Clinical review: Clinical management of new oral anticoagulants: a structured review with emphasis on the reversal of bleeding complications

New oral anticoagulants, including dabigatran, rivaroxaban, and apixaban, have been recently approved for primary and secondary prophylaxis of thromboembolic conditions. However, there is no clear strategy for managing and reversing their anticoagulant effects. We aimed to summarize the available evidence for clinical management and reversal of bleeding associated with new oral anticoagulants. Using a systematic review approach, we aimed to identify studies describing reversal strategies for dabigatran, rivaroxaban, and apixaban. The search was conducted using Medline, EMBASE, HealthSTAR, and grey literature. We included laboratory and human studies. We included 23 studies reported in 37 out of 106 potentially relevant references. Four studies were conducted in humans and the rest were in vitro and in vivo studies. The majority of the studies evaluated the use of prothrombinase complex concentrate (PCC), either activated or inactivated, and recombinant activated factor VII (rFVIIa). Other interventions were also identified. Laboratory studies suggest that hemostatic parameters and bleeding might be partially or completely corrected by PCC for rivaroxaban better than dabigatran. Studies in humans suggest that PCC might reverse the effects of rivaroxaban better than dabigatran assessed by hemostatic tests. We were not able to locate studies evaluating the clinical efficacy of these agents. The best available evidence suggests that PCC (activated or inactivated) might be the best option for reversing new anticoagulants. Evidence for rFVIIa is less compelling. There might be differences in the efficacy of reversing agents for different anticoagulants. Studies assessing the clinical efficacy of these reversal agents are urgently needed.


Clinical considerations in the management of new oral anticoagulants
Hemostasis is a cautiously balanced process that teeters between thrombus formation and degradation and that consists of a series of enzyme-mediated reactions in which thrombin plays a central role ( Figure 1). A basic understanding of the physiology of the hemostatic system might help when considering the selection of a reversal agent. It is theoretically possible that an agent such as activated prothrombinase complex concentrate (aPCC) is a better option when treating bleeding in patients who are receiving dabigatran, because it might provide a more effi cient activation of the coagulation cascade, although it is also possible that if dabigatran is still present in plasma, it could neutralize the newly generated thrombin. In the case of rivaroxaban, it is also conceivable that PCC or aPCC might be eff ective since they would bypass the eff ect of FXa inhibition. However, it should be underscored that these considerations remain theoretical and are clinically unproven. It is also possible that pro-hemostatic agents lessen NOAC-related bleeding by over coming the inhibitory eff ect of these drugs by means of providing massive amounts of factors II and X, either active or inactive. In general, it should be recognized that pro-hemostatic agents are not antidotes to NOACs and do not aff ect the ongoing inhibitory eff ect of these drugs on factors IIa and Xa. Furthermore, it should be noted that previous studies in patients with intracerebral hemorrhage have suggested that, although pro-hemo static agents can limit the extent of bleeding, their eff ect on mortality and disability appears minimal [5]. It should also be recognized that there is a small but clinically important risk when administering these agents as their use has been associated with an increased risk for venous and arterial thrombosis [6][7][8][9][10]. The coagulation response is triggered by tissue damage and exposure of components of the extracellular matrix, including collagen and tissue factor, which will activate factor VII and form a complex that triggers the activation of factors X and IX. This initiation phase will be rapidly inactivated by the tissue factor pathway inhibitor. The generation of small amounts of thrombin from prothrombin by the initial activation of factor X will in turn activate factor XI and the non-enzymatic factors VIII and V, resulting in a rapid amplifi cation of the coagulation response with the subsequent generation of large quantities of thrombin and the subsequent conversion of fi brinogen to fi brin. The steps of the coagulation system that are aff ected by the new anticoagulant drugs are shown in boxes. Roman numerals indicate coagulation factors; a subindex 'a' indicates an activated coagulation factor. TF, tissue factor.
Blood products such as cryoprecipitate and fresh frozen plasma are unlikely to result in a clinically signi ficant reversal. We were unable to fi nd evidence supporting the use of anti-fi brinolytic agents such as aminocaproic and tranexamic acid. A number of new agents are currently in diff erent phases of development. Th ey include monoclonal antibodies and small molecules with specifi c affi nity for the NOACs [11][12][13]. Clinical studies are eagerly awaited.
A number of practical considerations must be made when managing bleeding in patients on anticoagulants. It is necessary to understand the two diff erent types of bleeding events often reported in the literature evaluating anticoagulants for diff erent indications. A major bleed is usually defi ned as a fatal or symptomatic event that involves a critical site (intracranial, intraspinal, intraocular, retroperitoneal, intra-articular, pericardial, or intra muscular with compartment syndrome) or causes a hemoglobin level decrease of 20 g/L or more within 24 hours or results in transfusion of more than two units of packed red blood cells. Th e most common sites of major bleeding are gastrointestinal (approximately 80% of all bleeds), genitourinary, intracranial (including subdural, intracerebral (worst prognosis: 45% to 50% case fatality) [14,15], and subarachnoid), and soft tissue and intra-articular, which are typically traumatic. In contrast, minor bleedings are those that may require medical attention but, in general, do not require transfusion or hospitali zation, are usually self-limiting, and do not require interruption of anticoagulant medication, although this could be necessary in cases of non-self-limiting events.
In the case of patients requiring an invasive procedure, temporary interruption of NOACs might be considered, but this should be individualized according to the type of procedure required. It should be remembered that NOACs have much shorter half-lives than warfarin (38 to 42 hours): dabigatran (11 to 17 hours), rivaroxaban (5 to 9 hours), and apixaban (9 to 12 hours). Th us, their anticoagulant eff ects may be reduced by over 75% within 24 to 36 hours of the last NOAC dose, after two half-lives have elapsed. Th us, even if coagulation tests are normal, elective procedures should not be performed until at least 24 to 36 hours (or longer in renal dysfunction) after discontinuation. Paradoxically, however, their short halflives might put the patients at a higher risk of thrombosis in the case of interruptions, as was suggested by the slight increase in thrombotic events in patients enrolled in the ROCKET AF trial when transitioning from rivaroxaban to warfarin at the end of the study, although this remains largely a theoretical concern [16].
In the case of patients with kidney disease, dose reductions are required in patients with a creatinine clearance (CrCl) of 30 to 50 mL/min because renal impairment prolongs the half-lives of NOACs, especially dabigatran. Th e use of NOACs is contraindicated in patients with a CrCl of less than 30 mL/min in the case of dabigatran or a CrCl of less than 15 mL/min for rivaroxaban and apixaban [2,17,18]. For patients with a CrCl of between 15 and 30 mL/min, no clear dosing recommendations exist. In bleeding patients receiving dabigatran, hemodialysis may be considered in cases of acute renal failure and laboratory evidence of an excessive anticoagulant eff ect since 80% of dabigatran is not plasma-bound. Th ere is no available information for rivaroxaban or apixaban in this regard, although owing to their high protein binding, dialysis is unlikely to be very eff ective. One study showed that hemodialysis removes 62% to 68% of drug after a single 50 mg dabigatran dose in patients with end-stage renal disease [19]. Another study suggests that dabigatran could be removed with the use of dialysis and charcoal hemoperfusion columns [20]. However, it is not known whether the removal of dabigatran by dialysis results in better bleeding control clinically. Finally, NOACs should be avoided in patients with moderate-to-severe liver dysfunction because of the lack of data in such patients.
Concomitant use of P-glycoprotein and CYP3A4 inhibitors (for example, azole antifungals such as ketocona zole) or inducers (for example, rifampicin and antiepileptics) may increase or decrease both drugs' anticoagulant eff ect, and their concurrent use is contraindicated (Table 1).
Finally, caution should be exerted when interpreting coagulation tests since all NOACs may alter common coagulation laboratory tests ( Table 2). Although routine coagulation tests generally are not useful in evaluating the degree of anticoagulation in patients receiving NOACs, there is some evidence that, in a bleeding patient, the activated partial thromboplastin time (aPTT) may be useful in determining anticoagulant activity. Lindahl and colleagues [21] demonstrated that a prolonged aPTT of greater than 90 seconds and a prolonged quick prothrombin time (PT) may indicate overdosing or accumulation of dabigatran. However, it is very important to note that a normal aPTT or international normalized ratio result may not exclude the presence of clinically relevant levels of dabigatran and other NOACs. In addition, fi brinogen concentration can be falsely low in patients receiving dabigatran and this may lead to the unnecessary use of blood products such as cryoprecipitate [21]. Finally, although it has been noted that a normal thrombin time may indicate a lack of dabigatran activity, this test is extremely sensitive to very small amounts of the drug, rendering it unsuitable for clinical use. Owing to their predictable pharmacokinetic and pharma codynamic profi les, NOACs do not require routine laboratory monitoring, but this does not preclude the need to monitor specifi c coagulation-related conditions related to bleeding complications or changes in bleeding risk factors such as renal function.

Management of bleeding associated with new oral anticoagulants
To evaluate the interventions so far tested for managing bleeding in patients taking NOACs, we used a structured approach. We aimed to retrieve randomized controlled trials, cohort studies, case control studies, case seriescase reports (including self-poisoning), or studies conducted in animal models or in vitro. To be considered for inclusion, a study had to propose an intervention to reverse the anticoagulant eff ect of the drugs of interest and measure either clinical or laboratory indices of bleeding.
Th e search was conducted in November 2012 by using the exploded terms rivaroxaban, dabigatran, and apixaban and the textwords 'reverse' or 'reversal' . Th e search was conducted by using the OVID interface in the following electronic databases: MEDLINE, EMBASE, and Healthstar. Th e search was restricted to articles written in English or French. Th e retrieved references were assessed for possible inclusion on the basis of the evaluation of the title and the abstract (or the full text if no abstract was available). Letters to the editor, review articles, editorials, and commentaries were excluded. Grey literature was con sidered, and we included the electronic versions of the abstracts of the following major international meetings: the International Society on Th rombosis and Haemo stasis (1999 to 2011) and the American Society of Hema tology (1999 to 2012). In addition, the reference lists of the retrieved journal articles were reviewed to locate additional studies of potential interest. Th e search was limited to articles published after 1996 since no infor mation on these drugs was available prior to this date.
Potentially relevant studies were selected by one author and checked for accuracy by a second author. Owing to the heterogeneous nature of the studies included and to the lack of validated scales to assess the quality of the study designs included, we did not perform a quality assessment. Reports were divided into the following categories: (a) in vitro or in vivo studies (or both) in animal models and (b) studies in humans. Outcome was the eff ectiveness or potential eff ectiveness of the proposed interventions to revert NOACs. Owing to the heterogeneity among the studies, a formal meta-analysis was not conducted.
Th e search identifi ed 106 potentially relevant references, of which 35 were included in the fi nal review. Two additional references were identifi ed through the review of reference lists. Excluded references included two mechanistic studies and one letter, whereas the rest were reviews. One additional study was excluded because it did not provide suffi cient information [22]. We identifi ed 23 studies reported in 37 references [11][12][13]20,. Nine studies were reported in more than one reference, and only fi ve studies have been published in full, whereas the rest have been reported only in conference abstracts. Reversal of apixaban, dabigatran, and rivaroxaban has been evaluated in two, nine, and eight studies, respectively. Four studies have evaluated reversal of both rivaroxaban and dabigatran administered to human subjects, and only one has evaluated the administration of the proposed reversal agent to healthy volunteers. Nine studies included the evaluation of an in vivo model of bleeding in animals, whereas all studies evaluated diff erent hemostatic parameters. Th e interventions evaluated included recombinant human activated factor VII (rFVIIa) (NovoSeven), four-factor prothrombinase complex (PCC) (Beriplex, Octaplex, Kaskadil, Cofact, and Kanokad), activated prothrombinase complex (aPCC) (FEIBA), fi brinogen concentrate, fIX/fX concentrate and fresh frozen plasma (FFP), high-purity FX concentrate, activated charcoal hemoperfusion, hemodialysis, an antidabigatran antibody/Fab fragment, and a recombinant reconstructed mutant factor Xa (r-Antidote, PRT064445). Th e 23 included studies are summarized in Table 3. General principles should be observed for managing bleeding, including resuscitation, bleeding source control, drug cessation, and appropriate use of blood products. Activated charcoal can be considered if the off ending drug was taken within two hours [56]. In the case of life-threatening or intracranial bleeding, the use of pro-hemostatic agents should be considered. Table 4 summarizes the currently available information regarding the reversal of NOACs, all of which is based on limited studies and should be considered low-quality. Th e best evidence so far comes from four studies in healthy volunteers who were receiving NOACs. Th e fi rst study included healthy volunteers to whom rivaroxaban (20 mg twice a day) or dabigatran (150 mg twice a day) was administered for 2.5 days, after which they received a single dose of 50 IU/kg of four-factor (II, VII, IX, X) PCC, and its eff ects on PT, aPTT, thrombin time (TT), ecarin clotting time (ECT), and endogenous thrombin potential (ETP) were evaluated [26]. Th e study used Cofact, an agent available in Europe, but other similar four-factor PCCs (such as Octaplex or Beriplex) are available. Th e study showed that in subjects receiving rivaroxaban, both the PT and the ETP corrected to normal levels after administration of the PCC. In contrast, in patients receiving dabigatran, there were no corrections in the aPTT, ECT, ETP, or TT. In three other studies including healthy volunteers, participants received one or two doses of rivaroxaban (20 mg) or dabigatran (150 mg), and samples were obtained at diff erent times after administration [33]. To test the ex vivo eff ect of pro-hemostatic agents, patients' samples were incubated with diff erent concentrations of recombinant activated factor VII (rFVIIa, Novoseven), four-factor PCC (Kanokad, available in Europe; Beriplex; and Cofact), and activated fourfactor PCC (aPCC) (FEIBA; similar to four-factor PCC but with a more rapid eff ect because of the presence of 'activated' clotting factors). Coagulation was evaluated by measuring thrombin generation and other tests. Th e results of the studies showed that overall aPCC achieved the best correction of the coagulation parameters after rivaroxaban and dabigatran administration. Less effi cient corrections were achieved with PCC and rFVIIa. Interest ingly, one study suggested that lower doses of FEIBA might also be eff ective [33].
Bleeding has been assessed in some studies in animal models. A study evaluating the eff ects of rFVIIa and PCC in a rabbit bleeding model failed to show reversal of rivaroxaban-induced bleeding in spite of improvement in laboratory parameters [29]. However, other studies have reported reduced bleeding time in animal models of rivaroxaban-induced bleeding after administration of FEIBA [31,39]. In the case of dabigatran, animal models suggest that PCC, FEIBA, and rFVIIa could reduce bleeding [41,43,46]. Th ere are other agents, such as threefactor (II, IX, X) PCC (currently used in the US), but evidence supporting their use in bleeding patients receiving NOACs is lacking and their use cannot be supported on the basis of currently available evidence. Perhaps the most signifi cant in vivo animal study was published by Zhou and colleagues [46], who evaluated the eff ect of rFVIIA and PCC in an animal model of intracerebral bleeding in mice receiving dabigatran. Th e authors found that PCC at a dose of 100 IU/kg decreased the size and deterred the progression of the intraparenchymal hematoma and resulted in a lower mortality in the treated animals. Finally, only one study has described the experience with bleeding events in trials evaluating dabigatran for diff erent indications [22]. Th e study reviewed 1,121 reports of major bleeding events and found that patients receiving dabigatran who developed major bleeding were older, had a lower creatinine clearance, and had more concurrent use of antiplatelet agents than those receiving warfarin. However, only nine patients received PCCs and two received rFVIIA, but no details were available at the time of review, and nothing can be extrapolated to the clinical scenario on the basis of the available information.
Finally, it should be re-emphasized that all information on clinical management should be considered low-quality on the basis of currently available data. When considering the use of these reversal agents, clinicians should very carefully consider the trade-off between benefi ts and potential risk.

Conclusions
In spite of the emerging information, there is still a considerable lack of evidence to inform clinical practice, and most of the recommendations are based on expert opinion and interpretation of available evidence. Whereas experience today suggests that in the case of a catastrophic event such as an intracerebral bleeding the use of reversal agents is unlikely to result in major clinical benefi t, such agents still have a role in the case of patients requiring urgent surgical interventions which are potentially life-saving. Th e overall evaluation of the limited available studies suggests that the agents most likely resulting in benefi t are PCC and aPCC but evidence for rFVIIa is less consistent. However, it is essential to note that no clinical experience is available to date and that all studies have used surrogate measures such as coagu lation-based assays of hemostatic function or animal models. Th e actual eff ect of these agents in patients with bleeding events remains to be determined. In addition, the existing literature suggests that correction in hemostatic parameters does not necessarily correlate with an improvement in actual bleeding, and the currently available evidence is still confusing because of  the diversity of experimental models and assessments used and because of the lack of conclusive evidence relating the correction of laboratory parameters evaluating the hemostatic system with the eff ect on actual bleeding events. Given the lack of eff ective reversal agents, several new agents are currently being evaluated, the most promising of which are the monoclonal humanized anti-dabigatran antibody and Fab fragment and the r-Antidote (PRT064445) for direct Xa inhibitors. To the best of our knowledge, the latter drug is the only one that has achieved clinical development and is currently being evaluated in a phase II study, the results of which could be expected in late 2014. Further studies are urgently needed to elucidate the clinical eff ect of the currently available and prospective agents on patients with actual bleeding events and in particular on vital and functional outcomes. aPCC and rFVIIa showed reversal of both anticoagulants assessed by TG. Cofact showed improvement on TG but Beriplex did not.
anti-fXa, anti-activated factor X activity; aPCC, activated prothrombinase complex concentrate; aPTT, activated partial thromboplastin time; AUC, area under the curve; bid, twice daily; BT, bleeding time; ECT, ecarin clotting time; ETP, endogenous thrombin potential; FFP, fresh frozen plasma; fIX, coagulation factor IX; fX, coagulation factor X; INR, international normalized ratio; IV, intravenous; PCC, prothrombin complex concentrate; po, by mouth; PT prothrombin time; rFVIIA, recombinant activated factor VII; TAT, thrombin-antithrombin complex; TE, thromboelastography; TG, thrombin generation assay; TT, thrombin time. No study has assessed the clinical eff ect of these agents in patients with active bleeding events. The possible role of these agents is based on animal studies and human studies evaluating surrogate coagulation markers. All evidence should be considered low quality and the use of these agents should follow a careful consideration of risks and benefi ts. a No study has assessed the clinical eff ect of these agents in patients receiving apixaban. The possible role of these agents is theoretical and is based on extrapolation of evidence available for patients receiving rivaroxaban. b There is no available evidence regarding effi cacy or safety. Adapted from [23,26,29,30,33,39,41,43,46].