Comparison of dronedarone vs. flecainide in the maintenance of sinus rhythm, following electrocardioversion in adults with persistent atrial fibrillation: a systematic review and meta-analysis

Abstract

Aims

To compare flecainide and dronedarone for sinus rhythm (SR) maintenance following electrocardioversion of persistent atrial fibrillation (AF), in patients with minimal or no structural heart disease.

Methods and results

A systematic search of publications using EMBASE, CENTRAL, CINAHL, and MEDLINE (1989–2019), identified a total of 595 articles. No limitations were applied. Nine articles met the inclusion criteria [five randomized controlled trials (RCTs) and four cohort studies], encompassing 1349 persistent AF candidates. Two retrospective studies compared flecainide with dronedarone, indicating a 6% reduced risk of AF recurrence with flecainide; however, results were not statistically significant [risk ratio (RR) 0.94, 95% confidence interval (CI) 0.71–1.24; P = 0.66]. One RCT compared dronedarone to placebo, demonstrating a 28% reduced risk of AF recurrence at 6 months (RR 0.72, 95% CI 0.58–0.90; P = 0.004). Two RCTs compare flecainide to placebo, when a 16% decreased risk of AF recurrence at 6–12 months was indicated; however, these results were not statistically significant (RR 0.84, 95% CI 0.66–1.07; P = 0.16). Within a 6- to 12-month follow-up period, a combined recurrence rate of AF was examined, in which flecainide and dronedarone maintained SR in 50% and 42%, respectively. Four articles satisfied quality appraisal, one of which focused on flecainide data.

Conclusion

Dronedarone and flecainide displayed similar efficacy in maintaining SR in patients following electrocardioversion for persistent AF. The SR maintenance was numerically but not statistically significant in the flecainide group. Side effects uncovered similar pro-arrhythmic activity. However, in light of the deficiency of volume and quality of available evidence, the writer acknowledges the requirement for future research.

Introduction

As the most commonly sustained cardiac arrhythmia worldwide, atrial fibrillation (AF) holds the pre-eminent position, existing in one of every four adults.¹ A quarter of this population have persistent AF, in which the maintenance of sinus rhythm (SR) becomes the predominant objective. This is important in order to prevent the structural and electrophysiological remodelling precipitated by AF, rapidly leading to permanent AF and concomitant heart failure (HF).^2–5 Representing a significant economic burden, management of AF costs of €6.2 billion annually within five European countries, with a correlated two-fold mortality increase.⁶

AF promotes electrical, contractile, and structural remodelling within the atrial myocardium.⁷ Early initiation of SR can result in reversible damages, however, if left untreated, irreversible apoptosis, necrosis and fibrosis begins to unfold from as early as 1 month from AF occurrence.^8,9 To date, guidelines have recommended electrocardioversion as a first-line strategy to restore SR; however, relapse is common, with up to 84% of patients experiencing a relapse within 1 year.¹⁰ In persistent AF, the atrium is increasingly vulnerable during the ‘reversal remodelling’ phase, however, through initiation of antiarrhythmic drugs (AADs) following cardioversion, the maintenance of SR is doubled when compared to placebo.^11,12

Recommendations suggest Vaughn Williams classification III or IC AADs.^12,13 However, despite amiodarone demonstrating elite Class III efficacy, its toxic side effects limit its utilization for patients with structural heart disease (HD) alone.^12,14–18 Dronedarone was developed by alteration of the amiodarone compound, through the addition of a methane solfonyl group and removal of iodine, that precipitated amiodarone’s organ toxicity and tissue accumulation.^19,20 Dronedarone is the primary Class III agent advised for patients with minimal, or no structural HD, described as a pleiotropic agent²¹ as it enables blockage of multiple ion channels (Na, K, and Ca).^22,23 Flecainide is the comparative Class IC agent, a commonly prescribed agent since 1984 that manipulates the fast approaching Na channels, thus prolonging the action potential duration.²⁴

There is a delineated gap in knowledge regarding the choice of a first-line antiarrhythmic following cardioversion, for patients with persistent AF, with minimal, or no structural HD.^25–27 In light of this, a systematic review and meta-analysis was conducted, to establish if there is clear evidence as to which drug, either flecainide or dronedarone is more effective to maintain SR following electrocardioversion for persistent AF.

Methods

Design

A systematic review was undertaken following the guidance of PRISMA²⁸

Research question

The research question was formulated using the PICO model;

• Population of interest was adults with minimal or no structural HD, succeeding electrocardioversion for persistent AF, in SR prior to drug commencement.

• The Intervention was the Class IC AAD flecainide, and Comparison class III AAD dronedarone.

• The primary Outcome was SR maintenance, measured by the AF recurrence rate. Secondary outcomes studied were side effects of the AADs, including ventricular rate-control and pro-arrhythmic incidents.

Inclusion and exclusion criteria

Articles of interest were those investigating patients with persistent AF prior to cardioversion, with absence of, or minimal structural HD, that evaluated flecainide OR dronedarone for the purpose of SR maintenance, applied as a long-term rhythm control strategy.

Exclusion criteria were as follows: (i) candidates under 18 years; (ii) significant structural HD; (iii) qualitative methodologies; (iv) pharmacological cardioversion; (v) follow-up <6 months; and (vi) paroxysmal or permanent AF.

Search strategy

A database search using EMBASE, CENTRAL, CINAHL, and MEDLINE was executed from inception until March 2019, utilizing the following keywords in various combinations: ‘atrial fibrillation’ AND ‘cardioversion’ AND ‘dronedarone’ AND ‘flecainide’ (Figure 1). No limitations were applied to language, or publication date. Retrieved references were uploaded to endnote X7 and duplicate citations removed.

Data extraction

Selection of studies and subsequent data extraction were achieved based on a priori formulated criteria. Data extracted to a pre-designed data extraction table included study design, interventions, comparators, sample size and characteristics, inclusion and exclusion criteria, follow-up time, outcome measures, and results for the primary and secondary outcomes.

Quality assessment

Methodological quality of the included studies was assessed through RevMan software (version 5.3).²⁹ The EBL critical appraisal checklist³⁰ was applied to observational studies, and a score of ≥75% indicted study validity.

Statistical analysis

Where appropriate, data were analysed using the RevMan software (version 5.3).²⁹ To enable comparison of the two AADs, subgroups were created to compare the two drug categories separately within randomized controlled trials (RCTs). Data were analysed using risk ratio (RR) or hazard ratio (HR). Meta-analysis was undertaken using a pooled RR and 95% confidence interval (CI) employing a weighted random effects model. Results for the meta-analysis are presented using a forest plot. Heterogeneity among studies was tested and P-values were expressed to reflect statistical significance. Where statistical analysis of the results within studies was not possible, results are presented using a narrative summary.

Results of the search

The comprehensive database search yielded a total of 596 articles for review when duplicates were extracted. A further three full-text articles of relevance were retrieved from Google Scholar and following a rigorous inspection of chosen articles, the reference lists uncovered a further two articles which were ultimately excluded as not meeting the inclusion criteria. Finally, nine articles^31–39 met the inclusion criteria and formed the basis for this systematic review (see Figure 1 for an overview of the search process).

Participant characteristics and sample size

Table 1 outlines the characteristics of the included studies. Overall, nine studies (five RCTs^{31–33,35,36} and four cohort studies^34,37–39) were included. The studies involved a total of 1349 persistent AF candidates, with a follow-up 6–12 months. The majority (67%) of the participants were male, with a mean age of 63 years.

Included studies reflect an extensive geographical range, increasing the generalizability of findings; three studies conducted research in one country; Spain³⁹, UK,³⁴ and the Netherlands,³¹ two studies were carried out in Germany^35,37 and Korea,^36,38 and two studies independently covered a national and European level of 23 and 11 countries, respectively.^32–33

The European Society of Cardiology (ESC) guidelines defines persistent AF as ‘AF lasting longer than 7 days and less than 1 year’.¹² The definition of persistent AF varied within studies, furthermore, persistent AF was either not explicitly defined within studies^34,37–39 or was incorrectly categorized.^{31–33,35,36} Due to the limitation of trials examining a cohort of patients with absence of structural HD, candidates with minimal signs of structural HD was defined for the purpose of this review as a left ventricular ejection fraction (LVEF) ≥40% or New York Heart Association (NYHA) Class I or II.⁴⁰

Interventions, comparisons, and follow-up

In total, in the intervention groups, 26% (n = 352/1349) of the participants within five studies received flecainide and 35% (n = 477/1349) received dronedarone within six studies. In total, in the comparison groups, 12% (n = 156/1349) were enrolled to placebo within three studies, and 27% (n = 361/1349) were enrolled to a comparison AAD within three studies.

Dronedarone 400 mg twice daily (b.i.d.) orally was the independent dosage administered within the six studies investigating dronedarone.^{32,33,36–39} Three of these studies had a follow-up of 6 months^32,36,37 and the remainder administered the drug for 12 months.^33,38,39 Of the five studies investigating flecainide, one study did not report the dose³⁴ and the remainder report a varied dose regime ranging from 50 mg b.i.d.,³⁸ 50–100 mg b.i.d.,³⁹ 100 mg b.i.d./ three times daily (TDS),³⁵ to 100–150 mg b.i.d.³¹ Three of these studies had a 12-month follow-up^31,38,39 and the remainder administered the drug for 6 months.^34–35

Measurement of SR reflected various assessment methods throughout studies. Electrocardiogram (ECG) was the most common measurement employed (78%; n = 7),^{31,33,35–39} followed by telemetry (33%; n = 3),^32,35,37 and Holter monitoring (33%; n = 3).^31,35,39 One study adopted clinical evaluation³⁶ and one study did not outline the precise measurement used (Table 1).³⁴

Quality appraisal

Five studies were evaluated using the RevMan risk of bias table (Figure 2) and four studies were assessed utilizing the EBL quality appraisal tool.

Selection bias

Two studies were identified as having unclear bias, in this domain,^35,36 in light of their open-label design methods.

Performance bias

Two studies were identified as having high risk of performance bias^31,35 in light of the dose manipulation of flecainide during the study. Kirchhof et al.³⁵ initially administered 100 mg b.i.d. and increased to TDS at the ‘discretion of the investigator’, furthermore, the placebo arm was only investigated for 4-week duration, compared with 6 months for the intervention group. Van Gelder et al.³¹ initially administered a dose of 100 or 150 mg b.i.d., determined by body weight and renal function, and this was reduced to 150 or 200 o.d. if pro-arrhythmic side effects were observed. Chun et al.³⁶ was assigned high risk, in light of the open-label design and subsequent confounding factors, as a 19% increase of beta-blockers (P = 0.042) were observed within the comparison group.³⁶

Detection bias

Two studies were identified as having an unclear risk of detection bias.^32,33 AF is termed a complex condition in terms of its ‘silent’ episodes, resulting in potential asymptomatic occurrences, thus if a detection is based on a patient becoming ‘symptomatic’ this outcome measurement is classified as subjective.³⁶ Infrequent follow-up and outcome measurement were observed in two studies^31,36 ultimately reducing the sensitivity of event detection and subsequent detection bias. Stand-alone ECG’s detected AF, however, frequent follow-up methods could have captured more events.³³ Touboul et al.³² had reduced sensitivity and high specificity of AF reoccurrence detection, defined as: ‘AF episode lasting 10 minutes or more, captured on two distinct telemetry submissions, separated by the same time duration’.

EBL assessment

Population

In light of the sample size (n = 40/33³⁸ and n = 14/15³⁹), in two of the studies, there is a risk of under powering of the studies, where there may not be a sufficient sample size to detect differences between the study groups, should such differences exist. Retrospective studies and subsequent uncontrolled design measures^34,38,39 resulted in a high level of bias throughout these domains.

Data collection

Unclear methods of data collection and infrequent follow-up methods were observed within three studies.^34,38,39 Outpatient ECG monitoring may have missed the true volume of AF recurrences.³⁷

Study design

Retrospective design^34,38,39 resulted in limited data available to draw conclusions from the follow-up period.

Results

Confounding factors resulted in a low validity score for three studies (23%,³⁴ 52%,³⁸ and 42%³⁹). It was observed that 35% of the flecainide group had an increased beta-blocker utilization when compared with dronedarone (P = 0.032). Furthermore, there was a higher prevalence of structural HD in the dronedarone cohort (21% with left ventricular hypertrophy P = 0.01)³⁹ and flecainide dose variation was identified (50–100 mg b.i.d.).³⁹ Absence of dose reporting³⁴ threatened the external validity of this studies results. Gwag et al.³⁸ denote beta-blockers were more frequently used in combination with flecainide 66.7% (n = 22/33), compared with dronedarone 15% (n = 6/40).

Results for outcome: atrial fibrillation recurrence

Comparison 1: dronedarone vs. comparison 2 studies; 602 participants

Two studies were included for this comparison^33,36 and Figure 3 outlines the results of the meta-analysis. As can be seen, in the dronedarone group 43% (n = 127/298) had AF recurrence, conversely, in the comparison group (propafenone³⁶/amiodarone³³), 33% (n = 99/304) had AF recurrence. The RR is 1.30 (95% CI 1.07–1.59) indicating a 30% increased risk of AF recurrence in the dronedarone group (P = 0.008). These quantitative results suggest there is heterogeneity (I² = 86%).

Comparison 2: dronedarone vs. placebo 1 study; 102 participants

One study was included for this comparison³² and Figure 4 outlines the results of the analysis. As can be seen, in the dronedarone group 65% (n = 35/54) had AF recurrence, conversely, in the placebo group, 90% (n = 43/48) had AF recurrence. The RR is 0.72 (95% CI 0.58–0.90) indicating a 28% decreased risk of AF recurrence in the dronedarone group (P = 0.004). Similar results were observed in a prospective observational study,³⁷ when 73% (n = 52/71) of dronedarone treated subjects had AF recurrence at 6 months.

Comparison 3: flecainide vs. placebo 2 studies; 401 participants

Two studies were included for this comparison^31,35 and Figure 5 outlines the results of the meta-analysis. As can be seen, in the flecainide group 40% (n = 118/293) had AF recurrence, conversely, in the placebo group, 50% (n = 54/108) had AF recurrence. The RR is 0.84 (95% CI 0.66–1.07) indicating a 16% decreased risk of AF recurrence in the flecainide group (P = 0.16).

Comparison 4: flecainide vs. comparison pharmacotherapy 1 study; 71 participants

A retrospective study was included for this comparison³⁴ and Figure 6 outlines the results of the analysis. In the flecainide group, 55% (n = 6/11) had AF recurrence, conversely, in the comparison group (amiodarone/soltalol), 72% (n = 43/60) had AF recurrence. The RR is 0.76 (95% CI 0.43–1.34) indicating a 24% decreased risk of AF recurrence in the flecainide group (P = 0.34).

Comparison 5: flecainide vs. dronedarone 2 studies; 102 participants

Two retrospective studies were included for this comparison,^38,39 and Figure 7 outlines the results of the meta-analysis of this study. As displayed, in the flecainide group, 58% (n = 28/48) had AF recurrence, conversely, in the dronedarone group, 65% (n = 35/54) had AF recurrence. The RR is 0.94 (95% CI 0.71–1.24) indicating a 6% decreased risk of AF recurrence in the flecainide group (P = 0.66). Figure 8 displays the combined recurrence rate of AF examined throughout studies.

Secondary outcomes

Two RCTs investigated ventricular rate control as a secondary outcome,^32,36 four studies explored the AAD side effects,^32,33,37,39 five studies reported pro-arrhythmic events,^{31,33,35,37,39} and two studies did not report any secondary outcomes of interest.^34,38

Ventricular rate control

Administration of dronedarone yielded a median 6.5 heart beats/min reduction when compared with comparison (76.5 vs. 83 b.p.m.).³⁶ Similarly, when compared with placebo, ventricular rate at recurrence was reduced, on average by 13.2 heart beats/min (P = 0.0001).³² Furthermore, Said et al.³⁷ outlined that 16.8% of participants remained on dronedarone due to effective rate-control, despite AF recurrence.

Side effects

Incompletion of dronedarone pharmacotherapy over the study period was observed in Touboul et al.³², Le Heuzey et al.,³³ and Said et al.,³⁷ the drug was discontinued by 3.9% (n = 3/76), 10.4% (n = 26/249) and 22% (n = 42/191) (95% CI 17–27%) of the cohort respectively, as a result of side effects. Gastrointestinal (GI) disturbances were the prevailing side effects within dronedarone cohorts.^32,33,37,39

Within dronedarone cohorts, an increase in hepatic enzymes was observed in two studies^33,37 when 12% (n = 30/249)³³ and 2% (n = 3/191)³⁷ of the cohort developed abnormal liver function tests. The remaining studies investigating dronedarone, reported no significant liver function changes at follow-up.^32,39

Soriano et al.³⁹ reported 15.38% (n = 8/52) vs. 16.9% (n = 12/71) side effects regarding dronedarone and flecainide, respectively, ultimately revealing dronedarone showed a safety profile comparable to flecainide (HR 0.68, 95% CI 0.18–2.53; P = 0.566). Two deaths occurred during the treatment period of dronedarone 0.8% (n = 2/249) due to pulmonary embolism and unwitnessed death.³³ No deaths were reported in studies investigating flecainide. Discontinuation of flecainide treated subjects was reported in one study, when 9% (n = 5/58)³¹ of the cohort experienced nausea, headache, sinus arrest, rate-related left bundle branch block (LBBB) or symptomatic premature ventricular contractions (PVCs).

Proarrhythmic incidents

Dronedarone

QTc interval prolongation ≥500 ms was reported by Le Heuzey et al.³³ and Said et al.,³⁷ occurring in 11% (n = 27/249) (P = 0.0033) and 2.6% (n = 5/191) of the cohorts, respectively, with no subsequent torsades-de-pointes. Pro-arrhythmia was documented in 8.9% (n = 17/191), with one requiring resuscitation.³⁷ Furthermore, one serious episode of bradycardia was reported by Soriano et al.³⁹

Flecainide

Three studies explored the pro-arrhythmic profile of flecainide.^31,35,39 Of these, beta-blockers were concurrently administered with flecainide in 77%³⁵ and 52%³⁹ of the cohort. Soriano et al.³⁹ reported 9.86% (n = 7/71) pro-arrhythmic incidents, with one significant QRS widening. Similarly, Van Gelder et al.³¹ reported 9% (n = 5/58) pro-arrhythmic events including syncope, sinus arrest, high-grade AV block, rate related LBBB, and pro-arrhythmic PVC’s. Kirchhof et al.³⁵ documented 10.65% (n = 28/263) adverse incidents, including one episode of resuscitation and two sustained ventricular tachycardia (VT).

Discussion

It is evident that RCTs investigating a rhythm vs. rate control treatment in AF, disregard dronedarone as an AAD option.⁴¹ Two trials (RACE and AFFIRM) considered flecainide, however, a small subgroup of patients were enrolled to this pharmacology, with structural HD prevalent in around 80% of the subjects.^16,42 Studies that compare catheter ablation to AADs in persistent AF, neglect both flecainide and dronedarone following electrocardioversion, despite current guideline recommendations.^12,43 In light of this, the conclusions drawn from these studies are questioned, further increasing the gap in knowledge and consolidating the requirement for further research in this area. Furthermore, researching the efficacy of flecainide and dronedarone could redefine the balance towards a rhythm control strategy in persistent AF patients.

Kirchhof et al.³⁵ demonstrated that short-term treatment (4 weeks), although inferior to long-term (46% vs. 39% AF reoccurrence), may be appropriate within a selected cohort of patients who are at low risk of recurrent AF or high risk of pro-arrhythmia, as 80% of the short-term group carried a 6 month effect of the long-term treatment arm within their study. The first week following cardioversion, represents the most vulnerable period for reoccurrence.^11,44 It is not adequately elucidated however the exact length of vulnerability, that would give rise to an appropriate length of AAD administration, strengthening rationale for a 6-month follow-up duration. Berry et al.⁴⁵ demonstrated that SR at 1 month was an independent predicator of SR at 1 year of follow-up. This would be an interesting finding to support a shorter duration of AAD therapy, as remodelling has been reversed and recurrence of AF is less likely to occur. At baseline, dronedarone candidates had more prevalent structural HD presence,^32,33,39 when compared with flecainide.^31,34 As there were limitation of trials investigating persistent AF candidates with absence of structural HD, the trials included investigate a lower baseline HD profile than other studies.²⁷ With regard to secondary outcome reporting within four studies,^31,32,37,39 it is worthy of note that the side-effect profile of the drug is reported with a combined paroxysmal and persistent AF cohort, which ultimately overestimates the results.

Another limitation of this study is flecainide’s unreported³⁴ or variable dose regime throughout studies.^31,35,38,39 The ESC guidelines¹² denote that flecainide should be administered at 100 mg b.i.d. Despite this, the two exclusive trials that compare flecainide and dronedarone, constitute sub-optimal dosing.^38,39 Soriano et al.³⁹ denote the recommended dose may entail a risk to participants, therefore 25% of the cohort received 100 mg b.i.d.³⁹ Gwag et al.³⁸ increased to the therapeutic dose solely when there was an AF recurrence. Both studies published in 2018, following guideline recommendations, further accentuates the requirement for quality research evidence and the vitality of applying ESC guideline recommendations for maintaining SR within this patient cohort.

The increased combination of beta-blockers with flecainide, compared with dronedarone was observed.^35,38,39 ESC guideline recommendations explain that through pre-administering a beta-blocker with flecainide, high ventricular rates can be prevented resulting from the conversion of AF into Aflutter with 1:1 conduction.¹² It’s recognized that the high concomitant utility of beta-blockers with flecainide could explain why this side effect was not observed. Similarly, RCTs investigating flecainide provide further reassurance in the context of pro-arrhythmic activity and subsequent safety of dosage (100 mg b.i.d.),^46,47 with ongoing research anticipated to enlighten these findings.⁴⁸

As there is an absence of RCTs comparing both AADs, RCTs comparing the individual drug to placebo is the most effective approach to compare outcomes. Despite this, only three studies compared drug to placebo,^31,32,35 two of which were deemed valid following quality assessment.^32,35 In this case, at 6 months, a 28% reduced risk of AF recurrence was identified for dronedarone (RR 0.72, 95% CI 0.58–0.90; P = 0.004)³² and a 14% reduced risk of AF was revealed for flecainide (RR 0.86; 95% CI 0.65–1.14; P = 0.28).³⁵ Unequal assignment of candidates to placebo and intervention (n = 81 vs. n = 263), may have underestimated the statistical conclusion in this respect.³⁵ Two retrospective trials compare flecainide to dronedarone,^38,39 in which case flecainide reduced the risk of AF recurrence by 6% [RR 0.94 (95% CI 0.71–1.24 P = 0.66)], however lack of statistical significance and quality renders these outcomes invalid. The high risk of bias within the research was predominately driven by confounding factors, intervention and construct validity. The undefined categorization of persistent AF, dosage manipulation and structural HD presence, furthermore, limits the generalizability of these findings. As a result, a requirement for further research evidence exists.

Conclusion

Overall, flecainide demonstrated similar efficacy for SR maintenance when compared with dronedarone. The pro-arrhythmic profile was similar, however, dronedrone candidates reported an increased GI intolerance and reduction of ventricular rate-control at time of AF recurrence. There is a requirement for further research evidence, in light of the study limitations and certainty of the findings of this review. An RCT comparing flecainide to dronedarone, would most effectively determine a first-line drug choice within this patient cohort. Despite the limitation as outlined pertaining to study outcome certainty, this systematic review should facilitate grounds for generation of future research evidence.

Conflict of interest: Funding was received from the Mater Private Hospital in Dublin Ireland, for the completion of the MSc within the RCSI, however, the hospital had no involvement during the completion of this study. This study complies with the Declaration of Helsinki.

References