FormalPara Key Summary Points

Direct oral anticoagulants can treat thromboembolism, but their use is limited due to a dose-dependent increased bleeding risk.

Factor XIa inhibition has shown promise in laboratory and animal studies for disrupting thrombin formation without affecting hemostasis.

Milvexian phase I and II studies have reported dose-dependent efficacy and promising safety data.

Planned phase III trials are underway to examine the utility of milvexian in patients with acute stroke, acute coronary syndrome, and atrial fibrillation.

Introduction

The advent of direct oral anticoagulants (DOACs) introduced a novel medication class capable of prevention and treatment of venous and arterial thromboembolism without the inherent challenges of vitamin K antagonists. However, bleeding has remained a major side effect, and those deemed to be at an increased risk of bleeding often do not receive DOACs or are started on inappropriately low doses [1,2,3]. A need exists for an anticoagulant capable of both adequate anticoagulation and reduced bleeding risk compared with existing oral anticoagulants.

Factors in the intrinsic pathway are vital in the formation of thrombosis but have little role in hemostasis, and these factors have been the target of recent study. Factor XI was specifically identified to be of interest, as individuals with an inherent factor XI deficiency due to genetic mutations were found to have markedly lower rates of incident ischemic stroke and venous thromboembolism (VTE) [4,5,6]; yet without increased major bleeding. Furthermore, higher levels of factor XI were associated with increased risk of VTE [7]. The intriguing prospect of factor XIa inhibitor efficacy resulted in the development of a number of factor XI inhibitors, including small molecules (milvexian, asundexian), monoclonal antibodies (osocimab, abelacimab), antisense oligonucleotides, aptamers, and naturally occurring inhibitors [8]. This review will focus on the physiologic basis, mechanism of action, clinical trial data and results, and ongoing drug development of milvexian in addition to a brief overview of other factor XIa inhibitors in development. This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Anticoagulation Pathway

The coagulation cascade describes the nature of clotting factor interactions that allow for the resolution of spontaneous bleeding and the maintenance of hemostasis (Fig. 1). There are three major pathways in the coagulation cascade: the extrinsic, intrinsic, and common pathways [9]. The extrinsic pathway is activated through vessel injury and initiated through the exposure of tissue factor (TF) to blood, forming a TF and factor VIIa complex. This complex subsequently enters the common pathway and activates factor X, causing thrombus formation and a hemostatic plug to stop bleeding. The intrinsic pathway is activated through contact of factor XII with polyanionic compounds found in injured cells, activated platelets, or pathogens [10, 11]. Factor XII, through activation of factor XI, and subsequently factor XIa, activates factor X and the common pathway.

Fig. 1
figure 1

Overview of the coagulation cascade including intrinsic and extrinsic pathways, convergence into the common pathway, and various effects of anticoagulant and antithrombotic medications including milvexian. TF tissue factor

Full size image

The unique nature of factor XI can be explained through cross-pathway positive feedback (Fig. 2). Regardless of which pathway activated the common pathway, thrombin, a downstream product, reinforces the intrinsic pathway through positive feedback to factor XI [12]. Therefore, if contact between factor XII with polyanionic compounds activates the intrinsic pathway, inhibition of factor XI will interrupt the thrombin–factor XI feedback loop and prevent pathological formation of a thrombus. However, in the event of a bleed, explosive thrombin generation via the TF:factor VIIa complex in the extrinsic pathway leads to hemostasis; whereas feedback activation of factor XI by thrombin is of minor importance for amplification of thrombin generation. In other words, factor XI is dispensable for hemostasis but is central to thrombosis. Accordingly, inhibition of factor XI allows physiological hemostasis while preventing pathological thrombosis.

Fig. 2
figure 2

The mechanisms of hemostatic plug formation versus pathologic thrombi outline the differential role of factor XIa. Left: Outlines a physiological response to vessel trauma in which high tissue factor (TF) concentrations rapidly form a plug, and positive feedback to factor XIa plays only a minor role in maintaining hemostasis. Right: Low TF concentrations initiate thrombin formation. Thrombin combines with polyphosphates (polyP) and neutrophil extracellular traps (NETs) to activate factor XIa and create pathological thrombus buildup

Full size image

Factor XI interaction with platelets further highlights the downstream effects of factor XI in the coagulation pathway. Once platelets are activated, the conformational and cytoplasmic changes that occur foster contact pathway activation [13]. Specifically, the interactions between the platelet surface and factor XIa protect factor XI and enhance conversion to factor IX [14]. The potential but speculative impact on vascular walls is a direct result of clot formation within the blood vessel lumen. Should such clots form within the lumen, the ensuing pathologic inflammatory response and endothelial damage associated with intraluminal thrombus formation occur.

Current oral anticoagulants used in clinical practice are all target components of the common pathway. Warfarin and dabigatran indirectly or directly target thrombin; warfarin, rivaroxaban, and apixaban target factor X [15]. These medications targeting the common pathway decrease pathologic thrombi formation (“bad clots”) while impairing hemostasis during a bleed (“good clots”) [16,17,18,19,20,21]. The need to maximize good clots (hemostasis) while limiting the bad clots (thrombosis) has prompted investigation into factor XI inhibitors as a new class capable of addressing current limitations of anticoagulants.

Drug Characteristics

Milvexian is a small molecular, active-site inhibitor of factor XIa [8]. The chemical formula is C28H23Cl2F2N9O2 with an average weight of 626.45 g/mol [22]. The design of a factor XI inhibitor is more challenging than other coagulation cascade targets, as inhibitors must interact with both substrate and substrate prime (S′) binding pockets, as opposed to thrombin and factor Xa inhibitors that only interact with the former. The larger S′ regions require a large molecular inhibitor that binds to the pocket through polar interactions. Therefore, a balance of potency (more polarity to encourage stronger binding) and optimal pharmacokinetics (less polarity allowing detachment and a clinically appropriate half-life) is required. After numerous iterations, compound 17, the final compound, met criteria for advancement as it included a notably potent binding affinity, increased activated partial thromboplastin time (aPTT), showed promise for high oral bioavailability, and did not increase bleeding in animal models [23].

Milvexian has a high affinity for the active form of factor XIa. The compound’s Ki is equal to 0.11 nM and has an aPTT EC 1.5 of 0.50 uM. The specific pharmacokinetics of milvexian were explored in a subsequent study which found a single dose of the drug, at doses ranging from 4 to 500 mg, reached maximum plasma concentration (TMax) at 3 h and a time to half-life (T1/2) of approximately 8–14 h [8]. The presence of food, defined as a standard high-fat breakfast, both decreased the half-life of the drug by 1.5 h and increased the bioavailability by 1.4-fold to twofold for the 200 mg and 500 mg doses, respectively [8].

The effect of milvexian on aPTT was dose-dependent with a twofold increase in aPTT found at the highest tested dose of 500 mg. The aPTT prolongation peak aligned with the TMax of the drug. Other measures of the coagulation cascade, such as prothrombin time (PT), remained unaffected.

Clinical Trials and Data

Preclinical Studies

Early indications for the unique role factor XI may play in the coagulation cascade were observed in humans with inherited deficiencies. A study of those with normal, mild, and moderate–severe deficiencies in factor XI concluded that the moderate–severe cohort had significantly lower cardiovascular and VTE events [6]. Additionally, in those that had factor XI levels above the 90th percentile, there were significantly more VTE events than in the control group [7]. Analyses that stratified the study population into quartiles of factor XI levels and examined VTE rates showed a dose-dependent relationship between higher factor XI level quartiles and lower VTE incidence.

These findings prompted animal testing of factor Xia inhibition. A study conducted in the early 2000s sought to understand rates of venous thromboembolism in mice with knockout factor XI versus wild type [24]. After inducing vena cava thrombosis with iron chloride (FeCl3), the study concluded that mice without factor XI did not form thrombosis at lower doses of FeCl3, and this protective effect exceeded that of heparin, clopidogrel, and argatroban. Additionally, the study found bleeding risk was similar in knockout mice compared with wild type, providing a murine-tested confirmation for a hypothesized physiologic phenomenon.

Arterial thrombi formation in factor XI-deficient mice has also been studied. By inducing carotid artery damage with FeCl3, researchers found factor XI-deficient mice had significantly lower clotting rates than even those with factor VII deficiency, suggesting the intrinsic pathway plays a more critical role in clot formation than previously considered [25]. A second study, based on a similar carotid artery injury model, aimed to find the difference between factor XI and IX in murine models [26]. The study determined that while both factors were protective against clot formation, factor IX significantly prolonged bleeding time (5.8-fold increase) above the wild-type baseline. These biological studies corroborated the proposed effects of factor XI inhibition, which prompted further research into pharmacologic factor XI inhibitors.

Investigators introduced rabbit models to milvexian and studied the effect on thrombus formation in rabbit arteriovenous (AV) shunts [27]. Significant decreases in thrombus weight and dose-dependent increases in PTT were observed. Additional work examining the hemostatic changes observed with milvexian in rabbit arterial models found that milvexian in combination with aspirin did not increase bleeding time even when compared to monotherapy, with either aspirin or milvexian, alone [28].

Phase I

Phase I trials with milvexian began in 2015 (Table 1). The first trial examined the safety, tolerability, and pharmacokinetics of both single and multiple doses of the drug [29]. This randomized, parallel assignment, double-blinded trial included 104 participants. Participants were assigned to either the single ascending dose (SAD) or multiple ascending dose (MAD) arms with adverse effects monitored for a maximum of 21 days. The drug was generally well tolerated with the most common adverse event being headache and only minor bleeding events reported [8]. The bleeding events included one participant each with epistaxis, gingival bleeding, and petechia. None of the events were considered severe enough to discontinue treatment. The study concluded that SADs up to 500 mg and MADs of up to 200 mg twice daily (BID) were well tolerated. The half-life of the drug ranged from 8 to 14 h in SAD panels and 11–18 h in MAD panels. Little to no renal excretion was observed.

Table 1 Characteristics and key findings from major preclinical, phase I, phase II, and future phase III milvexian trials
Full size table

A second phase I trial began in 2021 to examine the relative oral bioavailability and food effect of both single and multiple doses of milvexian in healthy adults using various modes of drug delivery [30]. This randomized, crossover assignment, open-label trial included 114 participants and concluded enrollment and follow up in 2022. Results of the investigation have yet to be published.

Phase II

Axiomatic-TKR

Two phase II randomized controlled trials (RCTs) designed to evaluate milvexian have been completed. The first, AXIOMATIC-TKR, tested efficacy and adverse outcomes of the drug when compared to subcutaneous enoxaparin [31] in a multicenter international trial that enrolled 1242 participants aged 50 or older with elective unilateral total knee replacement. Participants were randomly assigned in a 1:1:1:1:1:1:2 approach to milvexian 25 mg BID, 50 mg BID, 100 mg BID, 200 mg BID, 50 mg daily, 200 mg daily, or enoxaparin 40 mg daily, respectively. After 252 patients had been enrolled, the independent safety committee recommended discontinuation of the milvexian 25 mg daily dose based on prespecified criteria for low efficacy.

The primary outcome was a composite of postoperative deep-vein thrombosis assessed by unilateral venography 10–14 days after knee replacement. The criteria for proof of efficacy were defined as either a dose–response trend with the twice-daily milvexian dosing or an incidence of VTE significantly lower than 30%, a conservative estimate of postoperative VTE rates in those undergoing TKR. Secondary outcomes included all-cause mortality, nonfatal proximal or distal deep-vein thrombosis, and nonfatal pulmonary embolism. The principal safety outcome was bleeding of any severity.

There was a significant dose-dependent relationship observed between increasing milvexian dosage and decreases in primary outcome events, with VTE rates of 21%, 11%, 9%, and 8% in the 50 mg, 100 mg, 200 mg, and 400 mg doses, respectively. VTE rates in the twice-daily milvexian arm were 12%; therefore, both primary outcome efficacy criteria were met. Additionally, there was no statistically significant increase in bleeding rates between milvexian and enoxaparin or with increasing doses of milvexian. Pharmacodynamic measures were as expected; milvexian increased the aPTT ratio in a dose-dependent manner, and neither of the study drugs increased the PT ratio. Therefore, the study concluded oral milvexian reduced the proportion of postoperative thromboembolism rates without a concurrent increase in bleeding risk when compared with enoxaparin in those undergoing knee arthroplasty.

Limitations of the study include a low number of clinically relevant non-major bleeding (CRNM) and major bleeding events for detection of differences in bleeding rates between milvexian doses and enoxaparin, an open-label design allowing study participants to know which drug they were assigned, and potential investigator bias in ascertaining bleeding rates. However, the dose of milvexian was blinded to study participants, and a blinded event committee reviewed all potential endpoint events.

Axiomatic-SSP

The second phase II trial, AXIOMATIC-SSP, specifically examined safety and efficacy of milvexian versus placebo in those with acute ischemic stroke or high-risk TIA [32,33,34]. This multicenter, international, double-blinded trial enrolled 2366 participants with acute ischemic stroke or TIA and randomly assigned to milvexian 25 mg daily, 25 mg BID, 50 mg BID, 100 mg BID, 200 mg BID, or matching placebo with a 1:1:1:1:1:2 ratio, respectively. Inclusion criteria for stroke and TIA characterization were based on a combination of acute neurological deficits, neuroimaging findings, and National Institutes of Health Stroke Scale (NIHSS) and ABCD [2] scores [35, 36]. All participants were to be treated with low-dose aspirin daily and clopidogrel loading dose following maintenance for 21 days and then aspirin alone for 90 days.

The primary outcome was symptomatic ischemic stroke and covert brain infarction characterized by comparison of magnetic resonance imaging (MRI) at 90 days. Secondary endpoints were the individual components and composite of symptomatic ischemic stroke, myocardial infarction (MI), and all-cause mortality, volume and number of new infarcts detected on MRI, and pharmacokinetic and pharmacodynamic measures. The principal safety outcome was major bleeding. The nature of the bleeding was characterized by both Bleeding Academic Research Consortium (BARC) classification and International Society on Thrombosis and Haemostasis and PLATelet inhibition and patient Outcomes (PLATO) criteria [37,38,39].

There was no difference in event proportion for the primary outcome between any milvexian dose and matching placebo. The primary efficacy outcome occurred in 17% of participants in the placebo group versus 18%, 14%, 15%, and 16% in the 25 mg, 50 mg, 100 mg, and 200 mg twice-daily groups, respectively. There was no significant dose-dependent response between milvexian and the primary efficacy outcome. Major bleeding rates across all trial arms were 1–2%. There was no significant dose-dependent response between milvexian and major bleeding rates. There were fewer symptomatic ischemic strokes at all milvexian doses except for 200 mg twice daily when compared to placebo. The study concluded that in those with ischemic stroke or TIA, oral milvexian did not reduce the composite outcome of brain infarction or ischemic stroke or increase bleeding risk at 90 days when compared with placebo on a background of dual antiplatelet therapy.

Phase III

The results of the phase I and II studies have provided a strong foundation for the LIBREXIA phase III program. The phase III program will include three separate, event-driven trials with common leadership and operations in acute stroke, acute coronary syndrome (ACS), and atrial fibrillation (AF).

LIBREXIA STROKE (NCT05702034), a randomized, parallel assignment, double-blinded study, has an enrollment target of 15,000 participants and an estimated completion date in 2026 [40] (p. 3). Patients with an acute ischemic stroke or high-risk TIA will be randomly assigned to oral milvexian 25 mg twice daily in addition to their standard-of-care antiplatelet therapy or placebo. The primary outcome is time to first occurrence of ischemic stroke. Key secondary outcomes include a composite of cardiovascular death (CVD), myocardial infarction (MI), or ischemic stroke; and a composite of CVD, MI, ischemic stroke, major adverse limb events, symptomatic pulmonary embolism, or deep-vein thrombosis.

LIBREXIA ACS (NCT05754957), a randomized, parallel assignment, double-blinded study, has an enrollment target of 16,000 participants and an estimated completion date in 2026 [41] (p. 3). Patients within 7 days of an ACS, either with or without percutaneous intervention, will be enrolled. Participants will be randomly assigned to oral milvexian 25 mg twice daily in addition to their standard-of-care antiplatelet therapy or placebo. The primary outcome is time to a composite of CVD, MI, or ischemic stroke. Key secondary outcomes include time to a composite of CVD, MI, ischemic stroke, major adverse limb events, or symptomatic venous thromboembolism; a composite of all-cause mortality, MI, or ischemic stroke; CVD; and all-cause mortality.

LIBREXIA AF (NCT05757869), a randomized, parallel assignment, double-blinded study, has an enrollment target of 15,500 participants and an estimated completion date in 2027 [42] (p. 3). Patients with non-valvular AF and appropriate for anticoagulant therapy will be enrolled. The participants will be randomly assigned oral milvexian 100 mg twice daily or apixaban (2.5 or 5 mg) twice daily. The trial is designed as a non-inferiority trial, and the primary outcome is occurrence of stroke or non-CNS systemic embolism. Key secondary outcomes include major bleeding; major or clinically relevant non-major bleeding; CVD, MI, stroke, or non-CNS embolism; CVD; all-cause death, MI, stroke, or non-CNS embolism; and CVD, MI, stroke, acute limb ischemia, or urgent hospitalization for vascular cause.

Discussion

The preclinical, phase I, and phase II trials provide confidence in the pharmacokinetic and pharmacodynamic profiles of milvexian, a clear dose–response curve for clinical efficacy from AXIOMATIC TKR, and no observed increase in bleeding in both phase II trials. The LIBREXIA phase III program will define the safety and efficacy of milvexian in acute stroke, ACS and AF populations.

The design of the LIBREXIA program is unique because three trials will be conducted in parallel rather than sequentially. In contrast, the development of the DOACs including apixaban, rivaroxaban, dabigatran, and edoxaban followed a traditional pathway of studies in orthopedic surgery, atrial fibrillation, acute coronary syndromes, and other patient populations. The investment to start three pivotal Food and Drug Administration (FDA)-regulated trials with nearly 50,000 patients is substantial.

A few attributes from the current data surrounding milvexian usage stand out. Given the half-life of milvexian, a twice-daily dosing lowers the peak-to-trough ratio relative to once-a-day dosing, and most dosing options explored in both AXIOMATIC-TKR and AXIOMATIC-SSP were twice-a-day dosing. AXIOMATIC-TKR supports the dose-dependent relationship of milvexian with the prevention of thrombotic events. However, AXIOMATIC-SSP found that risk of serious adverse events was increased in the 200 mg milvexian twice-daily arm. These findings, along with the reported lack of a dose-dependent relationship observed between drug dosage and bleeding risk and the use of apixaban as the comparator, provide the basis for exploration of 100 mg twice-daily milvexian to reduce bleeding risk while maintaining non-inferiority with regards to thrombotic event reduction compared to apixaban in the upcoming LIBREXIA AF trial. As there was no dose-dependent response for the secondary outcomes of MI and all-cause mortality or symptomatic ischemic stroke with the addition of milvexian to DAPT in AXIOMATIC-SSP, the LIBREXIA STROKE and ACS trials were designed with the 25 mg twice-daily dosing as the treatment arm. These are conceptually considered “safer efficacy” trials, in that the trials test for both non-inferiority of the respective main endpoints and superiority with respect to bleeding endpoints of milvexian versus the comparator arms. A comprehensive review of the rationale behind the LIBREXIA trials dose selection will be discussed separately. Success of the trials and their unique bifaceted conclusions may lead to possible changes in guideline-directed medical therapy.

There are a number of factor XIa inhibitors spanning oral, subcutaneous, and intravenous formulations, in various phases of clinical development. In Table 2, we describe the pharmacokinetic and pharmacodynamic properties of factor XIa inhibitors currently in development as well as relevant randomized control trials.

Table 2 Pharmacologic properties and primary studies of various factor XIa inhibitors in development
Full size table

A few key differences exist between milvexian and asundexian, two oral, factor XIa inhibitors. Pharmacologically, milvexian and asundexian are both daily-dosed factor XIa inhibitors; however, milvexian has both a faster time to peak aPTT prolongation and shorter half-life [43]. Milvexian is eliminated both by the kidneys and liver, and asundexian clearance occurs by first drug metabolism and the renal excretion. The three phase II PACIFIC trials assessed the safety and early efficacy of asundexian in the AF, stroke, and AMI populations and provide broad context for the different use cases between the two medications. In PACIFIC-AF, safety of the oral factor XIa inhibitor asundexian was compared with apixaban in patients with atrial fibrillation. Investigators enrolled 862 participants with AF and an increased bleeding risk to asundexian, a direct inhibitor of factor XIa, or apixaban [44]. The primary endpoint was composite of major or clinically relevant non-major bleeding, and the 20 mg and 50 mg asundexian regimen resulted in a lower bleeding rate compared to apixaban. PACIFIC-Stroke enrolled 1808 patients with an acute, non-cardioembolic ischemic stroke to asundexian or placebo on a background of usual antiplatelet therapy [45]. The primary outcome was the dose–response effect on both MRI-detected brain infarcts and recurrent symptomatic ischemic stroke, and the asundexian regimen did not reduce the composite outcome when compared with background antiplatelet therapy. PACIFIC-AMI randomized 1601 patients with recent MI to asundexian or placebo on a background of usual DAPT [46]. The primary outcome was a composite of CV death, MI, stroke, or stent thrombosis, and the asundexian regimen did not reduce the composite outcome when compared with DAPT. There was less or no change in bleeding rates in the study arm versus current DOAC therapy in the three trials. Following these results, a collection of randomized, phase III trials named the OCEANIC program were initiated. A study to test asundexian to prevent a clot-related stroke in participants after an acute ischemic stroke or high-risk TIA/mini-stroke (OCEANIC-Stroke) is estimated to enroll 9300 participants with acute non-cardioembolic stroke or high-risk TIA to asundexian or placebo on a background of standard antiplatelet therapy. The primary outcome is time to first occurrence of ischemia stroke and first major bleeding incident. The OCEANIC-AF trial enrolled patients with AF and an increased bleeding risk to asundexian or apixaban. The primary outcomes were time to first occurrence of stroke or systemic embolism and time to first occurrence of major bleeding. However, OCEANIC-AF was stopped early due to lack of efficacy as recommended by the Data Monitoring Committee [47]. No additional information has yet been presented or published. The OCEANIC-Stroke trial is still enrolling, and the LIBREXIA phase III trials also continue to recruit participants. There are a few key differences between the OCEANIC-AF and LIBREXIA AF trials, including the phase II program trials and findings, dose selected, and inclusion/exclusion criteria.

The translation of potential pharmacological advantages of factor XIa inhibition compared with current standard of care will require the completion of ongoing clinical outcome trials. The balance of risks and benefits will define the role of factor XIa inhibition compared with standard of care in the patient populations being evaluated. The potential clinical advantages and limitations of milvexian compared with other factor XI/XIa based on pharmacological properties (see Table 2) and completed clinical trials is speculative at this time.

Milvexian provides orally available factor XIa inhibition with a promising phase II clinical trial. The potential for milvexian to reduce thrombotic events without increased risk of bleeding will depend on the results of the ongoing phase III outcome studies. These three studies with nearly 50,000 patients will define the role of milvexian in patients with atrial fibrillation, secondary stroke prevention, and acute coronary syndromes.

Conclusion

Oral milvexian is a factor XIa inhibitor with promising preclinical and phase II results showing the prevention of thromboembolic events without increased bleeding. The results of the large ongoing LIBREXIA program with three concurrent phase III trials will allow researchers to prospectively characterize the safety and efficacy of milvexian in patients with acute stroke, acute coronary syndrome, and atrial fibrillation.