Management of rivaroxaban- or apixaban-associated major bleeding with prothrombin complex concentrates: a cohort study

Rivaroxaban and apixaban are direct acting factor Xa inhibitors. They have been evaluated in several large, randomized phase III trials for stroke prevention in patients with atrial fibrillation (SPAF)^1-3  and for the prevention and management of venous thromboembolism (VTE).^4-7  In these trials, rivaroxaban and apixaban demonstrated noninferior or superior efficacy with similar or superior safety to the current standard of care (warfarin with or without low–molecular weight heparin, depending on the indication). Consequently, rivaroxaban and apixaban are now approved in many countries for SPAF and for the management of patients with VTE.

Bleeding is a serious complication of treatment with anticoagulants.⁸  The reported annual rates of major bleeding with rivaroxaban or apixaban in the phase III trials for SPAF were 3.6% and 1.4%, respectively.^1-3  The corresponding rates in the phase III VTE trials during the average 6-month treatment period with rivaroxaban or apixaban were 1.1% and 0.6%, respectively.^4-7  The current strategies for the management of bleeding events in patients on these agents include the administration of activated charcoal and circulatory support through the administration of blood products until the anticoagulant effect weans off. In cases of more severe bleeding, the suggested strategies include the administration of nonactivated prothrombin complex concentrates (PCCs), activated prothrombin complex concentrates (APCCs), or recombinant factor VIIa (rFVIIa).⁹  Andexanet alfa, a specific reversal agent for factor Xa inhibitors, has shown promising results in a phase III trial, but was associated with a high incidence of thromboembolic events and is still not available for clinical use.¹⁰ 

Several guidelines recommend the administration of PCCs for the management of major bleeding events (MBEs) in patients on rivaroxaban or apixaban.^11,12  Animal studies have supported the effectiveness of PCCs in this setting.¹³  Studies on healthy volunteers have demonstrated the reversal of laboratory abnormalities and bleeding time induced by Xa inhibitors after the administration of PCCs.^14-16  However, apart from 1 small case series reporting on the outcome of 18 patients with rivaroxaban- or apixaban-related intracranial hemorrhage (ICH) treated with PCCs,¹⁷  there are no data to support the use of PCCs in patients with bleeding complications on these anticoagulants. Furthermore, the use of PCCs might tip the hemostatic balance toward hypercoagulability and might be associated with thromboembolic events. Other factors that might contribute to thromboembolic events in these patients include an increased background risk of thromboembolism, surgical trauma, and interruption of anticoagulation. This is described in the setting of warfarin reversal with PCCs,¹⁸  but comparable data on the safety of PCCs for the reversal of rivaroxaban or apixaban are lacking.

The aim of the UPRATE study (Unactivated Prothrombin complex concentrates for the Reversal of Anti-factor TEn inhibitors) was to assess the efficacy and safety of 4-factor PCCs for the reversal of the anticoagulant effect of rivaroxaban and apixaban in a multicenter, prospective setting.

The study was coordinated through the Coagulation Unit of the Karolinska University Hospital, and cases were recruited from several hospitals in Sweden where 4-factor PCCs were readily available. The study was approved by the Swedish Medical Products Agency and the regional ethics board in Stockholm, and conducted in accordance with the principles set out in the Declaration of Helsinki. The regional ethics board waived the requirement for written patient consent.

Patients were eligible for inclusion if they needed treatment with a PCC for the management of acute and active major bleeding while on rivaroxaban or apixaban (“active” indicating that there is evidence of blood loss or decreasing hemoglobin or, for ICH, deteriorating neurological symptoms during the past ≤48 hours with radiological evidence of ICH). The definition of MBE was according to the International Society of Thrombosis and Hemostasis (ISTH) definition for major bleeding in nonsurgical patients.¹⁹  The patients had to have taken the last dose of apixaban or rivaroxaban within 24 hours to be included in the study to ensure that the bleeding was related to circulating levels of the anticoagulant in the blood. The exclusion criteria were patients with a do-not-resuscitate order, patients with a drop of hemoglobin without an evident source of bleeding, preoperative reversal of rivaroxaban or apixaban anticoagulation effects, and patients with acute coronary syndrome or ischemic stroke within the past 30 days. Patients were excluded if they received other hemostatic agents (rFVIIa or activated PCC) prior to the administration of a PCC.

The Coagulation Unit at Karolinska University Hospital serves as a tertiary referral center for two-thirds of Sweden, providing advice on the management of cases with complex coagulation and thrombosis issues, including that of bleeding on rivaroxaban or apixaban. Consecutive patients from 25 hospitals between 1 January 2014 and 1 October 2016 were included prospectively. The Coagulation Unit was consulted for cases with MBEs on rivaroxaban or apixaban where prompt reversal of the anticoagulant effect was considered for bleeding control. Whenever the Coagulation Unit recommended treatment with a PCC, the patient was included in the study.

The investigators at Karolinska University Hospital recommended a PCC based on a predefined protocol: 1500 or 2000 IU were given for patients with body weights less or more than 65 kg, respectively. An additional dose of a PCC was allowed at the discretion of the treating physician if the bleeding did not stop after the first PCC dose. This protocol was based on an approximation of a PCC dose equivalent to 25 IU/kg.

At 30 days posttreatment with PCCs, the medical records of the patients were retrieved, including the medical notes, transfusion records, laboratory results, radiological examinations, and any other relevant details, including surgical notes. Data were collected on the patients’ demographics, the type of and indication for anticoagulation, the timing of the last dose, concomitant medication that could have influenced the outcome, length of stay, and discharge destination, the occurrence of thromboembolic events after PCC administration, and death.

The effectiveness assessment was based on the criteria published in the communication letter from the Standardization Subcommittee on the Control of Anticoagulation of the ISTH.²⁰  The criteria for the assessment of effectiveness of major bleeding management for 4 different bleeding types were presented: visible bleeding, nonvisible bleeding, musculoskeletal bleeding, and ICH. For non-ICH bleeding, data were collected on changes in the hemoglobin level, the need for transfusion of blood products, the need for intervention or surgery to stop the bleeding, and the administration of other hemostatic agents before or after PCCs. For ICH, follow-up computed tomography (CT) within 24 hours, when available, was compared with the initial CT when available. Furthermore, data on the change in the neurological status of the patients and the need for surgical intervention were also collected. The duration of hospital stay and discharge destinations were also recorded. These data were collected from medical notes, laboratory reports, and radiology reports covering the first 30 days from the bleeding event and treatment with PCCs.

The assessment of effectiveness was done continuously throughout the study when the notes were obtained. For each case, the assessment was performed independently by the same 2 coagulation specialists throughout the study to ensure consistency in the adjudication process.

Any discordance in the assessment was resolved by discussion. The details of the ISTH criteria used for the assessment of PCC effectiveness are presented in supplemental Table 1 (available on the Blood Web site).

The primary safety end point was the occurrence of an objectively verified arterial (stroke, myocardial infarction, or arterial thromboembolism) or venous thromboembolism (deep venous thrombosis or pulmonary embolism) after treatment with a PCC. Death was recorded up to 30 days after treatment with a PCC, and the cause of death was collected from medical notes as recorded by the treating physician. An interim analysis was done after 10 cases were included in the study by an independent data safety monitoring board to assess for the occurrence of serious adverse events, including thromboembolism and death.

Continuous variables are presented as medians and interquartile ranges (IQRs) and analyzed with the Mann-Whitney test for skewed distributions. Binomial data are presented as proportions or percentages and were compared by using χ² or Fisher’s exact test. A P value <.05 was considered statistically significant in a 2-sided test.

Altogether, 92 patients were included during the study period. Eight patients were excluded from further analysis, either because a PCC was given for anticoagulation reversal preoperatively (n = 4) or because the patients ultimately did not receive a PCC (n = 4). The basic characteristics of the remaining 84 patients are presented in Table 1. The median age of the study population was 75 years, and 42.9% were females. The indication for anticoagulation was SPAF in the majority of the patients (n = 63; 75.0%). There were 45 MBEs for patients on rivaroxaban and 39 MBEs for patients on apixaban. ICH was the most common type of bleeding requiring reversal with a PCC (n = 59; 70.2%), followed by gastrointestinal (GI) bleeding in 13 (15.5%) patients. The bleeding was classified as traumatic in 26 (31.0%) of the patients.

A total of 84 patients received PCCs for the urgent reversal of the anticoagulant effect of rivaroxaban or apixaban due to a MBE. The international normalized ratio (INR) and activated partial thromboplastin time (APTT) were measured in 71 and 61 patients, respectively, before treatment with a PCC. The APTT and INR were prolonged in 5 (17.2%) and 21 (61.8%) patients on apixaban and in 16 (50.0%) and 25 (67.6%) patients on rivaroxaban, respectively.

The median (IQR) time from the last dose of apixaban or rivaroxaban to treatment with a PCC was 12.5 hours (9-16 hours). The interval from the estimated bleeding time to PCC administration was 6 hours (2-10 hours). A PCC was administered at a median (IQR) dose of 2000 IU (1500-2000 IU) or 26.7 (21.4-29.9 IU/kg). A second PCC dose of 500-1500 IU was given in 3 cases due to an insufficient effect of the initial management with a PCC. The patients received one of the two 4-factor PCCs according to the availability at the treating site: Ocplex (elsewhere called Octaplex; Octapharma) was given to 46.4% of patients, and Confidex (previously known as Beriplex; CSL Behring) was given to 52.4% of patients (Table 2).

Additional management included the transfusion of red blood cell concentrates (before PCC administration: n = 12 patients; 14.3% after PCC administration: n = 18 patients; 21.4%), plasma (before PCC administration: n = 5; 5.9%, and after PCC administration: n = 8 patients; 9.5%), platelets, (after PCC administration: n = 10 patients; 11.9%), and tranexamic acid (n = 56 patients; 66.7%). Additionally, rFVIIa was administered in 1 patient with ICH due to an insufficient effect of initial management (including with a PCC).

After final adjudication, the hemostatic effectiveness of PCC for the management of a rivaroxaban- or apixaban-related MBE was assessed as effective in 58 patients (69.1%) and ineffective in 26 patients (30.9%) (Table 3).

When analyzed according to the bleeding site, the hemostatic effectiveness of PCC was assessed as effective in 43 of the 59 patients with ICH and 15 of the 25 patients with an extracranial bleeding location. No differences were found when comparing the effectiveness of treatment according to bleeding site (P = .30). In 16 patients with ICH, no control CT scan of the brain was done because the patient’s condition improved clinically, and PCC management was assessed as effective based on the clinical picture. Control CT scans were not performed in an additional 9 patients with ICH because the patient’s condition deteriorated quickly, leading to death, whereby the outcome of PCC management was assessed as ineffective.

In patients with traumatic bleeding, management with a PCC was assessed as effective in 19 patients (73.1%) compared with an effective outcome in 39 patients (67.2%) with nontraumatic bleeding, with no statistically significant difference when comparing the 2 groups (P = .8). Sixteen (61.5%) of the 26 patients with ineffective hemostasis after treatment with a PCC had ICH, and 5 (19.2%) had GI bleeding as the indication for reversal (Table 3).

Safety analysis could be performed for all the patients. An independent event adjudication committee assessed the occurrence of thromboembolic events after 10 included patients, and the decision was to continue the study due to the low rates of safety end points. Two patients developed ischemic stroke 5 and 10 days, respectively, after treatment with a PCC (Table 4). These were objectively confirmed with CT scans. The first patient was a 71-year-old female on 5 mg apixaban twice daily for SPAF and had a history of ischemic stroke. She received 2000 IU of a PCC (27 IU/kg) for a large subarachnoid hemorrhage, but developed an ischemic stroke after 5 days and died after 18 days. The cause of death was recorded as ICH and ischemic stroke. The other patient was on 20 mg rivaroxaban once daily for SPAF with a previous history of stroke and received 1500 IU of a PCC (29 IU/kg) for ICH, but developed an ischemic stroke after 10 days and ultimately died 112 days after the initial ICH due to a recurrent ischemic stroke. The characteristics of the patients with thromboembolic events are presented in Table 4.

A diagnosis of pulmonary embolism was suspected in a third patient with ICH (Table 4). Follow-up CT of the brain showed a decreased size of the ICH. However, the patient developed sudden shortness of breath, deteriorated quickly, and died 16 days after the ICH. The cause of death was reported as ICH and complicating pulmonary embolism, however, neither imaging nor autopsy were done to confirm the presence of a pulmonary embolism. Of the 3 previously mentioned patients, only the second patient was on thromboprophylaxis with low–molecular weight heparin at the time of the thromboembolic event.

Death occurred in 15 patients (18%) within the first week after the MBE, of which 13 patients (86.7%) had ICH. Three of these 15 patients had traumatic bleeding. The cause of death in 14 cases was attributed to the direct effect of the bleeding, and 1 patient died of multiorgan failure after GI bleeding. The characteristics of the patients who died within the first week are presented in Table 5. A total of 27 patients (32%) died within 30 days after the MBE, of which 20 (74.1%) had an ICH. Five of these 27 MBEs were classified as traumatic. For the 27 patients who died within 30 days, the cause of death was due to the direct effect of the bleeding in 18 patients (66.7%), sepsis and multiorgan failure in 7 patients (25.9%), and cardiac arrhythmia and arrest in 1 patient (3.7%). The cause of death in all these cases was assessed as not related to treatment with a PCC, apart from 1 patient (mentioned earlier under safety) who had an ischemic stroke 5 days after PCC administration for an ICH, where a possible relation to the PCC could not be excluded. In an additional patient, a suspected pulmonary embolism was reported as the contributing cause of death, although this was not verified, as detailed above.

In this article, we report the results from the hitherto largest cohort of patients treated with PCCs for the management of MBEs associated with rivaroxaban or apixaban. The main results are the overall good effectiveness of PCC in this setting and the low rates of serious adverse events, such as thromboembolic events.

MBEs on anticoagulants are associated with significant morbidity and mortality. Most bleeding events on direct oral anticoagulants can be managed conservatively by hemodynamic support and transfusion of blood products.^21,22  However, in some circumstances, a more active approach may be warranted. A specific reversal agent for anti-factor Xa anticoagulants is still not available in clinical practice, however, PCCs are currently used off-label in this setting. This is based on studies in healthy volunteers, animal models, and in vitro studies.^14-16,23-25  Although these studies suggest that PCCs could be effective in the management of bleeding on factor Xa inhibitors, at present, there is only 1 small clinical study that retrospectively evaluated the effect of PCCs in the management of ICH.¹⁷ 

We used an initial PCC dose of 25 IU/kg (approximated to 2000 IU for an average weight person of ∼80 kg) that could be repeated if necessary. Two studies in healthy volunteers have used higher PCC doses (50 IU/kg) for the reversal of the effect of rivaroxaban.^14,16  However, these volunteers where given double the recommended maintenance dose of rivaroxaban. Because patients on anticoagulant therapy have a higher risk of thromboembolism than healthy volunteers, the strategy used in this study could represent a balance between optimal bleeding control and the risk of adverse thromboembolic events.¹⁸ 

In our study, 30.9% of the patients had an ineffective hemostatic effect of PCC. Most of these patients (61.5%) suffered from an ICH, which is generally associated with poor outcomes and high mortality rates approaching 50%.²⁶  We observed ineffective hemostasis in 5 patients with GI bleeding in the setting of a bleeding vessel that had to be managed endoscopically or by coiling. Arterial bleeding is unlikely to be stopped by PCC infusion only. Therefore, the ineffective hemostatic outcome of PCCs observed in these 5 patients may not necessarily reflect the lack of effect of a given PCC.

The mortality rate reported in this study is higher than that in the Reversal Effects of Idarucizumab on Active Dabigatran (Reverse-AD)²⁷  and the Andexanet Alfa, a Novel Antidote to the Anticoagulation Effects of FXA Inhibitors (ANNEXA-4)²⁸  studies, which probably reflects the higher proportion of ICH in our cohort (70.2%) compared with the other 2 studies (20% and 43%, respectively). In addition, patients referred to our Coagulation Unit for consideration of treatment with a PCC for the reversal of factor Xa inhibitors have generally failed other supportive measures, resulting in patients with poorer prognoses included in our study and adding to the higher mortality rates that we observed.

A specific antidote for the reversal of factor Xa inhibitors, andexanet alfa, was studied in patients with MBE on a factor Xa inhibitor. An excellent-to-good effect was observed in 79% of patients (95% confidence interval, 64% to 89%),²⁸  which is comparable to the effective hemostatic outcome of PCCs observed in our study. Thromboembolic events were, however, observed in 18% of the patients treated with andexanet alfa for the management of MBEs on factor Xa inhibitors.²⁸  This might be explained partly by the higher baseline risk in the study population due to preexisting thrombogenic conditions and the activation of the coagulation system by the bleeding and trauma. The rates of thromboembolism observed in our study are low (2.4%) and are in line with other publications on PCCs for the reversal of vitamin K antagonists.¹⁸ 

In our study, the INR and APTT were within normal limits in 35% and 62% of patients, respectively, raising the question of whether the patients had therapeutic levels of rivaroxaban or apixaban at the time of PCC administration. Studies have shown that the APTT is less sensitive to the effect of rivaroxaban, and that the different prothrombin time (PT) assays showed marked differences.²⁹  Furthermore, APTT and PT were insensitive to the therapeutic effects of apixaban.³⁰  Therefore, normal APTT and INR should not be used as an indicator of the absence of significant levels of these drugs. Patients who were included in the study had to have received the last dose of apixaban or rivaroxaban within 24 hours of PCC administration, which means that measurable levels of these drugs would still be present in the circulation at the time of bleeding and treatment with PCC. Even though anti-factor Xa can be used to detect the residual anticoagulant effect of rivaroxaban or apixaban (and even PT in case of rivaroxaban), these test results were unfortunately not available for patients in this study.

The large number of patients, the prospective inclusion of the patients in the study, the inclusion of patients who received a dose of the anticoagulant within 24 hours before PCC administration and the use of standardized ISTH criteria for the assessment of the effectiveness of PCCs by 2 independent investigators adds to the strength of the study. The relatively large availability of PCCs, its good observed effectiveness, and the low rates of thromboembolic events mean that PCCs are a viable alternative for the management of MBEs on rivaroxaban or apixaban.

The results presented in this study need some careful consideration. First, because the study was an observational study, we were not able to perform radiological or laboratory examinations on some of the patients to support the assessment of effectiveness. As an example, a follow-up CT scan was not done for all patients with ICH, and plasma levels of rivaroxaban or apixaban were not obtained at the time of the bleeding event. In such situations, we relied on the other criteria in the ISTH document for the assessment of the effectiveness of PCCs.²⁰  Two independent investigators assessed the effectiveness of PCCs to add to the objectivity of the assessments. Second, we did not have a control group, managed conservatively by transfusion of blood products only, to be able to assess the additive effect of PCCs. Our study has, however, followed a design that is common to other studies that have investigated the effectiveness of other reversal agents for the management of bleeding events on anticoagulants.^27,28 

We found that the majority of patients treated with PCCs for the management of MBEs on rivaroxaban or apixaban achieved effective bleeding control, with few observed thromboembolic events. Based on the results of our study, we would suggest giving patients with MBEs on rivaroxaban or apixaban an initial PCC dose of 2000 IU, which may be repeated if the effect is suboptimal. Such an approach seems to be associated with an acceptable balance between efficacy and safety of PCCs. Our findings are, however, limited by the absence of a control group in the study.

This work was supported by an unrestricted grant from Octapharma.

The sponsor had no influence on the study design or conduct, the collection and analyses of data, and the final manuscript writing and approval.

Contribution: A.M. contributed to the study concept, drafting the study protocol, collection of data, assessment of management effectiveness, data analysis, and drafting the manuscript; A.Å. contributed to the study concept, collection of data, assessment of management effectiveness, and drafting the manuscript; M.H. contributed to the study concept, collection of data, and drafting the manuscript; M.B., J.O., and T.F. contributed to the collection of data and drafting the manuscript; R.C. assisted in the data safety monitoring board and drafting the manuscript; M.M. and E.-L.H. contributed to the collection of data and drafting the manuscript; and S.S. designed the study concept, assisted in drafting the study protocol, and assisted in the assessment of management effectiveness, data analysis, and drafting the manuscript.

Conflict-of-interest disclosure: A.M. reports receiving grant support/honoraria from Octapharma. S.S. received grant support/honoraria from Boehringer Ingelheim, Octapharma, Baxter, Bayer, Sanofi, and Bristol-Myers-Squibb. The remaining authors declare no competing financial interests.

Correspondence: Ammar Majeed, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Box 281, SE-171 77 Stockholm, Sweden; e-mail: ammar.majeed@ki.se.