Novel oral anticoagulants (NOACs) are safe and effective for the prevention of stroke or systemic embolism (S/SE) in atrial fibrillation. The efficacy and safety of NOACs compared with warfarin has not been systematically assessed in subjects with mild or moderate renal dysfunction. We performed a meta-analysis of the randomized clinical trials that compared efficacy and safety (major bleeding) outcomes of NOACs compared to warfarin for the treatment of nonvalvular atrial fibrillation and had available data on renal function. We estimated the pooled relative risk (RR) of S/SE and major bleeding in relation to renal function (assessed by baseline estimated glomerular filtration rate divided in 3 groups: normal [estimated glomerular filtration rate >80 ml/min], mildly impaired [50 to 80 ml/min], and moderate impairment [<50 ml/min]). We included 4 randomized clinical trials enrolling a total of 58,338 subjects. The RRs of S/SE and major bleeding were higher in subjects with renal impairment compared to normal renal function, independent of type of anticoagulant therapy. In subjects with normal renal function, no difference in the risk of S/SE was observed, whereas the risk of major bleeding was slightly lower for subjects taking NOACs (RR 0.87, 95% confidence interval [CI] 0.76 to 0.99). In subjects with mild or moderate renal impairment, NOACs were associated with a reduced risk of S/SE (RR 0.75, 95% CI 0.66to 0.85 and RR 0.80, 95% CI 0.68 to 0.94, respectively) and major bleeding (RR 0.87, 95% CI 0.79 to 0.95 and RR 0.80, 95% CI 0.71 to 0.91, respectively) compared to warfarin. The pooled analysis for major bleeding demonstrated significant heterogeneity. In conclusion, the use of NOACs was associated with a reduced risk of S/SE and reduced risk of major bleeding compared to warfarin in subjects with mild or moderate renal impairment suggesting a favorable risk profile of these agents in patients with renal disease.
Atrial fibrillation (AF) is the most common cardiac arrhythmia and is associated with fivefold increased risk of cardioembolic stroke. Warfarin reduces the risk of stroke or systemic embolism (S/SE) by approximately two-thirds and recent large-scale clinical trials have demonstrated that the novel oral anticoagulants (NOACs) are associated with a similar reduction in stroke with a favorable risk profile. Advantages of the NOACs over warfarin include the predictable pharmacokinetics, rapid onset of action, fewer drug–drug interactions and dietary restrictions, and convenience due to the lack of need for ongoing monitoring. Renal impairment is common in patients with AF and has an important impact on clinical outcomes. Patients with AF and renal dysfunction are at particularly high risk of thromboembolic complications as well as an increased risk of bleeding. Each of the NOACs has some degree of renal metabolism or excretion and their efficacy and safety across the range of mild-to-moderate renal impairment have not been fully studied. We performed a systematic review and meta-analysis of the safety and efficacy of the 4 Food and Drug Administration (FDA)–approved NOACs compared to warfarin in nonvalvular AF in relation to renal function.
Methods
We searched for randomized clinical studies that compared the risk of S/SE events in subjects with nonvalvular AF randomized to any of the 4 NOACs (dabigatran, rivaroxaban, apixaban, or edoxaban) versus warfarin and that reported outcomes in relation to renal function. We searched MEDLINE, EMBASE, Web of Science, and the Cochrane Database using keywords: “atrial fibrillation,” “dabigatran,” “rivaroxaban,” “apixaban,” “edoxaban,” “warfarin,” “new oral anticoagulant,” “oral factor Xa inhibitor,” and “oral thrombin inhibitor.” Studies were restricted to English language in peer-reviewed journals until January 2014. We searched the US FDA databases ( www.fda.gov ) to obtain data from the respective drug (NOACs) sponsor’s application pursuing FDA approval. Finally, principal investigators and/or sponsors of the trials were also contacted.
Data were independently reviewed and extracted. Studies considered in the present study were: (1) randomized clinical trials evaluating the safety and efficacy of any new oral anticoagulant compared to oral anticoagulation with therapeutic dose-adjusted warfarin, for prevention of S/SE in patients with nonvalvular AF, (2) evaluated renal function at baseline, and (3) reported outcomes in relation to renal function. For trials reporting more than one publication, data were extracted from the most complete publication, using the other reports to clarify or complement the information obtained. To assess the agreement between reviewers for study selection, we used the κ statistic (κ = 0.78).
Data included design, randomization protocol, allocation concealment, number of participants, mean age, gender, therapeutic interventions, type of drug, duration of treatment, and baseline renal function. Primary efficacy outcome: stroke (ischemic or hemorrhagic) or systemic embolism. Primary safety outcome: major bleeding as defined by the International Society of Thrombosis and Hemostasis criteria (bleeding causing a decrease in hemoglobin levels of >2 g/dl, bleeding leading to transfusion of >2 units of blood, symptomatic bleeding in a critical area [intracranial, intraspinal, intraocular, retroperitoneal, intrarticular, pericardial, or intramuscular with compartment syndrome], or bleeding events leading to death). Secondary outcomes: total mortality, cardiovascular mortality, ischemic stroke, intracranial bleeding, gastrointestinal bleeding, and myocardial infarction.
Information about these outcomes was stratified by renal function, according to baseline estimated glomerular filtration rate (eGFR) using the Cockcroft–Gault formula and classified in the following groups: eGFR <50 ml/min (moderate renal impairment), eGFR 50 to 80 ml/min (mild renal impairment), and eGFR >80 ml/min (normal renal function). Two reviewers independently assessed the quality of the included studies using the checklist of the Cochrane Database of Systematic Reviews that assesses the risk of bias in sequence generation, allocation concealment, blinding, attrition, selection, and other biases.
We calculated pooled relative risks (RRs) and corresponding 95% confidence intervals (CIs) for the primary efficacy (S/SE), safety (major bleeding), and when available for the secondary outcomes in relation to the 3 groups according to renal function. When possible, we did analysis with the intention-to-treat population. In the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) study, 2 doses of dabigatran were tested; we only included the higher dose. In the Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation-Thrombolysis in Myocardial Infarction 48 (ENGAGE AF-TIMI 48) trial, 2 doses of edoxaban were tested and we included data from the higher dose (60 mg/30 mg). A separate sensitivity analysis was carried out including only factor Xa inhibitors. Outcomes were then pooled and compared with a fixed effects model (Mantel–Haenszel method). A value of p <0.05 was considered statistically significant. All analyses were performed using Review Manager software (RevMan, version 5.2, from the Cochrane Collaboration). The appropriateness of pooling data across studies was assessed using Cochran Q and the I 2 test for heterogeneity.
Results
Of the 43 full-text studies reviewed, 36 studies were excluded (see study flow diagram, Figure 1 ), and of the excluded studies, 3 were randomized studies (1 study was excluded because it had no warfarin comparison group, 1 used different doses of dabigatran, and 1 randomized study was not blinded and not reporting outcomes in renal function subgroups). Six phase II trials were excluded because of no reporting baseline renal function and/or no outcomes in relation to renal function. Four phase III RCTs were included, information from 3 substudies reporting outcomes in relation to renal function were extracted to complete the information from the primary trials. Furthermore, data from the 4 drug sponsor’s application to the FDA were reviewed and used to complement other sources. All investigators of the primary trials and substudies were contacted but were not able to provide more detailed data about their studies.
The 4 included studies enrolled a total of 58,076 subjects. Of them, 29,066 subjects were randomized to an NOAC and 29,010 to warfarin. The studies characteristics are listed in Table 1 . All 4 trials were phase III randomized, 3 of them (ARISTOTLE, ROCKET-AF, and ENGAGE AF-TIMI 58 ) were double-blind and double-dummy studies. The RE-LY trial had the dabigatran arm blinded and the warfarin arm unblinded. The RELY and ARISTOTLE studies enrolled subjects at lower risk of stroke (CHADS 2 score of 2.1 both), whereas the ENGAGE AF-TIMI 58 trial enrolled subjects with CHADS 2 scores average of 2.8, and the ROCKET-AF trial had the subjects at highest risk (CHADS 2 score of 3.5 with no subjects with scores of 0 or 1). The number of subjects with previous warfarin use ranged from 50% to 60%. The time in therapeutic range in the warfarin arms was the lowest in the ROCKET-AF trial (mean 58%) and was similar in the other 3 trials (66% to 68%). All 4 studies had small fraction of subjects lost to follow-up (<0.1%). According to the Cochrane Database of Systematic Reviews assessment of the risk of bias, all 4 studies were deemed of high quality ( Table 2 ).
| RE-LY | ROCKET-AF | ARISTOTLE | ENGAGE-AF TIMI 48 | |||||
|---|---|---|---|---|---|---|---|---|
| Dabi 150 | Warf | Riva | Warf | Apixa | Warf | Edo 60 | Warf | |
| Study design | RCT phase III, unblinded warfarin | RCT phase III, double blind, double dummy | RCT phase III, double blind, double dummy | RCT phase III, double blind, double dummy | ||||
| Population | NVAF and ≥1 risk factor (stroke or TIA, CHF, age >75yo or 65-74 yo + DM or HTN or CAD | NVAF, CHADS score ≥2 | NVAF, ≥1 risk factor (age ≥75 y, stroke or TIA, CHF, DM, HTN | NVAF, CHADS ≥2 | ||||
| Number of randomized subjects | 6076 | 6022 | 7131 | 7133 | 9120 | 9081 | 7035 | 7036 |
| CHADS 2 | 2.2 | 2.1 | 3.5 | 3.5 | 2.1 | 2.1 | 2.8 | 2.8 |
| 0-1 | 32% | 31% | 0 | 0 | 34% | 34% | <1% | <1% |
| 2 | 35% | 37% | 13% | 13% | 36% | 36% | 46% | 47% |
| 3-6 | 33% | 32% | 87% | 87% | 30% | 30% | 54% | 53% |
| Age | 71 (mean) | 73 (median) | 70 (median) | 72 (median) | ||||
| Women | 37% | 37% | 40% | 40% | 36% | 35% | 39% | 38% |
| Previous warfarin use | 50% | 49% | 62% | 63% | 57% | 57% | 59% | 59% |
| Concomitant Aspirin use | 39% | 41% | 36% | 37% | 31% | 31% | 29% | 30% |
| Median follow-up (years) | 2.0 | 1.9 | 1.8 | 2.8 | ||||
| Warfarin group time in therapeutic range | 67 (54-78) | 58 (43-71) | 66 (52-77) | 68 (57-77) | ||||
| Random sequence generation (selection bias) | Allocation concealment (selection bias) | Blinding of participants and personnel (performance bias) | Blinding of outcome assessment (detection bias) | Incomplete outcome data (attrition bias) | Selective reporting (reporting bias) | Other bias | |
|---|---|---|---|---|---|---|---|
| RE-LY | Low | Low | Unclear | Low | Low | Low | Low |
| ROCKET-AF | Low | Low | Low | Low | Low | Low | Low |
| ARISTOTLE | Low | Low | Low | Low | Low | Low | Low |
| ENGAGE AF-TIMI 48 | Low | Low | Low | Low | Low | Low | Low |
As presented in Table 3 , the residual risk of S/SE in subjects receiving NOACs ranged from 2.16% for subjects with normal renal function to 3.8% for subjects with moderate renal impairment. Compared to subjects with normal renal function, the odds ratio of residual stroke was 1.3 times (95% CI 1.1 to 1.5) greater in subjects with mild renal impairment and 1.79 times (95% CI 1.5 to 2.2) greater in subjects with moderate renal impairment. In subjects receiving warfarin, the risk of residual S/SE was highest in the group with moderate renal function; in comparison to subjects with normal renal function, the residual risk of stroke was greater in subjects with moderate and mild renal impairment.
| Stroke or Systemic Embolism | Risk of Stroke | Odds Ratio | 95% CI | P value | |
|---|---|---|---|---|---|
| NOACs | |||||
| Normal Renal Function | 229/10,603 | 2.16% | Ref | Ref | NA |
| Mild Renal Impairment | 352/12,952 | 2.72% | 1.27 ∗ | 1.07-1.5 | <0.01 |
| Moderate Renal Impairment | 210/5,511 | 3.81% | 1.79 ∗ | 1.48-2.17 | <0.001 |
| Warfarin | |||||
| Normal Renal Function | 235/10,515 | 2.23% | Ref | Ref | NA |
| Mild Renal Impairment | 505/13,098 | 3.86% | 1.75 ∗ | 1.50-2.06 | <0.0001 |
| Moderate Renal Impairment | 261/5,397 | 4.84% | 2.22 ∗ | 1.86-2.66 | <0.0001 |
Similarly, the risk of bleeding was greater in subjects with renal impairment ( Table 4 ). In the group receiving NOACs, compared with subjects with normal renal function, the odds of major bleeding were greater in subjects with mild or moderate renal impairment. In subjects receiving warfarin, we observed greater risk of major bleeding in subjects with renal impairment.
| Major Bleeding | Risk of Major Bleed | Odds Ratio | 95% CI | P value | |
|---|---|---|---|---|---|
| NOACs | |||||
| Normal Renal Function | 397/10,606 | 3.74% | Ref | Ref | NA |
| Mild Renal Impairment | 734/12,957 | 5.66% | 1.54 ∗ | 1.36-1.75 | <0.0001 |
| Moderate Renal Impairment | 397/5,514 | 7.20% | 2.00 ∗ | 1.73-2.30 | <0.0001 |
| Warfarin | |||||
| Normal Renal Function | 457/10,512 | 4.35% | Ref | Ref | NA |
| Mild Renal Impairment | 840/13,096 | 6.41% | 1.51 ∗ | 1.34-1.70 | <0.0001 |
| Moderate Renal Impairment | 486/5,411 | 8.98% | 2.17 ∗ | 1.90-2.48 | <0.0001 |
In subjects with moderate renal impairment, S/SE occurred in 3.8% of subjects treated with NOACs and 4.8% of subjects treated with warfarin ( Figure 2 ). The use of NOACs was associated with a significant reduction of S/SE (RR 0.79, 95% CI 0.66 to 0.94). In subjects with mild renal impairment, S/SE occurred in 2.7% of subjects receiving NOACs and in 3.9% of subjects taking warfarin; the use of NOACs was associated with a significant reduction of S/SE (RR 0.71, 95% CI 0.62 to 0.81). In subjects with normal renal function, there was no difference in the risk of S/SE between NOACs or warfarin treatment (RR 0.96, 95% CI 0.81 to 1.15). When analyzing the differences among the groups, the test for interaction was significant (p = 0.02) with moderate degree of heterogeneity among the groups (I 2 = 54%).
In subjects with moderate renal impairment, major bleeding occurred in 7.20% of subjects receiving NOACs and in 8.98% patients receiving warfarin ( Figure 3 , Supplementary Figure 1 ). The use of NOACs was associated with a significant reduction in the risk of major bleeding compared to warfarin (RR 0.80, 95% CI 0.70 to 0.91). In subjects with mild renal impairment, major bleeding occurred in 5.66% of patients taking NOACs and in 6.41% of subjects taking warfarin. Treatment with NOACs was associated with a significant reduction of major bleeding compared to the use of warfarin (RR 0.88, 95% CI 0.80 to 0.97). In subjects with normal renal function, major bleeding occurred in 3.74% of subjects taking NOACs and in 4.35% of patients taking warfarin. The use of NOACs was associated with a significant reduction in the risk of major bleeding (RR 0.86 95% CI 0.75 to 0.98). There was a significant degree of heterogeneity (p = 0.03).
