Hemostatic abnormalities in patients with liver disease

INTRODUCTION — This topic discusses the hemostatic abnormalities in patients with liver disease and our approach to common clinical problems, including treatment of bleeding, invasive procedures, and management of venous thromboembolism (VTE).

Diagnosis and management of portal vein thrombosis, liver injury, cirrhosis, and liver disease from alcohol use are discussed in separate topic reviews:

●(See "Acute portal vein thrombosis in adults: Clinical manifestations, diagnosis, and management".)

●(See "Chronic portal vein thrombosis in adults: Clinical manifestations, diagnosis, and management".)

●(See "Surgical techniques for managing hepatic injury".)

●(See "Management of hepatic trauma in adults".)

●(See "Cirrhosis in adults: Overview of complications, general management, and prognosis".)

●(See "Hematologic complications of alcohol use".)

PHYSIOLOGIC EFFECTS OF HEPATIC DYSFUNCTION

Overview of abnormalities — Individuals with liver disease have a variety of hemostatic abnormalities, which are generally accepted to result in "rebalanced" hemostasis in which diminished hepatic function leads to both procoagulant and anticoagulant effects [1,2]. These changes increase risks of bleeding as well as thrombosis. They become especially important in patients with bleeding and thrombotic risks such as gastrointestinal varices and vascular stasis, respectively.

All stages of the hemostatic process may be abnormal, including primary hemostasis (platelet adhesion and activation), coagulation (generation and crosslinking of fibrin), and fibrinolysis (clot dissolution) [3,4]. Clinicians should not presume that patients with severe liver disease and abnormalities of coagulation testing are "auto-anticoagulated." There is overwhelming evidence that standard coagulation testing does not accurately assess bleeding or clotting risk in patients with cirrhosis. (See "Overview of hemostasis" and 'Prothrombotic changes' below and 'Altered fibrinolytic system' below.)

Factors that contribute to increased risks of both bleeding and thrombosis include altered blood flow, diminished numbers and function of platelets, and inflammatory alterations in endothelial cells [1,5]. These changes may result in a relatively balanced steady state in some patients, but it is generally accepted that susceptibility to both bleeding and thrombosis may be increased, with the relative balance or imbalance different for each patient and clinical circumstance. The risks of bleeding and thrombosis are not reflected in the conventional indices of coagulation status such as the prothrombin time (PT), international normalized ratio (INR), or activated partial thromboplastin time (aPTT). Of interest, elevation of von Willebrand factor (VWF) and the ratio of factor VIII to protein C carry independent prognostic importance in patients with cirrhosis [6]. The historical assumption that individuals with liver disease are "auto-anticoagulated" is now understood to be inaccurate.

Impaired hemostasis — Patients with liver disease are a heterogeneous population, and multiple abnormalities of hemostatic function may coexist in an individual patient leading to risks of hypo- and/or hypercoagulability.

Patients with cirrhosis can vary across a spectrum of disease stability ranging from well-compensated cirrhosis to advanced disease with multiorgan failure. For example, patients with acute-on-chronic liver failure (ACLF) may have altered hemostatic function relative to patients with compensated cirrhosis [7-10].

Coagulation factor deficiencies — The hepatocyte is the site of production of almost all of the numbered coagulation factors including fibrinogen (factor I), thrombin (factor II), and upstream factors V, VII, IX, X, and XI. Notable exceptions are factor VIII, which is produced in endothelial cells, and the factor XIII A-subunit, which is produced in megakaryocytes [11,12]. In addition to synthesizing coagulation proteins, hepatocytes also make post-translational modifications such as glycosylation and gamma-carboxylation of some factors. Both synthesis and post-translational modification may be impaired in liver disease, affecting coagulation factor abundance and function, respectively. (See "Vitamin K-dependent clotting factors: Gamma carboxylation and functions of Gla".)

In some patients with liver disease, particularly those actively using alcohol, vitamin K deficiency can further exacerbate deficiencies of vitamin K-dependent factors (II [prothrombin], VII, IX, and X) and/or lead to improper modifications (eg, under-gamma carboxylation of prothrombin) [13]. Qualitative defects in factors such as fibrinogen structure may further contribute to coagulopathy in cirrhosis [14].

Thrombocytopenia and platelet dysfunction — Patients with liver disease may have normal platelet counts (ie, ≥150,000/microL) or varying degrees of thrombocytopenia. Mild thrombocytopenia (eg, platelet count between 100,000 and 150,000/microL) has been reported in up to 75 percent of patients with chronic liver disease, and moderate thrombocytopenia (eg, between 50,000 and 100,000/microL) has been reported in approximately 13 percent of individuals with cirrhosis [15,16].

Secondary to rebalanced hemostasis in liver disease, there does not appear to be a strong correlation between thrombocytopenia and bleeding risk in these individuals, especially for platelet counts >50,000/microL.

The mechanism of thrombocytopenia in liver disease may include impaired platelet production, from decreased hepatic synthesis of thrombopoietin; bone marrow suppression, from hepatitis C virus (HCV) infection or alcohol use, other infection, or antiviral or antibiotic therapy; and increased platelet sequestration in the spleen, in the setting of portal hypertension and hypersplenism [15]. (See "Biology and physiology of thrombopoietin" and "Extrahepatic manifestations of hepatitis C virus infection" and "Splenomegaly and other splenic disorders in adults", section on 'Hypersplenism'.)

In addition to thrombocytopenia, individuals with advanced liver disease may have reduced platelet function due to coexisting acute kidney injury, infection, and/or endothelial abnormalities [17].

The overall incidence of infection in patients with liver disease has been estimated to be as high as 30 percent [18]. Overt sepsis or low levels of endotoxemia can impair platelet function in patients with cirrhosis. Moreover, infection is associated with increases in endogenous glycosaminoglycans known as heparinoids (eg, heparan sulfate, dermatan sulfate), which can act as anticoagulants; these increases may result from changes in nitric oxide metabolism or other endothelial changes [19,20].

Altered fibrinolytic system — The fibrinolytic system is altered in patients with cirrhosis and can have significant changes in pro- and antifibrinolytic factors [7,10]. Fibrinolysis (dissolution of the fibrin clot) is often increased in liver disease; however, evidence is emerging that this system can be rebalanced and place patients at risk for both bleeding and thrombosis.

Laboratory evidence of systemic fibrinolysis can be detected in patients with chronic liver disease. However, clinically evident hyperfibrinolysis is less common and has been estimated to occur in 5 to 10 percent of those with decompensated cirrhosis [21-23]. Hyperfibrinolysis promotes premature clot dissolution and interferes with clot formation due to the consumption of clotting factors.

Hyperfibrinolysis overlaps with a condition in cirrhosis that resembles disseminated intravascular coagulation (DIC), called "accelerated intravascular coagulation and fibrinolysis (AICF)," but it can be evident as a distinct clinical entity with intractable delayed bleeding from surgical interventions or dental extractions, or on occasion spontaneous bleeding without any recognizable trauma [24]. However, the lack of a commonly available means to clearly identify this condition (such as via the use of viscoelastic tests) often impedes the diagnosis of hyperfibrinolysis in patients with cirrhosis. Even with these global tests of clot formation and dissolution, widely accepted criteria for milder cases of hyperfibrinolysis are not yet established.

Hypofibrinolysis can occur in decompensated cirrhosis and lead to risks of thrombosis [25]. In a study involving 36 patients with acute-on-chronic liver failure (ACLF), the presence of hypofibrinolysis as diagnosed by viscoelastic testing was associated with a poor prognosis [8].

Hepatocytes and Kupffer cells are also responsible for clearing coagulation factors and products of fibrinolysis from the circulation [26]. Thus, chronically impaired liver function in cirrhosis may be associated with multiple mechanism(s) of altered fibrinolysis:

●Increased levels of tissue plasminogen activator (tPA), which generates plasmin [27].

●Decreased levels of alpha 2 antiplasmin, coagulation factor XIII [28], and thrombin-activatable fibrinolysis inhibitor (TAFI) [27].

●Elevated levels of fibrin degradation products such as D-dimer.

●Fibrinolytic activity of ascitic fluid that may be delivered to the systemic circulation via the thoracic duct [29].

These processes are discussed in more detail separately. (See "Overview of hemostasis", section on 'Clot dissolution and fibrinolysis'.)

Prothrombotic changes — The following changes may be seen:

●Reduced protein S, C, and antithrombin – In addition to synthesizing clotting factors, the liver also produces endogenous inhibitors of coagulation (eg, protein S, protein C, antithrombin [AT; formerly called AT III]) and fibrinolytic factors. Liver disease may contribute to a prothrombotic state because these natural inhibitors may be reduced [30-33].

●Increased VWF – Von Willebrand factor (VWF), derived from endothelial cells, is elevated in patients with cirrhosis and may contribute to prothrombotic changes that balance bleeding risks associated with thrombocytopenia [34,35].

●Increased PAI-1 – Liver disease may also be associated with increases in acute phase reactants, such as plasminogen activator inhibitor 1 (PAI-1), and decreased levels of the VWF-cleaving protease ADAMTS13, as well as inflammatory changes in endothelial cells that promote thrombosis [35,36]. (See "Overview of hemostasis", section on 'Control mechanisms and termination of clotting' and "Overview of hemostasis", section on 'Clot dissolution and fibrinolysis'.)

●Increased D-dimer – D-dimer may be increased due to altered fibrinolysis and/or decreased hepatic clearance [32]. (See 'Altered fibrinolytic system' above.)

●Reduced blood flow – Reduced vascular flow also contributes to local prothrombotic tendencies. Examples include stasis in the portal circulation and lower extremity venous stasis due to peripheral edema.

●Infection – Infections such as bacterial peritonitis or a chronic inflammatory state may further exacerbate these changes in endothelial reactivity and/or blood flow [37].

While these prothrombotic changes are increasingly appreciated, standard tests of coagulation do not effectively measure them.

Significance of type of liver disease — Many of the hemostatic abnormalities in liver disease are similar regardless of the underlying cause of liver injury [38]. However, some differences have been reported. As examples [39]:

●Cholestatic liver diseases such as primary biliary cholangitis (PBC) and primary sclerosing cholangitis (PSC) appear to have a less pronounced effect on anticoagulant than procoagulant mechanisms and may be at higher risk for portal vein thrombosis. This mild hypercoagulability may be mediated by changes in platelet activity [40-42].

●Nonalcoholic fatty liver disease (NAFLD) may confer a greater prothrombotic risk [43-45].

●Acute-on-chronic liver failure (ACLF) and acutely decompensated liver disease may represent a distinct population with unique coagulopathies [7-9,46,47].

●Acute liver failure may have a lower incidence of thrombocytopenia but more severe reductions in procoagulant and anticoagulant factors, compared with chronic hepatic insufficiency [48]. Clinically significant bleeding is nonetheless uncommon in patients with acute liver failure [49].

LABORATORY ABNORMALITIES

Coagulation testing

●PT and aPTT – Individuals with liver disease frequently have abnormalities in routine laboratory tests of coagulation, including prolongations of the prothrombin time (PT), international normalized ratio (INR), and activated partial thromboplastin time (aPTT), along with mild thrombocytopenia, elevated D-dimer, especially when liver synthetic function is more significantly impaired and portal pressures are increased [5]. However, these tests are very poor at predicting the risk of bleeding in individuals with liver disease because they only reflect changes in procoagulant factors [39].

We use the PT/INR as a prognostic measure (eg, in calculating the Model for End-stage Liver Disease [MELD] score) rather than a means of bleeding risk assessment. Prolongation of the INR probably does reflect instability of the overall hemostatic balance, but it does not indicate hypo- or hypercoagulability in cirrhosis.

●Fibrinogen – Plasma levels of fibrinogen can also be used to risk stratify bleeding risk in chronic liver disease patients. At sites of vessel injury, fibrinogen is converted by thrombin into fibrin for clot formation. Fibrinogen levels can be low in chronic liver disease, which may reduce clot formation potential. Measurement of fibrinogen and adequate repletion, if low, could prevent bleeding episodes in cirrhosis patients, and one study of variceal banding patients showed higher rates of bleeding in those patients with low fibrinogen levels, which may have been a reflection of the severity of portal hypertension [50]. However, efforts to replace fibrinogen with Cryoprecipitate do not appear to change bleeding outcomes [51]. Rather, low fibrinogen, similar to thrombocytopenia, may reflect critical illness and advanced liver disease more prone to bleeding or thrombosis.

●TEG and ROTEM – Viscoelastic testing such as thromboelastography (TEG) or thromboelastometry (ROTEM) is increasingly studied in patients with liver disease, but clinical use is mainly confined to perioperative transfusion guidance in patients undergoing liver transplantation [2,11,52-54].

TEG and ROTEM both produce a tracing that reflects dynamic changes in clot formation and lysis, assessed from the changes in torque between a pin and a cup that occur as blood clots; parameters of clot formation time and firmness can be read from the tracing (figure 1). In TEG, the cup rotates and the pin is stationary; in ROTEM, the pin rotates and the cup is stationary. Further information about parameters and interpretation of viscoelastic testing is presented separately. (See "Etiology and diagnosis of coagulopathy in trauma patients", section on 'Viscoelastic hemostatic assays' and "Platelet function testing", section on 'Viscoelastic testing (TEG and ROTEM)'.)

In general, studies of TEG and ROTEM in patients with liver disease have shown decreased use of blood products due to the ability of these devices to confirm relatively preserved hemostatic function despite thrombocytopenia and a prolonged INR [55-57].

However, study design limitations in this field have not established baseline bleeding risk in individuals who do not receive bleeding prophylaxis [58]. Therefore, evidence for the value of viscoelastic testing in liver disease remains very limited without clear established parameters, and the sensitivity of these tests to detect subtle coagulation changes is questionable.

•A retrospective review of TEG parameters in 344 individuals hospitalized with cirrhosis found that most of the patients had preserved clot initiation (R value (figure 1)) and propagation (K value and alpha angle), but clot strength (maximum amplitude [MA]) was impaired [59]. Reduced MA was largely attributed to thrombocytopenia. These parameters correlated with disease severity and overall survival but did not provide additional information over and above the model for end-stage liver disease (MELD) score. Another study using TEG found conventional parameters to be abnormal (INR and platelets), whereas TEG parameters were often within normal ranges [60].

•A small study comparing ROTEM parameters in blood samples from 51 patients with cirrhosis found that the parameters most likely to be abnormal in cirrhosis were the clot formation time and the maximum clot firmness; in contrast, the coagulation time was normal in more than one-half of the samples from cirrhotic individuals [61].

•A retrospective study in patients undergoing liver transplantation found patients with more advanced cirrhosis had hypocoagulable profiles on TEG [62]. There was a correlation between hypocoagulable TEG profile and risk of six-week incidence of variceal bleeding prior to liver transplantation.

•Additional small studies suggest that viscoelastic testing may be able to predict hypercoagulability in liver disease [63].

We typically do not use viscoelastic testing alone (as the sole basis) to guide decisions regarding transfusions for prophylaxis or rescue interventions. As noted below, we also incorporate information about disease state, comorbidities, and individual clinical judgment in making these decisions. Some recommendations support the use of viscoelastic testing in patients who are critically ill with acute on chronic liver failure [64]. While viscoelastic testing is a promising modality for hemostasis assessment prior to procedures in patients with cirrhosis, other societal guidelines acknowledge the need for further study to better define predictive capabilities and parameters for routine use [2,65]. (See 'General approach to managing bleeding' below.)

Interventions for hemostatic imbalance — In most patients, we do not intervene in the setting of asymptomatic laboratory changes (eg, elevations in the PT/INR or aPTT, decreases in the platelet count) that are thought to be due to the underlying liver disease.

In some cases, we give vitamin K to patients with suspected vitamin K deficiency. This includes individuals with suspected poor nutrition and cirrhosis, as well as those with cholestatic disease, diarrheal illness, or antibiotic use. The response can be useful in establishing the underlying degree of liver disease-related synthetic dysfunction but has not been shown to alter bleeding or thrombosis risk per se. Dosing is described below. (See 'General approach to managing bleeding' below.)

We do not administer Fresh Frozen Plasma (FFP) to "correct" an asymptomatic prolonged PT/INR, based on a large number of possible risks and costs of this approach (eg, transfusion reactions, volume overload, increased portal pressures) and the lack of good quality evidence that it provides any clinically important benefit [66]. (See "Clinical use of plasma components", section on 'Risks'.)

Cases that require further investigation or in which we may intervene include the following:

●Large or unexpected changes such as newly prolonged PT or aPTT or new decline in platelet count may require evaluation for the cause, which may include infection, new medication(s), or progression of the underlying liver disease. (See "Clinical use of coagulation tests", section on 'Evaluation of abnormal results'.)

●New acute severe thrombocytopenia, which can be associated with medication reaction, disseminated intravascular coagulation (DIC), thrombotic thrombocytopenic purpura (TTP) or other thrombotic microangiopathy, or heparin-induced thrombocytopenia (HIT) in patients receiving heparin. In the absence of bleeding or an imminent procedure, such cases warrant close observation and sometimes further investigation, but not platelet administration unless critical such as a platelet count <10,000 to 15,000/microL. (See "Diagnostic approach to thrombocytopenia in adults", section on 'Approach to evaluation'.)

●Active bleeding may require more aggressive correction of hemostatic abnormalities, although this should be based largely on a thorough clinical assessment rather than isolated laboratory testing such as the PT/INR, which does not provide a full picture of hemostasis in patients with liver disease. Bleeding prophylaxis prior to invasive procedures to prevent future bleeding is controversial, and data are too limited to support prophylactic plasma as a routine practice [2]. (See 'Bleeding' below and 'Invasive procedures' below.)

Liver disease versus DIC — Reduced synthesis of procoagulant and anticoagulant factors predominates in liver disease; the term "accelerated intravascular coagulation and fibrinolysis (AICF)" is used for these. (See 'Impaired hemostasis' above.)

In contrast, disseminated intravascular coagulation (DIC) is caused by ongoing intravascular thrombin generation with coagulation factor consumption. (See "Evaluation and management of disseminated intravascular coagulation (DIC) in adults".)

However, distinguishing liver disease from DIC can be challenging because the clinical and laboratory findings are similar. Therefore, clinical judgment plays a major role:

●Factor VIII activity levels generally are increased or normal in liver disease (factor VIII is produced in hepatic and non-hepatic endothelial cells) [67]. Therefore, a significant proportion of factor VIII production is unaffected by chronic liver disease. On the other hand, in DIC, consumption of factor VIII causes decreased factor VIII levels. (See "Evaluation and management of disseminated intravascular coagulation (DIC) in adults", section on 'Differential diagnosis'.)

●D-dimer levels are not useful in liver disease.

Importantly, liver disease and DIC may coexist, especially in the setting of infection/sepsis and/or malignancy. (See "Evaluation and management of disseminated intravascular coagulation (DIC) in adults".)

MEDICATION MANAGEMENT

Medications to avoid — Individuals with rebalanced hemostasis in the setting of liver disease have increased sensitivity to hemostatic challenges. Patients with well-compensated cirrhosis may tolerate anticoagulation and antiplatelet therapy with little increased risk of bleeding. However, in patients with more advanced liver disease or decompensated cirrhosis, sometimes avoiding medications that increase bleeding and/or thrombotic risk is prudent, unless the benefits clearly outweigh the risks (eg, acute treatment of a thromboembolic event).

●Anticoagulants in some individuals – We generally adhere to practice guideline recommendations when making decisions regarding venous thromboembolism (VTE) prophylaxis in patients with liver disease while hospitalized [68]. In patients with advanced cirrhosis and high risk for bleeding, it may be reasonable to avoid anticoagulation prophylaxis.

We generally avoid anticoagulants in individuals with decompensated Child-Pugh (CTP) class C cirrhosis [69]. Patients with decompensated cirrhosis who require anticoagulation (eg, for VTE or high-risk atrial fibrillation) are treated in an individualized manner with close collaboration between the hepatologist and the hematologist. (See "Cirrhosis in adults: Overview of complications, general management, and prognosis", section on 'Child-Pugh classification' and 'VTE therapy' below.)

●NSAIDs – We generally avoid nonsteroidal antiinflammatory drugs for routine treatment of pain and/or fever in all patients with cirrhosis. (See "Nonselective NSAIDs: Overview of adverse effects".)

●Anti-platelet drugs – For patients with decompensated cirrhosis, we generally avoid antiplatelet drugs such as aspirin or P2Y₁₂ receptor blockers such as clopidogrel, unless cardiovascular benefits exceed the bleeding risks of these agents.

●Over-the-counter medications, herbs, supplements – For all patients, we provide education regarding over-the-counter medications such as ginkgo biloba that may adversely affect hemostasis. (See "Clinical use of ginkgo biloba", section on 'Safety'.)

Discussions of over-the-counter hepatotoxins and medications that may exacerbate underlying liver disease are presented separately. (See "Drug-induced liver injury" and "Drugs and the liver: Metabolism and mechanisms of injury" and "Hepatotoxicity due to herbal medications and dietary supplements".)

Adjustments to anticoagulant dosing — Certain anticoagulants are predominantly metabolized by the liver.

●Warfarin – Warfarin is metabolized by the liver, but dosing is titrated to maintain a therapeutic international normalized ratio (INR). Difficulties can arise when the baseline INR in a cirrhosis patient is not within the normal range, and the target range is unclear. As indicated above, individuals with liver disease often have abnormalities in the INR measurement. (See 'Laboratory abnormalities' above.)

●DOACs – A percentage of hepatic metabolism occurs for all direct oral anticoagulants (DOACs). The product insert should be consulted for more details, as it may determine the optimal choice of anticoagulant for the patient.

For the direct factor Xa inhibitors (eg, apixaban, rivaroxaban, edoxaban) and dabigatran, product information suggests that these drugs should be used with caution or avoidance in patients with Child-Pugh class B or C. These issues are reflected in recommendations for dose adjustment from the US Food and Drug Administration (FDA) and the European Medicines Association (EMA), as illustrated in the table (table 1). Both clinical and translational data suggest that direct oral anticoagulants (DOACs) may have altered potency in more advanced liver disease, and caution with use in these populations is warranted [69-72].

Use of reversal agents for DOACs has been reported in patients with liver disease [73]. (See "Management of bleeding in patients receiving direct oral anticoagulants".)

●Heparins – Metabolism of heparins is more complex, but dosing may be simpler because unfractionated heparin is titrated to maintain the aPTT in a target range. Heparins require the function of antithrombin (AT) to inhibit thrombin and factor Xa (figure 2), and, since AT is synthesized in the liver, their activity may be reduced [74]. Anti-factor Xa levels appear to be less reliable in cirrhosis [75]. Low molecular weight (LMW) heparins are usually used for VTE prophylaxis, and prophylactic-dose LMW heparin has been safely used in individuals with cirrhosis [74].

BLEEDING

General approach to managing bleeding — Management of bleeding depends on the location and severity as well as the degree of hemostatic impairment. Variceal bleeding, which accounts for a high percentage of bleeding episodes overall in individuals with cirrhosis (approximately 80 percent), is considered to be predominantly a problem of increased portal pressure and resulting vascular abnormalities rather than a primary hemostatic defect [76]. (See 'Variceal bleeding' below.)

However, nonvariceal bleeding is common and was noted in one in five admissions for decompensated cirrhosis [77].

Our approach to management depends on how we answer the following questions [13]:

●Are other comorbidities present (eg, infection, kidney disease, malignancy)?

●What is the status of fibrinogen (level and function) and is hyperfibrinolysis clinically suggested (especially with delayed mucosal or wound oozing)?

●Is the patient on medications that alter the risk of bleeding or thrombosis?

We base our therapy on which hemostatic defect(s) is thought to be involved, as follows (figure 3):

●Comorbidities – Treatable comorbidities can contribute significantly to impaired hemostasis; thus, we address the following:

•Infection – Treat concomitant infections, which may impair hemostasis through effects on vascular function, endogenous heparinoids, and platelets.

•Acute and chronic kidney disease – Optimize kidney function, since uremia may impair normal platelet function. Data are emerging to suggest that acute kidney injury is associated with altered hemostatic function and may increase the risk of bleeding [17,78,79].

•Portal hypertension – Avoid raising portal pressures (eg, with aggressive transfusions in the absence of active bleeding or shock). Effective volume management including avoiding over-transfusion is essential.

•DIC – DIC is uncommon, but if suspected, identify and treat the underlying cause.

•Antithrombic medication – If the patient is receiving an antithrombotic medication (anticoagulant or antiplatelet agent), decisions about whether to stop and/or reverse the drug are individualized according to the underlying indication and the severity and site of bleeding.

We feel more strongly about discontinuation/reversal for individuals with more severe bleeding and less need for anticoagulation (eg, active life-threatening hemorrhage), and we feel more strongly about continuing the anticoagulant for individuals with a greater or obligate need for antithrombotic medications (eg, mechanical heart valve). Individual approaches are often necessary, and working closely with providers across disciplines is essential.

●Clot formation – Monitor platelets and fibrinogen levels and a global viscoelastic assay of hemostasis such as thromboelastography (TEG) or thromboelastometry (ROTEM) (see 'Coagulation testing' above), if available. Decide which hemostatic products to administer based on the mechanism identified or the deficiency that is contributing to impaired clot formation along with our clinical judgment, rather than a single laboratory measurement (figure 3). For the most part, management decisions are unaffected by results of prothrombin time (PT)/international normalized ratio (INR) and aPTT testing, and we avoid the use of isolated PT/INR testing for bleeding assessment. Consultation with a clinician with expertise in this area may be helpful.

The value of viscoelastic testing was illustrated in a trial that randomly assigned 96 individuals with cirrhosis and non-variceal upper gastrointestinal bleeding to transfusions guided by TEG or by the PT/INR and platelet count [80]. Both groups had similar control of bleeding and similar length of hospital stay, but compared with the PT/INR/platelet count group, the TEG group received significantly less blood products (median dose of plasma, 880 versus 440 mL; median number of platelet units, 2 versus 1) and experienced fewer transfusion-related adverse events. There was a trend towards better survival in the TEG group that did not reach statistical significance. There are some concerns about the trial design (eg, endoscopic interventions were not described, and transfusions were based on laboratory values rather than clinical endpoints of bleeding). Furthermore, the group selected as standard of care received plasma if their INR exceeded 1.8, a practice not supported by evidence in this population [2]. However, the overall results are encouraging and support the need for further study of this approach.

Another trial assessed the use of TEG to guide blood product transfusions compared with conventional markers of PT/INR and platelet count [81]. Only four patients in the TEG group received transfusions, versus 100 percent of the PT/INR/platelet count group. Outcomes including control of bleeding, rebleeding at five days, and mortality were similar between groups. In general, variceal-related bleeding is associated with portal hypertension rather than hemostatic failure, and standard approaches to portal hypertension management and endoscopic treatment are recommended. Routine transfusion of plasma or platelets based on coagulation testing results is not recommended in most cases of acute variceal hemorrhage. (See "Overview of the management of patients with variceal bleeding".)

•Vitamin K – For individuals with possible vitamin K deficiency (eg, suspected poor nutrition, cholestatic disease, diarrheal illness, antibiotic use), we administer vitamin K. Vitamin K can be administered (see 'Laboratory abnormalities' above) at a dose of 10 mg by intravenous infusion or 10 mg orally per day for three days. The intravenous infusion should be given when there is a concern about absorption and administered slowly (ie, no faster than 1 mg/minute) to reduce the rare but possible risk of anaphylaxis [82].

•Cryoprecipitate – If there is active or poorly controlled bleeding, it may be reasonable to administer a dose of Cryoprecipitate (eg, one bag per 10 kg of body weight) (table 2) while awaiting the results of laboratory testing for correction of presumptive hypofibrinogenemia or dysfibrinogenemia (abnormally functioning fibrinogen) [83].

For those who can wait for the results of the fibrinogen level, when this becomes available, administer a source of fibrinogen (preferably Cryoprecipitate, which creates a smaller volume load than Fresh Frozen Plasma [FFP]) to maintain a fibrinogen level ≥100 to 120 mg/dL [13].

Some experts also give Cryoprecipitate for patients with persistent bleeding despite a fibrinogen level ≥100 to 120 mg/dL, based on a presumptive diagnosis of dysfibrinogenemia (typical dose, one bag per 10 kg body weight).

The benefits of giving plasma products to individuals with an adequate fibrinogen level must be weighed against the potential for increasing portal pressure, especially in a patient with varices, even if these are not the site of bleeding.

We generally avoid more aggressive therapies with a greater risk of promoting thrombosis, such as fibrinogen concentrates, prothrombin complex concentrates (PCCs) (table 3), or recombinant activated factor VII (rFVIIa), because these products are costly and lack significant supportive evidence. However, these agents may be appropriately used on a case-by-case basis if bleeding continues despite other interventions with caution recognizing the risk from case reports mentioning thromboembolic events and potential DIC [84,85]. In vitro and clinical data are also emerging supporting the use of fibrinogen concentrates during liver transplantation [86,87].

We also generally avoid giving FFP due to the large volume load and potential adverse effect on portal pressures. The only instance in which we might consider FFP is for unexplained bleeding that persists after fibrinogen has been repleted with Cryoprecipitate or in patients requiring large numbers of red blood cell (RBC) transfusions.

●RBCs and platelets – Monitor complete blood count (CBC) with platelet count.

•In active bleeding not related to portal hypertension, transfuse platelets to maintain a platelet count >50,000 to 55,000/microL (>100,000/microL for active, severe, or central nervous system bleeding). Guidelines from various societies regarding platelet count thresholds are summarized in a 2022 review [88].

•Platelet dysfunction (eg, due to uremia or drugs) may be assessed somewhat by viscoelastic testing (eg, TEG, ROTEM); however, this has not been adequately studied in patients with liver disease. If bleeding is severe and platelet function is thought to be impaired, platelet transfusion may be appropriate. (See "Platelet transfusion: Indications, ordering, and associated risks", section on 'Platelet function disorders'.)

•Transfuse RBCs to maintain a hemoglobin level above 7 g/dL. The hemoglobin level may be used to guide RBC transfusions in individuals without ongoing, active bleeding. Higher target levels of hemoglobin could increase portal pressures but may be appropriate in certain circumstances such as with coexisting vascular disease. (See "Indications and hemoglobin thresholds for RBC transfusion in adults", section on 'Acute bleeding'.)

●Fibrinolysis – Maintain a high suspicion for hyperfibrinolysis, evidenced by delayed bleeding or persistent oozing from mucocutaneous sites. Hyperfibrinolysis may have been overestimated in the past due to reduced clearance of fibrinolytic markers, and its clinical bleeding implications still seem poorly understood [89].

Excess fibrinolytic activity has been reported in the past with saliva and ascites [90]. Thus, a high index of suspicion for hyperfibrinolysis is appropriate in individuals undergoing common dental procedures and with oral/mucosal bleeding. Testing for hyperfibrinolysis using viscoelastic testing such as TEG, ROTEM, or clot lysis times remains challenging. More often, a clinical diagnosis of this phenomenon is necessary.

For suspected hyperfibrinolysis based on clinical and/or laboratory features, it is appropriate to administer an antifibrinolytic agent. Available agents include tranexamic acid and epsilon aminocaproic acid. These both can be administered intravenously or orally, or in soaked gauze (eg, during dental procedures). Optimal dosing in the setting of liver disease has not been established; use of dosing extrapolated from other settings (eg, anticoagulant-associated bleeding) may be reasonable.

A randomized trial in 12,009 patients presenting with acute gastrointestinal bleeding (41 percent with liver disease) who were assigned to receive tranexamic acid infusion or placebo found similar five-day mortality rates in both groups [91]. However, there was an increased risk of venous thromboembolism (VTE) in the tranexamic acid group (RR 1.85, 95% CI 1.15-2.98), and on subgroup analysis, the elevated VTE risk was more pronounced in patients with suspected variceal bleeding or liver disease (RR 7.26, 95% CI 1.65-31.9). Caution is advised when using antifibrinolytic agents in patients with liver disease who are bleeding as they may be at risk to become hypercoagulable. (See "Management of bleeding in patients receiving direct oral anticoagulants", section on 'Antifibrinolytics and other pro-hemostatic therapies'.)

Variceal bleeding — Variceal bleeding is a common and potentially life-threatening complication of liver disease, occurring in 25 to 35 percent of individuals with cirrhosis, and accounting for the majority of cirrhotic bleeding episodes [76]. (See "Overview of the management of patients with variceal bleeding" and "Methods to achieve hemostasis in patients with acute variceal hemorrhage".)

Although abnormal coagulation testing is especially concerning in the setting of varices, the major cause of variceal bleeding is thought to be local vascular deformations and hemodynamic changes (eg, increased portal pressure) rather than a bleeding diathesis [39]. Hemostatic mechanisms appear to play only a transient role as evident in the unstable "platelet plug" or nipple sign, which is sometimes evident at the point of varix breach.

Thus, the most important interventions for prevention and treatment of bleeding involve reducing portal pressure and ligating bleeding lesions. Details are presented separately. (See "Methods to achieve hemostasis in patients with acute variceal hemorrhage".)

Interventions that may be helpful for excessively high INR or excessively low platelet count are noted above. (See 'Interventions for hemostatic imbalance' above and 'General approach to managing bleeding' above.)

Use of anticoagulation may be concerning in patients with liver disease presenting with active variceal bleeding. In a retrospective study involving 320 individuals with cirrhosis who were treated with an anticoagulant (mostly warfarin to reduce stroke risk in atrial fibrillation), there were 59 episodes of bleeding, and multivariate analysis identified esophageal varices as the greatest independent predictor of bleeding risk (odds ratio [OR] 5.7, 95% CI 1.8-17.7) [92]. Of 14 individuals with variceal bleeding, five (approximately one third) were not receiving a nonselective beta blocker to reduce bleeding risk. These data emphasize the importance of close monitoring for varices, use of interventions to reduce the development of varices, and, for those with varices, the use of prophylactic therapies to reduce the risk of bleeding such a beta blocker or endoscopic variceal ligation.

However, in a study that evaluated risk of upper gastrointestinal bleeding (variceal and non-variceal) in patients with cirrhosis on anticoagulation compared with a similar cohort of patients not on anticoagulation, outcomes were similar, including failure of endoscopic control of bleeding at five days, need for rescue therapy, and mortality at six weeks [93]. Another study analyzing bleeding risk from endoscopic variceal band ligation in patients with cirrhosis receiving low molecular weight (LMW) heparin found no significant increase in bleeding risk in the group treated with LMW heparin [94].

In patients with acute variceal bleeding, we recommend holding anticoagulation in the short term during bleeding events. Consideration for reversal of anticoagulation in individual circumstances may be necessary, but the risks and benefits must be weighed carefully. (See "Management of bleeding in patients receiving direct oral anticoagulants" and "Management of warfarin-associated bleeding or supratherapeutic INR" and "Heparin and LMW heparin: Dosing and adverse effects", section on 'Reversal'.)

The choice among prophylactic approaches is discussed separately. (See "Primary prevention of bleeding from esophageal varices in patients with cirrhosis".)

In patients with active bleeding despite these interventions, the use of pro-hemostatic agents may be appropriate as rescue agents in rare instances. However, potential benefits must be weighed against risks such as increased portal pressure with plasma transfusions and increased thrombotic risk with coagulation factor concentrates. We are unaware of studies that have adequately evaluated rFVIIa or PCC in uncontrollable hemorrhage, although they have occasionally been used to try to temporize an unstable situation in case reports. As mentioned above, the use of antifibrinolytic agents during gastrointestinal bleeding did not improve mortality and was associated with a higher risk of VTE.

INVASIVE PROCEDURES

General approach to invasive procedures — Management may be particularly challenging in individuals who require invasive procedures [2,5]. Individuals with liver disease may have an increased risk of bleeding with an invasive procedure; the mechanisms may be multifactorial and depend on the type of procedure, its level of bleeding risk, the degree of liver decompensation, and other comorbidities [5,95].

Assess bleeding risk factors — Bleeding risk and risk factors were illustrated in a prospective, multicenter observational study involving 1187 patients with cirrhosis who were admitted to the hospital for 3006 nonsurgical procedures [95]. Procedural-related bleeding was reported in 7 percent of admissions (3.0 percent of procedures). Major bleeding was rare, occurring in 2 percent of admissions (1 percent of procedures). In univariate analysis, pre-procedure prophylaxis (eg, platelet and plasma transfusion) was provided more commonly prior to procedures that resulted in bleeding (19 percent) than procedures without bleeding (7 percent; p=0.001). In multivariate analysis, higher rates of procedure-related bleeding occurred with:

●Higher-risk procedures: odds ratio (OR) 4.64, 95% CI 2.44-8.84

●Higher MELD score at admission: OR 2.37, 1.46-3.86

●Higher body mass index (BMI): OR 1.4, 95% CI 1.10-1.80

Pre-procedure international normalized ratio (INR) and platelet count were not predictive of bleeding [95]. Survival analysis indicated increased mortality strongly correlated with procedural related bleeding, particularly patients with major bleeding (HR 12.97, 95% CI 6.42-26.19).

Assess and optimize hemostasis — To address this paradigm shift in management of patients with cirrhosis undergoing procedures, international society guidance has emerged from several organizations, such as:

●AASLD – American Association for the Study of Liver Diseases (AASLD) [2]

●EASL – European Association for the Study of the Liver (EASL) [96]

●AGA – American Gastroenterological Association (AGA) [65]

In general, these guidance statements recommend against the use of conventional tests of hemostasis (eg, platelet count and INR) and against the use of pre-procedure bleeding prophylaxis in most cases. (See 'Society guideline links' below.)

Common themes that arise include the following, as summarized in guidelines from the AASLD published in 2021 [2]:

●Optimizing hemostasis – We treat comorbidities that can interfere with normal hemostasis, as done for patients with bleeding (see 'Bleeding' above).

Optimizing kidney function is an important intervention. (See "Overview of the management of chronic kidney disease in adults".)

Treating infections is also important, along with correcting vitamin K deficiency, and others, especially if there is active bleeding. (See 'General approach to managing bleeding' above.)

While viscoelastic testing does not clearly predict bleeding, we find use of global assays such as thromboelastography (TEG) or thromboelastometry (ROTEM) often can reassure operators prior to many procedures that hemostasis is intact. Of note, we do not administer Fresh Frozen Plasma (FFP) prior to procedures in individuals with liver disease, regardless of laboratory values. (See 'Laboratory abnormalities' above.)

●Assessment of hemostasis – Accurate assessment of hemostasis is limited using traditional and global measures of hemostasis. However, viscoelastic testing prior to procedures has been shown to reduce need for prophylaxis.

The value of using a measure of global clot formation to guide blood product use was demonstrated in a trial that randomly assigned 60 patients with cirrhosis who were undergoing an invasive procedure to be managed using thromboelastography (TEG) or according to standard hospital protocol [55]. Patients assigned to the TEG arm received FFP if the reaction time (r) was >40 minutes (normal range for this study, 12 to 26 minutes) and platelets if the maximum amplitude (MA) was <30 mm (figure 1 and table 4). Those assigned to standard protocol received FFP for international normalized ratio (INR) >1.8 and platelets for a count <50,000/microL. The overall use of blood products was lower in the TEG group (17 versus 100 percent). There was no bleeding in the TEG group and one episode of bleeding in the control group (hemoperitoneum following large-volume paracentesis). This study highlights the difficulty in assessing need for prophylactic therapy without a standardized and accurate measure of in vivo hemostasis [97,98].

In a series of 72 patients with decompensated cirrhosis, TEG findings accurately discriminated between those with major life-threatening bleeding (all had an MA <30 mm) and mild or no bleeding (MA >30 mm), whereas conventional coagulation testing and platelet counts were not useful in determining bleeding risk [99]. Two subsequent studies using viscoelastic testing in patients with cirrhosis also demonstrated that value of viscoelastic testing prior to central venous line placement and high-risk invasive procedures to reduce prophylactic transfusion; there were no bleeding events in either study [56,57]. Only one of these three trials incorporated a restrictive arm that did not use pre-procedure prophylaxis. Future studies should include restrictive arms with viscoelastic testing prior to determine the utility of these tests in predicting bleeding and to establish useful, clinically relevant parameters.

●Indications for platelet transfusion – Platelet transfusions are often given to individuals with severe thrombocytopenia (platelet count <50,000/microL) who require an urgent or emergency procedure with a high bleeding risk such as spinal surgery, cardiac surgery, liver biopsy, removal of a large gastrointestinal polyp, or transjugular intrahepatic portosystemic shunt (TIPS) procedure. In a study examining thrombin generation using a thrombin generation assay, platelet counts above 56,000 were associated with normal thrombin production [100]. Results from prior retrospective studies provided conflicting data regarding the risk of bleeding in the setting of thrombocytopenia [101,102]. A large prospective multicenter study from 2023 demonstrated that platelet count prior to a procedure was not a significant predictor of bleeding [95]. While prophylactic platelet transfusions were provided prior to only 2.4 percent of procedures, those procedures in which platelets were transfused were associated with more bleeding.

There is a clear confounding relationship between thrombocytopenia and the degree of hepatic decompensation, with progressive thrombocytopenia worsening in more advanced liver disease [95]. In addition, bleeding events with procedures are rare, and past studies have lacked control arms in which prophylactic platelet transfusions are omitted.

We advocate an individualized approach for each patient and the specific procedure type when planning for procedures. In general, AASLD, EASL, and AGA guidance recommends avoiding platelet transfusions or thrombopoietin receptor agonists (TPO-RAs) prior to most procedures [2,65,96]. However, EASL guidance recommends considering supportive approaches in patients with severe thrombocytopenia who are undergoing high risk procedures where local hemostasis cannot be achieved on a case-by-case basis [96]. Selected platelet count thresholds for specific invasive procedures are described separately. (See "Platelet transfusion: Indications, ordering, and associated risks", section on 'Preparation for an invasive procedure'.)

●Other interventions to raise the platelet count – Thrombopoietin receptor agonists (TPO-RAs) increase the platelet count by stimulating the TPO receptor. Their effect is transient and takes several days to be effective. We consider using TPO-RA in an individual with liver disease and severe thrombocytopenia who is undergoing a scheduled invasive procedure with a very high bleeding risk (eg, spinal surgery, cardiac surgery, large polypectomy), provided there is sufficient time (10 to 13 days) to start the drug and to document an increase in the platelet count to ≥50,000 microL.

Clinicians may reasonably use restrictive strategies without administering a TPO-RA in most scenarios, such as close observation, depending on the bleeding risk of the procedure [2]. As an example, we would not use a TPO-RA for a patient undergoing a low-risk procedure such as paracentesis, routine endoscopy, or central line placement, as the risks and burdens are likely to outweigh the benefits. More urgent or emergency procedures in the setting of severe thrombocytopenia are likely to require platelet transfusions, since TPO-RAs take several days to raise the platelet count. Recent guidelines recommend against routine use of TPO-agonists to correct thrombocytopenia in patients with cirrhosis prior to procedures, including patients critically ill in the intensive care unit setting [2,64,65,103].

Two TPO-RAs are approved for use in patients with chronic liver disease and thrombocytopenia prior to undergoing procedures. The clinical trials that led to their approval demonstrated clear efficacy and safety in raising platelet counts; however, the trials were not designed to assess overall bleeding risk in thrombocytopenia, as they lacked restrictive cohorts who did not receive a TPO-RA or platelet transfusion. Furthermore, bleeding events were rare, and most of the procedures carried a low risk for bleeding.

•Avatrombopag – In a pair of randomized trials that included 435 individuals with chronic liver disease and severe thrombocytopenia who required an invasive procedure, the TPO-RA avatrombopag was demonstrated to reduce the need for platelet transfusion (platelet transfusions given in 12 to 34 percent of avatrombopag-treated patients versus 64 to 66 percent of placebo-treated patients) [104]. There was no increase in complications (one case of portal vein thrombosis two weeks after completion of therapy in the avatrombopag group, one episode of pulmonary embolism and one myocardial infarction in the placebo group) or other serious safety signals. Avatrombopag was approved by the US Food and Drug Administration (FDA) in 2018 for the treatment of thrombocytopenia in adults with chronic liver disease who are undergoing a scheduled procedure [105]. The drug should be started 10 to 13 days before the procedure and taken for five days. The effect will wane over time and the platelet count is expected to gradually return to baseline once the drug is discontinued; thus, close attention should be paid to the timing of administration relative to the procedure. The dose of avatrombopag is based on the patient's platelet count (<40,000/microL: 60 mg; 40,000 to 50,000/microL: 40 mg).

•Lusutrombopag – Three randomized trials (as yet unpublished) evaluated use of the TPO-RA lusutrombopag in individuals with chronic liver disease and thrombocytopenia undergoing a scheduled invasive procedure. Compared with placebo, lusutrombopag greatly increased the likelihood of avoiding a platelet transfusion (13 to 29 percent avoided platelet transfusions with placebo versus 78 to 65 percent with lusutrombopag) [106]. The dose was 3 mg daily for up to seven days prior to the scheduled procedure. Lusutrombopag received approval in 2015 in Japan and in 2018 in the United States for this indication (raising the platelet count in individuals with severe liver disease and thrombocytopenia prior to an invasive procedure) [106,107].

Additional information about these and other TPO-RAs is presented separately. (See "Clinical applications of thrombopoietic growth factors".)

●Indications for plasma products – We do not routinely administer FFP or Cryoprecipitate to "correct" the PT/INR value prior to a procedure, because several large reviews of available evidence have shown no clinical benefit [3,13,108-111]. The underpinnings of INR as a target and measure of bleeding risk (outside of warfarin therapy) do not apply to patients with cirrhotic coagulopathy, which has a completely different pathophysiology than that seen in warfarin anticoagulation.

Additional risks and disadvantages of giving FFP include transfusion reactions, volume overload (and associated increases in portal pressure, especially in individuals with varices), and infection. Administration of FFP also may create a significant delay while awaiting repeated laboratory testing of the PT/INR [83].

●Reducing vascular engorgement – We avoid engorgement of the collateral bed as much as possible by limiting volume expansion [39]. Because of the mucosal engorgement of the capillary bed in the tributaries of the portal venous system, we recommend optimization of portal hypertension and careful attention specifically to procedures involving gut mucosa (eg, endoscopy).

Non-surgical invasive procedures — Procedural-related bleeding in cirrhosis is overall rare, and available tests of hemostasis do not accurately predict bleeding prior to procedures [95]. Investigators have routinely grouped low- and high-risk procedures together in studies assessing bleeding. When bleeding does occur, it is not clear that prophylaxis would have prevented it (eg, laceration of an artery during paracentesis).

We advocate for individual assessment of each patient's circumstances and bleeding risk specific to the procedure. In general, we favor approaches to minimize unnecessary use of prophylaxis except in high-risk situations.

In addition to addressing treatable comorbidities that may affect hemostasis, we use the following procedure-specific interventions:

Paracentesis — Diagnostic and therapeutic paracentesis carries a low risk for bleeding complications. In most cases, we do not use pre-procedure prophylaxis, which is consistent with guidelines in the United States [2,112,113]. Exceptions include patients who have bled before with paracentesis or who have known or suspected hyperfibrinolysis or some other overtly evident bleeding such as mucosal bleeding. With proper needle and cannula placement, bleeding in this situation is usually venous, from the collateral bed of the abdominal wall venous system. Thus, avoiding engorgement of the collateral bed (as with plasma transfusions) is important to minimize bleeding risk. Studies suggest that individualized approaches are necessary, bleeding risk is not completely uniform, and bleeding risk likely depends on dynamic factors for each patient [114]. (See 'Variceal bleeding' above.)

Thoracentesis — Thoracentesis is commonly performed in individuals with advanced liver disease and is considered a low-risk procedure [115]. Sparse data exist to establish overall bleeding risk, and we do not routinely recommend use of prophylaxis prior to thoracentesis. One retrospective study examined patients with perceived "coagulopathy" undergoing a total of 1009 thoracentesis (portion with liver disease not identified) and found no significant difference with or without prophylaxis (no bleeding events in the non-prophylaxis group) [116].

Liver biopsy — The use of liver biopsy for staging fibrosis is declining, but biopsy remains an essential diagnostic tool in many conditions. It always raises concerns about bleeding, especially in patients with suspected coagulopathy and low platelets, and even more so in those with added risks such as chronic kidney disease or increased risk of volume overload. However, a review of 200 patients in whom "liver bleeding time" was evaluated at the time of liver biopsy found that there was no correlation between the degree of bleeding and any measure of hemostasis tested (eg, prothrombin time [PT], platelet count, whole blood clotting time) [117].

There is no well-defined platelet cut-off that can be used to determine need for platelet correction prior to liver biopsy in patients with chronic liver disease. In cases where platelet correction is felt to be necessary, a careful assessment of patient medications and comorbidities is essential. Providing a TPO-RA or platelet transfusions to increase the platelet count to >50,000/microL can be considered in high-risk patients for whom there is concern for potential bleeding. Other considerations include transfusing Cryoprecipitate when fibrinogen is low and perceived bleeding risk is high, to raise the fibrinogen to >120 mg/dL.

Transvenous biopsy is an alternative approach in particularly high-risk patients, although the quality of specimens is sometimes limiting and overall complication rates are similar to percutaneous biopsy, albeit usually in a higher risk population. Our practice is consistent with guidance from the American Association for the Study of Liver Diseases, which suggests an individualized approach [2,118].

Dental extractions — Dental extractions can usually be performed using local hemostatic measures alone (without systemic interventions to enhance hemostasis) [119,120]. One trial randomly assigned 43 patients with cirrhosis awaiting liver transplantation who had an INR 2.0 to 3.0 and/or a platelet count ≤50,000/microL to receive intranasal DDAVP (300 mcg) or transfusions (FFP, 10 mL/kg and/or one unit of single-donor platelets, based on INR and platelet count) prior to the dental procedure [121]. Hemostasis was similar in the two groups, and adverse effects were less with DDAVP (no major events versus one bleeding and one allergic reaction in the transfusion group). This trial highlights the need for placebo arms to assess baseline bleeding risk. Use of topical gauze soaked in aminocaproic acid is a useful adjunct to prevent delayed bleeding.

Palliative drain placement — In some situations involving terminal care of a patient with cirrhosis, indwelling drains may be used to better control refractory ascites. In these situations, which are often accompanied by a degree of kidney failure, prophylactic therapy may be warranted, especially if a tunneled catheter is to be placed, due to increased risk of trauma to the portal venous collateral bed of the abdominal wall.

Major surgery — Surgical bleeding is separate from the above discussion of non-surgical, procedure-related bleeding. Interventions for patients who require major surgery, including liver transplant, must be individualized. Often surgical bleeding is directly instigated and managed by local surgical measures.

Increasingly, global measures of hemostasis such as TEG and ROTEM are being used by the anesthesiologist to guide optimal hemostatic measures. Global viscoelastic tests may assist with management of bleeding during the operation and in the post-operative period. (See 'General approach to managing bleeding' above.)

In the absence of these tests, we would recommend a similar approach as outlined above, which includes optimizing the platelet count (in severe thrombocytopenia), fibrinogen level, and comorbidities such as reduced kidney function; and avoiding the use of INR values to guide therapy.

As with all procedures, avoidance of volume expanding the portal collateral circulation is essential to optimize safety. In some cases, especially when there are abundant collaterals in the operative field (such as colon resection or uterine resection), preoperative transjugular intrahepatic portosystemic shunt (TIPS) may be appropriate in selected patients to decrease flow in portal collaterals. Individual assessment and subspeciality collaboration are recommended when this approach is sought, as data to support this practice are sparse. (See "Overview of transjugular intrahepatic portosystemic shunts (TIPS)".)

PORTAL VEIN THROMBOSIS (PVT) — Partially or completely occlusive thrombosis in the portal circulation occurs with an estimated incidence of approximately 5-16 percent per year in patients with stable cirrhosis without liver cancer [122,123]. PVT occurs more frequently in more advanced cirrhosis. PVT may be acute or chronic. Focal left or right branch PVT is relatively more common and often clinically silent, although its development may contribute to overall organ atrophy. (See "Epidemiology and pathogenesis of portal vein thrombosis in adults", section on 'Epidemiology'.)

The mechanism of PVT in liver disease is likely to involve a combination of factors, including reduction in natural anticoagulants, decreased portal blood flow, and endothelial dysfunction, along with other prothrombotic comorbidities in some individuals (eg, genetic predisposition, hepatocellular cancer) [124]. (See 'Prothrombotic changes' above and "Epidemiology and pathogenesis of portal vein thrombosis in adults", section on 'Pathogenesis'.)

The clinical significance of PVT continues to be debated. It is unclear whether development of PVT represents an epiphenomenon from progressive portal hypertension or a treatable condition in cirrhosis in the absence of overt symptoms. The need for therapy is clearer when there are obvious related symptoms or in the setting of liver transplantation candidacy. The role of the portosystemic collateral bed may also obscure the effects of PVT as slow flow due to shunting (steal) through large collaterals may be as significant whether or not associated with actual thrombosis [125].

PVT risk factors — The risk of PVT is increased in individuals with cirrhosis, and PVT occurs more frequently in those with greater disease severity (eg Child-Pugh class C) [126].

In addition to the severity of liver disease, other factors that increase PVT risk include the presence of hepatocellular cancer and genetic susceptibility factors (eg, inherited thrombophilia). Case-control studies have also noted that patients with cirrhosis and PVT have a higher incidence of factor V Leiden mutation. In the randomized trial of enoxaparin versus no therapy, 2 of 70 patients (3 percent) had a factor V Leiden and none had a prothrombin G20210A mutation [122]. The role of testing for genetic thrombophilic factors in this population is not well-defined and the role of testing should be determined on an individual basis [124,127].

PVT prophylaxis — Prevention of PVT in individuals with liver disease focuses on optimizing hepatic function, reducing portal venous pressure, and increasing portal flow, which diminishes stasis. (See "Cirrhosis in adults: Overview of complications, general management, and prognosis", section on 'General management' and "Overview of transjugular intrahepatic portosystemic shunts (TIPS)".)

The role of prophylactic anticoagulation is uncertain, and we do not use it outside of a separate indication. However, if the patient has a separate indication for anticoagulation, we do not consider cirrhosis to be a contraindication for the use of an anticoagulant, following assessment and treatment of high-risk varices if present.

A randomized trial assessed anticoagulation to prevent PVT was conducted using fixed-dose low molecular weight (LMW) heparin (enoxaparin 4000 international units subcutaneously once daily) versus no therapy for one year in 70 patients with advanced cirrhosis (Child-Pugh class B, mean model for end-stage liver disease [MELD] score 13) [122,128]. Portal vein patency was documented prior to entry. Findings included the following:

●A greater reduction in the incidence of PVT by ultrasound and computed tomography (the primary endpoint) was seen with enoxaparin compared with controls at all time points: one year (0 versus 17 percent), two years (0 versus 28 percent), or study end (approximately three and a half years; 9 versus 28 percent).

●Secondary endpoints including hepatic decompensation (eg, development of ascites, encephalopathy, peritonitis) were also less frequent with enoxaparin (12 versus 59 percent), and there was a modest but significant improvement in survival (8 versus 13 deaths).

●Complications of enoxaparin were minimal, with no episodes of severe bleeding and one episode of thrombocytopenia that led to discontinuation of enoxaparin. Two enoxaparin-treated patients and one control patient had variceal bleeding.

It should be noted that this trial was small, focused on patients with moderate disease (mostly Child-Pugh class B or early C), and had methodological concerns including premature data analysis, atypical patient population, and un-blinding [122,128,129]. No study has confirmed or refuted the findings of this single trial.

Further studies are essential to confirm these results in larger and more heterogenous cohorts; trials that may confirm or refute these results are enrolling [130].

As noted above, if a patient has another indication for anticoagulation, we typically counsel that cirrhosis in itself would not be a contraindication following assessment for risk factors (eg, severe thrombocytopenia) and treatment of high-risk varices if present.

PVT treatment — Treatment of PVT is discussed in detail in separate topic reviews. While controversial, evidence summarized in a 2017 meta-analysis suggests that anticoagulation is a safe and effective therapy for PVT in select patients with cirrhosis [131]. This subject is discussed in detail separately. (See "Acute portal vein thrombosis in adults: Clinical manifestations, diagnosis, and management" and "Chronic portal vein thrombosis in adults: Clinical manifestations, diagnosis, and management".)

VENOUS THROMBOEMBOLISM (VTE) — Management may be particularly challenging in individuals who require anticoagulant therapy [2,5]. Exclusion of patients with an underlying coagulopathy from the majority of VTE trials has led to an extreme paucity of data regarding risks, prevention, and treatment of peripheral deep vein thrombosis (DVT) and pulmonary embolism (PE) in patients with severe liver disease.

VTE risk — Several retrospective studies have evaluated the risk of VTE in individuals with chronic liver disease. Although the rates of VTE differ across studies, it is clear that patients with cirrhosis are at risk for VTE. Notably, VTE-associated mortality is significantly increased in patients with cirrhosis [132,133]. A single-center study found an increased risk of mortality in patients with cirrhosis who developed VTE compared with controls (7 versus 3 percent). The 30-day mortality among patients with cirrhosis was 7 percent for those with DVT and 35 percent for those with PE. Data from a large registry trial suggest patients with cirrhosis who are treated with anticoagulation have a high risk of VTE recurrence and a high risk of mortality related to bleeding [134].

●Any severity of liver disease – The risk of VTE was evaluated in a 2015 meta-analysis of 14 studies involving patients with liver disease that found a pooled incidence VTE of 1 percent (DVT: 0.6 percent; PE: 0.28 percent) [135]. In a 2016 meta-analysis focused solely on hepatitis C virus (HCV) infection that included over 100,000 HCV-infected individuals and over 8 million controls, the odds ratio (OR) for VTE was 1.9 (95% CI 1.4-2.6) [136]. In a 2009 retrospective study involving almost 9,000,000 patients hospitalized over an approximately 25-year period, the prevalence of VTE was 0.6 percent in those with alcoholic liver disease and 0.9 percent in those with nonalcoholic liver disease [137]. These were considered lower than seen in most patients hospitalized with other conditions.

●Cirrhosis – Patients with cirrhosis may have a higher risk of VTE. This was demonstrated in a meta-analysis of 11 studies involving patients with cirrhosis that found an increased risk of VTE compared with controls (OR 1.7, 95% CI 1.3-2.2) [138].

•In a 2011 database review of 449,798 hospitalizations for cirrhosis, 8231 were for VTE (1.8 percent) [139].

•In a 2015 review from an insurance database in Taiwan that compared 2223 patients with cirrhosis versus 22,230 matched controls, the adjusted hazard ratio (HR) of VTE in patients with cirrhosis was 1.71 (95% CI 1.05-2.78) [140]. The risk was especially high in those with advanced cirrhosis.

As expected, additional comorbidities and/or other VTE risk factors appear to further increase VTE risk. Patients with non-alcohol steatohepatitis-related cirrhosis may be at an increased risk to develop VTE [45].

VTE prophylaxis — VTE prophylaxis is appropriate for high-risk hospitalized medical and surgical patients with liver disease, similar to individuals hospitalized with other acute conditions [68,133]. Hospitalized patients with liver disease have multiple risk factors for VTE. Risk assessment models such as the Padua Prediction Score or IMPROVE score can be used to assess risk of VTE in hospitalized patients with liver disease [141,142].

VTE prophylaxis has historically been underused in this population, which is likely due to concerns about increased bleeding risk and/or the mistaken assumption that individuals with severe liver disease are "auto-anticoagulated" based on the finding of a prolonged prothrombin time (PT). Literature in this area is limited, and studies are small and underpowered to assess the efficacy and safety of VTE prophylaxis.

The safety of VTE prophylaxis in patients with cirrhosis was evaluated in a study of 235 patients with chronic liver disease (mean model for end-stage liver disease [MELD] score, 16.2) who were hospitalized 355 times and received VTE prophylaxis as part of an institution-wide protocol [143]. Approximately three-fourths received unfractionated heparin, and the remainder received a low molecular weight (LMW) heparin. The use of VTE prophylaxis was associated with nine episodes of gastrointestinal bleeding (2.5 percent), most of which were minor (<2 g/dL decline in hemoglobin), and two episodes of heparin-induced thrombocytopenia (HIT; 0.5 percent). Five patients (1.4 percent) developed a VTE despite prophylaxis.

A retrospective cohort study compared hospitalized patients with cirrhosis who did or did not receive VTE prophylaxis [144]. Out of 600 hospital admissions, VTE prophylaxis was used in 49 percent. The rates of bleeding and VTE were similar in the two groups, including in-hospital bleeding events (8 versus 6 percent), gastrointestinal bleeding (approximately 3 percent in both groups), and new VTE (approximately 2 percent in both groups). On subgroup analysis, patients receiving unfractionated heparin had a higher risk of bleeding (OR 2.38, 95% CI 1.15-4.94), whereas low molecular weight (LMW) heparin was not associated with bleeding.

Exceptions to the use of VTE prophylaxis may include the following:

●Low risk for VTE (eg, the individual is ambulatory)

●Increased risk for bleeding, such as severe thrombocytopenia (generally considered as a platelet count <50,000/microL; can be more or less depending on specific patient features)

●Active bleeding or other contraindication

Close discussion among treating clinicians is appropriate for individualizing therapy.

Lower extremity compression devices are less likely to be effective in individuals with significant peripheral edema, although this issue has not been studied.

Additional discussions regarding the indications, contraindications, timing, and dosing of prophylactic anticoagulation, and alternatives (eg, mechanical interventions), are presented in detail separately. (See "Prevention of venous thromboembolic disease in acutely ill hospitalized medical adults" and "Prevention of venous thromboembolic disease in adult nonorthopedic surgical patients".)

VTE therapy — Therapy for VTE may be required in patients with liver disease. This may be complicated by concerns about risk of bleeding, difficulty monitoring the degree of anticoagulation in an individual with abnormal clotting times at baseline, and the lack of evidence to support a specific anticoagulant, dose, or duration of therapy. The majority of studies examining therapeutic anticoagulation have been in patients with cirrhosis and portal vein thrombosis. (See 'Portal vein thrombosis (PVT)' above.)

In the absence of such evidence, our general practice is to balance the risks and benefits of anticoagulation overall in the patient. This may entail screening endoscopy if this has not been performed already.

●For individuals with varices and advanced cirrhosis, overall bleeding risk is likely increased. However, some individuals have a pressing indication for anticoagulation (eg, large pulmonary embolism) and require prompt therapy; these individuals should have proper variceal screening and bleeding risk reduction prior to the initiation of therapy. In a study that evaluated risk of upper gastrointestinal bleeding in patients with cirrhosis on anticoagulation compared with a similar cohort of patients not on anticoagulation, outcomes were similar, including failure of endoscopic control of bleeding at five days, need for rescue therapy, and mortality at six weeks [93]. Decisions regarding the timing of therapy and type of anticoagulation should be made with multidisciplinary discussions.

●For individuals without significant varices or high-risk contraindications to anticoagulation who develop a VTE, we generally use anticoagulation with LMW heparin, warfarin, or, increasingly, a direct acting oral anticoagulant (DOAC; eg, dabigatran, apixaban, rivaroxaban, edoxaban) [145-147]. The choice of anticoagulant and duration of therapy is usually determined through collaboration of hepatologists and hematologists. Individual characteristics of the patient, including but not limited to degree of hepatic decompensation (Child-Pugh score), comorbidities, platelet count, presence of varices, frailty, and risk of falls all factor into the decision of the optimal therapy. (See 'Adjustments to anticoagulant dosing' above.)

Monitoring depends on the anticoagulant used and the patient's baseline testing [124]. If their baseline international normalized ratio (INR) is within the normal range, then warfarin can be used conventionally. However, the INR often is outside of the normal range and obscures appropriate target ranges; this situation warrants consideration of the use of other anticoagulants.

●For patients receiving therapeutic doses of a LMW heparin, anti-factor Xa activity can be measured; however, this is controversial [148]. (See "Heparin and LMW heparin: Dosing and adverse effects", section on 'Heparin resistance/antithrombin deficiency'.)

●For those receiving warfarin, if the baseline is normal, then target INR levels should range from 2.0 to 3.0. We have used warfarin in patients with baseline INR prolongation (eg, patients with prosthetic heart valves or atrial fibrillation). However, data are lacking on target INR, and more study is needed to better guide therapy with warfarin and DOACs in this setting.

The relative safety of therapeutic anticoagulation for DVT and PE in patients with decompensated cirrhosis has not been determined. However, based on studies in individuals with portal vein thrombosis, bleeding risks appear to be relatively low in highly selected populations [149-152]. (See 'PVT treatment' above.)

Of note, some individuals with liver disease may require adjustment of doses of heparin products due to mild antithrombin deficiency. However, monitoring heparin efficacy in cirrhosis is often challenging and use of aPTT and anti-factor X levels are controversial. (See "Heparin and LMW heparin: Dosing and adverse effects", section on 'Heparin resistance/antithrombin deficiency'.)

ACUTE LIVER FAILURE — Hemostatic profiles in patients with acute liver failure (ALF) do not behave in a similar manner to those with chronic liver diseases. Studies have shown intact thrombin generation capacity in the setting of acute liver failure [153]. Despite an elevated international normalized ratio (INR), thromboelastography has demonstrated normal hemostasis in this population [48]. One study examined the rate of bleeding with intracranial pressure monitoring and found that bleeding rates were similar in the group that received INR correction with significant amounts of Fresh Frozen Plasma (FFP) transfused as the group that did not [154]. In an analysis of a cohort of 1770 patients from the ALF Study Group registry, 187 (11 percent) had bleeding events [49]. While the majority of events were spontaneous and unrelated to procedures, 20 intracranial hemorrhages occurred (10 spontaneous and 10 related to intracranial pressure monitor insertion). This study emphasizes the high mortality associated with ALF.

Optimizing hemostasis is one of many aspects of the management of acute liver failure; additional testing and interventions are presented in separate topic reviews:

●Children – (See "Acute liver failure in children: Etiology and evaluation" and "Acute liver failure in children: Management, complications, and outcomes".)

●Adults – (See "Acute liver failure in adults: Etiology, clinical manifestations, and diagnosis" and "Acute liver failure in adults: Management and prognosis".)

●Liver transplantation – (See "Liver transplantation in adults: Patient selection and pretransplantation evaluation".)

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Cirrhosis" and "Society guideline links: Gastrointestinal bleeding in adults" and "Society guideline links: Nonalcoholic fatty liver disease" and "Society guideline links: Alcohol-associated liver disease".)

INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5th to 6th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10th to 12th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.

Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)

●Basics topics (see "Patient education: Cirrhosis (The Basics)")

●Beyond the Basics topics (see "Patient education: Cirrhosis (Beyond the Basics)")

SUMMARY AND RECOMMENDATIONS

●Clinical features – Liver disease leads to a form of "rebalanced" hemostasis, in which diminished hepatic function leads to both procoagulant and anticoagulant effects. Patients with severe liver disease and abnormalities of coagulation testing should not be assumed to be "auto-anticoagulated," because standard coagulation testing does not assess prothrombotic and fibrinolytic changes. (See 'Physiologic effects of hepatic dysfunction' above.)

●Laboratory assessment – Individuals with liver disease frequently have abnormalities in routine laboratory tests of coagulation. If accurate testing of hemostasis is required, it might be more helpful to use comprehensive viscoelastic testing such as thromboelastography (TEG) or thromboelastometry (ROTEM) (figure 1) to guide therapy; clinically, this testing is generally confined to perioperative transfusion guidance in patients undergoing liver transplantation. (See 'Laboratory abnormalities' above.)

●Bleeding – Management of bleeding depends on the location and severity as well as the degree of hemostatic impairment. (See 'Bleeding' above.)

•Nonvariceal – For nonvariceal bleeding, our general approach includes addressing treatable comorbidities and assessing the platelet count, fibrinogen levels, and a global assay of hemostasis such as TEG or ROTEM. We give vitamin K if the patient history suggests possible deficiency, transfuse platelets to maintain a platelet count >50,000 in active bleeding and administer a source of fibrinogen (preferably Cryoprecipitate, which creates a smaller volume load than Fresh Frozen Plasma [FFP] (table 2)) to maintain a fibrinogen level ≥100 to 120 mg/dL, and administer an antifibrinolytic agent (tranexamic acid or epsilon aminocaproic acid) if excessive fibrinolysis is suspected (figure 3). (See 'General approach to managing bleeding' above.)

•Variceal – For variceal bleeding, the major cause is increased portal pressure rather than a bleeding diathesis, and the most important interventions for prevention and treatment of bleeding involve reducing portal pressure and ligating bleeding lesions. Hemostatic mechanisms appear to play only a transient role; however, anticoagulation may contribute to bleeding risk. (See 'Variceal bleeding' above and "Overview of the management of patients with variceal bleeding" and "Methods to achieve hemostasis in patients with acute variceal hemorrhage".)

●Invasive procedures – For patients undergoing invasive procedures, we optimize comorbidities such as kidney function, reduce vascular engorgement of the collateral bed (by limiting volume expansion), and treat comorbidities such as infections. We monitor the platelet count and rely on global measures of clot formation if available (or fibrinogen and platelet count if not). We do not routinely transfuse plasma for an increased prothrombin time and international normalized ratio (PT/INR) prior to a procedure. (See 'General approach to invasive procedures' above.)

For individuals with severe thrombocytopenia (platelet count <50,000/microL) who require an invasive procedure in which bleeding risk is considered to be high and for whom there is sufficient time to document an increased platelet count (eg, 10 to 13 days), we suggest considering use of a thrombopoietin receptor agonist (TPO-RA) (Grade 2B). Avatrombopag is approved in the United States and lusutrombopag is approved in the United States and Japan for this indication. Data support the efficacy of TPO-RAs efficacy in raising platelet counts, but these agents have not been demonstrated to reduce bleeding risk when given before procedures. Therefore, clinicians may reasonably choose close observation for procedures with a plan for rescue therapy should bleeding occur. Advice for specific procedures is presented above. (See 'Non-surgical invasive procedures' above and 'Major surgery' above.)

For urgent or emergency procedures requiring an increase in platelet count, platelet transfusions are used. When platelets are administered, they should be given as closely as possible to the start of the procedure. (See "Platelet transfusion: Indications, ordering, and associated risks".)

●Portal vein thrombosis – The clinical significance of portal vein thrombosis (PVT) continues to be debated. Prevention of PVT in individuals with liver disease focuses on optimizing hepatic function and reducing portal venous pressure. Further study is needed to clarify the role of prophylactic anticoagulation in this setting (without evident thrombus), and we do not routinely use it outside of clinical research. Treatment of PVT is discussed in detail in separate topic reviews. (See 'Portal vein thrombosis (PVT)' above and "Acute portal vein thrombosis in adults: Clinical manifestations, diagnosis, and management" and "Chronic portal vein thrombosis in adults: Clinical manifestations, diagnosis, and management".)

●Other sites of thrombosis – There is a paucity of data regarding other sites of venous thromboembolism (VTE) (eg, deep vein thrombosis [DVT], pulmonary embolism [PE]) risks, prevention, and treatment in patients with severe liver disease. VTE prophylaxis is appropriate for the majority of hospitalized medical and surgical patients with liver disease, similar to individuals hospitalized with other acute conditions. Therapy for VTE may be complicated by concerns about bleeding, difficulty monitoring the degree of anticoagulation, and lack of evidence to support a specific anticoagulant, dose, or duration of therapy. Dosing of anticoagulants in liver disease is summarized in the table (table 1). We individualize anticoagulant therapy in collaboration between hepatologists and hematologists. (See 'Venous thromboembolism (VTE)' above and "Prevention of venous thromboembolic disease in acutely ill hospitalized medical adults" and "Prevention of venous thromboembolic disease in adult nonorthopedic surgical patients".)

ACKNOWLEDGMENTS — The UpToDate editorial staff acknowledges Patrick G Northup, MD, MHES, and Stephen H Caldwell, MD, who contributed to an earlier version of this topic review.