The use of extracorporeal membrane oxygenation (ECMO) has increased exponentially in recent times [1], particularly following the H1N1 and COVID-19 pandemics [2]. It represents a modified cardiopulmonary bypass system that facilitates extracorporeal decarboxylation and oxygenation. The application of ECMO encompasses various scenarios involving refractory both heart and lung failure, and these indications have been established for adult and paediatric patients over recent decades, bringing a chance to improve the outcome of these patients [3]. The essential components of the ECMO circuit include a pump, a membrane oxygenator, a heat exchanger, a venous cannula, an arterial or venous infusion cannula, tubing, and connectors. It can be configured either in a venovenous (VV) mode, which is employed for the management of severe respiratory failure, or in a venoarterial (VA) mode, which provides simultaneous support for both respiratory and cardiac functions [4]. VV ECMO can be initiated using peripheral approaches involving two cannulae (femorojugular or femorofemoral) or it can be established using a double-lumen cannula inserted under ultrasound and echocardiographic guidance through the internal jugular vein. Alternatively, the drainage cannula is introduced through a venous access point in VA ECMO, while the return cannula is placed into an arterial access site. VA ECMO classification can be categorised as either central or peripheral, depending on the specific vessels used for cannulation [5]. In the central configuration, the drainage cannula can be directly inserted into the right atrium, with the return cannula placed in the ascending segment of the aorta. In the peripheral configuration, blood can be extracted through the femoral or jugular veins, and it is then returned to the patient through the carotid, axillary, or femoral arteries [5].

The anticoagulation strategy may differ between the VV and VA modalities. The VV ECMO may tend to have more bleeding complications at the insertion ^[8][10], likely related to a rapid decrease in pCO₂ together with a bolus anticoagulation with cannulation ^[9][13]. The VA ECMO may tend more towards thrombosis-related complications, especially with poor LV unloading and intracardiac thrombus formation, and during weaning of the VA modality due to low blood flow (down to 0.5 L/min) in the extracorporeal circuit and oxygenator ^[8][10]. Both modalities are fraught with cannulation-related deep venous thrombosis, i.e., related to the drainage cannulas and the return cannula of the VV ECMO ^[10][14]. The low-flow segment of the peripheral VA ECMO is between the return arterial cannula and the prograde superficial femoral artery cannula.

4. Considerations for Exhaustion of the Coagulation System

Exhaustion of the coagulation cascade typically results from a procoagulant state. A systemic inflammatory syndrome involves major triggers and regulators of the coagulation system. These are damaged endothelium by inflammation, activated intrinsic system with alteration in the protein C pathway, activated platelets, thromboplastin and activation of the extrinsic system, and finally involving the changes in fibrinolysis ^[11][15]. A typical example of coagulation activation is a sepsis-related disseminated inflammatory coagulopathy. Here, the mild prolongation of classic coagulation monitors like APTT, PT are accompanied by an increase in acute-phase proteins, including fibrinogen, and a rise in fibrin degradation products (D-dimers) with a D-dimer-related prolongation of the thrombin time ^[11][12][13][15,16,17]. An activation of platelets may be seen in the aggregometry tests or in the VETs. Specific tests may even show increased activity of certain factors like vWf. With progression of the systemic inflammation and exhaustion of the coagulation cascade, the thrombin activity decreases, accompanied by further extension of the coagulation times (APTT, PT, TT), decreasing fibrinogen, vWf, antithrombin III, and platelet count.

5. Special Considerations for COVID-19 and ECMO

Since the outbreak of SARS-CoV-2, an increased risk of thrombosis has been reported in patients with COVID-19 ^[14][19]. COVID-19-associated coagulopathy (CAC) manifests as micro- or macrothrombi formation, which can cause damage to multiple organs (lungs, heart, brain, kidneys) ^[14][19], resulting in an increase in morbidity and mortality. Compared to other viral infections, patients with COVID-19 have higher rates and severity of clotting events, which is underlined by laboratory findings of elevated plasma levels of D-dimer, C-reactive protein, fibrinogen and P-selectin ^[14][19]. Pathophysiological background includes a dysregulation among the inflammatory, immune, coagulation, fibrinolytic, complement and kallikrein–kinin systems. Thus, the term “immune thrombosis” has been established ^[14][19]. VV ECMO has been extensively used in the treatment of COVID-19-related ARDS. Less frequently, ECMO in VA configuration has been indicated for combined cardiac-respiratory failure ^[15][20]. Multiple studies showed that COVID-19 is connected to an increased risk of endothelial inflammation-mediated thrombosis ^[16][17][21,22]. A different methodology including only life-threatening bleeds, like in the ELSO registry or the International Society on Thrombosis and Haemostasis registry, may not elucidate the difference against the control population, because the bleeding events were independently associated with a higher in-hospital mortality on day 90. In addition, the rate of thrombotic complications was also high (36%), but thrombotic events were not associated with higher mortality rates ^[16][21]. These results differ from the large multicenter studies in the population of ECMO patients without COVID-19, which showed a significant increase in mortality due to thrombosis.  Currently, there is no consensus on an intensified anticoagulation practice for this particular patient population or those on only prophylactic anticoagulation, because the reported numbers of serious thrombotic events also related to ECMO are not negligible ^[18][29]. The increased prevalence of ICH highlights the need for a strictly controlled and individualised anticoagulation modality.

6. Unfractionated Heparin as Gold Standard

6.1. Introduction to Heparin

In contrast to most pharmacologic agents used in medicine, heparin is a heterogeneous compound consisting of molecules with variable molecular weights. The molecular weight of majority of the chains varies from 15,000 to 19,000 Daltons. This heterogeneity of heparin also leads to variable biologic effects. Therefore, the dose is usually not given in milligrams but in standardised international units—IU ^[19][20][30,31]. Heparin is used extensively because its effect can be easily titrated, the half-life is short, and its antidote (protamine) is widely accessible ^[21][32]. At its effect site, a bond with antithrombin (AT; previously antithrombin III) is created by the pentasaccharide segment of the molecule. The effect of AT is 1000 times multiplied when heparin is bound to its molecule. AT is a potent inhibitor of thrombin (factor II), factor Ixa, and Xa. Subsequently, a thrombin-dependent platelet activation and factors V, VIII, and XI are inhibited ^[19][22][30,33]. In conclusion, the mechanism of action of heparin is complex.

6.2. Monitoring and Target

Although UFH is the most frequently used anticoagulant in this specific group of patients, there are no detailed recommendations and a significant lack of data in scientific databases. According to the 2021 ELSO Adult and Paediatric Anticoagulation Guidelines ^[23][11], the monitoring of anticoagulation by ACT (in the range of 180–200 s ^[24][34]; with a low-concentration cartridge ^[25][35]) is considered a gold standard for the guidance of anticoagulation therapy in patients with ECMO. The widely spread standard laboratory-activated partial thromboplastin test (aPTT) is—owing to the guidelines—a gold standard assay for monitoring UFH. However, the laboratory aPTT assay may be influenced by multiple factors compared to the anti-Xa test, and recent data suggest purchasing both aPTT and the anti-Xa tests in parallel, especially in ECMO patients. This approach may allow for a calibration of the aPTT and prevent underdosing or overdosing with the unfractionated heparin modality ^[26][36]. ELSO suggested repeating the test every 1–2 h in case of ACT and every 6–12 h for monitoring the anticoagulation effect with aPTT. Great emphasis is placed on individual protocols of particular ECMO centres, which depend on local experiences. In the case of resistance to heparin, an examination with anti-Xa might be helpful. Routine supplementation with AT does not have scientific evidence. Under special circumstances, that is, spontaneously prolonged aPTT, VETs might give a reliable image of the actual coagulation status. Thus, hypercoagulability can be uncovered, and anticoagulation can be initiated regardless of spontaneously prolonged aPTT. In patients supported on ECMO, multiple studies evaluated the safety and feasibility of a VET-driven strategy to titrate heparin versus the “conventional” standard of care based on aPTT monitoring ^[24][34]. The improved trend of less bleeding complications was observed in a TEG^®-guided group (target 16–24 min of the R parameter). The rationale behind this precise anticoagulation titration is to keep the fragile balance between the two extreme situations. While bleeding and thrombosis remain common complications in ECMO patients, haemorrhagic events substantially influence mortality (emphasising fatal intra-cranial bleeding) ^[8][27][10,41]. According to the systematic review of Abruzzo et al., early computed tomography (CT) diagnosis of deep venous thrombosis was obtained in a number of patients as high as 71.4%, and pulmonary embolism was observed in 16.2% of the patients ^[10][14].

6.3. Heparin Resistance

This condition is defined as the need for an excessive dose of UFH to achieve the desired laboratory result. For some, this is defined as a dose >35,000 IU/24 h required to achieve a subtherapeutic range. One of the reasons for heparin resistance is considered to be AT deficiency ^[28][42]. Furthermore, other risk factors increase the risk of heparin resistance, including smoking, chronic obstructive pulmonary disease, liver dysfunction, nephropathies, hypoalbuminaemia, thrombocytosis, intra-aortic balloon pump, prolonged UFH treatment and presence of an ECMO circuit. AT deficiency might be acquired or (rarely) congenital ^[22][28][29][33,42,43].

7. Low-Molecular-Weight Heparins and Fondaparinux

Low-molecular-weight heparins (LMWH) are produced by depolymerisation of UFH to approximately one-third of the original size of the molecule (4000–6500 D), thus creating a shorter saccharide chain. Similarly to UFH, there is considerable variability in molecular weight and chemical properties ^[30][45]. Fondaparinux is a fully synthetic pentasaccharide, and its units are structurally similar to the cleaved monomeric units of UFH ^[31][46]. It was developed during an effort to solve problems associated with naturally sourced UFH (e.g., variability in composition, potential contamination) and eliminate side effects such as HIT ^[32][47]. Although LMWH has also been associated with HIT, the risk is considerably lower than with UFH and negligible in the case of fondaparinux ^[33][48]. Both LMWH and fondaparinux are eliminated by the kidneys, have a longer plasma half-life, better bioavailability, and in a standard setting, a more predictable dose response than UFH. However, they cannot be easily reversed by an antidote and their effect is prolonged in patients with renal failure ^[30][34][45,50].

7.1. Monitoring and Target

7.1.1. LMWH

The primary method of monitoring both LMWH and fondaparinux is by measuring the anti-Xa levels. The effect of fondaparinux is best measured using a specifically calibrated anti-Xa assay because the standard LMWH-calibrated assay overestimates the concentration of fondaparinux by about 20% ^[35][39]. Current recommendations for monitoring and anticoagulation targets in ECMO patients are outlined in the 2021 ELSO Adult and Pediatric anticoagulation guidelines. The guidelines suggest a target anti-Xa, albeit only in the context of UFH modality ^[23][11]. Concerning LMWH, published articles suggest that most centres opt for prophylactic dosage, especially in V-V ECMO ^[36][37][53,54]. This is in agreement with the current trend towards less or no anticoagulation for the V-V ECMO, supported mainly by retrospective studies and only one prospective pilot study ^[38][55]. Therefore, due to the lack of robust prospective data, the prophylactic approach is not included in the 2021 ELSO guidelines. LMWHs have gradually replaced UFH in many medical indications, and their superiority in prophylactic anticoagulation in critically ill patients has been consistently described ^[39][40][41][56,57,58]. Additionally, LMWHs have been used in prophylaxis in other types of extracorporeal circuits, for example, continuous renal replacement therapy (CRRT) ^[42][59]. Some data originating from a non-intensive care environment even suggest their possible superiority in therapeutic anticoagulation ^[43][44][60,61]. 

7.1.2. Fondaparinux

Similarly to LMWH, the evidence for using fondaparinux in patients with ECMO is scarce, albeit the published articles generally describe positive outcomes. Only several case reports and series on the topic were published, all with one common theme: patients on ECMO who developed HIT. The prophylactic dose once daily appears to be this group’s most prevalent anticoagulation strategy ^[45][46][47][65,66,67]. In most published case reports, only a prophylactic dose of fondaparinux was used, and anti-Xa was not measured ^[45][46][65,66].

7.2. Heparin-Induced Thrombocytopenia

A decrease in platelet count is a common complication of ECMO therapy (both in VV and VA) and occurs in up to 50% of subjects with ECMO. Post-cardiotomy ECMO patients represent a group with a higher risk of developing significant thrombocytopenia. There are many possibilities for the aetiology of thrombocytopenia in this particular group of patients: the foreign surface of the circuit, platelet activation, nonspecific activation of the inflammatory cascade, sepsis, medications (immunosuppressives, PDE III inhibitors), surgery, and bleeding. Often it is not possible to highlight one particular reason. Thus, aetiology is frequently considered multifactorial ^[48][49][50][68,69,70]. The prevalence of HIT in patients undergoing ECMO therapy ranges from 0.5% to 5.0% ^[50][51][70,71]. These numbers are probably underestimated, as HIT testing is not routinely performed in all patients. HIT is an antibody-mediated side effect of UFH administration that is characterised by the onset of thrombocytopenia, usually 5–10 days after the beginning of therapy. Typically, PF4 is stored in α-platelet granules (and released after adequate activation). The physiological function of PF4 is in binding to heparan sulphate, an endogenous substance very similar to UFH. However, its ability to bind with UFH shows a much higher affinity. Due to the newly created complex heparin–PF4 with the Fc site of the antibodies, the platelets and immune cells are activated. HIT-specific antibodies are examined by ELISA for the detection of the heparin–PF4 complex. This test shows a negative predictive value of 98–99%. This means that the negative result is very reliable for excluding the possibility of HIT. However, the positive predictive value is only 2–15%. Clinical suspicion and mentioned screening methods above predict the result of the confirmation methods well, so together they might often be sufficient.

8. Direct Thrombin Inhibitors

8.1. Bivalirudin

Bivalirudin is a DTI potentially used as an alternative anticoagulant in patients with ECMO. Bivalirudin is approved during percutaneous coronary intervention in patients with or at risk of HIT ^[52][75]. Its use in ECMO anticoagulation is considered off-label; however, it has been used as a primary choice for anticoagulant therapy in patients with heparin resistance, HIT, or those requiring surgery ^[52][75]. Its mechanism of action is based on reversible binding to the active site of thrombin with the onset of action within 4 min after bolus administration ^[53][74]. Bivalirudin has a relatively small volume of distribution, indicating that it remains primarily within the bloodstream. It does not bind extensively to plasma proteins and is predominantly metabolised by proteolytic cleavage via plasma carboxypeptidases, forming various metabolites with reduced anticoagulant activity ^[54][76]. Bivalirudin and its metabolites are primarily eliminated renally. The half-life of bivalirudin ranges from 25 to 80 min in patients with normal renal function ^[55][77].

8.2. Monitoring and Targets

The aPTT test is widely used to adjust and monitor the dosage of bivalirudin ^[55][77]. The same aPTT targets are commonly used for bivalirudin as for UFH, with the aim of a range of 1.5 to 2.5 times the patient’s baseline aPTT ^[56][78]. However, the correlation between the dosage and the response of aPTT is not flawless and has been described as weak to moderate ^[57][79]. Alternative tests may correlate strongly with bivalirudin doses, such as the ecarin chromogenic assay, the dilute thrombin time (dTT) and the chromogenic anti-IIa assay dTT, which involves plasma-diluted thrombin time and exhibits a better correlation with bivalirudin dosage compared to aPTT ^[58][80].

8.3. Argatroban

.3. Argatroban

Argatroban represents another DTI as an alternative anticoagulant to UFH in ECMO patients. Argatroban is metabolised in the liver through several pathways, including the cytochrome P450 enzyme ^[59][89], resulting in a final half-life of 45 min ^[60][90]. There are four metabolites, one partially active ^[61][91]. Therefore, in patients with liver failure, the half-life can be prolonged up to two- to threefold, requiring adequate dose reduction ^[60][90].

8.4. Monitoring and Target

Similarly to bivalirudin, aPTT represents the most common monitoring method for argatroban anticoagulation, with a target of 50–70 s reported in the literature ^[62][98]. ACT has also been used as a single parameter or combined with aPTT to guide argatroban anticoagulation with a wide range of 150 to 230 s ^[62][98]. Furthermore, the ecarin chromogenic assay (ECA), with its high specificity to monitor DTI or VETs, did not correlate with aPPT levels ^[63][97]. Interestingly, a significant proportion of patients who reached therapeutic levels of aPTT had normal values of ECA and VETs ^[64][99].

9. Conclusions

The balance of hemostatic mechanisms in ECMO patients is seriously disturbed from many points of view. First, the critical condition itself is a serious factor. Second, the unique influence of a particular underlying disease must be considered, i.e., COVID-19. Third, the disturbance caused by artificial materials of the extracorporeal circuit, oxygenator, and changes in blood homeostasis also play a significant role. Last but not least, the physician’s efforts to manage possible thrombotic or bleeding events, the choice of anticoagulant agent, and the targeting of the right numbers of the monitoring modality are all critical factors that should lead to better clinical outcomes.