Factor Xa Inhibitor Brain Bleeds: Reversal After Andexanet

A 74-year-old retired farmer on apixaban for atrial fibrillation arrives with right arm weakness and slurred speech that started 90 minutes ago. His wife watched him take his morning dose at 7 a.m. The noncontrast CT shows an 18 mL left basal ganglia hemorrhage, and his blood pressure is 186/98. The stroke order set still lists andexanet, which has not been sold in the United States since December 22, 2025.

Give four-factor prothrombin complex concentrate (4F-PCC) now, start smooth blood pressure lowering toward a systolic target of 140 mm Hg, kept between 130 and 150 mm Hg, and document the time of the last factor Xa inhibitor dose. Do not wait for an anti-factor Xa level.

Why andexanet left US formularies

ANNEXA-I was the randomized trial that was supposed to confirm andexanet's accelerated approval [1]. It enrolled 530 patients at 131 sites in 23 countries between 2019 and 2023. Eligible patients had an acute intracerebral hemorrhage while taking a factor Xa inhibitor, with the last dose within 15 hours before randomization. After protocol amendments, the hematoma had to be 0.5 to 60 mL, the NIHSS 35 or lower, and the time from symptom onset to baseline imaging 6 hours or less. Patients were excluded for a Glasgow Coma Scale score below 7, surgery planned within 12 hours, or a thrombotic event in the prior 2 weeks [1].

Patients received andexanet or usual care chosen by the local physician. In the 452-patient efficacy population, 85.5% of the usual-care group received prothrombin complex concentrate within 3 hours, at a median dose of 3,000 IU [1].

The primary end point was hemostatic efficacy, a composite the trial defined as hematoma expansion of 35% or less at 12 hours, an NIHSS increase of less than 7 points at 12 hours, and no rescue therapy between 3 and 12 hours. Andexanet achieved it in 67.0% versus 53.1% (adjusted difference 13.4 percentage points; 95% CI, 4.6 to 22.2). Thrombotic events occurred in 10.3% versus 5.6%, and ischemic stroke in 6.5% versus 1.5%. Death within 30 days was 27.8% versus 25.5%, with no appreciable difference in modified Rankin scores. The trial stopped early for efficacy at the interim analysis and was not powered for death or function [1].

On December 18, 2025, the FDA stated that andexanet's risks outweighed its benefits, citing thrombosis of 14.6% versus 6.9% and thrombosis-related death of 2.5% versus 0.9% at 30 days in its review of ANNEXA-I. The manufacturer ended US sales on December 22, 2025 [2].

What the 2026 guideline recommends

The Neurocritical Care Society and Society of Critical Care Medicine published a focused GRADE update on September 24, 2026 [3]. For factor Xa inhibitor-associated intracranial hemorrhage, the panel suggests 4F-PCC rather than andexanet. The recommendation is conditional, with moderate certainty for spontaneous hemorrhage and very low certainty for traumatic hemorrhage. The evidence base was 21 studies and 4,938 patients, with ANNEXA-I as the only randomized trial [3].

The guideline reports the effect separately by hemorrhage type. In spontaneous intraparenchymal hemorrhage, ANNEXA-I showed mortality of RR 1.09 (95% CI, 0.82 to 1.45) and poor function (mRS 4 to 6) of 72.0% versus 69.0% with andexanet. Thrombosis was RR 1.83 in the trial and RR 1.99 across three observational studies [3].

In traumatic hemorrhage, the only adjusted mortality estimate was RR 1.34 (95% CI, 0.67 to 2.69), and thrombosis did not differ (RR 1.14; 95% CI, 0.63 to 2.04). The panel still favored PCC there, because trauma patients already carry high venous thromboembolism risk [3].

The panel treated hemostatic efficacy as a surrogate end point. It weighed that surrogate against the thrombosis signal and against the absence of benefit in death and function, which it rated as the critical outcomes. It also noted that 15% of the ANNEXA-I usual-care arm received no hemostatic treatment at all, which biases the trial toward andexanet [3].

Giving 4F-PCC: criteria and dose

The 2026 update did not resolve PCC dosing. In the included studies, doses ranged from 25 to 50 units/kg. The 2016 guideline suggested 50 units/kg for factor Xa inhibitor-associated hemorrhage within 3 to 5 half-lives of drug exposure, or in liver failure, with low-quality evidence. No 4F-PCC product carries a labeled indication for this use, and some formulations contain heparin [3]. Use your institution's protocol dose, and know which product your pharmacy stocks.

The same 2016 recommendations, which the update did not revisit, call for stopping the anticoagulant. They also call for activated charcoal when ingestion was within 2 hours and aspiration risk is low [3]. ANNEXA-I enrolled only patients dosed within 15 hours, so the trial evidence applies to that window [1].

Routine coagulation tests are not reliable enough to gauge the level of anticoagulation, and specific anti-factor Xa levels are often unavailable or too slow in the emergency setting [4]. The patient's or family's account of the last dose is the clinical data point.

For small hemorrhages, the panel made no recommendation for or against treatment. It described small as generally under 5 to 10 mL depending on location. The factors it lists for that decision are onset-to-imaging delay, location, the specific anticoagulant and last-dose time, deficit severity, and the indication for anticoagulation [3].

Two operational points come from the update. Do not combine andexanet and PCC unless the benefit clearly outweighs the risk. Document exactly what was given and when, because doses given before a transfer are often lost between teams [3].

Who is most likely to expand

A 2026 secondary analysis of ANNEXA-I studied 459 patients with qualifying intracerebral hemorrhage. It defined expansion as growth of 12.5 mL or more, or 35% or more, at 12 hours [5]. Expansion occurred in 32.5%. Shorter onset-to-treatment time, larger baseline volume, higher diastolic pressure, and faster prescan growth rate predicted expansion. None of these predicted thrombotic events [5].

In the usual-care arm, expansion by quartile ran as follows [5]:

• Baseline volume: 27.9% at 3.5 mL or less, versus 53.6% above 22.4 mL.
• Onset to treatment: 48.8% at 3.3 hours or less, versus 24.6% beyond 5.4 hours.
• Prescan growth rate: 20.0% at 1.2 mL/h or less, versus 59.3% above 11.4 mL/h.
• Diastolic pressure: 29.5% at 73 mm Hg or less, versus 51.8% above 95 mm Hg.

The authors used these quartiles to argue for selecting patients for andexanet [5]. That debate continues in countries that still stock the drug. In the United States it does not apply. The predictors still matter here, because they mark the patient who most needs fast PCC and tight blood pressure control.

Blood pressure is the expansion variable you control. For mild to moderate hemorrhage with systolic pressure of 150 to 220 mm Hg, the 2022 AHA/ASA guideline supports lowering systolic pressure to a target of 140 mm Hg and keeping it between 130 and 150 mm Hg, with smooth, sustained titration. It rates an initial drop below 130 mm Hg as potentially harmful [4].

INTERACT3 tested a bundle in 7,036 patients with spontaneous hemorrhage presenting within 6 hours, at 121 hospitals in 10 countries, using a stepped-wedge cluster design [6]. The bundle combined a systolic target under 140 mm Hg, glucose control, treatment of fever, and warfarin reversal within 1 hour. It lowered the odds of poor 6-month function (common OR 0.86; 95% CI, 0.76 to 0.97) and reduced serious adverse events (16.0% versus 20.1%) [6]. The hospitals were in nine low- and middle-income countries and Chile, and none had a consistent hemorrhage protocol before the trial [6]. A department that already runs a protocol should expect a smaller gain.

Key Takeaways

• ANNEXA-I traded 13.4 more patients per 100 with hemostatic efficacy for 6.5% versus 1.5% ischemic stroke, with no difference in death at 30 days.
• The 2026 NCS/SCCM update suggests 4F-PCC over andexanet for both spontaneous and traumatic hemorrhage, with certainty moderate and very low respectively.
• PCC dose is unresolved. The studies used 25 to 50 units/kg, and the 2016 suggestion was 50 units/kg within 3 to 5 half-lives of the last dose.
• Expansion risk with usual care exceeded 50% for hematomas over 22.4 mL, growth over 11.4 mL/h, or presentation within 3.3 hours of onset.

FAQ

Is andexanet still available for apixaban bleeding?

Not in the United States. The FDA concluded that its risks outweighed its benefits, and US commercial sales ended December 22, 2025 [2]. It remains available in some other countries [3].

What dose of 4F-PCC reverses apixaban or rivaroxaban in brain hemorrhage?

No trial has settled the dose. Studies in the 2026 guideline review used 25 to 50 units/kg, and the 2016 guideline suggested 50 units/kg [3]. In ANNEXA-I, the median usual-care dose was 3,000 IU [1].

Should I wait for an anti-factor Xa level before giving PCC?

No. The 2022 AHA/ASA guideline notes these assays are not widely available and often cannot be run fast enough in the emergency setting [4]. Use the reported last-dose time and treat.

Dr. Chet's Take

I have worked as an ED attending for more than 25 years, and I direct TelEmergency for rural Mississippi emergency departments. This piece gets the central call right. In the United States the reversal choice is now settled by availability as much as by evidence, and the evidence leans the same way. The guideline did the harder thing by refusing to let a surrogate end point outrank stroke and death. What I would add is about the first ten minutes. In a small department the PCC lives in pharmacy, and pharmacy is not always staffed at 2 a.m. Someone has to know where the box is, who can mix it, and who documents the weight used. If the order set still names a drug you cannot buy, that order set is a delay waiting to happen.

That being said, I think the field is moving from one overconfidence to another. Andexanet was sold on a lab value. PCC is now winning on a comparison, and nobody has proven that PCC itself changes outcomes against no reversal. The honest answer is that we are giving a product off label at a dose nobody has settled, because it is the least harmful option on the shelf. I still give it, and I give it fast. Where I push back is on treating it as finished business. The patient who arrives at hour 14 with a tiny bleed and a pulmonary embolism diagnosed last month is a different decision than the farmer at hour 2. Those cases need a conversation with neurology, cardiology, and the family, and the chart should show that conversation happened.

If you are the attending, pull up your stroke order set this week and check whether it still points to andexanet. Replace that line with your PCC product, your protocol dose, and the phone number for after-hours pharmacy. Then ask your nurses to record the last anticoagulant dose time in the triage note, in hours and minutes, before the CT is read. When the patient transfers, put the PCC dose, product, and time on the first page of the transfer packet. Run the whole sequence as a drill with pharmacy and the CT technologist, and time the interval from the CT read to the PCC infusion running. Whatever that number turns out to be, it is the one your department owns. The first hour decides most of what the hematoma will do.

— Chester Shermer, MD, FACEP | Emergency Medicine, 25+ Years Clinical Experience | State Surgeon

Continue Your Training

If you're an emergency physician (or any clinician treating patients daily) trying to understand how AI will actually impact your clinical practice — not just the hype — I put together a free practical guide. You can download the AI in EM Survival Guide here.

Browse All Courses at Global MedOps Command

Relevant Reading on Global MedOps Command:

How to Avoid Becoming an AI Casualty — Dr. Shermer's guide to navigating AI tools in clinical and operational settings without compromising judgment or patient outcomes.

Emergency Department Efficiency Playbook — Practical systems for throughput, triage optimization, and operational efficiency.

Read more from Dr. Shermer on Medium

Connect with Dr. Shermer: LinkedIn — Chester "Chet" Shermer, MD, FACEP.

References

1. Connolly SJ, Sharma M, Cohen AT, et al. Andexanet for Factor Xa Inhibitor-Associated Acute Intracerebral Hemorrhage. N Engl J Med. 2024;390(19):1745-1755. PubMed
2. US Food and Drug Administration. Update on the Safety of Andexxa. FDA Safety Communication, December 18, 2025. FDA
3. Vallejo MC, Kennedy LS, Dengler B, et al. Treatment of Antithrombotic-Associated Intracranial Hemorrhage in Adults: A Focused Guideline Update from the Neurocritical Care Society and the Society of Critical Care Medicine. Neurocrit Care. 2026. DOI
4. Greenberg SM, Ziai WC, Cordonnier C, et al. 2022 Guideline for the Management of Patients With Spontaneous Intracerebral Hemorrhage: A Guideline From the American Heart Association/American Stroke Association. Stroke. 2022;53(7):e282-e361. PubMed
5. Shoamanesh A, Connolly SJ, Demchuk AM, et al. Predictors of Hematoma Expansion and Response to Andexanet in Patients With Intracerebral Hemorrhage: Secondary Analyses of the ANNEXA-I Randomized Clinical Trial. Stroke. 2026;57(7):1920-1928. PubMed
6. Ma L, Hu X, Song L, et al. The third Intensive Care Bundle with Blood Pressure Reduction in Acute Cerebral Haemorrhage Trial (INTERACT3): an international, stepped wedge cluster randomised controlled trial. Lancet. 2023;402(10395):27-40. PubMed

Related simulation training: EMS-MedSim · Books by Dr. Shermer