Apixaban is one of the new oral anticoagulants, which is prescribed as an alternative to vitamin K antagonists (VKAs). Concerns regarding its bleeding profile persist and require further evaluation. Therefore, we conducted a meta-analysis of randomized controlled trials (RCTs) to compare the risks of bleeding and all-cause mortality between apixaban and VKAs. The MEDLINE, EMBASE, and Cochrane Library of Clinical Trials databases were systematically searched for RCTs comparing the risks of bleeding and all-cause mortality of apixaban (2.5 or 5 mg twice daily) with those of VKAs. We included RCTs conducted in adults and published in English or French. Data were pooled across RCTs using random-effects meta-analytical models. Our systematic search identified 5 RCTs meeting our inclusion criteria (n = 24,435). They included patients with atrial fibrillation (n = 18,358), total knee replacement surgery (n = 458), and venous thromboembolism (n = 5,619). Data pooled across RCTs revealed that apixaban was associated with reduced risks of any bleeding (relative risk [RR] 0.73, 95% confidence interval [CI] 0.59 to 0.90) and a composite of major or clinically relevant nonmajor bleeding (RR 0.60, 95% CI 0.40 to 0.88). Apixaban was also associated with a lower risk of intracranial bleeding (RR 0.42, 95% CI 0.31 to 0.58) whereas analyses of major and minor bleeding were inconclusive. Moreover, apixaban was associated with decreased all-cause mortality (RR 0.89, 95% CI 0.81 to 0.99) although this finding was driven by the results of the ARISTOTLE trial. In conclusion, our meta-analysis found that apixaban is associated with a lower risk of bleeding than VKAs, providing some reassurance regarding its safety.
Highlights
- •
The results are based on data from 24,435 patients across 5 trials.
- •
Apixaban is associated with a lower risk of bleeding compared with vitamin K antagonists (VKAs).
- •
Apixaban may be associated with reduced all-cause mortality compared with VKAs.
- •
Apixaban may represent an effective and safe alternative to the standard VKA therapy.
The vitamin K antagonist (VKA) warfarin is the most commonly used anticoagulant. Its usage, however, is limited by several dietary restrictions, drug interactions, and required monitoring. Therefore, a large proportion of patients are unsuitable or unwilling to use it. The factor Xa inhibitor apixaban, one of the new oral anticoagulants (NOACs), is a potential alternative with fewer interactions and fixed-dose prescription. Apixaban was approved by the US Food and Drug Administration (FDA) for the reduction of the risk of stroke and systemic embolism in patients with nonvalvular atrial fibrillation (AF). In March 2014, the FDA expanded its approved indications to include prophylaxis of deep vein thrombosis after knee or hip replacement surgery. Despite its regulatory approval, concerns remain regarding apixaban’s potential risk of bleeding; as with other NOACs such as dabigatran and rivaroxaban, there is no rapid and efficient method to reverse its anticoagulation effects. The expansion of apixaban’s list of indications represents an opportune time to synthesize the available evidence regarding its bleeding risk. Thus, we conducted a meta-analysis of randomized controlled trials (RCTs) to compare the risks of bleeding and all-cause mortality between apixaban and VKAs.
Methods
We followed a prespecified protocol to conduct this meta-analysis and report it here as per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses. The MEDLINE (through Ovid), EMBASE (through Ovid), and Cochrane Library of Clinical Trials databases were systematically searched from inception to May 2014 using keywords, Medical Subject Headings terms, and Emtree terms for apixaban (“apixaban,” “Eliquis,” “BMS-562247,” “factor Xa inhibitor,” “direct factor Xa inhibitor,” or “oral factor Xa inhibitor”) and VKAs (“warfarin,” “acenocoumarol,” or “vitamin K antagonist”; Supplementary Data ). These search terms were combined with a modified version of the highly specific McMaster RCT hedge for the MEDLINE and EMBASE literature searches to restrict results to RCTs. References of relevant RCTs and previously published reviews were also hand-searched for additional RCTs not found through the electronic database searches. We only included RCTs randomizing participants to apixaban (2.5 or 5 mg twice daily [BID], as clinically recommended ) or to VKAs, involving adults aged ≥18 years, and published in English or French. Because VKA usage requires international normalized ratio (INR) monitoring, we included both blinded and open-label RCTs. We excluded placebo-controlled trials, trials involving patients with malignancies, reviews, editorials, commentaries, and letters to the editor.
Data on study- and patient-level characteristics and on the following bleeding outcomes were extracted: any, major/clinically relevant nonmajor (CRNM), major, CRNM, minor, critical, associated with a decrease in hemoglobin ≥2 g/dl, fatal, gastrointestinal, intracranial, intraocular, intra-articular, and intramuscular bleeding. We also extracted all-cause mortality as this clinically relevant end point was reported in all trials. However, given potential differences in causes of death in the variable patient populations in the included trials, this end point should be interpreted with caution. In RCTs with multiple apixaban arms, only those for which patients received the FDA-approved dosages (2.5 and 5 mg BID) were extracted. Data on bleeding were extracted from RCTs’ safety population (typically randomized patients who had received ≥1 dose of study treatment). Data on all-cause mortality were extracted from the RCTs’ efficacy (i.e., intention-to-treat) populations, except in 2 trials in which data were extracted from the safety populations. All data were extracted independently by 2 reviewers using standardized, pilot-tested data collection forms. Disagreements were resolved by consensus or, if necessary, by a third reviewer (KBF). We used the Cochrane Collaboration’s tool for assessing the risk of bias to assess the quality of included RCTs. All RCTs meeting our inclusion criteria were included regardless of their quality.
Relative risks (RRs) and their corresponding 95% confidence intervals (CIs) were generated for bleeding and mortality outcomes comparing patients randomized to apixaban with those randomized to VKAs using DerSimonian and Laird random-effects models with inverse-variance weighting. All analyses were stratified by patient population, and both population-specific and overall RRs were estimated. In our primary analysis, we pooled count data, as all RCTs presented results in this form. To estimate the proportion of the variance that was due to between-RCT heterogeneity, we calculated I 2 values. Pooled risk differences with corresponding 95% CIs were also calculated. In sensitivity analyses, data were pooled using fixed-effects models with inverse-variance weighting. In all analyses, we applied a 0.5 continuity correction to include data from RCTs with 0 events in 1 arm; if both arms had 0 events, the RCT was given 0 weight. To examine the influence of the largest included RCT (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation [ARISTOTLE] ), we conducted sensitivity analyses excluding it. All statistical analyses were performed using the program R, version 3.0.2 (R Core Team [2013], R Foundation for Statistical Computing, Vienna, Austria.)
Results
Our systematic search identified 1,009 potentially relevant publications ( Figure 1 ). Of those, 268 were duplicates that were subsequently removed, leaving 741 publications to screen by title and abstract. This process yielded 7 reports for full-text review. After an in-depth assessment of these 7 reports, 5 were identified as RCTs meeting all our prespecified inclusion criteria and included in our meta-analysis. No additional reports were identified from hand-searching reference lists of retrieved relevant RCTs and of previously published reviews.
A total of 25,576 patients were randomized in the 5 included RCTs, namely ARISTOTLE, Apixaban for the Initial Management of Pulmonary Embolism and Deep-Vein Thrombosis as First-Line Therapy (AMPLIFY), Apixaban PROphylaxis in Patients undergoing tOtal knee replacement Surgery (APROPOS), Botticelli, and ARISTOTLE-J ( Table 1 ). Patients were randomized to apixaban for AF, after total knee replacement (TKR), or for venous thromboembolism (VTE). Sample sizes of individual RCTs ranged from n = 222 to n = 18,201. Of the 25,576 randomized patients, 24,435 were included in our bleeding analyses because such analyses were restricted to the RCTs’ safety populations (i.e., all randomized patients who received ≥1 dose of the study drug), and we further restricted inclusion to the RCT arms in which patients were randomized to FDA-approved doses (i.e., 2.5 or 5 mg BID) of apixaban. In contrast, a total of 24,367 patients (n = 18,423 AF, 325 TKR, and 5,619 VTE) were included in our all-cause mortality analysis, which was based on the RCTs’ intention-to-treat or efficacy populations.
| Trial † | Year | Population | API Dose (mg) | Comparator | Target INR | Number of Patients (n) | Drug D/C (%) | Duration (Days) | Lost to Follow-up (%) | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Total Randomized | Safety Analysis | Tx | Follow- up | |||||||||||
| API | VKA | API | VKA | API | VKA | |||||||||
| ARISTOTLE | 2011 | AF | 5 ‡ | Warfarin | 2.0-3.0 | 18,201 | 9,088 | 9,052 | 25.3 | 27.5 | 657 § | 657 § | 0.4 | 0.4 |
| AMPLIFY | 2013 | VTE | 5 | Enoxaparin + warfarin || | 2.0-3.0 | 5,395 | 2,676 | 2,689 | 14.1 | 15.3 | 187 | 187 | 0.5 | 0.5 |
| APROPOS ¶ | 2007 | TKR | 2.5, 5 | Warfarin | 1.8-3.0 | 1,238 | 307 | 151 | NR | NR | 10-14 | 40-44 | NR | NR |
| Botticelli ¶ | 2008 | VTE | 5 | LMWH + VKA || # | 2.0-3.0 | 520 | 128 | 126 | 13.1 | 7.8 | 91 | 91 | NR | NR |
| ARISTOTLE-J | 2011 | AF | 2.5, 5 | Warfarin | 2.0-3.0 | 222 | 143 | 75 | 5.6 | 5.3 | 85 | 85 | NR | NR |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
