- Email: [email protected]
Bleeding Risk in Patients With Atrial Fibrillation
CHEST
Original Research ANTITHROMBOTIC THERAPY
Bleeding Risk in Patients With Atrial Fibrillation The AMADEUS Study Deirdre A. Lane, PhD; Pieter W. Kamphuisen, MD; Pascal Minini, PhD; Harry R. Büller, MD; and Gregory Y. H. Lip, MD
Objective: This study aimed to assess the impact of combination antithrombotic therapy on stroke and bleeding risk compared with anticoagulation therapy only in patients with atrial fibrillation (AF). Methods: Post hoc analysis of 4,576 patients with AF (mean ⫾ SD age, 70.1 ⫾ 9.1 years; men, 66.5%) enrolled in the Evaluating the Use of SR34006 Compared to Warfarin or Acenocoumarol in Patients With Atrial Fibrillation (AMADEUS) trial were randomized to receive either subcutaneous idraparinux (2.5 mg weekly) (n 5 2,283) or dose-adjusted vitamin K antagonists (VKAs) (international normalized ratio, 2.0-3.0) (n 5 2,293). Of these patients, 848 (18.5%) received antiplatelet therapy (aspirin, clopidogrel, ticlopidine, etc) in addition to anticoagulation treatment (combination antithrombotic therapy). Results: A total of 572 (15.3% per year) clinically relevant bleeding and 103 (2.6% per year) major bleeding events occurred. Patients receiving combination antithrombotic therapy had a 2.3- to 2.5-fold increased risk of clinically relevant bleeding events and major bleeding events, respectively, compared with those receiving anticoagulation therapy only. Multivariate analyses (hazard ratio, 95% CI) revealed that the risk of clinically relevant bleeding was significantly increased by age 65 to 74 years (1.44, 1.14-1.82) and ⱖ 75 years (1.59, 1.24-2.04, P 5 .001) and by combination antithrombotic therapy (2.47, 2.07-2.96, P , .0001). The same held true for major bleeding events, with analogous figures for age 65 to 74 years (2.26, 1.08-4.71) and ⱖ 75 years (4.19, 1.98-8.87, P 5 .0004) and for combination antithrombotic therapy (2.23, 1.49-3.34, P , .0001). Combination antithrombotic therapy was not associated with a decrease in ischemic stroke risk compared with anticoagulation therapy only (11 [1.4% per year] vs 22 [0.7% per year]; adjusted hazard ratio, 2.01; 95% CI, 0.94-4.30; P 5 .07). Conclusions: Combination antithrombotic therapy increases the risk of clinically relevant bleeding and major bleeding in patients with AF and does not appear to reduce the risk of stroke. CHEST 2011; 140(1):146–155 Abbreviations: ACS 5 acute coronary syndrome; AF 5 atrial fibrillation; AMADEUS 5 Evaluating the Use of SR34006 Compared to Warfarin or Acenocoumarol in Patients With Atrial Fibrillation; CAD 5 coronary artery disease; CHADS2 5 congestive heart failure, hypertension, age ⱖ 75 years, diabetes, and prior stroke or transient ischemic stroke; CIAC 5 Central Independent Adjudication Committee; INR 5 international normalized ratio; TIA 5 transient ischemic attack; VKA 5 vitamin K agonist
of antithrombotic therapy for atrial Prescription fibrillation (AF) must consider both the patient’s
stroke risk and his or her risk of bleeding while on anticoagulant therapy; that is, there must be a net clinical benefit to initiate anticoagulant therapy. Given that AF commonly coexists with associated atherosclerotic vascular disease, a significant proportion of patients with AF receiving anticoagulation therapy also receive concomitant antiplatelet therapy.1,2 The
benefit of combination antithrombotic therapy (anticoagulant plus antiplatelets) in patients with AF has yet to be elucidated from clinical trials, but combination antithrombotic therapy in this population has been shown to be associated with an increased risk of bleeding1,3-7 with no significant reduction in thromboembolic events.6 The present study is a post hoc analysis of predictors of major bleeding and clinically relevant bleeding
146
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
Original Research
in patients with AF receiving anticoagulation therapy only (either subcutaneous idraparinux or vitamin K antagonists [VKAs]) or combination antithrombotic therapy (one or more antiplatelet therapies together with anticoagulation) from the Evaluating the Use of SR34006 Compared to Warfarin or Acenocoumarol in Patients With Atrial Fibrillation (AMADEUS) trial cohort.8 The AMADEUS trial was halted prematurely after a mean ⫾ SD follow-up period of 10.7 ⫾ 5.4 months due to an excess of major bleeding and clinically relevant bleeding with idraparinux (346 cases, 19.7/100 patient-years) compared with VKA (warfarin or acenocoumarol) (226 cases, 11.3/100 patient-years, P , .001). However, idraparinux was noninferior to VKAs in the prevention of stroke or non-CNS systemic embolism (18 cases vs 27 cases; 0.9 vs 1.3/100 patient-years, respectively; hazard ratio [HR], 0.71; 95% CI, 0.39-1.30; P 5 .007 for noninferiority).8 Using the AMADEUS study cohort, we tested the hypothesis that patients with AF receiving combination antithrombotic therapy will experience a greater incidence of clinically relevant bleeding and major bleeding compared with patients receiving anticoagulation therapy only. The present study also examined predictors of bleeding in patients with AF receiving anticoagulation therapy only compared with those receiving combination antithrombotic therapy. Given the excess of clinically relevant bleeding with idraparinux, separate analyses were also undertaken to compare VKA alone to VKA plus antiplatelet therapy. Materials and Methods The details of the AMADEUS trial have been described previously.8 In summary, 4,576 patients with ECG-documented AF requiring long-term anticoagulation therapy (one or more risk factors for stroke [previous ischemic stroke, transient ischemic attack (TIA), or systemic embolism; aged ⱖ 75 years; hypertension requiring treatment, or left ventricular dysfunction] or those aged 65-75 years with either diabetes mellitus or coronary artery Manuscript received December 21, 2010; revision accepted February 9, 2011. Affiliations: From the University of Birmingham Centre for Cardiovascular Sciences (Drs Lane and Lip), City Hospital, Birmingham, England; Department of Vascular Medicine (Drs Kamphuisen and Büller), Academic Medical Centre, Amsterdam, The Netherlands; and Sanofi-Aventis Research and Development, Biostatistics and Programming (Dr Minini), ChillyMazarin, France. Funding/Support: The AMADEUS trial was funded by SanofiAventis. Correspondence to: Deirdre A. Lane, PhD, University of Birmingham Centre for Cardiovascular Sciences, City Hospital, Dudley Rd, Birmingham B18 7QH, England; e-mail: deirdrelane@ nhs.net © 2011 American College of Chest Physicians. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (http://www.chestpubs.org/ site/misc/reprints.xhtml). DOI: 10.1378/chest.10-3270
THE BOTTOM LINE How does this work advance the field? This large trial cohort study suggests that the risk of clinically relevant bleeding is significantly increased with age and the use of combination antithrombotic therapy. It provides strong evidence against concomitant prescription of oral anticoagulation plus antiplatelet therapy in stable patients with atrial fibrillation, given the lack of benefit for vascular events and the substantial increase in bleeding risk.
What are the clinical implications? Combination antithrombotic therapy increases the risk of clinically relevant bleeding and major bleeding in patients with atrial fibrillation and does not appear to reduce the risk of stroke.
disease [CAD]) were randomized to receive either idraparinux 2.5 mg once per week in prefilled syringes administered subcutaneously or dose-adjusted (international normalized ratio [INR] range, 2.0-3.0) open-label VKAs (warfarin or acenocoumarol) between September 2003 and July 2005. For patients with a baseline creatinine clearance level of 10 to 30 mL/min/1.73 mm2, the second and all subsequent doses of idraparinux were 1.5 mg weekly.8 Definition of Bleeding The main safety outcome was clinically relevant bleeding, which was further classified by a Central Independent Adjudication Committee (CIAC) into major bleeding (bleeding that was fatal, intracranial, or affected another critical anatomic site or overt bleeding with a drop of hemoglobin level ⱖ 2 g/dL or requiring transfusion of ⱖ 2 units of blood) and clinically relevant nonmajor bleeding (overt bleeding that did not satisfy the criteria for major bleeding but met defined criteria and included repetitive epistaxis for ⱖ 5 min at least twice in a 24-h period, hematuria [spontaneous or lasting . 24 h], hematemesis, and subcutaneous hematomas ⱖ 25 cm2 if spontaneous or . 100 cm2 if after trauma). All suspected safety or efficacy outcome events were adjudicated by a CIAC unaware of treatment assignment.8 For the present analyses, only events confirmed by the CIAC are described. The AMADEUS trial contained a subset of 848 (18.5%) patients receiving combination antithrombotic therapy (anticoagulant and antiplatelet therapy). In addition to anticoagulation therapy (either idraparinux or VKA), 716 (15.6%) patients took aspirin only, 34 (0.7%) received a thienopyridine (clopidogrel or ticlopidine) only, 12 (0.3%) received other antiplatelet therapy (one of the following: dipyridamole, cilostazol, eptifibatide, tirofiban hydrochloride, or anagrelide), and 86 (1.9%) received two or more antiplatelets. Statistical Analysis Data analyses were performed using SAS, version 8.02 (SAS Institute Inc; Cary, North Carolina) statistical software. All statistical tests were two tailed, with P ⱕ .05 considered statistically significant. Because antiplatelet therapy was not randomized in the AMADEUS trial, direct comparison of the treatment group may be misleading. In order to reduce potential bias due to imbalance between groups at baseline, the propensity score method9 was used. The propensity score was defined as the conditional probability of receiving antiplatelets given baseline characteristics
www.chestpubs.org
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
CHEST / 140 / 1 / JULY, 2011
147
of patients (Table 1) and estimated using a multivariate logistic regression. After selection of the most predictive factors using a stepwise procedure, nine covariates were included in the final propensity score model: age, sex, race, type of AF, previous ischemic stroke or TIA, previous symptomatic CAD, left ventricular dysfunction, diabetes mellitus, and previous VKA use. Patients were then grouped into five approximately equally sized strata defined according to the quintiles of the propensity scores. Comparison of the baseline characteristics between the two groups was performed before and after stratification on the propensity score to determine whether balance was achieved after stratification. The incidence of clinical events during the study was estimated from the event rate per 100 patient-years (annual rates). The effect of antiplatelet therapy in addition to anticoagulation therapy was assessed using a Cox proportional hazard model stratified on the propensity score in five categories. HRs, 95% CIs, and P values were determined from this model. In addition, univariate and multivariate analyses were performed in order to identify significant independent factors of clinical events. Each potentially predictive factor was classified in two or three categories, one of them being considered the reference category in order to aid interpretation. Significant factors in the univariate analysis were then introduced in the multivariate model. Note that in these analyses, combination antithrombotic therapy was defined as any antiplatelet medication received at any time after randomization and up to the end of the randomized treatment or up to the day before occurrence of a thromboembolic event. Patients who received antiplatelets after a thromboembolic event were analyzed in the anticoagulation-alone group.
Results Of the total cohort of 4,576 patients with AF (age, 70.1 ⫾ 9.1 years; men, 66.5%), 3,728 (81.5%) received anticoagulation therapy only (1,861 [40.7%] idraparinux and 1,867 [40.8%] VKA). In addition, 848 (18.5%) patients received combination antithrombotic therapy (Table 1). Combination antithrombotic therapy was more common with increasing age, particularly in patients aged ⱖ 75 years, with poorer renal function, a previous stroke or TIA, left ventricular dysfunction, diabetes mellitus (and consequently increasing CHADS2 [congestive heart failure, hypertension, age ⱖ 75 years, diabetes mellitus, and prior stroke or TIA] score), and symptomatic CAD (Table 1). In addition, patients with paroxysmal AF were more likely to receive combination antithrombotic therapy compared with those with persistent or permanent AF, whereas patients who had received VKA prior to enrollment were less likely to receive concomitant antiplatelet therapy. After stratification on the propensity score, there was no statistically significant difference between the two groups for any of the baseline characteristics. Within each propensity score quintile, the two groups were comparable. Similar results emerged when analyses were confined to those receiving VKA therapy only or in combination (data not shown), although there was no significant difference in age (P 5 .17) or in patients with poorer renal function (P 5 .052).
Thromboembolic Risk During follow-up, 43 (1.1% per year) strokes occurred, with 31 (1.0% per year) in patients receiving anticoagulation therapy only and 12 (1.5% per year) in those receiving combination antithrombotic therapy (Table 2). There was a trend toward more ischemic strokes occurring with combination antithrombotic therapy compared with anticoagulation therapy alone (HR, 2.01; 95% CI, 0.94-4.30; P 5 .073) (Fig 1). Similar results emerged when analyses were restricted to patients randomized to VKA therapy (7 [1.7% per year] vs 13 [0.8% per year]; HR, 2.25; 95% CI, 0.85-5.97; P 5 .10 stratified by propensity score). There was a threefold increased risk of myocardial infarction among patients with AF receiving combination antithrombotic therapy compared with anticoagulation therapy only (HR, 3.14; 95% CI, 1.45-6.78; P 5 .0036). In analyses confined to patients randomized to VKA therapy, there was an increased risk of myocardial infarction with combination antithrombotic therapy compared with anticoagulation therapy only (6 [1.4% per year] vs 7 [0.4% per year], respectively; HR, 3.45; 95% CI, 1.16-10.29; P 5 .03); however, stratifying by propensity score rendered this association not significant (HR, 2.70; 95% CI, 0.85-8.56; P 5 .09). Clinically Relevant Bleeding A total of 572 (15.3% per year) clinically relevant bleeding events occurred, with significantly more bleeding events among patients receiving combination antithrombotic therapy compared with anticoagulation therapy only (Fig 2, Table 2). Similar results emerged when analyses were confined only to patients randomized to VKA therapy (81 [21.7% per year] vs 145 [9.0% per year]; HR, 2.35; 95% CI, 1.77-3.13; P , .0001 stratified by propensity score). In univariate analyses of all randomized patients, age ⱖ 65 years, BMI , 25 kg/m2, poor renal function (baseline creatinine clearance level , 50 mL/min/1.73 m2), symptomatic CAD, VKA naivety at baseline, and combination antithrombotic therapy were significantly associated with clinically relevant bleeding (Table 3). Multivariate analyses revealed that age ⱖ 65 years (HR for age 65-75 years, 1.44; 95% CI, 1.14-1.82; HR for ⱖ 75 years, 1.59; 95% CI, 1.24-2.04; P 5 .001) and combination antithrombotic therapy (HR, 2.47; 95% CI, 2.07-2.96; P , .0001) predicted increased risk of clinically relevant bleeding (Table 4). When univariate analyses were restricted to patients taking VKA (n 5 2,293), only age 65 to 74 years (HR, 1.59; 95% CI, 1.12-2.27; P 5 .033), symptomatic CAD (HR, 1.58; 95% CI, 1.21-2.05; P 5 .0007), and combination antithrombotic therapy (HR, 2.44; 95% CI, 1.86-3.20; P , .0001) were significant
148
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
Original Research
Table 1—Patient Baseline Demographic and Clinical Characteristics by Antithrombotic Therapy Baseline Characteristic Age, y Age group , 65 y 65-74 y ⱖ 75 y Male sex Ethnicity White Black Asian Other BMI, kg/m2 Creatinine clearance level, mL/min/1.73 m2 Type of AF Paroxysmal AF Persistent Permanent Stroke risk factors Previous ischemic stroke Previous TIA Previous non-CNS systemic embolism Hypertension requiring drug treatment Left ventricular dysfunction Diabetes mellitus Symptomatic CAD CHADS2 score CHADS2 risk category Low risk (score 0-1) Moderate risk (score 2) High risk (score ⱖ 3) Randomized treatment group Idraparinux VKA Previous VKA use
Anticoagulation Therapy Only (n 5 3,728)a
Combination Antithrombotic Therapy (n 5 848)b
70.0 ⫾ 9.0
70.7 ⫾ 9.3
948 (25.4) 1,501 (40.3) 1,279 (34.3) 2,459 (66.0)
199 (23.5) 317 (37.4) 332 (39.2) 583 (68.8)
3,664 (98.3) 18 (0.5) 14 (0.4) 31 (0.8) 28.8 ⫾ 5.4 72.3 (55.8-92.4)
821 (96.8) 10 (1.2) 2 (0.2) 15 (1.8) 28.9 ⫾ 5.7 69.4 (52.2-90.1)
P Value .026 .017 … … .12 .01 … … … .51 .003
1,284 (34.6) 374 (10.1) 2,055 (55.3)
350 (41.5) 64 (7.6) 429 (50.9)
.0003 … …
447 (12.0) 372 (10.0) 129 (3.5) 2,895 (77.7) 805 (21.6) 686 (18.4) 980 (26.3) 1.9 ⫾ 1.1
133 (15.7) 106 (12.5) 27 (3.2) 635 (74.9) 267 (31.5) 211 (24.9) 428 (50.5) 2.2 ⫾ 1.2
.004 .03 .69 .08 , .0001 , .0001 , .0001 , .0001
1,599 (42.9) 1,175 (31.5) 951 (25.5)
278 (32.8) 273 (32.2) 297 (35.0)
… , .0001 …
1,861 (49.9) 1,867 (50.1) 2,939 (78.8)
422 (49.8) 426 (50.2) 537 (63.3)
.93 … , .0001
Data are presented as No. (%), mean ⫾ SD, or median (interquartile range). AF 5 atrial fibrillation; CAD 5 coronary artery disease; CHADS2 5 congestive heart failure, hypertension, age . 75 years, diabetes mellitus, prior stroke or transient ischemic attack; TIA 5 transient ischemic attack; VKA 5 vitamin K agonist. aAll patients randomized to idraparinux or VKA who received no antiplatelets during the randomized treatment period. bAll patients randomized to idraparinux or VKA who received one or more antiplatelet drug at any time after randomization and up to the end of the randomized treatment period or up to the day before a thromboembolic event.
predictors of clinically relevant bleeding. In multivariate analyses, only combination antithrombotic therapy was an independent predictor of clinically relevant bleeding (HR, 2.30; 95% CI, 1.73-3.04; P , .0001). Major Bleeding A total of 103 (2.6% per year) major bleeding events occurred and were more common with combination antithrombotic therapy than with anticoagulation therapy only (Fig 3, Table 2). However, in subgroup analyses confined to patients randomized to VKA therapy, combination antithrombotic therapy was not a significant predictor of major bleeding (9 [2.2% per year] vs 20 [1.2% per year]; HR, 1.46; 95% CI, 0.65-3.28; P 5 .36 stratified by propensity score).
In univariate analyses of all randomized patients, age ⱖ 65 years, poorer renal function (baseline creatinine clearance , 50 mL/min/1.73 mm2), BMI , 25 kg/m2, high CHADS2 stroke risk score, and combination antithrombotic therapy were significantly associated with major bleeding (Table 3). In multivariate analyses including all randomized patients, only age ⱖ 65 years (HR for 65-74 years, 2.26; 95% CI, 1.08-4.71; HR for ⱖ 75 years, 4.19; 95% CI, 1.98-8.87; P 5 .0004) and combination antithrombotic therapy (HR, 2.23; 95% CI, 1.49-3.34; P , .0001) were predictive of an increased risk of major bleeding. In univariate analyses excluding patients randomized to idraparinux (ie, the VKA-treated subgroup only), age ⱖ 75 years (HR, 4.00; 95% CI, 1.17-13.64; P 5 .037), previous ischemic stroke (HR, 2.47; 95% CI, 1.09-5.57; P 5 .030), and high CHADS2 stroke risk
www.chestpubs.org
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
CHEST / 140 / 1 / JULY, 2011
149
Table 2—Clinical Outcomes in Patients Receiving Anticoagulation Therapy Only or Combination Antithrombotic Therapy Antithrombotic Therapy
Clinical Outcomes Thromboembolic event Stroke or non-CNS systemic embolism Stroke Ischemic stroke Nonischemic stroke Non-CNS systemic embolism Cardiovascular death Myocardial infarction VTE event Bleeding event Any clinically relevant bleeding Major bleeding Intracranial hemorrhage Fatal bleeding Nonfatal bleeding into a critical organ Intracranial hemorrhage Retroperitoneal hemorrhage Pericardial hemorrhage Other bleeding into a critical organ Nonfatal bleeding associated with a fall in hemoglobin level ⱖ 2 g/dL and/or leading to a transfusion ⱖ 2 units of blood Mortality All-cause mortality Stroke Non-CNS systemic embolism Myocardial infarction VTE Bleeding Other cardiovascular cause Other known cause Not specified Composite end pointf
Nonstratified Analysis
Stratified Analysisc
Anticoagulation Therapy Only (n 5 3,728)a
Combination Antithrombotic Therapy (n 5 848)b
HR (95% CI)
P Value
HR (95% CI)
P Value
33 (1.0)
12 (1.5)
1.47 (0.76-2.85)
.25
1.26 (0.63-2.50)
.51
31 (1.0) 22 (0.7) 10 (0.3) 2 (, 0.1)
12 (1.5) 11 (1.4) 1 (0.1) 0
1.57 (0.80-3.05) 2.02 (0.98-4.18) 0.40 (0.05-3.09) …d
.19 .056 .38 …d
1.40 (0.70-2.80) 2.01 (0.94-4.30) 0.29 (0.04-2.30) …d
.34 .07 .24 …d
43 (1.3) 15 (0.5) 5 (0.2)
14 (1.7) 14 (1.7) 2 (0.2)
1.27 (0.70-2.33) 3.76 (1.81-7.79) 1.54 (0.30-7.94)
.44 .0004e .61
0.90 (0.49-1.68) 3.14 (1.45-6.78) 1.34 (0.24-7.61)
.75 .004e .74
364 (11.9)
208 (29.9)
2.52 (2.12-2.99)
, .0001e
2.50 (2.09-2.99)
, .0001e
64 (2.0) 21 (0.6) 10 (0.3) 17 (0.5)
39 (4.9) 9 (1.1) 5 (0.6) 10 (1.2)
2.44 (1.64-3.63) 1.68 (0.77-3.68) 1.96 (0.67-5.73) 2.32 (1.06-5.07)
, .0001e .19 .22 .035
2.26 (1.49-3.43) 1.61 (0.71-3.65) 1.58 (0.51-4.90) 2.23 (0.98-5.08)
.0001e .26 .43 .056
13 (0.4) 0
5 (0.6) 3 (0.4)
1.50 (0.54-4.22) …d
.44 …d
1.46 (0.50-4.28) …d
.49 …d
0 4 (0.1)
1 (0.1) 1 (0.1)
…d 1.04 (0.12-9.30)
…d .97
…d 1.17 (0.12-11.54)
…d .89
37 (1.1)
24 (3.0)
2.59 (1.55-4.34)
.0003e
2.43 (1.42-4.16)
.0012
86 (2.6) 1 (, 0.1) 0
37 (4.5) 0 0
1.68 (1.14-2.48) …d …d
.008e …d …d
1.33 (0.89-1.98) …d …d
.16 …d …d
4 (0.1) 1 (, 0.1) 10 (0.3) 37 (1.1) 29 (0.9) 4 (0.1) 171 (5.4)
2 (0.2) 0 5 (0.6) 12 (1.5) 17 (2.1) 1 (0.1) 86 (11.0)
2.01 (0.37-10.98) …d 1.96 (0.67-5.72) 1.26 (0.66-2.41) 2.33 (1.28-4.24) 0.86 (0.10-7.77) 2.06(1.59-2.66)
.42 …d .22 .49 .006 .90 , .0001e
1.36 (0.24-7.81) …d 1.58 (0.51-4.89) 0.90 (0.46-1.77) 2.09 (1.12-3.91) 0.76 (0.08-6.81) 1.79 (1.37-2.35)
.73 …d .43 .77 .02 .80 , .0001e
Data are presented as No. (% per year), unless otherwise indicated. HR 5 hazard ratio. See Table 1 legend for expansion of other abbreviations. aAll patients randomized to idraparinux or VKA who received no antiplatelets during the randomized treatment period. bAll patients randomized to idraparinux or VKA who received one or more antiplatelet drug at any time after randomization and up to the end of the randomized treatment period or up to the day before a thromboembolic event. cP value after stratification on propensity score in five categories based on quintiles. Propensity score was determined as the probability of receiving antiplatelets given age, sex, race, type of atrial fibrillation, previous ischemic stroke or TIA, left ventricular dysfunction, diabetes mellitus, previous symptomatic coronary artery disease, and previous VKA use. dNot estimatable. eSignificant at P ⱕ .05. fStroke, non-CNS systemic embolism, myocardial infarction, VTE, major bleeding, or death from any cause.
150
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
Original Research
Figure 1. Cumulative incidence of ischemic stroke for all randomized patients receiving anticoagulation therapy only or combination (combin) antithrombotic therapy.
Figure 2. Cumulative incidence of any clinically relevant bleeding for all randomized patients receiving anticoagulation therapy only or combination antithrombotic therapy. See Figure 1 legend for expansion of abbreviation.
score (HR, 3.02; 95% CI, 1.23-7.42; P 5 .024) were associated with major bleeding. In both univariate and multivariate analyses, combination antithrombotic therapy did not predict major bleeding (univariate HR, 1.82; 95% CI, 0.83-4.00; P 5 .14; multivariate HR, 1.61; 95% CI, 0.73-3.56; P 5 .24).
Even among patients randomized to receive VKA therapy (n 5 2,293), combination antithrombotic therapy doubled the risk of clinically relevant bleeding only. These results and those of other studies3,6,7,10 suggest that combination antithrombotic therapy in patients with AF increases the risk of major bleeding. However, the definition of bleeding used in studies and clinical trials varies considerably, thereby limiting direct comparisons of bleeding rates among studies; a universal bleeding definition would overcome this. Bleeding is multifactorial, and the benefit/risk ratio should be carefully considered prior to initiation of VKA or combining potent antithrombotic drugs in any patient. There are some instances, such as following an acute coronary syndrome (ACS) or percutaneous coronary intervention and stent implantation, where short-term concomitant antiplatelet therapy is still warranted.11 Of note, combination antithrombotic therapy was not associated with a reduction in arterial thromboembolic events but, instead, demonstrated an increased risk of myocardial infarction, although the actual number of events was small. However, patients at high cardiovascular risk may have been more likely to be prescribed antiplatelet drugs and, thus, may not benefit from combination therapy. There is also a common perception that antiplatelet drugs work on the platelet-rich thrombus (white clot) associated with atherosclerotic vascular disease, whereas warfarin is effective for the fibrin-rich thrombus (red clot) associated with AF.12 However, two trials (Antithrombotics in the Secondary Prevention of Events in Coronary Thrombosis-2 [ASPECT-2] and WarfarinAspirin Reinfarction Study II [WARIS-II]) comparing warfarin plus aspirin to warfarin only in patients with ACS also demonstrated that combination antithrombotic therapy did not confer a significant reduction in
Mortality A total of 123 (3.0% per year) patients died during follow-up, predominantly from other cardiovascular causes, bleeding, and other known causes. All-cause mortality was higher among patients receiving combination antithrombotic therapy than among those receiving anticoagulant therapy only, but this difference was not significant after stratification by propensity score (HR, 1.33; 95% CI, 0.89-1.98). In analyses restricted to patients randomized to VKA therapy, there was no significant difference in allcause mortality between those receiving VKA alone and those receiving combination antithrombotic therapy (16 [3.8% per year] vs 45 [2.6% per year], respectively; HR, 0.99; 95% CI, 0.55-1.78; P 5 .97 stratified by propensity score). Discussion The present post hoc analysis of the AMADEUS cohort demonstrates that age ⱖ 65 years and combination antithrombotic therapy were predictive of an increased risk of both clinically relevant and major bleeding among all randomized patients. Specifically, concomitant antiplatelet therapy in addition to anticoagulation (either idraparinux or VKA) independently predicted a 2.3- to 2.5-fold increased risk of major bleeding or clinically relevant bleeding, respectively, compared with anticoagulation therapy only. www.chestpubs.org
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
CHEST / 140 / 1 / JULY, 2011
151
Table 3—Univariate Predictors of Clinically Relevant Bleeding and Major Bleeding in All Patients Clinically Relevant Bleeding Patient Characteristic Age group , 65 ya 65-74 y ⱖ 75 y Sex Malea Female Ethnicity Whitea Nonwhite BMI, kg/m2 , 25a ⱖ 25 Baseline creatinine clearance level, mL/min/1.73 m2 , 50 ⱖ 50a Type of AF Paroxysmal Persistent Permanenta Previous stroke Yes Noa Previous TIA Yes Noa Previous non-CNS embolism Yes Noa Hypertension Yes Noa Left ventricular dysfunction Yes Noa Diabetes mellitus Yes Noa Symptomatic coronary artery disease Yes Noa CHADS2 risk category Lowa Moderate High Previous VKA (baseline) Previous VKA usea VKA naive Antithrombotic therapy Combination antithrombotic therapy Anticoagulation onlya
No.
No. Events (% Per Year)
HR (95% CI)
Major Bleeding P Value , .0001
No. Events (% Per Year)
HR (95% CI)
9 (0.8) 35 (1.9) 59 (3.7)
1.00 2.38 (1.14-4.94) 4.72 (2.34-9.52)
P Value , .0001b
b
1,147 1,818 1,611
101 (8.8) 237 (13.0) 234 (14.5)
1.00 1.47 (1.16-1.85) 1.70 (1.35-2.15)
3,043 1,533
365 (12.0) 207 (13.5)
1.00 1.12 (0.95-1.33)
.19 …
74 (2.4) 29 (1.9)
1.00 0.77 (0.50-1.18)
4,486 90
557 (12.4) 15 (16.7)
1.00 1.41 (0.84-2.35)
.19 …
102 (2.3) 1 (1.1)
1.00 0.50 (0.07-3.57)
.22 … … .49 …
1,050 3,526
151 (14.4) 421 (11.9)
1.00 0.81 (0.67-0.98)
… .026b
33 (3.1) 70 (2.0)
1.00 0.62 (0.41-0.94)
… .025b
819 3,757
122 (14.9) 450 (12.0)
1.30 (1.06-1.59) 1.00
.011b …
29 (3.5) 74 (2.0)
1.85 (1.20-2.84) 1.00
.005b …
1,634 438 2,504
207 (12.7) 53 (12.1) 312 (12.5)
1.09 (0.91-1.30) 1.04 (0.78-1.39) 1.00
.65 … …
32 (2.0) 10 (2.3) 61 (2.4)
0.86 (0.56-1.32) 0.99 (0.51-1.94) 1.00
.78 … …
580 3,996
80 (13.8) 492 (12.3)
1.15 (0.91-1.45) 1.00
.25 …
15 (2.6) 88 (2.2)
1.19 (0.69-2.06) 1.00
.53 …
478 4,098
63 (13.2) 509 (12.4)
1.07 (0.82-1.39) 1.00
.61 …
9 (1.9) 94 (2.3)
0.83 (0.42-1.65) 1.00
.60 …
156 4,420
19 (12.2) 553 (12.5)
1.00 (0.64-1.59) 1.00
.99 …
2 (1.3) 101 (2.3)
0.58 (0.14-2.35) 1.00
.44 …
3,530 1,046
435 (12.3) 137 (13.1)
0.99 (0.82-1.20) 1.00
.90 …
81 (2.3) 22 (2.1)
1.16 (0.72-1.86) 1.00
.54 …
1,072 3,504
140 (13.1) 432 (12.3)
1.03 (0.85-1.25) 1.00
.76 …
28 (2.6) 75 (2.1)
1.18 (0.76-1.82) 1.00
.46 …
897 3,679
110 (12.3) 462 (12.6)
0.97 (0.79-1.20) 1.00
.81 …
24 (2.7) 79 (2.1)
1.26 (0.80-1.98) 1.00
.33 …
1,408 3,168
202 (14.3) 370 (11.7)
1.21 (1.02-1.43) 1.00
.03b …
35 (2.5) 68 (2.1)
1.14 (0.76-1.71) 1.00
.54 …
1,880 1,448 1,248
222 (11.8) 174 (12.0) 176 (14.1)
1.00 1.02 (0.84-1.25) 1.23 (1.01-1.50)
.093b … …
34 (1.8) 26 (1.8) 43 (3.4)
1.00 1.00 (0.60-1.66) 1.95 (1.24-3.05)
.0037b … …
3,476 1,100
414 (11.9) 158 (14.4)
1.00 1.29 (1.07-1.55)
.007b …
72 (2.1) 31 (2.8)
1.00 1.45 (0.95-2.20)
.087 …
848
208 (24.5)
2.52 (2.12-2.99)
, .0001b
39 (4.6)
2.44 (1.64-3.63)
, .0001b
364 (9.8)
1.00
64 (1.7)
1.00
3,728
…
…
See Table 1 and 2 legends for expansion of abbreviations. aReference category. bSignificant at P ⱕ .05. 152
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
Original Research
Table 4—Multivariate Predictors of Clinically Relevant Bleeding and Major Bleeding in All Patients Clinically Relevant Bleeding Patient Characteristic Antithrombotic therapy Combination antithrombotic therapy Anticoagulation alonea Age group , 65 ya 65-74 y ⱖ 75 y BMI, kg/m2 . 25a ⱖ 25 Baseline creatinine clearance, mL/min/1.73 mm2 , 50 ⱖ 50a Previous VKA (baseline) Previous VKA usea VKA naive Symptomatic CAD Yes Noa CHADS2 risk category Lowa Moderate High
Major Bleeding
No. Events (% Per Year)
HR (95% CI)
P Value
No. Events (% Per Year)
HR (95% CI)
P Value
208 (24.5)
2.47 (2.07-2.96)
, .0001b
39 (4.6)
2.23 (1.49-3.34)
, .0001b
3,728
364 (9.8)
1.00
…
64 (1.7)
1.00
1,147 1,818 1,611
101 (8.8) 237 (13.0) 234 (14.5)
1.00 1.44 (1.14-1.82) 1.59 (1.24-2.04)
.001b … …
9 (0.8) 35 (1.9) 59 (3.7)
1.00 2.26 (1.08-4.71) 4.19 (1.98-8.87)
.0004b … …
1,050 3,526
151 (14.4) 421 (11.9)
1.00 0.89 (0.73-1.08)
.23 …
33 (3.1) 70 (2.0)
1.00 0.81 (0.52-1.25)
.34 …
819 3,757
122 (14.9) 450 (12.0)
1.00 (0.80-1.26) 1.00
.99 …
29 (3.5) 74 (2.0)
1.03 (0.64-1.67) 1.00
.89 …
3,476 1,100
414 (11.9) 158 (14.4)
1.00 1.09 (0.90-1.31)
.38 …
… …
… …
… …
1,408 3,168
202 (14.3) 370 (11.7)
0.96 (0.80-1.15) 1.00
.64 …
… …
… …
… …
1,880 1,448 1,248
… … …
… … …
… … …
34 (1.8) 26 (1.8) 43 (3.4)
1.00 0.69 (0.40-1.19) 1.18 (0.72-1.95)
.10 … …
No. 848
…
Only statistically significant variables in the univariate analysis (P ⱕ .05) were included in the multivariate analysis. See Table 1 and 2 legends for expansion of abbreviations. aReference category. bSignificant at P ⱕ .05.
thromboembolic and mortality risk compared with warfarin only, although there was a trend for a reduction in such events.13,14 In addition, the trend toward more ischemic strokes among patients receiving combination antithrombotic therapy may have been due
Figure 3. Cumulative incidence of major bleeding for all randomized patients receiving anticoagulation therapy only or combination antithrombotic therapy. See Figure 1 legend for expansion of abbreviation.
to a higher intrinsic risk of stroke (older, previous stroke/TIA, diabetes mellitus, left ventricular dysfunction, and symptomatic CAD), hence, the addition of antiplatelet therapy among these patients, when compared with those receiving oral anticoagulation alone (and this finding is despite the propensity score adjustments). A post hoc analysis of the combined Stroke Prevention Using Oral Thrombin Inhibitor in Atrial Fibrillation (SPORTIF) III and V data sets also revealed that combination antithrombotic therapy (warfarin or ximelagatran plus aspirin) significantly increased the risk of bleeding (major and minor) compared with treatment with either anticoagulant only.6 Combination therapy was not associated with a reduction in stroke or vascular events (death or myocardial infarction),6 but a 1.58-fold (95% CI, 1.01-2.49) increase in clinically relevant bleeding compared with treatment with warfarin alone was apparent, findings in keeping with the current study. In contrast to the SPORTIF cohort,6 the present study found that combination antithrombotic therapy did not independently increase the risk of major bleeding perhaps because of the smaller sample size in the VKA subgroup with fewer major
www.chestpubs.org
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
CHEST / 140 / 1 / JULY, 2011
153
bleeding events. Further, although halted early due to poor recruitment, the Fluindione, Fibrillation Auriculaire, Aspirin et Contraste Spontané (FFAACS) study, which compared fluindione (INR range, 2.0-2.6) plus placebo (anticoagulation alone) with fluindione (same INR range) plus aspirin (combination therapy), also demonstrated a higher bleeding event rate in the combination arm.4 Conversely, the National Study for Prevention of Embolism in Atrial Fibrillation (NASPEAF),10 which compared combination acenocoumarol (INR range, 2.0-3.0) and the antiplatelet agent triflusal, 600 mg/d, with either acenocoumarol or triflusal alone in patients at intermediate or high risk of thromboembolism, demonstrated that combination therapy was associated with an approximately 60% relative risk reduction in the composite outcome of vascular events and severe bleeding compared with either acenocoumarol or triflusal alone. A systematic review and meta-analysis of randomized controlled trials of combination antithrombotic therapy (oral anticoagulation plus aspirin) vs oral anticoagulation therapy only3 in patients at risk for vascular outcomes (including two trials in patients with AF: FFAACS4 and Second Copenhagen Atrial Fibrillation, Aspirin and Anticoagulant Therapy Study [AFASAK-2]5) also demonstrated that combination antithrombotic therapy did not reduce the risk of arterial thromboembolism or all-cause mortality but did significantly increase the risk of major bleeding. In the present analyses, patients receiving combination antithrombotic therapy were significantly less likely to have been prescribed a VKA prior to enrollment in the AMADEUS trial. This finding may have had an effect on the incidence of bleeding because patients with previous VKA experience (who by definition have not experienced significant problems with VKA, ie, “oral anticoagulation survivors”) may be less likely to experience problems with subsequent anticoagulation. Inception studies of oral anticoagulation therapy have demonstrated that the risk of bleeding is greatest during the first 90 days of VKA therapy.15 Therefore, those patients who experienced clinically relevant bleeding or major bleeding did so because they were “VKA naive” and because of the use of combination antithrombotic therapy. Indeed, nonuse of VKAs prior to randomization was a univariate predictor of clinically relevant bleeding in the present analyses. A retrospective analysis of VKAnaive patients commencing either acenocoumarol or phenprocoumon demonstrated that any concomitant antiplatelet therapy increased the risk of major bleeding, with the greatest risk of bleeding evident with concomitant clopidogrel use followed by aspirin, even after adjustment for use of nonsteroidal antiinflammatory drugs, antibiotics, steroids, and gastroprotective agents.16 Further, a recent Danish
registry of 85,541 patients experiencing first-time hospitalization for AF, with at least one prescription for warfarin, aspirin, or clopidogrel after discharge, demonstrated that the combined end point of hospitalization or death from bleeding or anemia was highest among those receiving dual (warfarin and clopidogrel) or triple therapy (15.5% and 17.5% per patient-year, respectively).7 The same investigators also reported a higher risk of bleeding in patients with acute myocardial infarction treated with different combinations of warfarin, aspirin, or clopidogrel.17 Nonetheless, these studies7,17 were based on a large prescriptions registry where treatments were nonrandomized and many clinical parameters of the patients are unknown; in contrast, the present analysis is based on a prospective trial setting where patients were carefully followed up. Limitations The present study has several limitations. First, the AMADEUS study was not originally powered to undertake these post hoc analyses, and analyses were run on parameters not taken into account in the randomization process. In clinical practice, many patients with AF are given combination warfarin and antiplatelet therapy,1 and patients with AF and an ACS, coronary stenting, or both often are prescribed dual or triple antithrombotic therapy, despite the current lack of evidence base to support this. Therefore, post hoc analyses of antithrombotic therapy in AF trial cohorts are currently the only way to try to answer the question of the effect of combination antithrombotic therapy in patients with AF. Second, the use of antiplatelet therapy was not randomized in the AMADEUS trial,8 and as we alluded to earlier, it is possible that significant differences between the groups in baseline demographic and clinical characteristics may confound the present findings. Although the use of propensity score and multivariate analyses can reduce the potential bias, some other factors not recorded in the AMADEUS study may affect the results of the analyses. For example, a history of bleeding, one of the most important predictors of recurrent bleeding, was not recorded in the AMADEUS cohort, and, therefore, it was not possible to examine the effect of this factor on bleeding risk in our multivariate analyses. However, patients with previous significant bleeding often are excluded from clinical trials of thromboprophylaxis, although previous major bleeding was not explicitly listed as an exclusion criterion in the AMADEUS cohort.8 There also exists the possibility of residual confounding factors behind a significant difference between the groups (eg, local practice, access to VKA, medical history).
154
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
Original Research
Conclusions Combination antithrombotic therapy increases the risk of clinically relevant bleeding and major bleeding in patients with AF and does not appear to reduce the risk of stroke. Therefore, combination antithrombotic treatment is justified only in patients with a clear indication that the benefit of adding antiplatelet therapy to anticoagulant treatment outweighs the increased risk of bleeding.
5.
6.
Acknowledgments Author contributions: Dr Minini had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr Lane: contributed design of the analyses, had full access to the results of the data analysis, and contributed to the preparation and revision of the manuscript. Dr Kamphuisen: contributed design of the analyses and the preparation and revision of the manuscript. Dr Minini: contributed preparation of the manuscript. Dr Büller: contributed design of the analyses and the preparation and revision of the manuscript. Dr Lip: contributed design of the analyses and preparation and revision of the manuscript. Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Lane is in receipt of an investigator-initiated educational grant from Bayer and has received industry-funded sponsorship for travel to educational meetings from AstraZeneca and Sanofi-Aventis. Dr Kamphuisen has received an independent research grant from the Ruitinga van Swieten Foundation, The Netherlands. He is a consultant for Bayer, Boehringer-Ingelheim, CSL Behring, and Ablynx NV and has received investigator-initiated research grants from Bayer, Leo Pharma Inc, Wyeth, and CSL Behring. Dr Minini is an employee of Sanofi-Aventis. Dr Büller has served as a consultant to Sanofi-Aventis, Bayer, Pfizer, GlaxoSmithKline, Astellas Pharma, Boehringer-Ingelheim, and Daiichi-Sankyo. Dr Lip has served as a consultant for Bayer, Astellas, Merck, AstraZeneca, SanofiAventis, Bristol-Myers Squibb/Pfizer, and Boehringher Ingelheim and has been on the speakers bureau for Bayer, Sanofi-Aventis, Bristol-Myers Squibb/Pfizer, and Boehringher Ingelheim. Role of sponsors: Sanofi-Aventis provided access to the AMADEUS data set for these analyses to be undertaken. Other contributions: We thank the AMADEUS Steering Committee for its comments and support of this subanalysis of the AMADEUS cohort.
7.
8.
9. 10.
11.
12. 13.
References 1. Shireman TI, Howard PA, Kresowik TF, Ellerbeck EF. Combined anticoagulant-antiplatelet use and major bleeding events in elderly atrial fibrillation patients. Stroke. 2004; 35(10):2362-2367. 2. Johnson SG, Witt DM, Eddy TR, Delate T. Warfarin and antiplatelet combination use among commercially insured patients enrolled in an anticoagulation management service. Chest. 2007;131(5):1500-1507. 3. Dentali F, Douketis JD, Lim W, Crowther M. Combined aspirin-oral anticoagulant therapy compared with oral anticoagulant therapy alone among patients at risk for cardiovascular disease: a meta-analysis of randomized trials. Arch Intern Med. 2007;167(2):117-124. 4. Lechat P, Lardoux H, Mallet A, et al; FFAACS (Fluindione, Fibrillation Auriculaire, Aspirin et Contraste Spontané)
14. 15.
16. 17.
Investigators. Anticoagulant (fluindione)-aspirin combination in patients with high-risk atrial fibrillation. A randomized trial (Fluindione, Fibrillation Auriculaire, Aspirin et Contraste Spontané; FFAACS). Cerebrovasc Dis. 2001;12(3): 245-252. Gulløv AL, Koefoed BG, Petersen P. Bleeding during warfarin and aspirin therapy in patients with atrial fibrillation: the AFASAK 2 study. Atrial Fibrillation Aspirin and Anticoagulation. Arch Intern Med. 1999;159(12):1322-1328. Flaker GC, Gruber M, Connolly SJ, et al; SPORTIF Investigators. Risks and benefits of combining aspirin with anticoagulant therapy in patients with atrial fibrillation: an exploratory analysis of stroke prevention using an oral thrombin inhibitor in atrial fibrillation (SPORTIF) trials. Am Heart J. 2006;152(5):967-973. Hansen ML, Sørensen R, Clausen MT, et al. Risk of bleeding with single, dual, or triple therapy with warfarin, aspirin, and clopidogrel in patients with atrial fibrillation. Arch Intern Med. 2010;170(16):1433-1441. Bousser MG, Bouthier J, Büller HR, et al; Amadeus Investigators. Comparison of idraparinux with vitamin K antagonists for prevention of thromboembolism in patients with atrial fibrillation: a randomised, open-label, non-inferiority trial [published correction appears in Lancet. 2008;372(9655): 2022]. Lancet. 2008;371(9609):315-321. D’Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med. 1998;17(19):2265-2281. Pérez-Gómez F, Alegría E, Berjón J, et al; NASPEAF Investigators. Comparative effects of antiplatelet, anticoagulant, or combined therapy in patients with valvular and nonvalvular atrial fibrillation: a randomized multicenter study. J Am Coll Cardiol. 2004;44(8):1557-1566. Lip GY, Huber K, Andreotti F, et al; European Society of Cardiology Working Group on Thrombosis. Management of antithrombotic therapy in atrial fibrillation patients presenting with acute coronary syndrome and/or undergoing percutaneous coronary intervention/stenting [published correction appears in Thromb Haemost. 2010;104(3):653]. Thromb Haemost. 2010;103(1):13-28. Lip GYH. Don’t add aspirin for associated stable vascular disease in a patient with atrial fibrillation receiving anticoagulation. BMJ. 2008;336(7644):614-615. van Es RF, Jonker JJ, Verheugt FW, Deckers JW, Grobbee DE; Antithrombotics in the Secondary Prevention of Events in Coronary Thrombosis-2 (ASPECT-2) Research Group. Aspirin and Coumadin after acute coronary syndromes (the ASPECT-2 study): a randomised controlled trial. Lancet. 2002;360(9327):109-113. Hurlen M, Abdelnoor M, Smith P, Erikssen J, Arnesen H. Warfarin, aspirin, or both after myocardial infarction. N Engl J Med. 2002;347(13):969-974. Hylek EM, Evans-Molina C, Shea C, Henault LE, Regan S. Major hemorrhage and tolerability of warfarin in the first year of therapy among elderly patients with atrial fibrillation. Circulation. 2007;115(21):2689-2696. Schalekamp T, Klungel OH, Souverein PC, de Boer A. Effect of oral antiplatelet agents on major bleeding in users of coumarins. Thromb Haemost. 2008;100(6):1076-1083. Sørensen R, Hansen ML, Abildstrom SZ, et al. Risk of bleeding in patients with acute myocardial infarction treated with different combinations of aspirin, clopidogrel, and vitamin K antagonists in Denmark: a retrospective analysis of nationwide registry data. Lancet. 2009;374(9706):1967-1974.
www.chestpubs.org
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
CHEST / 140 / 1 / JULY, 2011
155
Original Research ANTITHROMBOTIC THERAPY
Bleeding Risk in Patients With Atrial Fibrillation The AMADEUS Study Deirdre A. Lane, PhD; Pieter W. Kamphuisen, MD; Pascal Minini, PhD; Harry R. Büller, MD; and Gregory Y. H. Lip, MD
Objective: This study aimed to assess the impact of combination antithrombotic therapy on stroke and bleeding risk compared with anticoagulation therapy only in patients with atrial fibrillation (AF). Methods: Post hoc analysis of 4,576 patients with AF (mean ⫾ SD age, 70.1 ⫾ 9.1 years; men, 66.5%) enrolled in the Evaluating the Use of SR34006 Compared to Warfarin or Acenocoumarol in Patients With Atrial Fibrillation (AMADEUS) trial were randomized to receive either subcutaneous idraparinux (2.5 mg weekly) (n 5 2,283) or dose-adjusted vitamin K antagonists (VKAs) (international normalized ratio, 2.0-3.0) (n 5 2,293). Of these patients, 848 (18.5%) received antiplatelet therapy (aspirin, clopidogrel, ticlopidine, etc) in addition to anticoagulation treatment (combination antithrombotic therapy). Results: A total of 572 (15.3% per year) clinically relevant bleeding and 103 (2.6% per year) major bleeding events occurred. Patients receiving combination antithrombotic therapy had a 2.3- to 2.5-fold increased risk of clinically relevant bleeding events and major bleeding events, respectively, compared with those receiving anticoagulation therapy only. Multivariate analyses (hazard ratio, 95% CI) revealed that the risk of clinically relevant bleeding was significantly increased by age 65 to 74 years (1.44, 1.14-1.82) and ⱖ 75 years (1.59, 1.24-2.04, P 5 .001) and by combination antithrombotic therapy (2.47, 2.07-2.96, P , .0001). The same held true for major bleeding events, with analogous figures for age 65 to 74 years (2.26, 1.08-4.71) and ⱖ 75 years (4.19, 1.98-8.87, P 5 .0004) and for combination antithrombotic therapy (2.23, 1.49-3.34, P , .0001). Combination antithrombotic therapy was not associated with a decrease in ischemic stroke risk compared with anticoagulation therapy only (11 [1.4% per year] vs 22 [0.7% per year]; adjusted hazard ratio, 2.01; 95% CI, 0.94-4.30; P 5 .07). Conclusions: Combination antithrombotic therapy increases the risk of clinically relevant bleeding and major bleeding in patients with AF and does not appear to reduce the risk of stroke. CHEST 2011; 140(1):146–155 Abbreviations: ACS 5 acute coronary syndrome; AF 5 atrial fibrillation; AMADEUS 5 Evaluating the Use of SR34006 Compared to Warfarin or Acenocoumarol in Patients With Atrial Fibrillation; CAD 5 coronary artery disease; CHADS2 5 congestive heart failure, hypertension, age ⱖ 75 years, diabetes, and prior stroke or transient ischemic stroke; CIAC 5 Central Independent Adjudication Committee; INR 5 international normalized ratio; TIA 5 transient ischemic attack; VKA 5 vitamin K agonist
of antithrombotic therapy for atrial Prescription fibrillation (AF) must consider both the patient’s
stroke risk and his or her risk of bleeding while on anticoagulant therapy; that is, there must be a net clinical benefit to initiate anticoagulant therapy. Given that AF commonly coexists with associated atherosclerotic vascular disease, a significant proportion of patients with AF receiving anticoagulation therapy also receive concomitant antiplatelet therapy.1,2 The
benefit of combination antithrombotic therapy (anticoagulant plus antiplatelets) in patients with AF has yet to be elucidated from clinical trials, but combination antithrombotic therapy in this population has been shown to be associated with an increased risk of bleeding1,3-7 with no significant reduction in thromboembolic events.6 The present study is a post hoc analysis of predictors of major bleeding and clinically relevant bleeding
146
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
Original Research
in patients with AF receiving anticoagulation therapy only (either subcutaneous idraparinux or vitamin K antagonists [VKAs]) or combination antithrombotic therapy (one or more antiplatelet therapies together with anticoagulation) from the Evaluating the Use of SR34006 Compared to Warfarin or Acenocoumarol in Patients With Atrial Fibrillation (AMADEUS) trial cohort.8 The AMADEUS trial was halted prematurely after a mean ⫾ SD follow-up period of 10.7 ⫾ 5.4 months due to an excess of major bleeding and clinically relevant bleeding with idraparinux (346 cases, 19.7/100 patient-years) compared with VKA (warfarin or acenocoumarol) (226 cases, 11.3/100 patient-years, P , .001). However, idraparinux was noninferior to VKAs in the prevention of stroke or non-CNS systemic embolism (18 cases vs 27 cases; 0.9 vs 1.3/100 patient-years, respectively; hazard ratio [HR], 0.71; 95% CI, 0.39-1.30; P 5 .007 for noninferiority).8 Using the AMADEUS study cohort, we tested the hypothesis that patients with AF receiving combination antithrombotic therapy will experience a greater incidence of clinically relevant bleeding and major bleeding compared with patients receiving anticoagulation therapy only. The present study also examined predictors of bleeding in patients with AF receiving anticoagulation therapy only compared with those receiving combination antithrombotic therapy. Given the excess of clinically relevant bleeding with idraparinux, separate analyses were also undertaken to compare VKA alone to VKA plus antiplatelet therapy. Materials and Methods The details of the AMADEUS trial have been described previously.8 In summary, 4,576 patients with ECG-documented AF requiring long-term anticoagulation therapy (one or more risk factors for stroke [previous ischemic stroke, transient ischemic attack (TIA), or systemic embolism; aged ⱖ 75 years; hypertension requiring treatment, or left ventricular dysfunction] or those aged 65-75 years with either diabetes mellitus or coronary artery Manuscript received December 21, 2010; revision accepted February 9, 2011. Affiliations: From the University of Birmingham Centre for Cardiovascular Sciences (Drs Lane and Lip), City Hospital, Birmingham, England; Department of Vascular Medicine (Drs Kamphuisen and Büller), Academic Medical Centre, Amsterdam, The Netherlands; and Sanofi-Aventis Research and Development, Biostatistics and Programming (Dr Minini), ChillyMazarin, France. Funding/Support: The AMADEUS trial was funded by SanofiAventis. Correspondence to: Deirdre A. Lane, PhD, University of Birmingham Centre for Cardiovascular Sciences, City Hospital, Dudley Rd, Birmingham B18 7QH, England; e-mail: deirdrelane@ nhs.net © 2011 American College of Chest Physicians. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (http://www.chestpubs.org/ site/misc/reprints.xhtml). DOI: 10.1378/chest.10-3270
THE BOTTOM LINE How does this work advance the field? This large trial cohort study suggests that the risk of clinically relevant bleeding is significantly increased with age and the use of combination antithrombotic therapy. It provides strong evidence against concomitant prescription of oral anticoagulation plus antiplatelet therapy in stable patients with atrial fibrillation, given the lack of benefit for vascular events and the substantial increase in bleeding risk.
What are the clinical implications? Combination antithrombotic therapy increases the risk of clinically relevant bleeding and major bleeding in patients with atrial fibrillation and does not appear to reduce the risk of stroke.
disease [CAD]) were randomized to receive either idraparinux 2.5 mg once per week in prefilled syringes administered subcutaneously or dose-adjusted (international normalized ratio [INR] range, 2.0-3.0) open-label VKAs (warfarin or acenocoumarol) between September 2003 and July 2005. For patients with a baseline creatinine clearance level of 10 to 30 mL/min/1.73 mm2, the second and all subsequent doses of idraparinux were 1.5 mg weekly.8 Definition of Bleeding The main safety outcome was clinically relevant bleeding, which was further classified by a Central Independent Adjudication Committee (CIAC) into major bleeding (bleeding that was fatal, intracranial, or affected another critical anatomic site or overt bleeding with a drop of hemoglobin level ⱖ 2 g/dL or requiring transfusion of ⱖ 2 units of blood) and clinically relevant nonmajor bleeding (overt bleeding that did not satisfy the criteria for major bleeding but met defined criteria and included repetitive epistaxis for ⱖ 5 min at least twice in a 24-h period, hematuria [spontaneous or lasting . 24 h], hematemesis, and subcutaneous hematomas ⱖ 25 cm2 if spontaneous or . 100 cm2 if after trauma). All suspected safety or efficacy outcome events were adjudicated by a CIAC unaware of treatment assignment.8 For the present analyses, only events confirmed by the CIAC are described. The AMADEUS trial contained a subset of 848 (18.5%) patients receiving combination antithrombotic therapy (anticoagulant and antiplatelet therapy). In addition to anticoagulation therapy (either idraparinux or VKA), 716 (15.6%) patients took aspirin only, 34 (0.7%) received a thienopyridine (clopidogrel or ticlopidine) only, 12 (0.3%) received other antiplatelet therapy (one of the following: dipyridamole, cilostazol, eptifibatide, tirofiban hydrochloride, or anagrelide), and 86 (1.9%) received two or more antiplatelets. Statistical Analysis Data analyses were performed using SAS, version 8.02 (SAS Institute Inc; Cary, North Carolina) statistical software. All statistical tests were two tailed, with P ⱕ .05 considered statistically significant. Because antiplatelet therapy was not randomized in the AMADEUS trial, direct comparison of the treatment group may be misleading. In order to reduce potential bias due to imbalance between groups at baseline, the propensity score method9 was used. The propensity score was defined as the conditional probability of receiving antiplatelets given baseline characteristics
www.chestpubs.org
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
CHEST / 140 / 1 / JULY, 2011
147
of patients (Table 1) and estimated using a multivariate logistic regression. After selection of the most predictive factors using a stepwise procedure, nine covariates were included in the final propensity score model: age, sex, race, type of AF, previous ischemic stroke or TIA, previous symptomatic CAD, left ventricular dysfunction, diabetes mellitus, and previous VKA use. Patients were then grouped into five approximately equally sized strata defined according to the quintiles of the propensity scores. Comparison of the baseline characteristics between the two groups was performed before and after stratification on the propensity score to determine whether balance was achieved after stratification. The incidence of clinical events during the study was estimated from the event rate per 100 patient-years (annual rates). The effect of antiplatelet therapy in addition to anticoagulation therapy was assessed using a Cox proportional hazard model stratified on the propensity score in five categories. HRs, 95% CIs, and P values were determined from this model. In addition, univariate and multivariate analyses were performed in order to identify significant independent factors of clinical events. Each potentially predictive factor was classified in two or three categories, one of them being considered the reference category in order to aid interpretation. Significant factors in the univariate analysis were then introduced in the multivariate model. Note that in these analyses, combination antithrombotic therapy was defined as any antiplatelet medication received at any time after randomization and up to the end of the randomized treatment or up to the day before occurrence of a thromboembolic event. Patients who received antiplatelets after a thromboembolic event were analyzed in the anticoagulation-alone group.
Results Of the total cohort of 4,576 patients with AF (age, 70.1 ⫾ 9.1 years; men, 66.5%), 3,728 (81.5%) received anticoagulation therapy only (1,861 [40.7%] idraparinux and 1,867 [40.8%] VKA). In addition, 848 (18.5%) patients received combination antithrombotic therapy (Table 1). Combination antithrombotic therapy was more common with increasing age, particularly in patients aged ⱖ 75 years, with poorer renal function, a previous stroke or TIA, left ventricular dysfunction, diabetes mellitus (and consequently increasing CHADS2 [congestive heart failure, hypertension, age ⱖ 75 years, diabetes mellitus, and prior stroke or TIA] score), and symptomatic CAD (Table 1). In addition, patients with paroxysmal AF were more likely to receive combination antithrombotic therapy compared with those with persistent or permanent AF, whereas patients who had received VKA prior to enrollment were less likely to receive concomitant antiplatelet therapy. After stratification on the propensity score, there was no statistically significant difference between the two groups for any of the baseline characteristics. Within each propensity score quintile, the two groups were comparable. Similar results emerged when analyses were confined to those receiving VKA therapy only or in combination (data not shown), although there was no significant difference in age (P 5 .17) or in patients with poorer renal function (P 5 .052).
Thromboembolic Risk During follow-up, 43 (1.1% per year) strokes occurred, with 31 (1.0% per year) in patients receiving anticoagulation therapy only and 12 (1.5% per year) in those receiving combination antithrombotic therapy (Table 2). There was a trend toward more ischemic strokes occurring with combination antithrombotic therapy compared with anticoagulation therapy alone (HR, 2.01; 95% CI, 0.94-4.30; P 5 .073) (Fig 1). Similar results emerged when analyses were restricted to patients randomized to VKA therapy (7 [1.7% per year] vs 13 [0.8% per year]; HR, 2.25; 95% CI, 0.85-5.97; P 5 .10 stratified by propensity score). There was a threefold increased risk of myocardial infarction among patients with AF receiving combination antithrombotic therapy compared with anticoagulation therapy only (HR, 3.14; 95% CI, 1.45-6.78; P 5 .0036). In analyses confined to patients randomized to VKA therapy, there was an increased risk of myocardial infarction with combination antithrombotic therapy compared with anticoagulation therapy only (6 [1.4% per year] vs 7 [0.4% per year], respectively; HR, 3.45; 95% CI, 1.16-10.29; P 5 .03); however, stratifying by propensity score rendered this association not significant (HR, 2.70; 95% CI, 0.85-8.56; P 5 .09). Clinically Relevant Bleeding A total of 572 (15.3% per year) clinically relevant bleeding events occurred, with significantly more bleeding events among patients receiving combination antithrombotic therapy compared with anticoagulation therapy only (Fig 2, Table 2). Similar results emerged when analyses were confined only to patients randomized to VKA therapy (81 [21.7% per year] vs 145 [9.0% per year]; HR, 2.35; 95% CI, 1.77-3.13; P , .0001 stratified by propensity score). In univariate analyses of all randomized patients, age ⱖ 65 years, BMI , 25 kg/m2, poor renal function (baseline creatinine clearance level , 50 mL/min/1.73 m2), symptomatic CAD, VKA naivety at baseline, and combination antithrombotic therapy were significantly associated with clinically relevant bleeding (Table 3). Multivariate analyses revealed that age ⱖ 65 years (HR for age 65-75 years, 1.44; 95% CI, 1.14-1.82; HR for ⱖ 75 years, 1.59; 95% CI, 1.24-2.04; P 5 .001) and combination antithrombotic therapy (HR, 2.47; 95% CI, 2.07-2.96; P , .0001) predicted increased risk of clinically relevant bleeding (Table 4). When univariate analyses were restricted to patients taking VKA (n 5 2,293), only age 65 to 74 years (HR, 1.59; 95% CI, 1.12-2.27; P 5 .033), symptomatic CAD (HR, 1.58; 95% CI, 1.21-2.05; P 5 .0007), and combination antithrombotic therapy (HR, 2.44; 95% CI, 1.86-3.20; P , .0001) were significant
148
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
Original Research
Table 1—Patient Baseline Demographic and Clinical Characteristics by Antithrombotic Therapy Baseline Characteristic Age, y Age group , 65 y 65-74 y ⱖ 75 y Male sex Ethnicity White Black Asian Other BMI, kg/m2 Creatinine clearance level, mL/min/1.73 m2 Type of AF Paroxysmal AF Persistent Permanent Stroke risk factors Previous ischemic stroke Previous TIA Previous non-CNS systemic embolism Hypertension requiring drug treatment Left ventricular dysfunction Diabetes mellitus Symptomatic CAD CHADS2 score CHADS2 risk category Low risk (score 0-1) Moderate risk (score 2) High risk (score ⱖ 3) Randomized treatment group Idraparinux VKA Previous VKA use
Anticoagulation Therapy Only (n 5 3,728)a
Combination Antithrombotic Therapy (n 5 848)b
70.0 ⫾ 9.0
70.7 ⫾ 9.3
948 (25.4) 1,501 (40.3) 1,279 (34.3) 2,459 (66.0)
199 (23.5) 317 (37.4) 332 (39.2) 583 (68.8)
3,664 (98.3) 18 (0.5) 14 (0.4) 31 (0.8) 28.8 ⫾ 5.4 72.3 (55.8-92.4)
821 (96.8) 10 (1.2) 2 (0.2) 15 (1.8) 28.9 ⫾ 5.7 69.4 (52.2-90.1)
P Value .026 .017 … … .12 .01 … … … .51 .003
1,284 (34.6) 374 (10.1) 2,055 (55.3)
350 (41.5) 64 (7.6) 429 (50.9)
.0003 … …
447 (12.0) 372 (10.0) 129 (3.5) 2,895 (77.7) 805 (21.6) 686 (18.4) 980 (26.3) 1.9 ⫾ 1.1
133 (15.7) 106 (12.5) 27 (3.2) 635 (74.9) 267 (31.5) 211 (24.9) 428 (50.5) 2.2 ⫾ 1.2
.004 .03 .69 .08 , .0001 , .0001 , .0001 , .0001
1,599 (42.9) 1,175 (31.5) 951 (25.5)
278 (32.8) 273 (32.2) 297 (35.0)
… , .0001 …
1,861 (49.9) 1,867 (50.1) 2,939 (78.8)
422 (49.8) 426 (50.2) 537 (63.3)
.93 … , .0001
Data are presented as No. (%), mean ⫾ SD, or median (interquartile range). AF 5 atrial fibrillation; CAD 5 coronary artery disease; CHADS2 5 congestive heart failure, hypertension, age . 75 years, diabetes mellitus, prior stroke or transient ischemic attack; TIA 5 transient ischemic attack; VKA 5 vitamin K agonist. aAll patients randomized to idraparinux or VKA who received no antiplatelets during the randomized treatment period. bAll patients randomized to idraparinux or VKA who received one or more antiplatelet drug at any time after randomization and up to the end of the randomized treatment period or up to the day before a thromboembolic event.
predictors of clinically relevant bleeding. In multivariate analyses, only combination antithrombotic therapy was an independent predictor of clinically relevant bleeding (HR, 2.30; 95% CI, 1.73-3.04; P , .0001). Major Bleeding A total of 103 (2.6% per year) major bleeding events occurred and were more common with combination antithrombotic therapy than with anticoagulation therapy only (Fig 3, Table 2). However, in subgroup analyses confined to patients randomized to VKA therapy, combination antithrombotic therapy was not a significant predictor of major bleeding (9 [2.2% per year] vs 20 [1.2% per year]; HR, 1.46; 95% CI, 0.65-3.28; P 5 .36 stratified by propensity score).
In univariate analyses of all randomized patients, age ⱖ 65 years, poorer renal function (baseline creatinine clearance , 50 mL/min/1.73 mm2), BMI , 25 kg/m2, high CHADS2 stroke risk score, and combination antithrombotic therapy were significantly associated with major bleeding (Table 3). In multivariate analyses including all randomized patients, only age ⱖ 65 years (HR for 65-74 years, 2.26; 95% CI, 1.08-4.71; HR for ⱖ 75 years, 4.19; 95% CI, 1.98-8.87; P 5 .0004) and combination antithrombotic therapy (HR, 2.23; 95% CI, 1.49-3.34; P , .0001) were predictive of an increased risk of major bleeding. In univariate analyses excluding patients randomized to idraparinux (ie, the VKA-treated subgroup only), age ⱖ 75 years (HR, 4.00; 95% CI, 1.17-13.64; P 5 .037), previous ischemic stroke (HR, 2.47; 95% CI, 1.09-5.57; P 5 .030), and high CHADS2 stroke risk
www.chestpubs.org
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
CHEST / 140 / 1 / JULY, 2011
149
Table 2—Clinical Outcomes in Patients Receiving Anticoagulation Therapy Only or Combination Antithrombotic Therapy Antithrombotic Therapy
Clinical Outcomes Thromboembolic event Stroke or non-CNS systemic embolism Stroke Ischemic stroke Nonischemic stroke Non-CNS systemic embolism Cardiovascular death Myocardial infarction VTE event Bleeding event Any clinically relevant bleeding Major bleeding Intracranial hemorrhage Fatal bleeding Nonfatal bleeding into a critical organ Intracranial hemorrhage Retroperitoneal hemorrhage Pericardial hemorrhage Other bleeding into a critical organ Nonfatal bleeding associated with a fall in hemoglobin level ⱖ 2 g/dL and/or leading to a transfusion ⱖ 2 units of blood Mortality All-cause mortality Stroke Non-CNS systemic embolism Myocardial infarction VTE Bleeding Other cardiovascular cause Other known cause Not specified Composite end pointf
Nonstratified Analysis
Stratified Analysisc
Anticoagulation Therapy Only (n 5 3,728)a
Combination Antithrombotic Therapy (n 5 848)b
HR (95% CI)
P Value
HR (95% CI)
P Value
33 (1.0)
12 (1.5)
1.47 (0.76-2.85)
.25
1.26 (0.63-2.50)
.51
31 (1.0) 22 (0.7) 10 (0.3) 2 (, 0.1)
12 (1.5) 11 (1.4) 1 (0.1) 0
1.57 (0.80-3.05) 2.02 (0.98-4.18) 0.40 (0.05-3.09) …d
.19 .056 .38 …d
1.40 (0.70-2.80) 2.01 (0.94-4.30) 0.29 (0.04-2.30) …d
.34 .07 .24 …d
43 (1.3) 15 (0.5) 5 (0.2)
14 (1.7) 14 (1.7) 2 (0.2)
1.27 (0.70-2.33) 3.76 (1.81-7.79) 1.54 (0.30-7.94)
.44 .0004e .61
0.90 (0.49-1.68) 3.14 (1.45-6.78) 1.34 (0.24-7.61)
.75 .004e .74
364 (11.9)
208 (29.9)
2.52 (2.12-2.99)
, .0001e
2.50 (2.09-2.99)
, .0001e
64 (2.0) 21 (0.6) 10 (0.3) 17 (0.5)
39 (4.9) 9 (1.1) 5 (0.6) 10 (1.2)
2.44 (1.64-3.63) 1.68 (0.77-3.68) 1.96 (0.67-5.73) 2.32 (1.06-5.07)
, .0001e .19 .22 .035
2.26 (1.49-3.43) 1.61 (0.71-3.65) 1.58 (0.51-4.90) 2.23 (0.98-5.08)
.0001e .26 .43 .056
13 (0.4) 0
5 (0.6) 3 (0.4)
1.50 (0.54-4.22) …d
.44 …d
1.46 (0.50-4.28) …d
.49 …d
0 4 (0.1)
1 (0.1) 1 (0.1)
…d 1.04 (0.12-9.30)
…d .97
…d 1.17 (0.12-11.54)
…d .89
37 (1.1)
24 (3.0)
2.59 (1.55-4.34)
.0003e
2.43 (1.42-4.16)
.0012
86 (2.6) 1 (, 0.1) 0
37 (4.5) 0 0
1.68 (1.14-2.48) …d …d
.008e …d …d
1.33 (0.89-1.98) …d …d
.16 …d …d
4 (0.1) 1 (, 0.1) 10 (0.3) 37 (1.1) 29 (0.9) 4 (0.1) 171 (5.4)
2 (0.2) 0 5 (0.6) 12 (1.5) 17 (2.1) 1 (0.1) 86 (11.0)
2.01 (0.37-10.98) …d 1.96 (0.67-5.72) 1.26 (0.66-2.41) 2.33 (1.28-4.24) 0.86 (0.10-7.77) 2.06(1.59-2.66)
.42 …d .22 .49 .006 .90 , .0001e
1.36 (0.24-7.81) …d 1.58 (0.51-4.89) 0.90 (0.46-1.77) 2.09 (1.12-3.91) 0.76 (0.08-6.81) 1.79 (1.37-2.35)
.73 …d .43 .77 .02 .80 , .0001e
Data are presented as No. (% per year), unless otherwise indicated. HR 5 hazard ratio. See Table 1 legend for expansion of other abbreviations. aAll patients randomized to idraparinux or VKA who received no antiplatelets during the randomized treatment period. bAll patients randomized to idraparinux or VKA who received one or more antiplatelet drug at any time after randomization and up to the end of the randomized treatment period or up to the day before a thromboembolic event. cP value after stratification on propensity score in five categories based on quintiles. Propensity score was determined as the probability of receiving antiplatelets given age, sex, race, type of atrial fibrillation, previous ischemic stroke or TIA, left ventricular dysfunction, diabetes mellitus, previous symptomatic coronary artery disease, and previous VKA use. dNot estimatable. eSignificant at P ⱕ .05. fStroke, non-CNS systemic embolism, myocardial infarction, VTE, major bleeding, or death from any cause.
150
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
Original Research
Figure 1. Cumulative incidence of ischemic stroke for all randomized patients receiving anticoagulation therapy only or combination (combin) antithrombotic therapy.
Figure 2. Cumulative incidence of any clinically relevant bleeding for all randomized patients receiving anticoagulation therapy only or combination antithrombotic therapy. See Figure 1 legend for expansion of abbreviation.
score (HR, 3.02; 95% CI, 1.23-7.42; P 5 .024) were associated with major bleeding. In both univariate and multivariate analyses, combination antithrombotic therapy did not predict major bleeding (univariate HR, 1.82; 95% CI, 0.83-4.00; P 5 .14; multivariate HR, 1.61; 95% CI, 0.73-3.56; P 5 .24).
Even among patients randomized to receive VKA therapy (n 5 2,293), combination antithrombotic therapy doubled the risk of clinically relevant bleeding only. These results and those of other studies3,6,7,10 suggest that combination antithrombotic therapy in patients with AF increases the risk of major bleeding. However, the definition of bleeding used in studies and clinical trials varies considerably, thereby limiting direct comparisons of bleeding rates among studies; a universal bleeding definition would overcome this. Bleeding is multifactorial, and the benefit/risk ratio should be carefully considered prior to initiation of VKA or combining potent antithrombotic drugs in any patient. There are some instances, such as following an acute coronary syndrome (ACS) or percutaneous coronary intervention and stent implantation, where short-term concomitant antiplatelet therapy is still warranted.11 Of note, combination antithrombotic therapy was not associated with a reduction in arterial thromboembolic events but, instead, demonstrated an increased risk of myocardial infarction, although the actual number of events was small. However, patients at high cardiovascular risk may have been more likely to be prescribed antiplatelet drugs and, thus, may not benefit from combination therapy. There is also a common perception that antiplatelet drugs work on the platelet-rich thrombus (white clot) associated with atherosclerotic vascular disease, whereas warfarin is effective for the fibrin-rich thrombus (red clot) associated with AF.12 However, two trials (Antithrombotics in the Secondary Prevention of Events in Coronary Thrombosis-2 [ASPECT-2] and WarfarinAspirin Reinfarction Study II [WARIS-II]) comparing warfarin plus aspirin to warfarin only in patients with ACS also demonstrated that combination antithrombotic therapy did not confer a significant reduction in
Mortality A total of 123 (3.0% per year) patients died during follow-up, predominantly from other cardiovascular causes, bleeding, and other known causes. All-cause mortality was higher among patients receiving combination antithrombotic therapy than among those receiving anticoagulant therapy only, but this difference was not significant after stratification by propensity score (HR, 1.33; 95% CI, 0.89-1.98). In analyses restricted to patients randomized to VKA therapy, there was no significant difference in allcause mortality between those receiving VKA alone and those receiving combination antithrombotic therapy (16 [3.8% per year] vs 45 [2.6% per year], respectively; HR, 0.99; 95% CI, 0.55-1.78; P 5 .97 stratified by propensity score). Discussion The present post hoc analysis of the AMADEUS cohort demonstrates that age ⱖ 65 years and combination antithrombotic therapy were predictive of an increased risk of both clinically relevant and major bleeding among all randomized patients. Specifically, concomitant antiplatelet therapy in addition to anticoagulation (either idraparinux or VKA) independently predicted a 2.3- to 2.5-fold increased risk of major bleeding or clinically relevant bleeding, respectively, compared with anticoagulation therapy only. www.chestpubs.org
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
CHEST / 140 / 1 / JULY, 2011
151
Table 3—Univariate Predictors of Clinically Relevant Bleeding and Major Bleeding in All Patients Clinically Relevant Bleeding Patient Characteristic Age group , 65 ya 65-74 y ⱖ 75 y Sex Malea Female Ethnicity Whitea Nonwhite BMI, kg/m2 , 25a ⱖ 25 Baseline creatinine clearance level, mL/min/1.73 m2 , 50 ⱖ 50a Type of AF Paroxysmal Persistent Permanenta Previous stroke Yes Noa Previous TIA Yes Noa Previous non-CNS embolism Yes Noa Hypertension Yes Noa Left ventricular dysfunction Yes Noa Diabetes mellitus Yes Noa Symptomatic coronary artery disease Yes Noa CHADS2 risk category Lowa Moderate High Previous VKA (baseline) Previous VKA usea VKA naive Antithrombotic therapy Combination antithrombotic therapy Anticoagulation onlya
No.
No. Events (% Per Year)
HR (95% CI)
Major Bleeding P Value , .0001
No. Events (% Per Year)
HR (95% CI)
9 (0.8) 35 (1.9) 59 (3.7)
1.00 2.38 (1.14-4.94) 4.72 (2.34-9.52)
P Value , .0001b
b
1,147 1,818 1,611
101 (8.8) 237 (13.0) 234 (14.5)
1.00 1.47 (1.16-1.85) 1.70 (1.35-2.15)
3,043 1,533
365 (12.0) 207 (13.5)
1.00 1.12 (0.95-1.33)
.19 …
74 (2.4) 29 (1.9)
1.00 0.77 (0.50-1.18)
4,486 90
557 (12.4) 15 (16.7)
1.00 1.41 (0.84-2.35)
.19 …
102 (2.3) 1 (1.1)
1.00 0.50 (0.07-3.57)
.22 … … .49 …
1,050 3,526
151 (14.4) 421 (11.9)
1.00 0.81 (0.67-0.98)
… .026b
33 (3.1) 70 (2.0)
1.00 0.62 (0.41-0.94)
… .025b
819 3,757
122 (14.9) 450 (12.0)
1.30 (1.06-1.59) 1.00
.011b …
29 (3.5) 74 (2.0)
1.85 (1.20-2.84) 1.00
.005b …
1,634 438 2,504
207 (12.7) 53 (12.1) 312 (12.5)
1.09 (0.91-1.30) 1.04 (0.78-1.39) 1.00
.65 … …
32 (2.0) 10 (2.3) 61 (2.4)
0.86 (0.56-1.32) 0.99 (0.51-1.94) 1.00
.78 … …
580 3,996
80 (13.8) 492 (12.3)
1.15 (0.91-1.45) 1.00
.25 …
15 (2.6) 88 (2.2)
1.19 (0.69-2.06) 1.00
.53 …
478 4,098
63 (13.2) 509 (12.4)
1.07 (0.82-1.39) 1.00
.61 …
9 (1.9) 94 (2.3)
0.83 (0.42-1.65) 1.00
.60 …
156 4,420
19 (12.2) 553 (12.5)
1.00 (0.64-1.59) 1.00
.99 …
2 (1.3) 101 (2.3)
0.58 (0.14-2.35) 1.00
.44 …
3,530 1,046
435 (12.3) 137 (13.1)
0.99 (0.82-1.20) 1.00
.90 …
81 (2.3) 22 (2.1)
1.16 (0.72-1.86) 1.00
.54 …
1,072 3,504
140 (13.1) 432 (12.3)
1.03 (0.85-1.25) 1.00
.76 …
28 (2.6) 75 (2.1)
1.18 (0.76-1.82) 1.00
.46 …
897 3,679
110 (12.3) 462 (12.6)
0.97 (0.79-1.20) 1.00
.81 …
24 (2.7) 79 (2.1)
1.26 (0.80-1.98) 1.00
.33 …
1,408 3,168
202 (14.3) 370 (11.7)
1.21 (1.02-1.43) 1.00
.03b …
35 (2.5) 68 (2.1)
1.14 (0.76-1.71) 1.00
.54 …
1,880 1,448 1,248
222 (11.8) 174 (12.0) 176 (14.1)
1.00 1.02 (0.84-1.25) 1.23 (1.01-1.50)
.093b … …
34 (1.8) 26 (1.8) 43 (3.4)
1.00 1.00 (0.60-1.66) 1.95 (1.24-3.05)
.0037b … …
3,476 1,100
414 (11.9) 158 (14.4)
1.00 1.29 (1.07-1.55)
.007b …
72 (2.1) 31 (2.8)
1.00 1.45 (0.95-2.20)
.087 …
848
208 (24.5)
2.52 (2.12-2.99)
, .0001b
39 (4.6)
2.44 (1.64-3.63)
, .0001b
364 (9.8)
1.00
64 (1.7)
1.00
3,728
…
…
See Table 1 and 2 legends for expansion of abbreviations. aReference category. bSignificant at P ⱕ .05. 152
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
Original Research
Table 4—Multivariate Predictors of Clinically Relevant Bleeding and Major Bleeding in All Patients Clinically Relevant Bleeding Patient Characteristic Antithrombotic therapy Combination antithrombotic therapy Anticoagulation alonea Age group , 65 ya 65-74 y ⱖ 75 y BMI, kg/m2 . 25a ⱖ 25 Baseline creatinine clearance, mL/min/1.73 mm2 , 50 ⱖ 50a Previous VKA (baseline) Previous VKA usea VKA naive Symptomatic CAD Yes Noa CHADS2 risk category Lowa Moderate High
Major Bleeding
No. Events (% Per Year)
HR (95% CI)
P Value
No. Events (% Per Year)
HR (95% CI)
P Value
208 (24.5)
2.47 (2.07-2.96)
, .0001b
39 (4.6)
2.23 (1.49-3.34)
, .0001b
3,728
364 (9.8)
1.00
…
64 (1.7)
1.00
1,147 1,818 1,611
101 (8.8) 237 (13.0) 234 (14.5)
1.00 1.44 (1.14-1.82) 1.59 (1.24-2.04)
.001b … …
9 (0.8) 35 (1.9) 59 (3.7)
1.00 2.26 (1.08-4.71) 4.19 (1.98-8.87)
.0004b … …
1,050 3,526
151 (14.4) 421 (11.9)
1.00 0.89 (0.73-1.08)
.23 …
33 (3.1) 70 (2.0)
1.00 0.81 (0.52-1.25)
.34 …
819 3,757
122 (14.9) 450 (12.0)
1.00 (0.80-1.26) 1.00
.99 …
29 (3.5) 74 (2.0)
1.03 (0.64-1.67) 1.00
.89 …
3,476 1,100
414 (11.9) 158 (14.4)
1.00 1.09 (0.90-1.31)
.38 …
… …
… …
… …
1,408 3,168
202 (14.3) 370 (11.7)
0.96 (0.80-1.15) 1.00
.64 …
… …
… …
… …
1,880 1,448 1,248
… … …
… … …
… … …
34 (1.8) 26 (1.8) 43 (3.4)
1.00 0.69 (0.40-1.19) 1.18 (0.72-1.95)
.10 … …
No. 848
…
Only statistically significant variables in the univariate analysis (P ⱕ .05) were included in the multivariate analysis. See Table 1 and 2 legends for expansion of abbreviations. aReference category. bSignificant at P ⱕ .05.
thromboembolic and mortality risk compared with warfarin only, although there was a trend for a reduction in such events.13,14 In addition, the trend toward more ischemic strokes among patients receiving combination antithrombotic therapy may have been due
Figure 3. Cumulative incidence of major bleeding for all randomized patients receiving anticoagulation therapy only or combination antithrombotic therapy. See Figure 1 legend for expansion of abbreviation.
to a higher intrinsic risk of stroke (older, previous stroke/TIA, diabetes mellitus, left ventricular dysfunction, and symptomatic CAD), hence, the addition of antiplatelet therapy among these patients, when compared with those receiving oral anticoagulation alone (and this finding is despite the propensity score adjustments). A post hoc analysis of the combined Stroke Prevention Using Oral Thrombin Inhibitor in Atrial Fibrillation (SPORTIF) III and V data sets also revealed that combination antithrombotic therapy (warfarin or ximelagatran plus aspirin) significantly increased the risk of bleeding (major and minor) compared with treatment with either anticoagulant only.6 Combination therapy was not associated with a reduction in stroke or vascular events (death or myocardial infarction),6 but a 1.58-fold (95% CI, 1.01-2.49) increase in clinically relevant bleeding compared with treatment with warfarin alone was apparent, findings in keeping with the current study. In contrast to the SPORTIF cohort,6 the present study found that combination antithrombotic therapy did not independently increase the risk of major bleeding perhaps because of the smaller sample size in the VKA subgroup with fewer major
www.chestpubs.org
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
CHEST / 140 / 1 / JULY, 2011
153
bleeding events. Further, although halted early due to poor recruitment, the Fluindione, Fibrillation Auriculaire, Aspirin et Contraste Spontané (FFAACS) study, which compared fluindione (INR range, 2.0-2.6) plus placebo (anticoagulation alone) with fluindione (same INR range) plus aspirin (combination therapy), also demonstrated a higher bleeding event rate in the combination arm.4 Conversely, the National Study for Prevention of Embolism in Atrial Fibrillation (NASPEAF),10 which compared combination acenocoumarol (INR range, 2.0-3.0) and the antiplatelet agent triflusal, 600 mg/d, with either acenocoumarol or triflusal alone in patients at intermediate or high risk of thromboembolism, demonstrated that combination therapy was associated with an approximately 60% relative risk reduction in the composite outcome of vascular events and severe bleeding compared with either acenocoumarol or triflusal alone. A systematic review and meta-analysis of randomized controlled trials of combination antithrombotic therapy (oral anticoagulation plus aspirin) vs oral anticoagulation therapy only3 in patients at risk for vascular outcomes (including two trials in patients with AF: FFAACS4 and Second Copenhagen Atrial Fibrillation, Aspirin and Anticoagulant Therapy Study [AFASAK-2]5) also demonstrated that combination antithrombotic therapy did not reduce the risk of arterial thromboembolism or all-cause mortality but did significantly increase the risk of major bleeding. In the present analyses, patients receiving combination antithrombotic therapy were significantly less likely to have been prescribed a VKA prior to enrollment in the AMADEUS trial. This finding may have had an effect on the incidence of bleeding because patients with previous VKA experience (who by definition have not experienced significant problems with VKA, ie, “oral anticoagulation survivors”) may be less likely to experience problems with subsequent anticoagulation. Inception studies of oral anticoagulation therapy have demonstrated that the risk of bleeding is greatest during the first 90 days of VKA therapy.15 Therefore, those patients who experienced clinically relevant bleeding or major bleeding did so because they were “VKA naive” and because of the use of combination antithrombotic therapy. Indeed, nonuse of VKAs prior to randomization was a univariate predictor of clinically relevant bleeding in the present analyses. A retrospective analysis of VKAnaive patients commencing either acenocoumarol or phenprocoumon demonstrated that any concomitant antiplatelet therapy increased the risk of major bleeding, with the greatest risk of bleeding evident with concomitant clopidogrel use followed by aspirin, even after adjustment for use of nonsteroidal antiinflammatory drugs, antibiotics, steroids, and gastroprotective agents.16 Further, a recent Danish
registry of 85,541 patients experiencing first-time hospitalization for AF, with at least one prescription for warfarin, aspirin, or clopidogrel after discharge, demonstrated that the combined end point of hospitalization or death from bleeding or anemia was highest among those receiving dual (warfarin and clopidogrel) or triple therapy (15.5% and 17.5% per patient-year, respectively).7 The same investigators also reported a higher risk of bleeding in patients with acute myocardial infarction treated with different combinations of warfarin, aspirin, or clopidogrel.17 Nonetheless, these studies7,17 were based on a large prescriptions registry where treatments were nonrandomized and many clinical parameters of the patients are unknown; in contrast, the present analysis is based on a prospective trial setting where patients were carefully followed up. Limitations The present study has several limitations. First, the AMADEUS study was not originally powered to undertake these post hoc analyses, and analyses were run on parameters not taken into account in the randomization process. In clinical practice, many patients with AF are given combination warfarin and antiplatelet therapy,1 and patients with AF and an ACS, coronary stenting, or both often are prescribed dual or triple antithrombotic therapy, despite the current lack of evidence base to support this. Therefore, post hoc analyses of antithrombotic therapy in AF trial cohorts are currently the only way to try to answer the question of the effect of combination antithrombotic therapy in patients with AF. Second, the use of antiplatelet therapy was not randomized in the AMADEUS trial,8 and as we alluded to earlier, it is possible that significant differences between the groups in baseline demographic and clinical characteristics may confound the present findings. Although the use of propensity score and multivariate analyses can reduce the potential bias, some other factors not recorded in the AMADEUS study may affect the results of the analyses. For example, a history of bleeding, one of the most important predictors of recurrent bleeding, was not recorded in the AMADEUS cohort, and, therefore, it was not possible to examine the effect of this factor on bleeding risk in our multivariate analyses. However, patients with previous significant bleeding often are excluded from clinical trials of thromboprophylaxis, although previous major bleeding was not explicitly listed as an exclusion criterion in the AMADEUS cohort.8 There also exists the possibility of residual confounding factors behind a significant difference between the groups (eg, local practice, access to VKA, medical history).
154
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
Original Research
Conclusions Combination antithrombotic therapy increases the risk of clinically relevant bleeding and major bleeding in patients with AF and does not appear to reduce the risk of stroke. Therefore, combination antithrombotic treatment is justified only in patients with a clear indication that the benefit of adding antiplatelet therapy to anticoagulant treatment outweighs the increased risk of bleeding.
5.
6.
Acknowledgments Author contributions: Dr Minini had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr Lane: contributed design of the analyses, had full access to the results of the data analysis, and contributed to the preparation and revision of the manuscript. Dr Kamphuisen: contributed design of the analyses and the preparation and revision of the manuscript. Dr Minini: contributed preparation of the manuscript. Dr Büller: contributed design of the analyses and the preparation and revision of the manuscript. Dr Lip: contributed design of the analyses and preparation and revision of the manuscript. Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Lane is in receipt of an investigator-initiated educational grant from Bayer and has received industry-funded sponsorship for travel to educational meetings from AstraZeneca and Sanofi-Aventis. Dr Kamphuisen has received an independent research grant from the Ruitinga van Swieten Foundation, The Netherlands. He is a consultant for Bayer, Boehringer-Ingelheim, CSL Behring, and Ablynx NV and has received investigator-initiated research grants from Bayer, Leo Pharma Inc, Wyeth, and CSL Behring. Dr Minini is an employee of Sanofi-Aventis. Dr Büller has served as a consultant to Sanofi-Aventis, Bayer, Pfizer, GlaxoSmithKline, Astellas Pharma, Boehringer-Ingelheim, and Daiichi-Sankyo. Dr Lip has served as a consultant for Bayer, Astellas, Merck, AstraZeneca, SanofiAventis, Bristol-Myers Squibb/Pfizer, and Boehringher Ingelheim and has been on the speakers bureau for Bayer, Sanofi-Aventis, Bristol-Myers Squibb/Pfizer, and Boehringher Ingelheim. Role of sponsors: Sanofi-Aventis provided access to the AMADEUS data set for these analyses to be undertaken. Other contributions: We thank the AMADEUS Steering Committee for its comments and support of this subanalysis of the AMADEUS cohort.
7.
8.
9. 10.
11.
12. 13.
References 1. Shireman TI, Howard PA, Kresowik TF, Ellerbeck EF. Combined anticoagulant-antiplatelet use and major bleeding events in elderly atrial fibrillation patients. Stroke. 2004; 35(10):2362-2367. 2. Johnson SG, Witt DM, Eddy TR, Delate T. Warfarin and antiplatelet combination use among commercially insured patients enrolled in an anticoagulation management service. Chest. 2007;131(5):1500-1507. 3. Dentali F, Douketis JD, Lim W, Crowther M. Combined aspirin-oral anticoagulant therapy compared with oral anticoagulant therapy alone among patients at risk for cardiovascular disease: a meta-analysis of randomized trials. Arch Intern Med. 2007;167(2):117-124. 4. Lechat P, Lardoux H, Mallet A, et al; FFAACS (Fluindione, Fibrillation Auriculaire, Aspirin et Contraste Spontané)
14. 15.
16. 17.
Investigators. Anticoagulant (fluindione)-aspirin combination in patients with high-risk atrial fibrillation. A randomized trial (Fluindione, Fibrillation Auriculaire, Aspirin et Contraste Spontané; FFAACS). Cerebrovasc Dis. 2001;12(3): 245-252. Gulløv AL, Koefoed BG, Petersen P. Bleeding during warfarin and aspirin therapy in patients with atrial fibrillation: the AFASAK 2 study. Atrial Fibrillation Aspirin and Anticoagulation. Arch Intern Med. 1999;159(12):1322-1328. Flaker GC, Gruber M, Connolly SJ, et al; SPORTIF Investigators. Risks and benefits of combining aspirin with anticoagulant therapy in patients with atrial fibrillation: an exploratory analysis of stroke prevention using an oral thrombin inhibitor in atrial fibrillation (SPORTIF) trials. Am Heart J. 2006;152(5):967-973. Hansen ML, Sørensen R, Clausen MT, et al. Risk of bleeding with single, dual, or triple therapy with warfarin, aspirin, and clopidogrel in patients with atrial fibrillation. Arch Intern Med. 2010;170(16):1433-1441. Bousser MG, Bouthier J, Büller HR, et al; Amadeus Investigators. Comparison of idraparinux with vitamin K antagonists for prevention of thromboembolism in patients with atrial fibrillation: a randomised, open-label, non-inferiority trial [published correction appears in Lancet. 2008;372(9655): 2022]. Lancet. 2008;371(9609):315-321. D’Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med. 1998;17(19):2265-2281. Pérez-Gómez F, Alegría E, Berjón J, et al; NASPEAF Investigators. Comparative effects of antiplatelet, anticoagulant, or combined therapy in patients with valvular and nonvalvular atrial fibrillation: a randomized multicenter study. J Am Coll Cardiol. 2004;44(8):1557-1566. Lip GY, Huber K, Andreotti F, et al; European Society of Cardiology Working Group on Thrombosis. Management of antithrombotic therapy in atrial fibrillation patients presenting with acute coronary syndrome and/or undergoing percutaneous coronary intervention/stenting [published correction appears in Thromb Haemost. 2010;104(3):653]. Thromb Haemost. 2010;103(1):13-28. Lip GYH. Don’t add aspirin for associated stable vascular disease in a patient with atrial fibrillation receiving anticoagulation. BMJ. 2008;336(7644):614-615. van Es RF, Jonker JJ, Verheugt FW, Deckers JW, Grobbee DE; Antithrombotics in the Secondary Prevention of Events in Coronary Thrombosis-2 (ASPECT-2) Research Group. Aspirin and Coumadin after acute coronary syndromes (the ASPECT-2 study): a randomised controlled trial. Lancet. 2002;360(9327):109-113. Hurlen M, Abdelnoor M, Smith P, Erikssen J, Arnesen H. Warfarin, aspirin, or both after myocardial infarction. N Engl J Med. 2002;347(13):969-974. Hylek EM, Evans-Molina C, Shea C, Henault LE, Regan S. Major hemorrhage and tolerability of warfarin in the first year of therapy among elderly patients with atrial fibrillation. Circulation. 2007;115(21):2689-2696. Schalekamp T, Klungel OH, Souverein PC, de Boer A. Effect of oral antiplatelet agents on major bleeding in users of coumarins. Thromb Haemost. 2008;100(6):1076-1083. Sørensen R, Hansen ML, Abildstrom SZ, et al. Risk of bleeding in patients with acute myocardial infarction treated with different combinations of aspirin, clopidogrel, and vitamin K antagonists in Denmark: a retrospective analysis of nationwide registry data. Lancet. 2009;374(9706):1967-1974.
www.chestpubs.org
Downloaded From: http://journal.publications.chestnet.org/ by a Francis A Countway User on 06/28/2013
CHEST / 140 / 1 / JULY, 2011
155