It stands to reason that anticoagulation and its reversal should be firmly embedded in the repertoire of the intensivist, as we are frequently involved in doing invasive things to people who are heavily anticoagulated, or in the management of bleeding complications which tend to arise from it. Reversal of anticoagulation has been asked about in several past paper SAQs:
Most of these, except for the most recent, have been about warfarin. Only in 2022 did the college examiners decide to focus on other topics. Undoubtedly in the future all kinds of other permutations on this theme will eventually appear as SAQs, and in preparation for this the following list of reversal options is somewhat larger and richer in detail than would normally be expected. Taking pity on regular readers who understandably equate "rich detail" with "incoherent anacolutha and pleonasms", a quick table is also offered:
Anticoagulant Reversal agent options and their mechanisms Unfractionated heparin
- Binds covalently to heparin, forming an insoluble complex and removing it from the circulation.
- Dose is 1mg per 100 units heparin
- Indicated for emergency reversal of heparin effect when bleeding, reversal of regional heparinisation (eg. dialysis circuit), or reversal of procedural anticoagulation (eg. bypass circuit)
Low molecular weight heparin
- Binds covalently to LMWH, partially reversing the anti-Xa effect
- Dose is 1mg protamine for each 1mg of enoxaparin
- Replaces reduced vitamin K, bypassing the inhibited vitamin K epoxide reductase complex, and permitting the synthesis of active factors II, VII, IX, and X.
- Dose ranges from 1mg orally (if stable) up to 10mg IV (if bleeding)
- Indicated as sole agent for stable patients, or as an adjunct where the patient has lifethreatening bleeding
Prothrombin factor concentrate
- Replaces missing factors II, VII, IX, and X, bypassing their blocked synthesis
- Dose is 15-50 IU/kg
- Indicated where immediate reversal is required (for example, bleeding into a vital organ, eg. intracranial)
Fresh frozen plasma (FFP)
- Replaces numerous clotting factors, among them factors II, VII, IX, and X,
- Dose is up to 15ml/kg
- Reserved for situations when the prothrombinex is not readily available, or where volume resuscitation is desirable, or where acute haemorrhage has caused the consumption of all clotting factors (as a part of a massive transfusion)
- Monoclonal antibody fragment which binds dabigatran with 350 times more affinity than thrombin, neutralising its activity
- Dose is 5g, which is two 2.5g vials
- Reserved for life-threatening bleeding, eg. such that requires massive transfusion
Apixaban and rivaroxaban
- Decoy protein (recombinant modified Factor Xa which has no catalytic activity) that binds apixaban and rivaroxaban with a higher affinity than native Factor Xa
- Dose is 400mg bolus, followed by 4mg/min infusion for four hours (and there is a higher dose regimen which doubles this rate)
- Reserved for life-threatening bleeding, i.e. where simply stopping the DOAC is not an appropriate option
- Provides a source of unaffected platelets to be consumed in the coagulation cascade.
- Dose is about 30% of the remaining platelets, i.e. 6-8 units of apheresis platelets or 1unit of pooled platelets
- Improves platelet function by liberating stored vWF and Factor VIII from endothelial cells, as well as improving other aspects of platelet function
- Dose is 0.3 µg/kg.
- Replaces fibrinogen so that it can be used to reconstruct degraded fibrin in haemostatic thrombi
- Dose is 10 units of cryoprecipitate (to start with)
- Replaces platelets rendered dysfunctional by the release of D-dimers, which inhibit platelet aggregation
- Dose is 1-2 units of pooled platelets
- Competitive inhibitor of the activation of plasminogen to plasmin by binding to lysine residues on plasminogen (thus, prevents the degradation of fibrin by plasmin)
- Dose is 1g tds
In terms of peer-reviewed publications to aid revision, one needs to filter one's results by anything after 2015 or so, in order for their article to know about idarucizumab. Yee & Kaide (2019) is an excellent example of this, as their paper is concise, readable, and free via the actual publisher (bravo, westjem.com!). For something more specific, direct oral agents are reviewed briefly in Joseph & Garcia (2022), and you can find some guidance for reversing exotic parenteral agents (fondaparinux!) in the detailed guidelines by Hirsh et al (2008).
Unfractionated and low molecular weight heparin are serenaded in a chapter of their own in the CICM First Part Exam revision section. Without revisiting material already presented there, unfractionated heparin can be summarised as heterogeneous swarm of glycosaminoglycan molecules, all of different lengths but generally in the order of 3 to 30 kDa, which binds to antithrombin-III and increases its activity, which is to inactivate factors Xa and IIa (thrombin).
Protamine, the weird fish product, is the first-choice reversal agent for heparin. Its pharmacology is surprisingly difficult to track down in article form, and the best paper seems to be Pai & Crowther (2011) which is ostensibly about the neutralisation of heparin in a broader sense. In brief, it is a strongly alkaline polypeptide made up of four arginine-rich chains which binds to the strongly acidic heparin in an irreversible covalent way, forming an insoluble complex and removing it from the circulation. That insoluble product has no anticoagulant effect.
The modern indications for giving protamine are exclusively in the setting of reversing heparin anticoagulation:
Outside of heparin, protamine has no therapeutic effects. Modern, because at one stage, before becoming a reversal agent, it was used to stabilise insulin in its subcutaneous depot form (this weird concoction was in fact a revolutionary step for insulin delivery vehicles and was referred to as Hagedorn insulin, in honour of its creator). The normal human dose is 1mg for every 100 units, and no more than 50mg should be given at any one time, as there is said to be some sort of Factor V inhibitory effect in high doses. For the vast majority of non-emergency situations, there is no need to use protamine to reverse heparin, because the effect of heparinisation is typically short (though this is dose-dependent). It is often just safer to wait for the heparin to wear off, considering that protamine has non-trivial side effects.
Recombinant Platelet Factor 4 (rPF4), an analog of the endogenously available inhibitor of heparin, is a 7800 Da molecule that appeared in some trials in the mid-1990s, and appeared superior to protamine mainly because it had no major adverse effects. For whatever reason, interest in this molecule had died down, perhaps suppressed by the shadowy corporate influence of Big Protamine.
Factor VIIa is listed as one of the options for the reversal of heparin, and theoretically it should do this in record time and with few side-effects, as it is a powerful promoter of the coagulation cascade on one hand, and on the other, a direct activator of factor XI (thereby bypassing the heparin-deactivated thrombin-mediated part of the intrinsic pathway). Notably, according to Pai & Crowther (2011), it does not affect the anti-Xa activity, probably because it does not touch anything upstream of that pathway.
Random polycationic molecules of all persuasions have been trialled as heparin reversal agents, but the CICM trainee does not need to be aware of them, as most of them never even reached human studies, let alone the market. This list is a rogues' gallery of dangerous molecules that most reasonable people would be reluctant to infuse into their patients for this purpose, including substances like methylene blue, vancomycin, the nephrotoxic quaternary amine hexadimethrine bromide, heparinase produced by Flavobacterium heparinum, and weird concatemeric peptides made of high-affinity heparin-binding cysteine sequences.
Low molecular weight heparins, a group consisting of enoxaparin dalterparin tinzaparin and fondaparinux, are smaller molecules than unfractionated heparin (6000-1500 Da) and therefore cleared renally, rather than by the obscure black magic of the reticuloendothelium. Their main difference from the normal long heparin molecules is their failure to form a ternary complex with thrombin, which means their main activity involves inactivating factor Xa. Unlike nicely reversible UFH, these drugs do not really have an effective reversal agent, apart from probably protamine and activated Factor VII.
Protamine can partially reverse the dose of low molecular weight heparin. The dose depends on the sort of heparin being reversed - the more sulfate residues on the heparin molecule, apparently the more neutralisable it is, which means tinzaparin is the most reversible, followed by dalteparin, followed by enoxaparin. Within eight hours of enoxaparin administration, the dose of protamine is 1mg protamine per every 1mg of enoxaparin, according to Hirsh et al (2008). The reversal of the anti-Xa effect is variable and incomplete, with some doubt as to whether the clinical bleeding risk is reduced. For example, Bang et al (1991) used some kind of terrifying gastrically ulcerated rat model to demonstrate that bleeding time is reduced (from 20 to 9.5 minutes), but not completely abolished. There does not appear to be any human clinical trial evidence to support or refute this use case, but considering that the alternative (fVIIIa) is dressed in even skimpier evidence, most reasonable people will give protamine without hesitation when their back is to the wall.
Interestingly, it is worth noting that fresh frozen plasma is said to be contraindicated for the reversal of heparin. Consider: FFP contains antithrombin-III, the target molecule for heparins of all molecular weights, which means it would just feed the coagulopathy instead of reversing it. It is in fact used to treat heparin resistance, when heparin is not working hard enough. Following from this, one might conclude that giving FFP to a bleeding patient who is full of heparin will result in worsening bleeding and a paradoxical effect on the APTT. However, a perfunctory search has not revealed any case reports of catastrophic bleeding in response to FFP in an overheparinised patient, and FFP is commonly administered in cardiac surgery where the patient's anticoagulation is often incompletely reversed, raising the question whether this thing is for real. One might instead argue that the bleeding is consuming all clotting factors, and therefore all clotting factors require replacement, i.e. without FFP the patient is doomed, even if does make their antithrombin-III more aggressive. The author himself, as a trainee, has been the meat in a sandwich of conflicting opinions on this matter, with senior haematology and intensive care staff both putting forward some compelling mutually exclusive arguments.
Warfarin is a coumadin derivative, presented as a mixture of two active enantiomers. It exerts its action by inhibiting the vitamin K epoxide reductase complex, which is necessary for the activation of factors II, VII, IX, and X. In the presence of warfarin this system does not produce functional factors, because reduced vitamin K is not available to act as a co-factor for gamma glutamyl carboxylase (the enzyme that activates them). Logically, the problem is fixed by either more vitamin K (so more factors can be created), or by replacing the actual missing factors (as fresh frozen plasma or Prothrombinex concentrate, which contains Factors IX, II and X).
Vitamin K replaces the depleted reserves of reduced vitamin K and permits more activated factors to be produced by the liver. Because a change in protein synthesis is involved, the onset of effect is often slow, making this a poor choice for hosing haemorrhage. According to Polito et al (2019), some effect is seen within the first four hours after administration, and you can make this faster by giving more vitamin K, or by giving it intravenously.
Fresh frozen plasma contains factors II, VII, IX, and X, as well as most of the other factors (but not much fibrinogen, nor FVIII). The maximum dose is 15ml/kg, which would end up being about 1000ml for a normal sized person. Considering that this is a protein-rich product, this sort of volume could definitely produce transfusion-associated circulatory overload.
Interestingly, if you sample a unit of FFP and test its INR, it will be normal (~1.1), but if you infuse it into a person, you may never correct the INR to anything better than 1.6-1.7, provided you stick to reasonable recommended doses. The main reason for this is the finite factor content of this product, which ends up getting diluted in the bloodstream. Holland et al (2006) found that the relationship of FFP dose with INR correction is not linear, and that the last few points of INR are the ones you really have to pay for with large volumes. Most reasonable people would agree that the INR number itself is not the most important thing, particularly if the patient has stopped bleeding, but if you really insist on seeing a normal number, you should correct the INR with a targeted factor concentrate.
Prothrombinex, or more correctly Prothrombinex®-VF, is a commercial product containing lyophilised human clotting factor concentrate (Factors IX, II and X). The main advantage of this stuff is its small volume and precise predictable composition. No more guessing what vegan hell the random donors of your FFP have been cavorting in - the vial of Prothrombinex will have a reliable amount of the exact factors that are missing in overtherapeutic warfarin coagulopathy. The dose is 25 – 50 IU/kg, or something like 1000-3500 IU for a normal sized person. Each vial contains 500 IU and reconstitutes up to 50ml, which means instead of 1L of FFP you'd be giving your patient a comparatively conservative bolus of 350ml.
Indications for which of these agents to use depend on the severity of the coagulopathy, and to some extent on whether you are bleeding to death. The definition of "bleeding to death" is obviously a bit unofficial, and the availability of products and factor concentrates is clearly very regionally specific, which means most jurisdictions have their own local reversal policies. However the general trend is to follow various published consensus statements.
Dabigatran is a small nonpeptide direct thrombin inhibitor, which interacts with the active site of thrombin and reversibly inactivates it, thereby preventing the conversion of fibrinogen to fibrin (this mechanism of action is also shared by argatroban). This part of the pathway does not change the performance of the PT and APTT assays, which means the patient can be fully anticoagulated with normal coagulation studies (ecarin time is apparently a better test).
Idarucizumab is the main reversal agent for dabigatran. This is a monoclonal antibody fragment which has high affinity for dabigatran, and which reverses the anticoagulation within minutes. According to Pollack et al (2015), the affinity of this antibody for dabigatran is about 350 times greater than the affinity of thrombin, which means it is able to strip all of the dabigatran molecules off their thrombin binding sites. That sounds perfect, but unfortunately most health services would baulk at the expense, as each dose costs approximately $3.5K. The reader is invited to consider that, perhaps right now, in some dismal bureaucratic oubliette, a hospital administrator is calculating how many bags of blood products would need to be saved in order for this agent to become cost-effective, concluding that a burial would be cheaper. Ok, perhaps this ghoulish exaggeration is not completely accurate, but the expense is reflected in guidelines which reserve the use of this agent for patients with "life-threatening or uncontrolled bleeding". Some are as prescriptive as to specify exactly how life-threatening the bleeding needs to be before you qualify for this safe and immediately effective antidote (for example, across the sea, the guidelines recommend a Hb drop of 50g/L, or a transfusion requirement in excess of 4 pack red cell bags).
Theoretically, blood products (specifically FFP and prothrombinex) could be used to reverse some of the effects of a direct thrombin inhibitor by simply supplying more thrombin than can possibly be inhibited by the circulating dose of dabigatran. With the availability of idarucizumab being limited to well-resourced health systems, these may be the only options, though they are not well represented in the literature (for example, UpToDate authors scoff at the idea because of insufficient evidence). This practice is well described, if you know where to look, but the description usually takes the shape of case series and case reports. For example, Dumkow et al (2012) reported reversing the effects of supratherapeutic dabigatran using 16 units of FFP and 2000 units of prothrombinex-equivalent factor concentrate.
Apixaban and rivaroxaban are Factor Xa inhibitors, as the reader may have guessed from all the xa. Factor Xa is supposed to connect to Factor Va to form the prothrombinase complex, which then goes on to convert prothrombin into thrombin. To inhibit the formation of prothrombinase therefore prevents the formation of thrombin.
Andexanet alfa is the reversal agent for Factor Xa inhibitors. It is not a monoclonal antibody like idarucizumab, but it also acts as a "decoy" Factor Xa, binding the apixaban or rivaroxaban instead of the native factor, and with greater affinity. This trick was achieved by making a recombinant version of factor Xa, modified to have no catalytic activity of its own, but preserving the active binding site for the various xabans.
Additionally, and slightly off-topic, andexanet alfa should theoretically be able to effectively reverse the effects of all the low molecular weight heparins. Consider: by pretending to be Factor Xa, it should be able to bind to antithrombin-III. reducing its anti-Xa activity (and allowing more of the native Factor Xa to do its job). Lu et al (2013) tested this in vitro and found enoxaparin and fondaparinux were dose-dependently reversed. For a variety of reasons (expense, as well as lack of human data and the ubiquity of cheap plentiful protamine), this agent is not currently indicated for this use, but the application is worth knowing about.
As for rivaroxaban, theoretically fresh frozen plasma could be used to inundate the bloodstream with fresh Factor X to replace what has been inhibited, but in cases of (for example) overdose there will be many free apixaban molecules around in the circulation to disable the new factors. Ergo, the volume of FFP would likely need to be heroic, and the effect would be incomplete. Various haematologist consensus works tend to bluntly state that FFP just doesn't work.
A more elegant solution which does not include pulmonary oedema should be a factor concentrate that specifically contributes a massive amount of Factor X, or other activated factors that bypass its role in the clotting cascade (such as Factor VIIa). Practically, however, this does not seem to work very well. Schmidt et al (2019) gave what they thought were appropriate doses to blood collected from apixabanned patients and demonstrated that the factor concentrate only partially reversed the ROTEM findings, though rFVIIa was slightly better. Song et al (2017) had slightly better luck, probably because their factor concentrate contained more Factor X.
Ciraparantag is a novel anticoagulant reversal agent that has the potential to reverse anticoagulation with DOACS as well as heparin of whatever molecular weight. This agent is a small water-soluble molecule, or technically two L-arginine molecules connected via a piperazine. It was discovered among a throng of tested substances in the process of deliberate drug discovery, looking for something that would bind to unfractionated heparin (Ansell et al, 2021, detail this process in their excellent paper). As an afterthought, in the course of modelling the energy of its binding kinetics, the investigators were surprised to discover that high-affinity binding to dabigatran apixaban edoxaban and rivaroxaban was also predicted. In vitro experiments confirmed that reliable non-covalent binding to all the listed anticoagulants was taking place. This is excellent news, as ciraparang is a rather benign substance - it has a short half-life (12-19 minutes), is cleared almost completely in the urine, and does nothing pharmacodynamically exciting other than its intended effect. One envisions a bright future filled with completely reversed cardiac surgical patients smiling happily from their beds, and not a droplet of protamine in sight, consigned to the dust of history alongside aprotinin and epsilon-aminocapric acid.
This section should be short and easy to write, as there is basically no specific agent that counteracts the effects of platelet aggregation inhibitors. Practically speaking, we tend to give platelets and desmopressin (DDAVP) in these situations, holding our nose at the evidence, considering the harms are few while the potential benefits are many.
Platelet transfusion does not so much "reverse" the effects of the antiplatelet agents, as provides a source of unaffected platelets to be consumed in the coagulation cascade. For example, to restore platelet activity for a patient on aspirin, one would need to mix the patient's blood with enough donor platelets to achieve a 30% donor concentration, which means one or two units of apheresis platelets is usually enough (Li et al, 2011). This is a practice based on faith rather than data, as Nagalla & Sarode (2019) were forced to conclude after performing a thorough literature review. We get the impression that it helps to transfuse platelets because of studies where platelet function was tested before and after platelet transfusion in healthy subjects on antiplatelet agents, for example. Obviously nobody in their right mind will enrol a catastrophically bleeding emergency CABG patient into a study where one of the arms involves not giving the platelets, which means we are probably stuck with this level of evidence forever. Attempts to introduce something scientific into the debate have yielded weird results occasionally in direct conflict with the current practice. For example, when the PATCH trial tried to use platelet transfusion to put an end to intracranial haemorrhage, the platelet transfusion group had worse outcomes (though, too be fair, these were mainly thrombolysed patients, and the treatment group all had much larger haematomas).
DDAVP is also not a reversal agent in the traditional sense (as it does nothing to undo the molecular actions of the antiplatelet drugs), but it is recommended to improve platelet function in the face of platelet inhibition. The mechanism of action is detailed elsewhere, but briefly the effect is mainly due to the release of stored vWF and Factor VIII from endothelial cells, along with a series of other (possibly rather hypothetical) positive effects on platelet aggregation. The dose is usually much higher than what one might expect to use for its antidiuretic effects: Colucci et al (2014) recommended 0.3 µg/kg.
Catastrophic bleeding being the major complication of thrombolysis, methods of reversing fibrinolytic agents should be in the repertoire of every intensivist and anaesthetist, as ICU and theatres is where these patients typically end up. The potential therapies are unfortunately not particularly well-established. There are guidelines around, of course, and the CICM trainee is invited to become familiar with them. For example, the AHA has a statement on the reversal of alteplase following the haemorrhagic transformation of stroke (Yaghi et al, 2017). They are quite upfront about their recommendations mainly being sourced from expert opinion and small case series.
Fibrinogen concentrate, eg. cryoprecipitate, seems like a logical choice for the reversal of a fibrinolytic agent. Too much fibrin being destroyed? Throw more fibronogen in there to act as a fibrin precursor, to restore the integrity of those clots. The AHA guideline recommends 10 units as a starting dose, aiming for a fibrinogen level of something like 1.5g/L (i.e. basically a conventional low-normal value).
Platelet transfusion is mentioned as an option for the reversal of thrombolytic agents because of the theoretical concern that thrombolysis produces platelet inhibition effects. Both D-dimers and glycoprotein IIb/IIIa are released abundantly during the process of fibrinolysis, contributing to the coagulopathy. That the AHA do not specifically recommend this therapy ("controversial", they call it) suggests that they could not bring themselves to make a decision, considering the results of the abovementioned PATCH trial. If giving platelets is definitely your decision, give a lot- 6 to 8 standard apheresis units, or one whole unit of pooled platelets.
Tranexamic acid seems like a logical choice for the reversal of fibrinolysis, it being an antifibrinolytic agent. Its mechanism of action consists of competitively inhibiting the activation of plasminogen to plasmin, which means it should act as a direct antagonist for any tissue plasminogen activator. It has the other advantage of being cheap and instantly available (one does not need to thaw it like FFP, for example). Its safety is also proved through decades of use. It seems remarkable that this drug does not receive anything better than a lukewarm recommendation from the AHA, who noncommittally quip that it "may be considered in all patients".
Other reversal agents are also on the menu (because why not try everything), but their uses are limited to specific circumstances:
Indications for the use of these therapies, again drawing on the AHA guideline, are not fixed prescriptive values, but rather consist of a gestalt assessment of each individual situation, consisting of the following factors: