What is venous thrombosis?

Venous thrombosis is a blood clot that forms in a vein. It can happen in the leg — called phlebitis or deep vein thrombosis — or in the lungs, where it becomes a pulmonary embolism. Both are potentially serious and require prompt treatment.

The treatment: anticoagulant drugs

To dissolve these clots and prevent them from growing, doctors prescribe anticoagulants — drugs that "thin" the blood. The two most widely used today are apixaban (Eliquis) and rivaroxaban (Xarelto). Both are effective, but which causes less bleeding? That's exactly the question this study set out to settle.

What the study did

Researchers recruited 2,760 patients across several countries. Each patient had been diagnosed with acute phlebitis or a pulmonary embolism. They were randomly assigned one drug or the other for 3 months, and bleeding was monitored.

The result, plainly

In plain terms: with apixaban, about 1 patient in 30 had a significant bleed. With rivaroxaban, it's about 1 in 14. Apixaban reduces this risk by 54%. The difference is highly significant and unlikely to be due to chance.

Clinical background

Apixaban and rivaroxaban are both direct factor Xa inhibitors, reference oral anticoagulants in the management of proximal deep vein thrombosis (DVT) and symptomatic pulmonary embolism (PE). Although their anticoagulant efficacy is comparable, their dosing profile differs: apixaban is administered twice daily, rivaroxaban once daily in maintenance. Indirect data had suggested a bleeding advantage for apixaban for several years, but no head-to-head comparative trial had yet provided direct proof.

Trial design

The COBRRA trial (NCT03266783) is an international, randomized, open-label trial with blinded endpoint assessment (PROBE design). 2,760 patients with acute proximal DVT or symptomatic PE were randomized 1:1:

• Apixaban: 10 mg × 2/day for 7 days, then 5 mg × 2/day through month 3
• Rivaroxaban: 15 mg × 2/day for 21 days, then 20 mg × 1/day through month 3

The primary endpoint was clinically significant bleeding, defined per ISTH criteria (International Society on Thrombosis and Haemostasis), combining major bleeding and clinically relevant non-major bleeding.

Main results

The p-value for the primary endpoint is less than 0.001, indicating a highly significant difference. The mortality benefit is numerical but not significant (confidence interval very wide, as events are rare).

Pharmacological rationale

Apixaban and rivaroxaban are both direct reversible factor Xa inhibitors, orally active, requiring no biological monitoring. Their pharmacokinetic differences are nonetheless notable: apixaban has a half-life of ~12h, bioavailability of ~50%, predominantly hepatobiliary elimination (~75%) and lower renal dependence. Rivaroxaban has a half-life of ~5-9h, bioavailability of ~80-100% (food-dependent at the 20 mg dose) and ~33% renal elimination. These differences, combined with apixaban's more fractionated dosing, could explain a more favorable bleeding profile — a hypothesis COBRRA now confirms.

Trial methodology

PROBE (Prospective Randomized Open-label Blinded Endpoint) design — chosen because such different dosing schedules cannot be blinded, with central blinded adjudication of bleeding and thrombotic events. 1:1 randomization, stratified by center and by index event type (proximal DVT vs PE).

Dosing: apixaban 10 mg × 2/d × 7 d then 5 mg × 2/d; rivaroxaban 15 mg × 2/d × 21 d then 20 mg × 1/d. These doses exactly reflect current marketing-authorization recommendations for acute VTE.

Endpoints (ISTH definitions)

Composite primary endpoint: clinically significant bleeding = major bleeding (Hb drop ≥ 2 g/dL, transfusion ≥ 2 RBC units, bleeding in critical site or fatal) or clinically relevant non-major bleeding (CRNM: requiring medical intervention, unplanned hospitalization, or impairing daily activities).

Secondary endpoint: all-cause death.

Detailed results

Critical analysis and limitations

• Open-label (not double-blind) design: risk of detection bias for CRNM bleeding, mitigated by central blinded adjudication.
• No primary endpoint on thrombotic recurrence: the trial is not sized to demonstrate non-inferiority on anticoagulant efficacy — that was not its objective.
• Population: proximal DVT and symptomatic PE only; results are not directly generalizable to distal DVT or cancer-related VTE (excluded or underrepresented population).
• Duration limited to 3 months: no data on extended treatment.
• Mortality: events too rare (5 deaths total) for any conclusion — the 95%CI (0.03–2.26) is too wide.

Clinical implications

These data constitute the first direct evidence of apixaban's bleeding superiority over rivaroxaban in acute VTE during the initial treatment phase. This result is likely to influence future guidelines from learned societies (ESC, ASH, ACCP) which until now considered the two molecules equivalent on this point. The trial design, supported by the CIHR (Canadian Institutes of Health Research), confers high methodological robustness.