## Abstract

Aims In patients with persistent atrial fibrillation (AF), the efficacy and safety of two anti-arrhythmic drugs in preventing the recurrence of AF after successful direct current (DC) cardioversion was prospectively assessed in a multi-centre double-blind, placebo-controlled, randomised trial using daily trans-telephonic monitoring.

Methods and results 1182 patients with persistent AF were prospectively enrolled, 848 patients were successfully cardioverted and then randomised to either sotalol (383 patients), quinidine plus verapamil (377 patients) or placebo (88 patients). The primary outcome parameter was AF recurrence or death. All patients received an event recorder (Tele-ECG) and had to record and transmit via telephone at least one ECG per day during follow-up.

The mean follow-up period was 266 days. A total of 191,103 Tele-ECGs were recorded and transmitted. The primary outcome parameter (AF recurrence of any kind or death) was observed in 572 patients (67%) in whom at least one episode of AF recurrence was documented during follow-up, in 348 patients (41%) AF recurrence was persistent. The recurrence rates after one year for any AF were 83% for placebo, 67% for sotalol and 65% for quinidine plus verapamil, the latter being statistically superior to placebo but not different from sotalol. The recurrence rates for the secondary outcome parameter persistent AF were 77%, 49% and 38%, respectively. Quinidine plus verapamil was significantly superior to placebo and to sotalol. About 95% of all AF recurrences were initially detected in the daily Tele-ECG, about 70% of all AF recurrences occurred completely asymptomatic. Adverse events on sotalol and quinidine plus verapamil were comparable with the exception that all torsade de pointes tachycardias occurred on sotalol.

Conclusion Anti-arrhythmic treatment after DC cardioversion of persistent AF significantly decreases the recurrence rates of persistent AF compared to placebo with superiority of quinidine plus verapamil compared to sotalol. Symptoms were not reliable as clinical surrogates to detect episodes of AF.

## Introduction

In persistent atrial fibrillation (AF), present guidelines 1,2 recommend electrical cardioversion and treatment with anti-arrhythmic drugs to maintain sinus rhythm as a major approach. However, guidelines have become quite restrictive concerning the type of recommended drug. Their selection is mostly based on experience with the treatment of other types of arrhythmias. For AF, they often refer to either retrospective, open-labelled or uncontrolled studies or controlled trials with insufficient statistical power. In clinical routine, sotalol has become a frequently prescribed drug in many countries but quinidine is still used on a varying scale. However, to date there has been no prospective controlled trial with sufficient power comparing these two drugs in a well defined group of patients with persistent AF.

Use of quinidine often requires some additional medication to slow the conduction of the atrio-ventricular node. Therefore, it has frequently been combined with either digitalis or verapamil whereas the AV nodal blocking effects of sotalol would generally not necessitate additional drugs that slow the AV node. Earlier experimental work suggested 3 that the combination of quinidine and verapamil might be superior to quinidine alone. The same group has suggested that the addition of verapamil to quinidine might increase its clinically efficacy and lower the rate of side effects.4 Recent clinical observations 5 suggest that electrical remodelling of the atrium may be attenuated by verapamil. In addition, in the experimental setting, verapamil might reduce the propensity of quinidine to induce after-depolarisations.14–17 Based on these findings and due to safety concerns on the use of quinidine with verapamil, this trial was initiated in 1997 using a prospective, double-blind, randomised, multi-centre, parallel group comparison. Its aim was to assess both the frequency of recurrences of AF in cases with persistent AF following cardioversion during long-term treatment using more sensitive algorithms for detection of AF than previously used, and the efficacy and safety of two anti-arrhythmic drugs (sotalol and quinidine plus verapamil) in reducing the frequency of AF recurrence.

## Methods

PAFAC was a multi-centre double-blind randomised trial using daily trans-telephonic monitoring.

### Definitions

For the purpose of this trial, the following definitions were used:

For inclusion of the patients, persistent AF was defined as AF lasting more than seven days. Since at the time of the trial, 7-day Holter monitoring using 24-h technology was not considered to be feasible, evidence that AF was continuously present was obtained by history and by electrographic documentation. To meet this criterion, persistent AF was considered to be highly likely if two electrocardiographic documentations, including at least one 24-h long-term ECG of AF, on separate occasions during the 6 months before inclusion in the study were provided and no episode of sinus rhythm was reported.

During follow-up, the definition of AF had to take the new detection devices (Tele-ECG) into account. Since daily trans-telephonic ECG recordings (see below) had not been used in large multi-centre trials, special definitions for the follow-up of patients had to be used for the purpose of this trial. Therefore, AF recurrence was classified as follows: paroxysmal AF was any documented AF of 30 s duration or more,2 recorded either by a Tele-ECG, an ECG at rest or a Holter ECG. The aim was to exclude very short episodes of atrial tachyarrhythmias lasting only a few seconds.

According to present guidelines2 which were not available when this trial was designed, paroxysmal atrial fibrillation may last between 30 s and 7 days but the guidelines concede that most episodes of this type last less than 24 h. Although the guidelines state that persistent AF usually lasts more than 7 days, there is a period between day 2 when most episodes of paroxysmal AF terminate and day 7 where there is an uncertainty in the definitions proposed. According to the diagnostic methods available, our original protocol defined any AF lasting longer than 24 h as “chronic” which would now have to be replaced by “persistent”. Thus, our up-dated definition of persistent AF is any AF on a Holter ECG that lasts 24 h or more without spontaneous cessation. Any episode of persistent AF, or in the case that, paroxysmal AF had been documented in at least 3 Tele ECGs, was considered as a failure of study medication, which was then terminated. Such 3 episodes of paroxysmal AF were considered to be equivalent to persistent AF and were thus counted as such.

### Patient selection

Male and female patients between 18 and 80 years of age with documented persistent AF and a clinical indication for direct current (DC) cardioversion were included. Patients had to be adequately treated with anti-coagulants (INR levels above 2.0) at least in the three weeks prior to inclusion. The most important exclusion criteria were: heart failure NYHA IV, myocardial infarction or cardiac surgery within the past 3 months, valvular defects requiring surgery, implanted pacemakers or cardioverters/defibrillators, QTc prolongation

$$>$$
0.42 in the absence of anti-arrhythmic medication, congenital long QT syndrome, and serum K+ levels repeatedly
$${<}$$
4.5mval. All anti-arrhythmic agents, including beta-blockers, diltiazem, verapamil and gallopamil, digoxin, tricyclic antidepressants, antihistamines, and all drugs known to cause QT prolongation were prohibited as concomitant medication. Permitted drugs were digitoxin, anticoagulants, diuretics and medications to replace electrolyte deficiencies.

### Study procedures

After written informed consent, eligible patients were hospitalised for a minimum of three days. DC cardioversion with up to three attempts was performed without any class I or III anti-arrhythmic drugs (withdrawn for at least five half-lives). In case of successfully restored sinus rhythm, the patient was randomised on the same day after having been in sinus rhythm for at least 2 h without additional administration of anti-arrhythmic drugs.

Patients with failed cardioversion or with “immediate” recurrences of AF (within the first two hours) were not randomised. The active drugs were sotalol (80 mg tid beginning on the day after cardioversion and 160 mg bid thereafter) and the fixed combination of quinidine plus verapamil (

$$q$$
+
$$v$$
) (160 mg quinidine plus 80 mg verapamil bid on the day after cardioversion and tid thereafter). A third group received placebo in double-dummy technique. Amiodarone was not used as a comparator in this setting because of its different application patterns which would have caused a very difficult blinded randomisation. Patients were discharged on day 4, thereby entering the recurrence prevention phase. The minimal follow-up period in case of successful maintenance of sinus rhythm was one year. In case of sinus rhythm after one year, the patient was able to enter the extended follow-up period if both the patient and the physician agreed. Blinded medication was also provided for the extended follow-up period. If persistent AF or the third episode of paroxysmal AF occurred before one year of follow-up, study medication and follow-up were terminated. The status of those patients after one year was collected by questionnaire.

At baseline, and at each monthly follow-up visit, a physical examination, a specified interrogation for symptoms and an ECG at rest were obtained. Laboratory tests, an echocardiogram and a Holter ECG were performed at baseline. In addition, daily trans-telephonic ECG transmissions were obtained from each patient during the total follow-up period.

### Randomisation

The randomisation list was created by an independent organisation which was not involved in the conduct of the study. The sealed randomisation list was kept by the Drug Safety Department of Knoll Deutschland. Patients were randomised in blocks.

Each investigator received a set of sealed random code envelopes for the patients scheduled for enrolment at his/her site. Each envelope had a randomisation number and contained details of the study medication allocated to the patient assigned that number. In case of an adverse event, the random code envelope for a particular patient could only be opened if this was absolutely necessary for medical reasons. All random code envelopes were collected by the monitors at the end of the study. If the random code was broken, the date, time, reasons for opening, and name of the person responsible for opening the envelope had to be documented in the case report form and on the data sheet contained in the envelope.

### ECG monitoring

A credit card-sized Tele-ECG recorder (RhythmCardTM, Instromedix®, USA) was dispensed to each patient. All patients were asked to record and transmit at least one ECG at any time per day, independent of symptoms. One minute of a single-lead electrocardiogram was recorded and stored in the ECG recorder. The ECG could subsequently be transmitted to a central laboratory using any regular telephone. A custom-made computer-based automated system with voice control assisted the patient in the transmission process. It also asked for any symptoms or other remarks from the patient. The data were analysed by specially trained medical staff. In case of AF, a fax was immediately sent to the responsible study site to call the patient for an additional visit for detection of related symptoms and for the further specification of AF by a Holter ECG. In case of any other clinically relevant abnormalities detected in the Tele-ECG recording, the responsible study site was immediately informed using similar procedures.

### Outcome parameters

The primary outcome parameter was the time to the first recurrence of any type of AF or death. Secondary outcome parameters included the occurrence of and the time to persistent AF, and thus the time to discontinuation of treatment, the number of recurrences, and target parameter-associated symptoms (irregular heart beat, symptoms of tachycardia, dizziness, and syncope). In case of early termination of study medication without reaching a primary outcome parameter, e.g. due to adverse events, the patient was followed until a primary outcome parameter was reached. All patients who had experienced the end-point for study termination (i.e. recurrence of per study-defined persistent AF) before the planned 12-months period received a final questionnaire to assess their outcome after one year.

### Ethics

All procedures were followed in accordance with the ethical standards of the responsible committees on human experimentation and with the Helsinki Declaration. The trial was approved by the Ethics Committee of the institution of the principal investigator as well as by all Ethics Committees of participating institutions.

### Statistical analysis

This study was conducted as a prospective, double-blind, randomised, multi-centre, parallel group comparison with three treatment arms (q+v vs. sotalol vs. placebo). All variables are presented stratified by treatment group using descriptive summary tables. All time-to-event data are presented graphically (Kaplan–Meier estimation of the survival curves). All data are presented on an intention-to-treat basis. Data are given as mean (SD) or median values as appropriate.

For confirmatory analysis, three hypotheses were planned to be tested in a strictly hierarchical order to maintain a multiple significance level of 5%:6 H01 – the one year recurrence rate on

$$q+v$$
is equal to or higher than on placebo (test for superiority to placebo, one-sided,
$${\alpha}=0.025$$
); H02 – the one year recurrence rate on
$$q$$
+
$$v$$
is at least 10% higher than on sotalol (test for non-inferiority to sotalol, one-sided,
$${\alpha}=0.05$$
); H03 – the one year recurrence rate on
$$q$$
+
$$v$$
is equal to or higher than on sotalol (test for superiority to sotalol, one-sided,
$${\alpha}=0.05$$
). The time-to-event curves for the primary outcome parameter were analysed using the Cox proportional hazard model. The hazard ratio (HR) between placebo and
$$q+v$$
as well as the corresponding two-sided 95% confidence interval was calculated using its lower limit (LLplacebo,sotalol) to test H01 at the significance level of 2.5%. The HR between sotalol and
$$q$$
+
$$v$$
as well as the corresponding two-sided 90% confidence interval (CI) were calculated using its lower limit (LLsotalol, q+v) for the tests of H02 and H03 at the significance level of 5% each. Therefore, H01 was rejected when LLplacebo,sotalol
$$>$$
1. H02 was rejected when LLsotalol, q+v
$$>$$
lnon-inf. Under the proportional hazard assumption, the HR lnon-inf was determined by lnon-inf=ln(psotalol)/ln(psotalol-0.1) where psotalol is the event-free rate on sotalol at day 365 from the survival function and the pre-defined non-inferiority limit 0.1. H03 is rejected when LLsotalol, q+v
$$>$$
1. The sample size estimation was based on the following assumptions: recurrence rate at 12months on
$$q+v$$
50%, on sotalol 50%, and on placebo 75%, recruitment duration 2years, study duration 3years, non-inferiority margin between the active treatment arms 10%, level of significance
$${\alpha}$$
and power 1- for test1:
$${\alpha}=0.025$$
, 1-
$${\beta}=0.9$$
, for test 2 and test 3:
$${\alpha}=0.05,1{-}{\beta}=0.9$$
and a randomisation ratio for
$$q+v$$
:sotalol:placebo of 4:4:1. Assuming an annual dropout rate of 20%, the calculations revealed 760 patients to be randomised including a minimum of 84 patients on placebo.

## Results

### Patient population

A total of 1 182 patients were enrolled in 68 centres in Germany (756 pts) and in 31 centres in the Czech Republic (426 pts). 334 patients (28%) were not randomised due to unsuccessful cardioversion or recurrence of AF⩽2 h after cardioversion, 848 patients (72%) were randomised and received at least one dose of study medication: 88 patients (10%) were assigned to placebo, 383 patients (45%) received sotalol and 377 patients (45%) were treated with

$$q$$
+
$$v$$
. Demographic data are presented in Table 1 . No relevant differences in baseline data between the three treatment groups could be detected.

Table 1

Demographic and baseline data

Placebo
$$(n=88)$$

Sotalol
$$(n=383)$$

Quinidine + Verapamil
$$(n=377)$$

Total
$$(n=848)$$

Mean age±SD [years] 62±8.5 62±9.5 63±9.0 63± 9.2
Gender [male %/female %] 71/29 66/34 65/35 66/34
History of AF/AF symptoms
Mean±SD [days] 620±1415 412±816 487±938 467±949
Third quartile [days] 350 365 395 365
LA diameter [mean±SD (mm)] 45±6 44±6 45±6 45±6
FS [mean±SD (%)] 29±13 30±12 31±12 30±12
Previous DC cardioversions [pts. (%)] 11 (13) 38 (10) 43 (11) 92 (11)
Previous anti-arrhythmic agents of class I or III [pts. (%)] 18 (21) 64 (17) 55 (15) 137 (16)
Cardiac glycosides [pts. (%)] 52 (59) 250 (65) 267 (71) 569 (67)
History of rheumatic fever [pts. (%)] 4 (5) 4 (1) 10 (3) 18 (2)
Valvular regurgitation [pts. (%)] 33 (38) 149 (39) 179 (47) 361 (43)
Aortic valve regurgitation [pts. (%)] 5 (6) 27 (7) 35 (9) 67 (8)
Mitral valve regurgitation [pts. (%)] 18 (21) 82 (21) 97 (26) 197 (23)
Tricuspid valve regurgitation [pts. (%)] 10 (11) 40 (10) 47 (13) 97 (11)

Coronary artery disease
Angina pectoris [pts. (%)] 17 (19) 80 (21) 81 (22) 178 (21)
CAD confirmed by angiography 2 (2) 28 (7) 25 (7) 55 (7)
Previous MI [pts. (%)] 5 (6) 20 (5) 18 (5) 43 (5)

Heart failure 52 (59) 228 (60) 241 (64) 521 (61)
NYHA I [pts. (%)] 15 (17) 68 (18) 83 (22) 166 (20)
NYHA II [pts. (%)] 32 (36) 144 (38) 148 (39) 324 (38)
NYHA III [pts. (%)]

5 (6)

16 (4)

10 (3)

31 (4)

Placebo
$$(n=88)$$

Sotalol
$$(n=383)$$

Quinidine + Verapamil
$$(n=377)$$

Total
$$(n=848)$$

Mean age±SD [years] 62±8.5 62±9.5 63±9.0 63± 9.2
Gender [male %/female %] 71/29 66/34 65/35 66/34
History of AF/AF symptoms
Mean±SD [days] 620±1415 412±816 487±938 467±949
Third quartile [days] 350 365 395 365
LA diameter [mean±SD (mm)] 45±6 44±6 45±6 45±6
FS [mean±SD (%)] 29±13 30±12 31±12 30±12
Previous DC cardioversions [pts. (%)] 11 (13) 38 (10) 43 (11) 92 (11)
Previous anti-arrhythmic agents of class I or III [pts. (%)] 18 (21) 64 (17) 55 (15) 137 (16)
Cardiac glycosides [pts. (%)] 52 (59) 250 (65) 267 (71) 569 (67)
History of rheumatic fever [pts. (%)] 4 (5) 4 (1) 10 (3) 18 (2)
Valvular regurgitation [pts. (%)] 33 (38) 149 (39) 179 (47) 361 (43)
Aortic valve regurgitation [pts. (%)] 5 (6) 27 (7) 35 (9) 67 (8)
Mitral valve regurgitation [pts. (%)] 18 (21) 82 (21) 97 (26) 197 (23)
Tricuspid valve regurgitation [pts. (%)] 10 (11) 40 (10) 47 (13) 97 (11)

Coronary artery disease
Angina pectoris [pts. (%)] 17 (19) 80 (21) 81 (22) 178 (21)
CAD confirmed by angiography 2 (2) 28 (7) 25 (7) 55 (7)
Previous MI [pts. (%)] 5 (6) 20 (5) 18 (5) 43 (5)

Heart failure 52 (59) 228 (60) 241 (64) 521 (61)
NYHA I [pts. (%)] 15 (17) 68 (18) 83 (22) 166 (20)
NYHA II [pts. (%)] 32 (36) 144 (38) 148 (39) 324 (38)
NYHA III [pts. (%)]

5 (6)

16 (4)

10 (3)

31 (4)

Abbreviations: AF, atrial fibrillation; LA, left atrial; FS, fractional shortening of the left ventricle; DC, direct current; CAD, coronary artery disease.

### Efficacy

The median duration (3rd quartile in brackets) of follow-up on study medication was 16 (88) days in the placebo group, 102 (377) days for the patients treated with sotalol and 129 (478) days for the patients on

$$q$$
+
$$v$$
. A total of 191,103 Tele-ECGs were received with 95% valid for complete analysis.

The primary outcome parameter (AF of any kind or death) was observed in 572 patients (67%) within the first 12 months of follow-up: 73 patients (83%) in the placebo group, 255 patients (67%) on sotalol and 244 patients (65%) on

$$q$$
+
$$v$$
treatment. Death as part of the combined primary end-point occurred in 11 patients within the first 12 months (details are given in the safety section). The corresponding Kaplan–Meier estimates for the probability of remaining free of the primary outcome parameter are shown in Fig. 1 . The HR to reach the primary outcome parameter until month 12 for placebo versus
$$q$$
+
$$v$$
was 1.74 (95% confidence interval (CI): 1.34, 2.25) and for sotalol versus
$$q$$
+
$$v$$
0.97 (90% CI: 0.84, 1.12). As the survival curves question the proportional hazards assumptions, in particular for the two active treatment arms, a supportive analysis was performed based on the difference of the Kaplan–Meier estimates of event-free probabilities after 12 months of follow-up, leading to comparable results. The difference between
$$q$$
+
$$v$$
and placebo was 18% (95% CI: 9%, 27%), between
$$q$$
+
$$v$$
and sotalol 3% (90% CI: −3%, 9%). In accordance with the biometrical model of the trial, the superiority of
$$q$$
+
$$v$$
to placebo was statistically proven (the CI for the HR falls above 1, and the CI for the difference above 0%), as well as the non-inferiority to sotalol treatment (the CI for the HR falls above the non-inferiority margin 0.746, and the CI for the difference above the corresponding non-inferiority margin −10%). However, superiority of
$$q$$
+
$$v$$
to sotalol could not be shown. Multivariate analyses to assess risk factors and subgroup analyses as well as the analysis of time-treatment interactions of hazards will be reported elsewhere.

Fig. 1

Kaplan–Meier estimates for the probability of remaining free of first recurrence of atrial fibrillation or death. The numbers of patients at risk are stated in grey boxes for one and two years of follow-up.

Fig. 1

Kaplan–Meier estimates for the probability of remaining free of first recurrence of atrial fibrillation or death. The numbers of patients at risk are stated in grey boxes for one and two years of follow-up.

The probability of remaining free of persistent AF (=secondary outcome parameter) was much higher than for staying free of any recurrence of AF (Fig. 2 ). After one year, 62% of patients on

$$q$$
+
$$v$$
and 51% on sotalol were free of persistent AF compared to 23% on placebo.
$$q$$
+
$$v$$
had a 39% (95% CI: 28%, 50%) reduced probability of persistent AF compared to placebo, and a 12% (90% CI: 5%, 18%) reduced probability compared to sotalol. Hence,
$$q$$
+
$$v$$
were significantly superior to both placebo and sotalol. This combined secondary outcome parameter was based on paroxysmal AF (three recurrences in the Tele-ECG) in 3% of placebo patients (most developing persistent AF immediately if there was a recurrence) but in 16% and 14% of patients on sotalol and
$$q$$
+
$$v$$
, respectively. All other recurrences were due to persistent AF on Holter.

Fig. 2

Kaplan–Meier estimates for the probability of remaining free of persistent atrial fibrillation. The numbers of patients at risk are stated in grey boxes for one and two years of follow-up.

Fig. 2

Kaplan–Meier estimates for the probability of remaining free of persistent atrial fibrillation. The numbers of patients at risk are stated in grey boxes for one and two years of follow-up.

Another secondary outcome parameter was the occurrence of symptoms at the time of documented episodes of AF: only 181 of 605 patients (30%) with recurrences of any type of AF at any time during follow-up reported at least one symptom (Table 2 ), whereas in about 70% of documented recurrences, symptoms were not reported. No differences in the percentage of symptomatic and asymptomatic episodes of AF were found between the three groups.

Table 2

Symptoms and their relation to the occurrence of atrial fibrillation episodes

All patients

Placebo (
$$n=88$$
) (%)

Sotalol (
$$n=383$$
) (%)

Quinidine+verapamil (
$$n=377$$
) (%)

Total (
$$n=848$$
) (%)

Patients with available data

$$n=86$$

$$N=354$$

$$N=350$$

$$n=790$$

Recurrence of AF No Yes No Yes No Yes No Yes

Any symptom
No (%) 11 59 21 54 25 51 21 54
Yes (%)

2

28

2

23

3

21

2

23

All patients

Placebo (
$$n=88$$
) (%)

Sotalol (
$$n=383$$
) (%)

Quinidine+verapamil (
$$n=377$$
) (%)

Total (
$$n=848$$
) (%)

Patients with available data

$$n=86$$

$$N=354$$

$$N=350$$

$$n=790$$

Recurrence of AF No Yes No Yes No Yes No Yes

Any symptom
No (%) 11 59 21 54 25 51 21 54
Yes (%)

2

28

2

23

3

21

2

23

### Safety

A total of 1,682 adverse events occurred in 615 patients (73%). The majority of these cases were minor such as palpitations (21%), dizziness (15%), and symptoms of tachycardia (8%). The adverse events were equally distributed in the sotalol (73%) and the

$$q$$
+
$$v$$
(75%) groups. The placebo population experienced somewhat less adverse events (61%) but still to a marked extent. The following parameters were slightly but not significantly different between sotalol and
$$q$$
+
$$v$$
. Bradycardia:10% vs. 4%; QT interval prolongation: 2% vs. 6%; diarrhoea: 2% vs. 6%; peripheral oedema: 2% vs. 5%.

Serious adverse events (including new or prolonged hospitalisation as reason for notification) occurred with equal frequency (23% of patients on placebo, 25% on sotalol and 24% on

$$q$$
+
$$v$$
). Table 3 shows the incidence of “serious adverse events of special interest”. A total of 11 deaths of any kind occurred on study medication (i.e., in case of the termination of study medication within 3 half lives). Due to early recurrences of AF, the placebo group had a shorter observation interval until reaching persistent AF (mean number of days 16 versus 102 and 129 on sotalol or
$$q$$
+
$$v$$
, respectively). The mode of death was not significantly different between sotalol or
$$q$$
+
$$v$$
. The CEC considered 4 deaths on sotalol as possibly or probably related to the intake of study medication versus 5 deaths on
$$q$$
+
$$v$$
. Syncopes occurred more frequently on
$$q$$
+
$$v$$
than on sotalol. The most apparent difference was the occurrence of torsade de pointes tachycardia which all nine occurred in patients on sotalol. In case of more than one event, the most severe one was assigned. Thus, in the sotalol group, one patient experienced a torsade de pointes tachycardia resulting in unsuccessful resuscitation and death, this would increase the total number of documented torsade de pointes tachycardias to ten.

Table 3

Serious adverse events of special interest

Placebo (
$$n=88$$
) (%)

Sotalol (
$$n=383$$
) (%)*

Quinidine+verapamil (
$$n=377$$
) (%)

Total (
$$n=848$$
) (%)

Death – 6 (1.6) 5 (1.3) 11 (1.3)
Resuscitation – – 1 (0.3) 1 (0.1)
Syncope 1 (1.1) 3 (0.8) 6 (1.6) 10 (1.2)
TdP tachycardia – 9 (2.3) – 9 (1.1)
Ventricular fibrillation – 2 (0.5) 1 (0.3) 3 (0.4)
Ventricular tachycardia (run
$$>$$
10 beats)
2 (2.9) 2 (0.5) 4 (1.1) 8 (0.9)
Total

3 (3.4)

22 (5.7)

17 (4.5)

42 (5.0)

Placebo (
$$n=88$$
) (%)

Sotalol (
$$n=383$$
) (%)*

Quinidine+verapamil (
$$n=377$$
) (%)

Total (
$$n=848$$
) (%)

Death – 6 (1.6) 5 (1.3) 11 (1.3)
Resuscitation – – 1 (0.3) 1 (0.1)
Syncope 1 (1.1) 3 (0.8) 6 (1.6) 10 (1.2)
TdP tachycardia – 9 (2.3) – 9 (1.1)
Ventricular fibrillation – 2 (0.5) 1 (0.3) 3 (0.4)
Ventricular tachycardia (run
$$>$$
10 beats)
2 (2.9) 2 (0.5) 4 (1.1) 8 (0.9)
Total

3 (3.4)

22 (5.7)

17 (4.5)

42 (5.0)

As defined by the DSMC. All events belonging to this category are listed below. Abbreviations: DSMC, Data and Safety Monitoring Committee; TdP, Torsade de pointes.

The duration of drug use until a serious adverse event occurred, showed a broad distribution for death (median 107 days, range 3 to 529 days) and syncope (median 5 days, range 9 to 1,088 days). However, all 3 (100%) lethal episodes of ventricular fibrillation as well as 6 (67%) torsade de pointes tachycardias and 4 (50%) ventricular tachycardias with

$$>$$
10 echoes occurred until day 4 on drug therapy. Thus, 65% of all proarrhythmic episodes and potentially life-threatening events happened at the beginning of drug therapy.

### Long-term follow-up beyond study medication

Patients, in whom study medication had been stopped for any reason before one year after inclusion, received a questionnaire. Thirteen deaths had occurred of whom 2 patients had previously been on placebo, 7 on sotalol, and 4 on

$$q$$
+
$$v$$
. None of these deaths showed a temporal relationship to the previous intake of study medication. Three patients had died suddenly, one died after acute myocardial infarction, 3 due to terminal heart failure, 1 due to pulmonary embolism, 2 due to major bleedings, 2 related to cancer, and 1 due to a malignant histiocytoma. Eight (85%) of these deaths occurred free of anti-arrhythmic treatment, 3 patients died on anti-arrhythmic drugs other than the study medication (amiodarone, flecainide), in one patient the treatment prior to death was unknown. Thus, the total incidence of deaths while on study medication (11/848 = 1.3%) was comparable with the mortality rate after termination of study participation (13/577 = 2.3%), however, the duration during which patients were at risk differed.

## Discussion

To our knowledge, this is the largest trial comparing two anti-arrhythmic drugs versus placebo for the maintenance of sinus rhythm after electrical cardioversion using a multi-centre randomised, double-blind, parallel-group design. The high propensity to AF in the study population is evidenced by the high recurrence rate in the placebo group despite the fact that these patients had only relatively minor increases in left atrial size and a relatively low proportion of significant underlying heart disease (Table 1). Nevertheless, the recurrence rates in the placebo group were comparable to previous experience.

There were three major results emerging from this trial. First, the combination of

$$q$$
+
$$v$$
as well as sotalol was more effective than placebo in preventing any recurrences of type of AF although at a disappointingly low level. In contrast, regarding recurrence of persistent AF,
$$q$$
+
$$v$$
was more effective than placebo or sotalol. Of note, documented recurrences of AF occurred earlier and more frequently than indicated by corresponding symptoms. About 70% of all documented episodes of AF were asymptomatic and only detected in the daily Tele-ECG recording. This indicates the high sensitivity of this approach. These and other findings7–10 question the role of symptoms as clinical surrogate parameters for detection of AF which affects not only the definition of different types of AF but also to subsequent therapeutic strategies. For example, the decision for immediate DC cardioversion without sufficient anticoagulation in case of seemingly `recent onset AF' may be based on symptoms that occurred recently. This might ignore asymptomatic chronic AF that could have persisted for several weeks or even months. Hence, there is an urgent need for re-evaluation of clinical definitions and detection modes. Future trials should use a better documentation of relapses of AF if the arrhythmia itself is the target of evidence. Our experience shows that the use of daily telemetric event recording is highly accepted by the patient, feasible in the long term and affordable in large-scale trials.

The second major result was that

$$q$$
+
$$v$$
was more effective than sotalol for prevention of recurrences into persistent AF after DC cardioversion. Despite apparent differences in study design, from varying population characteristics and divergent follow-up periods to distinct application patterns, another possible explanation for divergent findings in the literature11,12 is the mode of documentation of recurrences by ECG recordings and symptoms. Slowing of ventricular rate in AF causes less symptomatic AF episodes.9,13 Thus, we may speculate that in trials that relied on symptoms for the detection of recurrences, it was more the occurrence of rapid irregular heart rates rather than the arrhythmia itself that was measured. As a consequence, the anti-arrhythmic effects of sotalol might have been over-estimated due to rate-controlling effects of the ß-blocking properties of the drug.

The third major result of this trial is that both active drugs had a comparable safety profile. The only exception was that all 10 episodes of torsade de pointes tachycardias occurred on sotalol. This result is noteworthy, as quinidine has become the most restricted anti-arrhythmic drug after the publication by Coplen et al.12 Our findings, however, demonstrate that the risk of torsade de pointes tachycardias was markedly higher on sotalol compared to

$$q$$
+
$$v$$
treatment. It has to be emphasised that in the present trial quinidine was administered as a fixed combination with verapamil and results can therefore not be extrapolated to quinidine alone. Experimental and clinical data14–17 suggest that verapamil is able to suppress after-depolarisations produced by anti-arrhythmic drugs of class I and III, which are likely to lead to torsade de pointes tachycardia. Based on these findings there is emerging evidence that verapamil might at least be partly responsible for the lower incidence of pro-arrhythmic events on the combination of
$$q$$
+
$$v$$
compared to quinidine therapy alone. It is noteworthy, that about 65% of all pro-arrhythmic and potentially life-threatening adverse events occurred during the first 4 days of treatment with anti-arrhythmic drugs. This finding raises the question whether observation with continuous electrocardiographic monitoring should be recommended as mandatory during initiation of anti-arrhythmic therapy. Comparable findings and suggestions have already been reported.18

## Limitations

When our trial was designed in 1996, today's commonly used definitions of AF (paroxysmal, persistent, permanent) had not been established.19 We have therefore now adapted the definitions used in our original protocol to those recommended in today's guidelines. Additionally, the use of daily ECG recordings required special definitions of AF recurrence that were not concordant with daily clinical routine. However, the modified definitions did not alter the results of the analysis.

In our trial, electrical cardioversion was performed without concomitant anti-arrhythmic drugs that facilitate cardioversion and stabilize sinus rhythm immediately afterwards. However, the use of anti-arrhythmic drugs for cardioversion would have necessarily led to a more complicated trial design with additional treatment arms to avoid a systematic bias based on differing responses to cardioversion.

## Conclusions

In patients with persistent AF, the combination of

$$q$$
+
$$v$$
proved more effective in preventing recurrence of persistent AF than sotalol or placebo. Addition of verapamil to class I anti-arrhythmic drugs may increase efficacy as has recently been suggested by a randomised open-label study that showed increased efficacy if verapamil was added to amiodarone or flecainide.20 This occurred at a comparable rate of side-effects of both meditations except for higher rates of occurrence of torsade de pointes tachycardias on sotalol. Thus, based on the results of this carefully performed multi-centre double-blind randomised trial, the use of quinidine in combination with verapamil should be re-considered. The detection modes for AF should also be re-considered.

## Appendix

### Committees

The Executive Committee of the Investigators (ECI) consisted of the representatives of the investigators and the sponsor. It was responsible for approval of the study protocol, the amendments and the analysis plan, and for all other decisions during the trial. The following two committees consisted of independent experts not participating in the trial. The Critical Event Committee (CEC) reviewed all serious adverse events notified by the local investigators on the basis of detailed patient records blinded to treatment group. The judgements were reported to the Data and Safety Monitoring Committee (DSMC) that monitored the progress of the trial, the consistency with the study protocol, and all safety aspects. The members of these committees are listed in the Appendix. The trial was assisted by the IKKF (Institute for Clinical Cardiovascular Research), Munich, Germany.

In this phase III trial, representatives of the sponsoring company could participate in the ECI meetings depending on the issue to be discussed; one representative is co-author (H.K.P.). However, at decision making only one person from the company was a voting member in the ECI, which therefore had a majority of investigators. The chairman of the ECI and the members of the writing committee were investigators who also developed and proposed the trial design and were responsible for the scientific leadership in all phases of the trial.

### Contribution to the report

The authors of this report are the members of the Executive Committee of the Investigators. Each of them has contributed to the planning, design, and leadership of the trial, to the interpretation of the results, and the writing of the manuscript.

### Conflict of interest statement

None of the authors except for H.P.K. has any conflict of interest to be disclosed that could inappropriately influence their work reported here. The members of the Executive Committee of Investigators received expense allowances for their work done as well as reimbursement of travel expenses for attending committee meetings from the sponsor. H.P.K. was employed by the sponsor and as such responsible as project manager for the trial. He participated as a voting member in the Executive Committee of the Investigators which therefore had a majority of investigators.

### Role of the funding source

The trial was entirely funded by Knoll Deutschland GmbH (now: Abbott GmbH & Co. KG). Independent monitoring was done by Staticon International, Munich. Auditing was performed at the contract research organisation Staticon International and the participating study centres by authorised staff from Knoll Deutschland or a contract research organization. All audits were documented and archived. Statistical analysis was done by Staticon International under control of the investigators (P.B.). Daily and trans-telephonic ECGs were received and analysed by the Biosignal Analysis Laboratory at the Department of Cardiology and Angiology and the Institute for Arteriosclerosis Research (BALCAM), University of Münster for which these institutions received a financial grant from the sponsor to cover costs for staff, telephone fees, and the custom-made equipment.

### Position, institution, and location of all authors

Dr. Thomas Fetsch, Universitätsklinik Münster, Münster, Germany (present address: Institut für klinisch-kardiovaskuläre Forschung (IKKF), München, Germany).

Prof. D. Peter Bauer, Institut für Medizinische Statistik, Universität Wien, Austria.

Prof. Dr. Rolf Engberding, M.D., Städt. Klinikum Wolfsburg, Wolfsburg, Germany.

Dr. Hans P. Koch, Abbott GmbH & Co. KG (formerly: Knoll Deutschland GmbH), Ludwigshafen, Germany.

Prof. Dr. Jan Lukl, FN s Poliklinikou, Olomouc, Czech Republic.

Prof. Dr. Thomas Meinertz, Universitätsklinikum Eppendorf, Hamburg, Germany.

Prof. Dr. Michael Oeff, Städtisches Klinikum Brandenburg, Brandenburg, Germany.

Prof. Dr. Ludger Seipel, Universitätsklinikum Tübingen, Germany.

Prof. Dr. Hans J. Trappe, Marienhospital I, Herne, Germany.

Prof. Dr. Norbert Treese, Marienkrankenhaus, Osnabrück, Germany.

Prof. Dr. Günter Breithardt, Universitätsklinik Münster, Münster, Germany.

### The Executive Committee of the Investigators (ECI)

G. Breithardt (Münster, Germany) – Chairman and Principal Investigator, T. Fetsch (Münster, Germany) – Head of BALCAM, R. Engberding (Wolfsburg, Germany), H. P. Koch (Ludwigshafen, Germany), J. Lukl (Olomuc, Czech Republic), H. W. Müller (Ludwigshafen, Germany), M. Oeff (Brandenburg, Germany), N. Treese (Osnabrück, Germany), H. J. Trappe (Herne, Germany).

### The Critical Event Committee (CEC)

L. Seipel (Tübingen, Germany) – Chairman, M. Meesmann (Würzburg, Germany), H. F. Pitschner (Bad Nauheim, Germany), E. Vester (Düsseldorf, Germany).

### The Data and Safety Monitoring Committee (DSMC)

T. Meinertz (Hamburg, Germany) – Chairman, P. Bauer (Vienna, Austria) – biometrical advisor, K. H. Kuck (Hamburg, Germany), M. Schlepper (Bad Nauheim, Germany), J. Witte (Berlin, Germany).

### Statistical center

P. Bauer (Vienna, Austria).

### The study centres in order of performance

#### Hospital name, city, investigator/co-investigator (number of patients randomised)

We thank Mrs. Gisela Burschel (MSc.), Mrs. Bärbel Gaede (MTA), and their co-workers at BALCAM (Biosignal Analysis Laboratory at the Department of Cardiology, Münster) for the excellent work in the co-ordination of the analysis of the Tele-ECGs and Holter recordings and evaluations. We also thank Mrs. Elisabeth Freund (MSc.) at Staticon International and her team for the excellent work they did in guiding and monitoring the trial throughout its course. We thank all members of the Data and Safety Monitoring Board and of the Critical Event Committee. We would also like to thank Wilfried Krümke, MD, at Knoll Deutschland GmbH for his continual interaction and support.

## References

[1]
Prystowsky
EN
, Benson DW Jr., Fuster V, et al. Management of patients with atrial fibrillation. A Statement for Healthcare Professionals. From the Subcommittee on Electrocardiography and Electrophysiology, American Heart Association.
Circulation
.
1996
;
93
:
1262
–1277
[2]
Fuster
V
, Ryden LE, Asinger RW, et al. ACC/AHA/ESC guidelines for the management of patients with atrial fibrillation. A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the European Society of Cardiology Committee for Practice Guidelines and Policy Conferences (Committee to develop guidelines for the management of patients with atrial fibrillation) developed in collaboration with the North American Society of Pacing and Electrophysiology.
Eur. Heart J
.
2001
;
22
:
1852
–1923
[3]
Guelker
H
, Bender F, Heuer H, et al. Enhancement of the antifibrillatory efficacy of quinidine in atrial fibrillation by verapamil.
Z Kardiol
.
1982
;
71
:
31
–34
[4]
Heilmann
E
, Bender F, Bachour G, et al. Combined treatment of auricular fibrillation and other tachycardiac arrhythmias using quinidine and verapamil.
Med. Welt
.
1972
;
23
:
1792
–1794
[5]
Tieleman
RG
, De Langen C, Van Gelder IC, et al. Verapamil reduces tachycardia-induced electrical remodeling of the atria.
Circulation
.
1997
;
95
:
1945
–1953
[6]
Bauer
P
, Roehmel J, Maurer W, et al. Testing strategies in multi-dose experiments including active control.
Statist. Med
.
1998
;
17
:
2133
–2146
[7]
Page
RL
, Wilkinson WE, Clair WK, et al. Asymptomatic arrhythmias in patients with symptomatic paroxysmal atrial fibrillation and paroxysmal supraventricular tachycardia.
Circulation
.
1994
;
89
:
224
–227
[8]
Anderson
JL
, Gilbert EM, Alpert BL, et al. Prevention of symptomatic recurrences of paroxysmal atrial fibrillation in patients initially tolerating antiarrhythmic therapy. A multicenter, double-blind, crossover study of flecainide and placebo with transtelephonic monitoring. Flecainide Supraventricular Tachycardia Study Group.
Circulation
.
1989
;
80
:
1557
–1570
[9]
Wolk
R
, Kuhakowski P, Karczmarewicz S, et al. The incidence of asymptomatic paroxysmal atrial fibrillation in patients treated with propranolol or propafenone.
Int. J. Cardiol
.
1996
;
54
:
207
–211
[10]
Page
RL
, Tilsch TW, Connolly SJ, et al. Asymptomatic or silent atrial fibrillation: frequency in untreated patients and patients receiving azimilide.
Circulation
.
2003
;
107
:
1141
–1145
[11]
Juul-Moller
S
, Edvardsson N, Rehnqvist-Ahlberg N. Sotalol versus quinidine for the maintenance of sinus rhythm after direct current conversion of atrial fibrillation.
Circulation
.
1990
;
82
:
1932
–1939
[12]
Coplen
SE
, Antman EM, Berlin JA, et al. Efficacy and safety of quinidine therapy for maintenance of sinus rhythm after cardioversion. A meta-analysis of randomized control trials.
Circulation
.
1990
;
82
:
1106
–1116
[13]
Murgatroyd
FD
, Gibson SM, Baiyan X, et al. Double-blind placebo-controlled trial of digoxin in symptomatic paroxysmal atrial fibrillation.
Circulation
.
1999
;
99
:
2765
–2770
[14]
Takanaka
C
, Singh BN. Barium-induced nondriven action potentials as a model of triggered potentials from early after depolarizations: significance of slow channel activity and differing effects of quinidine and amiodarone.
J. Am. Coll. Cardiol
.
1990
;
15
:
213
–221
[15]
Zhou
JT
, Zheng LR, Liu WY. Role of early after depolarization in familial long QTU syndrome and torsade de pointes.
Pacing Clin. Electrophysiol
.
1992
;
15
:
2164
–2168
[16]
Shimizu
W
, Ohe T, Kurita T, et al. Effects of verapamil and propranolol on early after depolarizations and ventricular arrhythmias induced by epinephrine in congenital long QT syndrome.
J. Am. Coll. Cardiol
.
1995
;
26
:
1299
–1309
[17]
Liao
WB
, Bullard MJ, Kuo CT, et al. Anticholinergic overdose induced torsade de pointes successfully treated with verapamil.
Jpn. Heart J
.
1996
;
37
:
925
–931
[18]
Maisel
WH
, Kuntz KM, Reimold SC, et al. Risk of initiating antiarrhythmic drug therapy for atrial fibrillation in patients admitted to a university hospital.
Ann. Int. Med
.
1997
;
127
:
281
–284
[19]
Gallagher
MM
, Camm J. Classification of atrial fibrillation.
Am. J. Cardiol
.
1998
;
82
:
18N
–28N
[20]
De Simone
A
, De Pasquale M, De Matteis C, et al. Verapamil plus antiarrhythmic drugs reduce atrial fibrillation recurrences after an electrical cardioversion (VEPARAF Study).
Eur. Heart J
.
2003
;
24
:
1425
–1429