QT as a predictor of recurrence after atrial fibrillation ablation and the impact of amiodarone: results from the placebo-controlled AMIO-CAT trial

Aims

Prolonged corrected QT interval (QTc) might be associated with arrhythmia recurrence after atrial fibrillation (AF) ablation. The effect of short-term amiodarone in this setting remains unknown. This study seeks to quantify short-term amiodarone’s impact on QTc, and to investigate QTc and amiodarone treatment as predictors of recurrence of arrhythmia after ablation.

Methods and results

The Short-term AMIOdarone treatment after CATheter ablation for atrial fibrillation (AMIO-CAT) trial randomized patients to 8 weeks of oral amiodarone or placebo following AF ablation. Scheduled and symptom-driven 12-lead electrocardiography and 3-day Holter-monitorings were performed. The endpoint was atrial fibrillation, atrial flutter or atrial tachycardia (AF/AT) lasting >30 s. The cut-off for prolonged QTc was 450 ms for men and 460 ms for women. A total of 212 patients were included, of which 108 were randomized to amiodarone and 104 to placebo. From baseline to 1 month QTc in the amiodarone group increased by 27 (±30) ms, while at 6 months QTc had normalized. After 3-months of blanking, new AF/AT recurrence was detected in 63% of patients with prolonged QTc vs. 41% of patients with normal QTc at baseline, and in multivariate Cox regression, prolonged QTc was associated with AF/AT recurrence [hazard ratio (HR) 2.19, P = 0.023]. Among patients with baseline QTc below median, amiodarone treatment decreased the rate of AF/AT recurrences (HR 0.43, P = 0.008).

Conclusions

Amiodarone increased QTc with 27 ms compared to placebo, and this effect decreased rapidly after drug discontinuation. Prolonged QTc at baseline independently predicted AF/AT recurrence, and baseline QTc identified patients who would possibly benefit from short-term amiodarone following ablation.

Introduction

Recurrence of arrhythmia remains a significant concern in patients undergoing catheter ablation for atrial fibrillation (AF).¹ In order to appropriately select patients for the procedure, previous studies have struggled to identify factors associated with poor ablation success.^2–4 Age, obesity, persistent AF, left atrial measures, and early recurrences have been suggested as important risk factors for long-term recurrence. However, early recurrences are highly common, and a 3-months blanking period is recommended before considering reablation.¹ Accordingly, at hospital discharge approximately two-thirds of patients are treated with anti-arrhythmic drugs, though long-term implications remain unclear.^1,5

Amiodarone is the most commonly used and most potent drug for maintaining sinus rhythm in these patients.^5,6 The drug has wide antiarrhythmic properties (Class I–IV), but the main effect is attributed to prolonging the repolarization by inhibition of potassium efflux (Class III), which in the electrocardiography (ECG) is observed as a an increase in QT duration.^7,8 However, knowledge about the extent to which oral amiodarone impacts the QT interval is limited, and the recovery after drug withdrawal remains uninvestigated. Furthermore, prolonged corrected QT interval (QTc) is a risk marker for new-onset AF,⁹ and recent smaller studies have suggested that QTc prolongation at baseline might be associated with recurrence after AF ablation in patients with hypertrophic cardiomyopathy or diabetes.^10,11

In the present analysis of the Short-term AMIOdarone treatment after CATheter ablation for atrial fibrillation (AMIO-CAT) trial,¹² we utilized the placebo-controlled design of the trial to quantify the impact on QTc from standard dosing of oral amiodarone during and after treatment. Furthermore, we sought to investigate QTc and amiodarone-mediated changes in QTc as predictive factors for recurrence after AF ablation.

Methods

Trial design

The design of the AMIO-CAT trial has been reported previously.¹² In brief, AMIO-CAT was a two-centre, double-blind, placebo-controlled, randomized clinical trial recruiting patients referred for catheter ablation of symptomatic paroxysmal or persistent AF. Exclusion criteria included previous amiodarone treatment within 3 months. All patients underwent pulmonary vein (PV) isolation by radiofrequency ablation, and the procedural endpoint was PV isolation by wide antral circumferential ablation assessed by meticulous remapping or by a circular mapping catheter. Additional substrate modification (linear ablations or complex fractionated electrogram-guided ablations) was left to the discretion of the operating electrophysiologist. Following AF ablation, patients were randomized to 8 weeks of amiodarone or matched placebo. Amiodarone was initiated with a loading dose of 400 mg on the night of ablation, followed by 400 mg b.i.d. for 13 days, 200 mg b.i.d. for 2 weeks, and 200 mg o.d. for 4 weeks. Other antiarrhythmic drugs were discontinued 3–5 days prior to ablation (except for rate controlling drugs given for other indications such as hypertension).

This investigator initiated trial was conducted in accordance with the Helsinki Declaration, approved by the Danish Regional Ethical Committee, the Danish Medicines Agency, and the Danish Data Protection Agency, was registered at clinicaltrials.gov prior to study start (https://clinicaltrials.gov/ct2/show/NCT00826826) and was monitored by the regional Good Clinical Practice unit, while all participants provided written informed consent. Enrolment began in February 2009 and was terminated in July 2013, while follow-up concluded in 2014. All data were gathered in a centralized electronic case report form system.

Baseline and follow-up measurements

The study visits before randomization and after 1, 3, and 6 months of follow-up included detailed medical history, routine physical examination, and 12-lead ECG. Three-day Holter-monitoring was performed before randomization and 8 weeks and 6 months after ablation. Furthermore, extra Holter-monitoring and/or 12-lead ECG was obtained in case of any symptoms possibly related to arrhythmia at any time after ablation. A transthoracic echocardiogram was performed during the 3 months prior to ablation in order to asses left ventricular and atrial function by means of left ventricular ejection fraction (LVEF), left atrial volume indexed by body surface area, left atrial volume at ventricular end systole (LAESV), and left atrial volume at ventricular end diastole.

Endpoints

The endpoint was defined as time to any documented atrial fibrillation, atrial flutter or atrial tachycardia (AF/AT) lasting >30 s, and was grouped as first AF/AT recurrence at any time during follow-up or first AF/AT recurrence occurring after a blanking period of 3 months post-ablation. AF/AT recurrences were evaluated by an adjudication committee before un-blinding the trial.

Corrected QT interval measurements

All ECGs were reviewed by an experienced physician blinded to the patient’s treatment group to accommodate for ECG findings inconsistent with a valid automatic measurement of the QT interval; i.e. electrical noise, bradyarrhythmias or tachyarrhythmias (heart rate <40 or >110 b.p.m.), bundle branch blocks, ventricular rhythms (QRS interval >120 ms), or multiple premature complexes. If AF was present on the ECG, QTc was calculated from a minimum of three consecutive cardiac cycles and the mean value was registered. Fridericia’s formula was used to calculate QTc as QT interval/RR interval^1/3. When applicable according to the above criteria, ECGs were processed using version 20.1 of the 12SL algorithm (GE CardioSoft V6.51) to automatically assess QTc. Gender-specific cut-offs for prolonged QTc of 450 ms for men and 460 ms for women, respectively, were applied.¹³

Statistics

Continuous variables were presented by means and standard deviations for normally distributed variables and medians and quartiles 1 and 3 (Q1;Q3) for non-normally distributed variables, while categorical variables were presented by frequencies and corresponding percentages. Groupwise comparisons were performed using independent samples t-tests or median tests for continuous variables and χ² tests or Fisher’s exact tests for categorical variables where appropriate. Analysis of covariance (ANCOVA) was utilized to investigate if increases in QTc were different across treatment groups independently of baseline QTc.

In survival analysis of the endpoints of first AF/AT recurrence at any time and first AF/AT recurrence after blanking, the Kaplan–Meier curves were grouped according to presence of prolonged QTc at baseline, and groupwise comparisons were made by log-rank tests. Multivariate Cox regression models were constructed to investigate the effect of prolonged QTc at baseline on AF/AT recurrence independently of the following variables: treatment group (amiodarone vs. placebo), gender, age, body mass index (BMI), LAESV, LVEF, diabetes, AF duration (years from diagnosis), and history of paroxysmal vs. persistent AF. These variables were chosen after groupwise comparisons and univariate Cox regression with the baseline variables in Table 1 and after consideration of known risk factors for recurrence after AF ablation.^2–4 To comply for possible interaction between treatment group and QTc at baseline, the analyses were repeated among the half of the population with the longest and shortest QTcs alone, and within each treatment group alone. Schoenfeld residuals were utilized to validate proportional hazards, and the linearity assumption for continuous variables was checked by Martingale residuals. Since amiodarone or placebo was given only for 8 weeks, the proportional hazard assumption was not met for the drug’s effect on first AF/AT recurrence at any time after ablation. Thus, this multivariate Cox regression model included two piecewise constant models for the effect of amiodarone.

The following supplementary analyses were performed: Instead of grouping by presence of prolonged QTc at baseline, log-rank tests and Cox regressions were performed according to prolonged QTc at 1 and 3 months, and presence of significant increase in QTc from baseline to 1 month (defined as mean increase in the amiodarone group; ΔQTc1) and significant decrease from 1 to 3 months (defined as mean decrease in the amiodarone group; ΔQTc3), respectively. Finally, these values of QTc and ΔQTc were analysed as continuous variables. The endpoint of these analyses was time to first AF/AT recurrence after blanking, and these analyses were repeated after stratification by treatment group.

The analyses were repeated with baseline QTc above or below median instead of above or below the gender-specific cut-off for prolonged QTc, and with Bazett’s formula for QT correction instead of Fridericia’s.

All analyses were performed with IBM SPSS Statistics 22 (IBM, Armonk, NY, USA), and the R software v3.4.1 (http://www.r-project.org/). A P-value of 5% was considered statistically significant.

Results

Population and follow-up

A total of 212 patients were included, of which 108 were assigned to amiodarone and 104 to placebo. Two patients were excluded due to withdrawal from the trial 1 day post-ablation (both placebo group). One patient discontinued study medication due to QTc prolongation >550 ms (amiodarone group), and one patient died during follow-up (placebo group), but these were included in the analyses. Consequently, the total population comprised 210 patients (Table 1).

In total, 121 of 210 patients (58%) had any documented AF/AT recurrence during follow-up. During the 3-months blanking period, 92 of 210 patients (44%) had AF/AT recurrence. Eight of the patients with recurrence within the blanking period were censored from the analysis of time to new AF/AT recurrence after blanking; four due to physician-prescribed open-label antiarrhythmic drug (three placebo, one amiodarone group), and four due to reablation within the blanking period (all placebo group). After blanking, 86 patients had new AF/AT recurrence.

Corrected QT interval changes during and after amiodarone treatment

No ECGs presented with bundle branch block or ventricular rhythms. Two patients had missing valid baseline QTc measurement. At baseline, QTc was mean 417 (±17), median 415 (401;434) ms, while 16 patients (7.6%) had prolonged QTc, and BMI and heart rate index differed between patients with prolonged and normal QTc (Table 1). After 1 month’s treatment, 36% of patients in the amiodarone group had prolonged QTc, compared to 3% in the placebo group, P < 0.0001 (Figure 1). From baseline to 1 month, QTc in the amiodarone group had in average increased by 27 (±30) ms to a mean value of 444 (±28) ms (Figure 2). At 3 months, QTc in the amiodarone group had decreased by 17 (±20) ms compared to 1 month, and at 6 months, no differences from baseline persisted, and there was no difference between amiodarone and placebo. In ANCOVA, amiodarone increased QTc from baseline to 1 month independently of baseline QTc, P < 0.0001.

Corrected QT interval and predictive factors for recurrence of arrhythmia

Baseline characteristics of patients with and without prolonged QTc at baseline are presented in Table 1. After the 3-months blanking period, new recurrence of AF/AT was detected in 63% of patients with prolonged QTc at baseline vs. 41% of patients with normal QTc (P = 0.10). Figure 3A and B present Kaplan–Meier curves for time to first AF/AT recurrence at any time and after blanking, respectively, according to presence of prolonged QTc at baseline. In the multivariate Cox regression model of time to first AT/AF recurrence at any time, amiodarone treatment decreased recurrences with hazard ratio (HR) 0.45, P < 0.001 (Table 2). For first recurrence after blanking, baseline QTc prolongation was significantly associated with increased recurrence rate with HR 2.06, P = 0.035, while none of the other pre-specified variables remained significant in prediction of recurrence among all patients (Table 3).

When the model was applied only to the 102 patients with baseline QTc below median, the 8 weeks of amiodarone treatment was associated with decreased rate of AF/AT recurrences after blanking with HR 0.45, P = 0.01 (Table 3). In this sub-population, 55% had been randomly assigned to amiodarone and 45% to placebo (P = 0.2). When the model was applied to the amiodarone group only, prolonged QTc at baseline still predicted recurrence after blanking with HR 5.04, P = 0.002 (Table 3). When considering only patients in the placebo group, or only with baseline QTc above median, respectively, QTc prolongation was not associated with recurrences. The results did not change after further adjustment for heart rate at baseline.

Regarding the supplementary analyses, neither presence of prolonged QTc at 1 or 3 months, nor significant ΔQTc1, nor significant ΔQTc3, were associated with recurrence after blanking, in log-rank tests (Supplementary material online, Figures S1–S4) or in Cox regressions. The same was true when only the amiodarone group was considered, and when the values of QTc and ΔQTc were analysed as continuous variables instead of categorical variables. Finally, when analysed with baseline QTc above or below median instead of above or below the cut-off for prolonged QTc, and with Bazetts’s formula for QT correction instead of Fridericia’s, the findings were identical in terms of statistical significance.

Discussion

The present study is the first to examine the impact of amiodarone on QTc as well as the recovery of ECG alterations after drug discontinuation. Furthermore, we present the first investigation of QTc as a marker for patient selection in AF ablation.

Our main findings are: (i) QTc increased with a mean of 27 ms at 1 month after standard loading and continuous doses of oral amiodarone, and normalized completely during 4 months after drug discontinuation. (ii) Prolonged QTc at baseline was independently associated with AF/AT recurrence rate, and when considering only the patients with baseline QTc below median, 8 weeks of amiodarone treatment post-ablation was effective in halving the rate of AF/AT recurrence after blanking.

Utilizing the placebo-controlled design of the AMIO-CAT trial, we sought to quantify the extent of amiodarone-mediated changes in QTc. Previously, Hohnloser et al.⁷ investigated intravenous administration of amiodarone in 77 patients during the first 4 days following coronary artery bypass surgery and found that QTc increased by nearly 20%. This is more than three times what we find in 1 month of oral dosing. However, this study utilized Bazett’s formula for QTc calculation, which leads to overestimation of QTc during faster heart rates, keeping in mind that the amiodarone patients also had significant changes in heart rate during treatment. Comparable to our findings, the EMIAT investigators reported a 20 ms increase in QTc after amiodarone using a dosing regimen similar to AMIO-CAT in more than 800 patients with myocardial infarction. However, QTc was measured by automated Holter-recordings which might not be comparable to the 12-lead ECGs at baseline.⁸

In the AMIO-CAT trial, we hypothesized that repellence of early recurrences by short-term amiodarone after ablation would decrease long-term recurrences in accordance with the ‘AF begets AF’-principle.¹⁴ Although the trial found that 8 weeks of oral amiodarone decreased AF/AT recurrences and AF-related hospitalizations during blanking, this effect ceased after the first 3 months.¹² In the present analysis, we demonstrate that when only patients with baseline QTc below the median of 415 ms were considered, amiodarone did in fact decrease the risk of long-term recurrences, even though statistical power was halved. This finding was independent of known risk factors for AF/AT recurrences.^2–4 In this manner, the QT interval identified patients with higher likelihood of benefitting from short-term amiodarone therapy following AF ablation.

Although amiodarone decreased recurrences in these patients while lengthening the QT interval, baseline QTc prolongation was itself a predictor of recurrence, independently of LVEF as well as known risk factors for recurrence. In this regard, two retrospective studies from the Beijing Anzhen Hospital recently reported that QTc prolongation at baseline was associated with recurrence rate after AF ablation in 39 patients with hypertrophic cardiomyopathy and 134 patients with diabetes mellitus, respectively.^10,11 However, these studies only included 24-h Holter monitorings, and all patients received various anti-arrhythmic drugs following ablation, though this was not adjusted for in the analyses. Furthermore, QTc was calculated by Bazett’s formula, and the studies did not include timewise changes in QTc.

Whether QTc is associated with AF has also been investigated in other populations: longer QTc has been associated with incident AF in patients with cryptogenic stroke,¹⁵ and in a large population-based study the extremes of short and prolonged QTc were both associated with incident AF.⁹

We found that the extent of amiodarone-mediated QTc prolongation was not associated with AF/AT recurrence after blanking, though patients receiving amiodarone had both significant QTc prolongation and less recurrence within the blanking period. Together with the effect from baseline prolonged QTc this emphasizes that an underlying QTc prolongation, though merely reflecting ventricular repolarization, might be useful as a marker of electrical/structural remodelling in the atria,^9,15 while also being a marker of poor response to ablation and amiodarone treatment.

Contrary to our finding, Huang et al.¹⁶ reported that the actual extent of QTc prolongation was associated with decreased recurrences after pharmacological cardioversion by another Class III antiarrhythmic drug, dofetilide, in patients with persistent AF. However, this was a retrospective single-centre study utilizing short-term in-hospital QTc alterations as the predictor variable, while half of the study population discontinued dofetilide before discharge or during follow-up. Previous smaller studies have shown no effect from baseline QTc or pharmacologically prolonged QTc on AF recurrence after cardioversion.^17,18

Limitations

First, the trial was not designed to investigate QTc as a predictor of outcome, and hence the findings are exploratory and should be interpreted as such. We attempted to accommodate potential bias by adjusting for factors known to increase the risk of arrhythmia recurrence, as well as possible confounders for QTc prolongation such as LVEF. Second, as follow-up for endpoints included serial ECGs and serial 3-day Holter-monitorings, we are limited by not having continuous rhythm monitoring in this population and thus additional asymptomatic recurrences could have happened without our knowledge. However, the patients in the trial had a history of symptomatic AF and symptom-driven extra monitoring was obtained in all patients, increasing the chance of confirming AF/AT recurrence. Third, we were not able to adjust for atrial fibrosis, inflammation, or atrial hormones; factors that might be associated with outcome after AF ablation.

We chose Fridericia’s formula for QT correction because this better predicts clinical outcomes,¹⁹ and Fridericia’s formula has more conservative heart rate correction than Bazett’s, keeping in mind that this study included patients with a history of AF and were randomized to a treatment with significant impact on heart rate. We furthermore chose to control that our findings were independent of heart rate. We finally repeated our analyses with Bazett’s formula and found identical results in terms of significance, though then 29 patients (14%) fulfilled the criteria for prolonged QTc at baseline. When appropriate, we utilized automatic QTc measurements to decrease interobserver variability.²⁰

Conclusions

In this study of 210 patients referred for AF ablation, we found that short-term oral amiodarone increased QTc with nearly 30 ms compared to placebo, and ECG alterations decreased rapidly after drug discontinuation. Presence of prolonged QTc before ablation doubled the risk of AF/AT recurrence independently of other risk factors. Among patients with shorter QTc at baseline, 8 weeks of amiodarone treatment was associated with decreased rate of recurrences during 6 months of follow-up. Thus, QTc might represent a marker for improved patient selection and antiarrhythmic drug therapy in AF ablation.

Acknowledgements

We thank Tor Biering-Sørensen, MD, PhD, for assistance with echocardiographic analysis and our study nurses for helping with patient follow-up.

Funding

AMIO-CAT was supported by the Danish Heart Foundation, Copenhagen, Denmark (09-04-R72-A2408-22545, 10–04-R78-A2929-22588, and 11–04-R84-A3230-22650) and The Heart Centre Research Committee at Rigshospitalet, Copenhagen, Denmark.

Conflict of interest: none declared.

References