Proarrhythmic events in atrial fibrillation: A multivariate analysis of the AFFIRM study

Proarrhythmia has become increasingly recognized as a hazard of antiarrhythmic drug therapy.1 Certain populations are at higher risk when exposed to certain types of drugs. Women are more likely than men to develop prolonged QT intervals and torsades de pointes ventricular tachycardia.2 Patients with advanced structural heart disease are at high risk for proarrhythmia, especially when exposed to sodium channel blocking medications.3,4 Awareness that certain types of patients are at high risk when exposed to specific types of drugs has allowed clinicians to exercise caution when choosing antiarrhythmic drugs for their patients. We analyzed the risk of proarrhythmia in patients with atrial fibrillation who were enrolled in the rhythm-control arm (exposure to antiarrhythmic drugs for the maintenance of sinus rhythm) of the Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) study to determine the risk of proarrhythmia when using today’s safety guidelines.5-7

Patients and methods

Details about patient selection for AFFIRM have been published previously.5-7 Patients randomized to the rhythm-control strategy form the basis of this analysis. Strategy assignment was open-label: patients received antiarrhythmic drug therapy at the choice of their physicians, provided that this choice fell within the safety guidelines specified by the AFFIRM protocol, based on standard safe clinical practice (Table). Doses were selected based on renal and hepatic function, and electrocardiograms were monitored for changes. Choices of antiarrhythmic drugs included quinidine (Cardioquin, Quinaglute), procainamide (Procan SR, Pronestyl), disopyramide (Norpace), flecainide (Tambocor), propafenone (Rythmol), moricizine (Ethmozine), sotalol (Betapace, Sorine), amiodarone (Cordarone, Pacerone), and when it became available toward the end of the study, dofetilide (Tikosyn). Some patients received ibutilide (Corvert) briefly in the hospital for attempted chemical cardioversion. If a patient failed a drug because of recurrent atrial fibrillation or intolerance, his or her clinician could choose another agent.

Data collected at follow-up visits included clinical characteristics, antiarrhythmic drug exposures, electrocardiographic measurements, and arrhythmic events. Multivariate analysis was used to determine the effects of clinical characteristics on the risk of proarrhythmia. Some analyses were performed only on subjects exposed to the drugs thought to be most likely to cause proarrhythmia (quinidine, procainamide, disopyramide, sotalol, ibutilide, and dofetilide).

Results

Of the 2,033 patients randomized to rhythm control, the mean age was 69.7 ± 9 years; 71% had hypertension, 39% had a history of coronary artery disease, 23% had a history of heart failure, and 19% had diabetes. The ejection fraction was normal in 73%, mildly abnormal in 13%, and below 40% in 14% of patients. Patients were followed up for an average of 3.5 years. Their exposure to antiarrhythmic drugs at any time during follow-up was as follows: amiodarone, 63%; sotalol, 41%; propafenone, 14%; flecainide, 8%; procainamide, 8%; quinidine 7%; disopyramide, 4%; moricizine, 2%; dofetilide, 0.6%; and ibutilide, 0.2%.

The cumulative incidence of arrhythmic events (cardiac arrest, torsades de pointes ventricular tachycardia, sustained ventricular tachycardia, and arrhythmic death) was 7% at 6 years. The Figure shows the time to first ventricular arrhythmic event in the 1,047 patients who were exposed to high-risk drugs, including quinidine, procainamide, disopyramide, sotalol, dofetilide, and ibutilide. At 5 years, the Kaplan-Meier estimate of the incidence of proarrhythmia was 5% in women and 5% in men. Multivariate analysis showed that heart failure increased the risk of proarrhythmia at 5 years (9.7% compared with 4.0% in those without heart failure). Mitral regurgitation and age older than 65 years were also associated with increased risk. After adjusting for significant covariates, female sex did not increase the risk of proarrhythmia. Torsades de pointes ventricular tachycardia occurred in 1.0% of women and 0.4% of men. This arrhythmia was usually associated with marked QT interval prolongation and often with bradycardia, hypokalemia, or hypomagnesemia.

Discussion

In this study, the incidence of proarrhythmia was low (5% at 5 years in the patients taking high-risk medications), representing only a small portion of the adverse outcomes (total mortality in the rhythm-control arm was 27%). The 0.6% incidence of torsades de pointes ventricular tachycardia was much lower than the 2% to 10% incidence reported in the earlier literature.8-13 Our findings likely reflect the careful drug-for-patient selection, electrocardiographic monitoring, and dose adjustments performed as part of the AFFIRM protocol.5

One reason for the low overall number of proarrhythmic events may have been the common use of amiodarone, which is rarely associated with proarrhythmia, and the relatively infrequent use of Vaughan Williams Class I drugs.14 Sotalol, dofetilide, and the Vaughan Williams Class IA drugs have substantial risk of torsades de pointes ventricular tachycardia. Furthermore, Class I agents are associated with increased risk of mortality in patients with underlying structural heart disease.3,4,15 In the AFFIRM population, Class I agents were used in only about one third of patients, and the protocol specified which groups of patients could not receive them. For example, patients with coronary artery disease or left ventricular dysfunction were prohibited from receiving Class IC drugs, and physicians were discouraged from using any Class I drugs in patients with low ventricular ejection fractions.5

Torsades de pointes ventricular tachycardia occurred in 7 of 771 women (1.0%) and in 5 of 1,262 men (0.4%). Importantly, 9 of these 12 events occurred late (not within 72 hours of drug initiation). This finding highlights the importance of avoiding late development of hypokalemia, hypomagnesemia, bradycardia, or metabolically

mediated (eg, due to worsening renal function or drug interaction) increases in antiarrhythmic drug levels. Although we found no statistically significant increased risk of torsades de pointes in women, the low number of torsades events may have prevented the difference from becoming statistically significant.

Conclusions

In the rhythm-control arm of the AFFIRM study, high-risk patients with atrial fibrillation were exposed to antiarrhythmic medications with proarrhythmic potential. The AFFIRM protocol specified guidelines for the safe use of these drugs, including drug selection, electrocardiographic monitoring, and dose adjustments based on hepatic and renal function. The overall risk of proarrhythmia was low. Although the risk of proarrhythmia continues to be an important cause for concern, clinicians can reduce this risk through informed, appropriate selection of the right drug for the right patient.