PDF: ANNEXA-I
(2024)Objective
To assess the efficacy and safety of andexanet alfa versus usual care for hemostatic control in patients with factor Xa inhibitor-associated acute intracerebral hemorrhage (ICH).
Study Summary
• Median reduction in anti-factor Xa activity from baseline to 1-2 h nadir: 94.5% with andexanet vs 26.9% with usual care (median percent change -94.5% vs -26.9%; p<0.001)
• 30-day thrombotic events: 10.3% vs 5.6% (difference 4.6 pp, 95% CI 0.1 to 9.2; p=0.048); ischemic stroke 6.5% vs 1.5%
• 30-day death: 27.8% vs 25.5% (adjusted difference 2.5 pp, 95% CI -5.0 to 10.0; p=0.51); mRS 0-3 at 30 days (post hoc): 28.0% vs 31.0%
Intervention
Multicenter, randomized (1:1), open-label, blinded-endpoint (PROBE) trial comparing intravenous andexanet alfa vs usual care (which could include prothrombin complex concentrate) for reversal of factor Xa inhibitor anticoagulation in acute ICH.
Inclusion Criteria
Adults with acute ICH as the main bleeding event (not subdural or subarachnoid alone; hemorrhage-preceded-by-trauma patients were included — 11.6% andexanet, 14.5% usual care), factor Xa inhibitor use (apixaban, rivaroxaban, or edoxaban) with most recent dose within 15 hours before randomization, hematoma volume 0.5-60 mL on CT/MRI performed within 2 hours before randomization, NIHSS ≤35, GCS ≥7, and ≤6 hours from symptom onset to baseline imaging (per protocol amendment; initial eligibility included any acute FXa-associated intracranial hemorrhage with a 12-hour window).
Study Design
Arms: Andexanet Alfa vs. Usual Care
Patients per Arm: Extended safety population: Andexanet 263, Usual Care 267. Primary efficacy interim analysis: Andexanet 224, Usual Care 228.
Outcome
• Median reduction in anti-factor Xa activity (baseline to 1-2 h nadir): 94.5% vs 26.9% (median percent change -94.5% vs -26.9%; p<0.001).
• 30-day thrombotic events: 10.3% vs 5.6% (p=0.048); ischemic stroke 6.5% vs 1.5%.
• Death at 30 days: 27.8% vs 25.5% (p=0.51); mRS 0-3 at 30 days (post hoc): 28.0% vs 31.0%.
Bottom Line
Among patients with acute intracerebral hemorrhage receiving factor Xa inhibitors, andexanet alfa improved the primary composite hemostatic efficacy endpoint at 12 hours (67.0% vs 53.1% with usual care; adjusted difference 13.4 percentage points, 95% CI 4.6-22.2; P=0.003) and rapidly reduced anti-factor Xa activity, but was associated with a significantly higher rate of thrombotic events including ischemic stroke, with no appreciable difference in 30-day death or modified Rankin scale distribution.
Major Points
- 550 patients underwent randomization (263 andexanet, 267 usual care in the extended safety population of 530 after exclusion of 20 with deferred/emergency consent issues); primary efficacy analysis based on prespecified interim analysis of the first 452 patients (224 andexanet, 228 usual care).
- The prespecified interim analysis performed on the first 452 patients enrolled met the criterion for efficacy and the DSMB recommended stopping; however, 78 additional patients were enrolled between the interim-analysis database lock and the stopping recommendation, so the final safety database included 530 patients.
- Primary composite hemostatic efficacy (≤35% hematoma volume expansion at 12 h, NIHSS increase <7 points at 12 h, and no rescue therapy 3-12 h): 150/224 (67.0%) with andexanet vs 121/228 (53.1%) with usual care; adjusted difference 13.4 percentage points (95% CI 4.6 to 22.2); P=0.003.
- Secondary endpoint (median percent change in anti-factor Xa activity from baseline to 1-2 h nadir): -94.5% (IQR -96.6 to -88.9) with andexanet vs -26.9% (IQR -54.2 to -9.5) with usual care; corresponds to a median reduction of 94.5% vs 26.9%; P<0.001.
- Thrombotic events at 30 days: 27/263 (10.3%) with andexanet vs 15/267 (5.6%) with usual care; difference 4.6 percentage points (95% CI 0.1 to 9.2); P=0.048. Ischemic stroke: 17 (6.5%) vs 4 (1.5%); difference 5.0 pp (95% CI 1.5 to 8.8).
- Death from any cause at 30 days: 73/263 (27.8%) with andexanet vs 68/267 (25.5%) with usual care; adjusted difference 2.5 percentage points (95% CI -5.0 to 10.0); P=0.51.
- mRS 0-3 at 30 days (post hoc analysis): 69/246 (28.0%) with andexanet vs 79/255 (31.0%) with usual care — no functional-outcome benefit.
Study Design
- Study Type
- Multicenter, randomized (1:1), open-label, blinded-endpoint (PROBE) trial
- Randomization
- Yes
- Blinding
- Open-label treatment assignment with independent central adjudication of hemostatic efficacy, thrombotic events, and NIHSS-related outcomes by committee members blinded to group assignments.
- Sample Size
- 530
- Follow-up
- 30 days (primary hemostatic endpoint assessed at 12 hours; safety endpoints assessed at 30 days)
- Centers
- 131
- Countries
- 23 countries
Primary Outcome
Definition: Hemostatic efficacy at 12 hours after randomization — composite requiring ALL of: (1) hematoma volume expansion ≤35% at 12 h vs baseline, (2) NIHSS score increase <7 points at 12 h, and (3) no receipt of rescue therapy (andexanet, PCC, or decompressive surgery) between 3 and 12 hours.
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| 121/228 (53.1%) | 150/224 (67.0%) | - (4.6 to 22.2 (percentage points)) | 0.003 |
Limitations & Criticisms
- Trial stopped early at the prespecified interim analysis when the efficacy criterion for the primary endpoint was met, limiting statistical power for secondary and safety analyses.
- Significantly higher rate of thrombotic events (10.3% vs 5.6%; P=0.048) including ischemic stroke (6.5% vs 1.5%) in the andexanet group is a major safety concern.
- Open-label treatment assignment (though hemostatic efficacy and thrombotic events were adjudicated by a blinded committee).
- Primary composite endpoint was driven mainly by the hematoma volume-expansion component; the two other components (NIHSS change, rescue therapy) did not differ appreciably between groups.
- No difference in 30-day death (27.8% vs 25.5%; P=0.51) or in mRS distribution at 30 days (mRS 0-3: 28.0% vs 31.0%); no functional-outcome benefit demonstrated.
- Trial was designed primarily by the first author and two Portola employees and funded by the manufacturer (Portola → Alexion → AstraZeneca), a potential source of bias despite independent analysis by the Population Health Research Institute.
Citation
N Engl J Med 2024;390:1745-55. DOI: 10.1056/NEJMoa2313040