There will be blood

Clinical overt bleeding is the most common complication of ECMO, affecting nearly 60% of patients during ECMO treatment.

Approximately half of these cases meet the criteria for major bleeding as established by the International Society on Thrombosis and Haemostasis (Kalbhenn and Zieger, 2022). Retrospective data showed comparable major bleeding rates between V-A and V-V ECMO patients (Nguyen et al, 2022).  

ECMO significantly perturbs the normal balance of haemostasis through multiple mechanisms (Kalbhenn and Zieger, 2022): 

• Thrombocytopenia is common due to circuit induced platelet activation and aggregation 
• Platelet dysfunction due to loss of key platelet surface molecules and P-selectin due to circuit 
• Shear mediated loss of high molecular weight von Willebrand multimers and acquired von Willebrand syndrome 
• Circuit induced diffused activation of the coagulation system leads to waste coagulopathy, which is associated with hyperfibrinolysis and factor XIII deficiency 
• Heparin anticoagulation – recommended by ELSO to be continuously administered during ECMO
• Coagulopathy due to underlying pathology, e.g., post-cardiac arrest syndrome.

Common sites of bleeding include (Kawauchi et al, 2022; Nguyen et al, 2022) :

• ECMO cannulation sites 
• Intrapulmonary 
• Gastrointestinal tract 
• Nasopharyngeal region 
• Intracranial

Patients on ECMO commonly develop anaemia even without bleeding. The many causes of anaemia include:

• Decreased production
  • Malnutrition 
  • Bone marrow suppression (e.g., due to medications or critical illness)
• Increased destruction
  • Haemolysis 
• Increased loss
  • ECMO-related bleeding complications 
  • Post-operative blood loss 
  • Coagulopathy 
  • Other
    • Stress induced GI bleeding 
    • Frequent blood sampling

Consider the following (Alfred Health, 2020; Kalbhenn and Zieger, 2022):

• Local actions
  • Reinforce dressing/dressing care 
  • Pressure bandage 
  • Surgical revision 
  • Endovascular treatment ie stent graft or embolisation (Zheng et al, 2019)
  • Re-advancing of the arterial return cannula
    • This is rarely performed due to the risks associated with it such as cannulation site infection, bloodstream infection, haematoma and arterial dissection. 
    • This may be performed under strict sterile condition if cannula continues to migrate out with worsening bleeding despite other measures
  • Note: suturing of the site is generally advised against, due to risk of damage to cannulae 
• Systemic actions
  • Pause heparin 
  • Reverse acidosis and maintain normothermia 
  • Replace clotting element deficiencies and optimise coagulation profile – Platelet >80, fibrinogen >1.5, INR <1.3 
  • Tranexamic Acid 
  • Consider careful use of protamine or recombinant factor VII if uncontrollable bleeding despite optimisation of above
    • This therapy especially has to be weighed carefully against the risk of acute clot formation within the extracorporeal system and possible thrombotic pump failure. 

Reasonable advice includes:

• Pause heparin until bleeding controlled
• Vigilant monitoring of circuit health: D-dimer, fibrinogen, plasma free Hb, transmembrane pressure, oxygenation
• Once bleeding controlled, re-introduce heparin at 500 IU/h and uptitrate as per local protocol 
• May aim for the lower end of therapeutic APTT range, e.g.; 50-60.

Commonly, clinicians use a trigger of 80g/L (Tenure et al, 2020).

Blood transfusion is common in ECMO patients, as evidenced by almost 90% of all patients receiving at least one packed red blood cell transfusion during VV-ECMO support (Kalbhenn and Zieger, 2022). Despite this, transfusion thresholds for patients on ECMO support are currently subject to variation due lack of expert consensus and lack of robust, high-quality evidence. Despite a recent meta-analysis indicating statistically significant benefits associated with restricted transfusion, like reduced risks of acute kidney injury and mortality, these findings were notably influenced by considerable heterogeneity, methodological limitations, and publication bias in the analyzed studies (Abbasciano et al, 2020).