VE-1902—A direct thrombin inhibitor with reversible covalent mechanism of action shows efficacy with reduced bleeding in rodent models of thrombosis

Abstract

Introduction:

High incidence of bleeding events remains a key risk for patients taking anticoagulants, especially those in need of long-term combination therapy with antiplatelet agents. As a consequence, patients may not receive clinically indicated combination antithrombotic therapy. Here, we report on VE-1902, a member of a novel class of precision oral anticoagulants (PROACs) that combines effective anticoagulation with reduced bleeding in preclinical testing.

Methods and results:

Acting through covalent, reversible active-site modification of thrombin similar to a previously described molecule [1], VE-1902 shows potency and selectivity for thrombin inhibition in human plasma comparable to clinically relevant direct thrombin inhibitors (DTI) such as argatroban and dabigatran (thrombin generation assay ETP EC₅₀ = 1.3 μM compared to 0.36 μM and 0.31 μM for argatroban and dabigatran; > 100-fold selectivity against related serine proteases). Unlike the current anticoagulants, VE-1902 does not significantly inhibit thrombin-mediated platelet activation in in vivo models of thrombosis. In the thrombin generation assay, the compound inhibits thrombin formation without significantly delaying the initiation phase of the clotting cascade. These features are possibly responsible for the observed reduced bleeding in tail bleeding and saphenous vein bleeding models. Consistent with this novel pharmacological profile, VE-1902 shows efficacious anticoagulation in several fibrin-driven animal models of thrombosis (arteriovenous shunt, venous stasis thrombosis, and thrombin-induced thromboembolism models), whereas it does not significantly prevent arterial occlusion in the platelet dependent FeCl₃ model.

Conclusions:

By leaving platelet activation following vascular injury mostly unaffected, VE-1902, and the PROACs more generally, represent a new generation of precision anticoagulants with reduced bleeding risk.

1. Introduction

Thrombotic diseases are the cause of widespread morbidity, mortality, and clinical disability, with myocardial infarction and stroke representing the leading causes of death in the United States [2]. To reduce the risk of such events in cardiovascular patients, anticoagulants and antiplatelet agents are the mainstay of long-term prevention and therapeutic treatments.

Anticoagulants have proven effective for treating and preventing deep vein thrombosis and for reducing cardioembolic risk from clot formation in atrial fibrillation (AF). Oral anticoagulants (e.g., dabigatran, apixaban, rivaroxaban) have largely replaced warfarin for long-term therapy, offering healthcare providers more convenient and safer treatment options. However, bleeding is still the most common treatment-related side effect [3] and bleeding risks prevent the use of anticoagulation in up to a third of patients with AF and stroke risk [4]. Antiplatelet agents (e.g., acetylsalicylic acid (ASA), clopidogrel, ticagrelor) are most commonly used for reducing the risk of events in atherosclerotic disease, where the formation of platelet-rich plaques in arterial vasculature can predispose patients to myocardial infarction and stroke [5]. However, both single and dual anti-platelet therapies are inferior to anticoagulation in preventing stroke in patients with AF [6].

While anticoagulants and antiplatelet agents have different indications for use, both reduce the risk of thromboembolic events by inhibiting mediators of endogenous clotting, thereby conferring some risk of treatment-related bleeding. This risk is even greater for patients receiving combination treatment with both anticoagulation and antiplatelet agents [7]. Notably, an estimated 34% of AF patients also suffer from coronary artery disease (CAD) and would potentially benefit from long-term combined antiplatelet-anticoagulant treatment to reduce their risk of adverse cardiovascular events, provided that the combinations avoid clinically unacceptable bleeding [8]. In addition, anticoagulated AF patients who undergo stent implantation require the addition of antiplatelet agents to inhibit stent and culprit lesion thrombosis for up to 1 year after the procedure [9–11]. The ATLAS ACS-2, TIMI 51 phase 3 trial showed a lower rate of cardiovascular mortality, myocardial infarction, or stroke compared to placebo in ACS patients taking low-dose rivaroxaban in addition to dual antiplatelets but with increased rates of major bleeding, limiting the application of this combined therapy in clinical practice [12]. In the APPRAISE-2 trial, apixaban in combination with ASA resulted in an increased risk of bleeding compared with ASA plus a P2Y12-receptor antagonist, leading to the premature discontinuation of the trial [13]. To mitigate the risk of bleeding, the current European Society of Cardiology guidelines recommend no > 6–12 months of combined anticoagulant-antiplatelet therapy [14], with shorter durations of combined therapy in those at higher perceived bleeding risk. In fact, a recent study (AFIRE) [15] of patients with atrial fibrillation for more than one year on rivaroxaban or on combination therapy with rivaroxaban and an antiplatelet agent showed noninferiority of the monotherapy versus the combination therapy however the safety margin was better with monotherapy.

Furthermore, recent results from the COMPASS trial showed the potential for combining a reduced dose of oral anticoagulant (one-quarter of the full stroke prevention dose of rivaroxaban, i.e., 2.5 mg bd) with antiplatelet therapy (100 mg ASA) to reduce the risk of adverse cardiovascular events in patients with chronic atherosclerosis [16]. These findings suggest a potential benefit of combined anticoagulant-antiplatelet therapy for patients traditionally maintained on antiplatelet agents alone. However, despite the reduced dose of anticoagulant, combination therapy was estimated to cause a 70% relative increase in bleeding. Overall, the significant risk of bleeding complications has limited the uptake of antithrombotic agents even among those at high risk of thrombotic events and stroke [17]. There is therefore still an ongoing effort to find anticoagulants with improved bleeding profile. From epidemiological and animal model data FXI inhibition is thought to have effective anticoagulation with low bleeding risk [18] which could make them useful in combination therapies. A number of large and small molecule FXI/a inhibitors are being pursued in the clinic [19,20].

Here we describe the preclinical profile of VE-1902, the lead clinical candidate in a new class of precision oral anticoagulants (PROACs) with a reversible covalent mechanism of action [1]. Associated with this mechanism of action are two key features: retention of platelet activity and lower bleeding in preclinical models. This profile makes VE-1902 a promising candidate for a new generation of safer anticoagulants. Such agents may be suitable for prolonged co-administration with antiplatelet agents.

2. Materials and methods

2.1. In vitro studies

2.1.1. Fibrinogen cleavage

Inhibition of thrombin in the presence of a native substrate, fibrinogen, was performed as previously described [1]. Briefly, the assay was performed with human platelet-poor plasma (20% final concentration), following addition of the test compound in DMSO, human α-thrombin was added and incubated at room temperature for 30 min. The reaction was quenched and centrifuged to pellet the proteins. The supernatant was then analyzed for quantitation of fibrinopeptide A by HPLC.

2.1.2. Protein C activation and thrombin-activatable fibrinolysis inhibitor (TAFI) activation by thrombin

Thrombomodulin is a cofactor in the thrombin-induced activation of protein C and TAFI [21]. The assays were performed as described previously [1]. Briefly, a thrombin/thrombomodulin complex was first created by premixing α-thrombin with thrombomodulin. This complex was then incubated with compound at room temperature for 10 min. Then, protein C or TAFI was added and incubated at 37 °C for 20 min. Subsequently, thrombin activity was completely inhibited by addition of PPACK, and reporter substrate for activated protein C or TAFI substrate was added and the absorbance at 405 nm or 336 nm, respectively, was recorded after incubation at room temperature for 60 min.

2.1.3. In vitro platelet activation in platelet-rich plasma (PRP)

Upon platelet activation, dense granules release molecules such as ADP, ATP, serotonin, and PF4. The release of ATP by activated platelets can be measured using the Luciferase/Luciferin system by chemiluminescence [22,23]. To this end, PRP was prepared from fresh blood collected in citrate vials by centrifugation at 100× RCF for 10 min at RT. The top layer and as much as possible of the particulate matter on top of the buffy coat were collected. The assay was performed by mixing PRP derived from mouse, rat, or human whole blood (20% PRP) with assay buffer (10 mM Hepes, 150 mM NaCl, 1 mM MgCl₂, 3 mM KCl, pH 7.4), the peptide GPRP (to prevent clotting), DMSO at 2% in buffer with or without test compound, and finally thrombin (10/20/30 nM for mouse/human/and rat thrombin; these concentrations yielded about 90% platelet activation). The reaction was incubated for 2 min, then 10 μL of the reagent ChronoLume (Chrono-log Corp, Havertown, PA) was added and the luminescence read after incubation for another 5 min.

2.1.4. In vitro platelet activation in whole blood

Mouse, human and rat whole blood were collected in citrate vials, and the platelet activation assay was performed with 20% whole blood in assay buffer with GPRP and DMSO (2% in buffer). DMSO was diluted before addition to the blood to prevent hemolysis. For human blood, the assay was also performed at 76% whole blood. 76% whole blood mimics in vivo conditions as closely as possible while 20% allowed comparison to the mouse and rat whole blood assays. Thrombin (final concentration 10 nM for mouse and human, 20 nM for rat) from the respective species was added for a total assay volume of 100 μL. For activation with tissue factor, Dade Innovin (Siemens, Tarrytown, NY) at a final dilution of 1/5000 and calcium chloride at 10 mM (final) were used. For activation with ADP, a final concentration of 10 μM was used. For activation with collagen (rat collagen from Chrono-log Corp, Havertown, PA) at 50 μg/mL was used. Immediately, 2 μL anti-CD61 (APC) and 2 μL anti-CD62P (PE) were added, mixed thoroughly, and incubated for 10 min at RT for thrombin and 15 min for the other agonists. The samples were then diluted 100× in PBS and then immediately read in a flow cytometer (CytoFlex, Beckman Coulter, Brea, CA). The platelet population was identified using the CD61 antibody and the percentage of activated platelets was measured using the CD62P antibody.

2.1.5. Thrombin generation assay (TGA)

The TGA was performed as previously described [1]. Briefly, the assay was performed with human platelet-poor plasma (70% final concentration). The assay also contained phospholipids and Dade Innovin and was initiated by the addition of a solution containing substrate and CaCl₂. The reaction was monitored kinetically by fluorescence at 37 °C for 90 min. The data were analyzed as described [24]. Essentially, the plot of fluorescence intensity versus time is converted to a thrombogram showing the amount of free thrombin produced (in nM) versus time.

2.2. In vivo studies

All in vivo studies were performed in accordance with Verseon Corporation institutional animal care and use committee (IACUC) protocols. Mouse studies were performed with male CD-1 mice and rat studies with male SD rats.

For VE-1902, the IV formulation was N, N-dimethylacetamide (DMA):kolliphor HS15:water in the ratio 5:15:80. The IV formulations used for argatroban, dabigatran, and apixaban were saline, slightly acidified saline, and N,N-dimethylacetamide:propylene glycol:water in the ratio 10:30:60, respectively. We note that the difference in in vivo data for saline and acidified saline is statistically non-significant. Hence, the two vehicles are not explicitly distinguished in the data shown below.

2.2.1. Arteriovenous shunt (AV shunt) rat model of thrombosis [25]

For the arteriovenous (AV) shunt thrombosis model, male SD rats were anesthetized and the common carotid artery and jugular vein of each rat were exposed and cannulated. An external plastic tube (shunt) containing a silk thread soaked in tissue factor was connected to each cannula [26]. After dosing of test article, the shunt was opened and blood was allowed to flow for 15 min, allowing time for thrombus formation. The weight of the thrombus formed on the silk thread was measured at the end of the test.

2.2.2. Platelet status in AV shunt model

The AV shunt model was performed as described above. At different time points throughout the experiment, a few drops of blood from the arterial side of the shunt were collected into EDTA vials. The blood was then analyzed for platelet status by flow cytometry similar to the method described above. The percentage of activated platelets was measured at 1, 5, 8, 10, 12, and 15 min after connecting the shunt. The percentage of activated platelet was then averaged over the time points.

2.2.3. Inferior vena cava (IVC) stasis model in rats [25,27,28]

The inferior vena cava (IVC) is first prepared by closing the side branches from the vena cava and putting in place methods to allow rapid closure of the vena cava itself. The vena cava side branches are ligated between the left renal and iliac veins by tying off the branches with suture silk. A proximal suture is inserted under the vena cava just below the left renal vein. The distal IVC is dissected away from the abdominal aorta and surrounding facia just above the iliac bifurcation to allow placement of a vascular clamp. The test compound is injected by i.v. bolus into the left saphenous vein (via a catheter). One minute following injection of the test compound (to allow distribution), the proximal inferior vena cava is ligated using the suture placed earlier. Thrombus formation is induced by the injection via the left saphenous vein catheter of Thromborel S. 30 s later, the clamp above the iliac bifurcation of the IVC is closed to prevent further blood flow into the vena cava. The resultant thrombus was retrieved after 10 min of stasis and its mass determined.

2.2.4. Thrombin-induced thromboembolism mouse model [29–32]

Mice were anesthetized with a 1:1 mixture of ketamine:xylazine. Test compound was introduced via tail-vein injection, followed by slow injection of human α-thrombin in PBS (final concentration 1.75 U/g mouse body weight) or PBS alone after 30 s. The animal was then observed for 30 min for mortality. Animals that survived the entire 30 min were euthanized by cervical dislocation immediately after the observation period. The results are summarized as percent survival, with animals surviving to the maximal observation time of 30 min marked as survivors.

2.2.5. Platelet status after thrombin induced-thromboembolism [33–35]

To assess platelet status, whole blood samples were drawn 2 min after thrombin/PBS injection and then stained with CD61 (APC). Flow cytometry was then used to count the total number of platelets in the sample. Plots of APC versus side scatter (SSC) were used to measure platelet count. Typically, 100,000 events were measured in these experiments. The ex vivo platelet count in blood following thrombin administration is a known measure of platelet activation in vivo. Injection of agonists such as thrombin, collagen or tissue factor leads to an immediate drop in platelet count which is associated with platelet activation in vivo and adherence to clots and the vasculature.

2.2.6. Mouse iron-chloride-induced arterial thrombosis model [36,37]

Mice were anesthetized by intraperitoneal injection with 300 mg/kg of tribromoethanol and secured in supine position on a Hallowell heated hard pad at 37 °C. Following a midline incision, the carotid artery was surgically isolated from the vagus nerve, jugular vein, and branches. A Doppler flow probe was used to establish baseline blood flow after 3–10 min of equilibration. An intravenous bolus injection of the test article or vehicle was administered and thrombosis was induced by application of a 4 mm × 2 mm piece of filter paper saturated with 3% aqueous ferric chloride onto the adventitial surface of the carotid artery. Occlusion time, defined as the time from the application of the filter paper to the cessation of blood flow, was recorded.

2.2.7. Tail bleeding model in mice [38–40]

In the tail bleeding time test, the tail veins of CD-1 mice were injected with test article, anesthesia was applied (300 mg/kg dose of tribromoethanol given IP), and 2-mm sections of their tail tips were transected and immediately placed into a 15 mL conical vial containing PBS at 37 °C. Bleeding duration and blood cell loss were measured for 25 min.

2.2.8. Saphenous vein bleeding model in mice

In the saphenous vein bleeding time test, the tail veins of CD-1 mice were injected with test article, anesthesia was applied, and the saphenous vein was exposed and pierced. Blood was wicked away until hemostasis occurred. The clot was then gently mechanically disrupted, and the process was repeated. The time to hemostasis was recorded over a 30-min time interval.

3. Results

3.1. In vitro studies

VE-1902 is a potent inhibitor of thrombin (IC₅₀ = 4 nM) with typically > 1000-fold selectivity against related serine proteases in in vitro biochemical assays (see Supplemental materials, Potency and Selectivity). Like other PROACs, VE-1902 inhibits thrombin in a reversible, covalent manner (see Supplemental materials, Release of inactive carrier moiety; Supplemental materials, Kinetics). The inactivation rate of VE-1902 is k_inact = 0.58 ± 0.05 s⁻¹ which translates to a half-life of t_½ = 1.2 s and the inhibition constant is K_i = 1.3 ± 0.2 μM. The inhibition is reversible with a half-life of 4.5 h.

3.1.1. Inhibition of the thrombin/thrombomodulin complex and fibrinogen cleavage by thrombin

The ability of VE-1902 to inhibit the activity of the thrombin/thrombomodulin complex acting on its endogenous substrates Protein C and TAFI is shown in Table 1. In this assay VE-1902 is preincubated with the thrombin/thrombomodulin complex followed by addition of Protein C and TAFI. VE-1902’s low-nanomolar IC₅₀ against protein C and TAFI activation indicates that it is a potent inhibitor of the thrombin/thrombomodulin complex (see Supplemental material, Potency and Selectivity). Dabigatran and argatroban too are potent inhibitors of the thrombin/thrombomodulin complex.

Table 1.

IC₅₀ values for VE-1902, dabigatran, and argatroban against the endogenous substrates of thrombin. Values shown are mean ± SD.

(μM)	Fibrinogen cleavage	Activation of protein C	Activation of TAFI
VE-1902	1.6 ± 0.9 (n = 8)	0.005 ± 0.002 (n = 2)	0.014 ± 0.006 (n = 3)
Dabigatran	0.13 ± 0.05 (n = 6)	0.005 ± 0.003 (n = 4)	0.22 ± 0.04 (n = 3)
Argatroban	0.4 ± 0.2 (n = 6)	0.008 ± 0.001 (n = 3)	0.19 ± 0.03 (n = 3)

We tested the potency of VE-1902 against cleavage of fibrinogen by thrombin. To be physiologically relevant the assay was performed in plasma without preincubating VE-1902 with thrombin prior to exposure to plasma. The weaker activity of VE-1902 against fibrinogen cleavage (IC50 = 1.6 μM) is due to the lack of preincubation and exposure to high concentration of plasma proteins.

3.1.2. Inhibition of platelet activation

Inhibition of platelet activation is monitored by adding the agonist (thrombin or tissue factor) to a solution of plasma or blood containing the test compound. Argatroban, dabigatran, and rivaroxaban have previously been shown to be potent inhibitors of thrombin-mediated platelet activation [41,42]. Following stimulation with either thrombin or tissue factor (TF), the platelet inhibition EC₅₀s for these anticoagulants ranged from 7 nM to 50 nM (Table 2). In contrast, VE-1902 was a less potent inhibitor of platelet activation, requiring > 100-fold concentration to achieve an equivalent level of inhibition across almost all conditions tested especially in the assays using the more physiological relevant agonist tissue factor. As a selective direct thrombin inhibitor, VE-1902 did not inhibit platelet activation by the agonists ADP or collagen up to 50 μM.

Table 2.

EC₅₀ values for the inhibition of thrombin- or TF-mediated platelet activation in plasma or whole blood of different species. 20% plasma or whole blood concentration was used unless stated otherwise. Values shown are mean ± SD.

(μM)	VE-1902	Dabigatran	Argatroban	Rivaroxaban
Mouse plasma—Thrombin	15 ± 2 (n = 5)	0.007 ± 0.003 (n = 4)	0.007 ± 0.005 (n = 45)	NA
Rat plasma—Thrombin	4.7 ± 0.8 (n = 2)	0.033 ± 0.02 (n = 2)	0.024 ± 0.01 (n = 4)	NA
Human plasma—Thrombin	3.7 ± 1.7 (n = 9)	0.031 ± 0.02 (n = 4)	0.033 ± 0.02 (n = 4)	NA
Mouse blood—Thrombin	6.4 ± 0.02 (n = 2)	0.021 ± 0.01 (n = 2)	0.047 ± 0.02 (n = 7)	NA
Rat blood—Thrombin	5.5 ± 3.1 (n = 5)	0.027 ± 0.01 (n = 4)	0.018 ± 0.004 (n = 3)	NA
Human blood—Thrombin	3.2 ± 0.6 (n = 8)	0.035 ± 0.01 (n = 5)	0.049 ± 0.018 (n = 5)	NA
Human blood (76%)—Thrombin	5.19 ± 0.15 (n = 7)	0.014 ± 0.001 (n = 2)	0.017 ± 0.01 (n = 4)	NA
Human blood—TF	6.22 ± 1.1 (n = 3)	0.023 ± 0.006 (n = 3)	0.022 ± 0.006 (n = 4)	0.031 ± 0.014 (n = 3)
Human blood (76%)—TF	8.02 ± 1.3 (n = 3)	0.029 ± 0.003 (n = 2)	0.024 ± 0.01 (n = 6)	0.012 ± 0.003 (n = 4)

3.1.3. In vitro functional efficacy—thrombin generation assay (TGA)

The TGA [1] measures the ability of the coagulation cascade to generate thrombin under conditions similar to the physiological conditions involved in blood clotting. It proceeds by an initial lag (initiation) phase followed by clotting and the rapid generation of thrombin (propagation phase).

TGA results for VE-1902, the DTIs dabigatran and argatroban, and the FXa inhibitors apixaban and rivaroxaban are presented in Fig. 1 and Table 3. The inhibition of thrombin generation by VE-1902 and the established DTIs was reflected by decreased peak amplitude in the thrombograms with increasing inhibitor concentration.

Fig. 1.

Thrombin generation assay. Thrombin activity, as measured by chromogenic substrate, was measured over time following the addition of tissue factor and CaCl₂ to platelet-poor plasma. While all tested anticoagulants showed dose-dependent inhibition of thrombin (decreased peak heights; C_max), only VE-1902 did not significantly increase the lag time (T_max EC_2x).

Table 3.

Thrombin generation assay parameters. Values shown are mean ± SD.

(μM)	VE-1902 (n = 33)	Apixaban (n = 6)	Argatroban (n = 1400)	Dabigatran (n = 11)	Rivaroxaban (n = 5)
ETP EC50	1.3 ± 0.3	0.47 ± 0.01	0.36 ± 0.162	0.31 ± 0.24	0.31 ± 0.14
Cmax EC50	0.8 ± 0.2	0.1 ± 0.1	0.35 ± 0.14	0.45 ± 0.45	0.1 ± 0.1
Tmax EC2x	34 ± 13	0.07 ± 0.04	0.454 ± 0.24	0.24 ± 0.24	0.1 ± 0.03

VE-1902 differed from the other tested anticoagulants in that it showed a very high T_max EC_2x (concentration of inhibitor required to double the time at which C_max occurs) while the comparators showed large, concentration-dependent increases in TGA lag time, as previously reported [43,44]. For instance, the T_max EC_2x for dabigatran was 0.24 μM compared to 34 μM for VE-1902, a > 100-fold difference, see Table 3. Similar TGA results were observed using platelet rich plasma (data not shown).

3.2. In vivo thrombosis studies

3.2.1. Rat AV shunt model of arterial thrombosis

In this model, administration of VE-1902 resulted in a dose-dependent reduction in thrombus size indicative of in vivo anticoagulant efficacy (Fig. 2, left). VE-1902 showed an ED₅₀ of 2.9 mg/kg and an ED₉₀ of 10 mg/kg after IV administration. Dabigatran showed an EC₅₀ of 0.01 mg/kg (four point dose response with n = 6 at each dose; see supplemental - Rat AV shunt model of thrombosis) compared to ~0.04 mg/kg [45], rivaroxaban showed an EC50 of 0.1 mg/kg (five point dose response with n = 6 for the lowest dose and n = 3 for other doses; see supplemental - Rat AV shunt model of thrombosis) compared to literature value of 0.3 mg/kg [46] and apixaban inhibited about 90% of thrombus formation at 3 mg/kg infusion (n = 6; see supplemental – Apixaban AV Shunt and Rat IVC Stasis models) similar to literature value (~80% inhibition at 3 mg/kg/infusion) [47].

Fig. 2.

Dose-response of VE-1902 in the AV shunt model (left) and IVC stasis model (right) in rats. Thrombus weight is shown as a function of dose after IV dosing (n = 12 for AV shunt; n = 2, 4, or 6 for IVC stasis).

3.2.2. Rat IVC stasis model of venous thrombosis

In this model, the extent of clotting was measured by recovering and weighing the ensuing clot after 10 min of stasis. Inhibition of clotting by anticoagulants was compared by administration of the test compounds prior to the initiation of stasis. For model validation, dabigatran was tested and showed an ED₅₀ of 0.01 mg/kg (four point dose with n = 4 at each dose) after IV administration, as compared to the value published by the manufacturer of 0.033 mg/kg [28]. Further, Apixaban at 3 mg/kg/infusion (n = 6; see supplemental – Apixaban AV Shunt and Rat IVC Stasis models) showed 90% inhibition of thrombus formation compared to about 60% inhibition seen by Shumacher et al. 2010 [47]. Results for VE-1902 in the IVC stasis model showed strong concordance with those in the AV shunt model, (IV ED₅₀ = 2.43 mg/kg, cf. Fig. 2, right). Additionally, a dose-response relationship of VE-1902 across more than two orders of magnitude was observed.

3.2.3. Mouse thrombin-induced thromboembolism model

While the AV shunt and IVC stasis models focus on quantitative measurements of the formation of focal, initiated clots in specific vasculature, the thrombin-induced thromboembolism model replicates a systemic thrombosis in vivo. Clotting was initiated by injection of human α-thrombin directly into the circulatory system of mice via the tail vein. Untreated mice in the saline and vehicle groups experienced 100% mortality within 15 min due to pulmonary embolism. VE-1902 dosed IV at 10 mg/kg was able to prevent mortality in 88% of the treated mice. Argatroban dosed at 4.5 mg/kg IV was able to prevent mortality in 91% of treated mice (n = 12 for each treatment group).

3.2.4. Mouse iron-chloride-induced thrombosis model

In the FeCl₃-induced thrombosis model, VE-1902 increased average occlusion time (13.1 ± 8.4 min) compared to vehicle (approx. 4 min) while dabigatran (0.2 mg/kg, n = 10), argatroban (4 mg/kg, n = 6), and apixaban (6 mg/kg infusion, n = 4) increased the occlusion time to > 30 min. However, for VE-1902 in five out of six measurements, blood flow ceased within 30 min, whereas both DTI controls extended occlusion time beyond 30 min for all animals. In fact, the left panel of Fig. 3 shows an example in which cessation of blood flow did not even occur within 60 min for dabigatran. Hence, while VE-1902 extends occlusion time, it is not an effective inhibitor of clot formation in the ferric chloride thrombosis model.

Fig. 3.

Blood flow over the course of the ferricchloride-induced thrombosis model. Representative measurements for dabigatran (left) and VE-1902 (right) are shown. After the baseline was established, mice were treated with either vehicle or test compound. The timepoint at which thrombosis was initiated by applying filter paper soaked in ferric chloride solution on the carotid artery is indicated by an arrow. Occlusion was then measured by cessation of blood flow.

3.3. In vivo bleeding studies

3.3.1. Mouse tail-bleeding model

In the tail-bleeding test, two metrics were assessed: the time until bleeding cessation and the total number of blood cells lost. Increased bleeding time and overall blood loss in the mouse tail-bleeding model are preclinical predictors of potential bleeding risk for anticoagulants.

Doses for the tail transection tests were chosen based on efficacy (ED₉₀) from in-house data in the rat IVC stasis and AV shunt models for VE-1902, argatroban, and dabigatran and from the literature for apixaban [47]. To translate the rat efficacious doses to doses appropriate for administration in mice, simple allometric scaling was applied to the ED₉₀s by multiplying by two, leading to the following IV efficacious doses in mice: 4 mg/kg for argatroban, 0.2 mg/kg for dabigatran, 6 mg/kg infusion for apixaban, and 20 mg/kg for VE-1902.

IV administration resulted in a statistically significant (p < 0.0001) extension of bleeding time compared to vehicle controls for all tested anticoagulants except VE-1902, see Fig. 4, left. In contrast, VE-1902 caused no statistically significant prolongation of bleeding time compared to vehicle.

Fig. 4.

Tail-bleeding test. Left panel: Time to bleeding cessation for argatroban (n = 12), dabigatran (n = 15), apixaban (n = 8), VE-1902 (n = 10), and their respective vehicles. Right panel: Total blood cell loss for the same compounds and vehicles. Significance was determined using one-way ANOVA (p < 0.0001 for both datasets) followed by post hoc tests, as indicated, using the Bonferroni method. *: p < 0.1, ***: p < 0.001; ****: p < 0.0001, ns: not statistically significant. Bars and error bars represent mean ± SEM.

Administration of the DTI controls also led to a statistically significant increase in total blood cells lost (p < 0.1 for dabigatran, p < 0.001 for argatroban, and p < 0.0001 for apixaban). In contrast, administration of VE-1902 did not cause a statistically significant difference in total blood cells lost compared to vehicle, see Fig. 4, right.

3.3.2. Mouse saphenous vein bleeding model

In this test, two correlated metrics were measured: the average time to bleeding cessation and the number of times bleeding stopped (disruptions) during the observation period of 30 min. Increased bleeding times and decreased number of disruptions in the saphenous vein bleeding model are preclinical predictors of potential bleeding risk for anticoagulants.

All compounds were administered IV at the same doses as described above for the tail-bleeding model. Administration of argatroban or apixaban resulted in a statistically significant increase in the average bleed time compared to their vehicle controls (p < 0.1 and p < 0.0001, respectively). Administration of dabigatran led to a modest observed increase in bleeding time, which, however, was not statistically significant. VE-1902 did not substantially increase bleeding time, see Fig. 5, left.

Fig. 5.

Saphenous vein bleeding model. Left panel: Time to bleeding cessation for argatroban (n = 12), dabigatran (n = 11), apixaban (n = 12), VE-1902 (n = 12), and corresponding vehicles. Right panel: Number of disruptions required to restart bleeding during the 30 min duration of the experiment. Significance was determined using one-way ANOVA (p < 0.0001 for both datasets) followed by post hoc tests, as indicated, using the Bonferroni method. *: p < 0.1, ***: p < 0.001, ****: p < 0.0001, ns: not statistically significant. Bars and error bars represent mean ± SEM.

Furthermore, administration of VE-1902 did not increase the number of bleeding disruptions compared to vehicle. In contrast, the administration of argatroban, dabigatran, or apixaban resulted in a statistically significant decrease in the number of bleeding disruptions compared to their vehicle controls (p < 0.0001, p < 0.0001, and p < 0.001, respectively), see Fig. 5, right.

3.4. In vivo platelet activation status studies

3.4.1. Platelet status in mouse thrombin-induced thromboembolism model

To test the hypothesis that the lower bleeding observed in the in vivo models discussed above is due to VE-1902’s ability to preserve platelet activity at its efficacious anticoagulant concentration (approx. 90% survival), in vivo platelet activation was monitored in the thrombin-induced thromboembolism model at the efficacious doses as described above.

Injection of thrombin in mice induces in vivo platelet activation. When activated, platelets adhere to the vasculature, adhere, and form a thrombus in the animal. This results in platelets being retained in the animal and not being released in the extracted blood [34]. Thrombin-induced clotting in animals not treated with an anticoagulant hence results in lower platelet counts while administration of an anticoagulant that inhibits thrombin-mediated platelet activation results in higher platelet counts. Fewer platelets in the extracted blood is, therefore, an indication of in vivo activation.

Flow cytometry was used to measure platelet count, see Supplemental material, Platelet Count Method and Fig. 6 for a quantification. Administration of either argatroban or VE-1902 without thrombin challenge did not affect the platelet count compared to their vehicle controls (data not shown). Administration of argatroban prior to thrombin challenge did not affect platelet count compared to nothrombin saline control. Administration of VE-1902 or vehicle control preceding thrombin challenge, however, reduced the platelet count to approximately the same levels, indicating that VE-1902 did not inhibit the activation of platelets at a dose that prevents mortality in the thrombin-induced thromboembolism model.

Fig. 6.

Quantification of platelet count in the thrombin-induced thromboembolism model. The percentage of platelets in mouse whole blood immediately after treatment with argatroban or VE-1902 in the thrombin-induced thromboembolism model is shown for different treatment scenarios. Significance was determined using one-way ANOVA (p < 0.0001, n = 18 for each treatment) followed by post hoc tests, as indicated, using the Bonferroni method. **: p < 0.01, ****: p < 0.0001, ns: not statistically significant. Bars and error bars represent mean ± SEM.

3.4.2. Platelet status in rat AV shunt model

In order to further validate the platelet-activity preserving effect of VE-1902 in vivo, the status of platelet activation was determined in blood drawn from rats at different timepoints during the AV shunt thrombosis test. Due to high variability, percent platelet activation was then averaged over all timepoints as well as over multiple animals. Administered IV doses for this study were 0.1 mg/kg, 2 mg/kg, and 10 mg/kg for dabigatran, argatroban, and VE-1902, respectively, based on the respective ED₉₀s in the AV shunt model.

For saline-treated animals (the vehicle for argatroban and dabigatran) and those treated with VE-1902 vehicle, approximately 20% of platelets were found to be activated (Fig. 7). Administration of dabigatran and argatroban significantly (p < 0.01 and p < 0.0001, respectively) reduced platelet activation to < 10%. Administration of VE-1902, however, did not significantly reduce platelet activation, showing a reduced impact of VE-1902 on platelet function compared to the controls, despite effective thrombus reduction.

Fig. 7.

Platelet activation status in the AV shunt model. Blood was drawn at several timepoints throughout the AV shunt model and platelet status was determined after treatment with either argatroban (n = 35), dabigatran (n = 6), VE-1902 (n = 20), or the corresponding vehicles. The average percentage of activated platelets over the duration of the experiment is shown. Significance was determined using one-way ANOVA (p < 0.0001) followed by post hoc tests, as indicated, using the Bonferroni method. **: p < 0.01, ****: p < 0.0001, ns: not statistically significant. Bars and error bars represent mean ± SEM.

4. Discussion

We have developed a new class of precision oral anticoagulants (PROACs) that are potent and selective inhibitors of thrombin. PROACs avoid the inhibition of thrombin-induced platelet activity seen with current anticoagulants and hence have the potential to reduce bleeding. In this work, we have described the in vitro and in vivo pharmacology of the lead clinical candidate, VE-1902, which showed efficacy in preclinical models and lower in vivo bleeding than current anticoagulants.

It was previously demonstrated that PROACs inhibit thrombin by acylating the active-site serine in a reversible covalent manner [1]. VE-1902, like the previously described PROAC, showed a slow time dependent on-rate for thrombin inhibition with a half-life of 1.2 s. Once covalently bound and inhibited, inhibition is reversible with a half-life of 4.5 h (see Supplemental materials, Kinetics). However, the reversibility may not be physiologically relevant as thrombin is inactivated by antithrombin in a matter of minutes.

VE-1902 is a low nanomolar inhibitor against thrombin/thrombomodulin acting on its physiological substrates (protein C, and TAFI) indicating that VE-1902 is as effective at inhibiting thrombin or the thrombin/thrombomodulin complex similar to the other known DTI’s arhatroban and dabigatran. VE-1902 was assayed against the cleavage of fibrinogen in plasma with thrombin without preincubation in order to keep it physiologically relevant. VE-1902 showed an IC50 of 1.6 μM in this assay which is weaker than argatroban or dabigatran never the less it is an effective inhibitor of fibrinogen cleavage.

The Thrombin Generation Assay (TGA) is possibly the most physiologically relevant assay for measuring inhibition of the coagulation cascade. VE-1902 is potent in this assay as seen by the ETP EC50 of 1.3 μM which is similar to the other know anticoagulants. A difference between VE-1902 and the competitive, noncovalent inhibitors argatroban, dabigatran, apixaban, and rivaroxaban was the timing of peak thrombin activity in the TGA. Toward the end of the lag phase, a small amount of thrombin was generated and fibrinogen clot formation was initiated [48]. The presence of the competitive, noncovalent inhibitors resulted in a dose-dependent increase in lag time before the onset of thrombin generation. In contrast, lag time was virtually unchanged for VE-1902, except well above the ETP EC₅₀. This may allow coagulation to start promptly without affecting the initiation phase of thrombin generation, after which VE-1902 potently limited thrombin production.

This distinctive profile can likely be attributed to the relatively slow inactivation time of thrombin by VE-1902 compared to the rapid inactivation by noncovalent DTIs. The weak inhibition of the TGA initiation phase by VE-1902 may also explain the very weak activity in the clotting assays PT and aPTT (see Supplemental material, PT and aPTT), because these assays essentially measure initial clot formation but not subsequent thrombin activity and clot strengthening.

The key difference between VE-1902 and other anticoagulants is the markedly lower impact on thrombin-mediated platelet activation even at efficacious doses. Under all conditions tested, VE-1902 was found to be a significantly weaker in vitro inhibitor of platelet activation than the standard of care anticoagulants, while showing similar potency against other activities of thrombin, particularly fibrinogen cleavage and the inhibition of thrombin growth in the TGA. Like the other DTIs and FXa inhibitors, VE-1902 did not inhibit platelet activation by ADP or collagen [28,49,50].

VE-1902’s weaker activity against thrombin-mediated platelet activation in vitro was also confirmed in vivo at efficacious doses. In the thrombin-induced thromboembolism model, administration of argatroban prior to thrombin challenge kept platelet count comparable to vehicle without thrombin, indicating that argatroban inhibits the activation of platelets. In contrast, administration of VE-1902 preceding thrombin challenge reduced platelet count by similar amounts to vehicle after thrombin challenge, showing that VE-1902 did not inhibit the activation of platelets in vivo. These data indicate that VE-1902 has an antithrombotic effect sufficient to reduce thromboembolism while leaving platelet function largely unaffected. Similarly, VE-1902 showed excellent efficacy on par with standard-of-care anticoagulants in the AV shunt and IVC models, two widely accepted models of arterial and venous thrombosis [25]. However, VE-1902 did not inhibit thrombin-mediated platelet activation despite reducing thrombus size by 90% in the AV shunt model. These results indicate that VE-1902 is potent against a spectrum of physiologically observed thrombi while preserving thrombin-mediated platelet activity.

Importantly, VE-1902 did not effectively prevent arterial occlusion at 10 mg/kg in the mouse FeCl₃ model of arterial thrombosis. This result is not unexpected, as thrombi resulting from FeCl₃ treatment are rich in platelets [37,51] and VE-1902 is only a weak inhibitor of platelet activation.

The difference between VE-1902 and other anticoagulants with respect to platelet activation may also result in the observed superiority of VE-1902 in the bleeding tests discussed below. Based on the in vitro and in vivo data discussed here, VE-1902 is considered a ‘precision’ anticoagulant that effectively inhibits fibrinogen cleavage but does not disrupt thrombin-mediated platelet activation.

The sparing of thrombin-mediated platelet activation by VE-1902 may have important physiological implications. Upon vascular injury, platelets form a critical component of the hemostatic clot [52,53], with thrombin-mediated PAR activation being one of the most potent platelet activators. In human platelets, thrombin cleaves PAR-1 and PAR-4, with PAR-1 appearing to be the dominant pathway since antagonists of PAR-1 signaling show favorable anticoagulant properties [54]. As highlighted by the in vitro and in vivo results presented here, VE-1902 spares this key platelet activation pathway.

The slow thrombin inhibition kinetics of VE-1902 provide a possible explanation for the weak inhibition of both platelet activation and clot initiation during the initial phase of the coagulation cascade, while allowing the compound to inhibit subsequent thrombin generation. Platelet activation is a rapid process and platelet receptors PAR1 and PAR4 can be activated with a small amount of active thrombin. Thus an inhibitor that allows a small amount of thrombin generation in the initiation phase will not prevent platelet activation. It will however prevent fibrin clot formation since it effectively inhibits the exponential thrombin production as seen in the thrombin generation assay.

In particular, the weak inhibition of thrombin-mediated platelet activation by VE-1902 suggests a possible mechanism for the observed lower bleeding in two rodent models wherein the activated platelets form a ‘platelet plug’ that restricts blood flow. In the tail bleed experiment, VE-1902 at 20 mg/kg showed a more desirable profile compared to argatroban, dabigatran, and apixaban with no statistically significant extension of bleeding time or increase in blood cell loss compared to vehicle. Similarly, in the saphenous vein bleeding model, VE-1902 was found to be non-inferior to dabigatran and superior to argatroban and apixaban with respect to average bleeding time, and superior to all three tested standard-of-care anticoagulants with respect to number of bleeding disruptions.

A limitation of this study is that a direct causal link between the weak platelet inhibitory effect of VE-1902 and the lower bleeding has not been clearly established. This would require testing the bleeding effects of a series of related compounds with different platelet inhibitory potentials but with the same reversible covalent mechanism of action. This is clearly beyond the scope of this paper. An additional limitation is that all of the in vivo studies were performed using male rodents. Previous studies have reported higher susceptibility to thrombosis in male mice than females [55]. Hence applicability of the results to female animals is unknown.

Antiplatelet agents, particularly in combination with anticoagulants, have also shown benefits for patients predisposed to platelet-rich thrombosis, for example after myocardial infarction and stent implantation. However, these treatments are also associated with major bleeding events [56]. Similarly, a number of studies have shown some benefit in preventing vascular events in stroke patients, but there was also an associated risk of life-threatening major bleeds [5,57,58]. Phase 3 clinical trials studying apixaban or rivaroxaban in combination with dual antiplatelet agents in the APPRAISE-2 and ATLAS ACS 2-TIMI 51 trials showed dose-dependent increases in minor and major bleeding events, including life-threatening intracranial bleeding [59,60]. For patients with atherosclerotic disease, the COMPASS trial suggested that combined anticoagulant-antiplatelet treatment (low-dose rivaroxaban, 2.5 mg bd, and ASA) may lead to better cardiovascular outcomes [16]. However, although rivaroxaban was dosed at just a quarter of the full stroke-prevention dose in AF, this trial reported a significantly increased incidence of bleeding events compared to ASA alone.

There is large-scale observational evidence that anticoagulant use in combination with antiplatelet agents are limited by measured and perceived bleeding risks. Anticoagulants with reduced in-vivo bleeding have the potential to address this deficit. In this context, precision anticoagulants such as VE-1902 may play a significant clinical role. In the context of chronic anticoagulation for AF, treatment aims at preventing the formation of generally fibrin-rich clots [61,62]. Therefore, preferentially targeting thrombin-mediated fibrinogen cleavage is of great relevance. Conventional anticoagulants, however, may predispose patients to significant bleeding risk by inhibiting the major downstream effectors of thrombin signaling, including thrombin-mediated platelet activation.

5. Conclusions

VE-1902’s is a selective and potent direct thrombin inhibitor. VE-1902 has unique pharmacology likely because of its covalent mechanism of action wherein it selectively inhibits thrombus formation while leaving thrombin-mediated platelet function largely unaffected and further unlike the other known anticoagulants in the thrombin generation assay it inhibits thrombin formation without significantly delaying the initiation phase of the clotting cascade. We postulate that the ability to maintain platelet activation while inhibiting thrombus formation is responsible for the reduced bleeding in the models tested. VE-1902, and the PROACs in general, may provide the ability to modulate coagulation and platelet activity more precisely than current anticoagulants, thereby reducing bleeding risk. In particular, PROACs may achieve improved cardiovascular outcomes without the concomitant bleeding risk of current therapies especially in situations where both an anticoagulant and antiplatelet agent maybe beneficial. VE-1902 has the appropriate preclinical properties suitable for human dosing and is currently in a first-in-human phase 1 clinical trial.