[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Table 1.  
Baseline Characteristics of the Patients
Baseline Characteristics of the Patients
Table 2.  
Changes in Metabolite Levels After 24 Weeks With Sildenafil Treatment
Changes in Metabolite Levels After 24 Weeks With Sildenafil Treatment
Table 3.  
Multiple Linear Regression for Metabolite Association With Composite Clinical Rank Score Adjusted for Age, Sex, Race/Ethnicity, Smoking Status, New York Heart Association Class, and Treatment Groups
Multiple Linear Regression for Metabolite Association With Composite Clinical Rank Score Adjusted for Age, Sex, Race/Ethnicity, Smoking Status, New York Heart Association Class, and Treatment Groups
1.
Zouein  FA, de Castro Brás  LE, da Costa  DV, Lindsey  ML, Kurdi  M, Booz  GW.  Heart failure with preserved ejection fraction: emerging drug strategies.  J Cardiovasc Pharmacol. 2013;62(1):13-21.PubMedGoogle ScholarCrossref
2.
Redfield  MM, Chen  HH, Borlaug  BA,  et al; RELAX Trial.  Effect of phosphodiesterase-5 inhibition on exercise capacity and clinical status in heart failure with preserved ejection fraction: a randomized clinical trial.  JAMA. 2013;309(12):1268-1277.PubMedGoogle ScholarCrossref
3.
Shah  SH, Hauser  ER, Bain  JR,  et al.  High heritability of metabolomic profiles in families burdened with premature cardiovascular disease.  Mol Syst Biol. 2009;5(1):258.PubMedGoogle Scholar
4.
Benjamini  Y, Hochberg  Y.  Controlling the false discovery rate: a practical and powerful approach to multiple testing.  J R Stat Soc B. 1995;57(1):289-300. doi:10.2307/2346101Google Scholar
5.
Gottlieb  SS, Harris  K, Todd  J,  et al.  Prognostic significance of active and modified forms of endothelin 1 in patients with heart failure with reduced ejection fraction.  Clin Biochem. 2015;48(4-5):292-296.PubMedGoogle ScholarCrossref
6.
Koves  TR, Ussher  JR, Noland  RC,  et al.  Mitochondrial overload and incomplete fatty acid oxidation contribute to skeletal muscle insulin resistance.  Cell Metab. 2008;7(1):45-56.PubMedGoogle ScholarCrossref
7.
Muoio  DM, Noland  RC, Kovalik  J-P,  et al.  Muscle-specific deletion of carnitine acetyltransferase compromises glucose tolerance and metabolic flexibility.  Cell Metab. 2012;15(5):764-777.PubMedGoogle ScholarCrossref
8.
Shah  SH, Kraus  WE, Newgard  CB.  Metabolomic profiling for the identification of novel biomarkers and mechanisms related to common cardiovascular diseases: form and function.  Circulation. 2012;126(9):1110-1120.PubMedGoogle ScholarCrossref
9.
Shah  SH, Sun  J-L, Stevens  RD,  et al.  Baseline metabolomic profiles predict cardiovascular events in patients at risk for coronary artery disease.  Am Heart J. 2012;163(5):844-850.e1.PubMedGoogle ScholarCrossref
10.
Kraus  WE, Muoio  DM, Stevens  R,  et al.  Metabolomic quantitative trait loci (mQTL) mapping implicates the ubiquitin proteasome system in cardiovascular disease pathogenesis.  PLoS Genet. 2015;11(11):e1005553.PubMedGoogle ScholarCrossref
11.
Zordoky  BN, Sung  MM, Ezekowitz  J,  et al; Alberta HEART.  Metabolomic fingerprint of heart failure with preserved ejection fraction.  PLoS One. 2015;10(5):e0124844.PubMedGoogle ScholarCrossref
Views 645
Citations 0
Brief Report
August 2017

Sildenafil Treatment in Heart Failure With Preserved Ejection FractionTargeted Metabolomic Profiling in the RELAX Trial

Author Affiliations
  • 1Duke Molecular Physiology Institute, Duke University School of Medicine, Durham, North Carolina
  • 2Department of Biostatistics and Bioinformatics, Duke University School of Medicine, Durham, North Carolina
  • 3Duke Clinical Research Institute, Duke University School of Medicine, Durham, North Carolina
  • 4Division of Cardiology, Department of Medicine, Duke University School of Medicine, Durham, North Carolina
  • 5Department of Medicine, Mayo Clinic, Rochester, Minnesota
JAMA Cardiol. 2017;2(8):896-901. doi:10.1001/jamacardio.2017.1239
Key Points

Question  What are the underlying mechanisms and biomarkers associated with sildenafil treatment in the PhosphodiesteRasE-5 Inhibition to Improve Clinical Status and Exercise Capacity in Diastolic Heart Failure with Preserved Ejection Fraction (RELAX) trial, which compared sildenafil with a placebo for treating patients with heart failure with preserved ejection fraction?

Findings  In this metabolomic profiling study using samples from the RELAX trial, levels of circulating amino acids, short-chain dicarboxylacylcarnitines, and long-chain acylcarnitines increased significantly with sildenafil but not with a placebo among patients with heart failure with preserved ejection fraction. Higher baseline levels of short-chain dicarboxylacylcarnitine metabolite 3-hydroxyisovalerylcarnitine/malonylcarnitine and asparagine/aspartic acid were associated with worse clinical rank scores in groups treated with sildenafil and a placebo.

Meaning  Circulating metabolites may contribute to the underlying mechanisms associated with sildenafil and serve as biomarkers for outcomes in heart failure with preserved ejection fraction.

Abstract

Importance  Phosphodiesterase-5 inhibition with sildenafil compared with a placebo had no effect on the exercise capacity or clinical status of patients with heart failure with preserved ejection fraction (HFpEF) in the PhosphodiesteRasE-5 Inhibition to Improve Clinical Status and Exercise Capacity in Diastolic Heart Failure with Preserved Ejection Fraction (RELAX) clinical trial. Metabolic impairments may explain the neutral results.

Objective  To test the hypothesis that profiling metabolites in the RELAX trial would clarify the mechanisms of sildenafil effects and identify metabolites associated with clinical outcomes in HFpEF.

Design, Setting, and Participants  Paired baseline and 24-week plasma samples of 160 stable outpatient individuals with HFpEF enrolled in the RELAX clinical trial were analyzed using flow injection tandem mass spectrometry (60 metabolites) and conventional assays (5 metabolites).

Interventions  Sildenafil (n = 79) or a placebo (n = 81) administered orally at 20 mg, 3 times daily for 12 weeks, followed by 60 mg, 3 times daily for 12 weeks.

Main Outcomes and Measures  The primary measure was metabolite level changes between baseline and 24 weeks stratified by treatments. Secondary measures included correlations between metabolite level changes and clinical biomarkers and associations between baseline metabolite levels and the composite clinical score.

Results  No metabolites changed between baseline and 24 weeks in the group treated with a placebo; however, 7 metabolites changed in the group treated with sildenafil, including decreased amino acids (alanine and proline; median change [25th-75th], −38.26 [−100.3 to 28.19] and −28.24 [−56.29 to 12.08], respectively; false discovery rate–adjusted P = .01 and .03, respectively), and increased short-chain dicarboxylacylcarnitines glutaryl carnitine, octenedioyl carnitine, and adipoyl carnitine (median change, 6.19 [−3.37 to 14.18], 2.72 [−3 to 12.57], and 10.72 [−11.23 to 29.57], respectively; false discovery rate–adjusted P = .01, .04, and .05, respectively), and 1 long-chain acylcarnitine metabolite (palmitoyl carnitine; median change, 7.83 [−5.64 to 26.99]; false discovery rate–adjusted P = .03). The increases in long-chain acylarnitine metabolites and short-chain dicarboxylacylcarnitines correlated with increases in endothelin-1 and creatinine/cystatin C, respectively. Higher baseline levels of short-chain dicarboxylacylcarnitine metabolite 3-hydroxyisovalerylcarnitine/malonylcarnitine and asparagine/aspartic acid were associated with worse clinical rank scores in both treatment groups (β, −96.60, P = .001 and β, −0.02, P = .01; after renal adjustment, P = .09 and .02, respectively).

Conclusions and Relevance  Our study provides a potential mechanism for the effects of sildenafil that, through adverse effects on mitochondrial function and endoplasmic reticulum stress, could have contributed to the neutral trial results in RELAX. Short-chain dicarboxylacylcarnitine metabolites and asparagine/aspartic acid could serve as biomarkers associated with adverse clinical outcomes in HFpEF.

Introduction

Heart failure with preserved ejection fraction (HFpEF) is a complex syndrome associated with many diagnostic and therapeutic challenges.1 A recent clinical trial, Phosphodiesterase-5 Inhibition to Improve Clinical Status and Exercise Capacity in Heart Failure with Preserved Ejection Fraction (RELAX),2 did not demonstrate improvements in exercise capacity or clinical status of patients with HFpEF who were treated with sildenafil compared with participants treated with a placebo. In this study, we performed quantitative, targeted metabolomic profiling in RELAX samples to examine the underlying mechanisms modified by sildenafil that could have mediated the neutral trial results and to identify biomarkers associated with clinical outcomes.

Methods
Study Cohort

The RELAX trial2 was a multicenter, double-blind trial conducted through the Heart Failure Clinical Research Network from 2008 to 2012 on outpatients with HFpEF who were randomized to be treated with a placebo (n = 103) or with sildenafil (n = 113) for 24 weeks. The study was approved by the institutional review board at all participating sites, and all participants provided informed written consent. The primary trial end point was a change in peak oxygen consumption (change in oxygen consumption = 24-week−baseline oxygen consumption). Secondary end points included changes in 6-minute walk distance, a composite clinical rank score (based on time to death and cardiovascular or cardiorenal hospitalization), and clinical biomarkers (creatinine, cystatin C, N-terminal pro-B-type natriuretic peptide, and endothelin-1). The current biomarker substudy had 160 paired plasma samples available (79 samples from patients treated with sildenafil, 81 samples from patients treated with a placebo).

Metabolomic Profiling

Plasma metabolomic profiling (45 acylcarnitines, 15 amino acids, 5 conventional) was conducted using tandem flow injection mass spectrometry (Acquity TOD Triple Quadrupole; Waters) with stable-isotope–labeled internal standards3 and by conventional means3 (eMethods and eTable 1 in the Supplement).

Statistical Analyses

Baseline patient characteristics and metabolite levels between treatment groups were compared using the Wilcoxon rank sum test (continuous variables) and the χ2/Fisher exact test (categorical variables). An exploratory analysis of changes in metabolite levels was first performed using the Wilcoxon signed rank test, stratified by treatment received. A linear regression model was then used with changes in metabolite regressed on treatment received and adjusted for age, race/ethnicity, sex, baseline metabolite level, and creatinine. Correlations between changes in metabolite levels and clinical biomarkers were assessed (Spearman coefficient) for the full cohort and by treatment received. Baseline metabolite levels were assessed for association with the change in oxygen consumption, the change in 6-minute walk distance, and the composite clinical score using multivariable linear regression adjusted for age, sex, race/ethnicity, smoking status, the New York Heart Association class, and the treatment received. Sensitivity analyses adjusted for baseline creatinine levels were also performed. A 2-tailed α of .05 was used for statistical significance, and the false discovery rate controlling procedure was used to adjust for testing multiple hypotheses 4 for paired analyses. The analyses of clinical outcomes and correlations were unadjusted for multiple comparisons. All statistical analyses were conducted using R, version 3.2 (The R Foundation).

Results

Baseline clinical characteristics and metabolite profiles did not differ significantly between treatment groups, except for a lower prevalence of hypertension among the group treated with sildenafil, which was similar to results of the full trial2 (Table 1, eTable 2 in the Supplement).

No metabolites changed significantly with treatment with a placebo (eTable 3 in the Supplement). The levels of 7 metabolites changed significantly with sildenafil treatment (false discovery rate–adjusted): alanine, proline, and lactate decreased; and 3 short-chain dicarboxylacylcarnitines (SCDAs) (glutaryl carnitine, octenedioyl carnitine, and adipoyl carnitine) and 1 long-chain acylcarnitine (LCAC), palmitoyl carnitine, increased (Table 2). Consistent with the full trial,2 our analysis also showed an association between increased clinical biomarkers creatinine, cystatin C, and endothelin-1 and sildenafil treatment (Table 2). In the multivariable regression adjusted for age, sex, race/ethnicity, and baseline metabolite level and creatinine, alanine and proline changes were negatively associated with treatment, while changes in SCDA and LCAC levels were positively associated (eTable 4 in the Supplement). Correlations between changes in metabolites and clinical biomarkers were most significant between alanine and lactate (Spearman rank correlation coefficient, 0.52; P = 1.1 × 10−10); SCDA levels (glutaryl carnitine, methylmalonyl carnitine/succinyl carnitine, 3-hydroxyisovalerylcartine/malonylcarnitine, octenedioyl carnitine, 3-hydroxy-cis-5-octenoyl carnitine) and creatinine/cystatin C, and LCACs and endothelin-1 (eFigure and eTable 5 in the Supplement).

Similar to the full trial, our analyses did not reveal any difference in the primary (the change in oxygen consumption) or secondary end points (the change in 6-minute walk distance and composite clinical score) between treatment groups (P = .80, .98, and .75, respectively). Higher levels of 7 baseline metabolites were associated with worse clinical scores: SCDA 3-hydroxyisovalerylcartine/malonyl carnitine, methylmalonyl carnitine/succinyl carnitine, and glutaryl carnitine (β, −96.60, −50.24, and −31.35, respectively; P = .001, .02, and .03, respectively); short-chain acylcarnitines butyryl carnitine/isobutyryl carnitineand tigyl carnitine (β, −5.60 and −85.59; P = .02); and amino acids asparagine/aspartic acid and citrulline (β, −0.02 and −0.08; P = .01 and .03, respectively) (Table 3). After adjusting for baseline creatinine, asparagine/aspartic acid remained significantly associated with the clinical score (P = .02), while 3-hydroxyisovalerylcartine/malonylcarnitine showed a trend toward association (P = .09).

Discussion

Using targeted metabolomic profiling in paired samples from the RELAX trial, we found circulating metabolites among patients with HFpEF that were modified by treatment with sildenafil: decreased alanine and proline, and increased LCACs and SCDAs. The increases in LCAC and SCDA metabolites correlated with worsening collagen metabolism biomarkers (endothelin-1) and renal function (creatinine and cystatin C) biomarkers, respectively. Additionally, baseline levels of the SCDA metabolite 3-hydroxyisovalerylcartine/malonylcarnitine and asparagine/aspartic acid were associated with a worse composite clinical rank score at 24 weeks.

Long-chain acylcarnitine metabolites have been implicated in mitochondria-mediated inflammation and cellular stress promotion. In this study, LCAC metabolite levels increased with sildenafil treatment and exhibited a significant correlation with endothelin-1, an established prognostic biomarker for mortality and morbidity in heart failure.5 While the exact mechanism of sildenafil-associated elevation in LCAC metabolites is unclear, our findings suggest that the role of mitochondrial dysfunction and alterations in the endothelin-1 signaling pathway are potential explanations.

The accumulation of SCDA metabolites reports on dysregulated endoplasmic reticulum stress of the ubiquitin-proteasome system, which induces an autophagy to remove damaged cells.6,7 In this study, increased SCDAs by sildenafil treatment may be explained by this pathway activating, which could lead to a disrupted cellular homeostasis, dysregulated autophagy, and worse long-term outcomes. This negative consequence may have contributed to the lack of benefits received by those treated with sildenafil therapy.

In this study, higher baseline levels of the SCDA 3-hydroxyisovalerylcartine/malonylcarnitine and amino acids asparagine/aspartic acid were associated with worse clinical scores. The SCDA result is consistent with previous studies that demonstrated an association between SCDA metabolites and cardiovascular events.8,9 Notably, this association was attenuated after adjusting for baseline renal function, which could be because of an impaired clearance of SCDAs with impaired renal function or an underlying pathophysiology affecting renal function and SCDA metabolism concordantly. In prior studies, SCDAs were associated with incident cardiovascular events even after adjusting for baseline estimated glomerular filtration rates,10 suggesting that a complex relationship exists between SCDAs, renal function, and cardiovascular disease. The amino acid result is consistent with recent studies showing higher levels of asparagine/aspartic acid in HFpEF compared with patients who had not experienced heart failure,11 highlighting the importance of underlying alterations in amino acid metabolism in HFpEF. The internal standards used to quantify asparagine/aspartic acid do not discriminate these 2 metabolites, and thus we were unable to determine the specific biologic pathway represented.

Limitations

This study has several limitations. Patients were selected who could perform the peak oxygen consumption test, which may have introduced biases. Additionally, our results of metabolite association with clinical end points were attenuated when adjusting for baseline renal function, suggesting that a complex relationship exists between SCDAs, renal function, and cardiovascular disease.

Conclusions

Among patients with HFpEF, phosphodiesterase type 5 inhibition with sildenafil treatment resulted in significantly increased LCAC and SCDA metabolites that paralleled increases in adverse clinical markers. These results suggest that there is a potential role of mitochondrial dysfunction and endoplasmic reticulum stress among patients with HFpEF who are treated with sildenafil and provide potential candidates for biomarkers associated with clinical outcomes.

Back to top
Article Information

Corresponding Author: Svati H. Shah, MD, MS, MHS, Duke Molecular Physiology Institute, 300 N Duke St, Durham NC 27701 (svati.shah@duke.edu).

Accepted for Publication: March 17, 2017.

Published Online: May 10, 2017. doi:10.1001/jamacardio.2017.1239

Author Contributions: Drs Wang and Shah had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Wang, Bain, Hernandez, Felker, Redfield, Shah.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Wang.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Wang, Anstrom, Shah.

Obtained funding: Hernandez, Shah.

Administrative, technical, or material support: Newgard, Hernandez, Shah.

Supervision: Newgard, Felker, Shah.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Newgard is a member of the Pfizer CVMED scientific advisory board. Dr Hernandez reports receiving consultant fees/honoraria from BMS, Boston Scientific, Gilead, Janssen, and Novartis. Dr Felker reports receiving consultant fees/honoraria from Amgen, Novartis, Trevena, Singulex, and Medtronic. Dr Redfield reports receiving royalties from Annexon. Dr Shah reports holding a patent on a related finding. No other disclosures were reported.

Funding/Support: Funding support was provided by grants HL127009 and T32HL007101 (principal investigator [PI], Dr Shah) from the National Institutes of Health and the American Heart Association Strategically Focused Research Network Heart Failure Grant (PIs, Drs Shah, Felker, and Hernandez).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

References
1.
Zouein  FA, de Castro Brás  LE, da Costa  DV, Lindsey  ML, Kurdi  M, Booz  GW.  Heart failure with preserved ejection fraction: emerging drug strategies.  J Cardiovasc Pharmacol. 2013;62(1):13-21.PubMedGoogle ScholarCrossref
2.
Redfield  MM, Chen  HH, Borlaug  BA,  et al; RELAX Trial.  Effect of phosphodiesterase-5 inhibition on exercise capacity and clinical status in heart failure with preserved ejection fraction: a randomized clinical trial.  JAMA. 2013;309(12):1268-1277.PubMedGoogle ScholarCrossref
3.
Shah  SH, Hauser  ER, Bain  JR,  et al.  High heritability of metabolomic profiles in families burdened with premature cardiovascular disease.  Mol Syst Biol. 2009;5(1):258.PubMedGoogle Scholar
4.
Benjamini  Y, Hochberg  Y.  Controlling the false discovery rate: a practical and powerful approach to multiple testing.  J R Stat Soc B. 1995;57(1):289-300. doi:10.2307/2346101Google Scholar
5.
Gottlieb  SS, Harris  K, Todd  J,  et al.  Prognostic significance of active and modified forms of endothelin 1 in patients with heart failure with reduced ejection fraction.  Clin Biochem. 2015;48(4-5):292-296.PubMedGoogle ScholarCrossref
6.
Koves  TR, Ussher  JR, Noland  RC,  et al.  Mitochondrial overload and incomplete fatty acid oxidation contribute to skeletal muscle insulin resistance.  Cell Metab. 2008;7(1):45-56.PubMedGoogle ScholarCrossref
7.
Muoio  DM, Noland  RC, Kovalik  J-P,  et al.  Muscle-specific deletion of carnitine acetyltransferase compromises glucose tolerance and metabolic flexibility.  Cell Metab. 2012;15(5):764-777.PubMedGoogle ScholarCrossref
8.
Shah  SH, Kraus  WE, Newgard  CB.  Metabolomic profiling for the identification of novel biomarkers and mechanisms related to common cardiovascular diseases: form and function.  Circulation. 2012;126(9):1110-1120.PubMedGoogle ScholarCrossref
9.
Shah  SH, Sun  J-L, Stevens  RD,  et al.  Baseline metabolomic profiles predict cardiovascular events in patients at risk for coronary artery disease.  Am Heart J. 2012;163(5):844-850.e1.PubMedGoogle ScholarCrossref
10.
Kraus  WE, Muoio  DM, Stevens  R,  et al.  Metabolomic quantitative trait loci (mQTL) mapping implicates the ubiquitin proteasome system in cardiovascular disease pathogenesis.  PLoS Genet. 2015;11(11):e1005553.PubMedGoogle ScholarCrossref
11.
Zordoky  BN, Sung  MM, Ezekowitz  J,  et al; Alberta HEART.  Metabolomic fingerprint of heart failure with preserved ejection fraction.  PLoS One. 2015;10(5):e0124844.PubMedGoogle ScholarCrossref
×