Research ArticleCardiovascular Disease

Oral thrombin inhibitor aggravates platelet adhesion and aggregation during arterial thrombosis

See allHide authors and affiliations

Science Translational Medicine  30 Nov 2016:
Vol. 8, Issue 367, pp. 367ra168
DOI: 10.1126/scitranslmed.aad6712

The hidden side of an anticoagulant

For patients with atrial fibrillation and other conditions that predispose them to thrombosis, long-term anticoagulation treatment is the norm. Surprisingly, oral thrombin inhibitors, one of the types of anticoagulants used in humans, slightly increased the risk of acute coronary syndromes in clinical studies. To understand this apparent paradox, Petzold et al. compared the effects of treatment with oral thrombin inhibitors, treatment with vitamin K antagonists (another class of anticoagulants), or no treatment at all in patients’ blood and in mouse models of arterial thrombosis, confirming the observations from clinical studies and identifying some of the underlying mechanisms.

Abstract

In patients with atrial fibrillation, oral anticoagulation with oral thrombin inhibitors (OTIs), in contrast to vitamin K antagonists (VKAs), associates with a modest increase in acute coronary syndromes (ACSs). Whether this observation is causatively linked to OTI treatment and, if so, whether OTI action is the result of a lower antithrombotic efficacy of OTI compared to VKA or reflects a yet undefined prothrombotic activity of OTI remain unclear. We analyzed platelet function in patients receiving OTI or dose-adapted VKA under static and flow conditions. In vivo, we studied arterial thrombosis in OTI-, VKA-, and vehicle-treated mice using carotid ligation and wire injury models. Further, we examined thrombus formation on human atherosclerotic plaque homogenates under arterial shear to address the relevance to human pathology. Under static conditions, aggregation in the presence of ristocetin was increased in OTI-treated blood, whereas platelet reactivity and aggregation to other agonists were only marginally affected. Under flow conditions, firm platelet adhesion and thrombus formation on von Willebrand factor, collagen, and human atherosclerotic plaque were increased in the presence of OTI in comparison to VKA. OTI treatment was associated with increased thrombus formation in injured carotid arteries of mice. Inhibition or ablation of GPIbα-thrombin interactions abolished the effect of OTI on thrombus formation, suggesting a mechanistic role of the platelet receptor GPIbα and its thrombin-binding site. The effect of OTI was also abrogated in the presence of aspirin. In summary, OTI treatment has prothrombotic activity that might contribute to the increase in ACS observed clinically in patients.

INTRODUCTION

Oral anticoagulation with vitamin K antagonists (VKAs) is efficient in the prevention of stroke due to atrial fibrillation (AF) and in the treatment of venous thromboembolism. Non-VKA oral anticoagulants (NOACs), comprising oral thrombin inhibitors (OTIs) and factor Xa antagonists, overcome some of the limitations of VKAs, including various drug and food interactions, a delayed onset of action, and the requirement for frequent monitoring. Therefore, they have become an attractive alternative to VKAs for specific indications, including nonvalvular AF as well as treatment and prevention of venous thromboembolism (15).

Among the NOACs, dabigatran acts as a direct univalent, reversible thrombin inhibitor, which is administered orally as the prodrug dabigatran etexilate. The prodrug is rapidly hydrolyzed by ubiquitous esterases, yielding a maximum anticoagulant capacity 0.5 to 2 hours after uptake (6). The Randomized Evaluation of Long-Term Anticoagulant Therapy (RE-LY) trial, comparing the OTI dabigatran to VKA (dose-adjusted) for prevention of stroke in patients with AF, demonstrated that OTI treatment was superior in the higher dose [150 mg twice daily (bid)] and noninferior in the lower dose (110 mg bid) compared to VKA treatment, whereas bleeding risk was similar (150 mg bid) or reduced (110 mg bid). However, uncertainty arose regarding its safety profile, because OTI treatment in the RE-LY trial was associated with a small increase in myocardial infarction (2, 7). Similarly, other randomized clinical trials comparing OTI with standard regimens for the prevention of deep vein thrombosis in knee replacement (8), for the treatment of deep vein thrombosis (9), and as an add-on anticoagulant after percutaneous coronary intervention (PCI) in acute coronary syndrome (ACS) (10) revealed an increase in ACSs in patients receiving OTI. In a meta-analysis including 30,514 participants from seven randomized clinical trials, OTI treatment with dabigatran was associated with a 0.27% absolute and a 33% relative risk increase for acute coronary events (11), a finding that was confirmed recently (12). A more recent meta-analysis including follow-up data of the RE-LY trial as well as two additional studies calculated an even higher relative and absolute risk increase of 41 and 0.53%, respectively, yielding a number needed to treat of 188 to observe one additional myocardial infarction (13). This study also suggested a group-specific effect for OTIs, because ximelagatran and AZD0837 also showed increased numbers for acute coronary events (13).

Notably, the frequency of these adverse events is low, and OTI treatment shows a clear overall clinical net benefit compared to VKA treatment (2, 14, 15). However, it remains unclear whether the increase in acute coronary events is causatively linked to OTI treatment and, if so, whether OTI facilitates arterial thrombotic events (prothrombotic, increasing thrombotic responses compared to both vehicle and VKA) or, rather, is less efficacious than VKAs in preventing thrombosis (reduced antithrombotic activity of OTI, suppressing thrombosis compared to vehicle albeit to a lesser extent than VKA) (16). In particular, the impact of OTI on specific functions of platelets and the coagulation system during arterial thrombosis remains insufficiently understood. Therefore, we analyzed platelet function and coagulation parameters in patients with nonvalvular AF receiving either OTI or VKA (dose-adjusted).

RESULTS

Ristocetin-induced whole blood aggregation is increased under OTI treatment

A total of 95 consecutive patients with nonvalvular AF or flutter receiving chronic oral anticoagulation with OTI (n = 41) twice daily or dose-adapted VKA (n = 54) were included in our study. We enrolled patients only if they had not received any antiplatelet or heparin therapy in the previous 2 weeks. Both patient groups showed similar baseline characteristics (table S1). Assessment of international normalized ratio (INR) values, activated partial thromboplastin time (aPTT), thrombin time, and OTI (dabigatran) plasma concentrations showed the expected differences between groups (Fig. 1, A to D). OTI plasma concentrations [148 ± 109 ng/ml (mean ± SD)] were similar to the plasma concentrations in the RE-LY trial (17).

Fig. 1. Ristocetin-induced whole blood aggregation is increased under OTI treatment.

Basal coagulation parameters of patients receiving VKA or OTI. (A) Measurements of OTI (dabigatran) plasma concentrations using a diluted thrombin time assay (n = 46 VKA, 35 OTI). (B) aPTT (n = 54 VKA, 41 OTI), (C) thrombin time (n = 42 VKA, 33 OTI), and (D) INR (n = 54 VKA, 41 OTI) were measured using routine laboratory assays. BLD, below level of detection. P values were determined by two-tailed, unpaired t test (B) or Mann-Whitney test (C and D) for nonnormally distributed values. (E and F) Whole blood aggregation upon stimulation with ADP or TRAP (E) or ristocetin (F) was measured by impedance-based aggregometry and expressed as area under the curve (AUC) or units, respectively (ADP: n = 30 VKA, 25 OTI; TRAP: n = 15 VKA, 15 OTI; ristocetin: n = 11 VKA, 8 OTI). P values were determined by two-tailed, unpaired t test. (G) vWF plasma concentrations were determined by enzyme-linked immunosorbent assay (ELISA) (n = 11 VKA, 8 OTI). Platelet activation and reactivity in diluted whole blood from patients under VKA or OTI treatment were assayed by flow cytometry. (H and I) GPIIbIIIa activation (H) and P-selectin surface expression (I) were quantified under resting conditions and after stimulation with the indicated agonists (n = 14 VKA, 14 OTI). MFI, mean fluorescence intensity. All data are from individual blood donors with means ± SEM.

To determine whether OTI treatment affects platelet aggregation in whole blood, we performed impedance-based aggregation experiments. We found similar aggregation responses to adenosine 5′-diphosphate (ADP) and thrombin receptor activating protein (TRAP) (Fig. 1E), whereas the response to ristocetin, which mimics shear-induced binding of von Willebrand factor (vWF) to GPIbα (18), was increased (Fig. 1F), despite similar vWF plasma concentrations (Fig. 1G). Next, we investigated platelet activation in response to different agonists in blood from OTI- and VKA-treated patients using flow cytometry. P-selectin expression and GPIIbIIIa integrin activation were similar in VKA- and OTI-treated patients under resting conditions and upon activation with thrombin, ADP, and collagen-related peptide (CRP) (Fig. 1, H and I, and fig. S1, A and B). Further, no significant differences were seen compared to nonanticoagulated controls (fig. S1, C and D).

OTI-treated blood shows increased platelet adhesion, aggregation, and thrombus formation under flow

Next, we analyzed platelet reactivity in whole blood under arterial flow conditions. Consistent with enhanced ristocetin-induced aggregation (Fig. 1F), we found increased platelet adhesion and aggregate formation on vWF after 3 min in response to OTI treatment compared to the VKA group or untreated controls (Fig. 2, A and B). Firm platelet adhesion to collagen was significantly increased (P = 0.022) after 1 min (Fig. 2, C and D) of flow in the OTI group compared to the VKA group, which resolved after 3 min of constant flow (Fig. 2, C and E), suggesting an enhanced thrombotic response at the onset. This increase in adhesion was associated with the formation of larger aggregates under flow (Fig. 2, D and E), which were categorized into three adhesion size categories (fig. S2A). Platelet activation induced by CRP was not altered (fig. S1, A and B).

Fig. 2. Platelet adhesion and aggregate formation are augmented in OTI-treated blood on vWF and collagen under flow conditions.

Blood from VKA- or OTI-treated patients and healthy controls was perfused at a shear rate of 3000/s for 3 min through a vWF-coated flow chamber. (A) Representative fluorescent images after 1 and 3 min of flow. Arrows indicate the direction of flow and represent 120 μm. (B) Platelet adhesion and aggregate formation quantified by total surface coverage (n = 11 VKA, 9 OTI, 9 control). P values were determined by one-way analysis of variance (ANOVA) testing followed by Tukey’s test. (C to E) Platelet adhesion and aggregate formation of VKA- and OTI-treated blood on collagen after 1 min (D) and 3 min (E) of flow (1000/s). Platelets and platelet aggregates were categorized into three groups by their individual adhesion area (<10, 10 to 83, and >83 μm2). The surface area covered is shown per category (n = 14 VKA, 13 OTI). P values were determined by two-tailed, unpaired t test. All data are from individual blood donors with means ± SEM.

To place our findings into a clinical context, we next used a human atherosclerotic plaque homogenate, which is composed of a variety of different matrix proteins that are exposed upon plaque rupture in humans and orchestrates atherothrombosis in a two-step model in vitro (19). In this model, exposure of highly thrombogenic subendothelial matrix proteins, including collagens and vWF, triggers the initial platelet recruitment and aggregate formation, whereas plaque tissue factor–induced thrombin and fibrin formation occurs later (1921). Thrombus formation on the human atherosclerotic plaque homogenate was enhanced in whole blood from OTI-treated patients compared to VKA-treated patients or healthy controls (Fig. 3, A and B, and movie S1). Along with the increase in thrombus area, three-dimensional (3D) thrombus volumes were also larger in the presence of OTI (Fig. 3, C and D, and movie S2). To exclude potential effects of differences in ionized Ca2+ concentrations in recalcified citrated blood, we repeated the experiments using corn trypsin inhibitor (CTI)–anticoagulated blood, which does not affect ionized Ca2+ concentrations (fig. S3A). Again, OTI treatment showed higher surface coverage compared to VKA treatment (fig. S3, B and C, and movie S3).

Fig. 3. OTI-treated blood shows increased platelet adhesion and thrombus formation on the human atherosclerotic plaque homogenate under flow.

Blood from VKA- or OTI-treated patients and healthy controls was perfused at a shear rate of 1000/s for 3 min through a flow chamber covered with human atherosclerotic plaque homogenate. (A) Representative fluorescent images after 1 and 3 min of flow. Arrows indicate the direction of flow and represent 120 μm. (B) Platelet adhesion was quantified as total surface coverage. P values were determined by one-way ANOVA testing followed by Tukey’s test. Data are from individual blood donors with means ± SEM (1 min: n = 7 VKA, 9 OTI, 10 control; 3 min: n = 8 VKA, 9 OTI, 10 control; for 1 VKA patient, the 1-min time point was not acquired). (C) Representative 3D reconstructions of platelet aggregates and thrombi on the atherosclerotic plaque homogenate after 3 min of flow (1000/s). Scale bars, 25 μm. (D) Flow chamber experiments [shown in (A)] were analyzed for surface coverage area versus thrombus volume after 3 min of flow. Data are from individual blood donors, and crosses show means ± SD (n = 5 VKA, 7 OTI, 6 control).

Owing to its pharmacokinetic properties, OTI may have variable intraday plasma concentrations, reaching a peak about 2 to 3 hours after ingestion. To address the potential impact of the timing of drug intake on the parameters of platelet adhesion and thrombus formation, we performed flow experiments on a human atherosclerotic plaque homogenate using blood obtained immediately before and 2 hours after OTI administration. We did not observe any difference in surface coverage (fig. S3D), and there was no correlation with OTI plasma concentrations (fig. S3E). Therefore, OTI treatment increases platelet adhesion and thrombus formation under flow in vitro, exerting a prothrombotic effect.

OTI augments thrombus formation and thrombus stability during arterial thrombosis in mice

Next, we investigated whether our in vitro findings were of relevance for arterial thrombosis in two mouse models, which have been shown to replicate clinically relevant findings (22, 23). First, we analyzed platelet recruitment and clot formation early during arterial thrombosis in a mouse model of carotid ligation injury (24) using intravital microscopy. Mice received a single dose of OTI, a single dose of dimethyl sulfoxide (DMSO), or a 2-day treatment with VKA by oral gavage. Compared to control (vehicle), OTI treatment yielded a significant (P = 0.0001) increase in aPTT (Fig. 4A). VKA treatment produced an increase in INR values compared with OTI (Fig. 4B), which has been reported to plateau at 48 hours of treatment (25). The different effects of VKA and OTI on prothrombin time (PT) and aPTT assays arise from different sensitivities of both tests for quantifying the anticoagulant capacity of the drugs in patients (26, 27). Therefore, our results are in line with an earlier dose-finding study in mice, in which OTI doses between 37.5 and 112.5 mg/kg did not affect PT but resulted in a dose-dependent increase in aPTT values (28).

Fig. 4. OTI augments thrombus formation and thrombus stability during arterial thrombosis in mice.

(A and B) Mouse coagulation parameters. (A) aPTT was determined after application of a single dose of OTI or vehicle (DMSO) in wild-type (WT) mice. Data show individual animals’ means ± SEM (n = 10 vehicle, 9 OTI). (B) INR values were determined after application of a single dose of OTI or 2 days (48 hours) of VKA treatment. Data show individual animals’ means ± SEM (n = 11 VKA, 11 OTI). P values were determined by two-tailed, unpaired t test. (C to E) WT mice experienced carotid artery injury 30 min after application of OTI or vehicle (DMSO), or after 2 days (48 hours) of VKA treatment. In the carotid artery ligation model, platelet adhesion (C) and aggregate formation (D) were measured at the injury site at indicated time points. Data are from individual animals expressed as a percentage of the vehicle treatment group at the 5-min time point with means ± SEM (n = 6 VKA, 9 OTI, 9 vehicle). P value was determined by the Kruskal-Wallis test [15-min time point in (D)] followed by Dunn’s test. (E) Occlusive thrombus formation and presence of reflow were assessed by intravital microscopy at the indicated time points in the carotid artery wire injury model. Data are presented as a Kaplan-Meier curve (n = 6 vehicle, 3 VKA, 9 OTI). P value was calculated by the log-rank (Mantel-Cox) test.

Compared to controls or VKA-treated animals, OTI-treated animals had normal platelet adhesion at the site of injury (Fig. 4C). However, aggregate formation was significantly (P = 0.02) increased in OTI-treated mice compared to vehicle-treated mice (Fig. 4D). These findings further substantiate our flow chamber results and point to an agonistic prothrombotic effect of OTI on platelet aggregate formation during early arterial thrombosis. Next, we sought to delineate the impact of OTI on occlusive thrombus formation at later stages of arterial thrombosis. We performed wire injury in the common carotid artery, which causes rapid thrombotic vessel occlusion that partially resolves over time, thereby reestablishing blood flow. Next, we used intravital microscopy to determine whether occlusive thrombi were formed and we measured the time to recurrence of blood flow, which reflects thrombus stability. In the OTI group, all vessels were occluded, and blood flow was restored within 30 min in only one of nine animals, whereas all vehicle-treated animals (n = 6) reestablished blood flow in the occluded vessels within 15 min (Fig. 4E). VKA-treated animals did not show any occlusive thrombus formation, similar to the anticoagulation with heparin (29) reported earlier. Together, these data indicate that OTI treatment has a mild but consistent prothrombotic effect with augmented platelet aggregation and thrombus formation.

The prothrombotic action of OTI is not a result of thrombin-driven coagulation

Recent evidence suggests increased thrombin formation during OTI treatment (30). We examined potential mechanisms behind the proatherothrombotic effects of oral thrombin inhibition by investigating thrombin formation after OTI or VKA treatment under steady-state conditions. Plasma concentrations of prothrombin fragments F1/F2 were significantly (P = 0.0001) elevated in response to OTI treatment, and thrombin–antithrombin III (TAT) complexes appeared to increase slightly but did not reach statistical significance (Fig. 5, A and B). Thrombin protein concentrations were reduced by up to 80% in blood from VKA-treated patients compared with OTI-treated patients (Fig. 5C), although thrombin plasma concentrations were similar between OTI-treated patients and healthy controls (Fig. 5D). However, this did not translate into an increase in thrombin enzymatic activity (Fig. 5E). In line with this latter finding, plasma D-dimer concentrations did not differ between OTI- and VKA-treated patients (Fig. 5F). Together, these findings indicate that OTI treatment is less efficient than VKA treatment in suppressing thrombin formation under chronic treatment but does not increase plasma thrombin compared to healthy controls. Hence, differences in plasma thrombin alone are unlikely to explain the differences in platelet adhesion and aggregation in OTI-treated patients compared to VKA-treated patients and healthy controls.

Fig. 5. The prothrombotic action of OTI is not a result of thrombin-driven coagulation.

(A and B) Plasma concentrations of prothrombin fragments F1+F2 (A) and thrombin–antithrombin III complexes (B) were analyzed by ELISA-based assays. P value was determined by two-tailed unpaired t test. Data are from individual blood donors with means ± SEM (n = 15 VKA, 10 OTI). (C) Western blot of thrombin from plasma of individual blood donors. Thrombin concentrations are shown as multiples of a reference plasma (Ref. value = 1, which corresponds to one VKA-treated patient). Relative thrombin concentrations are indicated. Data show individual blood donors with means ± SEM (n = 14 VKA, 13 OTI). P value was determined by two-tailed, unpaired t test. A representative Western blot for thrombin and albumin (protein loading control) is shown. A.U., arbitrary units. (D) Thrombin and albumin (protein loading control) from plasma of OTI patients and nontreated controls are shown as multiples of reference plasma (a value of 1 corresponds to one OTI-treated patient). n.d., not determined. (E) Thrombin activity was quantified by ELISA. (F) D-dimer levels were determined by routine laboratory assays. Data in (D) and (E) are from individual donors with means ± SEM (n = 15 VKA, 10 OTI). P values were determined by Mann-Whitney test.

Prothrombotic effects of OTI can be reversed by aspirin treatment

To address whether the prothrombotic effects are mediated by alterations of platelet function, we treated patient whole blood with 1 mM aspirin, the most widely used clinical antiplatelet drug. As expected, aspirin strongly reduced whole blood aggregation upon stimulation with arachidonic acid, a cyclooxygenase-1 substrate (Fig. 6, A and B). On human atherosclerotic plaque, the excess prothrombotic effect observed with OTI compared to both VKA-treated patients and controls was abolished in the presence of aspirin. Consequently, platelet surface coverage under flow (Fig. 6, C and D, and movie S4) and thrombus volume (Fig. 6E) were not significantly different between groups in the presence of aspirin. These data indicate that OTI exerts its prothrombotic effects through alterations of platelet function that are sensitive to aspirin. Unfortunately, concomitant treatment with OTI and aspirin in mice resulted in a bleeding phenotype, which precluded an in-depth analysis of aspirin sensitivity of thrombus formation in response to OTI in vivo.

Fig. 6. Prothrombotic effects of OTI can be reversed by aspirin treatment.

(A and B) Whole blood aggregation upon stimulation with arachidonic acid after aspirin (1 mM) or vehicle treatment was measured by impedance-based aggregometry. (A) Aggregation responses after 6 min are shown (1 U corresponds to 10 AUC). Data are from individual donors with means ± SEM (n = 8 VKA, 9 OTI, 6 control). (B) Representative aggregation tracing of aspirin or vehicle treatment of OTI-treated blood. (C to E) Effect of aspirin on blood from VKA- or OTI-treated patients and healthy controls perfused at a shear rate of 1000/s through a flow chamber covered with the human atherosclerotic plaque homogenate. (C) Representative fluorescent images after 1 and 3 min of flow. Arrows indicate the direction of flow and represent 120 μm. (D) Platelet adhesion and thrombus formation were analyzed by quantification of total surface coverage (μm2). Data show individual blood donors with means ± SEM (n = 8 VKA, 9 OTI, 9 control). (E) Surface coverage versus thrombus volume after 3 min of flow. Data show individual blood donors with means ± SD (n = 5 VKA, 7 OTI, 5 control).

An altered thrombin-GPIbα interaction mediates the prothrombotic effects of OTI

To further dissect the underlying molecular mechanism, we determined whether activation of platelet thrombin receptors contributes to enhanced thrombus formation in response to OTI. Platelets express several thrombin surface receptors, including the family of protease-activated receptors (PARs) and GPIbα (31). Together with GPIbβ, GPV, and GPIX, GPIbα forms the vWF receptor complex, which mediates initial platelet recruitment. Inhibition of GPIbα-vWF interaction with blocking antibody 6B4 drastically reduced platelet recruitment and adhesion to human atherosclerotic plaque homogenates in control blood (fig. S4).

To evaluate the potential GPIbα dependence of the prothrombotic action of OTIs, we used two different in vivo approaches. In one approach, we generated bone marrow chimeras whose megakaryocytes/platelets express a mutant IL-4R/GPIbα-tg fusion protein in which the GPIbα extracellular domain is replaced by the human interleukin-4 receptor (IL-4R) extracellular domain linked to a functional GPIbα intracellular domain (32). IL-4R/GPIbα-tg chimeras have normal platelet counts, and no GPIbα surface expression was detected by flow cytometry, indicating complete chimerism in peripheral blood (fig. S5, A and B). In these animals, thrombus formation after wire injury was absent after either vehicle or OTI treatment (Fig. 7A). In a second set of experiments, WT C57Bl/6J mice were treated with the Xia.B2 GPIbα antibody Fab fragment (2.5 mg/kg) (33) to block vWF-GPIbα interaction plus OTI or vehicle. Again, no occlusive thrombus formation was observed in either group (Fig. 7A). Thus, OTI does not exert prothrombotic actions in the absence of platelet GPIbα.

Fig. 7. An altered thrombin-GPIbα interaction mediates the prothrombotic effects.

(A) Wire injury of the common carotid artery was inflicted in WT mice and IL-4R/GPIbα-tg chimeric animals 30 min after application of a single dose of OTI or vehicle (DMSO). Half of the WT animals received the Xia.B2 Fab fragment (2.5 mg/kg) or the control Fab fragment intravenously (i.v.). Occlusive thrombus formation and presence of reflow were assessed by intravital microscopy at the indicated time points. (B to D) Flow chamber perfusion of blood from VKA- or OTI-treated patients in the absence or presence of VM16D GPIbα–blocking antibody (20 μg/ml). Chambers coated with the human plaque homogenate were perfused at a shear rate of 1000/s for 3 min. (B) Representative images after 3 min of blood flow. Arrows indicate the direction of flow and represent 120 μm. (C) Platelet adhesion and thrombus formation were analyzed as total surface coverage (μm2) for individual patients (n = 5 VKA, 4 OTI). (D) Platelet GPIbα surface expression was determined in whole blood of VKA- and OTI-treated patients. Data are from individual blood donors with means ± SEM (n = 14 VKA, 13 OTI). (E) Blood from OTI-treated patients was perfused in the absence or presence of hirudin (26 μg/ml). Total platelet surface coverage (μm2) is shown for individual patients (n = 3). (F) Blood from untreated controls was treated with buffer, the thrombin inhibitor hirudin (26 μg/ml), or the PAR-1 antagonist SCH79797 (2.5 μM) and perfused for 3 min over the plaque homogenate at a shear rate of 1000/s. Platelet deposition was analyzed as a percentage of surface coverage. Data from individual blood donors are shown with means ± SEM.

Thrombin binding to GPIbα drives platelet aggregation and induces intracellular Rho-associated protein kinase signaling during platelet shape change even in the absence of PAR-1 and PAR-4 receptor signaling and independently of vWF (34). Recently, vWF-independent activation of GPIbα was shown to trigger thromboxane A–sensitive paracrine platelet activation (35). Therefore, we tested whether GPIbα was directly involved in the prothrombotic effects of the OTI. We performed flow chamber experiments in the presence of the antibody VM16D, which specifically blocks GPIbα-thrombin interactions (36). VM16D but not the Ct6 isotype control antibody reduced dynamic thrombus formation on collagen (fig. S6, A to C) and on human atherosclerotic plaque homogenates exposed to arterial shear (Fig. 7, B and C, and movie S5). Notably, there were no significant differences in platelet GPIbα surface expression between OTI- and VKA-treated patients (Fig. 7D), indicating that thrombus formation in OTI patients is facilitated by an interaction between thrombin and the thrombin-binding site of GPIbα. Consistently, hirudin (26 μg/ml), an inhibitor of thrombin catalytic activity, did not affect the impact of OTI treatment on thrombus formation in the blood of OTI patients (Fig. 7E). Similar results were found for control blood treated with hirudin and a PAR-1 antagonist (SCH79797; 2.5 μM) (Fig. 7F). Thus, enhanced GPIbα-thrombin interactions, but not PAR-1 or the catalytic activity of thrombin, contribute to the increase in flow-induced platelet adhesion and thrombus formation in response to OTI treatment.

DISCUSSION

Although OTI treatment shows a clear overall clinical net benefit compared to VKA treatment in recent clinical trials, it is unclear whether the reported numeric increase in myocardial infarction in patients with nonvalvular AF receiving OTI is due to chance or is specific to OTI. Here, we demonstrate that OTI facilitates platelet adhesion and thrombus formation on mouse and human subendothelial matrix proteins, including human atheromatous plaque material, compared to both the VKA group and healthy controls. Our data suggest that OTI exerts mild prothrombotic effects that might contribute to the increased frequency of myocardial infarction under OTI. Patients at risk could potentially benefit from an individualized anticoagulation therapy.

Low doses of aspirin ablated the prothrombotic effects of OTI, suggesting that platelet activation in response to OTI contributed to our findings. However, in line with earlier studies (3740), we observed that OTI did not alter platelet aggregability alone or in response to ADP, TRAP, or CRP under static conditions. However, OTI increased aggregation in response to ristocetin, which triggers vWF binding to platelet GPIbα. Moreover, compared to the VKA group and healthy controls, OTI treatment enhanced platelet recruitment and aggregate formation on vWF under flow conditions. In addition, OTI facilitated platelet adhesion and aggregation on fibrillar collagen, a substrate that also triggers vWF-driven platelet recruitment (21, 41). To model the state of atherosclerotic plaque rupture in ACSs, we used a human atherosclerotic plaque homogenate (19). Notably, plaque-induced thrombus formation strongly depends on the vWF receptor GPIbα in this setting (21). Again, we noted a marked increase in platelet adhesion and thrombus formation with OTI-treated blood compared to VKA- or vehicle-treated blood. Our in vitro data were supported by in vivo findings in mice, which revealed an increase in thrombus formation and thrombus stability in two different models of GPIbα-dependent arterial thrombosis in OTI-treated animals compared to VKA- and vehicle-treated animals. Together, these findings indicate that a GPIbα-dependent pathway contributes to the prothrombotic effects observed in the presence of OTI.

As we further explored the underlying mechanisms, we found that thrombin signaling via platelet thrombin receptor(s) facilitated thrombus formation in response to OTI. Thrombin, the target of OTI, mediates platelet activation through various mechanisms. Enzymatically active thrombin is one of the most potent platelet stimuli, which cleaves platelet PARs, yielding full-fledged platelet activation (42). However, hirudin had no effect on OTI-triggered thrombus formation, thus excluding a major contribution of catalytic thrombin signaling via PAR. In addition to PAR, thrombin can also signal in a noncatalytic fashion through GPIbα (34), a major adhesion receptor involved in thrombus formation under flow in our in vitro and in vivo assays. Data from crystal structure studies suggested a clustering and bridging function of thrombin between two neighboring GPIbα receptors on the same or adjacent platelets through exosites (34, 43, 44), of which exosite II seems to play the predominant role (45). Here, we show that antibody blockade of thrombin-GPIbα interactions abolished the prothrombotic effect of OTI. Compared to the VKA group and healthy controls, OTI therefore supported an interaction between thrombin and the platelet GPIbα receptor, which in turn facilitated platelet adhesion and aggregation under flow.

Several mechanisms likely contributed to enhanced thrombin-GPIbα signaling in the presence of OTI. Whereas GPIbα receptor expression did not differ between groups, thrombin plasma concentrations were suppressed only in patients receiving VKA but not in OTI-treated patients or healthy volunteers. Hence, differences in plasma thrombin might have contributed to the prothrombotic effects of OTIs compared to VKA, although they cannot explain the differences in platelet adhesion/aggregation observed between OTI patients and healthy controls. Although thrombin concentrations prevailing in the local microenvironment of a developing thrombus may vary between OTI and healthy controls despite similar plasma concentrations, OTI binding might also increase thrombin’s affinity to the GPIbα receptor. Our observation that the platelet response to thrombin (P-selectin exposure) is mildly increased in the presence of OTI compared to healthy controls and VKA could support the concept that OTIs shift thrombin’s affinity toward platelet agonist receptors, including GPIbα.

OTIs do not cause spontaneous platelet activation, and two additional conditions, namely, shear stress and exposure of subendothelial (atheromatous) matrix, are required for enhanced platelet adhesion and aggregation to occur in the presence of OTI. This is consistent with the clinical scenario (13), where OTIs predispose to thrombotic events under high-shear conditions prevailing within coronary arteries, and they do so in a cohort of patients characterized by a higher incidence of coronary artery disease (46). A case in point is a recent randomized clinical trial, which replaced conventional periprocedural unfractionated heparin by preprocedural OTI in patients undergoing PCI for stable angina (47). Although underpowered to allow end point–driven conclusions, the OTI group was associated with a numerical increase in periprocedural abrupt vessel closure that required bailout antithrombotic therapy. Nevertheless, in the context of our study, these latter findings have to be interpreted cum grano salis because the cause of coronary thrombosis during or early after PCI is multifactorial and comprises multiple mechanisms unrelated to subendothelial matrix exposure.

Hence, the mild prothrombotic effects upon OTI treatment identified here likely contributed to the increased frequency of clinically relevant myocardial ischemia reported in recent clinical trials for patients receiving OTI. Therefore, an assessment of the individual coronary artery disease and myocardial infarction risk should be performed before OTI treatment is initiated.

Reproducing the interwoven complexity of spatially and temporarily regulated mechanisms during arterial thrombus formation in vivo remains challenging. Therefore, our study has an inherent limitation, because it is difficult to directly translate data obtained in vitro and in animal models into the clinical scenario of myocardial infarction. We addressed this key shortcoming by recapitulating our in vitro findings using human atheromatous plaque homogenates. Our study also has specific methodological limitations. In particular, not all patients could be tested in every experimental setup because of logistical limitations. Further, blood donors were not randomized to VKA or OTI treatment, which might have resulted in a selection bias. However, the baseline characteristics were similar between groups, allowing comparisons of OTI- and VKA-treated patients. In animal experiments, we used IL-4R/GPIbα-tg chimeric mice as well as antibody treatment to demonstrate that the prothrombotic effects of OTI were not found in the absence of the extracellular domain of GPIbα. Although this is in line with our concept that OTI facilitated thrombosis via enhanced thrombin-GPIbα signaling, we cannot completely rule out that our models overestimated the GPIbα dependency of OTI-mediated prothrombotic effects in vivo, because the absence of functional GPIbα has previously been shown to protect from thrombus formation induced by ferric chloride in mesenteric arteries (48).

Our study identified a prothrombotic effect as a likely mechanism underlying the observed increase in the rate of myocardial infarctions in patients receiving OTI. However, we cannot exclude that, in addition to the platelet agonistic effects reported here, a reduced antithrombotic efficacy of OTI compared to VKA, which may have been masked by OTI’s prothrombotic effects, may have contributed to the clinical observation. Two clinical observations are consistent with a broader anticoagulant activity of VKA compared to OTI: (i) In patients with mechanical heart valves, VKAs are more efficient than OTIs in preventing thromboembolic complications (49); (ii) VKAs show a higher bleeding rate compared to OTIs (12). Additional evidence might come from the ongoing OAC-ALONE (Optimizing Antithrombotic Care in Patients with AtriaL fibrillatiON and Coronary stEnt) study, which compares different anticoagulation regimens with or without additional antiplatelet therapy after coronary intervention (ClinicalTrials.gov Identifier: NCT01962545).

Finally, the translational implications emerging from our data regarding the risk of myocardial infarction under NOAC treatment are limited to patients receiving OTI treatment and do not allow conclusions regarding the effect of factor Xa inhibitors on atherothrombosis. Further experimental studies and clinical randomized trials comparing different NOACs in a head-to-head design might help to address this important clinical question.

In conclusion, our translational study provides in vitro and in vivo evidence suggesting that OTI treatment exerts a mild prothrombotic effect by triggering thrombin-GPIbα signaling in platelets. These findings might contribute to the increase in ACS observed clinically in patients receiving OTI.

MATERIALS AND METHODS

Study design

Our experimental study aimed to address whether the increased number of acute coronary events observed in patients who underwent OTI treatment in randomized clinical trials is causatively linked to OTI treatment. Therefore, we analyzed the effects of OTI as well as VKA on platelet function and coagulation parameters in patients with AF. Eligible patients were at least 18 years old and treated with oral anticoagulants because of AF or flutter. Patients received either the OTI dabigatran (150 mg bid) or VKA (last reported INR values between 2 and 3) without any additional antiplatelet medication or heparin therapy for at least 2 weeks. Patients with known hematological disorders, cancer, or renal insufficiency were excluded from the study. Healthy volunteers, who did not receive any anticoagulants or antiplatelet therapy, were also recruited as blood donors and served as controls. Patients receiving the indicated treatment were continuously recruited in our clinic and were therefore not randomized for the treatment regime.

Patient numbers for primary study end point (surface coverage on human atherosclerotic plaque material after 3 min) were calculated to establish statistical certainty with the probability of a type I error of P = 0.05 (α), the probability of a type II error of P = 0.2 (power, 0.8), a minimum detectable difference of 30%, and an expected SD of 15%. Owing to organizational limitations, not all patients could be tested in every experimental setup. Data were analyzed by staff blinded to the type of treatment. All individual data are shown, and no data outliers were dropped. After clinical file review, two patients were excluded because they had received heparin treatment within the last 2 weeks before presentation.

The study was approved by the institutional ethics committee of the Ludwig-Maximilian University of Munich (LMU Munich) and the Technical University of Munich (Germany). Written informed consent was obtained from all blood donors before inclusion in our study, and investigations were conducted according to the principles expressed in the Declaration of Helsinki.

Flow chamber experiments

Flow chambers (μ-Slide VI0.1 or glass slides stuck to Sticky-Slide I0.1 Luer; Ibidi) were coated with vWF (20 μg/ml) (Enzo), fibrillar collagen (250 μg/ml) (Takeda-Nycomed), or homogenized human atherosclerotic plaque material, which was obtained by endarterectomy from carotid arteries (50), as described earlier. These slides were blocked with 4% human serum albumin (Sigma-Aldrich) and rinsed with phosphate-buffered saline (PBS). Citrate-anticoagulated blood was stained with 0.01% (final concentration) rhodamine 6G chloride (Life Technologies) for 15 min at 37°C in the dark. The anti-CD42b mouse antibody (VM16D, Thermo Scientific) with a final concentration of 20 μg/ml was added. VM16D specifically inhibits GPIbα-thrombin interactions at low thrombin concentrations (<0.05 U/ml) without affecting GPIbα-vWF interactions (36). For some experiments, recombinant hirudin (26 μg/ml; Refludan), a PAR-1 antagonist (2.5 μM SCH79797; Abcam), or the GPIbα-vWF interaction blocking antibody 6B4 was added to the blood. Immediately before starting blood perfusion, 6 mM CaCl2 was added to reach a (free) calcium concentration of about 0.5 mM. For collagen and plaque material, we used 1000/s as the shear rate, because this is generally accepted to adequately mimic arterial shear conditions in vitro. For studies on vWF, a shear rate of 3000/s and citrated blood were used. The blood was perfused using a syringe infusion pump (KD Scientific), followed by a 3-min PBS rinsing step. Alternatively, blood was anticoagulated with CTI with a final concentration of 32 μg/ml. Experiments with CTI-anticoagulated blood were performed within 15 min after blood drain without addition of CaCl2. A detailed description of materials and methods is found in the Supplementary Materials.

Statistical analysis

Statistical analysis was performed using Prism 5 software (GraphPad Software). Normality testing was carried out using the D’Agostino-Pearson omnibus test, where applicable, or the Kolmogorov-Smirnov test. Data were then analyzed by unpaired two-sided Student’s t test or Mann-Whitney test (nonnormally distributed), as appropriate. For multiple comparisons, one-way ANOVA or the Kruskal-Wallis test (nonnormally distributed) was performed, followed by Tukey’s or Dunn’s (nonnormally distributed) post hoc analysis, respectively. Wire injury data are presented as Kaplan-Meier curves and were analyzed by the log-rank (Mantel-Cox) test. P values <0.05 were considered significant.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/8/367/367ra168/DC1

Materials and Methods

Fig. S1. Platelet reactivity is not altered under OTI compared to VKA treatment and nonanticoagulated blood.

Fig. S2. Adhesion size categories are characterized by area size and thrombus volume.

Fig. S3. Prothrombotic effects of OTI are found under CTI anticoagulation and are independent from OTI plasma concentration.

Fig. S4. Inhibition of GPIbα-vWF interaction reduces thrombus formation on plaque material under arterial flow conditions.

Fig. S5. IL-4R/GPIbα-tg chimeras have normal platelet counts and lack GPIbα surface expression.

Fig. S6. Platelet adhesion and thrombus formation depend on GPIbα-thrombin interaction but not catalytically active thrombin.

Table S1. Hematological and clinical baseline characteristics are similar between treatment groups.

Movie S1. Flow chamber experiment on a plaque homogenate.

Movie S2. Confocal microscopy 3D reconstructions of flow chamber experiments.

Movie S3. Flow chamber experiment using CTI-inhibited blood.

Movie S4. Flow chamber experiment on the plaque homogenate after aspirin treatment.

Movie S5. Flow chamber experiment in the presence of VM16D GPIbα–blocking antibody.

Reference (51)

REFERENCES AND NOTES

  1. Acknowledgments: We thank F. Gärtner for helpful discussions regarding technical issues. We also thank C. Kieu, E. Raatz, G. Baleanu, and C. Sonne for technical assistance and help with patient recruitment. We thank H. Deckmyn for providing the GPIbα antibody (6B4). Funding: This work was supported by the DZHK (German Centre for Cardiovascular Research), the Deutsche Forschungsgemeinschaft (SFB 1123 project A07, B06, and B08 and SFB 914), the BMBF (German Ministry of Education and Research), the FöFoLe program (LMU Munich), and the August-Lenz-Stiftung. Author contributions: T.P. designed and performed experiments and wrote a draft of the manuscript. I.K., A.E., I.S., and S.C. performed and analyzed animal experiments. M.T., J.J., R.R., M.L., and B.H. performed and analyzed flow chamber experiments and confocal microscopy. R.B. extracted and provided human atherosclerotic plaques. D.B., C.L., and C.K. collected and analyzed patient data and helped to interpret experimental data. S.B. determined the coagulation parameter. W.S. contributed in the design of the experiments and in the drafting of the manuscript. C.S. and S.M. wrote the manuscript and approved the final version of the manuscript. Competing interests: S.B. received lecture fees from Boehringer Ingelheim in 2013. C.K. received lecturer fees, travel support, or research sponsoring from Biotronik (advisory board), Boston Scientific, Medtronic, St. Jude Medical, and Sorin-LivaNova (advisory board). C.L. received lecturer fees or travel support from Biotronik, St. Jude Medical, Sorin, and Medtronic. D.B. received lecturer fees from Abbott Vascular. No other conflicts of interest in connection with the submitted article exist. Data and materials availability: Material sources are indicated within the text.
View Abstract

Navigate This Article