Abstract

Aims

It is uncertain whether the benefit from apixaban varies by type and duration of atrial fibrillation (AF).

Methods and results

A total of 18 201 patients with AF [2786 (15.3%) with paroxysmal and 15 412 (84.7%) with persistent or permanent] were randomized to apixaban or warfarin. In this pre-specified secondary analysis, we compared outcomes and treatment effect of apixaban vs. warfarin by AF type and duration. The primary efficacy endpoint was a composite of ischaemic or haemorrhagic stroke or systemic embolism. The secondary efficacy endpoint was all-cause mortality. There was a consistent reduction in stroke or systemic embolism (P for interaction = 0.71), all-cause mortality (P for interaction = 0.75), and major bleeding (P for interaction = 0.50) with apixaban compared with warfarin for both AF types. Apixaban was superior to warfarin in all studied endpoints, regardless of AF duration at study entry (P for all interactions >0.13). The rate of stroke or systemic embolism was significantly higher in patients with persistent or permanent AF than patients with paroxysmal AF (1.52 vs. 0.98%; P = 0.003, adjusted P = 0.015). There was also a trend towards higher mortality in patients with persistent or permanent AF (3.90 vs. 2.81%; P = 0.0002, adjusted P = 0.066).

Conclusion

The risks of stroke, mortality, and major bleeding were lower with apixaban than warfarin regardless of AF type and duration. Although the risk of stroke or systemic embolism was lower in paroxysmal than persistent or permanent AF, apixaban is an attractive alternative to warfarin in patients with AF and at least one other risk factor for stroke, regardless of the type or duration of AF.

See page 2429 for the editorial comment on this article (doi:10.1093/eurheartj/eht205)

Introduction

One of the most serious complications of atrial fibrillation (AF) is stroke.1 Not only is stroke prevalent in patients with AF, but when it occurs, it is either fatal or associated with substantial disability in the majority of patients.2–5 Whether the risk of stroke is affected by the type, duration, and frequency of AF has been debated for years. Although a few studies have suggested a lower risk of stroke in patients with paroxysmal AF than those with persistent or permanent AF, a pooled analysis of the Stroke Prevention in Atrial Fibrillation (SPAF) trials demonstrated a comparable risk of stroke in paroxysmal and permanent AF.6–8 Practice guidelines on AF management made identical recommendations regarding stroke prevention for all types of AF, based on known risk factors for stroke in this patient population.9,10 These recommendations are further supported by data from the Euro Heart Survey and a pooled analysis of the SPORTIF (Stroke Prevention with the Oral Thrombin Inhibitor in Atrial Fibrillation) III and V randomized clinical trials that showed a similar risk of stroke in paroxysmal AF and persistent or permanent AF.11,12

Until recently, oral anticoagulation with vitamin K antagonists (e.g. warfarin) was the only effective treatment available for stroke prevention in AF. Over the past few years, several new oral anticoagulants that specifically inhibit thrombin or factor Xa have emerged.13–16 In the ARISTOTLE (Apixaban for Reduction in Stroke and Other Thrombo-embolic Events in Atrial Fibrillation) trial, apixaban was superior to warfarin at reducing the risk of stroke or systemic embolism, bleeding, and mortality in patients with AF and at least one risk factor for stroke or systemic embolism.14,17 Whether the effect of apixaban varies by type and duration of AF is uncertain. We conducted this pre-specified secondary analysis of the ARISTOTLE trial to compare outcomes and treatment effect of apixaban vs. warfarin by AF type and duration.

Methods

The ARISTOTLE trial

The details of the ARISTOTLE trial have been published.14,17 In brief, ARISTOTLE was a multicentre, double-blind, double-dummy clinical trial that randomized patients with AF and at least one risk factor for thrombo-embolism as determined by the CHADS2 score to dose-adjusted warfarin and apixaban placebo (n = 9081) vs. apixaban and warfarin placebo (n = 9120). The primary objective of the trial was to study whether apixaban is non-inferior to warfarin at reducing the risk of stroke or systemic embolism. If non-inferiority is demonstrated, the trial was powered to detect superiority of apixaban over warfarin in its effect on stroke or systemic embolism followed by bleeding followed by mortality.14,17

Patient population

To be eligible for the trial, patients had to be in AF or atrial flutter documented on two separate occasions at least 2 weeks apart in the 6 months prior to enrolment. Atrial fibrillation or atrial flutter may be documented by an electrocardiogram (ECG) or as an episode at least 1 min in duration on a rhythm strip of a Holter recording. In addition, patients had to have at least one of the following risk factors for stroke: age ≥75 years; prior stroke, transient ischaemic attack (TIA), or systemic embolism; symptomatic heart failure within 3 months or left ventricular ejection fraction ≤40%; diabetes mellitus; and hypertension requiring pharmacological treatment. Patients were ineligible for enrolment if they had moderate or severe mitral stenosis, AF due to a reversible cause, a contraindication to oral anticoagulation due to an increased bleeding risk, conditions other than AF that required anticoagulation such as prosthetic heart valve, persistent uncontrolled hypertension, or planned AF or atrial flutter ablation among other criteria listed in the original manuscript.14,17

The definitions of paroxysmal, persistent, and permanent AF used in the trial were in accordance with the definitions used in the 2006 American College of Cardiology/American Heart Association/European Society of Cardiology AF guidelines.9 Specifically, paroxysmal AF was defined as recurrent AF that terminates spontaneously, persistent AF was defined as AF that is sustained beyond 7 days, and permanent AF was defined as long-standing AF in which restoring and/or maintaining sinus rhythm has failed or has been foregone. The prior need for a cardioversion did not automatically lead to classifying AF as persistent; the subsequent occurrence of self-terminating paroxysms of AF in such patients led to classifying them as having paroxysmal AF. The type of AF was classified by the sites at randomization when they had to specify whether a particular patient had paroxysmal or persistent/permanent (combined into one category) AF or atrial flutter documented by an ECG at the time of enrolment or whether a patient had AF or atrial flutter documented on two separate occasions at least 2 weeks apart in the 6 months prior to enrolment.

Treatment

Apixaban or matching placebo was given twice daily with apixaban administered in 5 mg pills, or 2.5 mg doses for patients with two or more of the following factors: age ≥80 years, body weight ≤60 kg, and serum creatinine ≥1.5 mg/dL (133 µmol/L). Warfarin (or matching placebo) was administered in 2 mg tablets and was adjusted to achieve a target international normalized ratio (INR) of 2.0–3.0. International normalized ratios were monitored using a blinded, encrypted, point-of-care INR device. For patients who were lost to follow-up or withdrew consent, attempts were made to ascertain vital status at the end of the trial. The median time in therapeutic range (TTR) among patients on warfarin was 66.0% and the mean time was 62.2%. Therapies for rate and rhythm control were left to the discretion of the treating physician.14,17

In this analysis, rhythm control during the follow-up was defined as the use of an antiarrhythmic medication, cardioversion, and/or ablation procedure for AF at any point during the study. Rate control was defined as the use of none of the above and use of beta-blockers, calcium channel blockers, and/or digoxin.

Endpoints

Using pre-specified criteria, all endpoints were adjudicated by a clinical events committee blinded to treatment assignment.14,15 The primary efficacy endpoint for this analysis was a composite of ischaemic or haemorrhagic stroke or systemic embolism. Stroke was defined as a non-traumatic, focal neurological deficit of at least 24 h duration and was classified as ischaemic (with or without haemorrhagic transformation), haemorrhagic, or uncertain type (for patients without brain imaging or autopsy). The secondary efficacy endpoint was all-cause mortality.14,17

The primary safety endpoint for this analysis was International Society of Thrombosis and Hemostasis (ISTH) major bleeding. Per the ISTH criteria, major bleeding was defined as clinically overt bleeding accompanied by a decrease in the haemoglobin level of ≥2 g/dL, transfusion of ≥2 units of packed red blood cells, occurring in a critical site, or resulting in death.

Statistical analysis

Continuous variables are presented as means and standard deviations as well as medians and 25 and 75th percentiles, and categorical variables are presented as percentages. We examined the baseline characteristics of patients by type of AF and compared the two groups using t-tests for continuous variables and the χ2 tests for categorical variables.

For each type of AF at baseline, we determined the frequency of follow-up ECGs and calculated the proportion of patients with AF on all subsequent ECGs and no documentation of sinus rhythm. Subjects with no ECG during the follow-up (<10%) were excluded from this subanalysis. We also calculated the proportion of patients in AF (not sinus rhythm) at the 1-year ECG and end of study ECG. Patients with no ECG at these time points were excluded, and the resulting denominator was reported.

We calculated the TTR using the method of Rosendaal.18–20 Briefly, this method linearly interpolates INR values for each day between measured INRs, and then calculates TTR as the proportion of days with an INR between 2 and 3. International normalized ratios were monitored on Day 4, twice a week for 2 weeks, once a week for 2 weeks, and monthly thereafter once a stable INR was obtained.

Efficacy and safety endpoints are presented using rates per 100 patient-years of follow-up. The interaction between the randomized treatment and type of AF was tested in a Cox proportional hazards model including the randomized treatment, type of AF, and the interaction. Hazard ratios for the randomized treatment within categories of baseline AF were derived from this model with interaction. To illustrate the interaction, event rates were estimated as 1 minus the Kaplan–Meier survival probability and plotted by AF categories in combination with treatment.

Atrial fibrillation duration, defined as time from first documented occurrence of AF to randomization, was tested for association with outcomes and interaction with randomized treatment by inclusion in unadjusted Cox proportional hazards models. Outliers having AF duration in excess of 30 years were set to missing and excluded from the analysis (0.5% upper tail). Linear and restricted cubic splines were used to account for non-linearity in the relationship between AF duration and outcomes. There was strong evidence of non-linearity in the relationship between duration of AF and death, and weak evidence for other endpoints. The pattern of non-linearity was similar across all endpoints with decreasing risk associated with increased duration until 2 years, increased risk associated with increased duration between 2 and 10 years, and no apparent relationship beyond 10 years. Thus, we fit a linear spline to AF duration and reported a different hazard ratio in each range (0–2, 2–10).

Baseline data were balanced by randomized treatment within types of AF. Hazard ratios [95% confidence intervals (CI)] comparing the categories of baseline AF were derived from a Cox proportional hazards model. Hazard ratios are reported unadjusted first and then after adjustment for age; sex; region; prior myocardial infarction; heart failure; prior stroke; TIA or systemic embolism; diabetes; hypertension; prior clinically relevant or spontaneous bleeding; time from first documented occurrence of AF to randomization (i.e. duration of AF); whether a vitamin K antagonist was used prior to enrolment; vascular disease; and baseline use of aspirin, statins, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, and amiodarone. The components of the CHADS2 score were included individually as covariates in the model, rather than as an aggregate score, to allow more comprehensive adjustment. Non-linearity in continuous covariates was handled by fitting restricted cubic splines. Adjusted event rates were estimated and plotted for each type of AF according to the average survival method, in which model predictions for each type of AF are averaged over all covariate values in the sample. We checked for violations of the proportionality assumption and found none that impact the interpretation of the results.

All analyses were performed using SAS software version 9.2 (SAS Institute, Inc., Cary, NC, USA). All tests were two-sided, and a P–value <0.05 was considered statistically significant. The institutional review board of the Duke University Health System approved this study. The authors had full access to the data and take responsibility for its integrity. All authors have read and agreed to the manuscript as written.

Results

Baseline characteristics

At study entry, 2786 (15.3%) patients had paroxysmal AF and 15 412 (84.7%) had persistent or permanent AF. The median duration of AF prior to study entry for the overall population was 840 days (25th, 75th percentiles: 140, 2269); 489 (145, 1721) for patients with paroxysmal AF, and 914 (139, 2372) for patients with persistent or permanent AF. The baseline characteristics by type of AF are presented in Table 1. New onset AF within 14 days before randomization was present in 31 (1.1%) patients with paroxysmal AF and 824 (5.4%) patients with persistent or permanent AF. Compared with patients with paroxysmal AF, those with persistent or permanent AF were older and were less often women, more often had a history of heart failure and a higher CHADS2 score, and less often had a history of prior myocardial infarction, hypertension, and vascular disease. There was a trend towards a higher rate of prior stroke, TIA, or systemic embolism among patients with paroxysmal AF. Prior clinically relevant or spontaneous bleeding was evenly distributed between the two groups.

Table 1

Baseline characteristics

Characteristic Paroxysmal AF (n = 2786) Persistent or permanent AF (n = 15 412) P-value 
Age, median (25th, 75th), years 69 (61, 75) 70 (63, 76) <0.001 
Female sex, n (%) 1166 (41.9) 5249 (34.1) <0.001 
Time from first documented occurrence of AF to randomization, mean (SD), months 39.7 (53.3) 52.4 (62.4) <0.0001 
Median (25th, 75th) 16.0 (4.8, 56.5) 30.0 (4.6, 77.9)  
New onset AF within 14 days before randomization, n (%) 31 (1.1) 824 (5.4) <0.0001 
Systolic BP, mean (SD), mmHg 132 (16) 131 (16) <0.0001 
Diastolic BP, mean (SD), mmHg 78 (9) 79 (10) <0.001 
Weight, mean (SD), kg 84.3 (19.5) 84.0 (20.9) 0.504 
Prior MI, n (%) 444 (15.9) 2141 (13.9) 0.005 
Chronic heart failure, n (%) 680 (24.4) 4859 (31.5) <0.001 
Prior stroke, TIA, or systemic embolism, n (%) 578 (20.7) 2959 (19.2) 0.058 
Diabetes, n (%) 659 (23.7) 3887 (25.2) 0.079 
Hypertension, n (%) 2509 (90.1) 13 405 (87.0) <0.001 
History of vascular disease, n (%) 783 (28.10) 3717 (24.12) <0.001 
Prior clinically relevant or spontaneous bleeding, n (%) 442 (15.9) 2598 (16.9) 0.200 
Warfarin naïve, n (%) 1334 (47.9) 6465 (41.9) <0.001 
CHADS2 score, mean (SD) 2.0 (1.1) 2.1 (1.1) <0.001 
CHADS2 score, n (%) 
 1 1070 (38.4) 5112 (33.2) <0.001 
 2 928 (33.3) 5587 (36.3) 
 ≥3 788 (28.3) 4713 (30.6) 
Therapies at time of randomization, n (%) 
 ACE inhibitor or ARB 1975 (72.2) 10 855 (71.6) 0.507 
 Beta-blocker 1598 (58.4) 9884 (65.2) <0.001 
 Aspirin 978 (35.1) 4654 (30.2) <0.001 
 Clopidogrel 65 (2.3) 272 (1.8) 0.041 
 Digoxin 341 (12.5) 5487 (36.2) <0.001 
 Calcium channel blocker 834 (30.5) 4731 (31.2) 0.461 
 Statins 1299 (47.5) 6173 (40.7) <0.001 
 Antiarrhythmic medications 1326 (48.5) 1876 (12.4) <0.001 
 Prior electrical cardioversion 695 (25.0) 2403 (15.6) <0.001 
 Prior AF ablation 62 (2.2) 216 (1.4) 0.001 
Characteristic Paroxysmal AF (n = 2786) Persistent or permanent AF (n = 15 412) P-value 
Age, median (25th, 75th), years 69 (61, 75) 70 (63, 76) <0.001 
Female sex, n (%) 1166 (41.9) 5249 (34.1) <0.001 
Time from first documented occurrence of AF to randomization, mean (SD), months 39.7 (53.3) 52.4 (62.4) <0.0001 
Median (25th, 75th) 16.0 (4.8, 56.5) 30.0 (4.6, 77.9)  
New onset AF within 14 days before randomization, n (%) 31 (1.1) 824 (5.4) <0.0001 
Systolic BP, mean (SD), mmHg 132 (16) 131 (16) <0.0001 
Diastolic BP, mean (SD), mmHg 78 (9) 79 (10) <0.001 
Weight, mean (SD), kg 84.3 (19.5) 84.0 (20.9) 0.504 
Prior MI, n (%) 444 (15.9) 2141 (13.9) 0.005 
Chronic heart failure, n (%) 680 (24.4) 4859 (31.5) <0.001 
Prior stroke, TIA, or systemic embolism, n (%) 578 (20.7) 2959 (19.2) 0.058 
Diabetes, n (%) 659 (23.7) 3887 (25.2) 0.079 
Hypertension, n (%) 2509 (90.1) 13 405 (87.0) <0.001 
History of vascular disease, n (%) 783 (28.10) 3717 (24.12) <0.001 
Prior clinically relevant or spontaneous bleeding, n (%) 442 (15.9) 2598 (16.9) 0.200 
Warfarin naïve, n (%) 1334 (47.9) 6465 (41.9) <0.001 
CHADS2 score, mean (SD) 2.0 (1.1) 2.1 (1.1) <0.001 
CHADS2 score, n (%) 
 1 1070 (38.4) 5112 (33.2) <0.001 
 2 928 (33.3) 5587 (36.3) 
 ≥3 788 (28.3) 4713 (30.6) 
Therapies at time of randomization, n (%) 
 ACE inhibitor or ARB 1975 (72.2) 10 855 (71.6) 0.507 
 Beta-blocker 1598 (58.4) 9884 (65.2) <0.001 
 Aspirin 978 (35.1) 4654 (30.2) <0.001 
 Clopidogrel 65 (2.3) 272 (1.8) 0.041 
 Digoxin 341 (12.5) 5487 (36.2) <0.001 
 Calcium channel blocker 834 (30.5) 4731 (31.2) 0.461 
 Statins 1299 (47.5) 6173 (40.7) <0.001 
 Antiarrhythmic medications 1326 (48.5) 1876 (12.4) <0.001 
 Prior electrical cardioversion 695 (25.0) 2403 (15.6) <0.001 
 Prior AF ablation 62 (2.2) 216 (1.4) 0.001 

ACE, angiotensin-converting enzyme; AF, atrial fibrillation; ARB, angiotensin receptor blocker; BP, blood pressure; MI, myocardial infarction; SD, standard deviation; TIA, transient ischaemic attack.

Treatment

Of patients with paroxysmal AF, 1347 (7.55%) were randomized to apixaban and 1412 (7.76%) to warfarin. Of patients with persistent or permanent AF, 7744 (42.55%) were randomized to apixaban and 7668 (42.14%) to warfarin. Of patients with paroxysmal AF, 47.9% were vitamin K antagonist-naïve vs. 41.9% of patients with persistent or permanent AF. In the warfarin group, patients with paroxysmal AF had a mean TTR of 61.7 and a median TTR of 65.0 and patients with persistent or permanent AF had a mean TTR of 62.3 and a median TTR of 66.0 during the trial (P-value for difference of means = 0.30).

Table 1 shows medications and procedures by type of AF at baseline. Compared with patients with paroxysmal AF, those with persistent or permanent AF less often received amiodarone, any antiarrhythmic medication, aspirin, clopidogrel, and statins. Patients with persistent or permanent AF more often received a beta-blocker, digoxin, and warfarin. The use of an angiotensin-converting enzyme inhibitor or an angiotensin receptor blocker and a calcium channel blocker was evenly distributed between the two groups. Patients with paroxysmal AF more often had a prior history of electrical cardioversion and AF ablation. During the follow-up, 627 (22.5%) patients with paroxysmal AF were managed with rhythm control and 2158 (77.5%) were managed with rate control. Of patients with persistent or permanent AF, 1020 (6.6%) were managed with rhythm control and 14 391 (93.4%) were managed with rate control.

Outcomes

As shown in Figure 1, there was a consistent reduction in stroke or systemic embolism (P for interaction 0.71), all-cause mortality (P for interaction 0.75), and major bleeding (P for interaction 0.50) with apixaban compared with warfarin for both types of AF. Longer duration of AF was associated with a piecewise linear reduction in the risk of death [P-value for death <0.001; HR: 0.83, 95% CI (0.76, 0.90) per 1 year increase in duration between 0–2 years; HR: 1.01, 95% CI (0.99, 1.04) beyond 2 years]. There was no association between AF duration and any other endpoint. This association was sustained after controlling for age (P-value <0.001). The benefit from apixaban was consistent compared with warfarin in all of the studied endpoints, regardless of the duration of AF at study entry (P-value for all interactions >0.13).

Figure 1

Outcomes by type of atrial fibrillation and study treatment.

Figure 1

Outcomes by type of atrial fibrillation and study treatment.

As shown in Table 2 and Figure 2, the rates of stroke or systemic embolism, all-cause mortality, and the composite endpoint of stroke or systemic embolism, all-cause mortality, and major bleeding were significantly lower in patients with paroxysmal AF than patients with persistent or permanent AF [HR: 0.65, 95% CI (0.48, 0.87), P = 0.003; HR: 0.72, (95% CI 0.61, 0.85), P = 0.0002; HR: 0.77, (95% CI: 0.68, 0.87), P < 0.0001, respectively]. There was also a trend towards a lower risk of major bleeding in the paroxysmal AF group [HR: 0.83, 95% CI (0.68, 1.02), P = 0.078]. Adjusting for potential confounders, the risk of stroke or systemic embolism was still significantly lower for paroxysmal AF than for persistent or permanent AF [HR: 0.70, 95% CI (0.51, 0.93), P = 0.015]. There was a trend towards a higher risk of all-cause mortality and the composite endpoint of stroke or systemic embolism, all-cause mortality, and major bleeding in patients with persistent or permanent AF (P = 0.066 and P = 0.068, respectively).

Table 2

Outcomes by type of atrial fibrillation

Outcome No. of patients Paroxysmal AF No. of events (%/100 patient-years) Persistent or permanent AF No. of events (%/100 patient-years) Unadjusted HR (95% CI) Unadjusted P-value 
Stroke or systemic embolism 18 198 51 (0.98) 426 (1.52) 0.65 (0.48, 0.87) 0.003 
All-cause mortality 18 198 149 (2.81) 1123 (3.90) 0.72 (0.61, 0.853) 0.0002 
Major bleeding 18 137 104 (2.22) 685 (2.68) 0.83 (0.68, 1.02) 0.078 
Composite of stroke or systemic embolism, all-cause mortality, major bleeding 18 198 272 (5.31) 1905 (6.91) 0.77 (0.68, 0.87) <0.0001 
Outcome No. of patients Paroxysmal AF No. of events (%/100 patient-years) Persistent or permanent AF No. of events (%/100 patient-years) Unadjusted HR (95% CI) Unadjusted P-value 
Stroke or systemic embolism 18 198 51 (0.98) 426 (1.52) 0.65 (0.48, 0.87) 0.003 
All-cause mortality 18 198 149 (2.81) 1123 (3.90) 0.72 (0.61, 0.853) 0.0002 
Major bleeding 18 137 104 (2.22) 685 (2.68) 0.83 (0.68, 1.02) 0.078 
Composite of stroke or systemic embolism, all-cause mortality, major bleeding 18 198 272 (5.31) 1905 (6.91) 0.77 (0.68, 0.87) <0.0001 

AF, atrial fibrillation; CI, confidence interval; HR, hazard ratio.

Figure 2

Outcomes by type of atrial fibrillation. (A) Stroke or systemic embolism. (B) Major bleeding, (C) All-cause mortality (D) the composite endpoint of stroke or systemic embolism, all-cause mortality, and major bleeding.

Figure 2

Outcomes by type of atrial fibrillation. (A) Stroke or systemic embolism. (B) Major bleeding, (C) All-cause mortality (D) the composite endpoint of stroke or systemic embolism, all-cause mortality, and major bleeding.

Follow-up electrocardiogram data

The majority of patients had data entered on ≤2 ECGs during the follow-up; 10 696 (69.4%) in the persistent or permanent AF group and 1877 (67.4%) in the paroxysmal AF group. The proportion of patients with AF on all subsequent ECGs was 81.6% in the persistent or permanent AF group and 17.5% in the paroxysmal AF group (P < 0.0001). The proportion of patients in AF at the 1-year ECG was 87.8% in the persistent or permanent AF group vs. 23.6% on the paroxysmal AF group (P < 0.0001). Of patients with paroxysmal AF at study entry, 700 (30.9%) had AF on their last trial ECG; and of patients with persistent or permanent AF at study entry, 10 733 (87.3%) had AF on their last trial ECG.

Discussion

This study has three important findings. First, there was a consistent reduction in stroke or systemic embolism, all-cause mortality, and major bleeding with apixaban compared with warfarin regardless of the type of AF. Secondly, the superiority of apixaban over warfarin was consistent regardless of AF duration, defined as time from first documented occurrence of AF to randomization. Finally, in this study, persistent or permanent AF was associated with a higher risk of stroke or systemic embolism and a trend towards a higher risk of all-cause mortality than paroxysmal AF. These differences were maintained even after adjustment for differences in baseline characteristics that could influence the risk of these events.

Despite differences in baseline characteristics and outcomes by type of AF, there was a consistent reduction in stroke or systemic embolism, all-cause mortality, and major bleeding with apixaban compared with warfarin for both AF types. In an analysis of ACTIVE W (Atrial Fibrillation Clopidogrel Trial With Irbesartan for Prevention of Vascular Events), compared with the combination of aspirin and clopidogrel, warfarin resulted in a lower risk of stroke or systemic embolism and a higher risk of bleeding, irrespective of the AF type.21 In RE-LY (Randomized Evaluation of Long-Term Anticoagulation Therapy), and compared with warfarin, the higher studied dose of dabigatran showed superior efficacy for stroke prevention with comparable risk of major bleeding in patients with paroxysmal as well as patients with persistent and permanent AF.22 Thus, it seems likely that the results on stroke prevention with the new anticoagulants compared with warfarin will be consistent regardless of the type of AF.

In this study, apixaban was superior to warfarin, on average, and was consistent regardless of AF duration, defined as time from first documented occurrence of AF to randomization. To our knowledge, our study is the first to examine the effect of an anticoagulant by duration of AF in the context of a large, multicentre, randomized clinical trial. Similar analyses of ACTIVE W and RE-LY did not examine the benefit of study drug by duration of AF at study entry.21,22 Because of the potential clinical importance of this factor, it should be examined in future studies including secondary analyses of large contemporary clinical trials of AF. The association between longer duration of AF and reduction in the risk of death is intriguing; however, it likely reflects selection bias that resulted from inclusion in a clinical trial. In other words, patients with longer duration of AF have typically passed the initial first months of AF during which the risk of stroke and major bleeding is high and are more commonly warfarin experienced and more commonly had not had side-effects and complications on warfarin as such patients would have been excluded from participating in a randomized clinical trial.

Although some studies suggested lower event rates in patients with paroxysmal AF than patients with persistent or permanent AF, data from major clinical trials of warfarin as well as other anticoagulants failed to show any difference in clinical events related to AF by AF type.6–8,11,12,21,22 Our findings are in stark contrast with the findings of those studies and raise questions about potential reasons for these differences. Except for RE-LY, previous studies included substantially fewer patients than our study.8,12,21,22 Specifically, the pooled analysis of the SPAF trials included only 460 patients with paroxysmal AF and 1552 patients with sustained AF, the pooled analysis of the SPORTIF trials included 836 patients with paroxysmal AF and 6493 subjects with persistent AF, and the more recent ACTIVE W trial included 1202 patients with paroxysmal AF and 5495 patients with persistent or permanent AF.8,12,21 By including more patients (2786 patients with paroxysmal AF and 15 412 patients with persistent or permanent AF), our study had more statistical power to determine whether AF type is an independent risk factor for stroke or systemic embolism. Also, ARISTOTLE enrolled a greater proportion of patients with persistent/permanent AF than RE-LY (15 412 vs. 12 164). This may explain why our findings are different from the findings of RE-LY, as patients with persistent or permanent AF are typically more morbid than patients with paroxysmal AF.22 Although largely speculative, differences between our findings and those of RE-LY may relate to differences in patient characteristics, definitions of types of AF, and outcomes or overall management of patients. Analyses of outcomes by AF type in other large contemporary randomized clinical trials, such as ROCKET AF (Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation) and ENGAGE-AF-TIMI 48 (Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation-Thrombolysis in Myocardial Infarction Study 48), may help address this controversy.15,23

Several caveats should be kept in mind when assessing our results. First, classifying AF is challenging especially given the nuances of the definitions of AF types and the transitioning of some patients from one type of AF to another at different time points. Secondly, although this was a pre-specified secondary analysis of the ARISTOTLE trial and several statistical models were used to adjust for confounding, residual confounding cannot be ruled out. However, our adjusted analysis was quite comprehensive, and therefore, our findings cannot be explained by imbalances in measured covariates. Although there may be unmeasured patient characteristics associated with type of AF that would explain the observed discrepancies, we adjusted for well-known comorbidities that are likely to influence outcomes. Thirdly, we were unable to analyse persistent and permanent AF separately because these were combined into one category during data collection. Because patients with these types of AF may be inherently different, combining them may have introduced some bias. In addition, the distinction between paroxysmal and non-paroxysmal AF is not always clear cut. Fourthly, because the majority of patients in our study had two or fewer ECGs reported in follow-up, these data could not be used to discern type of AF during the follow-up. However, it is somewhat reassuring that the proportion of patients with AF on all subsequent ECGs was substantially higher in the persistent/permanent AF group than in the paroxysmal AF group (81.6 vs. 17.5%). Fifthly, we were neither able to capture the effect of all rhythm-control therapies on AF type nor to assess the AF burden in the follow-up. Our results originated from a large clinical trial population. Notwithstanding the large number of included patients and countries, our results should be validated in other databases.

Conclusions

In conclusion, in patients with AF and at least one risk factor for stroke, the risk of stroke or systemic embolism, mortality, and major bleeding was lower with apixaban than with warfarin regardless of the AF type and duration. Therefore, although the risk of stroke or systemic embolism was lower in paroxysmal than persistent or permanent AF, it seems equally important to consider the use of apixaban as an alternative to warfarin regardless of AF type or duration.

Funding

This work was supported by Bristol-Myers Squibb, Co Princeton, NJ and Pfizer, Inc., New York, NY and coordinated by the Duke Clinical Research Institute, USA and Uppsala Clinical Research Center, Sweden.

Conflict of interest: S.M.A.-K. served as a member of the Steering committee for this trial. Currently, she has nothing to disclose. L.W.: research grants from AstraZeneca, Merck/Schering-Plough, Boehringer Ingelheim, Bristol-Myers Squibb/Pfizer, GlaxoSmithKline; consultant for Merck/Schering-Plough, Regado Biosciences, Evolva, Portola, C.S.L. Behring, Athera Biotechnologies, Boehringer Ingelheim, AstraZeneca, GlaxoSmithKline, and Bristol-Myers Squibb/Pfizer; lecture fees from AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb/Pfizer, GlaxoSmithKline, Schering-Plough; honoraria from Boehringer Ingelheim, AstraZeneca, Bristol-Myers Squibb/Pfizer, GlaxoSmithKline, and Schering-Plough/Merck. R.D.L.: grants from Bristol-Myers Squibb, AstraZeneca, Boehringer Ingelheim, and Daiichi-Sankyo and consulting fees from Bristol-Myers Squibb. B.G.: consultant for Ortho-McNeil-Janssen, GE Healthcare, St Jude Medical, Abbott Laboratories, Medispec, Merck, and Boston Scientific. D.G.: consulting fees from Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi-Sankyo, Pfizer, and CSL Behring; and fees for the development of educational materials from Boehringer Ingelheim. J.E.: consulting fee and travel support from Bristol-Myers Squibb/Pfizer; research support, honoraria, and payments for continuing medical education events from AstraZeneca, Amgen, Abbott, Servier, Johnson & Johnson, Pfizer, and Bristol-Myers Squibb. M.A.: travel support from Bristol-Myers Squibb; board member for Bayer, Boehringer Ingelheim, MSD, and Sanofi-Aventis; lecture fees from Bayer, Boehringer Ingelheim, MSD, and AstraZeneca; development of educational presentations for Boehringer Ingelheim; and travel support from St Jude Medical and Boston Scientific. J.H.A.: grants from Bristol-Myers Squibb, Merck/Schering-Plough, and Regado Biosciences; travel support from Bristol-Myers Squibb; consulting fees from Bristol-Myers Squibb, Pfizer, Merck/Schering-Plough, AstraZeneca, Boehringer Ingelheim, Ortho-McNeil-Janssen, PolyMedix, Regado Biosciences, Bayer, and Daiichi-Sankyo. G.F.: research grant, consulting fees, travel support, committee member fees from BMS/Pfizer; and compensation for development of educational presentations from Pfizer. M.H.: employee of Bristol-Myers Squibb and receiving stock as a part of compensation. C.B.G.: grants from Bristol-Myers Squibb, Boehringer Ingelheim, Astellas, AstraZeneca, GlaxoSmithKline, Medtronic foundation, Merck, Sanofi-Aventis, and the Medicine's Company; consulting fees from Bristol-Myers Squibb, Pfizer, Boehringer Ingelheim, AstraZeneca, Lilly, GlaxoSmithKline, Hoffman-La Roche, Novartis, Otsuka, Sanofi-Aventis, and the Medicines Company; travel support from Pfizer, Novartis, Otsuka, and Hoffmann-La Roche. L.T. and H.Y.: Duke received funding from BMS for the statistical analysis and manuscript preparation performed by L.T. and H.Y.

References

1
Wolf
PA
Abbott
RD
Kannel
WB
Atrial fibrillation as an independent risk factor for stroke: the Framingham Study
Stroke
 , 
1991
, vol. 
22
 (pg. 
983
-
988
)
2
Lin
HJ
Wolf
PA
Kelly-Hayes
M
Beiser
AS
Kase
CS
Benjamin
EJ
D'Agostino
RB
Stroke severity in atrial fibrillation. The Framingham Study
Stroke
 , 
1996
, vol. 
27
 (pg. 
1760
-
1764
)
3
Hylek
EM
Go
AS
Chang
Y
Jensvold
NG
Henault
LE
Selby
JV
Singer
DE
Effect of intensity of oral anticoagulation on stroke severity and mortality in atrial fibrillation
N Engl J Med
 , 
2003
, vol. 
349
 (pg. 
1019
-
1026
)
4
Jørgensen
HS
Nakayama
H
Reith
J
Raaschou
HO
Olsen
TS
Acute stroke with atrial fibrillation. The Copenhagen Stroke Study
Stroke
 , 
1996
, vol. 
27
 (pg. 
1765
-
1769
)
5
Lopes
RD
Shah
BR
Olson
DM
Zhao
X
Pan
W
Bushnell
CD
Peterson
ED
Antithrombotic therapy use at discharge and 1 year in patients with atrial fibrillation and acute stroke: results from the AVAIL Registry
Stroke
 , 
2011
, vol. 
42
 (pg. 
3477
-
3483
)
6
Aboaf
AP
Wolf
PS
Paroxysmal atrial fibrillation. A common but neglected entity
Arch Intern Med
 , 
1996
, vol. 
156
 (pg. 
362
-
367
)
7
Petersen
P
Godtfredsen
J
Embolic complications in paroxysmal atrial fibrillation
Stroke
 , 
1986
, vol. 
17
 (pg. 
622
-
626
)
8
Hart
RG
Pearce
LA
Rothbart
RM
McAnulty
JH
Asinger
RW
Halperin
JL
Stroke with intermittent atrial fibrillation: incidence and predictors during aspirin therapy. Stroke Prevention in Atrial Fibrillation Investigators
J Am Coll Cardiol
 , 
2000
, vol. 
35
 (pg. 
183
-
187
)
9
ACC/AHA Task Force on Practice Guidelines and the ESC Committee for Practice Guidelines (Writing Committee to Revise the 2001 Guidelines for the Management of Patients With Atrial Fibrillation)
ACC/AHA/ESC 2006 Guidelines for the Management of Patients With Atrial Fibrillation—Executive Summary
J Am Coll Cardiol
 , 
2006
, vol. 
48
 (pg. 
854
-
906
)
10
ACCF/AHA Task Force
2011 ACCF/AHA/HRS focused updates incorporated into the ACC/AHA/ESC 2006 guidelines for the management of patients with atrial fibrillation
Circulation
 , 
2011
, vol. 
123
 (pg. 
e269
-
367
)
11
Nieuwlaat
R
Dinh
T
Olsson
SB
Camm
AJ
Capucci
A
Tieleman
RG
Lip
GY
Crijns
HJ
Should we abandon the common practice of withholding oral anticoagulation in paroxysmal atrial fibrillation?
Eur Heart J
 , 
2008
, vol. 
29
 (pg. 
915
-
922
)
12
SPORTIF Investigators
Stroke event rates in anticoagulated patients with paroxysmal atrial fibrillation
J Intern Med
 , 
2008
, vol. 
264
 (pg. 
50
-
61
)
13
Connolly
SJ
Ezekowitz
MD
Yusuf
S
Eikelboom
J
Oldgren
J
Parekh
A
Pogue
J
Reilly
PA
Themeles
E
Varrone
J
Wang
S
Alings
M
Xavier
D
Zhu
J
Diaz
R
Lewis
BS
Darius
H
Diener
HC
Joyner
CD
Wallentin
L
Dabigatran versus warfarin in patients with atrial fibrillation
N Engl J Med
 , 
2009
, vol. 
361
 (pg. 
1139
-
1151
)
14
Granger
CB
Alexander
JH
McMurray
JV
Lopes
RD
Hylek
EM
Hanna
M
Al-Khalidi
HR
Ansell
J
Atar
D
Avezum
A
Bahit
MC
Diaz
R
Easton
JD
Ezekowitz
JA
Flaker
G
Garcia
D
Geraldes
M
Gersh
BJ
Golitsyn
S
Goto
S
Hermosillo
AG
Hohnloser
SH
Horowitz
J
Mohan
P
Jansky
P
Lewis
BS
Lopez-Sendon
JL
Pais
P
Parkhomenko
A
Verheugt
FW
Zhu
J
Wallentin
L
Apixaban versus warfarin in patients with atrial fibrillation
N Engl J Med
 , 
2011
, vol. 
365
 (pg. 
981
-
992
)
15
Patel
MR
Mahaffey
KW
Garg
J
Pan
G
Singer
DE
Hacke
W
Breithardt
G
Halperin
JL
Hankey
GJ
Piccini
JP
Becker
RC
Nessel
CC
Paolini
JF
Berkowitz
SD
Fox
KA
Califf
RM
Rivaroxaban versus warfarin in nonvalvular atrial fibrillation
N Engl J Med
 , 
2011
, vol. 
365
 (pg. 
883
-
891
)
16
Al-Khatib
SM
Alexander
JH
Lopes
RD
Mahaffey
K
Patel
M
Granger
CB
Promise of factor Xa inhibition in atrial fibrillation
Curr Cardiol Reports
 , 
2012
, vol. 
14
 (pg. 
70
-
78
)
17
Lopes
RD
Alexander
JH
Al-Khatib
SM
Ansell
J
Diaz
R
Easton
JD
Gersh
BJ
Granger
CB
Hanna
M
Horowitz
J
Hylek
EM
McMurray
JJ
Verheugt
FW
Wallentin
L
Apixaban for stroke prevention in atrial fibrillation (ARISTOTLE) trial: design and rationale
Am Heart J
 , 
2010
, vol. 
159
 (pg. 
331
-
339
)
18
Rosendaal
FR
Cannegieter
SC
van der Meer
FJ
Briet
E
A method to determine the optimal intensity of oral anticoagulant therapy
Thromb Haemost
 , 
1993
, vol. 
69
 (pg. 
236
-
239
)
19
van Leeuwen
Y
Rombouts
EK
Kruithof
CJ
van der Meer
FJ
Rosendaal
FR
Improved control of oral anticoagulant dosing: a randomized controlled trial comparing two computer algorithms
J Thromb Haemost
 , 
2007
, vol. 
5
 (pg. 
1644
-
1649
)
20
Cannegieter
SC
Rosendaal
FR
Wintzen
AR
van der Meer
FJ
Vandenbroucke
JP
Briet
E
Optimal oral anticoagulant therapy in patients with mechanical heart valves
N Engl J Med
 , 
1995
, vol. 
333
 (pg. 
11
-
17
)
21
Hohnloser
SH
Pajitnev
D
Pogue
J
Healey
JJ
Pfeffer
MA
Yusuf
S
Connolly
SJ
Incidence of stroke in paroxysmal versus sustained atrial fibrillation in patients taking oral anticoagulation or combined antiplatelet therapy: an ACTIVE W Substudy
J Am Coll Cardiol
 , 
2007
, vol. 
50
 (pg. 
2156
-
2161
)
22
Flaker
G
Ezekowitz
M
Yusuf
S
Wallentin
L
Noack
H
Brueckmann
M
Reilly
P
Hohnloser
SH
Connolly
S
Efficacy and safety of dabigatran compared to warfarin in patients with paroxysmal, persistent, and permanent atrial fibrillation results from the RE-LY (Randomized Evaluation of Long-Term Anticoagulation Therapy) Study
J Am Coll Cardiol
 , 
2012
, vol. 
59
 (pg. 
854
-
855
)
23
Ruff
CT
Giugliano
RP
Antman
EM
Crugnale
SE
Bocanegra
T
Mercuri
M
Hanyok
J
Patel
I
Shi
M
Salazar
D
McCabe
CH
Braunwald
E
Evaluation of the novel factor Xa inhibitor edoxaban compared with warfarin in patients with atrial fibrillation: design and rationale for the Effective aNticoaGulation with factor xA next GEneration in Atrial Fibrillation-Thrombolysis In Myocardial Infarction study 48 (ENGAGE AF-TIMI 48)
Am Heart J
 , 
2010
, vol. 
160
 (pg. 
635
-
641
)

Comments

2 Comments
Take heed of the confidence intervals
22 September 2013
oscar,m jolobe

Inspection if the confidence intervals in Fig 1(Outcomes by type of atrial fibrillation and study treatment)shows that, in spite of the assertion that "there was [also] a trend towards a lower risk of major bleeding in the paroxysmal AF group...", and the assertion that "the risk of stroke or systemic embolism was still significantly lower for paroxysmal AF than for persistent or permanent AF"(1), the confidence intervals for both outcomes were much wider for paroxysmal AF than for persistent of permanent AF(1). This outcome might signify that, for patients with paroxysmal AF characterised by high arrhythmia burden(2), the benefit vs risk profile might be higher than for those with low arrhythmia burden. Attention to confidence intervals is a useful strategy for raising the index of suspicion for the confounding effect of subgroups , as was the case in revealing the impact of the Stroke Prevention in Atrial Fibrillation Study(which used an aspirin dose of 325 mg/day) on the wide confidence intervals(2% to 38%) which generated the widely quoted 22% reduction in stroke incidence attributed to the use of a wide range of aspirin doses in thromboprophylaxis of nonvalvular atrial fibrillation(3). The consequences of ignoring the wide confidence intervals in that study were pivotal in generating what some of us consider to be a highly prevalent and misplaced belief in the thromboprophylactic benefit of a wide range of aspirin doses in NVAF. There is a risk of repeating that "genre" of errors of judgement if we do not take heed of the confounding effect of arrhythmia burden when thromboprophylaxis generates an outcome associated with wide confidence intervals in paroxysmal AF References

(1) Al-Khatib S., Thomas L., Wallentin L et al Outcomes of apixaban vs warfarin by type and duration of atrial fibrillation:results from the ARISTOTLE trial Eur Heart J 2013;34:2461-2471

((2)Capucci A., Santini M., Padeletti L et al Monitored atrial fibrillation duration predicts arterial embolic events in patients suffering from bradycardia and atrial fibrillation implanted with antitachycardia pacemakers J Amer Coll Cardiol 2005;46:1913-1920

((3) Hart RG., Benavente O., McBride R., Pearce |LA Antithrombotic therapy to prevent stroke in patients with atrial fibrillation: A meta-analysis Ann Intren Med 1999;131:492-501

Conflict of Interest:

None declared

Submitted on 22/09/2013 8:00 PM GMT
The Totality of the Evidence and the Consistency of the Findings Are Also Important
22 September 2013
Sana M. Al-Khatib (with Laine Thomas, Lars Wallentin, Renato Lopes, John Alexander and Christopher Granger)

Our study shows that patients with paroxysmal atrial fibrillation (AF) have lower event rates than those with persistent or permanent AF.(1) The differences between the 2 groups are striking for both mortality and stroke and barely non-significant for bleeding. The consistency across multiple endpoints suggests that this is not a chance finding, but a reflection of generally better health state. Although there is uncertainty about how much lower the risks of stroke or systemic embolism, mortality and particularly bleeding are among patients with paroxysmal versus permanent AF, we do not find the confidence intervals in Table 2 to be unusually wide. It is not clear how this should impact assessment of benefit vs. risk of oral anticoagulation since these patients have lower risk of stroke or systemic embolism (less to gain) and lower risk of bleeding (less to lose). However, our focus in the paper is to compare treatments (apixaban vs. warfarin) directly in the populations of interest. The results in Figure 1 show no evidence of heterogeneity in the treatment effect when comparing patients with paroxysmal AF with patients with persistent or permanent AF. This is a different issue from "confounding", whereby some causal effect of interest is obscured by unmeasured factors. In the paper, we do not attempt to isolate a causal mechanism; therefore, the term "confounding" is out of context. In the noted example of aspirin, the point estimates with wide confidence intervals fell on both sides of the unity line; however, in our paper, all the point estimates and confidence intervals favor apixaban.(2) We agree that it is not helpful to fixate on the point estimate within the smaller subgroups; however, various ways to address the question, including interaction tests, send a similar message in favor of apixaban vs. warfarin for reduction of both stroke or systemic embolism and bleeding. Therefore, we agree that confidence intervals are important to illustrate uncertainty around a point estimate, and we recommend readers interpret our results with the standard caution applied to any subgroup analysis, but the overall trial results provide the best estimate for major subgroups and in this case that is reassuring for important benefits on stroke and bleeding with apixaban.

References: 1) Al-Khatib S., Thomas L., Wallentin L et al. Outcomes of apixaban vs warfarin by type and duration of atrial fibrillation:results from the ARISTOTLE trial Eur Heart J 2013;34:2461-2471. 2) Hart RG., Benavente O., McBride R., Pearce LA. Antithrombotic therapy to prevent stroke in patients with atrial fibrillation: A meta-analysis. Ann Intren Med 1999;131:492-501

Conflict of Interest:

All authors' interests are disclosed in the original manuscript.

Submitted on 22/09/2013 8:00 PM GMT