The COBRRA Trial — Ending the Venous Thromboembolism Safety Toss-up
Lisa K. Moores
N Engl J Med 2026;394:1123-1124
DOI: 10.1056/NEJMe2600525
The landscape of treatment options for venous thromboembolism was transformed by the arrival of direct oral anticoagulants. Large-scale trials showed that the direct oral anticoagulants apixaban and rivaroxaban were noninferior to vitamin K antagonists in terms of efficacy and had a better safety profile. However, for over a decade, a critical question remained: which direct oral anticoagulant is safest?
In this issue of the Journal, Castellucci and colleagues1 report the results of the COBRRA (Comparison of Bleeding Risk between Rivaroxaban and Apixaban) trial. This landmark trial provides the first evidence from a randomized, head-to-head comparison to guide clinicians in choosing between the two most prescribed anticoagulants for acute venous thromboembolism.
Until now, the preference for one direct oral anticoagulant over another was based largely on indirect comparisons. In the registration trials comparing the two direct oral anticoagulants with vitamin K antagonists, the risk of clinically relevant bleeding (major and nonmajor) was 4.3% with apixaban in the AMPLIFY (Apixaban for the Initial Management of Pulmonary Embolism and Deep-Vein Thrombosis as First-Line Therapy) trial,2 8.1% with rivaroxaban in the EINSTEIN-DVT trial,3 and 10.3% with rivaroxaban in the EINSTEIN-PE trial.4 Although these baseline risks suggested a safety advantage with apixaban over rivaroxaban, differences in the trial designs and heterogeneity in the patient populations made direct conclusions about safety difficult. Recent large-scale evidence from an analysis of multiple databases that included more than 160,000 patients5 and a network meta-analysis6 reinforced this apparent difference, consistently showing that apixaban appeared to be associated with a lower risk of major bleeding than rivaroxaban. Clinical practice guidelines nevertheless maintained a neutral stance.7-9 It is notable that the 2021 American College of Chest Physicians guidelines acknowledged emerging observational data, stating that limited direct comparisons suggest that the risk of bleeding may be lower with apixaban, but this observation was not included in their formal recommendations.9
The COBRRA trial was a pragmatic, international trial with a PROBE (prospective, randomized, open-label, blinded end-point) design.1 A total of 2760 patients with acute symptomatic pulmonary embolism or proximal deep-vein thrombosis were assigned to receive 3 months of treatment with apixaban (at a dose of 10 mg twice daily for 7 days followed by 5 mg twice daily) or rivaroxaban (at a dose of 15 mg twice daily for 21 days followed by 20 mg daily). The results are striking. The primary outcome, as adjudicated by an independent central adjudication committee, was clinically relevant bleeding, a composite of major bleeding or clinically relevant nonmajor bleeding. During the treatment period, a primary-outcome event occurred in 3.3% of the patients in the apixaban group as compared with 7.1% of those in the rivaroxaban group (relative risk, 0.46; 95% confidence interval [CI], 0.33 to 0.65; P<0.001). This translates to a decrease of more than 50% in the risk of bleeding with apixaban.
The safety advantage of apixaban over rivaroxaban was driven by the lower risks of major bleeding (0.4% vs. 2.4%; relative risk, 0.16; 95% CI, 0.06 to 0.40) and clinically relevant nonmajor bleeding (2.9% vs. 4.9%; relative risk, 0.59; 95% CI, 0.40 to 0.86). It is important to indicate that the significant safety benefit did not come at the cost of efficacy: the risk of recurrent symptomatic venous thromboembolism was nearly identical in the apixaban and rivaroxaban groups (1.1% vs. 1.0%; relative risk, 1.08; 95% CI, 0.52 to 2.23).
A key discussion point arises from differences in the duration of the loading-dose phase with each drug. Apixaban was given at a loading dose of 10 mg twice daily for 7 days, whereas rivaroxaban was given at a loading dose of 15 mg twice daily for 21 days. Data from the trial suggest that this prolongation of the loading-dose phase by 2 weeks in the rivaroxaban group was a primary driver of the disparity in safety outcomes. The Kaplan–Meier curves for clinically relevant bleeding show an early and sharp separation between the trial groups during the first 21 days of treatment. This observation suggests that the extended high-dose phase of rivaroxaban might have resulted in a higher burden of bleeding without a corresponding increase in protection against thrombosis. Essentially, the extra 14 days of high-intensity therapy in the rivaroxaban group appears to have intensified the antithrombotic effect, leading to an increased risk of bleeding without enhancing efficacy.
Fluctuations in pharmacokinetic measures may offer an additional explanation of the difference in bleeding risk. Pharmacokinetic data have shown that the 20-mg once-daily maintenance dose of rivaroxaban results in peak plasma concentrations that are higher than those resulting from the 5-mg twice-daily maintenance dose of apixaban.10 These higher peaks may reach a critical drug concentration above which anticoagulant activity in mucosal tissues becomes sufficient to cause clinically significant bleeding, a hypothesis that is supported by the higher risks of hematuria (0.3% vs. 1.3%) and gastrointestinal bleeding (0.6% vs. 1.0%) with rivaroxaban than with apixaban.
Despite its rigor, the current trial has limitations. The trial was not powered to assess the significance of differences in the risk of recurrent venous thromboembolism. Furthermore, the trial population lacked diversity according to race and ethnic group, with approximately 90% of the participants reporting White race. The results may also not be generalizable to high-risk groups excluded from the trial, such as patients with cancer-associated thrombosis, clinically significant renal or hepatic dysfunction, or a body weight of more than 120 kg. Finally, because data cover only the initial 3 months of treatment, whether these safety differences persist during an extended period of secondary prevention remains unknown.
Despite these limitations, the trial provides vital evidence for the treatment of venous thromboembolism. By confirming the safety advantage of apixaban over rivaroxaban in a randomized trial, these results should prompt a reevaluation of current guidelines. For many patients with acute venous thromboembolism, the choice of anticoagulant is no longer a toss-up. Apixaban is a safer first-line option than rivaroxaban for minimizing the risk of bleeding without compromising the prevention of recurrent thrombosis.
