Atrial fibrillation (AF) is a kind of arrhythmia, and there are about 8 million patients with AF in China, 2.2 million in the United States and 4.5 million in Europe Union.1 It is known that AF is one of the independent risk factors for stroke and might increase its risk by up to 5-fold. The AF-related strokes are associated with higher risk of mortality and morbidity than non-AF strokes.2 For a half-century, clinicians have prescribed aspirin or vitamin K antagonist (most commonly warfarin) for the patients with AF. Warfarin is always a preferred option to prevent stroke or systemic embolism events.3,4 Although with the proven efficacy, its use is limited by some drawbacks including the narrow therapeutic window requiring frequent international normalized ratio (INR) monitoring and a high risk of bleeding.5,6
Therefore, as a result, some new oral anticoagulants (NOACs) inhibiting either activated factor X (factor Xa) or thrombin have been developed to provide a more promising option.7,8 The 4 large phase III randomized controlled trials (RCTs) RE-LY (Dabigatran randomized evaluation of long-term anticoagulant therapy), ROCKET-AF (Rivaroxaban once daily oral direct factor Xa inhibition compared with vitamin K antagonism for prevention of stroke and embolism trial in atrial fibrillation), ARISTOTLE (apixaban for reduction in stroke and other thromboembolic events in atrial fibrillation), and ENGAGE AF-TIMI 48 (Edoxaban once daily to prevent stroke or systemic embolism)9 have separately examined the long-term effect of NOACs compared with warfarin to prevent stroke and systemic embolism in patients with AF. Although these NOACs have been proved more efficacious than warfarin for the primary efficacy endpoint of stroke and systemic embolism, the conclusion for the secondary efficacy endpoints and safety endpoints are heterogeneous.10,11 Therefore, it is important to comprehensively compare efficacy and safety of NOACs versus warfarin in the patients with AF.
Several meta-analyses for NOACs versus warfarin12,13 have been published, but these publications included only 3 phase 3 studies: RE-LY, ROCKET-AF, and ARISTOTLE. In this article, except the above 3 studies, we added 22,385 patients by including J-ROCKET AF (Rivaroxaban Clinical Trial in Japan),14 and ENGAGE AF-TIMI 48 published recently. This allowed us to perform a more comprehensive comparative analysis of NOACs versus warfarin. Moreover, different from previous analyses, we performed separate meta-analyses for the high-dose groups of RE-LY (150 mg twice daily) and ENGAGE AF-TIMI 48 (60 mg twice daily) combined with the single dose studies ARISTOTLE, ROCKET-AF, and J-ROCKET AF and the low-dose groups of RE-LY (110 mg twice daily) and ENGAGE AF-TIMI 48 (30 mg twice daily), respectively. This will not merge the benefit and risk of different doses.
Although with a reduced INR target level of 1.6–2.6 in J-ROCKET AF than the regular therapeutic range 2.0–3.0,15 the design was similar and the results were consistent with those of ROCKET AF for the primary efficacy endpoint, principal safety outcome, and major bleeding. Therefore, to include J-ROCKET AF in the meta-analysis would not introduce bias.
We systematically searched the publications of RCTs comparing NOACs to warfarin in patients with AF from the Cochrane Library, Embase, MEDLINE, Science Citation Index Expanded, and ProQuest in December 2013. The keywords or medical terms included “new oral anticoagulants,” “oral thrombin inhibitors,” “oral factor Xa inhibitors,” “DABIGATRAN,” “RIVAROXABAN,” “APIXABAN,” “EDOXABAN,” “BETRIXABAN,” “YM-150,” “RE-LY,” “LY-517717,” “ENGAGE AF-TIMI 48,” and “WARFARIN.” We only searched clinical trials from the Embase and MEDLINE. The Science Citation Index Expanded and ProQuest searches were limited to the reports with full-text available. In addition, we manually searched the clinical databases such as the website of ClinicalTrials, related review, and reports for further eligible studies. Two authors selected the studies independently, and the disagreements were resolved by discussion among all the authors.
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement for reporting systematic reviews and meta-analyses of RCTs was used for this analysis.16 The criteria for studies included in our analysis were as following: (1) they were RCTs between NOACs and warfarin, (2) all the patients were randomized to warfarin or NOACs, and (3) the target population was the patients with AF. Except double-blinded, the open-label studies were also included in the search because of the need of frequent INR monitoring for warfarin. The data extracted from these studies included patients' age, median follow-up time, mean CHADS2 [congestive heart failure, hypertension, age ≥75 years, biabetes mellitus, stroke (doubled)] scores,17 gender, mean time in the therapeutic range of warfarin, and some specific medical history. For all the included studies, the primary efficacy endpoint was composite of stroke and systemic embolism. The secondary efficacy endpoints included ischemic stroke, hemorrhagic stroke, all-cause mortality, and myocardial infraction. For safety evaluation, the main endpoint was major bleeding defined as fatal bleeding or bleeding in a critical site, and the secondary endpoint included gastrointestinal and intracranial bleeding. We only considered the studies approved or in development so the studies for ximelagatran, which had been withdrawn,18 and studies for darexaban, which is no longer in development19 were excluded from our analysis.
The Cochrane Collaboration's tool was used to conduct quality assessment to risk of bias.20 It evaluated bias in a RCT within the following domains: sequence generation; allocation concealment; blinding of participants, personal, and outcome assessors; incomplete outcome data; selective outcome reporting; and other potential threats to validity. In each domain, the risk of bias was classified into high, low, or unclear for each RCT.
The events including primary efficacy endpoint (stroke and systemic embolism), ischemic stroke, hemorrhagic stroke, all-cause mortality, and myocardial infarction were used for efficacy analysis, and those including major bleeding, intracranial hemorrhage, and gastrointestinal bleeding were used for safety. The analysis populations were intention-to-treat for efficacy analysis and safety population for bleeding endpoints analysis. In RE-LY and ENGAGE AF-TIMI 48, there were 2 NOACs doses regimens compared with warfarin, respectively. Instead of combining both doses to 1 meta-analysis, which might merge the benefit and risk of different doses, the high-dose groups of RE-LY (150 mg twice daily) and ENGAGE AF-TIMI 48 (60 mg twice daily) were combined with the single dose studies ARISTOTLE, ROCKET-AF, and J-ROCKET. A separate meta-analysis was done for the low-dose groups of RE-LY (110 mg twice daily) and ENGAGE AF-TIMI 48 (30 mg twice daily).
Based on the random-effects model by DerSimonian and Laird,21 we calculated the pooled relative risks (RRs) and their corresponding 95% confidence intervals (CIs). The presence of between-study variability was assessed by the Q statistic (P < 0.10 was used as indicator of statistically significant result), and the proportion of heterogeneity was assessed by the I2 index. All the analyses were conducted in statistical software Stata 11.0 (StataCorp LP, College Station, TX).
Eight hundred nine publications were identified though database search (Fig. 1). By study selection process, the 5 studies from 56 publications (ARISTOTLE, ENGAGE AF-TIMI 48, RE-LY, ROCKET-AF, and J-ROCKET) assessing the efficacy and safety of NOACs compared with warfarin in patients with AF were evaluated for eligibility. The primary objective was to determine whether the new drug was noninferior to warfarin about the composite of stroke and systemic embolism.
Forty-three thousand fifty patients received NOACs and 29,911 received warfarin. Baseline characteristics of the populations are listed in Table 1. The average age and proportion of female were similar between NOACs and warfarin groups among the 5 studies. The underlying risk for stroke indicated by the proportion of patients with CHADS2 was significantly different across studies. It can be observed the studies ENGAGE AF-TIMI 48, ROCKET-AF, and J-ROCKET enrolled more patients with higher risk of stroke than ARISTOTLE and RE-LY. The median follow-up time in all studies ranged from 1.6 to 2.8 years, and the median time in therapeutic range of patients assigned to warfarin ranged from 44% to 68%.
The risk of bias was assessed by Cochrane Collaboration's tool. In RE-LY, dabigatran was administered in a blinded fashion, whereas warfarin was administered in an unblinded fashion to locally adjust INR within the therapeutic range 2.0–3.0. However, all the investigators, coordinating center members, the steering committee, and the sponsor were kept blinded during event ascertainment and analyses process.22 Therefore, the risk of bias due to blind was low for RE-LY. For ROCKET-AF, the efficacy was analyzed based on intention-to-treat and bleeding on safety analysis data sets (1 site including 93 patients was excluded due to good clinical practice violation).10 This should not introduce additional selection bias.
For the high-dose regimen analysis, NOACs demonstrated noninferior to warfarin in the prevention of stroke and systemic embolism in each study. NOACs further demonstrated superior to warfarin at the above endpoint in ARISTOTLE and RE-LY (Fig. 2). In respect to the major bleeding prevention, NOACs showed superiority to warfarin in ARISTOTLE and ENGAGE AF-TIMI 48, whereas other studies showed comparable risk of major bleeding (Fig. 3). The pooled risk of stroke and systemic embolism in the patients randomized to NOACs was 20% lower (RR = 0.80; 95% CI, 0.71–0.91) than those randomized to warfarin. This benefit was mostly driven by the large reduction of hemorrhagic stroke (RR = 0.50; 95% CI, 0.41–0.62) and the reduction of all-cause mortality (RR = 0.90; 95% CI, 0.85–0.95). For safety, the pooled risk of major bleeding events in the patients randomized to NOACs was reduced by 14% (RR = 0.86; 95% CI, 0.74–0.99) compared with the risk of those on warfarin because of the large reduction of intracranial hemorrhage (RR = 0.48; 95% CI, 0.41–0.56) (Fig. 4).
For the low-dose regimen meta-analysis, NOACs demonstrated similar efficacy to warfarin for prevention of stroke and systemic embolism in each study. The similar conclusion was drawn for the pooled group (RR = 1.03; 95% CI, 0.84–1.27). If differentiated by stroke types, the large reduction in the risk of hemorrhagic stroke (RR = 0.33; 95% CI, 0.23–0.46) was offset by the increase in ischemic stroke (RR = 1.31; 95% CI, 1.14–1.49). For low-dose regimens, there was also a reduction in the risk of all-cause mortality (RR = 0.89; 95% CI, 0.83–0.96) but a higher risk of myocardial infarction (RR = 1.25; 95% CI, 1.04–1.50). For safety, although with the significant difference for the risk of major bleeding events in each study, the pooled RR of NOACs compared with warfarin (RR = 0.63; 95% CI, 0.38–1.04) was inconclusive. When differentiated by bleeding types, there was a large risk of decrease in intracranial hemorrhage (RR = 0.31; 95% CI, 0.24–0.41) and also a risk of decrease in gastrointestinal bleeding (RR = 0.85; 95% CI, 0.72–1.00) (Fig. 5).
From the combined results of the high-dose and low-dose regimens, the inclusion of low-dose diminished the magnitude of risk reduction in stroke and systemic embolism (RR = 0.86; 95% CI, 0.75–0.99) by NOACs but resulting in lower risk of major bleeding events (RR = 0.78; 95% CI, 0.64–0.94).
Based on the design of each study, in this article, we considered meta-analyses for high-dose regimen, low-dose regimen, and their combination. Five phase III RCTs including ARISTOTLE, ENGAGE AF-TIMI 48, RE-LY, ROCKET-AF, and J-ROCKET were included in our analysis. It was found that randomization to NOACs reduced the risk of stroke and systemic embolism compared with warfarin. This benefit was mainly driven by the substantial reduction of hemorrhagic stroke. Although as a part of the efficacy assessment of NOACs, hemorrhagic stroke is also a complication of anticoagulant treatment.23 The rough half reduction in risk of hemorrhagic stroke by NOACs indicated the benefit of the treatment. NOACs were also found to be associated with lower risk of hemorrhagic stroke and all-cause mortality compared with warfarin. A lower risk in ischemic stroke (RR = 0.87; 95% CI, 0.77–0.99) between NOACs and warfarin was reported by Miller et al,12 which was different from our inconclusive result. However, for the prevention of ischemic stroke, the NOAC is effective due to the reduction of the risk by two thirds compared with placebo.5 The pooled risk of major bleeding events was similar between NOACs and warfarin. The combined results of efficacy and safety support use of the NOACs as alternatives to warfarin for long-term prevention in the patients with AF.
The separate meta-analyses of the high-dose and low-dose regimens showed that the high-dose regimen has better performance than low dose in efficacy. Although with similar risks of stroke and systemic embolism and major bleeding events, low-dose regimen was found to significantly reduce the risk of hemorrhagic stroke, all-cause mortality, and intracranial hemorrhage. Consequently, the low-dose regimen could be an appealing option for the patients with high risk of bleeding with full-dose anticoagulation therapy.
Warfarin has been underused due to concerns over the need of frequent INR monitoring and the risk of bleeding.24 Only about a half of patients with AF received warfarin. Furthermore, the patients receiving it might have 30%–50% of time not within the therapeutic range 2.0–3.0. Therefore, the development of NOACs seems necessary. Moreover, there are already 2 compounds dabigatran and rivaroxaban approved by the US Food and Drug Administration.25,26 It is necessary to compare them with warfarin on efficacy and safety by meta-analysis so that the informed clinical decisions can be made.
From the meta-analysis, we observed that the estimates are inconsistent across all the studies. We believe that these large phase III studies have enough power to evaluate the primary efficacy but may not have sufficient power to compare the secondary endpoints. Especially for the inconclusive results from individual study, they cannot always show us a real conclusion. One example is: the conclusive result for all-cause mortality is only benefited by apixaban and low-dose edoxaban. This also indicates the ability of meta-analysis in the assessment of the relative benefits of NOACs compared with warfarin. Another example is: different studies enrolled different proportions of patients with CHADS2 scores. It would have been difficult to provide an overall description for NOACs compared with warfarin on their efficacy and safety without performing a meta-analysis.
Actually, by working mechanisms, rivaroxaban, apixaban, and edoxaban should be factor Xa inhibitors, whereas dabigatran is the thrombin inhibitor. A sensitivity analysis was done by excluding dabigatran from the high-dose regimen. The risk of stroke and systemic embolism (RR = 0.85; 95% CI, 0.77–0.93) was similar but the risk of major bleeding (RR = 0.84; 95% CI, 0.69–1.01) became inconclusive. However, the 5 studies could be pooled together for below reasons: they are all phase 3 warfarin-controlled trials with similar designs, and they are NOACs by specific inhibitors of important factors in the coagulation cascade, and previous meta-analyses took a similar approach.27
There were 2 potential limitations for our meta-analyses: First, the patients in clinical trials were often at lower risk of adverse events than those seen in routine clinical practice. This might affect the generalizability of the results based on the clinical trials.28 Second, the patients taking warfarin in routine clinical practice often had less time of INR well-controlled in the therapeutic range.29 This variability could not be applied to the NOACs in the meta-analysis.
In general, our meta-analyses have shown the balance between safety and efficacy of the NOACs compared with warfarin. NOACs demonstrated promising alternatives to warfarin in prevention of stroke in patients with AF.
1. Fuster V, Rydén LE, Cannom DS, et al.. 2011 ACCF/AHA/HRS focused updates incorporated into the ACC/AHA/ESC 2006 guidelines for the management of patients with atrial fibrillation: a report of the American College of Cardiology Foundation/American Heart Association Task Force on practice guidelines. J Am Coll Cardiol. 2011;57:e101–e198.
2. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke. 1991;22:983–988.
3. Camm AJ, Kirchhof P, Lip GY, et al.. Guidelines for the management of atrial fibrillation: the task force for the management of atrial fibrillation of the European Society of Cardiology (ESC). Eur Heart J. 2010;31:2369–2429.
4. Piccini JP, Lopes RD, Mahaffey KW. Oral factor Xa inhibitors for the prevention of stroke in atrial fibrillation. Curr Opin Cardiol. 2010;25:312–320.
5. Hart RG, Pearce LA, Aguilar MI. Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med. 2007;146:857–867.
6. Lam YY, Ma TK, Yan BP. Alternatives to chronic warfarin therapy for the prevention of stroke in patients with atrial fibrillation. Int J Cardiol. 2011;150:4–11.
7. Pisters R, Lane DA, Nieuwlaat R, et al.. A novel user- friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest. 2010;138:1093–1100.
8. Connolly SJ, Ezekowitz MD, Yusuf S, et al.. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009;361:1139–1151.
9. Giugliano RP, Ruff CT, Braunwald E, et al.. Edoxaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2013;369:2093–2104.
10. Patel MR, Mahaffey KW, Garg J, et al.. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365:883–891.
11. Granger CB, Alexander JH, McMurray JJ, et al.. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365:981–992.
12. Miller CS, Grandi SM, Shimony A, et al.. Meta-analysis of efficacy and safety of new oral anticoagulants (dabigatran, rivaroxaban, apixaban) versus warfarin in patients with atrial fibrillation. Am J Cardiol. 2012;110:453–460.
13. Capodanno D, Capranzano P, Giacchi G, et al.. Novel oral anticoagulants versus warfarin in non-valvular atrial fibrillation: a meta-analysis of 50,578 patients. Int J Cardiol. 2013;167:1237–1241.
14. Hori M, Matsumoto M, Tanahashi N, et al.; J-ROCKET AF study investigators. Rivaroxaban vs. warfarin in Japanese patients with atrial fibrillation. Circ J. 2012;76:2104–2111.
15. Van Walraven C, Jennings A, Oake N, et al.. Effect of study setting on anticoagulation control: a systemic review and meta regression. Chest. 2006;129:1155–1166.
16. Liberati A, Altman DG, Tetzlaff J, et al.. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med. 2009;6:e1000100.
17. Gage BF, Waterman AD, Shannon W, et al.. Validation of clinical classification schemes for predicting stroke: results from the National registry of atrial fibrillation. JAMA. 2001;285:2864–2870.
18. Keisu M, Andersson TB. Drug-induced liver injury in humans: the case of ximelagatran. Handb Exp Pharmacol. 2010;196:407–418.
19. Apostolakis S, Lip GY. Novel oral anticoagulants: focus on the direct factor Xa inhibitor darexaban. Expert Opin Investig Drugs. 2012;21:1057–1064.
20. Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0. The Cochrane Collaboration; 2011. http://www.cochrane-handbook.org.htm
. Accessed January 10, 2014.
21. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7:177–188.
22. Ezekowite MD, Connolly S, Parckh A, et al.. Rationale and design of RE-LY: randomized evaluation of long-term anticoagulant therapy, warfarin, compared with dabigatran. Am Heart J. 2009;157:805–810.
23. Dogliotti A, Paolasso E, Giugliano RP. Novel oral anticoagulants in atrial fibrillation: a meta-analysis of large, randomized, controlled trials vs warfarin. Clin Cardiol. 2013;36:61–67.
24. Connolly SJ, Eikelboom J, O'Donnell M, et al.. Challenges of establishing new antithrombotic therapies in atrial fibrillation. Circulation. 2007;116:449–455.
27. Dentali F, Riva N, Crowther M, et al.. Efficacy and safety of the novel oral anticoagulants in atrial fibrillation: a systematic review and meta-analysis of the literature. Circulation. 2012;126:2381–2391.
28. Cabral KP, Ansell J, Hylek EM. Future directions of stroke prevention in atrial fibrillation: the potential impact of novel anticoagulants and stroke risk stratification. J Thromb Haemost. 2011;9:441–449.
29. Morgan CL, McEwan P, Tukiendorf A, et al.. Warfarin treatment in patients with atrial fibrillation: observing outcomes associated with varying levels of INR control. Thromb Res. 2009;124:37–41.