Reversing Rivaroxaban Anticoagulation as Part of a Multimodal Hemostatic Intervention in a Polytrauma Animal Model

The methods for this study were similar to those of several previous studies.^16–18  An overview of the study stages is presented in figure 1.

This study was performed at the RWTH Aachen University Hospital, Aachen, Germany, following German legislation governing animal studies following the Guide for the Care and Use of Laboratory Animals.¹⁹  The government office approved the protocol for animal care and use (Landesamt für Natur, Umwelt und Verbraucherschutz, Recklinghausen, Germany). Before surgery, the pigs were housed in ventilated rooms and allowed to acclimatize to their surroundings for a minimum of 7 days. Examination by a veterinarian ensured the good health of all animals.

Forty-eight (n = 48) male pigs were anesthetized, prepared, and monitored as described (see Supplemental Digital Content, https://links.lww.com/ALN/C667, for additional details of animal preparation and anesthesia).²⁰  Concealment of allocation was carried out as the animal entered the trial. After line placement, a midline laparotomy with cystostomy was performed. Subsequently, a 30-min infusion of rivaroxaban (1 mg/kg; Bayer Healthcare, Germany) was started. After this infusion of 30 min, trauma was induced. Therapy was started 12 min after the infliction of injury. Animals (n = 8/group) were allocated to receive normal Ringer’s solution (control) or 12.5, 25, or 50 U/kg PCC (Beriplex P/N; CSL Behring, Germany; U.S. brand name Kcentra). In the second part of this study, the animals were allocated to receive tranexamic acid (20 mg/kg; Cyclocapron, Pfizer, USA) plus PCC (12.5 U/kg) or tranexamic acid (20 mg/kg) plus PCC (12.5 U/kg) plus fibrinogen concentrate (80 mg/kg; Haemocomplettan P, CSL Behring).

After localizing the femur with a needle, a captive bolt gun (Karl Schermer and Co., Germany) was used to fracture both femurs and create a soft tissue injury midshaft. Next, a reproducible blunt liver injury was induced using a custom-made instrument described elsewhere.²¹  Five minutes after injury and after the onset of hemorrhagic shock, the animals were resuscitated with Ringer’s solution (1 l over 5 min). Twelve minutes after injury, blood loss was measured by suctioning intraperitoneal blood. Study treatments were then administered. Animals surviving for the 240-min observation period after injury were euthanized with pentobarbital. Immediately after death, total postinjury blood loss was determined, and the internal organs (heart, lungs, liver, and kidneys) were examined macroscopically and histologically by a blinded pathologist for the occurrence of thromboembolic events. The tissue sections were embedded in paraffin and stained by hematoxylin and eosin based on a standard protocol.^22,23 

The blood samples were obtained before infusion of rivaroxaban; after a 30-min infusion of rivaroxaban; and 12, 30, 60, 120, 180, and 240 min after injury. For animals dying before 240 min postinjury, the last sample was obtained immediately after death. Details of the tests performed on the blood samples are provided in the Supplemental Digital Content (https://links.lww.com/ALN/C667).

In total, 48 animals were included in this study, and each group (groups 1 to 6) consisted of 8 animals. The sample size was based on previous experiments performed by our group with a similar animal model that investigated PCC monotherapy for dabigatran reversal.¹⁶  Therefore, no a priori statistical power calculation was conducted. One-way ANOVA was used to assess the differences in total blood loss between groups, which was the primary endpoint of the study. Post hoc comparisons of all means were performed using a Tukey–Kramer test. A repeated-measures ANOVA (mixed model with restricted maximum likelihood) was used to evaluate the effects of treatments on coagulation parameters, blood cell count, and hemodynamic variables, representing the secondary endpoints. The model included intervention as a group factor and time as a repeated factor. The Sidak method was applied for α error correction of multiple comparisons in this case. A Shapiro–Wilk test was performed to test for normality. Survival was analyzed by applying the pairwise log-rank tests. All data are reported as the means ± SD, the significance was set at P < 0.05, and early animal death before the end of the experiment time was the only reason for missing data. In the control group, one animal died before 60 min; thus, the number of animals was seven at 120 min and five at 180 min, and it was reduced to one animal at 240 min posttrauma. In the 12.5 U/kg PCC plus tranexamic acid group, there were seven animals at 120 min and five at 240 min after trauma induction. Two-tailed statistical analyses were performed using SPSS Statistics software (version 25.0, SPSS, USA), and GraphPad Prism (version 8.2.1, GraphPad Software, Inc., USA) was used for graphical purposes.

Forty-eight pigs were included in this study. Until the time of death, complete data were available for all animals for all study variables. Baseline laboratory and hemodynamic parameters were comparable between the groups before the injury.

After the 30-min infusion of rivaroxaban, the overall mean plasma rivaroxaban concentration (all animals) was 693 ± 91 ng/ml (mean ± SD), and both prothrombin time and extrinsic thromboelastometry clotting time were prolonged (table 1; fig. 2, A and B). Plasma concentration of rivaroxaban 12 min after trauma induction and before the administration of the study treatments was 368 ± 98 ng/ml. Rivaroxaban did not affect hemoglobin, platelet count, nor fibrinogen concentration (table 1; fig. 3, A and B). A further effect of rivaroxaban was measured in the reduction of thrombin generation (fig. 4). Four hours after rivaroxaban administration, the plasma concentrations of rivaroxaban were reduced to less than 50 ng/ml in all treatment groups (table 1).

In the control group (rivaroxaban plus placebo), postinjury blood loss was 3,313 ± 634 ml compared with 1,541 ± 269 ml and 1,464 ± 108 ml in the 25 and 50 U/kg PCC groups, respectively (P < 0.0001 vs. 25 and 50 U/kg PCC; fig. 5). The mortality rate was 100% in the control group, and the mean survival time was 131 min (range, 50 to 200 min). All animals that received 25 U/kg PCC or 50 U/kg PCC survived the 240-min observation period (P < 0.01 vs. control). Hemodynamic variables and lactate levels were stabilized (see table, Supplemental Digital Content, https://links.lww.com/ALN/C667, hemodynamic variables).

Treatment with 12.5 U/kg PCC (blood loss, 2,671 ± 334 ml; P = 0.934 vs. control) and 12.5 U/kg PCC plus tranexamic acid (blood loss, 2,910 ± 856 ml; P = 0.931 vs. control) did not significantly reduce blood loss compared with control, whereas the mean survival time was 234 min (range, 195 to 240 min) and 186 min (range, 46 to 240 min), respectively. The mortality rates were 88 and 50% in the 12.5 U/kg PCC and 12.5 U/kg PCC plus tranexamic acid groups, respectively.

All animals in the 12.5 U/kg PCC plus tranexamic acid and fibrinogen concentrate group survived, and the total postinjury blood loss was reduced to 1,836 ± 556 ml (all P < 0.001 vs. control, 12.5 U/kg PCC, and 12.5 U/kg PCC plus tranexamic acid groups; fig. 5). Stabilization of hemodynamic variables and lactate levels was also observed in this group (see table S1, Supplemental Digital Content, https://links.lww.com/ALN/C667, hemodynamic variables).

Despite fluid resuscitation, control animals developed severe shock after trauma, with low mean arterial pressure and cardiac output attributable to ongoing blood loss (see table S1, Supplemental Digital Content, https://links.lww.com/ALN/C667, hemodynamic variables). Control animals also had the lowest platelet counts at 120 min posttrauma compared to 12.5 U/kg PCC plus tranexamic acid and fibrinogen concentrate and 50 U/kg PCC groups (all P < 0.001) and the lowest hemoglobin levels of all treatment groups at 60 min after trauma induction (table 1). Furthermore, these animals developed severe coagulopathy with deterioration in all coagulation parameters over time (fig. 3 and 4; table 1). Extrinsic thromboelastometry maximum clot firmness was significantly lower in this group at 2 h posttrauma than the 12.5 U/kg PCC (P = 0.014 vs. control), 25 U/kg PCC, 50 U/kg PCC, and 12.5 U/kg PCC plus tranexamic acid and fibrinogen concentrate groups (all P < 0.0001) and at 3 h posttrauma compared with the PCC 12.5, PCC 25, PCC 50, 12.5 U/kg PCC plus tranexamic acid, and 12.5 U/kg PCC plus tranexamic acid and fibrinogen concentrate groups (all P < 0.0001 vs. control; fig. 2, C and D). Additionally, the plasma fibrinogen concentrations were significantly lower than in the 12.5 U/kg PCC plus tranexamic acid and fibrinogen concentrate group at 30 min posttrauma (P < 0.0001 vs. control; fig. 3B). The impairment of thrombin generation caused by rivaroxaban was sustained among control animals throughout the observation period (fig. 4).

After PCC therapy, prothrombin time was reduced dose-dependently in these three groups at 60 min after trauma (all P < 0.01 vs. control; table 1). Thrombin generation increased dose-dependently immediately after PCC administration. At 30 min after trauma, endogenous thrombin potential was 419 ± 60 nM min in the 12.5 U/kg PCC group, 655 ± 89 nM min in the 25 U/kg PCC group, and 981 ± 159 nM min in the 50 U/kg PCC group, whereas the peak height was 28 ± 6, 31 ± 7, and 49 ± 11 nM, respectively (fig. 4). Endogenous thrombin potential remained significantly greater than baseline throughout the remainder of the observation period in both the 25 and 50 U/kg PCC groups (all P < 0.0001 vs. control). Peak height was below the baseline value in all PCC groups and subsequently increased over time while maintaining baseline value 240 min posttrauma. Additionally, peak height in the 50 U/kg PCC group of treated animals was significantly higher 30 min after trauma and over the observation time compared to all other groups (all P < 0.0001; fig. 4, C and D).

Corresponding with significant blood loss reductions, hemodynamic parameters remained stable in the 25 and 50 U/kg PCC groups after fluid resuscitation and trauma (see table S1, Supplemental Digital Content, https://links.lww.com/ALN/C667, hemodynamic variables). Due to ongoing blood loss, endogenous thrombin potential in the 12.5 U/kg PCC group decreased slightly over time, and platelets were significantly lower than the 50 U/kg PCC group at 120 min posttrauma (P = 0.008; fig. 4A; table 1). The reduction in hemoglobin and platelet counts continued until the end of the observation period compared to the 25 and 50 U/kg PCC groups (table 1).

Animals in the 12.5 U/kg PCC or 12.5 U/kg PCC plus tranexamic acid groups developed severe shock after injury, with low mean arterial pressure attributable to ongoing blood loss 2 h after trauma induction (See table S1, Supplemental Digital Content, https://links.lww.com/ALN/C667, Hemodynamic variables). Animals in the 12.5 U/kg PCC or 12.5 U/kg PCC plus tranexamic acid groups also had lower platelet counts than animals in the 12.5 U/kg PCC plus tranexamic acid and fibrinogen concentrate group (table 1). Endogenous thrombin potential in these animals was restored above baseline immediately after PCC administration but decreased slightly over time (fig. 4B).

After administering fibrinogen concentrate (80 mg/kg) in the 12.5 U/kg PCC plus tranexamic acid and fibrinogen concentrate group, plasma fibrinogen concentrations were restored to baseline and remained significantly higher than in all other groups up to 240 min postinjury (all P < 0.0001; fig. 3B). Similarly, extrinsic thromboelastometry maximum clot firmness was restored close to the baseline (fig. 2D). Although the total blood loss was lower significantly in the 12.5 U/kg PCC plus tranexamic acid and fibrinogen concentrate group compared with the other groups in the second part of the study (P < 0.0001 vs. control, P = 0.001 vs. 12.5 U/kg PCC, P < 0.001 vs. 12.5 U/kg PCC plus tranexamic acid; fig. 5), endogenous thrombin potential and peak height were comparable with the 12.5 U/kg PCC or 12.5 U/kg PCC plus tranexamic acid groups over time. However, peak height was significantly lower at 240 min after trauma in animals treated with 12.5 U/kg PCC compared with those in the 12.5 U/kg PCC plus tranexamic acid and fibrinogen concentrate (P = 0.007; fig. 4, B and D).

The histopathologic examination of injured liver sections revealed homogenous tissue damage and comparable lacerations in all animals. No thromboembolic or other remarkable pathologic changes were present in kidneys, lungs, heart, or nontraumatized liver tissue.

This study shows that PCC at 25 and 50 U/kg can effectively control bleeding and restore hemostasis under rivaroxaban anticoagulation in a clinically relevant experimental polytrauma animal model. Administration of PCC at a lower dose of 12.5 U/kg, even though it reduced the blood loss postinjury compared with control animals, was not effective in reversing the anticoagulant effects of rivaroxaban either as monotherapy (12.5 U/kg) or in combination with tranexamic acid. These results imply that a threshold level of coagulation factors or restoration of thrombin generation is needed to overcome rivaroxaban-induced thrombin inhibition. This level depends on the concentration of rivaroxaban and PCC. The addition of fibrinogen plus tranexamic acid to a low dose (12.5 U/kg) of PCC could compensate for this lack of efficacy.

PCC, at appropriate doses (25 or 50 U/kg), was effective in treating coagulopathy without excessive activation of coagulation at the level of anticoagulation used in this model. This is in agreement with several previous animal studies, as well as limited experience in humans.^5,24–26  In trauma-associated bleeding with rivaroxaban anticoagulation, PCC-mediated increase in factor X and factor II levels are increasing to the point that thrombin generation should, theoretically, overcome the inhibitory effects of rivaroxaban. This observation is supported by the inefficacy of the lower dose of PCC (12.5 U/kg) to overcome the rivaroxaban-induced coagulopathy in our porcine model. However, the exact mechanism whereby PCC may improve hemostasis under factor Xa inhibition is not fully understood. The mechanism of action can probably not be explained by a simple stoichiometric reaction. On a molar basis, the concentration of factor X in PCC is not sufficient to overcome the antagonizing effect of the factor Xa inhibitor. Instead, PCCs may restore hemostasis in patients on direct oral anticoagulants because they increase not only factor X levels but also factor II, VII, and IX concentrations at the site of injury and may exert a compensatory prohemostatic effect with increased thrombin generation potential. However, PCC’s efficacy to restore hemostasis is also dependent on different factors such as clinical status, the hemostatic system of the species under investigation, trauma severity, blood loss, and coagulopathy status.

PCC’s benefit for managing factor Xa inhibitor–associated bleeding was explored in animal models of acute hemorrhage using different injury approaches and spiking experiments in blood from human volunteers.^25–33  These studies have also shown a dose-dependent effect of PCC on the reversal of factor Xa inhibitors. The observational multicenter cohort study Unactivated Prothrombin complex concentrates for the Reversal of Anti-factor Ten inhibitors (UPRATE), which was performed in Sweden and Canada, investigated the efficacy of a four-factor PCC at a median dose of 25 U/kg to restore hemostasis in rivaroxaban- or apixaban-anticoagulated patients with major acute bleeding (mainly intracerebral hemorrhage and gastrointestinal bleeding) defined according to the International Society on Thrombosis and Haemostasis.^34,35  In both cohorts, 68 and 69% of patients showed “effective” hemostasis after four-factor PCC therapy. However, due to the lack of a control group and the lack of factor Xa activity measurements, it is difficult to draw definitive conclusions about PCC therapy’s efficacy and appropriate PCC dosing.

In our model, the causes of coagulopathy included consumption and dilution of coagulation factors, platelets, and impaired thrombin generation. As reported above, a PCC dose of 25 U/kg was shown to restore hemostasis effectively. Similarly, the combination of tranexamic acid and fibrinogen concentrate plus a low dose of PCC at 12.5 U/kg restored hemostasis. In contrast, tranexamic acid plus a low dose of PCC at 12.5 U/kg had no impact on blood loss. From a mechanistic perspective, administration of fibrinogen concentrate increased the plasma fibrinogen levels to enable fibrin clot formation.³⁶  At the same time, tranexamic acid inhibits the fibrinolysis pathway and attenuates clot degradation.^36,37 

In major blood loss, fibrinogen is one of the first coagulation factors decreasing below the critical level (100 mg/dl).^38,39  The critical minimum concentration of fibrinogen to maintain hemostasis is a matter of debate. The recent European guidelines recommend threshold concentrations of 150 to 200 mg/dl relating to perioperative bleeding and trauma.⁴  Plasma concentrations of fibrinogen after trauma were below this threshold in our trauma model. Therefore, when treating trauma patients who are receiving rivaroxaban therapy, it is possible that the early administration of fibrinogen and antifibrinolytic medication could reduce the dose of PCC that is required for restoring hemostasis.

The propensity of PCCs to increase thrombin generation is the key to their efficacy but may also increase the risk of thromboembolic events. Kinetic parameters of thrombin generation potential (endogenous thrombin potential and peak height) and coagulation (D-dimers) were increased above baseline after PCC therapy. However, no adverse clinical signs or histopathologic signs of prothrombotic changes were observed in the current study. Excessive concentrations of prothrombin (factor II) to antithrombin have been suggested to cause thromboembolic complications.⁴⁰  In an experimental animal model of liver injury without anticoagulation, it was reported that 50 U/kg PCC might increase the risk of disseminated intravascular coagulation, attributable to an imbalance of procoagulant and anticoagulant proteins.^22,40  These data are in line with those from an observational, clinical study of four-factor PCC in severely injured patients.¹⁰  In contrast, in experiments in a similar animal model of acute injury, PCC doses as high as 100 U/kg were well tolerated in the presence of dabigatran anticoagulation. These data imply that PCC’s impact might be different in clinical situations with and without previous anticoagulation.

Similarly, in eight uncontrolled studies using PCC for factor Xa inhibitor–associated major bleeds, the crude pooled incidence of thromboembolic events was 5%.⁴¹  Therefore, it seems likely that an intrinsic thromboembolic event rate of 4 to 5% in these specific patients has to be accepted when oral anticoagulants are reversed for major bleeding. Conclusively, there is reason to believe that the use of 25 to 50 U/kg PCC for management of factor Xa inhibitor–associated bleeding has a positive effect without significantly increasing the risk of thromboembolic events. The current study suggests a comparable effect and safety profile of either treatment option. Follow-up investigations are needed to determine whether a multimodel therapy may provide an improved safety profile as compared to higher doses of PCC monotherapy during extended follow-up periods.

At present, no preclinical or clinical studies have investigated the efficacy of PCC versus andexanet alfa to control severe bleeding and/or restore hemostasis under direct oral anticoagulant therapy in a direct head-to-head comparison. Previous studies in the pig polytrauma model, however, evaluated the effect of andexanet alfa versus vehicle control (bolus dose of 1,000 mg, or bolus of 1,000 mg plus infusion of 1,200 mg over 2 h) to reverse apixaban-associated bleeding.⁴²  It could be shown that andexanet alfa significantly reduced total blood loss posttrauma and achieved a 100% survival rate compared with control animals, similar to the results seen in the current study. A more detailed cross-study comparison to the effects of PCC versus andexanet alfa is hindered by key differences in study designs, including oral versus systemic administration and anticoagulation via apixaban versus rivaroxaban. Follow-up studies are therefore required to allow direct statistical comparisons. Due to the high costs of andexanet alfa, limited availability, and improvement of hemostasis not yet being established,⁴³  clinical studies, such as A Phase 4 Randomized Clinical Trial Of AndexXa (Portola Pharmaceuticals, Inc., USA) In Acute Intracranial Hemorrhage In Patients Receiving An Oral Factor Xa Inhibitor (ANNEXA-I), investigating andexnat alfa versus PCC, are urgently needed to guide clinical therapy.

This animal model has its limitations. In this study, we used only males to keep the groups as homogenous as possible at baseline to minimize the SD. The injury was induced in anesthetized healthy animals. To obtain therapeutic plasma concentrations of rivaroxaban, intravenous infusion of rivaroxaban was administered for 30 min. Therefore, the observed results may not be directly applicable in clinical settings. Because our study used a standardized trauma model, it may not show the differences between bleeding sites, patient’s physiologic responses, and drug–drug interactions. The absence of thromboembolic events is limited by the observation time of 4 h.

In conclusion, this study investigated effective hemostatic control mediated by either monotherapy of PCC or a multimodal regimen with low-dose PCC plus additional tranexamic acid and fibrinogen concentrate in the presence of rivaroxaban anticoagulation in a severe bleeding polytrauma setting. The primary outcomes of this study demonstrates that the utility of a high dose of PCC greater than 25 U/kg may not have any additional advantage in significantly reducing the bleeding in trauma complications. Additionally, coadministration of a low dose of 12.5 U/kg PCC with tranexamic acid and fibrinogen concentrate compared with a higher dose of PCC can be considered as an alternative therapeutic option. The findings also confirm that PCC may be more efficient as part of a multimodal therapy than monotherapy in restoring hemostasis and balancing thrombin generation.

The authors thank David Coppenheur, M.Sc., and Kelly Waagemeester, M.Sc. (Department of Anesthesiology, RWTH Aachen University Hospital, Aachen, North Rhine-Westphalia, Germany), for laboratory assistance. In addition, the assistance provided by professional biostatisticians Marcel Mischnik, Ph.D. (Department of Research, CSL Behring GmbH, Marburg, Germany), and Oliver Ghobrial, Ph.D. (Department of Pharmacology, CSL Behring, King of Prussia, Pennsylvania), in reviewing this study’s statistical analysis is greatly appreciated.

Supported by Bayer AG (Wuppertal, Germany) and CSL Behring (Marburg, Germany).

Dr. Grottke has received research funding from Bayer AG (Wuppertal, Germany), Biotest (Dreieich, Germany), Boehringer Ingelheim (Ingelheim, Germany), CSL Behring (Marburg, Germany), Octapharma (Lachen, Switzerland), Novo Nordisk (Bagsværd, Denmark), Nycomed (Konstanz, Germany), and Werfen (Munich, Germany). He has also received honoraria for lectures and consultancy support from Baxalta (Unterschleißheim, Germany), Bayer AG, Boehringer Ingelheim, Ferring (Kiel, Germany), CSL Behring, Octapharma, Pfizer (Berlin, Germany), Takeda (Zurich, Switzerland), Portola (San Francisco, California), and Sanofi (Berlin, Germany). Dr. Schöchl has received speaker fees from CSL Behring, TEM International (Munich, Germany), Haemonetics (Signy-Centre, Switzerland), and Stago (Asnieres, France). Dr. Rossaint has received honoraria for lectures and consultancy from CSL Behring, Bayer AG, Boehringer Ingelheim, and Novo Nordisk (Mainz, Germany). Dr. Herzog is an employee of CSL Behring (King of Prussia, Pennsylvania). Dr. Heitmeier is an employee of Bayer AG. The other authors declare no competing interests.