Current Concepts in Coagulation Profile in Cirrhosis and Acute‐on‐Chronic Liver Failure

Abstract

Watch a video presentation of this article Patients with cirrhosis have profoundly altered hemodynamics and hemostatic pathways with procoagulant and anticoagulant mechanisms, resulting in a tenuous "rebalanced" state in the setting of portal hypertension. This balance could be tipped toward either a procoagulant or anticoagulant phenotype by superimposed conditions.1 Specific coagulation factor (F) V (FV), FVII, FIX, FX, FXI, prothrombin, protein C, and protein S are reduced with concomitant increased FVIII and von Willebrand factor (vWF) activity in cirrhosis.2 Thrombocytopenia, increased nitric oxide, and prostacyclin inhibit platelet function (PF), and higher vWF and FVIII activity support platelet aggregation.3 Thrombocytopenia is the result of splenic sequestration in portal hypertension, decreased hepatic thrombopoietin synthesis, and immune-mediated platelet destruction due to glycoprotein IIb/IIIa platelet surface antigen-antibody interaction triggered by inflammation or sepsis.4 Patients with acute liver failure (ALF) have a markedly prolonged international normalized ratio (INR) but preserved thrombin generation potential, and they tend not to bleed. In cirrhosis, INR increases modestly with synthetic dysfunction and reflects short-term patient mortality, but not bleeding risk, during invasive procedures because of adequate thrombin generation potential.5 Acute-on-chronic liver failure (ACLF) is associated with organ failure, progressive clinical course, and high 28-day mortality.6 In ACLF, coagulation switches from a procoagulant to an anticoagulant phenotype with the onset of systemic inflammation and endothelial activation, which is compounded by sepsis.7-9 In these situations, critical scenarios, such as variceal bleeding, the need for invasive procedures, central line placement, venous thromboembolism, portal vein thrombosis (PVT), and secondary organ failure, often supervene. Standard coagulation tests (SCTs), such as prothrombin time (PT), activated partial thromboplastin time (aPTT), INR, and bleeding time, do not measure the risk for bleeding in liver disease.3 Hence inappropriate use of blood products to correct these parameters is often ineffective and leads to volume overload and transfusion-related acute lung injury (TRALI).10, 11 The traditional chemical enzymatic pathways are now superseded by the cell-based model of hemostasis in which activated platelets are the primary effectors of clot initiation making a platelet plug. A crucial cellular element is endothelial injury, which is a driver for bleeding and thrombosis with activation of the inflammatory cascade. Primary hemostasis is mediated by platelets, whereas secondary hemostasis is characterized by formation of the fibrin mesh with plasma procoagulant proteins.4 Finally, when the vascular repair is complete, plasma anticoagulant proteins break down this fibrin mesh in a third process known as fibrinolysis. In cirrhosis, bleeding is often due to portal hypertension per se rather than being coagulopathy related, like hyperfibrinolysis. The cell-based model also explains why local hemostatic changes at the site of injury do not override the systemic hemostatic balance, and conventional tests of coagulation remain the same in a patient with liver disease with clinically apparent bleeding.1, 2 The clinician's ability to detect the coagulation defect at the site of portal hypertensive bleeding, such as variceal or mucosal or vessel injury (e.g., after biopsy), remains elusive because conventional tests reveal the global rather than the local defect.12 Patients with compensated liver cirrhosis have a rebalanced coagulation profile. In patients with cirrhosis, bleeding (mostly variceal) is usually the consequence of portal hypertension (gastropathy/variceal bleeding) or due to vessel injury (biopsy site/paracentesis site) bleeding.3 This implied that an elevated INR is not a predictor of bleeding, nor is a therapeutic correction target in a bleeding patient.5 Progression of chronic liver failure is associated with derangement of coagulative balance, and the balance shifts from a procoagulant to anticoagulant phenotype. Platelet count (PLT) alone is an incomplete test of coagulation and indirectly correlates with the degree of portal hypertension, splenomegaly, and hepatic decompensation. However, studies have shown that a target PLT of 56 × 109/L suffices to control variceal bleeding because of intact platelet-dependent thrombin generation in cirrhosis.13, 14 A special scenario is liver transplantation (LT) in cirrhosis, where excess transfusion of platelet concentrate (PC), cryoprecipitate, or fresh frozen plasma (FFP) may contribute to the development of hepatic arterial or venous thrombosis.15, 16 Recent studies have sparked interest in coagulation in ACLF.3, 7-9, 11 Fisher et al.8 showed lower thrombin generation in patients with ACLF compared with those with acute decompensation (AD). Bedreli et al.9 demonstrated that rotational thromboelastometry (ROTEM) may reduce substitution of coagulation factors in patients with ACLF. Blasi et al.7 described the ROTEM profile in ACLF as compared with cirrhosis. Finally, our group prospectively studied the dynamic global coagulation dysfunction in patients with ACLF in relation to the development of systemic inflammatory response syndrome (SIRS) and sepsis, and found that worsening coagulation failure often preceded the development of sepsis and mortality in these critically ill patients.3, 17 Table 1 summarizes available studies on coagulation in cirrhosis and liver failure.3, 5, 7, 8, 18-24 Figure 1 shows our understanding of this dynamic model of coagulation in cirrhosis and ACLF. Preprocedure correction of SCTs is a double-edged sword. Assessment of baseline bleeding risk, limitations of prediction by conventional tests, and the benefits of prophylactic correction are to be weighed (Table 2). Correction of the PT/INR may lead to paradoxical bleeding by increasing blood flow in the collateral beds.25 Standard doses of blood components, such as FFP, can only partially correct SCTs. Common dose ranges are from 15 to 30 mL/kg; however, these data have been extrapolated from critically ill patients with trauma and coagulopathy.26, 27 Increased volumes of FFP infusion predispose to transfusion-associated circulatory overload and TRALI. For every 100-mL rapid expansion of blood volume, portal pressure increases by 1 mm Hg, which can subsequently cause portal collateral-related bleeding.28 Prothrombin complex concentrates (PPCs) and recombinant FVIIa (rFVIIa) have been designed for specific situations. The use of cryoprecipitate, a small-volume product having vWF, fibrinogen, and fibronectin without need for cross-matching, is an attractive option. PCCs are available as three-factor (II, IX, X) and four-factor (II, IX, X, with the addition of VII) products. These products have additional factors (protein C, protein S, antithrombin [AT] III) with or without heparin. The use of PCCs has been standardized only for vitamin K antagonist (VKA)-treated patients, and the dose is based on body weight, INR, and FIX content. rFVIIa has been studied in acute variceal bleeding; however, no mortality benefit was demonstrated29 (Table 3). The thrombopoietin receptor agonist eltrombopag can increase PLTs to a variable degree after a 10- to 20-day course but bears a risk for PVT.30 Two new congeners, avatrombopag and lusutrombopag, are reportedly bereft of this complication, making them of clinical interest.31, 32 Global viscoelastic (VE) tests provide a more physiological assessment of coagulation and should be considered to guide blood transfusion requirements in LT and other major surgeries. Its application in patients with ACLF or in a critical care setting requires more data. VE tests, which include TEG, ROTEM, and Sonoclot, offer a means of assessing the activity of procoagulant and anticoagulant pathways, hyperfibrinolysis, and excessive clot lysis.3, 33 A major fallacy in the interpretation of these tests is that they are in vitro assays and cannot assess the in vivo hemostatic milieu of the endothelium, tissue factor, and portal pressure and flow. Assessment of clot formation can be performed in 10 to 20 minutes; however, assessment of clot lysis takes 30 to 60 minutes.34 VE testing in LT has been described in one randomized controlled trial and several retrospective studies on the basis for which VE test guided transfusion algorithms are recommended by major guidelines.35 Patients with cirrhosis are at an increased risk (0.5%-2%) for venous thromboembolism.36 Rates of PVT have been reported as approximately 8% per year, with morbidity and mortality at 1 year impacted by prophylactic anticoagulation.37 Anticoagulation demonstrates the most utility in patients with more extensive portal vein and mesenteric thrombosis in the absence of other risk factors for bleeding. Low-molecular-weight heparin (LMWH) does not appear to increase risk for variceal bleeding and is likely the safest choice.38 Thromboembolic events occur in 2% to 5% of patients without cirrhosis who receive thrombopoietin receptor agonists, and the risk is apparently higher in patients with cirrhosis.31 Acute variceal bleeding and portal hypertensive bleeding from gastropathy, vascular ectasia, and colopathy are often challenging in patients with cirrhosis.11 Another scenario encountered by the clinician is risk assessment and prophylactic correction of coagulation defects before an invasive procedure.19 Conversely, patients with cirrhosis and those with liver failure may have bleeding at one site and thrombosis at another. An example of this situation is a patient with acute PVT with extension of thrombus to the superior mesenteric vein causing mesenteric ischemia in the setting of high-grade esophageal varices. Tables 3 and 4 show these clinical scenarios and the clinical strategy to manage the condition. Most coagulation enzymes are also components of the inflammatory cascade. The cell-based model suggests that coagulation is one aspect of systemic inflammation. The interaction of platelets and inflammatory cytokines released from damaged endothelium triggers the adhesion of neutrophils and macrophages, and activation of inflammation. Hence patients with cirrhosis or ACLF with SIRS or sepsis are bound to have coagulation defects with an anticoagulant or procoagulant or mixed phenotype. The changes in the coagulation profile are dynamic, with resolution of global coagulation defects once sepsis resolves. Recurrence of a coagulation defect, such as platelet dysfunction, or occurrence of a new defect, such as fibrinolysis, is seen with new-onset sepsis. Prior studies have described reduced coagulation FV, FVII, FIX, FX, FXI, and prothrombin, and increased FVIII activity in cirrhosis.3, 7-9 Two other divergent agents of the cascade affected by sepsis are tissue plasminogen activator (tPA) and plasminogen activator inhibitor (PAI). In liver failure, tPA is elevated because of endothelial activation and reduced hepatic clearance. Levels of PAI, a blocker of fibrinolysis, are also increased, but to a lesser extent than tPA.12 Low fibrinogen is associated with a 29% increase in mortality for every 1-g/L (100-mg/dL) reduction in decompensated cirrhosis.39 Hypofibrinogenemia predisposes to increased risk for bleeding after prophylactic endoscopic variceal band ligation.40 Dysfibrinogenemia (i.e., an altered fibrinogen molecule) may cause decreased permeability of the formed clot compared with controls and may even confer hypercoagulable features, causing scenarios such as portal vein or deep vein thromboses.41 The development of sepsis in patients with cirrhosis changes the coagulation factors and cellular elements, including endothelium, macrophages, and lymphocytes. Endotoxins inhibit PF by increased prostacyclin and nitric oxide production.12 SIRS and sepsis trigger the release of endogenous heparinoids, or a heparin-like effect, caused by small endothelium/mast cell–derived glycosaminoglycans, which can be detected on heparinase-treated VE assays. One study demonstrated the presence of heparan sulphate–like molecules in patients with variceal bleeding.42 Senzolo et al.33 showed that these glycosaminoglycans affect hemostasis in patients with cirrhosis and bacterial infection as seen by a hypocoagulable native thromboelastography (TEG). This can be corrected by using heparinase-treated TEG. Bacterial sepsis predisposes patients to new bleeding events and results in difficult-to-control bleeds or endoscopic treatment failure.11 Endotoxin also activates the clotting cascade and results in disseminated intravascular coagulation (DIC). After appropriate treatment, the coagulation profile resolves and endogenous heparinoids are cleared, emphasizing the complementary association of the coagulation cascade and the inflammatory pathways.17 SCTs, such as INR and PT, are determined by liver-synthetized coagulation FI, FII, FV, FVII, and FX. The anticoagulation system (low protein C and protein S) is not tested, which may result in a hypercoagulable state despite increased INR in ACLF. Effective thrombin generation potential has been demonstrated to be normal in cirrhosis, AD of cirrhosis, and even uncomplicated ACLF because of higher levels of FVIII and low protein C.3, 7-9 VE assays provide a physiological estimation of coagulation; however, their use has been validated only in the setting of LT. New research has demonstrated the use of global coagulation tests, such as TEG, ROTEM, and Sonoclot, in the setting of cirrhosis, ALF, AD of cirrhosis, and ACLF; thus, VE tests are finding mention in newer guidelines for hemostatic resuscitation, but more validation is needed before they can be recommended as standard of care.3, 5-9, 25 Table 4 shows the therapeutic targets and agents available for evidence-based management of bleeding or thromboses in liver disease. Figure 2 shows a clinical decision algorithm to determine coagulation correction targets using VE assays and SCTs. Key elements of the whole blood TEG include the reaction time (R time), which reflects the quantity of available factors; clot formation time (K); alpha angle (α) reflecting the rate of clot formation and indirectly indicating fibrinogen levels; maximum amplitude (MA), which is an indicator of platelet activity; and finally, a measure of clot lysis.3 Thrombin generation assays provide the rate of thrombin formation in the presence of a triggering agent, such as phospholipid, a tissue factor that indicates a preserved coagulation milieu in cirrhosis or liver failure.13 The dogmatic practice of conventional testing and treatment of coagulation defects is being challenged in light of new clinical evidence of the cell-based hemostasis model in cirrhosis and liver failure. The use of global coagulation tests has provided more credible data for practical use of blood component therapy. Appreciation of the role of fibrinolysis in bleeding and effective use of antifibrinolytic agents to control mucosal bleeding are of interest in clinical practice, reducing the need for component transfusions. The pitfalls of the use of INR to guide transfusion strategies are well recognized. Hypercoagulable states in the setting of cirrhosis and liver failure are often undiagnosed. Prospective studies are needed to examine the utility of current risk models in patients with liver disease with a goal of improving knowledge of the incidence and natural history of venous thromboembolism.

Topics & Concepts