsinki; all patients provided written informed consent.
Using a targeted, quantitative tandem flow injection mass spectrometry-based approach, we determined levels of 45 acylcarnitines and 15 amino acids in both study populations. Proteins were first removed by precipitation with methanol; aliquoted supernatants were dried and esterified with hot acidic methanol (acylcarnitines) and n-butanol (amino acids). For the analysis, we used tandem mass spectrometry with a Quattro Micro instrument (Waters Corp., Milford, Massachusetts), and the addition of internal standards enabled quantitative assessment of metabolites. Testing for all of the assays was done in random batch order by the Metabolomics/Biomarker Core Laboratory of the Duke Molecular Physiology Institute at Duke University; testing personnel were blinded to the clinical status of patients, and samples were randomly distributed without knowledge of event status.
In the HF-ACTION cohort, the associations of metabolite component factors with peak Vo2 and with clinical outcomes were assessed. Paralleling outcomes from the main clinical trial, the primary clinical outcome was the composite variable of all-cause mortality or all-cause hospitalization; secondary clinical outcomes included all-cause hospitalization, cardiovascular death or cardiovascular hospitalization, or cardiovascular death or heart failure exacerbation.
In the HF-ACTION cohort, principal components analysis (PCA) with varimax rotation was used to reduce the large number of correlated metabolites into uncorrelated factors, as we have done previously (6,11). Multiple regressions were used to evaluate the association of baseline PCA-derived factor levels with baseline peak Vo2, and Cox proportional hazards regression modeling was used to assess the relation between factors and clinical outcomes. Models were conducted in 3 stages: 1) unadjusted; 2) adjusted for age, sex, and body mass index (BMI); and 3) adjusted for all known predictors of each outcome, which had been previously identified in the full HF-ACTION cohort (12). In each model, we considered all factors simultaneously and used stepwise selection to select the set of significant factors. For the peak Vo2 model, covariates were age, sex, race, region, BMI, diabetes, peripheral vascular disease, New York Heart Association functional class, LVEF, ventricular conduction, and test modality. Covariates included in clinical outcomes models were age, sex, race, geographic region, LVEF, blood urea nitrogen, presence of severe mitral regurgitation, medications, symptom scores, and measures from the baseline cardiopulmonary exercise testing. The proportional hazards assumption was checked for significant factors in each Cox model, and was found to be met in all cases. Kaplan-Meier methods were used to generate time-to-event curves for significant metabolite factors. Baseline characteristics and individual metabolites were compared between the HF-ACTION and end-stage HF LVAD groups using Pearson chi-square tests for categorical variables and Wilcoxon rank sum tests for continuous variables. The authors had full access to and take full responsibility for the integrity of the data. All analyses were performed with SAS version 9.2 (SAS Institute Inc., Cary, North Carolina) and R 2.15.3 (R Development Core Team, Vienna, Austria). A p value ≤0.05 was considered statistically significant for all analyses.
Baseline patient characteristics
Per the baseline characteristics of the chronic HF (HF-ACTION) and end-stage HF (LVAD) patient populations (Table 1), the median age was 59 years in the chronic HF group and 68 years in the end-stage HF group. The patients were similarly distributed with regard to sex, race, BMI, and comorbidities (including hypertension, hyperlipidemia, and diabetes), as well as objective laboratory measures. The groups differed in objective measures of cardiovascular fitness: peak Vo2 of 14.3 ml/kg/min in chronic HF compared with 12.5 ml/kg/min in end-stage HF; Ve-VCO2 slope of 32.3 in chronic HF compared with 42.1 in end-stage HF; and N-terminal pro–B-type natriuretic peptide of 823 ng/l in chronic HF and 3,108 ng/l in end-stage HF. LVEF, as ascertained by echocardiogram, was not statistically significantly different: 25% in chronic HF versus 20% in end-stage HF. Baseline characteristics of the subset of HF-ACTION patients used for this study were not substantially different from the overall cohort (Online Table 1).
Baseline metabolomic factors
PCA identified 13 metabolite factors grouping in biologically consistent pathways (Online Table 2) similar to our previous studies (5,13). In the original HF-ACTION trial, baseline peak Vo2 was the most significant predictor of mortality in this population (chi-square = 153) and we sought to determine if there was a similar association with metabolite profiles. As shown in Table 2, factor 1 (medium-chain acylcarnitines), factor 2 (long-chain dicarboxylacylcarnitines), factor 4 (branched amino acids and related catabolites), factor 5 (long-chain acylcarnitines), and factor 8 (short-chain dicarboxylacylcarnitines) were all associated with peak Vo2 in the fully adjusted model.
Table 3 shows the association of principal component factors with clinical outcomes. For the fully adjusted model, factors 5 (long-chain acylcarnitines) and 7 (medium-chain acylcarnitines) were associated with an increase in risk of the primary outcome of all-cause mortality or hospitalization (hazard ratio [HR]: 1.24; 95% confidence interval [CI]: 1.09 to 1.42 and HR: 1.16; 95% CI: 1.02 to 1.31, respectively). Factor 9 (amino acids) was associated with decreased risk of the primary outcome (HR: 0.88; 95% CI: 0.78 to 0.99). With regard to secondary outcomes, factor 5 was associated with a greater risk of all-cause-hospitalization (HR: 1.42; 95% CI: 1.16 to 1.74), and cardiovascular death or cardiovascular hospitalization (HR: 1.22; 95% CI: 1.06 to 1.39). Of note, neither factor 7 nor 9 has associations with risk of hospitalization or cardiovascular death. Figure 1 shows Kaplan-Meier curves of the association between tertiles of factor 5 for the primary endpoint of the trial; the highest tertile showed a greater rate of death or hospitalization compared with the middle and lowest tertiles.
Given the independent association between factor 5 and peak Vo2 as well as all clinical outcomes, we examined whether its constituent metabolites (i.e., metabolites with the highest factor load in the factor) changed significantly with LVAD support (Figure 2). As shown, C16, C18:1, and C18:2 were significantly higher at baseline in patients with end-stage HF prior to LVAD placement and decreased after support. The other major components of factor 5—arginine, C18, and C20:4—did not change significantly with LVAD support.
This study examined the association of baseline metabolomic profiles with measures of cardiorespiratory fitness and clinical outcomes in 453 ambulatory patients with chronic systolic HF. We found a metabolite factor (factor 5), composed mostly of long-chain acylcarnitines, that was independently associated with lower peak Vo2 as well as both the primary and secondary clinical endpoints of the parent trial (HF-ACTION). Of the major components of this factor, levels of C16, C18:1, and C18:2 acylcarnitines were significantly higher in patients with end-stage HF prior to LVAD implantation but decreased with circulatory support. The direction of this improvement would have predicted better outcomes in HF-ACTION. This pattern of metabolite abnormalities suggested impaired mitochondrial FA oxidation, a finding previously described in HF. This further confirms that HF is characterized by dysfunction in a central pathway of energy utilization by the heart and peripheral musculature, that the measured abnormalities possess prognostic importance, and that use of mitochondrial-based therapeutics might hold promise in treating chronic systolic HF (3,14).
The mammalian heart has a unique ability to switch between fuel sources to adapt to changing physiological or dietary conditions—so-called metabolic flexibility. Healthy myocardium primarily meets its requirements for energy through the oxidation of long-chain fatty acids (LCFA), where carnitine plays a key role as a carrier (15) (Central Illustration). LCFAs are activated by esterification to coenzyme A (CoA) at the outer mitochondrial membrane. The inner mitochondrial membrane is impermeable to the acyl-CoA esters. The “carnitine shuttle,” which regulates the flux of acyl-CoA esters into the mitochondria requires the use of 3 major proteins: carnitine palmitoyltransferase (CPT) I, carnitine-acylcarnitine translocase (CACT), and CPT II.
Many clinical and experimental studies have demonstrated that the failing heart undergoes metabolic remodeling and develops a metabolic inflexibility, switching to glucose utilization at the expense of FA oxidation (3). Although the molecular changes underlying the change in fuel utilization are complex and still incompletely understood, mitochondrial dysfunction remains a common pathological theme (16). In our study, we observed that elevated plasma levels of key long-chain acylcarnitines (C16 and C18) were independently associated with impaired cardiorespiratory capacity and also increased risk of all adverse clinical outcomes. These C16 and C18 acylcarnitines are derivatives of the most abundant dietary fatty acids, palmitate and oleate, respectively. Patients with end-stage as compared with chronic systolic HF demonstrated significantly higher levels of C16 and C18, which decreased with LVAD support. These findings are consistent with the notion that the syndrome of HF may be characterized by a general state of metabolic inflexibility and mitochondrial inefficiency that leads to accumulation of metabolic intermediates of FA oxidation such as the long-chain acylcarnitines. Moreover, our findings indicate that these metabolic changes have distinct prognostic implications.
The precise mechanisms underlying the association of these long-chain acylcarnitines is unclear. One explanation may derive from abnormalities seen in rare Mendelian disorders. Irregularities in plasma levels of these molecules are characteristic of disorders of the carnitine shuttle, specifically, CPT II and CACT deficiencies, both of which are associated with skeletal and cardiac myopathy (17). CPT II deficiency can present with extreme phenotypic variability, with muscle weakness and cardiomyopathy being common associated findings. A far rarer and lethal disease, CACT deficiency results in early infant demise from skeletal muscle damage and cardiomyopathy. Our findings suggest that chronic systolic HF recapitulates a milder form of these basic metabolic defects, that the degree of these defects worsen in patients with more advanced disease, and that they may be reversible with cardiac support.
In addition to defects in the carnitine shuttle leading to mitochondrial dysfunction and impaired FA oxidation, the accumulation of circulating long-chain acylcarnitines may reflect a shift toward increased myocardial glucose oxidation with down-regulation of fatty acid oxidation that has been described in the progression of HF. To support this hypothesis, a recent study in murine models of HF demonstrated progressive down-regulation of genes involved in myocyte FA oxidation and transport and a coordinated increase in myocardial long-chain acylcarnitine species that marked the transition from compensated hypertrophy to HF (18). Moreover, increased rates of FA oxidation may produce a “bottleneck” of substrate flux into the Krebs cycle, leading to the accumulation of oxidative intermediates such as acylcarnitines, mitochondrial dysfunction, and the depletion of adenosine triphosphate needed for contractile function (19). The possibility also exists that these changes in plasma metabolites reflect impaired peripheral metabolism. Another recent animal study demonstrated that increased skeletal muscle mitochondrial efficiency, as reflected by more complete FA oxidation, may underlie changes in exercise capacity (peak Vo2) (20). Thus, the changes in plasma metabolites observed in our study may indicate impaired peripheral fatty acid oxidation and thus impaired peak Vo2, which is a major predictor of adverse outcomes in individuals with heart failure. Overall, more work utilizing metabolic flux techniques in cellular and animal models may help to unravel the mechanisms that link elevated circulating acylcarnitines and HF.
Although our primary hypothesis was that elevations in long-chain acylcarnitine levels simply signal mitochondrial dysfunction, there is a distinct possibility that they may also contribute to disease progression. Studies have demonstrated that cardiac myocytes exposed to hypoxia exhibit rapid accumulations in long-chain acylcarnitines, and these amphiphilic molecules have been shown to inhibit excitatory sodium currents in vitro (21). Furthermore, long-chain acylcarnitines increase calcium efflux in a concentration-dependent manner in isolated cardiac sarcoplasmic reticulum vesicles (21). This might predispose HF patients to malignant arrhythmias (22). Long-chain acylcarnitines are also associated with insulin resistance; these reside in cell membranes where they can potentially interfere with insulin signaling directly within the cell membrane (13). This might explain the noted state of insulin resistance seen in patients with HF (23). If true, these data will provide additional support for developing therapeutic strategies to improve these functions in the myocardium. Furthermore, there may be a role for testing the efficacy of currently available mitochondrial-based therapies such as mitoprotective agents and L-carnitine supplementation in HF (24).
The improvement with LVAD support we noted indicates that these molecules may also play a role in monitoring the efficacy of current and novel therapeutics in HF (25). Several strategies have been proposed for specific molecular targets for modifying mitochondrial function: micronutrient supplementation, increasing mitochondrial biogenesis, decreasing production of reactive oxygen species, and improvement of cellular iron homeostasis (24,26,27). Of particular interest: a novel class of compounds that selectively target cardiolipin on the inner mitochondrial membrane to optimize efficiency of the electron transport chain and thereby restore cellular bioenergetics (9,28). A key next step might be to evaluate changes in cardiorespiratory capacity and metabolite profiles as biomarkers of response during trials of these and other agents in clinical trials.
Our patient population was a subset of the total HF-ACTION study; however, there were no meaningful differences in key baseline characteristics between these patients and the overall trial. The LVAD patient population was relatively small and was from a single center. These results should be confirmed in another similar population. Our current study should therefore be considered hypothesis-generating; further studies are required to confirm the findings. Last, we report on peripheral metabolite profiles and therefore cannot assume that they represent myocardial metabolism, but rather reflect global changes in metabolism.
We found that greater circulating levels of long-chain acylcarnitines were independently predictive of functional status and mortality in patients with chronic systolic HF. The abnormalities were modifiable with LVAD support in end-stage HF patients. These findings suggest a potentially novel way to prognosticate and manage HF patients in clinical practice while providing an impetus for pharmacological targeting of the mitochondria for treatment of HF.
COMPETENCY IN MEDICAL KNOWLEDGE: Circulating levels of long-chain acylcarnitine metabolite in patients with HF are associated with adverse clinical outcomes.
TRANSLATIONAL OUTLOOK: The metabolic processes that regulate blood levels of mitochondrial FA metabolites are potential targets for diagnostic or therapeutic modalities in patients with worsening HF.
For supplemental tables, please see the online version of this article.
This study was performed with a grant from the Daland Fellowship in Clinical Investigation. The HF-ACTION study was funded by grants from the National Heart, Lung, and Blood Institute. Dr. Ahmad has received consulting fees from Roche. Dr. Patel has served as a consultant for Thoratec Corp. and HeartWare Inc. Dr. Milano has served as a consultant for HeartWare, Inc. Dr. Felker has received grant support from Roche Diagnostics; and has served as a consultant for Singulex. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Evelyn Horn, MD, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- fatty acid
- heart failure
- left ventricular assist device
- principal components analysis
- Received June 8, 2015.
- Revision received September 7, 2015.
- Accepted October 22, 2015.
- American College of Cardiology Foundation
- Ahmad T.,
- Wang T.,
- O'Brien E.C.,
- et al.
- Fillmore N.,
- Lopaschuk G.D.
- Shah S.H.,
- Bain J.R.,
- Muehlbauer M.J.,
- et al.
- Shah S.H.,
- Kraus W.E.,
- Newgard C.B.
- Cheng M.L.,
- Wang C.H.,
- Shiao M.S.,
- et al.
- Dai W.,
- Shi J.,
- Gupta R.C.,
- Sabbah H.N.,
- Hale S.L.,
- Kloner R.A.
- Shah S.H.,
- Newgard C.B.
- Schooneman M.G.,
- Vaz F.M.,
- Houten S.M.,
- Soeters M.R.
- McGarrah R.W.,
- Ahmad T.,
- Koeberl D.D.,
- Patel C.B.
- Stanley W.C.,
- Recchia F.A.,
- Lopaschuk G.D.
- Longo N.,
- Amat di San Filippo C.,
- Pasquali M.
- Lai L.,
- Leone T.C.,
- Keller M.P.,
- et al.
- Kalim S.,
- Clish C.B.,
- Wenger J.,
- et al.
- Bonnet D.,
- Martin D.,
- Pascale De L.,
- et al.
- Ashrafian H.,
- Frenneaux M.P.,
- Opie L.H.
- Soukoulis V.,
- Dihu J.B.,
- Sole M.,
- et al.
- Bayeva M.,
- Gheorghiade M.,
- Ardehali H.
- Dai D.F.,
- Hsieh E.J.,
- Chen T.,
- et al.
- Dai D.F.,
- Chen T.,
- Szeto H.,
- et al.