Author + information
- Received June 23, 2003
- Revision received October 1, 2003
- Accepted October 20, 2003
- Published online April 7, 2004.
- Craig M Pratt, MD, FACC*,* (, )
- Steven N Singh, MD, FACC†,
- Hussein R Al-Khalidi, PhD‡,
- Jose M Brum, MD‡,
- Michael J Holroyde, PhD‡,
- Stephen R Marcello, MD, FACC‡,
- Peter J Schwartz, MD, FACC§,
- A.John Camm, MD, FACC∥,
- ALIVE Investigators
- ↵*Reprint requests and correspondence:
Dr. Craig M. Pratt, Baylor College of Medicine, 6565 Fannin Street, F1001, Houston, Texas 77030, USA.
Objectives The purpose of this study was to assess the effect of oral azimilide dihydrochloride (AZ) 100 mg versus placebo on the onset, termination, and prevalence of atrial fibrillation (AF) in a subpopulation of patients in the Azimilide Postinfarct Survival Evaluation (ALIVE) trial.
Background Previous clinical trials have demonstrated the antiarrhythmic effects of AZ in patients with AF. Azimilide was investigated for its effects on mortality in patients with depressed left ventricular (LV) function after recent myocardial infarction (MI) and in a subpopulation of patients with AF.
Methods A total of 3,381 post-MI patients with depressed LV function were enrolled in this randomized, placebo-controlled, double-blind study of AZ 100 mg on all-cause mortality. A total of 93 patients had AF on the baseline 12-lead electrocardiogram (ECG). An additional 27 patients developed AF after initially being in sinus rhythm at randomization. These patients were identified through 12-lead ECGs obtained during routine visits at week 2, months 1, 4, 8, and 12.
Results Patients with AF at baseline had a higher mortality than those without AF (p = 0.0006). Among AF patients, there was no difference in mortality between AZ patients and placebo patients (p = 0.82). Fewer AZ patients developed AF than placebo patients (p = 0.04). More AZ patients than placebo patients converted to sinus rhythm, but this difference did not achieve statistical significance (p = 0.076). Over one-year follow-up, more AZ patients were in sinus rhythm than placebo patients (p = 0.04).
Conclusions Azimilide was safe and effective AF therapy in patients with depressed LV function after an MI.
The clinical consequence of atrial fibrillation (AF) is amplified in patients with left ventricular (LV) dysfunction and clinical congestive heart failure (1). In patients with a low LV ejection fraction (LVEF), AF is associated with a higher incidence of congestive heart failure and decreased survival, in both males and females (2). Such high-risk patients are frequently more resistant to pharmacologic or electrical cardioversion and more prone to AF recurrence (3). The use of many antiarrhythmic drugs is not indicated in such patients either because of the lack of efficacy data or concern regarding proarrhythmic risk and/or negative inotropic effects (4).
Azimilide dihydrochloride (AZ) is a Vaughan-Williams class III antiarrhythmic drug, which is being developed primarily for AF (5)and also as adjunctive antiarrhythmic therapy in patients with implantable cardioverter defibrillators (6). Azimilide has a unique electrophysiologic profile possessing both IKrand IKschannel blocking properties. In previous trials, AZ at doses of 100 mg and 125 mg daily prolonged the time to first symptomatic recurrence of AF and atrial flutter (5,7).
In general, the clinical experience with antiarrhythmic drugs in patients with AF associated with moderate-to-severe heart failure is limited (8). One approach to increase the knowledge of specific antiarrhythmic drugs in these high-risk patients is to conduct subpopulation analyses of safety and efficacy in AF patients participating in large, randomized, placebo-controlled mortality trials. The goal of this paper is to expand this knowledge to include AZ, by analysis of a recently completed mortality trial.
The Azimilide Postinfarct Survival Evaluation (ALIVE) trial (9)was a placebo-controlled, randomized, prospective trial of patients with a recent acute myocardial infarction (MI) and in whom LV systolic dysfunction was present. The primary end point was total mortality focused on high-risk patients. The results showed a neutral association of AZ to mortality (primary efficacy analysis: p = 0.74; hazard ratio [HR] = 0.95; 95% confidence interval [CI] = 0.71 to 1.27). An important prespecified objective of the ALIVE trial was to assess the effect of AZ 100 mg versus placebo on the onset, termination, and prevalence of AF.
This report concerns an analysis of a subgroup of patients in the ALIVE trial who had AF at randomization or developed AF during the one-year follow-up. A total of 3,717 patients were enrolled in the ALIVE trial. Briefly, the inclusion criteria for ALIVE required a measured LVEF of 15% to 35% in patients five to 21 days after acute MI. The randomized groups included: placebo (n = 1,690); AZ 100 mg (n = 1,691); and an additional lower AZ dose group of 75 mg (n = 336), which was discontinued due to enrollment difficulties. These latter patients are not considered further in the analyses presented, which focused on the remaining 3,381 patients followed for one year.
At randomization, a total of 93 patients had AF on the baseline 12-lead electrocardiogram (ECG). They are the primary focus of this analysis. An additional 27 patients developed AF after initially being in sinus rhythm at randomization. These patients were identified through 12-lead ECGs that were obtained during routine visits at week 2 as well as months 1, 4, 8, and 12. The AF case report form was not designed to capture symptoms.
There were several prespecified analyses conducted for patients with documented AF captured by 12-lead ECG at randomization as well as those who were free of AF at baseline, but developed the arrhythmia during the trial, which were assessed through routine 12-lead ECGs. The primary prespecified end points were: 1) for those not in AF at baseline, time-to-first documented AF event during any visit (n = 27); 2) for those in AF at baseline (n = 93), time-to-first spontaneous conversion to sinus rhythm captured by the 12-lead ECG during any visit; and 3) the percent of patients with AF at baseline who were in sinus rhythm at each subsequent visit. The latter analysis considers the prevalence of AF over time in each randomized study group at defined intervals throughout the one-year trial.
Analyses are reported according to the intention-to-treat principle. For time-to-first event analysis, Kaplan-Meier estimates were used with a log-rank test to compare the probability of freedom from recurrent AF, and the Cox proportional hazards model to calculate hazard ratios and 95% CI (10). Repeated measure analysis utilizing generalized estimating equations (11)was used for the prevalence of AF among treatment groups during the one-year follow-up.
In addition, the Cox proportional hazards regression analysis of total mortality was used among all patients in AF (at randomization vs. those patients not in AF) to explore the effect of AF on all-cause mortality, after adjustment was made for age, sex, beta-blocker, and angiotensin-converting enzyme inhibitor usage, LVEF, congestive heart failure, diabetes, hypertension, and New York Heart Association (NYHA) class.
Of the total of 3,381 patients randomized in the ALIVE trial, 120 had AF during the trial, 93 patients at randomization, and an additional 27 patients developed AF during the trial.
The baseline demographic characteristics of the ALIVE patients in sinus rhythm at randomization, compared with those patients with AF, are presented in Table 1. Atrial fibrillation patients were older (mean age, 68 vs. 60 years); more likely to have symptomatic (NYHA class II/III) congestive heart failure (72% vs. 52%); had a higher frequency of hypertension (69% vs. 55%); and less beta-blocker (60% vs. 74%) and aspirin use (73% vs. 89%) (Table 1).
The influence of AF on total mortality is presented by Kaplan-Meier estimate in Figure 1A. Cox proportional hazards model was used to estimate the hazard ratio and the 95% CIs before and after statistical adjustment for relevant covariates (age, gender, hypertension, heart failure, LVEF, diabetes, NYHA class, beta-blocker, and angiotensin-converting enzyme use) in Figure 1B. In both analyses, the patients with AF had a higher mortality (unadjusted: log-rank p = 0.0006, HR = 2.1, 95% CI = 1.4 to 3.3; adjusted: p = 0.075, HR = 1.5, 95% CI = 0.96 to 2.34). Although, the adjusted analysis achieved borderline statistical significance, the AF patients after adjustment had a 50% higher chance of death than patients not in AF at baseline.
Among AF patients, there was no difference in mortality between placebo and AZ patients (p = 0.82; HR = 1.1; 95% CI = 0.5 to 2.7). This was consistent with the overall neutral mortality results of the ALIVE trial (p = 0.74; HR = 0.95; 95% CI = 0.71 to 1.27) (Fig. 2).
Several prespecified analyses were used to investigate the effect of 100 mg AZ compared with placebo on AF. A demographic comparison of the randomized patients with AF is presented in Table 2. The two treatment groups were generally well balanced. However, the AZ group was characterized by a higher percentage of diabetes and hypertension. Despite these trends, none of these characteristics achieved statistical significance.
Figure 3depicts the time-to-new documented onset of AF among patients not in AF at baseline. Fewer AZ patients developed AF than placebo patients (8 of 1,630 [0.5%] vs. 19 of 1,648 [1.2%], respectively; p = 0.04; HR = 0.43; 95% CI = 0.19 to 0.99).
Further analysis considers the time-to-first spontaneous conversion to sinus rhythm, which occurred in 15 of 56 (26.8%) of AZ patients compared with 4 of 37 (10.8%) placebo patients (p = 0.076, HR = 2.63, 95% CI = 0.87 to 7.93) (Fig. 4).
An additional analysis included the prevalence of the arrhythmia at designated time intervals during the trial after randomization among those patients in AF at baseline. This analysis is depicted in Figure 5. Over one-year follow-up of patients in AF at baseline, a repeated-measure analysis of the AF prevalence showed more AZ patients compared with placebo patients were in sinus rhythm (p = 0.04).
Serious adverse events were similar in AZ and placebo patients (Table 3). The pattern of adverse events did not differ from the ALIVE trial as a whole. However, among patients not in AF at baseline, there were fewer AZ patients with new or worsening heart failure than placebo patients (97 of 1,630 [6%] vs. 127 of 1,648 [8%], respectively; p = 0.05). In the total ALIVE population, 5 of 1,691 (0.3%) AZ patients and 1 of 1,690 (0.06%) placebo patients developed torsade de pointes based on event committee assessment. Severe neutropenia (absolute neutrophil count <500 cells/μl) occurred in 15 of 1,691 (0.9%) AZ patients and 4 of 1,690 (0.2%) placebo patients. Azimilide-associated neutropenia had a predictable onset time (19 to 48 days) and reversed after drug withdrawal. No patients developed life-threatening infections. There was only one torsade de pointes case (AZ patient) and no severe neutropenia cases among AF substudy.
The ALIVE trial was a multinational, randomized, double-blind, prospective trial of AZ versus placebo in surviving MI patients with moderate-to-severe LV systolic dysfunction, over half of whom had NYHA II/III congestive heart failure. As such, it included patients with concomitant AF at randomization as well as patients who developed AF during the one-year trial.
The major conclusions based on the analysis of these high-risk patients with AF at baseline or those who developed AF during the trial are: 1) fewer AZ patients developed AF than placebo patients (p = 0.04); 2) more AZ patients converted from AF to sinus rhythm than placebo patients, but this difference did not achieve statistical significance (p = 0.076); and 3) over one-year follow-up of patients in AF at baseline, a repeated-measure analysis of the AF prevalence showed more AZ patients were in sinus rhythm compared with placebo patients (p = 0.04). These observations are consistent with data from previous AZ randomized studies in AF population (5).
In the ALIVE trial, the analysis of AF was considered an important and prespecified primary subpopulation analysis. These results can be compared with those of dofetilide and amiodarone and their respective mortality trials, the Danish Investigators of Arrhythmia and Mortality on Dofetilide trial (DIAMOND) (dofetilide) (12–14)and the Survival Trial of Antiarrhythmic Therapy in Congestive Heart Failure (CHF-STAT) (amiodarone) (15,16). These analyses were pivotal as the basis for recommending only these two drugs as initial therapy for AF in patients with LV systolic dysfunction and clinical congestive heart failure in the current AF management guidelines (17). All these substudies are comprised of AF patients with cardiomyopathy and clinical congestive heart failure. The placebo-subtracted efficacy rates in these three trials were remarkably similar for maintenance of sinus rhythm: AZ, 28%; dofetilide, 25%; amiodarone, 23%. Direct comparisons are imperfect with some differences in the severity of heart failure and mean follow-up, but the extent of LV dysfunction is similar.
The increased risk of AF and mortality in congestive heart failure patients is well established (2,18). In the Framingham experience, heart failure increased the risk of developing AF 4.5-fold in men and 5.9-fold in women, after multivariate adjustments (1). Additionally, in the DIAMOND trials, AF was >3 times more prevalent (25.8% vs. 7.6%) in the DIAMOND congestive heart failure population than in the DIAMOND MI population. Atrial fibrillation was also associated with an increased risk of death, by approximately 25% in DIAMOND congestive heart failure patients, compared with patients in sinus rhythm (12–14).
The potential reasons for this increased mortality with AF include hemodynamic compromise due to lack of atrial contraction, the potential effect of tachyarrhythmic cardiomyopathy, and embolic consequences. The mortality results of ALIVE are consistent with these estimates, with the AF population having a higher risk of death, even after statistical adjustment for relevant covariates affecting mortality. Importantly, mortality in AF patients on AZ was comparable with placebo.
The ALIVE population is similar to DIAMOND (dofetilide) and CHF-STAT (amiodarone), with a mean LVEF of 29%, more than half of whom had NYHA II/III clinical congestive heart failure and a substantial (15% at one year) placebo mortality.
To clarify the risk/benefit analysis, one should consider the incidence of torsade de pointes ventricular tachyarrhythmia reported on these three drugs were: dofetilide, 32 of 1,511 (2.12%); amiodarone, 0 of 336 (0%); and AZ, 5 of 1,691 (0.3%). An additional risk of severe neutropenia occurred more frequently in AZ (0.9%) compared with placebo patients (0.2%) in ALIVE. However, no patients developed life-threatening infections.
In conclusion, this analysis demonstrated that the efficacy of 100 mg AZ in AF patients in the ALIVE trial is consistent with the results of a meta-analysis of AZ in the management of symptomatic, recurrent AF from four clinical trials with 1,380 patients (7)showing that AZ doses of 100 and 125 mg/day significantly prolonged the time to arrhythmia recurrence. Furthermore, in the AF trials of AZ, those patients with a history of ischemic heart disease or heart failure had a greater treatment effect from AZ than patients without a history of those conditions (5).
Additionally, the mortality in AF patients on AZ and placebo was comparable and consistent with results of the ALIVE trial (5). Furthermore, AZ patients in the ALIVE trial were less likely than placebo patients to develop new or worsening congestive heart failure. This finding is consistent with the results of a study in patients with implantable cardioverter defibrillators (mean LVEF = 30.6%) showing no effect of AZ on LVEF (6,9). The torsade de pointes and severe neutropenia occurred more frequently in AZ patients compared with placebo patients (0.3% vs. 0.06% and 0.9% vs. 0.2%, respectively) as reported in previous studies. Torsade de pointes was associated with well-established risk factors such as female gender, low potassium, and diuretic therapy. Severe neutropenia had a predictable early onset time (19 to 48 days) and could be identified by weekly blood monitoring during the first two to three months of dosing.
The results of ALIVE, together with the subpopulation analyses of AF patients, indicate that AZ is an effective and safe therapy for AF in patients with depressed LV function and an MI.
The authors thank the 489 clinical investigators who provided and cared for study patients.
☆ Supported by a grant from the Health Care Research Center, Procter & Gamble Pharmaceuticals, Inc., Cincinnati, Ohio. Drs. Pratt and Camm are consultants for the scientific and regulatory development of azimilide and served as co-chair and chair, respectively, on the ALIVE Steering Committee. Dr. Singh was a member of the ALIVE Executive Medical Committee, and Dr. Schwartz was a member of the ALIVE Steering Committee.
Presented as an abstract at the 75th Annual American Heart Association's Scientific Sessions, November 2002.
- atrial fibrillation
- Azimilide Postinfarct Survival Evaluation
- azimilide dihydrochloride
- Survival Trial of Antiarrhythmic Therapy in Congestive Heart Failure
- confidence interval
- Danish Investigators of Arrhythmia and Mortality on Dofetilide trial
- hazard ratio
- left ventricle/left ventricular
- left ventricular ejection fraction
- myocardial infarction
- New York Heart Association
- Received June 23, 2003.
- Revision received October 1, 2003.
- Accepted October 20, 2003.
- American College of Cardiology Foundation
- Benjamin E.J.,
- Wolf P.A.,
- D'Agostino R.B.,
- et al.
- Vaziri S.M.,
- Larson M.G.,
- Benjamin E.J.
- Prystowsky E.N.,
- Benson D.W. Jr.,
- Fuster V.,
- et al.
- Pritchett E.L.C.,
- Page R.L.,
- Connolly S.J.,
- et al.
- ↵Singer I, Al-Khalidi H, Niazi I, et al. Azimilide decreases recurrent ventricular tachyarrhythmias in patients with ICDs. J Am Coll Cardiol. In Press.
- Slavik R.S.,
- Tisdale J.E.,
- Borzak S.
- Collett D.
- Davis C.S.
- Køber L.,
- Bloch-Thomsen P.E.,
- Møller M.,
- et al.,
- Danish Investigators of Arrhythmia and Mortality on Dofetilide (DIAMOND) Study Group
- Pedersen O.D.,
- Bagger H.,
- Keller N.,
- et al.
- Fuster V.,
- Ryden L.E.,
- Asinger R.W.,
- et al.
- Khairy P.,
- Nattel S.