FXIa inhibitors vs DOACs: bleeding risk and efficacy

Why the coagulation target matters: intrinsic vs. common pathway

Direct oral anticoagulants (DOACs) like apixaban and rivaroxaban block the common coagulation pathway — the final convergent route where both the intrinsic and extrinsic arms meet — by directly inhibiting Factor Xa, which catalyses the conversion of prothrombin to thrombin. Thrombin is the central enzyme that converts fibrinogen to fibrin, activates platelets via protease-activated receptors (PARs), and provides positive feedback amplification by activating Factors V, VIII, and XI. Because thrombin generation is equally critical for physiological haemostasis — stopping bleeding from any injury anywhere in the body — inhibiting Factor Xa or thrombin inevitably impairs this vital process. This is the fundamental source of the significant bleeding liability that defines DOACs and vitamin K antagonists (VKAs).

The intrinsic — or contact activation — pathway is initiated when blood contacts negatively charged surfaces such as collagen, polyphosphates, or artificial medical surfaces. Its key components include Factor XII (FXII), prekallikrein, high-molecular-weight kininogen, Factor XI (FXI), and Factor IX (FIX). Factor XIa (FXIa), activated primarily by FXIIa, activates FIX to FIXa, which in complex with FVIIIa activates Factor X, feeding into the common pathway. Crucially, the intrinsic pathway is not the only route to clot formation — the extrinsic pathway, driven by Tissue Factor, can independently initiate haemostasis. This distinction is the mechanistic foundation for the entire FXIa inhibitor hypothesis.

The FXIa hypothesis: amplifier inhibition and the haemostasis disconnect

The scientific case for targeting FXIa rests on a critical observation: the contact activation pathway appears more important for pathological thrombosis than for stopping bleeding from everyday injuries. FXIa functions primarily as a major amplifier of the coagulation cascade. Thrombin generated initially via the extrinsic Tissue Factor pathway feeds back to activate FXI, which then generates more FIXa, FXa, and thrombin — a self-reinforcing loop driving explosive pathological clot growth. Inhibiting FXIa dampens this amplification loop without eliminating the initial thrombin burst needed for wound sealing.

This epidemiological observation is the cornerstone of the FXIa inhibitor hypothesis. It suggests that, unlike Factor Xa or thrombin, Factor XIa is not indispensable for physiological haemostasis. Animal model data reinforce this: genetic deletion or pharmacological inhibition of FXI or FXII protects against thrombosis in arterial, venous, and artificial surface-induced models without causing spontaneous bleeding or significantly impairing haemostasis after injury. According to NIH-supported haematology research, the distinction between haemostatic and thrombotic roles of individual coagulation factors is a central focus of next-generation anticoagulant development.

“FXIa inhibitors target an upstream amplifier pathway that appears crucial for pathological clot propagation but less critical for stopping bleeding from typical injuries — offering the potential for dissociation between antithrombotic efficacy and bleeding risk.”

The extrinsic pathway, driven by Tissue Factor and FVIIa, is sufficient to initiate haemostasis at sites of vascular injury. Because this route remains fully functional under FXIa inhibition, patients treated with asundexian or milvexian retain their primary haemostatic mechanism. The intrinsic/contact pathway becomes relevant primarily when blood encounters negatively charged artificial surfaces, atherosclerotic plaque rupture debris, or conditions of stasis — the very scenarios that drive pathological thrombosis in atrial fibrillation, venous thromboembolism, and device-associated clotting. This mechanistic selectivity is what the entire drug class is built upon.

Patent landscape: what Bayer, BMS and J&J are protecting

The patent landscapes for asundexian and milvexian reveal strategic, layered protection reflecting both the commercial stakes and the breadth of potential clinical applications. Core compound patents — including WO2015173161A1 and WO2016044443A1 for milvexian analogs, and WO2018020004A1 and WO2019072830A1 for asundexian — establish foundational IP by claiming specific chemical scaffolds (pyridazinones and triazolopyridazines), their salts, and crystalline forms. As tracked through PatSnap’s patent intelligence platform, these foundational filings were followed by a systematic build-out across multiple IP dimensions.

Formulation and composition patents detail optimised oral formulations — for example, WO2020234320A1 for milvexian — covering bioavailability, stability, and patient compliance. Combination therapy patents are among the most commercially significant filings: WO2021255169A1 protects milvexian combined with antiplatelet agents, and WO2020152320A1 covers asundexian with antiplatelets. These reflect the clinical strategy of deploying FXIa inhibitors on top of existing antiplatelet regimens in post-stroke and post-ACS settings without the additive bleeding burden that would accompany DOAC/antiplatelet combinations. Specific dosing regimens are also claimed, with strategies often targeting lower doses than initial Phase II testing to optimise the efficacy/bleeding window based on emerging pharmacokinetic and pharmacodynamic data.

Therapeutic indication claims span venous thromboembolism (DVT/PE), ischaemic stroke, cardioembolic and cryptogenic stroke, atrial fibrillation, acute coronary syndrome, and complications related to medical devices such as LVADs, catheters, and extracorporeal circuits. Some patents also cover conditions involving contact activation specifically, including heparin-induced thrombocytopenia (HIT). A subset of patents covers biomarker and diagnostic methods — including FXI activity levels and aPTT monitoring — to guide therapy or patient selection, signalling an ambition to position FXIa inhibitors within a precision medicine framework. The breadth of these filings, monitored through PatSnap’s pharmaceutical IP solutions, underscores the industry’s conviction that FXIa inhibition can address therapeutic niches currently inaccessible to DOACs because of bleeding risk.

Clinical trial evidence: bleeding signals and efficacy results

Phase II trials for asundexian and milvexian have provided the first robust human data testing the mechanistic hypothesis, and the bleeding results in particular have been striking. Across multiple trials and patient populations — including post-stroke, post-MI, and VTE settings — FXIa inhibitors consistently demonstrate significantly lower rates of bleeding compared to expectations based on DOACs and historical controls, particularly when added to antiplatelet therapy.

PACIFIC programme (asundexian, Bayer)

PACIFIC-STROKE enrolled 1,807 patients with recent ischaemic stroke and added asundexian (20mg or 50mg once daily) to background antiplatelet therapy versus placebo. Asundexian did not significantly reduce covert brain infarcts — the primary efficacy endpoint — at 6–12 months. However, it demonstrated significantly lower rates of major and clinically relevant non-major (CRNM) bleeding (HR 0.57, 95% CI 0.36–0.91) and any bleeding (HR 0.71, 95% CI 0.58–0.86) for pooled asundexian versus placebo. PACIFIC-AMI enrolled over 4,000 post-MI patients adding asundexian (10mg, 20mg, or 50mg once daily) to dual antiplatelet therapy (DAPT) versus placebo. No significant reduction was observed in the composite efficacy endpoint of cardiovascular death, MI, stroke, or stent thrombosis. However, significantly lower ISTH major or CRNM bleeding (HR 0.83, 95% CI 0.69–1.00) and any bleeding (HR 0.89, 95% CI 0.81–0.98) were observed for pooled asundexian versus placebo.

AXIOMATIC programme (milvexian, BMS/J&J)

AXIOMATIC-SSP enrolled 2,366 patients with recent ischaemic stroke or TIA and tested milvexian at multiple doses (25mg, 50mg, 100mg twice daily; 50mg, 100mg once daily) added to aspirin/clopidogrel versus placebo. Milvexian showed a dose-dependent reduction in the composite efficacy endpoint of ischaemic stroke, covert brain infarct, and MI, with HR 0.78 (95% CI 0.58–1.04) for 100mg twice daily, achieving statistical significance on ordinal analysis. ISTH major bleeding increased dose-dependently (HR 1.65, 95% CI 0.96–2.84 for 100mg twice daily), but absolute rates remained low at 1.1% versus 0.7% for placebo. ISTH CRNM bleeding also increased dose-dependently. Separately, milvexian demonstrated non-inferiority to enoxaparin for VTE prevention after knee replacement, with a trend towards lower bleeding.

The tradeoff in focus: what Phase III must prove

The clinical evidence to date strongly validates the core mechanistic hypothesis: FXIa inhibition significantly reduces bleeding risk — particularly major and intracranial bleeding — compared to common pathway inhibition. The near-absence of any intracranial haemorrhage signal across all reported Phase II trials, even when FXIa inhibitors are combined with antiplatelet agents, represents a potential paradigm shift in anticoagulation safety. According to EMA guidance on anticoagulant benefit-risk assessment, intracranial haemorrhage is weighted as the most serious bleeding outcome in regulatory evaluations. The fact that FXIa inhibitors have not increased this risk in Phase II is therefore clinically meaningful.

The efficacy picture is more nuanced. The neutral results from PACIFIC-STROKE and PACIFIC-AMI raise legitimate questions about achieving sufficient antithrombotic effect in certain settings — particularly when added to potent background antiplatelet therapy in ACS and stroke populations. Dose selection in the PACIFIC programme may have been insufficient for optimal efficacy in those contexts. By contrast, milvexian showed a promising dose-dependent reduction in ischaemic events in AXIOMATIC-SSP, and asundexian demonstrated strong efficacy in VTE prevention and reduction. The biology clearly supports the mechanism’s ability to prevent thrombosis; the clinical question is one of dose, patient population, and endpoint sensitivity.

Phase III programmes are designed to resolve these questions. The OCEANIC programme (Bayer) includes OCEANIC-AF — testing asundexian versus apixaban in atrial fibrillation — alongside OCEANIC-STROKE and OCEANIC-AMI. Notably, OCEANIC-STROKE uses asundexian 50mg, a higher dose than many Phase II arms. The LIBREXIA programme (J&J/BMS) includes LIBREXIA-STROKE (milvexian versus placebo post-stroke/TIA), LIBREXIA-AF (milvexian versus apixaban in AF), and LIBREXIA-VTE. According to ClinicalTrials.gov, these are among the largest anticoagulant trials currently recruiting globally. The head-to-head comparisons against apixaban in AF — the condition where DOACs are most established — will be the most commercially and scientifically decisive readouts.

Patient population selection will be critical to Phase III success. Populations where the intrinsic/contact pathway plays a dominant role — cryptogenic stroke (where contact activation at cardiac surfaces or atherosclerotic plaque may be the predominant mechanism), device-related thrombosis, and VTE — may show stronger efficacy signals than settings dominated by platelet-driven thrombosis where antiplatelet agents already provide substantial coverage. The bleeding safety advantage, if confirmed at scale in Phase III, would also open FXIa inhibitors to populations currently excluded from DOAC therapy: the elderly and frail, patients with prior intracranial haemorrhage, and those with cancer-associated thrombosis where bleeding risk is a dominant concern. The WHO estimates that cardiovascular disease remains the leading cause of death globally, and the need for safer long-term anticoagulation in high-risk populations is substantial. The next few years of Phase III readouts will determine whether asundexian and milvexian fulfil the mechanistic promise that their patents, Phase II data, and biological rationale collectively support.