ANNEXA-1: Andexanet Alfa Associated with Harm in DOAC Reversal

Background: In May of 2018, Andexanet alfa gained accelerated approval by the FDA for the reversal direct oral anticoagulants (DOACs) despite a lack of robust evidence for use. The 2022 AHA/ASA guidelines give the drug a level 2A recommendation and recommend it over the use of 4F-PCC (Greenberg 2022). FDA approval alongside guideline endorsement has led to the drug seeing a remarkable growth in use without a single high-quality study to support its use. The available data reports good hemostatic control: a subjective measure that is highly biased by unblinding and selection bias. More importantly, there are no studies comparing andexanet alfa to 4F-PCC or even placebo looking at important, patient-centered outcomes.

Article: Connolly SJ et al. Andexanet for Factor Xa Inhibitor-Associated Acute Intracerebral Hemorrhage (ANNEXA-1). NEJM 2024; 390(19): 1745-55. PMID: 38749032

Clinical Question: Does the use of andexanet alfa in patients on DOACs with intracerebral hemorrhage improved hemostatic efficacy?

Population: Patients > 18 years of age on a factor Xa inhibitor (taken within 15 hours of randomization) with an acute intracerebral hemorrhage.


  • Primary: Hemostatic efficacy assessed at 12 hours after randomization. Hemostatic efficacy was defined as:
    • Excellent hemostatic efficacy: Change in hematoma volume < 20%
    • Good hemostatic efficacy: Change in hematoma volume < 35%
    • Increase in NIHSS < 7 points at 12 hours
    • No receipt of rescue therapies within 3-12 hours from randomization
    • No surgery to decompress the hematoma within 3-12 hours from randomization.
  • Secondary: Percent change from baseline in anti-factor Xa activity during the first 2 hours from randomization
  • Safety Endpoints (assessed at 30 days)
    • Thrombotic events (ischemic stroke, myocardial infarction, VTE).
    • Death

Intervention: Andexanet alfa high-dose or low-dose bolus followed by infusion depending on time and dose from last DOAC use.

Control: Usual care

Design: Non-blinded, randomized controlled trial performed at 131 centers across 23 countries over 4 years.


  • GCS < 7 at the time of consent
  • NIHSS > 35
  • Surgery planned within 12 hours of enrollment
  • Thrombotic event within 2 weeks of enrollment
  • Time from symptom onset > 6 hours
  • Pregnancy


  • Primary results
    • 581 patients were assessed for eligibility across 131 sites over 4 years
      • 31 excluded prior to randomization
      • 20 excluded after randomization due to consent issues
    • 530 analyzed for the safety outcomes
      • 263 patients assigned to andexanet alfa arm
      • 267 patients assigned to usual care arm
    • 452 patients were analyzed for the primary outcome
    • 85.5% (195/228) patients in the usual care arm received 4F-PCC
    • 78.1% (175/224) patients in the andexanet arm received the low-dose regimen

Critical Results

Andexanet alfa

Usual Care

Difference (95% CI)

P Value

Primary Outcome

Hemostatic Efficacy

67% (150/224)

53.1% (121/228)

13.4 (4.6 – 22.2)


NIHSS change < 7 points

87.9% (188/214)

83.0% (181/218)

4.6 (-2.0 – 11.2)

Secondary Outcome

Anti-Factor Xa % Change

-94.5% (-96.6 – 88.9)

-26.9% (-54.2 – -9.5)

Safety Outcome

Thrombotic Events



4.6 (0.1 – 9.2)





Ischemic Stroke



Myocardial Infarction









Arterial Embolism








  • This is the first randomized trial comparing andexanet alfa to standard care in this patient group.
  • Multicenter, multinational study increasing applicability of findings.
  • Outcome assessors were blinded to treatment arm.
  • Hematoma measurements were made with a standard protocol and central site adjudication.
  • 12 hour NIHSS assessments were performed by health care professionals who were unaware of group assignments


  • Study funded, designed, and supervised by AstraZeneca Pharmaceuticals the maker of Andexanet alpha.  Although, this does not refute the findings of this study, it should make readers skeptical.
  • Clinicians were not blinded to the treatment arm patients were randomized to. This may introduce bias particularly in terms of subsequent treatments (treatments outside of reversal are not detailed in the study).
  • Primary endpoint is not patient centered.
  • Convenience sample of patients which introduces bias.
  • There are some baseline differences between groups and it’s hard to say how this may have influenced the results.
  • Exclusion criteria are likely to be difficult for clinicians to assess real time leading to protocol violation (particularly items like planned surgery and recent thrombotic event).
  • Dose adjustment for time from ingestion likely to lead to protocol violation as this info difficult to assess.
  • Exclusion criteria: Removed the sickest patients.


  • The positive primary and secondary outcomes
    • Both the primary (hematoma expansion) and secondary (anti-factor Xa reduction) outcomes were better in the andexanet group.
    • Unfortunately, these are disease-oriented outcomes instead of patient centered outcomes: the patient doesn’t care if their hematoma expands by 20% or 25% or 30%. They care about clinically important outcomes like disability or death.
    • The authors note that in other studies, hematoma expansion has been associated with worse outcomes, but this was clearly not demonstrated in this study as 90d mRS and death were the same between groups.
    • Bottom line is that there wasn’t even a hint of improved clinical outcomes in the andexanet group.
  • Safety outcomes favored the usual care group
    • In general, larger studies or registries of patients are required to determine safety of a treatment.
    • In this study, however, there is a clear signal for harm even with a small group of patients under ideal circumstances (ie enrolled within a study).
    • Though death was not statistically different, the raw numbers favor usual care.
    • Thrombotic events were clearly increased in the andexanet group.
    • Across a larger group of patients outside of the pristine setting of a study, it is likely that we would see an increase in thrombotic events and death.
  • Only 85.5% of patients in the usual care group received 4F-PCC
    • Though there isn’t abundant evidence for the use of 4F-PCC in this setting, it does represent standard practice.
    • The authors do not report about the subgroup of patients who did not receive 4F-PCC and their outcomes.
      • If this data shows worse outcomes with no reversal treatment, it would suggest that usual care with 4F-PCC may be superior to andexanet alfa for clinical outcomes.
      • If this data shows improved outcomes with no reversal treatment, it would suggest that specific reversal agents aren’t necessary.
  • There were multiple protocol changes during the study. Typically, protocols should not be changed while the study is enrolling patients. This is often done to try to steer the data towards benefit.
  • Initial power calculation was for 900 patients to achieve a 90% power to detect and absolute difference of 10% points in terms of hemostatic efficacy but then made an addendum to the protocol to stop after 450 patients.
    • After this stop point, the safety and monitoring board recommended the trial be stopped.
    • Though the authors state they had no knowledge of the effect prior, there is no clear explanation given for this change and it raises the possibility that the trial was stopped prior to additional data showing harm was collected.
  • Drug cost
    • Andexanet alfa costs between $30 – 50,000/treatment. This only takes into account drug costs (ie not monitoring, nursing costs etc).
    • 4F-PCC costs around $5-6,000/treatment.

Author Conclusion: “Among patients with intracerebral hemorrhage who were receiving factor Xa inhibitors, andexanet resulted in better control of hematoma expansion than usual care but was associated with thrombotic events, including ischemic stroke.”

Clinical Take Home Point: The authors conclusions are correct. However, they don’t properly stress the findings.

Treatment of patients with intracerebral hemorrhage on a DOAC with Anexanet alfa did not improve clinical outcomes when compared to usual care. Based on safety data, andexanet alfa resulted in increased harm to patients. Andexanet alfa should not be part of the standard treatment in this scenario based on the available evidence.


  1. Greenberg SM et al. 2022 Guidelines for the Management of Patients with Spontaneous Intracerebral Hemorrhage: A Guideline from the American Heart Association/American Stroke Association. Stroke 2022; 53(7). PMID: 35579034
  2. Connolly SJ et al. Andexanet for Factor Xa Inhibitor-Associated Acute Intracerebral Hemorrhage (ANNEXA-1). NEJM 2024; 390(19): 1745-55. PMID: 38749032

For More Thoughts on This Topic Checkout:

Post Peer Reviewed By: Salim R. Rezaie, MD (Twitter/X: @srrezaie)

Cite this article as: Anand Swaminathan, "ANNEXA-1: Andexanet Alfa Associated with Harm in DOAC Reversal", REBEL EM blog, May 23, 2024. Available at:

Like this article?

Share on Facebook
Share on Twitter
Share on Linkdin
Share via Email

Want to support rebelem?