Introduction

Anticoagulant agents are often used for prevention and treatment of a wide range of cardiovascular diseases. Most frequently used anticoagulants are heparin or its derivatives, vitamin K antagonists (such as warfarin or coumadin) and antiplatelet agents, including aspirin and thienopyridine derivatives, such as clopidogrel. A myriad of clinical studies demonstrate that these agents (alone or in combination) can prevent or treat acute or chronic thrombo-embolic complications, such as in patients with atrial fibrillation or prosthetic heart valves, after myocardial infarction or ischemic stroke, and in patients with venous thrombosis or pulmonary embolism [1]. The most important complication of treatment with anticoagulants is hemorrhage, which may be serious, may cause long-term debilitating disease, or may even be life-threatening [2]. In well-controlled patients in clinical trials treatment with vitamin K antagonists increases the risk of major bleeding by 0.5%/year and the risk of intracranial hemorrhage by about 0.2%/year [3]. In a very large series of 34,146 patients with acute ischemic coronary syndromes, anticoagulant-associated bleeding is associated with a fivefold increased risk of death during the first 30 days and a 1.5-fold higher mortality between 30 days and 6 months [4]. Major bleeding is an independent predictor of mortality across all subgroups that are analyzed. In some clinical situations the incidence of serious bleeding complications may annihilate or even overwhelm the efficacy of antithrombotic agents, as has been shown in the secondary prevention of patients with ischemic stroke by vitamin K antagonists [5]. Nevertheless, in many situations, clinical studies show a favorable balance between efficacy and safety in favor of anticoagulant treatment. However, if severe bleeding occurs, or if a patient needs to undergo an urgent invasive procedure, such as emergency surgery, it may be required to reverse the anticoagulant effect of the various agents. Depending on the clinical situation, i.e., the severity of the bleeding or the urgency and estimated risk of the invasive procedure, this reversal may take place in a few h, but in some cases immediate reversal is necessary (Table 1). Generally, each (immediate) reversal of anticoagulant treatment needs also to take into consideration the indication for the antithrombotic agents. For example, the interruption of combined aspirin and clopidogrel treatment in a patient in whom an intracoronary stent has been recently inserted will markedly increase the risk of acute stent thrombosis with consequent downstream cardiac ischemia or infarction. Likewise, in a patient with a prosthetic mitral valve and atrial fibrillation, interruption of vitamin K antagonists may increase the risk of valve thrombosis and cerebral or systemic embolism. Each of these specific clinical situations requires a careful and balanced assessment of the benefits and risks of reversing anticoagulants (and potential strategies to keep the period of reversal as short as possible). In this article, we will describe the various strategies to reverse the anticoagulant effect of currently most widely used antithrombotic agents and some new anticoagulants.

Table 1 Commonly used anticoagulants and their antidotes
Full size table

Heparin and low molecular weight (LMW) heparin

Herparin and heparin derivatives act by binding to antithrombin, and thereby about 1,000-fold potentiating the anticoagulant effect of this endogenous inhibitor towards thrombin and factor Xa (and some other coagulation factors). Heparin has a relatively short half-life of about 60–90 min, and therefore the anticoagulant effect of therapeutic doses of heparin will be mostly eliminated at 3–4 h after termination of continuous intravenous administration [6]. The anticoagulant effect of high dose subcutaneous heparin, however, will take a longer time to cease. If a more immediate neutralization of heparin is required, intravenous protamine sulphate is the antidote of choice. Protamine, derived from fish sperm, binds to heparin to form a stable biologically inactive complex. Each milligram of protamine will neutralize approximately 100 units of heparin. Hence, the protamine dose in a patient on a stable therapeutic heparin dose of 1,000–1,250 U/h should be about 25–30 mg (sufficient to block the amount of heparin given in the last 2–3 h). The maximum dose of protamine is 50 mg. Since the half-life of protamine is only about 10 min, the reversal of a therapeutic dose of subcutaneous heparin requires a repeated infusion of protamine sulphate (e.g., repeated after 1 h). The effect of protamine can be monitored by measuring the activated partial thromboplastin time (aPTT), which should normalize after the protamine administration.

The reversal of LMW heparin is more complex, as protamine sulphate will only neutralize the anti-factor IIa activity, and has no or only partial effect on the smaller heparin fragments causing the anti-factor Xa activity of the compound [7, 8]. The net effect of protamine reversal of LMW heparin is not completely clear. There are no clinical studies that have systematically studied this, and small case series and experimental animal studies show contradictory results [8–10]. As the aPTT is not useful as a monitoring assay when using LMW heparin, it can also not be used for the monitoring of the neutralizing effect of protamine. Given the relatively long half-life of LMW heparin, the lack of an adequate strategy to reverse its anticoagulant action may sometimes cause a problem in clinical situations. A practical approach is to give 1 mg of protamine per 100 anti-factor Xa units of LMW heparin given in the last 8 h (where 1 mg of enoxaparin equals 100 anti-factor Xa units). If bleeding continues, a second dose of 0.5 mg per 100 anti-factor Xa units can be given.

The most important adverse effect of protamine is an allergic response, including hemodynamic and respiratory problems [11]. Most adverse reactions can be prevented or minimized by slowing the rate of administration of the drug or by pretreatment with steroids and antihistamines. Risk factors for an adverse reaction are sensitivity to fish (as may occur in traditional fishermen who are often exposed to fish proteins when cutting themselves), a history of vasectomy (which may demolish the blood–testis barrier with consequent formation of anti-semen antibodies) and a history of receiving protamine sulphate containing insulin. Initial reports that the use of protamine sulphate could lead to an increased risk of rebound thrombosis, in particular ischemic stroke [12, 13] are not confirmed in a recent randomized controlled study [14].

There are some other strategies to reverse (mostly unfractionated) heparin, such as platelet factor-4, heparanase, or extracorporeal heparin-removal devices, but none of these approaches have been properly evaluated, and they are not currently approved for clinical use [15–17].

Pentasaccharides

Pentasaccharides are recently developed synthetic compounds that effectively bind and potentiate antithrombin to block factor Xa. Since they lack the additional glycosaminoglycan saccharide residues to bind to thrombin, it has an effect on factor Xa exclusively. The prototype pentasaccharide (and the only one approved for clinical use so far) is fondaparinux. Another pentasaccharide that is currently under study is idraparinux. The main difference between these two agents is in the elimination of half-life, which is 15–20 h for fondaparinux and 5½ days for idraparinux. This means that idraparinux can be administered once weekly, which renders the subcutaneous route of administration less cumbersome. Pentasaccharides are effective in the prophylaxis and treatment of venous thromboembolism, and are currently being evaluated in other types of thrombosis [18]. The (very) long half-life of pentasaccharides necessitates the availability of a suitable antidote if major bleeding complicates the treatment, which may especially occur in patients who are treated with therapeutic doses of this type of anticoagulation. So far, there is no antidote for pentasaccharides that has been studied in controlled clinical studies [19]. The only agent that has been systematically evaluated to reverse the anticoagulant effect of pentasaccharides is recombinant factor VIIa (rVIIa). Two randomized placebo-controlled studies in healthy volunteers have tested the hypothesis that rVIIa may be useful as a suitable antidote for pentasaccharide anticoagulation [20, 21]. In the first study 16 subjects were treated with therapeutic doses of the pentasaccharide fondaparinux and after 2 h (at the time of maximal anticoagulation) challenged with rVIIa or placebo. Injection of rVIIa (90 μg/kg) after fondaparinux normalizes the prolonged aPTT and prothrombin time (PT) and reverses the decrease in prothrombin activation fragments 1 + 2 (F(1 + 2)), as observed with fondaparinux alone. Thrombin-generation time and endogenous thrombin potential, which were inhibited by fondaparinux, normalize in up to 6 h after rVIIa injection. In the second study, 12 subjects received a single subcutaneous dose of 7.5 mg idraparinux, (which is threefold higher than the currently recommended dose). The inhibition of thrombin generation by idraparinux, as reflected by an increased thrombin generation time (TGT) and decreased level of prothrombin fragment 1 + 2 (F1+2), is partially reversed by injection of rVIIa 3 h after idraparinux administration. The administration of rVIIa 1 week after treatment with idraparinux (when much lower, though still therapeutic, doses of the pentasaccharide are present) results in a nearly complete reversal of anticoagulation, reflected by normalization of thrombin generation time and other markers of thrombin generation. As mentioned earlier, there are no controlled trials in patients who present with pentasaccharide-induced bleeding, but there is some anecdotal experience suggesting that rVIIa may indeed be able to stop bleeding in patients anticoagulated with fondaparinux.

Vitamin K antagonists

Vitamin K antagonists interfere with the γ-carboxylation of glutamate residues on vitamin K-dependent proteins, which therefore are not capable of a calcium-dependent conformational change by which they can bind to phospholipid surfaces, resulting in a strongly reduced coagulant activity [22]. There are several vitamin K antagonists available, of which warfarin is most widely used, but also the coumadin derivatives acenocoumarol and phenprocoumon are frequently prescribed. The most important difference between these three agents is their half-life, which is 9 h for acenocoumarol, 36–42 h for warfarin, and 90 h for phenprocoumon, respectively. This variation in half-lives may have important consequences for the optimal strategy to reverse each of these agents. The time to reversal of anticoagulation with vitamin K antagonists will not only be affected by their half-life after cessation of treatment, but also by the time it will take to produce properly carboxylated coagulation factors that have not been affected by vitamin K antagonists.

Vitamin K antagonists are potent anticoagulants and it is recommended to not perform major surgical procedures when the patient is fully anticoagulated. In these situations the anticoagulant treatment needs to be interrupted or its intensity should be considerably reduced. Perioperative prophylaxis of venous thromboembolism can be achieved by administration of subcutaneous (LMW) heparin.

The most straightforward intervention to counteract the effect of vitamin K antagonists is the administration of vitamin K [23, 24]. Whereas there is quite some debate on the use of vitamin K in patients with a too high intensity of anticoagulation (i.e., a too high INR) but no signs of bleeding, in patients with clinically significant bleeding administration of vitamin K is crucial to reverse the anticoagulant effect of warfarin or coumadin derivatives. Vitamin K can be given orally and intravenously, whereas despite a reasonably quick and good systemic bioavailability of oral vitamin K, the parenteral route has the advantage of a more rapid onset of the treatment [25–31]. After the administration of intravenous vitamin K, within 2 h the INR will start to drop and will be completely normalized within 12–16 h [31], whereas after oral administration it will take up to 24 h to normalize the INR [23, 32]. Intramuscular injections of vitamin K should be avoided in patients who are anticoagulated (since they may cause muscle bleeding) and subcutaneous administration of vitamin K results in a less predictable bioavailability [25, 27]. When the INR is below seven, a dose range of 2.5–5 mg vitamin K has been advocated to completely counteract the anticoagulant effect, whereas with higher INRs a dose of 5–10 mg is required to correct the INR. Higher doses of vitamin K are equally effective but may lead to warfarin or coumadin resistance for more than a week, which may hamper long-term management of these patients [33]. Of note, when reversal of anticoagulation has to be sustained for a longer period, repeated administration of vitamin K may be required, especially in patients who use vitamin K antagonists with a long half-life, such as phenprocoumon. A potential concern with the use of parenteral vitamin K is the occurrence of anaphylactic reactions, although the incidence of this complication is very low, in particular with the more modern micelle preparations [34–36]. To avoid this, a low infusion rate (e.g., 1 mg/3 min) has been advocated [23].

In case of very serious or even life-threatening bleeding, immediate correction of the INR is mandatory, and can be achieved by the administration of vitamin K-dependent coagulation factors. Theoretically, these factors are present in fresh frozen plasma; however, the amount of plasma that is required to correct the INR is very large, carries the risk of fluid overload, and will probably take hours to administer [37, 38]. Therefore, prothrombin complex concentrates (PCCs), containing all vitamin K-dependent coagulation factors, are more useful. In a small study in patients using vitamin K antagonists who needed urgent reversal of anticoagulation because of major bleeding or emergency invasive procedures, PCCs at a fixed dose of 500 U are sufficient to correct INR values <5.0. However, at INR values >5 higher doses are required [39]. Although PCCs can indeed be given using fixed dose schemes [40], it has been shown that individualized dosing regimens based on INR at presentation and body weight are more effective [41]. In a prospective cohort study of patients who presented with major bleeding associated with the use of vitamin K antagonists patients were treated with PCCs at a relatively high dose of 25–50 U/kg (exact dose per patient was “based on the INR and severity of bleeding”), which was effective in reducing the INR below 2 in 56 out of 58 patients [42, 43]. Another prospective study in patients using vitamin K antagonist and presenting with bleeding or the need for an urgent invasive procedure necessitating normalization of the INR also finds that doses of PCCs in a similar dose range are effective in reducing the INR below 1.3 in 93% of patients and results in at least satisfactory and sustained hemostasis in 98% [44]. In this study, the dose of PCCs was tailored to the INR at presentation (INR 2–4 25 U/kg, INR 4–6 35 U/kg, and INR >6 50 U/kg, respectively).

PCCs can be given over a short time frame, have an immediate effect, and the efficacy of the reversal of anticoagulation can be monitored by measuring the INR. PCCs are shown to be safe in a series of 14 retrospective and prospective cohort studies encompassing 460 patients [45]. Thrombo-embolic complications occurred in seven patients although in most cases this could have been explained by the underlying clinical situation and co-morbidity as well. In recent years the safety of PCCs, in particular regarding the transmission of blood-borne infectious diseases, has markedly improved owing to several techniques, such as pasteurization, nanofiltration, and addition of solvent detergent. The often stated risk of disseminated intravascular coagulation (DIC) due to traces of activated coagulation factors in PCCs comes from older literature, and modern PCCs seem not to be associated with eliciting or aggravating DIC [46].

Another option for immediate correction of the INR in patients using vitamin K antagonists is the administration of rVIIa, although this treatment is not officially approved for this indication. In healthy volunteers who were treated with the vitamin K antagonist acenocoumarol, the prolongation of the international normalized ratio (INR) above 2.0 was normalized with the administration of rVIIa at doses between 5 and 320 μg/kg [47]. The duration of the INR correction is dependent on the dose of rVIIa, whereby doses of rVIIa larger than 120 μg/kg result in an INR normalization that lasts longer than 24 h. Correction of the INR after administration of rVIIa may be a “cosmetic” effect since the prothrombin time is very sensitive towards traces of VIIa in plasma; however, a small clinical study finds that rVIIa at a dose of 16 μg/kg results in satisfactory hemostasis in 14 out of 16 patients who presented with major bleeding while using vitamin K antagonists [48]. Another study demonstrates that reversal of warfarin anticoagulation in a series of 13 patients undergoing invasive procedures results in a normalization of the prothrombin time and prevents bleeding in all subjects [49] and a series of six patients with central nervous system bleeding due to treatment with vitamin K antagonists showed successful reversal of anticoagulation, arrest of bleeding and uncomplicated drainage in all patients after administration of rVIIa [50].

New anticoagulants

In recent years a large number of new antithrombotic agents has been developed and tested in clinical trials, and many of these new agents will become available for clinical practice in the very near future [51]. The need for new anticoagulant agents is quite obvious; first, the current agents are insufficiently effective. For example 10–15% of patients undergoing major orthopedic surgery develop venous thromboembolism, despite prophylaxis with low molecular weight (LMW) heparin [52]. Furthermore, the available anticoagulants are relatively unsafe, mostly due to the occurrence of bleeding as discussed earlier. Last, current anticoagulant agents are often cumbersome in regard to their clinical use, requiring repeated laboratory control and frequent dose adjustments. Increasing knowledge on the function of the hemostatic system in vivo has resulted in a new generation of anticoagulant agents. Despite the fact that most of these agents indeed have a somewhat more advantageous benefit-risk profile and are much more easy to use with no need for monitoring or dose adjustments, bleeding remains the most important adverse effect of any type of anticoagulant treatment [53, 54].

An important group of new anticoagulants is the class of direct thrombin inhibitors. Thrombin is the central enzyme in the coagulation process, not only mediating the conversion of fibrinogen to fibrin, but also being the most important physiological activator of platelets and various other coagulation factors. Inhibition of thrombin can be achieved by administration of heparin, but in view of the limited capability of the heparin–antithrombin complex to inhibit surface-bound thrombin, new antithrombin-independent anticoagulants have been developed [55]. Prototype of these thrombin inhibitors is hirudin, originally derived from the saliva from leeches (Hirudo medicinalis), and nowadays produced by recombinant technology. Melagatran is a synthetic thrombin inhibitor, which has predictable pharmacokinetic properties and can thus be used in a fixed dose [56]. Moreover, the pro-drug ximelagatran is relatively quickly absorbed after oral ingestion and results in a sufficient systemic availability, rendering this agent suitable for long-term use as oral anticoagulant. Clinical trials on prevention and treatment of venous thromboembolism and in patients with atrial fibrillation showed promising efficacy of (xi)melagatran; however, due to the occurrence of enhanced liver enzymes in 6–7% of patients using melagatran, the compound has been withdrawn by the manufacturer. Nevertheless, there are many similar compounds under evaluation in clinical trials at this moment [57]. None of these compounds has an established antidote available in case of serious bleeding complicating the anticoagulant treatment. In a controlled clinical study in healthy subjects the melagatran-induced effects on aPTT, thrombin generation, and platelet activation are not affected by the administration of rVIIa [58]. Based on these results it seems that rVIIa is not effective in reversing direct thrombin inhibition. Since, however, rVIIa was able to correct the melagatran-induced prolongation of the prothrombin time and increased thrombin precursor protein concentrations, it might be that higher doses of rVIIa will have some effect in this situation, but this needs to be studied in future experiments.

Another class of new anticoagulants is directed at factor Xa. Prototype of these agents are rivaroxaban and apixaban, which have shown promising results in initial experimental and clinical studies [59, 60] There is no evidence of any antidote towards the anticoagulant effect of any of these orally available factor Xa inhibitors so far. Based on the experience with rVIIa in the reversal of the anticoagulant effect of fondaparinux, one can postulate that rVIIa may be an effective antidote for these agents; however, direct proof has not been demonstrated.

In view of the central role of the tissue factor–factor VIIa pathway in the initiation of blood coagulation, novel therapeutic strategies aimed at inhibiting this catalytic complex are currently being evaluated. One of these inhibitors is recombinant Nematode Anticoagulant Protein c2 (rNAPc2), a potent and selective inhibitor of the tissue factor–factor VIIa complex in the presence of factor Xa [61]. This compound shows promising antithrombotic properties in clinical trials of prevention and treatment of venous thromboembolism and percutaneous coronary interventions [62, 63]. Recent reports from in vitro studies indicate that therapeutic doses of recombinant factor VIIa (rVIIa) may be able to induce activation of coagulation resulting in subsequent thrombin generation, possibly independent of tissue factor activity [64, 65]. In a double-blind, randomized, placebo-controlled crossover study, administration of rNAPc2 causes a prolongation of the prothrombin time, which is immediately and completely corrected by the subsequent injection of rVIIa [66]. Administration of rVIIa in the presence of NAPc2 results in a marked generation of thrombin, as reflected by plasma levels of prothrombin fragment F1+2 and thrombin–antithrombin levels, respectively.

Aspirin

Aspirin is effective in the secondary prevention of atherothrombotic disease, in particular coronary artery disease, cerebrovascular thromboembolism and peripheral arterial disease [67]. As a consequence, aspirin is one of the most widely used agents in the Western world. Aspirin increases the risk of bleeding, in particular gastro-intestinal bleeding, and has been associated with a small but consistent increase in intracerebral hemorrhage. In addition, it has been shown that the use of aspirin is associated with increased perioperative blood loss in major procedures, although this does not necessarily translate into clinically relevant endpoints, such as the requirement for transfusion or re-operation [68]. Over the last few years the approach to the patient who uses aspirin and who presents with bleeding or needs to undergo an invasive procedure has changed considerably. In fact, in current clinical practice bleeding can almost always be managed with local hemostatic procedures or conservative strategies without interrupting aspirin. Also most invasive procedures do not require the cessation of aspirin when adequate attention is given to local hemostasis. In contrast, interruption of aspirin has been associated with an increased risk of thromboembolic complications, potentially due to a rebound hypercoagulability [69]. Obviously, in special clinical circumstances, such as intracranial bleeding or the need to undergo a neurosurgical or ophthalmic procedure, the anti-hemostatic effect of aspirin needs to be reversed immediately. The most rigorous measure to achieve that is the administration of platelet concentrate after cessation of aspirin. Another approach is the administration of de-amino d-arginin vasopressin (DDAVP, desmopressin). DDAVP is a vasopressin analogue that despite minor molecular differences has retained its antidiuretic properties, but has much less vaso-active effects [70]. DDAVP induces release of the contents of the endothelial cell associated Weibel Palade bodies, including von Willebrand factor. Hence, the administration of DDAVP results in a marked increase in the plasma concentration of von Willebrand factor (and associated coagulation factor VIII) and (also by yet unexplained additional mechanisms) a remarkable augmentation of primary hemostasis as a consequence. DDAVP is effective in patients with mild hemophilia A or von Willebrand’s disease and in patients with qualitative platelet defects, such as in uremia or liver cirrhosis. DDAVP seems also capable of correcting the aspirin-induced platelet dysfunction, although large clinical studies employing relevant outcome parameters are missing [71]. The combined effect of platelet concentrate and subsequent administration of DDAVP has also been advocated to correct the aspirin effect on platelets. The standard dose of DDAVP is 0.3–0.4 μg/kg in 100 ml saline over 30 min and its effect is immediate.

Thienopyridine derivatives and other antiplatelet agents

Clopidogrel belongs to the class of thienopyridine derivatives, which act by blocking the ADP receptor on the platelet. Clinical studies have shown that clopidogrel is as good as aspirin in the secondary prevention of atherothrombotic events [72]. However, the combination of aspirin and clopidogrel is superior to aspirin alone in patients who have received intracoronary stents or in other patients with high risk coronary artery disease. The increased efficacy of the combined use of aspirin and clopidogrel is also associated with a higher bleeding risk [73]. The decision whether to interrupt or even reverse antithrombotic treatment with clopidogrel and aspirin in case of serious bleeding or the need to perform an invasive procedure will depend not only on the specific clinical situation but also on the indication for the antithrombotic treatment (see above). Especially in patients with recent implantation of an intracoronary stent (in the last 6–12 weeks), cardiologists will often not or only reluctantly agree with cessation of treatment [74]. In this period re-endothelialization of the stent has not yet occurred and the patient is very vulnerable to acute thrombotic occlusion of the stent. In patients with drug-eluting stents this period may be even longer. If, however, the decision is made to stop and even reverse the treatment with aspirin and clopidogrel, administration of platelet concentrate is probably the best way to correct the hemostatic defect [75]. In addition, DDAVP was shown to correct the defect in platelet aggregation caused by clopidogrel, so this may be another option [76].

Other antiplatelet agents, which are predominantly used in interventional cardiology, include IIb–IIIa receptor antagonists, such as abciximab or eptifibatide. These agents have a very potent antiplatelet effect and although rare, may carry the risk of thrombocytopenia. If in these situations serious bleeding occurs, administration of platelet concentrate, possibly in combination with DDAVP, will correct the hemostatic defect [77].

Conclusion

Conventional anticoagulant treatment can be reversed by specific interventions when the clinical situation requires immediate correction of hemostasis. For the new generation of anticoagulants, no specific antidotes are available; although some interventions are promising they need further evaluation. Antiplatelet therapy with aspirin can be reversed, but this is often not required. More potent antiplatelet strategies can be counteracted with correcting strategies as well; however, this may in some cases not be desirable in view of the indication for this treatment.