Conclusion: Use of andexanet in patients with ICH while taking a factor Xa inhibitor resulted in better control of haematoma expansion compared to usual care, however the effect on patient-important outcomes such as morbidity and mortality is unclear. There was an increased incidence of thrombotic events associated with use of andexanet.
My take away: The evidence if starting to shift more and more to favouring the use of Andexanet in life-threatening haemorrhage associated with Factor Xa inhibitor use. However, the cost-benefit payoff of this highly expensive medication remains unclear and the risk of increased thrombotic events is troublesome. More data from non-industry-funded studies is needed.
- Intracerebral haemorrhage is one of the most concerning complications of oral anticoagulant use and is associated with poor outcomes
- While the use of oral factor Xa inhibitors (princially rivaroxaban and apixaban) has become common place for the management of conditions such as atrial fibrillation, pulmonary and venous thromboembolism due to their favourable risk-benefit profile and lack of therapeutic monitoring requirements, the reversal of these agents in the setting of life-threatening haemorrhage remains a common problem facing clinicians
- Haematoma expansion has been shown to portend worse outcomes with the INTERACT1 trial (Intensive Blood Pressure Reduction in Acute Cerebral Hemorrhage Trial) reporting a 5% increased risk of death or dependency at 90 days with each 1mL of haematoma
- Rapid reversal of anticoagulation may reduce the risk of haematoma expansion and to date has been attempted principally with the use of prothrombin complex concentrate
- Andexanet alpha is a modified recombinant inactive form of human factor Xa which binds and sequesters factor Xa inhibitor molecules which rapidly reduces anti-factor Xa inhibition and restores thrombin generation
- ANNEXA-4 (single-centre, open-label, single-group study) previously demonstrated use of Adexanet resulted in markedly reduced anti-factor Xa activity (no patient orientated outcomes). A thrombotic rate of 10% was reported in this study
- This was an randomised, open-label trial of 550 patients with acute intracranial haemorrhage and use of a factor Xa inhibitor within the last 15 hours presenting to 131 sites across 23 countries
- Patients were assigned to receive andexanet (either high-dose or low-dose bolus over 15-30mins followed by continuous infusion over 2 hours – dose closen was dependent on the time since last factor Xa inhibitor dose) or usual care determined at physician discretion
- There were 2 protocol amendments over the course of the trial with restriction to patients with an acute ICH as the main bleeding event (not SDH or SAH) and estimated 0.5-60ml blood volume and maximum NIHSS of 35, followed by further time restriction to patients with onset of bleeding symptoms to imaging from 12 to 6 hours. An interim analysis was planned at 450 patients following the first protocol change and the trial was stopped early after this analysis due to reported benefit
- Patients were excluded if GCS < 7, NIHSS > 35, surgery was planned within 12 hours after enrolment or if they had a thrombotic event within 2 weeks of enrolment
- The primary outcome was haemostat efficacy assessed at 12 hours after randomisation, defined as change in haematoma volume of 20% or less (excellent) or 35% or les (good), an increased in NIHSS <7 points at 12 hours and no receipt of rescue therapies such as andexanet, PCC or surgery within 12 hrs
- Secondary outcome was the percentage change from baseline to nadir in anti-Xa activity during the first 2 hours after randomisation
- Groups were well matched at baseline (mean age 78.9, intracerebral haemorrhage 88.4% andexanet group vs 93.9%, apixaban use 62.5% vs 59.2%, rivaroxaban use 28.6% vs 28.5%)
- Results: 452 patients were analysed in a pre-specified interim analysis with 224 assigned to andexanet; haemastatic efficacy occurred in 67% of the andexanet group compared to 53.1% of the usual care group (95% CI 4.6-22.2, p=0.003)
- Most patients who met the criteria for haemastatic efficacy had a haematoma volume expansion of 20% or less (76.7% had expansion of 35% or less) and this appeared to be the predominate factor driving the difference between groups qs there was no significant difference in change in NIHSS or need for rescue therapy
- 85.5% of patients in the usual care group received PCC (median dose 3000iu, IQR 2000-3500)
- The median change in anti-factor Xa activity was -94.5% with andexanet and -26.9% with usual care (p<0.001)
- Thrombotic events occurred in 10.3% of those receiving andexanet and 5.6% receiving usual care (95% CI 0.1-9.2, p=0.048). There were mostly ischaemic stroke (6.55%) and myocardial infarction (4.2%) as opposed to DVT/PE (0.8%).
- Death within 30 days occurred in 27.8% vs 25.5% (-5.0 to 10.0, p=0.051)
- In a post-hoc analysis mRankin scores of 0-3 (favourable neurologic outcome) were 28% in the andexanet group and 31% in the usual care group
- Strengths:
- This is the first randomised trial comparing andexanet to usual care
- The multinational and multi-centre design of the trial adds to external validity
- Limitations:
- This was an manufacturer-funded trial
- The trial was stopped early following a pre-planned interim analysis which tends to favour treatment effect
- Neither the primary or secondary outcomes were patient oriented and so it remains to be seen if there are clinically important benefits from andexanet in terms of morbidity and mortality
- The trial excluded the sickest patients in whom reversing anticoagulation may be the most important (those GCS<7 or requiring urgent neurosurgical intervention)
- Andexanet remains expensive and this trial doesn’t answer the question of whether there is sufficient clinical benefit to justify it’s cost/use, especially in light of an apparent increased thrombosis risk
Reference:
Connolly SJ, Sharma M, Cohen AT et al; ANNEXA-I Investigators. Andexanet for Factor Xa Inhibitor-Associated Acute Intracerebral Hemorrhage. N Engl J Med. 2024 May 16;390(19):1745-1755. doi: 10.1056/NEJMoa2313040. PMID: 38749032.
