Heart Failure in the Era of Precision Medicine: A Scientific Statement From the American Heart Association

Authors, Journal, Affiliations, Type, DOI

• Authors: Sharon Cresci (Chair), Naveen L. Pereira (Vice Chair), Ferhaan Ahmad, Mirnela Byku, Lisa de las Fuentes, David E. Lanfear, Carolyn M. Reilly, Anjali T. Owens, Matthew J. Wolf; on behalf of the AHA Council on Genomic and Precision Medicine
• Journal: Circulation: Genomic and Precision Medicine
• Type: AHA Scientific Statement
• DOI: 10.1161/HCG.0000000000000058
• Published: October 2019

Overview

This AHA Scientific Statement synthesises six "omics" domains as they relate to heart failure precision medicine: genomics, pharmacogenomics, epigenomics, proteomics, metabolomics, and microbiomics. One in five individuals will develop HF in their lifetime; 6.5 million Americans currently have HF with 1 million new diagnoses annually and >30% 1-year mortality in Medicare-hospitalised patients. The statement argues that genomic, epigenetic, proteomic, metabolomic, and microbiome variability drives heterogeneity in treatment response, and that multi-omics integration — aided by machine learning — is the future of HF management.

Keywords

Heart failure, precision medicine, genomics, pharmacogenomics, epigenomics, proteomics, metabolomics, microbiomics, omics, preventive medicine, early diagnosis

Key Takeaways

Genomics of HF

HF Is a Heritable Trait

• Common/idiopathic HF heritability estimated at ~18%; arises from cumulative effects of multiple variants in multiple genes, each with small-to-modest effect sizes (vs. inherited cardiomyopathies with Mendelian inheritance).
• GWAS have identified variants in HSPB7, FRMD4B, BAG3, ZBTB17, CMTM7, USP3, and SLC25A46 associated with HF and non-familial DCM.

Clinical Genetic Testing

• Pathogenic/likely pathogenic variant detection frequency: HCM 60–70%, DCM 30–40%, ARVC 50–60%.
• In idiopathic DCM with concomitant neuromuscular disease, pathogenic variant yield rises to 62%.
• Multidisciplinary genetics clinics (cardiologist + geneticist/genetic counselor) are recommended for cardiac genetic testing.
• Variant classification: ACMG 5-category framework (pathogenic, likely pathogenic, VUS, likely benign, benign).
• Two key management-altering exceptions for genetic testing in HF probands: (1) phenocopy diagnosis (amyloid, Danon, Fabry disease) alters management; (2) LMNA pathogenic variants require enhanced ICD vigilance due to high arrhythmia risk that can precede HF symptoms.
• Major utility of genetic testing: enables cascade family screening; genotype-positive phenotype-negative individuals benefit from heightened surveillance. Negative result in family member may allow dismissal from follow-up (major cost and psychological benefit).

Genetics Insights Into Pathophysiology

• TTN: Truncating variants (TTNtv) cause 20–25% of familial DCM and a subset of peripartum CMP. Rat models show decreased titin protein levels, metabolic shift from fatty acid oxidation to glycolysis, and mTORC1 pathway activation — mirroring clinical DCM. Individuals with TTNtv may be in a compensated state until triggered by stress (pregnancy, alcohol exposure).
• RBM20: Variants cause familial and idiopathic DCM with a higher ventricular arrhythmia burden than TTN variants at equivalent LVEF. RBM20 regulates alternative splicing of TTN, CAMK2D, and other sarcomere genes. Knockout mouse cardiomyocytes show abnormal calcium handling and increased spontaneous sarcoplasmic reticulum calcium release — mechanistic basis for arrhythmia excess.
• BAG3: Mutations cause familial and idiopathic DCM. BAG3 protein (cochaperone, HSP70 ATPase domain interactor) maintains protein homeostasis, mitochondrial integrity, and myofibrillar function. Disruption → proteotoxicity → mitochondrial fragmentation → cardiomyopathy.
• LMNA: Accounts for ~8% of DCM; also seen in HCM and ARVC. Lamins A/C provide nuclear stability, chromatin organisation, gene expression regulation, and DNA damage repair. LMNA mutations → cardiac fibrosis, abnormal mechanical stress response, defective electrical signalling, impaired autophagy, and upregulation of MAPK and mTOR signalling pathways.

Pharmacogenomics of HF

Beta-Blockers

• ADRB1 Arg389Gly (BEST Trial): Homozygous Arg389 patients had 38% reduction in mortality with bucindolol; Gly389 carriers had no significant benefit. This association has not been replicated with other beta-blockers.
• GRK5 Leu41 variant: Carriers showed no mortality benefit from beta-blocker therapy — predominantly a Black patient finding (allele frequency 0.23 in Blacks vs. 0.02 in whites). GRK5 Leu41 may provide an intrinsic beta-blocker-like effect through enhanced desensitisation of cardiac beta-adrenergic receptors.
• GNB3 Thr825 (A-HeFT Trial): Homozygous Thr825 carriers had significantly greater event-free survival with fixed-dose hydralazine/isosorbide therapy vs. GNB3 Cys825 variant carriers — a pharmacogenomic predictor of H-ISDN response in African Americans.

CYP450 Metabolism

• CYP2D6 and beta-blockers: Poor metabolizers (PM) of metoprolol have heart rates 8.5 bpm lower and BP 5 mmHg lower vs. extensive metabolizers (EM); 4-fold increased bradycardia risk. Royal Dutch Pharmacists Association recommends bisoprolol (non-CYP2D6 substrate) or 50–75% dose reduction for IMs/PMs; ultrarapid metabolizers require dose increase up to 250%.
• CYP3A5 and tacrolimus (post-heart transplant): Clinical Pharmacogenetics Implementation Consortium recommends 1.5–2× higher initial tacrolimus dose for EMs/IMs; reduces transplant rejection risk (OR 1.32 for under-dosing in EMs/IMs with standard dosing). EHR-embedded preemptive genotyping facilitates implementation.

Clinical Implementation Barriers

• Lack of robust statistical associations, absent replication cohorts, and inadequate sample size in appropriately phenotyped HF cohorts.
• Emerging strategy: preemptive genotyping at enrolment, results embedded in EHR, triggered at point of prescription.

Epigenomics of HF

DNA Methylation

• CpG hypomethylation of satellite repeat elements in end-stage HF LV tissue → up to 27-fold upregulation of corresponding transcripts → chromosomal instability as an unrecognised HF progression mechanism.
• LY75 (lymphocyte antigen 75) promoter CpG hypermethylation in idiopathic DCM → decreased expression → confirmed cardiac dysfunction in zebrafish ablation model.
• 59 differentially methylated CpGs in DCM LV myocardium (30 hypomethylated, 29 hypermethylated); patterns conserved in peripheral blood — potential non-invasive epigenetic biomarkers.

Histone Modifications

• Histone acetyltransferases p300/CBP: necessary and sufficient for pathological cardiac hypertrophy via acetylation of MEF2 and GATA4.
• Class I HDAC inhibitors: attenuate cardiac fibrosis and hypertrophy in multiple animal models; clinical cardiovascular toxicity in oncology trials mostly limited to mild ECG changes.
• Class IIa HDACs (HDAC5, HDAC9): inhibit hypertrophy by repressing MEF2 transcription; export from nucleus on phosphorylation releases MEF2 → pro-hypertrophic signalling.
• Sirtuins (SIRT1, SIRT6, SIRT7): Class III HDACs, NAD⁺-dependent; protective against cardiac hypertrophy and HF at moderate expression levels. SIRT6/SIRT7 deficiency → cardiac hypertrophy, HF, and apoptosis in murine models.
• LV assist device support in DCM reverses histone demethylation patterns in cardiomyocytes.

Non-Coding RNAs (ncRNAs)

• miRNAs as biomarkers: miR-423-5p diagnostic for HF in the dyspnoea setting; miR-519e identifies high-risk patients for decompensation, transplantation, or CV death; elevated plasma miR-30d predicts CRT response better than QRS duration alone.
• Distinguishing HFrEF from HFpEF: Two studies identified distinct miRNA signatures separating HFrEF from HFpEF.
• lncRNA LIPCAR level predicts 3-year cardiovascular mortality.
• Therapeutic ncRNA modulation: miR-208a inhibition and miR-133 overexpression show preclinical efficacy. Delivery mechanisms: viral vectors, miRNA mimics, antagomirs (antimiRNAs), GapmeRs for lncRNA knockdown.

Proteomics of HF

Established Biomarkers

• Natriuretic peptides (BNP/NT-proBNP): Class I indication for HF diagnosis and prognosis; strongest biomarkers across HFpEF and HFrEF.
• sST2 (soluble ST2): Independent mortality predictor in stable and acute HF; threshold 35 pg/mL (HFrEF) / 32 pg/mL (HFpEF) discriminates high vs. low risk; uniquely unaffected by age, renal function, or BMI (unlike NPs). Serial changes over 12 months predict outcomes.
• High-sensitivity troponin (hsTnT): Correlates with hibernating myocardium amount in ischemic cardiomyopathy; predicts myocardial recovery with NT-proBNP.
• Multimarker approaches: PRIDE cohort 5-biomarker score (NT-proBNP, CRP, ST2, haemoglobin, BUN) shows direct correlation with 1-year mortality.

Biomarker-Guided Therapy

• NT-proBNP-guided HF therapy tested in GUIDE-IT (multicenter RCT, HFrEF post-hospitalisation); stopped early for futility — NT-proBNP-directed intensification did not reduce outcomes vs. standard care.
• Ongoing marker-directed trials: STADE-HF (sST2-guided), HOMAGE (galectin-guided MRA for HFpEF prevention), TROUPER (copeptin-guided vasopressin antagonist).

High-Throughput / Unbiased Proteomics

• Multiplexed approaches (Olink dual antibody-nucleotide proximity extension assay, aptamer arrays assaying >1300 proteins, mass spectrometry): enable unbiased discovery of novel HF biomarkers.
• Olink platforms identified PCSK9 and other novel prognostic biomarkers in HF cohorts.
• Integrating genomics + proteomics: genetic loci associated with interindividual variability in NT-proBNP, ceruloplasmin, and soluble IL-2Rα levels identify novel HF pathways.

Metabolomics of HF

Metabolic Substrate Shift

• Early HF: normal to slightly elevated fatty acid oxidation. Advanced HF: downregulation of fatty acid oxidation → shift to glucose utilisation (metabolic reprogramming mirrors foetal gene programme).

Predicting HF Development

• ARIC Study (Black patients, 20-year follow-up): elevated branched chain amino acid catabolites (hydroxyleucine, hydroxyisoleucine) + decreased very-long-chain PUFA (dihydroxydocasatrenoic acid) independently predict incident HF hospitalisation.
• Plasma branched chain amino acids and acylcarnitines elevated in Stage A HF vs. healthy controls.
• Plasma ceramide ratios (C24:0/C16:0) inversely associated with incident HF in FHS and SHIP cohorts.

HFrEF Metabolomic Profiles

• Elevated 3-hydroxybutyrate, acetone, and succinate in HFrEF; linearly related to metabolic energy expenditure (outcome predictor).
• Metabolite panel (histidine, phenylalanine, spermidine, phosphatidylcholine C34:4) has diagnostic value comparable to BNP and reverses with functional improvement.
• 4-metabolite prognostic panel (dimethylarginine/arginine ratio, spermidine, butyrylcarnitine, total essential amino acids) correctly classifies 85% of events vs. 74% with BNP alone.
• Ketone bodies elevated in early HF but lower in severe LV dysfunction — metabolomic profiles are stage/severity-specific.

HFpEF vs HFrEF Metabolomic Differences

• HFpEF: higher serum acylcarnitines, carnitine, creatinine, betaine, and amino acids vs. controls.
• Medium- and long-chain acylcarnitines and ketone bodies are higher in HFpEF than in HFrEF — shared underlying mechanism of dysregulated fatty acid oxidation across both phenotypes.

Clinical Translation Barriers

• Lack of standardised quantification across laboratories; unclear effects of diet and comorbidities on metabolite measurement.

Microbiomics of HF

Gut Dysbiosis in HF

• HF → reduced intestinal blood flow → intestinal acidosis and hypoxia → sodium/fluid retention, increased intestinal permeability, bacterial translocation, increased pathogenic flora (Campylobacter, Shigella, Salmonella, Yersinia enterocolitica, Candida).
• Pathogenic flora levels and intestinal permeability correlate directly with right atrial pressure and clinical HF severity.
• Bacterial endotoxin (lipopolysaccharide) entry into systemic circulation → TNF-α and IL-6 release → systemic inflammation (elevated CRP in HF).

TMAO as a Biomarker

• TMAO (trimethylamine N-oxide): Gut-derived metabolite from dietary choline/phosphatidylcholine/carnitine via intestinal bacteria → hepatic FMO3 N-oxidation → renal excretion.
• Plasma TMAO elevated in HF vs. healthy controls.
• Independently predicts in-hospital and 1-year mortality in acute HF and 5-year all-cause mortality in stable HF, even after full adjustment for cardiorenal and traditional risk factors.
• Also predicts chronic kidney disease development and severity in HF patients.

Other Uremic Toxins

• Indoxyl sulfate (from tryptophan via gut microbes + liver): in DCM patients, elevated serum indoxyl sulfate independently predicts HF hospitalisation, cardiac death, or both over 5-year follow-up (adjusted for clinical factors, BNP, eGFR).
• p-Cresyl sulfate (from phenylalanine): elevated in dysbiosis; renally cleared; contributes to accelerated HF and CKD progression.

Therapeutic Interventions

• Mediterranean diet: Reduces HF risk by 21% over 10 years (RR 0.79, P=0.004) in a large population cohort without known CVD; reduces trimethylamine-producing substrates → lower TMAO.
• Probiotics: Saccharomyces boulardii for 3 months in HFrEF → decreased creatinine, uric acid, CRP + improved LA diameter and LVEF vs. controls.
• AST-120 (oral charcoal absorbent): Selectively binds gut microbial metabolites; reduced rehospitalisation and length of stay in a small trial (n=20, HF + moderate CKD).
• Replication in RCTs required before clinical recommendations can be made for microbiome-targeted interventions in HF.

Future Directions: AI and Integrated Omics

• Multi-omics integration (genomic + epigenomic + proteomic + metabolomic + microbiome data) + EHR phenotypic data + mobile/sensor technology + machine learning = future HF precision medicine framework.
• Supervised machine learning: models outcome prediction in HF using multiple independent omics + clinical variables.
• Unsupervised machine learning: HF subtyping for precision treatment stratification.
• All of Us Research Program: NIH-funded ($130 million) initiative enrolling ≥1 million Americans with comprehensive omic and health data to advance precision medicine.
• Key barriers: patient data privacy, database standardisation and searchability, statistical methodology for multidimensional omic data, and adequate omic-based clinical HF trial infrastructure.

Limitations of the Document

• Published in 2019; pharmacogenomics, AI/ML-in-HF, and microbiome landscape have evolved significantly since.
• Most pharmacogenomic associations lack replication due to limited cohorts with appropriate phenotype and sample size.
• Metabolomics limited by absence of standardised quantification across laboratories and incomplete understanding of dietary/comorbidity confounders.
• Microbiome-targeted interventions for HF are based on very small or non-HF trials; no RCT evidence supports clinical recommendations.
• GWAS HF/DCM findings predominantly based on European-ancestry cohorts — limited ethnic diversity in discovery populations.

Key Concepts Mentioned

• concepts/Genetic-Testing-in-Cardiomyopathy — yields by cardiomyopathy type; ACMG classification framework
• concepts/Titin-Isoform-Switch — RBM20 as master TTN isoform switch; abnormal splicing in RBM20-DCM
• concepts/Titin-PTMs — mTORC1 activation in TTNtv rat models; metabolic shift to glycolysis
• concepts/Gene-Silencing-Therapy — ncRNA therapeutic strategies (antagomirs, GapmeRs, miRNA mimics)
• concepts/HFpEF — metabolomic and proteomic profiles; acylcarnitine and ketone body distinctions from HFrEF
• concepts/Pharmacogenomics-in-HF — ADRB1/GRK5/GNB3 beta-blocker genetics; CYP2D6/CYP3A5 metabolism
• concepts/Gut-Microbiome-in-HF — gut dysbiosis mechanism; TMAO biomarker; uremic toxins; dietary/probiotic interventions

Key Entities Mentioned

• entities/Heart-Failure — precision medicine framework across all omics domains; omics landscape overview
• entities/TTN — TTNtv in 20–25% familial DCM; metabolic shift; mTORC1 activation; peripartum CMP subset
• entities/LMNA — nuclear stability + chromatin functions; MAPK/mTOR upregulation in mutations; ~8% DCM
• entities/DCM — genomic drivers (RBM20, BAG3, LMNA, TTN); GWAS variants; proteogenomics integration

Wiki Pages Updated

• wiki/sources/HF-Precision-Medicine-AHA-2019.md (created)
• wiki/sourceindex.md (updated)
• wiki/wikiindex.md (updated — new concept pages added)
• wiki/concepts/Pharmacogenomics-in-HF.md (created)
• wiki/concepts/Gut-Microbiome-in-HF.md (created)
• wiki/entities/Heart-Failure.md (updated — precision medicine/omics section)
• wiki/entities/TTN.md (updated — mTORC1 + metabolic shift from this source)
• wiki/entities/LMNA.md (updated — nuclear stability + MAPK/mTOR pathway)