For physicians, three new oral anticoagulants offer potential alternatives to warfarin for reducing the risk of stroke in patients with atrial fibrillation. But in the absence of head-to-head comparisons, how can clinicians and payors choose? Focusing their analysis of clinical trial data on dabigatran, rivaroxaban and apixaban on high-risk patients, researchers found that no one drug stood out for efficacy but that apixaban produced fewer major hemorrhages.
Sebastian Schneeweiss, MD, ScD, of the division of pharmacoepidemiology and pharmacoeconomcs at Brigham and Women’s Hospital in Boston, and colleagues used a method called adjusted indirect comparison to try to tease out differences among these three drugs. In their study published online July 10 in Circulation: Cardiovascular Quality and Outcomes, they noted that Lip et al had conducted a similar analysis of the three anticoagulants using the same method (J Am Coll Cardiol May 2012 online), but they argued the study missed the mark because it failed to adjust for differences in patient populations in the drugs’ three clinical trials: ARISTOTLE, RE-LY and ROCKET-AF.
ARISOTLE, RE-LY and ROCKET-AF compared apixaban 5 mg BID, dabigatran 150 mg BID and rivaroxaban 20 mg QD with warfarin, respectively, but only ROCKET-AF required participants to have a CHADS 2 score of 2 or greater, they wrote. Consequently, the proportion of patients with stroke risk factors was higher in both the rivaroxaban and warfarin patient groups in ROCKET-AF. To address that difference, Schneeweiss and colleagues compared all patients in the ARISTOTLE and RE-LY trials as well as a subgroup of patients with an elevated risk of stroke to make results more comparable to ROCKET-AF.
Their analyses included 44,535 patients from all three trials. The primary efficacy endpoint was stroke or systemic embolism. The safety endpoint was major hemorrhage. They found no statistical difference in efficacy among the drugs but apixaban had a lower risk of major hemorrhage.
“Among all participants, apixaban had a lower risk of major bleeding than dabigatran, with no significant differences in efficacy,” Schneeweiss and colleagues wrote. “Among patients with CHADS 2 3 or greater, dabigatran and apixaban trended toward greater efficacy than rivaroxaban, but the differences were not significant. Major bleeding risk on apixaban was significantly less than that on rivaroxaban or dabigatran.”
Despite the adjustment for the patient populations, their study had limitations related to differences in the three clinical trial designs. For instance, the three trials used different protocols for blinding and different warfarin dosing, and ARISTOTLE and RE-LY lacked complete intention-to-treat safety data. But the authors added that when done carefully, adjusted indirect treatment comparisons can provide reliable estimates.
“In the absence of head-to-head comparisons from randomized trials or effectiveness studies using routine care data, adjusted indirect comparisons using data from RCTs [randomized clinical trials] can provide useful approximations for clinical decision makers, information that can be generated even before a drug is authorized for marketing for a given indication, as with apixaban.”
Apixaban (Eliquis, Bristol-Myers Squibb and Pfizer) has yet to be approved for the prevention of stroke in patients with atrial fibrillation in any country, making supplemental information limited, the authors wrote. The FDA approved dabigatran (Pradaxa, Boehringer Ingelheim) for that indication in 2010 and rivaroxaban (Xarelto, Janssen Research & Development) in 2011. The authors pointed out that the relative effectiveness and safety of these anticoagulants could change in real-world clinical practice, and should be monitored in observational claims-based studies.
“Because such studies can themselves take some time to organize and conduct, rigorous comparisons of existing trial data—if feasible and responsibly conducted—may offer the best basis for evidence-based decision making when new drugs are first brought to market,” they concluded.