LAUSR

Patients with endstage liver disease are at risk for both bleeding and thrombotic complications.1 Although the paradoxical risk for both bleeding and thrombotic events has been ascribed to the complex haemostatic changes that occur in patients with cirrhosis, this explanation may be wrong. It is now well accepted that patients with endstage liver disease largely remain in a haemostatic balance due to simultaneous changes in proand antihaemostatic pathways.2 The increased thrombotic risk is likely related to specific prothrombotic changes in the haemostatic system such as an imbalance in the von Willebrand factor/ADAMTS13 axis, decreased hepatic production of the natural anticoagulants associated with enhanced thrombin generating capacity, a prothrombotic clot structure and resistance to fibrinolysis in the sickest patients.3– 5 In contrast, the spontaneous or procedurerelated bleeding risk of patients with endstage liver disease is at least in part unrelated to the haemostatic changes in these patients.1 For example, multiple lines of evidence suggest that variceal bleeding, one of the most common bleeding complications, is fully unrelated to haemostatic failure but is altogether attributable to portal hypertension and local vascular abnormalities. Indeed, the cornerstone of the prevention and variceal bleeding treatment consists of medication (nonselective betablockers, vasoactive drugs) or procedures (TIPS) that decrease portal pressure. Prohaemostatic therapy, either with blood component transfusion or by infusion of pharmacological agents, is therefore not indicated. Nevertheless, despite clear expert opinion6 and societal recommendations,1,7 patients with variceal bleeding frequently receive blood products or other prohaemostatic therapy during initial resuscitation in line with general recommendations on treatment of massive bleeding. Although societal guidance documents indicate that transfusion of fresh frozen plasma (FFP) or platelet concentrate likely is ineffective in acute variceal bleeding and is associated with adverse events such as infection, transfusionrelated acute lung injury or transfusionassociated circulatory overload, the risk/benefit of blood component transfusion in acute variceal bleeds has not been formally assessed. In this issue of Liver International, Mohanty and coworkers provide compelling evidence that transfusion of FFP during acute variceal bleeding is ineffective and does harm.8 Specifically, in multivariable analysis, FFP transfusion was associated with an increased risk of failure to control bleeding at 5 days and with mortality at 42 days. Why is FFP still commonly given to patients with cirrhosis and acute variceal bleeding, even though societal guidelines advise against this treatment? First, in the setting of acute variceal bleeding, red blood cell (RBC) transfusion might become necessary either because of anaemia or because of the development of hypotension, with evidence from randomized trials favouring a low haemoglobin threshold for RBC transfusion (7 g/dL).9 When RBCs are transfused, clinicians are inclined to transfuse balanced amounts of FFP and platelet concentrate as per general major bleeding protocols. Indeed, in the study by Mohanty, almost all patients that received FFP had also received RBC transfusion. Secondly, patients with advanced chronic liver disease frequently have a prolonged prothrombin time or associated International Normalised Ratio (INR), and it is (at least at first sight) a logical clinical response to try to correct a prolonged INR in a bleeding patient. Interestingly, a proportion of patients in the Mohanty study received FFP in the absence of a substantially prolonged INR (<1.5), suggesting that these patients were given FFP not to counteract their coagulopathy but as a general response to RBC transfusion. Why is FFP ineffective as a haemostatic agent in the setting of acute variceal bleeding? The most likely explanation is that variceal bleeding is truly a nonhaemostatic bleed.1 The risk of variceal bleeding is not enhanced by anticoagulant therapy,10 and the severity and outcome of variceal bleeding are not worse in patients using anticoagulant drugs at the time of the bleed.11 If anticoagulants do not make variceal bleeding worse, it seems plausible that it is unlikely that procoagulants are effective in stopping the bleed. Indeed, recombinant factor VIIa, a low volume prohaemostatic, had little effect in randomized studies on variceal bleeding.12 In addition, the coagulation system in patients with endstage liver disease is intact, or even hypercoagulable,2 and one study published only in abstract form has demonstrated intact thrombin generating capacity even in patients during an acute variceal bleed.13 The 'don't fix anything that isn't broken' principle is another reason why FFP is likely ineffective during bleeding in patients with endstage liver disease. In vitro and ex vivo studies have shown that although FFP improves the INR, it hardly enhances thrombin generating capacity,14– 16 although one recent study suggested FFP to have prothrombotic effects in a reallife clinical setting by increasing the thrombin generation to levels higher than healthy controls and by demonstrating an increase in levels of markers of coagulation activation.17 Importantly, the number of units