The coagulopathy of cirrhosis

The concept that liver failure results in a profound coagulopathy is firmly established among hepatologists.1 The origin of the coagulopathy is commonly believed to be a defective hepatic synthesis of clotting factors, which is aggravated by the thrombocytopenia caused by portal hypertension.1 Hyperfibrinolysis may further contribute to a bleeding tendency.2, 3 In addition, many studies have shown a decreased hepatic synthesis of coagulation factors, reflected globally by the prothrombin time (PT). PT is an excellent marker of liver failure and a strong independent prognostic indicator of survival in patients with chronic liver disease.4 As such, new and old models of prognosis for patients with advanced liver disease include PT as one of its components.4, 5 PT, prothrombin time; APTT, activated partial thromboplastin time; ETP, endogenous thrombin potential. In this issue of HEPATOLOGY, the very interesting paper by Tripodi et al.6 challenges these concepts. Patients with cirrhosis, in spite of abnormal standard coagulation tests (PT and activated partial thromboplastin time, or APTT), may generate adequate amounts of thrombin in an in vitro assay (the endogenous thrombin potential, or ETP) modified by the addition of thrombomodulin. In their elegant study, Tripodi et al. show this is mainly due to the fact that these patients, in addition to the diminished hepatic synthesis of clotting factors, also have a profound deficit of natural anticoagulants, mainly of protein C (a protein synthetized by the liver), and also of antithrombin, which may counterbalance the bleeding tendency caused by the deficiency in procoagulants. Based on their findings, Tripodi et al.6 suggest that the feared coagulopathy of cirrhosis is more a myth than a reality and that patients with cirrhosis are probably not at an increased risk of bleeding from other causes than the existence of portal hypertension and its local consequences in the gastrointestinal tract (gastroesophageal varices, portal hypertensive gastropathy, and colopathy). In our view, although this study supports the view that the coagulopathy of cirrhosis is less important than what suggested by measuring the PT or APTT, the conclusion that the bleeding tendency in cirrhosis is merely due to the presence of portal hypertension goes beyond what is shown by their data: (1) First, they studied a series of stable patients with cirrhosis that included only 14 patients in Child-Turcotte-Pugh class C. Thus, the number of patients with advanced liver failure (at risk of exhibiting a severe "coagulopathy") was probably too small to establish in this group whether thrombin generation was adequate or not. This is especially the case in view of the very large range of values obtained in control subjects, suggesting the need for a better standardization of the test to avoid its large variability.7 The argument that their findings are in keeping with the fact that patients with cirrhosis do not exhibit soft-tissue hematomas and hemarthroses is not entirely correct; in fact, spontaneous hematomas or after minimal or inadvertent trauma are extremely common in patients with advanced liver failure and can be quite severe in some cases.8 (2) Second, the potential contribution of thrombocytopenia (and/or platelet dysfunction) and hyperfibrinolysis to such a bleeding diathesis is not commented upon, in spite of documenting a marked reduction of platelet counts and prolonged bleeding time in their patients. Thrombocytopenia in cirrhosis can be very profound, and in these cases there is an increased risk of bleeding complications in invasive procedures and surgery, even without a marked deficit of clotting factors. It should be noted in this regard that the endogenous thrombin potential was measured in platelet free plasma; as a result, possible differences caused by the platelet number or function were not evaluated. This is relevant because it has been shown in previous studies9 that the amount of thrombin generated depends clearly on platelet count and function, both of which are frequently altered in cirrhosis.10 (3) Third, the possible role of these coagulation abnormalities in worsening the bleeding due to portal hypertension is not recognized, despite the clinical evidence indicating that variceal bleeding is more severe, more difficult to control, and more likely to recur in patients with more advanced liver failure (with more prolonged PT, and sometimes, but not always, with lower platelet counts) compared with those with relatively preserved liver function.4, 11 We entirely agree with Tripodi et al. that it is very unlikely that these coagulation abnormalities play any role in initiating variceal bleeding. However, we believe they can be clinically relevant, contributing to aggravation of the portal hypertensive bleeding and facilitating recurrence. This view is compatible with the exploratory findings of a recent double-blind controlled multicenter trial reporting a favorable effect of rFVIIa compared with placebo. rFVIIa improved the initial control of variceal bleeding achieved with state-of-the-art medical therapy and allowed a higher rate of survival free of early rebleeding in patients with advanced liver failure (those of Child-Turcotte-Pugh class B and C; failure rates 8% in rFVIIa-treated patients vs. 23% in those receiving placebo).12 It should also be highlighted that to obtain a normal, or even increased, thrombin generation in the referred assay the authors modified the test by adding exogenous thrombomodulin. This is an elegant way to unmask the effect on coagulation of the reduction of the activity of protein C system. However, the amount of thrombomodulin required and the possible different behavior of soluble thrombomodulin compared with the protein located in cell surfaces have not been adequately investigated. Interestingly, with regards to the hyperfibrinolysis frequently reported in patients with cirrhosis (with some parallelism with the findings by Tripodi et al.), a recent study using a global plasma-based assay failed to note an increased fibrinolytic potential in these patients, suggesting that the reduction of profibrinolytic factors observed in liver disease may be compensated by a concomitant reduction in antifibrinolytics.13 All these new data highlight the complexity of the different mechanisms influencing the severity of bleeding episodes in these patients, involving both hemodynamic changes, thrombocytopenia and impairment of platelet function, disturbances in fibrinolysis, and decreased coagulation factors, although the latter may contribute less than was previously assumed. A practical take-home message is that when considering an invasive procedure or surgical operation in patients with cirrhosis, in addition to a history of hemorrhagic disturbances, clinicians should probably pay more attention to disturbances in platelet count and function (which is rarely investigated in patients with cirrhosis) than to abnormal PT and APTT. The data of Tripodi et al.6 clearly show that PT and APTT do not adequately reflect coagulation abnormalities in patients with cirrhosis. After their study, these tests should be interpreted with great caution as an index of the coagulation status in such patients (if used at all). Thus, we agree with the suggestion that ETP may be much more reliable to explore the role of coagulopathy in patients with cirrhosis in whom this is suspected to be clinically significant. At present, there are still clear limitations to introduce the thrombin generation test in the routine clinical evaluation of patients with cirrhosis, including its technical complexity and cost. However, commercially available simplified and automated versions of the endogenous thrombin potential assay will be soon available. It appears that the unquestioned use of the prothrombin time to assess the impaired hemostasis in patients with cirrhosis has come to an end. But does it mean that PT should no longer be used in patients with liver failure? As commented above, we do not think this is the case: we will continue to use it, not to estimate bleeding tendency, but as a good, simple, inexpensive, quantitative and accurate prognostic marker in patients with cirrhosis. We are most grateful to Dr J.G. Abraldes for his useful comments. Supported in part by grants from Instituto de Salud Carlos III and Fondo de Investigación Sanitaria (C03/02 and 04/0655).