AF is more prevalent in patients with a concomitant diagnosis of heart failure (HF) and has been linked to worsened cardiovascular outcomes including HF-related decompensation, hospitalisation and death.1–16 The influence of one pathological process on the other is complex, with cardiovascular outcomes made further divergent based on the type of AF and type of HF.15,16 In addition, there remains a subset of patients diagnosed with HF with recovered ejection fraction (HFrecEF), having initial reductions in their left ventricular ejection fraction (LVEF) solely and secondary to AF with rapid and irregular ventricular rates. Rhythm control in these patients has equated to improved cardiovascular outcomes.17–19 These findings pose the question of which patients, of those with HF and AF, would benefit most from rhythm control in addition to the use of guideline-directed medical therapy (GDMT).20
This article discusses rhythm and rate control in the treatment of patients with AF and HF with reduced ejection fraction (HFrEF), and those with HF with preserved ejection fraction (HFpEF), separately. Difference in outcomes using catheter ablation versus antiarrhythmic therapy in these HF subtypes are discussed in addition to the potential antiarrhythmic effect of GDMT on AF. This paper serves to provide guidance on the treatment of AF in both HFrEF and HFpEF.
Heart Failure with Reduced Ejection Fraction
Although there is ample room for further discrimination of outcomes based on aetiology or duration of HFrEF, LVEF, New York Heart Association (NYHA) class, and type of AF, the consensus and committee-based recommendations are to pursue rhythm control of AF in patients with HFrEF, and specifically via catheter ablation (CA).21–23 This recommendation is based on randomised and observational literature, including sub-group and meta-analyses, demonstrating significant reductions in cardiovascular outcomes and/or mortality when CA is pursued.24–31 Further delineation of patients most likely to benefit from ablation may be based on the integration of additional markers such as late gadolinium enhancement from cardiac MRI, or the summative burden of low-voltage zones with endocardial electroanatomical mapping.32 The strongest data supporting catheter ablation as front-line therapy in AF and HFrEF come from the results of the CASTLE-AF trial, an observational study mirroring the findings of CASTLE-AF, CABANA and the CABANA AF-HF sub-study.24,25,33,34
Options for rhythm control, besides CA, in patients with HFrEF are limited to amiodarone or dofetilide and cardioversion.22,23 Amiodarone and dofetilide are recommended based on the safety of using these drugs in HF patients with minimal effects on haemodynamics. Increased mortality has been associated with other antiarrhythmic agents.35–37
Studies randomising HFrEF and AF patients to amiodarone or dofetilide fail to consistently show mortality benefit, influenced by a low efficacy in maintaining sinus rhythm.27,38–40 Amiodarone was 34% as effective as rhythm control over a 2-year period in the AATAC trial.27 In contrast, patients randomised to CA in AATAC were 71% more likely to remain in sinus rhythm. In the congestive HF survival trial of antiarrhythmic therapy (CHF-STAT) amiodarone was compared with rate control in patients with HFrEF and AF.38 There was no difference in survival and a 31% efficacy of amiodarone in maintaining sinus rhythm. Patients who maintained sinus rhythm had significantly higher survival than those who remained in AF. The DIAMOND-CHF study compared dofetilide with rate control in patients with AF and HFrEF.39 Mortality did not significantly differ between arms. By 12 months, dofetilide had converted 44% of the patient arm, 61–79% if including dofetilide with electrical cardioversion.39,40 Considering cardioversion alone, the RACE study focused on patients with mild to moderate HF (NYHA II/III), demonstrating benefit for rhythm control but again a weak efficacy in maintaining sinus rhythm. Patients who converted (36%) had no cardiovascular mortality and reduced progression of HF over a follow up of approximately 2 years.41
β-blockers are primarily recommended for rate control in patients with AF regardless of LVEF.22,23 In the CAPRICORN trial, carvedilol treatment resulted in a lower incidence of AF or atrial flutter (AFL) over an average follow up of 1.3 years.42 That trial randomised patients to placebo or carvedilol after MI and was not limited to patients with HFrEF. The absolute number of patients with AF or AFL was low in that study and, as a result, the confidence interval is wide. Importantly, however, outside of any potential antiarrhythmic effect, there is questionable benefit of β-blockers in reducing mortality in HFrEF with AF.43–46 These analyses note that the mortality reduction of β-blockers in HFrEF may be reduced in patients with AF regardless of ventricular rate control, and benefits may be seen only in patients who maintain sinus rhythm.
The Dig-trial, which pre-dated current GDMT, primarily enrolled patients with NYHA class II–III HF and showed that treatment with digoxin for 2–5 years had no effect on mortality but modestly reduced the combined risk of death and hospitalisation.47 Non-dihydropyridine calcium channel blockers should be used with caution in HFrEF. Dronedarone should not be administered to patients with decompensated HF based on the ANDROMEDA study.22,23,48 Atrioventricular nodal (AVN) ablation with biventricular pacing can be considered for AF patients with uncontrolled ventricular rates and in whom a rhythm control strategy is ineffective.49 Findings from the PABA-CHF trial noted no significant improvement in LVEF, in contrast to CA.50
GDMT is mandatory for the treatment of HFrEF.20,51 Angiotensin receptor–neprolysin inhibitors (ARNIs), angiotensin-converting enzyme inhibitors and angiotensin receptor blockers reduce morbidity and mortality in patients with HFrEF. In the SOLVD trial, enalapril prevented the occurrence of AF, with similar findings in a post-MI population with left ventricular (LV) dysfunction using trandolopril.4,52 In Val-HeFT, valsartan prevented the occurrence of AF; similar findings were noted with candesartan in the CHARM trial.5–7 There are no data to show that ARNIs prevent AF. In a meta-analysis, sacubitril–valsartan did not result in a lower incidence of AF compared with either enalapril or valsartan, and in PARAGON-HF sacubitril–valsartan did not result in a lower incidence of AF among those without a previous history of AF.53,54 Eplerenone, a mineralocorticoid receptor antagonist, reduced the occurrence of AF compared with placebo in a secondary analysis of the EMPHASIS-HF trial.9 However, in TOPCAT post hoc analysis, spironolactone did not reduce AF occurrences.55 In the DECLARE-TIMI 58 trial, the sodium– glucose cotransporter-2 inhibitor dapagliflozin reduced AF in patients with and without HF but had no effect on AF in the DAPA-HF trial.56,57
As GDMT for HFrEF evolved to more current class 1 recommendations, via the use of β-blockers and aldosterone antagonists, rhythm versus rate control would again be explored in the AF-CHF and CAFE-II trials.58,59 In the AF-CHF trial, largely an amiodarone trial, no mortality difference was found.58 Hospitalisations were additionally greater in the rhythm control group due to arrhythmia-related reasons. In that study, most patients (58%) would have AF recurrence on amiodarone and for an unspecified period. Follow-up analysis of the AF-CHF trial using rhythm as a time-dependent covariate found that AF did not result in increased mortality or HF, even after comparing those with a high versus low prevalence of sinus rhythm.60 CAFE-II, also using amiodarone, focused principally on persistent AF and HFrEF. With a primary outcome of change in quality of life, no significant difference was noted.59
Overall, there is no certain mortality benefit of pharmacological rhythm control over adequate rate control (Table 1). Furthermore, pharmacological rhythm control may be associated with increased healthcare costs via arrhythmia-related hospitalisations, drug toxicity or repeated need for cardioversions.58–61 However, most studies attest that maintaining sinus rhythm can result in mortality reduction, identifying the limitation of antiarrhythmics: they do not alone sufficiently convert or maintain sinus rhythm long enough to demonstrate benefit.38,39 Patients should not have interruption or removal of HF GDMT to enable initiation or maintenance of antiarrhythmic therapy. Considering these points, rhythm control should be primarily pursued via CA in patients with HFrEF. Furthermore, benefit of ablation seems to be evident and persistent despite AF subtype, with greater absolute change in patients with non-paroxysmal AF.24,50
There are some patients with HFrEF who may not benefit from rhythm control with CA. Patients with advanced HF and higher NYHA class garner uncertain benefit via ablation.29 Those with permanent AF or evidence of severe atrial myopathy, particularly if secondary to valvular disease, may not benefit from rhythm control, although these patients are usually excluded from randomised trials. It is the opinion of the authors that unless there is contraindication or safety concerns, rhythm control via CA should be pursued in patients with AF and HFrEF given the limited pharmacological options and benefit of maintaining sinus rhythm (Tables 1 and 2).61 In patients for whom a rhythm control strategy is not an option and pharmacological rate control is not achieved, AVN ablation with biventricular or physiological pacing should be considered. Figure 1 outlines a treatment algorithm for patients with HFrEF and AF.
Heart Failure with Preserved Ejection Fraction
The definition, diagnosis and treatment of HFpEF remain variable with no pharmacological or intervention consistently proven to reduce cardiovascular mortality, as seen in patients with HFrEF.62,63 This non-standardised method of diagnosing HFpEF, in addition to limited GDMT for this subtype of HF, creates significant outcome heterogeneity in observational and randomised studies comparing rhythm with rate control in patients with AF.
The European Society of Cardiology (ESC) guidelines currently recommend β-blockers, diltiazem, verapamil and digoxin for rate control (<100–110 BPM) in patients with HFpEF.23 IV amiodarone may be appropriate only in the acute setting. There are otherwise no consistent data to show that GDMT reduces AF occurrences in patient with HFpEF. The EMPEROR Preserved Trial is studying whether empagliflozin will be effective in suppressing AF in this patient population.64 In the DELIVER trial, dapagliflozin improved outcomes across the patient population with AF but did not suppress the occurrence of AF.65 In the TOPCAT trial, spironolactone did not reduce the new onset or recurrence of AF or AFL in patients with symptomatic HFpEF.66 Finally, although eplerenone decreased AF occurrence in patients with renal disease, this has not been shown in patients with HFpEF.67
Presently, there are three published randomised trials comparing rhythm control via ablation with rate control in patients with AF and (largely) HFpEF.34,68,69 In the RAFT-AF trial, patients with HFpEF were defined as those with LVEF >45%, an elevated N-terminal pro-brain natriuretic peptide level, and NYHA class II–III HF.68 Approximately 40% of the patient population studied had HFpEF. Comparing ablation with rate control, there was no significant difference in the primary outcome of cardiovascular mortality or HF events in all patients over a minimum follow-up of 2 years. These findings persisted with subgroup analysis limited to those with HFpEF. Subgroup analysis of the CABANA trial is limited in that definition and diagnosis of HF were reliant on clinical interpretation with no specific diagnostic criteria.34 However, when limiting analysis to patients with a diagnosis of HF, of which most patients included had EF >40%, the primary composite outcome of death, stroke, bleeding or cardiac arrest was significantly in favour of ablation versus drug therapy (pharmacological rhythm or rate control). This was reproducible when analysing patients in CABANA who had a high modified H2FPEF score.70 The H2FPEF score is a weighted scoring system to determine the probability of HFpEF.
In the study by Chieng et al., the diagnosis of HFpEF was much more stringent, requiring elevations in pulmonary capillary wedge pressure (PCWP) at rest or with exercise in addition to an EF ≥50%.69 That study compared ablation with pharmacological rhythm or rate control; the rate control group was defined as having at least adequate rate control of <100 ventricular BPM in AF and a resting systolic blood pressure of <160 mmHg. The primary endpoint of that 6-month follow-up study was the difference in PCWP. Although it was a small study of 31 patients, there remained a significant reduction in PCWP on follow-up in those randomised to ablation. No significant difference was seen in those maintained on medical therapy. Furthermore, there was sufficient improvement in the exercise right heart catheterisation data that would suggest reversal of an HFpEF diagnosis in half of patients randomised to ablation on follow-up testing (Table 3).69
The EAST-AFNET4 trial did not use ablation as the only rhythm control option but did compare early rhythm control with rate control in patients with AF diagnosed at or under 1 year before enrollment.71 The trial is notable in that early rhythm control was associated with significant reduction in death from cardiovascular cause, stroke, hospitalisation for HF or acute coronary syndrome. Most patients were treated with various classes of antiarrhythmics and not ablation. In a follow-up analysis focusing on those with a diagnosis of HFpEF there was no significant difference between treatment arms.72
Meta-analyses including randomised and observational studies focusing on HFpEF found a significant reduction in all-cause and cardiovascular mortality in favour of rhythm control, be it catheter ablation or antiarrhythmic.73,74 Finally, a cohort study by Xie et al. noted a significant reduction in HF hospitalisations in patients assigned to ablation when propensity score matched patients were assigned to medical therapy.75 There was no difference in all-cause death. Additional thorough review of ablation of AF in HFpEF reinforces favourable outcomes without compromise in safety.76
Although mortality data are not as consistent or robust for rhythm control in patients with AF and HFpEF versus those with HFrEF, this should not negate the potential for benefit of use. Rhythm control, be it via CA or pharmacological therapy, shows evidence for morbidity reduction when compared with rate control of AF in patients with HFpEF. Furthermore, remember that the definition and inclusion of patients with HFpEF, in these studies, remain varied. Further clarity, standardisation of the diagnosis and causes of HFpEF are needed, given that it is possible that outcomes may differ. For example, it is reasonable to consider that rhythm control in the patient with AF-induced HFpEF would be significantly superior to rate control across several outcomes, whereas such differences may not occur in the patient with HFpEF secondary to multiple, influential comorbidities. The severity of HFpEF will also need further exploration, with separation of groups based on LVEF range. Future literature will also need to further stratify outcomes based on subtype of AF.
It is the opinion of the authors that rhythm control should be pursued, when feasible and safe, in patients with HFpEF. Rhythm control can be attempted via the use of antiarrhythmic medications or ablation, although ablation remains most efficacious for durable maintenance of sinus rhythm. In addition to amiodarone and dofetilide, dronedarone can be considered in HFpEF patients with AF in the absence of recent HF decompensation. The ATHENA trial had many such patients who were elderly, hypertensive and with diastolic dysfunction.77 GDMT may have secondary advantages in controlling hypertension and minimising left atrial dilation that may be helpful in the long-term in suppressing AF.67,78 Figure 2 outlines a treatment algorithm for HFpEF and AF.
Conclusion
In patients with HFrEF, rhythm control strategies, specifically CA, should be prioritised along with GDMT to reduce cardiovascular mortality. Conversely, there is no confirmed mortality benefit of rhythm control in HFpEF patients with AF, but there is morbidity reduction. The greatest benefit of rhythm control seems to occur with earlier diagnosis and treatment, highlighting a potential window of reversibility of HF and the overlapping pathogenesis between AF and HF, regardless of subtype. CA offers the greatest chance for durable rhythm control and can be safely performed with minimal risk. This may additionally avoid or reduce the use of various antiarrhythmic drugs that can interact with or limit the therapeutic dosing of other medications as indicated for the HF patient.
Clinical Perspective
- Rhythm control of AF is proven to reduce mortality in patients with heart failure with reduced ejection fraction.
- Rhythm control of AF reduces morbidity in patients with heart failure with preserved ejection fraction.
- Rhythm control is most successful via catheter ablation.
- Guideline-directed medical therapy for heart failure may help to reduce AF but is not an effective lone strategy for durable rhythm control.