New oral anticoagulants, including apixaban, dabigatran, and rivaroxaban, have been developed as alternatives to warfarin, the standard oral anticoagulation therapy for patients with atrial fibrillation (AF). A systematic review and meta-analysis of randomized controlled trials was performed to compare the efficacy and safety of new oral anticoagulants to those of warfarin in patients with AF. The published research was systematically searched for randomized controlled trials of >1 year in duration that compared new oral anticoagulants to warfarin in patients with AF. Random-effects models were used to pool efficacy and safety data across randomized controlled trials. Three studies, including 44,563 patients, were identified. Patients randomized to new oral anticoagulants had a decreased risk for all-cause stroke and systemic embolism (relative risk [RR] 0.78, 95% confidence interval [CI] 0.67 to 0.92), ischemic and unidentified stroke (RR 0.87, 95% CI 0.77 to 0.99), hemorrhagic stroke (RR 0.45, 95% CI 0.31 to 0.68), all-cause mortality (RR 0.88, 95% CI 0.82 to 0.95), and vascular mortality (RR 0.87, 95% CI 0.77 to 0.98). Randomization to a new oral anticoagulant was associated with a lower risk for intracranial bleeding (RR 0.49, 95% CI 0.36 to 0.66). Data regarding the risks for major bleeding (RR 0.88, 95% CI 0.71 to 1.09) and gastrointestinal bleeding (RR 1.25, 95% CI 0.91 to 1.72) were inconclusive. In conclusion, the new oral anticoagulants are more efficacious than warfarin for the prevention of stroke and systemic embolism in patients with AF. With a decreased risk for intracranial bleeding, they appear to have a favorable safety profile, making them promising alternatives to warfarin.
New oral anticoagulants are categorized, on the basis of their targets, as direct thrombin or factor Xa inhibitors. Direct thrombin inhibitors include AZD0837 and dabigatran, and direct factor Xa inhibitors include apixaban, betrixaban, edoxaban, LY-517717, rivaroxaban, and ym-150. Recently, 3 large phase III randomized controlled trials (RCTs), the Apixaban for Reduction of Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trial, the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial, and the Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared With Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET AF), have examined the long-term use of new oral anticoagulants. Although these trials established that new oral anticoagulants were more efficacious than warfarin with respect to the primary end point of combined stroke and systemic embolism, their results pertaining to important secondary efficacy end points as well as safety outcomes were inconclusive or heterogenous. We therefore performed a systematic review and meta-analysis to examine the long-term efficacy and safety of the new oral anticoagulants compared to warfarin in preventing stroke and systemic embolism in patients with atrial fibrillation (AF).
Methods
We systematically searched the published medical research for RCTs comparing new oral anticoagulants to warfarin in patients with AF. The Cochrane Library, Embase, MEDLINE, Science Citation Index Expanded, and ProQuest’s Dissertations and Theses databases were searched from inception through July 2011 without language restriction. The following were used as Medical Subject Heading terms and/or keywords: “new oral anticoagulants,” “oral thrombin inhibitors,” “oral factor Xa inhibitors,” “dabigatran,” “rivaroxaban,” “apixaban,” “edoxaban,” “betrixaban,” “ym-150,” and “LY-517717.” We did not restrict our search to studies conducted in patients with AF, to avoid excluding trials that reported subgroup data on patients with AF. The Embase and MEDLINE searches were limited to clinical trials, and the Embase search was further limited to studies performed in humans. The Science Citation Index Expanded and ProQuest searches were limited to full-text reports. Clinical trial databases, relevant reviews, and the reference lists of retrieved reports were hand searched for potentially relevant studies not identified in our electronic database search.
The PRISMA statement for reporting systematic reviews and meta-analyses of RCTs was used for the method of this study.
Studies were included if (1) they were RCTs, (2) they randomized subjects to warfarin or to non–vitamin K antagonist oral anticoagulants, (3) they were conducted in patients with AF, and (4) they were published in peer-reviewed journals. Studies examining ximelagatran were excluded because it has since been removed from the market because of hepatotoxicity. Conference abstracts and presentations were also excluded, because their results may not be final, and such publications undergo more limited peer review. Open-label and blinded studies were included, because warfarin’s need for monitoring makes blinding difficult. Finally, to assess the long-term efficacy and safety of these agents, only RCTs with follow-up durations of >1 year were included.
Two reviewers independently extracted data from the RCTs. Disagreements were resolved by consensus or, if necessary, by a third party. Data extracted from each RCT included patient- and study-level characteristics as well as outcomes. Extracted patient- and study-level characteristics included average age, median follow-up times, discontinuation rates, mean CHADS 2 scores, gender distribution, mean time in the therapeutic range of warfarin, and proportion of patients with relevant co-morbidities present at baseline. The main efficacy outcome of interest was a composite end point of stroke (including hemorrhagic stroke) and systemic embolism. Other efficacy outcomes were ischemic and unidentified stroke, hemorrhagic stroke, all-cause mortality, vascular mortality, and myocardial infarction. The main safety outcome of interest was major bleeding. Other safety outcomes were gastrointestinal bleeding and intracranial bleeding.
Quality assessment of included trials was conducted using the Cochrane Collaboration’s tool for assessing risk for bias. This assessment tool evaluates bias in an RCT within the following domains: sequence generation; allocation concealment; blinding of participants, personnel, and outcome assessors; incomplete outcome data; selective outcome reporting; and other potential threats to validity. The risk for bias in each domain was classified as high, low, or unclear for each RCT.
We estimated pooled relative risks (RRs) and corresponding 95% confidence intervals (CIs) using DerSimonian and Laird random-effects models, which account for within- and between-study variability. The presence of between-study variability was assessed using the Q statistic (with p <0.10 considered significant), and the proportion of heterogeneity due to between-study variability was estimated using the I 2 index. All analyses were conducted using Stata version 11.0 (StataCorp LP, College Station, Texas).
Results
Our electronic search identified a total of 3,167 reports ( Figure 1 ). After removing duplicates, we screened titles and abstracts, and the full text of 44 publications was retrieved and evaluated for eligibility. Three trials that met our inclusion criteria were identified and included in the present study. One trial was published as an original report with a follow-up report providing additional data. The other 2 trials were presented as ClinicalTrials.gov entries and were subsequently published in peer-reviewed journals. No additional studies were identified from Cochrane systematic reviews, manual searches of the reference lists of retrieved reports, relevant reviews, or clinical trial databases.
The 3 included trials assessed the relative efficacy and safety of a new oral anticoagulant, apixaban, dabigatran, or rivaroxaban, compared to warfarin in patients with AF ( Table 1 ). They were each designed to determine if the study drug was noninferior to warfarin with respect to the composite end point of all stroke and systemic embolism.
| Characteristic | Trial | |||||
|---|---|---|---|---|---|---|
| ARISTOTLE | RE-LY | ROCKET AF | ||||
| Apixaban (n = 9,120) | Warfarin (n = 9,081) | Dabigatran (n = 6,076) | Warfarin (n = 6,022) | Rivaroxaban (n = 7,131) | Warfarin (n = 7,133) | |
| Dose (mg) | 5 twice daily ⁎ | Target INR of 2.0–3.0 | 150 twice daily | Target INR of 2.0–3.0 | 20/day † | Target INR of 2.0–3.0 |
| Drug discontinuation rate | 25.3% | 27.5% | 21.2% | 16.6% | 23.7% | 22.2% |
| Median length of follow-up (days) | 657 | 730 | 707 | |||
| Mean time in therapeutic range | NA | 62% | NA | 64% | NA | 55% |
| Age (years), mean ± SD or median (IQR) | 70 (63–76) | 70 (63–76) | 71.5 ± 8.8 | 71.6 ± 8.6 | 73 (65–78) | 73 (65–78) |
| Women | 35.5% | 35.0% | 36.8% | 36.7% | 39.7% | 39.7% |
| AF type | ||||||
| Persistent or permanent | 84.9% | 84.4% | 67.4% | 66.2% | 81.1% | 80.8% |
| Paroxysmal | 15.1% | 15.5% | 32.6% | 33.8% | 17.5% | 17.8% |
| New | NA | NA | 1.4% | 1.4% | ||
| CHADS 2 score, mean ± SD | 2.1 ± 1.1 | 2.1 ± 1.1 | 2.2 ± 1.2 | 2.1 ± 1.1 | 3.5 ± 0.94 | 3.5 ± 0.95 |
| 0 or 1 | 34.0% | 34.0% | 32.2% | 30.9% | ∼0% | ∼0% |
| 2 | 35.8% | 35.8% | 35.2% | 37.0% | 13.0% | 13.1% |
| 3–6 | 30.2% | 30.2% | 32.6% | 32.1% | 87.0% | 86.9% |
| Previous VKA use ‡ | 57.1% | 57.2% | 50.2% | 48.6% | 62.3% | 62.5% |
| Previous stroke or transient ischemic attack § | 19.2% | 19.7% | 20.3% | 19.8% | 54.9% | 54.6% |
| Heart failure ∥ | 35.5% | 35.4% | 31.8% | 31.9% | 62.6% | 62.3% |
| Diabetes mellitus | 25.0% | 24.9% | 23.1% | 23.4% | 40.4% | 39.5% |
| Hypertension | 87.3% | 87.6% | 78.9% | 78.9% | 90.3% | 90.8% |
⁎ Subjects who, at baseline, were ≥80 years of age, had body weights ≤60 kg, or had serum creatinine levels ≥1.5 mg/dl (133 μmol/L) received apixaban 2.5 mg twice daily.
† Subjects with creatinine clearance of 30 to 40 ml/min at baseline received rivaroxaban 15 mg/day.
‡ Defined as patients who used VKAs for ≥61 and 30 days in RE-LY and ARISTOTLE, respectively. Unspecified for ROCKET AF.
§ For ROCKET AF and ARISTOTLE, these data include patients who had systemic embolisms.
∥ For ROCKET AF and ARISTOTLE, these data include patients with left ventricular ejection fractions <35% and <40%, respectively.
In ARISTOTLE, 18,201 patients with nonvalvular AF were randomized to either apixaban 5 mg twice daily or to warfarin. In RE-LY, 18,113 patients with nonvalvular AF were randomized to 1 of 3 treatment arms: dabigatran 110 mg twice daily, dabigatran 150 mg twice daily, or warfarin. The 150-mg dose was used in our analysis because it is the dose administered to patients with AF. ROCKET AF compared a 20 mg/day dose of rivaroxaban to warfarin in 14,264 patients with nonvalvular AF.
These 3 trials randomized a total of 44,563 patients, 22,327 to new oral anticoagulants and 22,236 to warfarin. The mean length of follow-up ranged from 657 to 730 days, and the average age ranged from 70 to 73 years. Mean CHADS 2 scores were between 2.1 and 3.5. Women constituted 35% to 40% of the study populations, and the mean time in the therapeutic range of warfarin ranged from 55% to 64%.
Quality assessment of included trials was conducted using the Cochrane tool for assessing risk for bias. In RE-LY, patients were unblinded with respect to dabigatran or warfarin assignment. However, all investigators, coordinating center members, the steering committee, the event adjudication committee, and the sponsor were blinded during event ascertainment and analyses. As such, the risk for bias for RE-LY was described as low for the domain of blinding.
ROCKET AF, a trial whose primary objective was one of noninferiority, performed its efficacy and safety analyses on the basis of per protocol and as-treated populations rather than the intention-to-treat population. Although such analyses are appropriate for noninferiority designs, they disturb the integrity of randomization, leading to potential confounding and selection bias. Although an intention-to-treat analysis was provided for the primary efficacy outcome of combined stroke and systemic embolism, it was not provided for other efficacy and safety outcomes. Consequently, ROCKET AF was considered unclear in the domain of other sources of bias. Importantly, the as-treated population excluded only 28 of 14,264 patients included in the intention-to-treat analyses.
ARISTOTLE, another noninferiority trial, used an intention-to-treat analysis for all efficacy outcomes but not for safety outcomes, for which only patients who received ≥1 dose of study drug were considered. It was therefore considered unclear in the domain of other sources of bias. These safety analyses excluded 61 of 18,201 patients included in the intention-to-treat population.
In each trial, the new oral anticoagulants were found to be at least noninferior to warfarin for the composite end point of stroke (including hemorrhagic stroke) and systemic embolism ( Table 2 ). ARISTOTLE and RE-LY further demonstrated the superiority of apixaban and dabigatran, respectively, to warfarin with respect to this composite end point. All 3 drugs were associated with a significantly decreased risk for hemorrhagic stroke compared to warfarin. RRs of other secondary efficacy outcomes, including ischemic stroke, all-cause mortality, vascular mortality, and myocardial infarction, were comparable or inconclusive in ≥2 of the 3 trials. With regard to safety, dabigatran and rivaroxaban were found to have comparable risks for major bleeding to warfarin, while apixaban demonstrated superiority for this outcome. Gastrointestinal bleeding data were heterogenous among the RCTs. The new oral anticoagulants were each associated with a decreased risk for intracranial bleeding compared to warfarin.
| Outcome ⁎ | Trial | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| ARISTOTLE † | RE-LY | ROCKET AF ‡ | |||||||
| Apixaban (n = 9,120) (%/year) | Warfarin (n = 9,081) (%/year) | HR (95% CI) | Dabigatran (n = 6,076) (%/year) | Warfarin (n = 6,022) (%/year) | HR (95% CI) | Rivaroxaban (n = 7,081) (%/year) | Warfarin (n = 7,090) (%/year) | HR (95% CI) | |
| All stroke or systemic embolism | 1.27% | 1.60% | 0.79 (0.66–0.95) | 1.11% | 1.71% | 0.65 (0.52–0.81) | 2.1% | 2.4% | 0.88 (0.75–1.03) |
| Ischemic or unspecified stroke § | 0.97% | 1.05% | 0.92 (0.74–1.13) | 0.92% | 1.20% | 0.76 (0.60–0.98) | NR | NR | NR |
| Hemorrhagic stroke | 0.24% | 0.47% | 0.51 (0.35–0.75) | 0.10% | 0.38% | 0.26 (0.14–0.49) | 0.26% | 0.44% | 0.59 (0.37–0.93) |
| Myocardial infarction | 0.53% | 0.61% | 0.88 (0.66–1.17) | 0.81% | 0.64% | 1.27 (0.94–1.71) | 0.91% | 1.12% | 0.81 (0.63–1.06) |
| All-cause mortality | 3.52% | 3.94% | 0.89 (0.80–0.998) | 3.64% | 4.13% | 0.88 (0.77–1.00) | 1.87% | 2.21% | 0.85 (0.70–1.02) |
| Vascular mortality | 1.80% | 2.02% | 0.89 (0.76–1.04) | 2.28% | 2.69% | 0.85 (0.72–0.99) | 1.53% | 1.71% | 0.89 (0.73–1.10) |
| Major bleeding ∥ | 2.13% | 3.09% | 0.69 (0.60–0.80) | 3.32% | 3.57% | 0.93 (0.81–1.07) | 3.6% | 3.4% | 1.04 (0.90–1.20) |
| Gastrointestinal bleeding | 0.76% | 0.86% | 0.89 (0.70–1.15) | 1.51% | 1.02% | 1.50 (1.19–1.89) | NR ¶ | NR | NR |
| Intracranial bleeding | 0.33% | 0.80% | 0.42 (0.30–0.58) | 0.30% | 0.74% | 0.40 (0.27–0.60) | 0.5% | 0.7% | 0.67 (0.47–0.93) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
