INTRODUCTION — The clotting process is a dynamic, highly interwoven array of multiple processes, which can be viewed as occurring in four phases.
- Initiation and formation of the platelet plug
- Propagation of the clotting process by the coagulation cascade
- Termination of clotting by antithrombotic control mechanisms
- Removal of the clot by fibrinolysis
Healthy individuals possess adequate amounts of clotting factors, regulatory proteins, and platelets to achieve optimal clot formation, clot limitation, and dissolution. Patients with liver disease, on the other hand, have a disturbed balance of pro-coagulant and anti-coagulant factors deviating from the normal coagulation cascade, with little in the way of “reserve” [1-4]. (See "Tests of the liver's biosynthetic capacity (eg, albumin, coagulation factors, prothrombin time)".)
Abnormalities of these coagulation pathways in patients with liver disease will be discussed here. An overview of coagulation and the clinical use of coagulation tests are presented separately. (See "Overview of hemostasis" and "Clinical use of coagulation tests".)
PROBLEM OVERVIEW — Various factors contribute to the abnormalities of coagulation seen in patients with liver disease:
- Increased bleeding risk — Decreased production of non-endothelial cell-derived coagulation factors (eg, factors II, V, VII, IX, X, XI, XIII) is only one component of the coagulation process that disrupts hemostasis. Thrombocytopenia, altered platelet function, platelet inhibition by nitric oxide, abnormalities of fibrinogen, and decreased thrombin activatable fibrinolysis inhibitor (TAFI) all contribute to an increased bleeding risk.
- Increased thrombotic risk — Decreased levels of the liver-synthesized natural anticoagulant proteins C and S, decreased antithrombin levels, decreased plasminogen, and elevated levels of endothelial cell-derived factor VIII and von Willebrand factor (vWF) favor thrombosis formation.
The varying degrees of disruption of these opposing pathways lead to different and potentially changing hemostatic activity for individual patients with cirrhosis. The relative balance or imbalance in these patients is not reflected in conventional indices of coagulation, such as the prothrombin time (PT), activated partial thromboplastin time (aPTT) or INR [3]. There is now convincing evidence disproving the common misconception that liver disease patients are “auto-anticoagulated” [5]. (See "Tests of the liver's biosynthetic capacity (eg, albumin, coagulation factors, prothrombin time)", section on 'International normalized ratio'.)
HYPOCOAGULABILITY IN CIRRHOSIS — As noted above, multiple factors can result in a hypocoagulable state in patients with liver disease. Decreased levels of all liver synthesized pro-coagulant factors, including the vitamin K dependent clotting factors (II, VII, IX, and X) in patients with cirrhosis are widely recognized. In addition, the liver modifies these vitamin K dependent factors for proper physiologic function, a function which may be deranged in patients with vitamin K deficiency as well as in those with hepatocellular carcinoma. (See "Tests of the liver's biosynthetic capacity (eg, albumin, coagulation factors, prothrombin time)", section on 'Des-gamma-carboxy prothrombin' and "Vitamin K, gamma carboxyglutamic acid, and the function of coagulation and other proteins", section on 'Plasma proteins of blood coagulation'.)
Other liver-synthesized clotting factors that may contribute to hypocoagulability include fibrinogen, factors V, XI, XII, prekallikrein, kininogen, and plasminogen [6]. Predicting the occurrence of bleeding based on factor levels or composites of factor activity such as the prothrombin time is difficult, due to the occurrence of counter forces, which favor thrombosis (see below) and the development of super-imposed conditions such as infection [7]. However, this is only a fraction of the disturbances present in the coagulation pathways in patients with liver disease, as will be discussed immediately below [6,8]. (See "Disorders of fibrinogen", section on 'Liver disease'.)
Abnormalities of platelet number and function — Thrombocytopenia and platelet-endothelial adhesion dysfunction affect clot formation and contribute to a relative hypocoagulable state, especially because platelet levels are pivotal in the initial coagulation response. The surface phospholipids of platelets provide the platform for factor complexes, amplification, and propagation of clot formation. (See "Overview of hemostasis", section on 'Multicomponent complexes'.)
Many theories exist regarding the genesis of thrombocytopenia in patients with liver disease. Decreased thrombopoietin (TPO) levels [9], splenic sequestration of platelets due to portal hypertension, auto-antibody destruction of platelets [10], and bone marrow suppression due to underlying liver disease can all contribute to thrombocytopenia. Low levels of TPO measured in patients with cirrhosis resolved after undergoing liver transplantation, reinforcing the role of TPO levels in liver disease thrombocytopenia [9]. (See "Congenital and acquired disorders of platelet function", section on 'Liver disease'.)
Platelet dysfunction is also a contributing aspect of decreased clot formation. Defective interactions between platelets and the endothelium result in sub-optimal clot formation that can be corrected by the addition of recombinant factor VIIa in an in vitro model [11]. In conditions mimicking intravascular flow, low hematocrit and low platelet counts contributed to decreased adhesion of platelets to endothelial cells, although increased vWF may offset this change in patients with cirrhosis [12]. In addition, patients with liver disease and concomitant renal insufficiency (such as different forms of hepatorenal syndrome) may have platelet dysfunction due to uremia and to changes in vessel wall endothelial function. This is a common and often overlooked aspect of impaired hemostasis in cirrhosis. Effective dialysis in this sense can be sometimes viewed as in effect a hemostatic intervention. (See "Congenital and acquired disorders of platelet function", section on 'Uremia'.)
Thrombin generation is also directly affected by the platelet count in patients with cirrhosis [13]. Consistent with the cell-based model of coagulation, studies of thrombin generation capacity in these patients have demonstrated the importance of platelets in potentiating the clotting cascade, leading ultimately to thrombin (factor IIa) generation, which then converts fibrinogen to fibrin. Platelet levels of around 50,000 to 60,000/microL or more were associated with adequate thrombin production, while levels closer to 100,000/microL were associated with optimal thrombin production. Accordingly, platelet transfusions in some cirrhotic patients with significant thrombocytopenia and clinical bleeding may augment thrombin generation in addition to correcting the low platelet count.
Infection and endogenous heparinoids — The overall incidence of infection in patients with liver disease has been estimated to be as high as 30 percent [14]. Overt sepsis or low levels of endotoxemia can affect platelet function, production, and adhesion in cirrhotics. Moreover, infection has been associated with increased detection of endogenous heparinoids, probably as a reflection of endothelial dysfunction and associated changes in the endothelial glycocalyx, possibly mediated by infection-related changes in nitric oxide metabolism [15].
Glycosaminoglycans (heparinoids) found in the vessel wall are bound by the endothelium and help to maintain hemostatic balance by preventing clot formation and facilitating blood flow within a vessel. Two specific glycosaminoglycans (GAGs), heparan sulphate and dermatan sulphate, have anticoagulant properties similar to heparin. Increased levels of these proteins, measured by thromboelastography, were found in patients with cirrhosis [16], and even higher concentrations were found in subjects with cirrhosis and active bleeding or infection. Further exploration of the effect of infection-related GAGs on coagulation in patients with liver disease is necessary and may guide focused pharmacologic therapies for bleeding.
Hyperfibrinolysis — Evidence of systemic fibrinolysis can be detected in 30 to 46 percent of patients with chronic liver disease and parallels the degree of liver dysfunction. However, clinically evident hyperfibrinolysis is less common and has been estimated to occur in 5 to 10 percent of those with decompensated cirrhosis [17-19].
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 bleeding following puncture wounds or dental extractions, or on occasion without any recognizable trauma [20]. However, the lack of readily available means to clearly identify this condition (such as via the use of thromboelastography) impedes the evaluation of hyperfibrinolysis in patients with cirrhosis. (See "Thrombotic and hemorrhagic disorders due to abnormal fibrinolysis".)
Hyperfibrinolysis promotes premature clot dissolution and interferes with clot formation due to the consumption of clotting factors and decreased platelet aggregation due to the degradation of vWF and fibrinogen platelet receptors [18]. The mechanisms involved hinge on dysregulation of a complex set of interacting factors including tPA (tissue plasminogen activator) and PAI-1 (plasminogen activator inhibitor 1), neither of which is liver synthesized, although both have altered clearance in cirrhosis, and TAFI (thrombin-activatable fibrinolysis inhibitor), which is liver synthesized [21].
Hyperfibrinolysis may play a role in several discrete clinical situations:
- It has been implicated as an exacerbating factor in variceal rupture by disrupting the hemostatic plug that forms with primary hemostasis.
- High fibrinolytic activity in saliva likely potentiates bleeding following dental extractions in patients with cirrhosis and fibrinolysis in other normal body cavities, such as the biliary tree, and urinary bladder could potentiate bleeding from these sites in the setting of cirrhosis.
- Ascites fluid fibrinolytic activity has been proposed as an initiator of systemic hyperfibrinolysis in cirrhosis due to increased thoracic duct delivery of ascites fluid to the systemic circulation [22].
- During liver transplantation, high levels of tPA in the recipient may result in hyperfibrinolysis [23].
HYPERCOAGULABILITY IN CIRRHOSIS — Hypercoagulability is an increasingly recognized aspect of liver disease. The lack of proper measurement tools to identify those patients who are prone to develop clots, and reliance on clinical endpoints (eg, deep vein thrombosis, portal vein thrombosis) likely lead to an under-estimation of this problem [5,24]. Based upon the National Hospital Discharge Survey, the prevalence of VTE was 0.6 and 0.9 percent for those with chronic alcoholic liver disease and chronic nonalcoholic liver disease, respectively [25].
- In one study, hospitalized patients with liver disease were found to have a 0.5 percent incidence of deep venous thromboembolism despite a “coagulopathic profile” by conventional parameters in many of the patients [24].
- In a second study performed in 73 patients with cirrhosis, in vitro assays indicated significantly increased thrombin generation in subjects with cirrhosis when compared with 38 normal controls [26].
As a result of such observations and improved understanding of the pathophysiology of clot formation in liver disease, there is increasing awareness that the old dogma of “auto-anticoagulation” represented by an elevated INR in patients with cirrhosis is unfounded.
The reduction of procoagulant factors in cirrhosis, reflected in a prolonged PT and INR, is offset by decreased levels of anti-coagulant factors (ie, protein C, protein S, anti-thrombin) of potentially equal or greater magnitude, but not measured by conventional parameters such as the PT, INR, or aPTT [27].
Several endothelium-derived procoagulant factors are increased in cirrhosis, including factor VIII and von Willebrand Factor (vWF). The former, an acute phase reactant, is known to be prothrombotic, while increased levels of vWF favor platelet adhesion. Changes in the latter result from compensatory increased production, decreased levels of a vWF cleaving protein in cirrhotics (ADAMTS13), and an increased presence of vWF on endothelial walls due to a chronic inflammatory state [28]. Together, these variables contribute to a new and somewhat precariously rebalanced hemostatic system in cirrhosis, which may tilt in either direction [29]. A major clinical challenge is to determine how best to measure this as a guide to optimal management [4]. (See "Causes of thrombotic thrombocytopenic purpura-hemolytic uremic syndrome in adults", section on 'ADAMTS13 activity in other conditions'.)
Hypercoagulability in cirrhosis can lead to macro- and micro-thrombi production, resulting in various complications.
- Macro-thrombotic complications include portal vein thrombosis, deep vein thrombosis, and pulmonary embolism. These clinical outcomes present complex management dilemmas in patients with cirrhosis. Traditional warfarin-based anticoagulation is titrated to measurements of the INR. However, in patients with liver disease the varying INR values lead to unclear targets. (See "Tests of the liver's biosynthetic capacity (eg, albumin, coagulation factors, prothrombin time)", section on 'International normalized ratio'.) At this time, no clear evidence-based recommendations can be made for managing DVT in patients with cirrhosis and therapy must be individualized. (See 'Venous thromboembolism prophylaxis in hospitalized patients with cirrhosis' below.)
- Potential micro-thrombotic complications of cirrhosis are subtle and typically run a chronic course. Intra-hepatic microthrombi can cause localized ischemia, which leads to scarring and accelerated development of cirrhosis in a process known as parenchymal extinction [30]. This is thought to be one of the mechanisms leading to liver atrophy, which contributes to the transition to severely decompensated cirrhosis. Platelet aggregation and activation in these microscopic lesions may also contribute to fibrosis and possibly to thrombocytopenia. A similar process in the lung microvasculature appears to play a role in a serious complication of cirrhosis known as portopulmonary hypertension. The resultant remodeling of the pulmonary vasculature can have important ramifications in transplant eligibility, outcomes and survival [31]. Whether or not anticoagulation therapy prevents development of these micro-thrombotic conditions remains to be fully evaluated. (See "Portopulmonary hypertension".)
- Non-alcoholic fatty liver disease (NAFLD), which is associated with obesity, insulin resistance, diabetes, and dyslipidemia, presents a unique and problematic intersection of chronic liver disease and thrombotic disorders, including coronary vascular disease. Reduced sensitivity to insulin, increased fatty acid levels forming thromboxane A2, and higher low-density-lipoprotein (LDL) levels all contribute to increased platelet aggregation. Increased plasminogen activator inhibitor-1 (PAI-1) also contributes to hypercoagulability in the metabolic syndrome which is almost a universal finding in patients with NAFLD. These problems complicate management of patients with co-existing nonalcoholic steatohepatitis (NASH)-related cirrhosis and coronary disease, especially when coronary re-vascularization is undertaken in the setting of increased risk of portal hypertensive bleeding. Such patients are among the most clinically challenging [32].
Portal vein thrombosis — Broadly speaking, portal vein thrombosis (PVT) can be described in terms of patients with or without cirrhosis, the latter including idiopathic PVT and PVT related to thrombophilic factors. We will restrict our comments to PVT in the setting of cirrhosis in this section. The reader is directed to other sections of UpToDate for discussion of non-cirrhotic PVT as well as the Budd-Chiari syndrome. (See "Extrahepatic portal vein obstruction (portal vein thrombosis)" and "Etiology of the Budd-Chiari syndrome".)
PVT affects about 10 to 20 percent of all patients with cirrhosis, but the prevalence varies with disease severity, being much lower in Child-Pugh A patients and increasing step-wise in Child B and C patients [33]. Contributing factors include stasis due to portal hypertension, development of hepatocellular cancer (HCC), and sometimes to genetic predisposition to hypercoagulability. To what extent this development reflects relative hypercoagulability due to changes in hemostatic pathways discussed above is not known [34].
Focal left or right branch PVT is relatively more common and often clinically silent, although its development may contribute to overall organ atrophy. Extrahepatic portal vein thrombosis (EPVT) is relatively less common but has increased risk of decompensation due to variceal bleeding or portosystemic shunting. The development of any of these forms warrants a careful examination for hepatocellular carcinoma, while EPVT especially should prompt consideration of a more exhaustive search for additional factors, such as prothrombin gene mutation, factor V Leiden, or a myeloproliferative neoplasm (eg, polycythemia vera), especially if there is mesenteric extension. (See "Etiology of the Budd-Chiari syndrome", section on 'Myeloproliferative disorders' and "Etiology of the Budd-Chiari syndrome", section on 'Other hypercoagulable states'.)
Treatment — Anticoagulation to treat PVT may be considered in limited cases, especially with extensive clot burden or the presence of secondary factors such as the prothrombin gene mutation. However, the risk benefit is uncertain due to the risk of portal hypertensive bleeding. Anticoagulation has been reported to result in recanalization in up to 40 percent and to reduce post-operative complications in patients awaiting liver transplantation, especially those with extensive clot, although this remains a debated approach [35]. (See "Extrahepatic portal vein obstruction (portal vein thrombosis)", section on 'Management'.)
Prevention — Preliminary results are available from a prospective, randomized trial of the safety and efficacy of the low molecular weight heparin enoxaparin (4000 units/day versus placebo for 12 months) for the prevention of PVT in cirrhosis [36]. During the one-year study period, PVT occurred significantly more often in those taking placebo (6 of 36, 17 percent) than in those taking enoxaparin (zero of 34). No relevant side effects, in particular no hemorrhagic events, were attributed to the use of enoxaparin. Enoxaparin treatment was also associated with a lower number of patients developing hepatic decompensation (53 versus 12 percent), suggesting its use for that indication, as well. The final report of this study is eagerly awaited.
COAGULOPATHY IN ACUTE LIVER FAILURE — Coagulopathy is by definition a key component of the diagnostic criteria of acute liver failure (ALF), which is defined as the development of clinically evident hepatic encephalopathy and laboratory evidence of coagulopathy (usually prolongation of the PT and/or INR) within 24 weeks of the new onset of acute liver disease with no history of prior liver abnormalities.
Prolongation of the PT and INR in this setting is due to impaired synthesis of key factors measured by the prothrombin time, including especially factors VII and V, which have the shortest half-lives of liver-synthesized procoagulant factors [37]. The biological half-life of factor V is as long as 36 hours but that of factor VII is only 4 to 6 hours (table 1). Thus, significant and progressive hepatic necrosis will lead to relatively rapid depletion of these factors and prolongation of the PT and the INR.
Whether or not prolongation of the PT and INR in the setting of ALF reflects an increased bleeding risk has not been nearly as well studied as the coagulopathy of cirrhosis. One study followed 51 patients with acute liver injury/failure who underwent global hemostatic testing via thromboelastography (TEG). Despite a mean INR of 3.4, most patients had preserved hemostasis by TEG [38]. It was concluded that hemostasis was maintained via a combination of increased clot strength with increasing severity of liver injury, increased factor VIII levels, and a commensurate decline in both pro- and anticoagulant proteins.
When present, bleeding is usually limited to capillary and mucosal bleeding [39]. Although this may be relatively minor, the need for multiple invasive procedures increases the concern regarding iatrogenic bleeding.
Because of prolongation of the PT and INR, patients with ALF may be treated with fresh frozen plasma (FFP), especially prior to invasive procedures. However, the utility of this intervention has not been clearly demonstrated. Moreover, high INR values may require a significant volume of FFP (much more than the typical doses of two to four units) to correct the laboratory value. This is likely detrimental because large volumes of plasma may worsen body edema and intracranial hypertension due to the remarkably large volumes of plasma needed to correct the INR. (See "Clinical use of plasma components", section on 'Fresh frozen plasma'.)
In part because of this, recombinant activated factor VII (rFVIIa) has become a conventional means of correcting the INR especially prior to placement of an intracranial pressure monitor [40]. With rFVIIa, INR can be seen as reflecting an appropriate rFVIIa dose response, although whether or not this reflects a decreased risk of bleeding in ALF patients remains to be established. As with cirrhosis, very little scrutiny of existing literature is needed to see that this area requires further clinical research to determine which of the various available interventions will improve patient care [41]. (See "Endoscopic procedures in patients with disorders of hemostasis", section on 'Liver failure' and "Therapeutic uses of recombinant coagulation factor VIIa", section on 'Coagulopathy of liver dysfunction'.)
TESTS OF COAGULATION IN LIVER DISEASE
Prothrombin time (PT) and the International Normalized Ratio (INR) — The prothrombin time (PT) is used to assess the extrinsic pathway of clotting, which consists of tissue factor and factor VII, and coagulation factors in the common pathway (prothrombin (factor II), factors V and X, and fibrinogen). The sensitivity of the PT to reduced activity of the vitamin K-dependent factors within this pathway comprises the rationale for the use of the PT to monitor warfarin therapy. (See "Clinical use of coagulation tests", section on 'Prothrombin time'.)
In order to promote standardization of the PT for monitoring oral anticoagulant therapy, the World Health Organization (WHO) developed an international reference thromboplastin, currently recombinant tissue factor, and recommended that the PT ratio be expressed as the International Normalized Ratio or INR. This allows values of the PT from various locations to be directly compared, as may happen when a patient taking warfarin has blood sampled at different laboratories. Because this test was developed to assess dysfunction in vitamin K dependent coagulation factors II, VII, IX, and X during warfarin therapy, rather than in patients with cirrhosis, its extrapolation to the liver disease population is under increasing critical scrutiny [42]. (See "Tests of the liver's biosynthetic capacity (eg, albumin, coagulation factors, prothrombin time)", section on 'International normalized ratio'.)
INR variation in liver disease — A major shortcoming of utilizing the INR in patients with liver disease is the wide inter-laboratory variation, which depends at least in part on the type of thromboplastin used in the test [43]. This degree of variation can have substantial clinical implications in assessing bleeding risk, and in calculating the patient’s MELD score (model of end stage liver disease) [44]. The MELD score is calculated from three components, total bilirubin, creatinine, and the INR, and since 2002 it has become the standard score used to list and prioritize patients for liver transplant [45].
The reagent-dependent INR can lead to drastically different MELD scores [46-48]. As a result of these issues, there have been attempts to develop an alternative form of the INR based on a standard of patients with liver disease, the “INR liver”, as well as calls to dispense with the INR in patients with cirrhosis and use instead the prothrombin time as a measure of percentage of normal activity [43,49]. A more practical use of this measurement is still being investigated [50]. (See "Model for End-stage Liver Disease (MELD)".)
Platelet level and function — The platelet count, which is often included in a complete blood count (CBC), is the quantitative measurement of circulating platelets. It is often depressed in cirrhosis due to the combination of hypersplenism (figure 1), bone marrow depression, and changes in thrombopoietin metabolism. (See "Extrinsic nonimmune hemolytic anemia due to mechanical damage: Fragmentation hemolysis and hypersplenism", section on 'Extravascular nonimmune hemolysis due to hypersplenism'.)
Along with the INR, platelet levels have been a long standing convention in assessing cirrhosis-related bleeding risk. However, in contrast to empiric cut-offs for the INR, more reliable in vitro data exist to support specific levels of circulating platelets as more effectively promoting hemostasis in cirrhosis. Using an assay that resembles in vivo thrombin (factor II) generation, adequate thrombin production was demonstrated with platelet counts of around 56,000/microL and optimal thrombin production at levels closer to 100,000/microL. These findings support the relevance of traditional targets for low-moderate risk procedures and high-risk procedures.
However, in cirrhosis, platelet function remains an inadequately addressed aspect of platelet activity in the context of the cell-based mechanism of hemostasis (see above). Different methods of measuring platelet aggregation are available. Commercial tests designed to mimic in vivo hemostasis include the Clot Signature Analyzer, the Thrombotic Status Analyzer, and the Platelet Function Analyzer. All of these modalities measure platelet plug formation within capillary tubes designed to mimic vessels. (See "Platelet function testing".)
However, none of these tests of platelet function has undergone adequate translational testing to understand the relevance to bleeding in patients with cirrhosis. Such studies are essential to determine this aspect of platelet function in cirrhosis. Increased circulating vWF and systemic changes in cell membrane phospholipid composition upon which the essential elements of the clotting cascade occur are necessary to know whether or not aggregation and activity are impaired or possibly enhanced in cirrhosis. Renal failure (hepatorenal syndrome) and uremic platelet dysfunction is another important but understudied aspect of reduced platelet function in cirrhosis. (See "Congenital and acquired disorders of platelet function", section on 'Acquired platelet functional disorders'.)
Bleeding time in cirrhosis — The bleeding time is an indirect measure of platelet function. It is variably reported as being prolonged in cirrhosis. The variable results in cirrhosis may reflect counteracting forces, including not only platelet number and activity but also factors such as vascular smooth muscle dysfunction as seen in the “hyperdynamic state” (diffuse vasodilation) in people with portal hypertension and secondary portosystemic shunting. Test results and the range of normal values are very user dependent, which further lessens its predictive value. (See "Platelet function testing", section on 'The in vivo bleeding time'.)
Effective shortening of the bleeding time with DDAVP (desmopressin) has been demonstrated in cirrhosis (see below). A related measure called the “liver bleeding time” has also been reported in a number of studies that examined the relationship between liver surface bleeding and peripheral indices of hemostasis following laparoscopic biopsy and in studies of procoagulant factor administration.
Fibrinogen and individual factor levels — Measurement of individual factor levels is sometimes useful clinically.
- Factor VIII levels may help to distinguish superimposed disseminated intravascular coagulation (DIC) from liver failure. The former is characterized by severely decreased factor VIII levels, while the latter is associated with significantly increased levels.
- Factor V and VII levels are used prognostically in acute liver failure and may help to distinguish vitamin K deficiency from liver failure. Proportional reduction in both factors suggests liver failure, whereas a greater reduction in factor VII than in factor V favors vitamin K deficiency.
- Fibrinogen levels can help guide use of fibrinogen-rich cryoprecipitate, as levels below 120 mg/dL are associated with diminished clot formation and possibly to resistance to procoagulants such as recombinant factor VIIa.
- The ratio of factor VIII to protein C may help to identify patients with well preserved thrombin production [29]. Individual values of protein C and S as well as antithrombin, sometimes measured in the setting of deep vein thrombosis or portal vein thrombosis, can be difficult to interpret in cirrhosis and requires consideration of relative declines in each of the factors.
Fibrin degradation products and D-dimer — Increased fibrinolytic activity, decreased coagulation factors and decreased platelets can be a component of cirrhosis or coexisting disseminated intravascular coagulation (DIC). The overlap in laboratory abnormalities for cirrhosis and DIC makes these entities sometimes difficult to differentiate. In patients with stable liver disease, levels of thrombin generation, fibrin degradation products (FDP), D-dimer levels, and prothrombin fragments were found to be similar to normal controls, resulting in less overlap with the findings in DIC [51]. (See "Clinical features, diagnosis, and treatment of disseminated intravascular coagulation in adults", section on 'Diagnosis'.)
However, in decompensated liver disease, many of these values may mirror those found in DIC. While DIC and liver disease both cause elevations in D-dimer levels, low levels of factor VIII may suggest a predominance of DIC. High D-dimer levels, low factor VIII levels, and end organ micro-thrombotic disease should significantly raise the suspicion of DIC.
Fibrinogen abnormalities — Dysfunction of the fibrinogen molecule can cause an inherited or acquired coagulation disorder. In cirrhosis, acquired dysfibrinogenemia results in a prolonged thrombin time in the setting of normal fibrinogen levels [52]. (See "Clinical use of coagulation tests", section on 'Thrombin time'.)
The dysfibrinogen produced in this setting resembles fetal fibrinogen and presumably reflects the liver’s response to injury and the formation of regenerating nodules in the cirrhotic liver. The laboratory diagnosis of dysfibrinogenemia involves tests that are not widely available (eg, fibrin clot formation, thrombin time, reptilase time, and fibrinogen clotting activity-antigen ratio). The significance of dysfibrinogenemia in liver disease has not been fully established [53]. (See "Disorders of fibrinogen", section on 'Liver disease'.)
Thrombin generation testing — One of the many problems associated with the use of standard coagulation tests (ie, PT, aPTT) to evaluate hemostasis in general, including in patients with liver disease, is that they only measure part of the process of thrombin generation. The conventional clotting times typically measure the time it takes from the initiation of clotting to the initial generation of trace amounts of thrombin, ie, the lag time in the thrombin generation curve. It does not take into account the rate of the subsequent burst of thrombin generation, the peak thrombin concentration achieved, the total amount of thrombin formed (the endogenous thrombin potential), and the eventual inhibition of thrombin by the natural anticoagulants. Thus, a number of key factors are not taken into account, such as the following [3] (see "Overview of hemostasis", section on 'Thrombin generation' and "Overview of hemostasis", section on 'Activated protein C and protein S'):
- Platelets, which support thrombin generation by assembling activated coagulation factors on their surface
- Thrombomodulin, a protein situated on vascular endothelial cells, which is the main physiologic activator of protein C, a strong inhibitor of thrombin generation
Accordingly, the addition of platelets and/or thrombomodulin to the reaction mixture in thrombin generation tests has been able to indicate the important function of these other factors in contributing to the coagulation imbalance noted in patients with liver disease [3,4,13,26,29]. (See 'Platelet level and function' above.)
Thromboelastography and thromboelastometry — Originally developed over 50 years ago, there has been renewed interest in these tests as functional measures of clot formation and stability in whole blood samples from cirrhosis patients [54]. Although point of care testing may offer the greatest advantage, citrated samples with subsequent recalcification allow a delay in assay performance for one to two hours after collection.
- Thromboelastography (TEG) involves the measurement of torque on a pin as a cup containing whole blood is rotated at a constant rate. As clot formation begins, the torque increases, and then with clot breakdown from fibrinolysis, the torque decreases (figure 2).
- Thromboelastometry (ROTEM) is similar to the TEG but involves a stationary cup and rotating pin.
Both assays detect dynamic aspects of clot formation and lysis. Parameters include initiation time, the rate of rise in clot strength, maximal clot strength, and the rate of clot decline as lysis ensues. An overall measurement calculated from these variables gives a coagulation index and a measure of clot stability. (See "Coagulopathy associated with trauma", section on 'Thromboelastography' and "Platelet function testing", section on 'Instruments measuring physical properties of the clot'.)
In one study, the use of clot stability as measured by the TEG with conventional measures of platelets and INR correlated to blood product use in patients with cirrhosis better than conventional measures alone [55].
Sonorheometry — A novel method of using ultrasound to measure whole blood coagulation is currently being investigated. While TEG employs changes in torque to measure clot formation, sonorheometry utilizes acoustic radiation forces to assess hemostasis. This technique is currently under preliminary investigation, and has not been applied to patients with liver disease [56].
THERAPEUTIC AGENTS FOR BLEEDING — The proper use of blood products and their efficacy have not been established in patients with liver disease [57]. It is still relatively controversial due to the increased risk of transfusion related injury, unwanted volume expansion and associated increases in portal pressure [58-60], unnecessary cost of blood typing and administration of the infusion, in addition to the cost of the blood products themselves. Nevertheless, with very few alternatives, the focused use of blood products in the proper clinical scenario is one of the only options currently available for the patient with liver disease and bleeding.
Vitamin K — Malnutrition in patients with liver disease or cholestatic disease can lead to Vitamin K deficiency, although this can be difficult to diagnose in the setting of cirrhosis-related alterations in coagulation. Vitamin K deficiency is more likely in primary cholestatic diseases such as primary biliary cirrhosis and primary sclerosing cholangitis. The relative decline in factor V versus factor VII is sometimes helpful in questionable cases, since a proportionally greater decline in factor VII than in factor V suggests vitamin K deficiency. If suspected, repletion of vitamin K (dose range: 1 to 25 mg) can be accomplished by oral, subcutaneous, or intravenous administration. Because of edema and possible gut malabsorption in severely jaundiced patients, the intravenous route is sometimes preferred, although this can be associated with anaphylactic reactions. Overcorrection, which may be clinically important in patients being treated with a vitamin K antagonist for thrombotic illnesses, is generally not an issue, but balancing procoagulant activity with thrombotic risk in bedbound patients is challenging. (See "Overview of vitamin K".)
Fresh frozen plasma — Although fresh frozen plasma (FFP) is often used to treat coagulopathy in patients with liver disease, its benefit is still unclear. As an example, when 80 such patients were studied, conventional doses of FFP reduced their prothrombin time (PT) to within 3 seconds of normal in only 10 to 12 percent of recipients [61]. Moreover the amount of plasma needed to more consistently reduce the PT (10 to 15 mL/kg is the recommended full dose) clearly exposes patients to an increased risk of volume overload (leading to increased portal pressures) [2,58-60], as well as transfusion-related injury lung injury (TRALI) [62]. Furthermore, as mentioned above, it is unclear if the target PT or INR is a true reflection of bleeding risk in those with cirrhosis. Thus, because of uncertainty on the risk-benefit ratio in cirrhosis, these products should be used with caution. (See "Clinical use of plasma components", section on 'Fresh frozen plasma'.)
Cryoprecipitate — Cryoprecipitate contains the cold-insoluble residue that remains when FFP is thawed to a temperature of 4°C. It contains fibronectin, fibrinogen, and von-Willebrand factor (vWF) in a volume-reduced solution. Although it is understudied, there is a general consensus that in the setting of cirrhotic bleeding, fibrinogen levels should be measured and repleted if low. As a rationale, proponents indicate fibrinogen’s role as the last step toward fibrin clot formation in the clotting cascade.
The optimal fibrinogen level is uncertain, but a level of at least of 120 mg/dL has been suggested. Surprisingly, many patients maintain a fibrinogen level higher than this even with advanced disease, although the quality of the fibrinogen (ie, the presence or absence of dysfibrinogenemia) is often unmeasured. (See 'Fibrinogen abnormalities' above.) Recommended dosing is one bag of cryoprecipitate per 10 kg of body weight. (See "Clinical use of plasma components", section on 'Cryoprecipitate'.)
Platelet transfusion — Generally, transfusions of platelets result in an immediate increase in the platelet counts, maximal at about one hour after completion of the transfusion. In stable non-refractory patients, a six pack of pooled platelets or one unit of apheresis platelets should raise the platelet levels by about 30,000/microL, although this is highly variable. (See "Clinical and laboratory aspects of platelet transfusion therapy", section on 'Expected and observed increase in platelet counts and use of the CCI'.)
Patients with cirrhosis usually demonstrate a blunted response due to splenic sequestration, and the optimal target for platelet therapy in cirrhosis has not been agreed upon [1]. In vitro, studies have shown adequate thrombin production with platelet levels from 50,000/microL and optimal levels at platelet levels closer to 100,000/microL [13]. From these data, it is reasonable to aim for levels above 50,000/microL for moderate risk procedures (eg, liver biopsy [63]) and above 100,000/microL for high risk procedures [64].
Recombinant factor VIIa — Many studies have examined the use of recombinant human factor VIIa (rFVIIa) in liver disease patients, but no conclusive guidelines have been established. rFVIIa rapidly normalizes elevated prothrombin times in cirrhotic patients, but the efficacy in reducing bleeding risk remains unclear [65]. Depending on the dose, the effect on INR is evident for at least one to two hours. The most common use of rFVIIa in patients with liver disease is in the setting of intracranial pressure (ICP) monitor placement in acute liver failure, where a dose of 40 mcg/kg is usually employed [66]. (See "Therapeutic uses of recombinant coagulation factor VIIa", section on 'Coagulopathy of liver dysfunction' and "Acute liver failure: Prognosis and management", section on 'Coagulopathy'.)
The advantages of rFVIIa include low volume, relative safety, and rapid correction of the PT/INR. The major disadvantage is overall cost of this product and the possibility of inducing a hypercoagulable state. Since rFVIIa does not have a clearly defined role, it should be reserved for rescue therapy in active hemorrhage or as prophylaxis in very high risk invasive procedures such as ICP monitor placement.
Anti-fibrinolytics — Anti-fibrinolytics prevent clot lysis and are most useful in the setting of hyperfibrinolysis [67]. The hyperfibrinolytic states are difficult to diagnose with conventional testing, and should be considered especially in the setting of delayed bleeding following a procedure or intractable oozing from a puncture wound. Bodily fluids such as ascites or saliva have increased fibrinolytic properties. Therefore in the setting of patients with cirrhosis and bleeding from peritoneal sources or dental extractions, anti-fibrinolytics should be considered.
Currently, there are no randomized clinical trials showing any benefit of anti-fibrinolytics in patients with cirrhosis and upper gastrointestinal bleeding [68]. However, in certain settings, especially when hyperfibrinolysis is proven or suspected, these agents (eg, aminocaproic acid, tranexamic acid) may be effective and appear safe in cirrhosis [18,20,69,70]. Dosing of anti-fibrinolytic agents varies among various studies, and most populations consisted of patients without liver disease.
- Aminocaproic acid loading dose ranges from 1 to 15 g (usually given intravenously at a dose of 4 to 5 g for the first hour followed by 1 g/hr for up to 8 hours) and a maximal total dose of 30 g. For dental procedures, aminocaproic acid can be given as a 10 mL (4 g) oral rinse solution that is held in the mouth for 2 minutes and then spit out, repeated every 6 hours for up to 2 days or added to a gauze pad as a local compress.
- Intravenous tranexamic is administered with a 10 mg/kg loading dose and repeated 3 to 4/day for a total of 2 to 8 days [71].
Prothrombin complex concentrates — Prothrombin complex concentrates (PCCs) contain human-derived concentrated vitamin K-dependent factors usually including factors II, IX, and X and variable amounts of factor VII in their inactive forms along with (in some preparations) the natural anticoagulant factors such as protein C and protein S. These agents are often used to reverse warfarin-associated bleeding, such as warfarin-associated intracerebral hemorrhage. (See "Management of warfarin-associated intracerebral hemorrhage", section on 'Unactivated prothrombin-complex concentrates' and "Correcting excess anticoagulation after warfarin", section on 'Significant or life-threatening bleeding'.)
However, very little data exist regarding the efficacy of PCCs in cirrhosis-related bleeding, although the theoretical advantage of a markedly lower infused volume have stimulated interest as substitutes for fresh frozen plasma in active bleeding. Studies are currently planned to study their role in liver disease patients. (See "Plasma derivatives and recombinant DNA-produced coagulation factors", section on 'Prothrombin complex concentrates'.)
Desmopressin — As an analogue of the antidiuretic hormone vasopressin, desmopressin (dDAVP) increases levels of factor VIII and vWF, although the exact mechanism for its potential use in cirrhosis is uncertain, as both of these factors are usually elevated. Nonetheless, intravenous doses of 0.3 mcg/kg in patients with cirrhosis improve bleeding times with little demonstrable effect on systemic hemodynamics [72,73]. The use of intranasal dDAVP (300 mcg) was compared with blood product transfusions in patients with cirrhosis undergoing dental extractions. DDAVP had comparable results to infusion of fresh frozen plasma or platelets, but was felt to be more convenient, better tolerated, and less expensive compared with the use of blood products [74]. Side effects are usually mild, but may include increases in blood pressure, facial flushing, headache, and hyponatremia. Further studies are needed to clarify the exact role of dDAVP for prophylaxis in patients with liver disease.
Blood product usage — There is a lack of evidence-based guidelines regarding optimal use of blood products in patients with liver disease. Based on understanding of the pathophysiology of coagulation in cirrhosis and the limitations of conventional risk indices [42], it is likely that blood products are overused in this population, with resultant unnecessary risk exposure.
A single center study showed that on three separate days, patients with liver disease consistently utilized one-third of the total fresh frozen plasma infused [75]. The disproportionate use of blood products in this group reinforces the need for further clinical studies to help develop evidence-based guidelines for appropriate resource utilization of blood products. Non-evidence based societal guidelines regarding specific cut-offs for conventional tests such as INR should be discouraged.
Renal support — Renal support therapy (ie, dialysis) has a complex relationship to hemostatic pathways in cirrhosis. There is little doubt that effective dialysis can potentiate platelet function and improve volume status in severe renal failure complicating decompensated cirrhosis, although definitive data such as pre- and post-dialysis thromboelastography are lacking. However, the mechanical aspects of renal replacement therapy can also lead to platelet activation through turbulence and shear [64]. The formation of thrombin and glycoprotein receptor activation can initiate the coagulation cascade [31]. Subsequent platelet aggregation may dictate the need for anticoagulation during dialysis. Conversely, platelet destruction during the contact time with bioartificial membranes and increased turnover when in contact with dialysis tubing can lead to worsening of the patient’s thrombocytopenia.
VENOUS THROMBOEMBOLISM PROPHYLAXIS IN HOSPITALIZED PATIENTS WITH CIRRHOSIS — Little is known about peripheral (non-portal) venous thromboembolic events (VTE) in patients with cirrhosis and there have been no randomized controlled trials for the treatment or prophylaxis of deep vein thrombosis (DVT) or pulmonary embolism (PE) in patients with cirrhosis. In fact, all patients with underlying coagulopathy were excluded in the landmark clinical trials studying VTE so there is an extreme paucity of data from which to draw conclusions.
As stated above, there are a growing number of single center [24,76,77] and population-based [78] studies showing a prevalence of DVT or PE in hospitalized patients with cirrhosis between 0.5 to 6.3 percent, which is in the range commonly expected in the general internal medicine population. (See 'Hypercoagulability in cirrhosis' above and "Overview of the causes of venous thrombosis", section on 'Acquired thrombophilia'.)
It is clear that patients with cirrhosis are exposed to many risk factors for VTE including cancer, older age, surgical procedures, prolonged hospitalization, and inactivity. Improvements in the treatment of liver disease over the recent decades have led to an increased life expectancy in these patients and thus led to an increased exposure to these risk factors. There is also increasing awareness that the coagulopathy of liver disease is not protective from VTE in patients with cirrhosis.
The most obvious fear regarding VTE prophylaxis or treatment in patients with cirrhosis is hemorrhage, especially variceal bleeding from the associated portal hypertension. This is a reasonable concern in the patient admitted with clinical bleeding or with high risk esophageal varices but less clear in the patient with no or low risk varices. Based on the general internal medicine literature and practice, the current clinical practice in many hospitals with experience in treating patients with cirrhosis is to use VTE prophylaxis in those with minimal varices and no evidence of clinical bleeding, especially those with high risk VTE conditions.
A study reported in abstract form compared GI bleeding and mortality rates in 235 cirrhotic patients receiving VTE prophylaxis and 141 cirrhotic untreated controls. GI bleeding rates were similar between the treated and control groups (2.54 versus 4.96 percent, respectively), with no significant difference in hospital death rates (3.94 versus 10.64 percent, respectively) [79].
The use of anti-Xa levels as a means of monitoring prophylaxis with low molecular weight (LMW) heparin and similar agents is also under study in some centers. (See "Therapeutic use of heparin and low molecular weight heparin", section on 'Anti-factor Xa testing'.)
Preliminary data suggest that low levels of liver-derived antithrombin (AT, previously called antithrombin III) may cause these patients to be resistant to LMW heparin. Lower extremity compression devices have not been studied in patients with cirrhosis and are likely to be of little benefit in the patient with massive peripheral edema. Further clinical studies are needed in this area. (See "Therapeutic use of heparin and low molecular weight heparin", section on 'Heparin resistance'.)
Anti-platelet and anticoagulant therapy — Therapeutic anticoagulation is occasionally needed in patients with cirrhosis and coagulopathy although the management of therapy with warfarin can be difficult because of the baseline coagulopathy as measured by the INR. The best indicator of therapeutic anticoagulation levels in patients with cirrhosis is probably measurement of anti-Factor Xa activity following the use of heparin products. However, as noted above, low levels of antithrombin may cause resistance to heparin-like agents.
While the safety of therapeutic anticoagulation in patients with decompensated cirrhosis has not been determined, some conclusions can be drawn from sparse data related to anticoagulation in those with acute and chronic portal vein thrombosis.
- One study in non-cirrhotic portal hypertension (including esophageal varices) showed good recanalization rates and no excess bleeding events [80].
- A study in 39 patients with cirrhosis and chronic portal vein thrombosis including 50 percent who presented with variceal bleeding showed a 75 percent complete restoration of portal venous flow using therapeutic enoxaparin after complete eradication of varices by band ligation and no significant bleeding complications in the treatment group [81].
- Another study showed an overall survival benefit with therapeutic anticoagulation in patients with splanchnic venous thrombosis awaiting liver transplantation without excess bleeding complication [35].
While the total number of patients in these reports is small, the safety of therapeutic anticoagulation in selected patients with cirrhosis but without high risk esophageal varices appears comparable to other general medical patients. Finally, as mentioned above, intrahepatic thrombosis may play a role in parenchymal extinction leading to liver atrophy in cirrhosis. Studies are under consideration to determine the possible utility of low dose anticoagulation or platelet inhibition in stable cirrhosis patients. Enoxaparin use in the previously described abstract to prevent PVT also showed a decrease in the occurrence of hepatic decompensation in cirrhotic patients; more studies will be necessary to validate this clinical finding [36]. (See 'Prevention' above.)
CLINICAL APPROACH
Bleeding risk assessment, prophylaxis, and active bleeding — From the foregoing discussion, it is evident that measuring bleeding risk is one of the most challenging areas in the clinical care of liver disease patients. Given the uncertain relationship between conventional tests and hemostasis, it is inevitable that prophylactic measures usually lack strong supportive evidence. Nonetheless, some guidance can be offered, as outlined in the following paragraphs.
Platelet counts parallel thrombin production, with adequate production around levels of 55,000/microL and near normal production at levels closer to 100,000/microL. Therefore, it is reasonable to aim for platelet counts of at least 55,000/microL during moderate risk procedures or interventions and platelet counts closer to 100,000/microL in high risk situations or in the presence of active bleeding. (See 'Platelet level and function' above.)
Because of inter-laboratory variation and the limited relationship of the INR to parameters such as thrombin production, it is difficult to endorse a specific INR cutoff applicable to either risk reduction or intervention in patients with liver disease and active bleeding. With this in mind, and because of the adverse effects of plasma infusion on portal vein pressures and collateral vessel flow, caution is warranted in the use of plasma. A judicial approach is further supported by the very limited effect of the traditional dose of two units of plasma, as this dose is unlikely to significantly alter coagulation factor levels.
An alternative approach in high risk or actively bleeding cases is to measure fibrinogen levels and replace with cryoprecipitate for fibrinogen levels less than 120 to 150 mg/dL. This avoids the use of large volumes associated with plasma infusion. Raising fibrinogen increases the likelihood that a “rescue” agent such as rFVIIa would be effective should there be uncontrollable hemorrhage. (See "Disorders of fibrinogen", section on 'Cryoprecipitate and FFP' and "Therapeutic uses of recombinant coagulation factor VIIa", section on 'Conditions affecting rFVIIa activity'.)
Other general measures include maintaining the hematocrit at 25 percent or more, controlling infection, and treating uremia to minimize platelet dysfunction. Whether or not measurement of other factors such as factor VIII or protein C and whether thromboelastography (TEG) in the clinical setting are helpful remain to be determined.
A high index of suspicion for hyperfibrinolysis (eg, delayed but persistent oozing from puncture wounds or dental work), supported where possible by laboratory studies such as TEG or clot lysis times, should be maintained because of the potential efficacy of anti-fibrinolytic agents in this setting. (See "Thrombotic and hemorrhagic disorders due to abnormal fibrinolysis", section on 'Laboratory diagnosis of abnormal fibrinolysis' and "Thrombotic and hemorrhagic disorders due to abnormal fibrinolysis", section on 'Treatment of abnormal fibrinolysis'.)
Risk assessment when thrombosis is present — There are no clear guidelines to guide treatment of venous thromboembolism (VTE) in patients with liver disease. Aggressive intervention in this setting carries some risk for bleeding that is difficult to measure because of the underlying liver disease and the effects of anticoagulation on the conventional INR. Heparin therapy and prophylactic strategies may also be blunted due to diminished levels of anti-thrombin.
After the diagnosis of an episode of VTE, we recommend determining if the patient is at high risk for the development of gastrointestinal bleeding. This is accomplished via a screening endoscopy for gastric or esophageal varices.
- In the presence of varices, the patient is deemed high risk for anticoagulation. For those with peripheral deep vein thrombosis, placement of a vena caval filter (although controversial) is one option in order to balance these opposing problems.
- If the INR is elevated, there is no accurate reference range to guide traditional warfarin therapy. This should entail discussion with the patient and hematology service to determine the proper course of anticoagulation with a non-warfarin-based regimen.
SUMMARY AND RECOMMENDATIONS
Overview — The fundamental problem underlying the coagulopathy in patients with liver disease is a lack of protein production by the poorly functioning liver. A number of factors should be kept in mind when dealing with such patients. (See 'Problem overview' above.)
- Because both procoagulant and anticoagulant proteins are diminished in unpredictable ratios, the predilection for either bleeding or clotting in an individual patient is not predictable with current testing paradigms.
- Traditional measures of coagulation based on warfarin therapy (eg, use of the prothrombin time, INR, activated partial thromboplastin time) do not generally apply to the coagulopathy of liver disease. These tests should not be the sole source of information in predicting bleeding events, especially post-procedural complications.
- The use of traditional blood products (eg, fresh frozen plasma, plasma concentrates) is not helpful in reversing the coagulopathy of liver disease for more than a transient period of time.
Diagnosing the problem — The following battery of tests is helpful in determining the coagulation status of the patient with liver disease. (See 'Tests of coagulation in liver disease' above.)
- Platelet count
- Prothrombin time
- Activated partial thromboplastin time
- Thrombin time
- Fibrinogen level
- Fibrin D-dimer
Measurement of individual coagulation factors (eg, factors V, VII, VIII) may be helpful in determining vitamin K status as well as the presence or absence of disseminated intravascular coagulation.
Treatment of bleeding — A number of therapeutic options are available for the patient with liver disease and bleeding. However, none has been adequately studied. The proper treatment requires answers to the following questions. (See 'Hypocoagulability in cirrhosis' above and 'Therapeutic agents for bleeding' above and 'Bleeding risk assessment, prophylaxis, and active bleeding' above.)
- Is vitamin K deficiency present
- Is disseminated intravascular coagulation present
- Is the platelet count adequate
- What is the status of fibrinogen (level and function) and is fibrinolysis present
- Is there danger of volume overload if fresh frozen plasma is used
- Is there oral mucosal bleeding which can be corrected via the use of antifibrinolytic agents
Prevention and treatment of thrombosis — Prophylactic or therapeutic anticoagulation and the use of antiplatelet agents are difficult issues to deal with in patients with liver disease, especially since the usual coagulation tests do not adequately assess thrombotic risk in these patients, who are also prone to develop treatment-associated bleeding. This area has also not been well studied, although the cautious use of unfractionated or low molecular weight heparins, with monitoring via measurement of anti-factor Xa levels, is worthy of consideration. (See 'Risk assessment when thrombosis is present' above.)
Searching for
