Patients with severe liver disease are at increased risk for bleeding either spontaneous or induced by drugs or invasive procedures (i.e., liver biopsy, paracenthesis). The bleeding in chronic liver diseases usually comes from esophageal or gastric varices, and less commonly from congested gastric mucosa or gastro-duodenal ulcers. Systemic bleedings are rare and usually manifest from skin, mouth, nose as well as from sites of intravenous catheterization or endotracheal tubes. In many patients with portal hypertension bleeding is aggravated by coexisting clotting alterations commonly present in advanced liver diseases. Platelet count <50.000/μl and PT prolongation of>3 seconds are relative contraindications to elective surgery (1, 2).

Treatment of coagulative abnormalities in liver disease is complex and challenging, since it is influenced on one hand by the type of coagulopathy and on the other, by relative unreliability of routine coagulation tests like prothrombin time (PT), international normalization ratio (INR) or partial thromboplastin time (PTT) (3).

Vitamin K should be given by parenteral way if there is cholestasis related to biliary obstruction or an evidence of bacterial overgrowth, malnutrition or prolonged use of antibiotics. Generally, vitamin K has poor efficacy in treatment of coagulopathies caused by parenchymal liver injury without cholestasis. Anyway, the dose of 10 mg of vitamin K may be given intravenously for 3 days to ensure that vitamin K deficiency is not contributing to haemostatic disorders.

Plasma frozen within 8 hours of collection (fresh frozen plasma; FFP) is the main product used in patients with hepatic-related coagulopathies. Common indication for FFP infusion is persistent bleeding in patients with INR>2 or PT prolongation over 4 sec of normal value (4). The recommended starting dose is 15 ml/kg body weight, however, in practice often larger amounts of FFP are needed to effectively correct haemostatic defect (up to 25 ml/kg).

In majority of cirrhotic patients normally used doses of FFP are inadequate to restore deficient clotting factors. For example, 4 units of FFP increase levels of most clotting factors by only 10%, leaving bleeding patients with increased INR (5). The efficacy of FFP in bleeding patients with severe liver disease should be assessed clinically, because PT or PTT may be not credible indicators of haemostasis in these circumstances (6, 7).

FFP administration is associated with many complications, of which most important are transfusion-related lung injury (TRALI), allergic reactions and transfusion-associated circulatory overload (TACO). Table 1 shows diagnostic criteria and treatment approaches in TRALI and TACO. Biologically active molecules being present in FFP, such like cytokines and antibodies, activate leukocytes, which damage endothelial cells in pulmonary microvessels, leading to development of vascular-related pulmonary edema (8). The treatment of TRALI includes oxygenation and ventilatory support. Diuretics should not be used in this setting. Cautious use of larg amounts of FFP is the most important measure to prevent TRALI. Interestingly, plasma from male donors, who have not been transfused with blood products before donation, contains fewer leucocytes activating molecules than female donated plasma. TRALI remains the most serious complication of FFP transfusion, occurring 12-folds more commonly than it happens with other blood products and carrying a 5-10% mortality (9, 10).

Allergic reactions related to FFP infusion are reported in 1-3% of cases and can be life-threatening in multitransfused patients. Transmission of prions still raises concern regarding FFP therapy (8).

TACO is mainly seen in patients with cardiac failure. Treatment measures are compatible with management of pulmonary edema, but patients may poorly and slowly respond to diuretics. Isovolemic exchange transfusion is the best preventive measure (9). Generally, the volume of transfused FFP should not exceed 2000 ml, and in patients requiring fluid restriction no more than 600-800 ml of FFP may be infused (tab. 2).

Cryoprecipitate (CRYO) is not expensive concentrated source of fibrinogen. Apart from fibrinogen CRYO contains factor VIII, von-Willebrand factor (VWF), fibronectin and factor XIII. It should be remembered that CRYO is depleted of some coagulation factors.

CRYO is recommended in patients with disseminated intravascular coagulation (DIC) or active fibrinolysis with severe hypofibrinogenemia (<50-100 mg/dl). Haemostatic level of fibrinogen is believed to be higher than 100 mg/dl, however, many patients with plasma fibrinogen levels below 50 mg/dl may not present bleeding. An advantage of CRYO in patients with late-stage liver cirrhosis or acute liver failure is small volume of infusion (1 unit has a volume of 10-15 ml). The clinical experience with CRYO in hepatic coagulopathies is limited.

CRYO is transfused within 10-15 minutes after thawing. A single unit of CRYO given per 10 kg body weight is expected to increase fibrinogen concentration by 50 mg/dl in the absence of significant endogenous consumption of fibrinogen or bleeding (11). CRYO replacement therapy does not correct the underlying hyperfibrinolytic defect, but temporary improves fibrinogen concentration. In the randomized comparative clinical trial performed in haemorrhagic patients with acute liver failure, 5 units of CRYO were less effective in improving PT than 4 units of FFP, but one patient in the FFP group developed pulmonary edema (12).

Volume expansion is sometimes required in patients with cirrhosis who are bleeding, have bacterial infection or developed renal failure. Albumin is the most effective volume expander with long half-life. Albumin has shown benefit in maintaining renovascular homeostasis while having minimal effect on portal haemodynamics and haemostasis (13). Volume expansion should be performed with caution, since fluid overload may augment portal pressure, increasing the likelihood of variceal bleeding.

Recombinant factor VIIa (rVIIa) is currently a second-line haemostatic drug characterized by short biological half-life (2-3 hours). A single dose of 4.8 mg should be administered within 10-15 minutes as intravenous transfusion (14) (tab. 2). In study by Bosch et al. (15) 242 cirrhotic patients with gastrointestinal bleeding were randomized to two treatment arms i.e., eight doses of rVIIa 100 μg/kg and placebo. All patients were receiving standard therapy to achieve haemostasis. Benefit in control of bleeding with rVIIa treatment was observed only in patients with advanced cirrhosis (Child Pugh class C). In another study from the same center low and high doses of rVIIa were compared with respect to control esophageal varices bleeding in patients with advanced liver cirrhosis (16). The study showed no difference in controlling bleeding within first 24 hours and no difference in rebleeding and mortality rates at day 5. Shami et al. (17) retrospectively compared eight patients with acute hepatic failure who had been treated with FFP alone with seven patients with the same disease who had been given FFP and low dose rVIIa (40 ?g/kg). The target of this study was correction of INR before placement of intracranial pressure sensor. Only patients treated with FFP plus rVIIa achieved normal INR values, while in those who received FFP alone no correction of the INR has been noticed. Again no differences were seen between both groups regarding bleeding complications, survival time or need for liver transplantation.

The efficacy of rVIIa is limited in patents with low fibrinogen (<100 mg/dl) when concomitant administration of FFP was required (18). Sequential treatment with rVIIa (40-80 ug/kg) and FFP may be useful in case of resistant coagulopathy in acute liver failure. Thrombotic events infrequently seen during therapy with rVIIa and high cost of treatment (dose of 40 ug/kg costs 4000-6000 US dollars) make considerations regarding the clinical efficacy even more difficult (19).

Epsilon aminocaproic acid (EACA) and tranexamic acid (TA) have been widely used to prevent blood loss during liver transplantation; however, experience with these products is mainly based on clinical practice. TA given at high dose (40 mg/kg/hour) showed efficacy in reduction of bleeding (20), but was not effective at lower doses (21). EACA used in 37 cirrhotic patients with subcutaneous haemorrhages caused more than 90% resolution of bleeding in the setting of hyperfibrinolysis (22). DIC needs to be excluded before initiation therapy with antifibrinolytic drugs (tab. 2).

EACA and TA are useful in hyperfibrinolysis or dysfibrinogenemia. These conditions should be suspected in the presence of mucosal bleeding or late bleeding (hours postbiopsy or line placement), suggesting that clot has formed but was prematurely dissolved. Most body fluids, such as saliva, ascitic fluid and urine show their own fibrinolytic activity, thus bleeding in these areas may be amenable to therapy with antifibrinolytic agents (22). In non-cirrhotic patients TA is given at dose 3-10 g, with a loading dose ranging from 2 to 7 g. EACA has been used with an average dose of 10-30 g/24 hours, with loading dose ranging from 1 to 15 g (23). Thrombotic complications may occur at high dose regimens. There are case reports of thrombotic complications involving pulmonary, hepatic or graft vasculature. TA causes nausea and was reported to be responsible for renal dysfunction and skeletal muscle weakness. Optimal doses of these drugs in cirrhotic patients are still unknown.

The use of EACA and TA may be beneficial in patients with liver disease bleeding into body cavities, where natural fibrinolytic activity remains high (12). Both drugs, however, are not commonly used in liver failure due to unsatisfactory efficacy of lower doses, risk of thrombotic complications and difficulties in exclusion of DIC in these patients.

Prothrombin complex concentrate (PCC) contains many coagulation factors that are deficient in patients with liver disease. Twenty one adults with severe liver disease (Quick test <50%) received PCC at a median dose of 25.7 IU/kg for bleeding requiring urgent surgery or endoscopic intervention. The mean value of Quick test increased from 39% to 65% that was associated with significant increase of plasma level of factor VIIa, prothrombin fragment 1+2 and thrombin-antithrombin complex without concurrent thrombotic complications. Clinical efficacy was judged as very good in 16 and good in 5 patients while no PCC related adverse reaction was noted. In all bleeding patients complete cessation of digestive haemorrhage was seen after initial endoscopic treatment and no rebleeding occurred. There was no increased or prolonged bleeding in 18 patients who underwent diagnostic and therapeutic procedures. Among seven adverse events two were fatal, but were judged to be related to the underlying disease (liver failure and septic shock).

There is a concern regarding safety of PCC (12). The thrombogenic risk associated with use of this preparation is well known (24). Thrombosis presumably develops as consequence of introduction into circulation trace amounts of activated coagulation factors (factors IXa and Xa) not cleared by the diseased liver and ATIII deficiency, commonly present in patients with liver failure (25).

The concentration of coagulation factors in PCC is much higher than in FFP, hence a considerable smaller volume of PCC can be used to correct the haemostatic defect. Thus, correction of haemostatic defect with PCC is faster, allowing early initiation of surgery (26-28). Whereas the thawing and infusion of FFP requires at least 60-90 minutes, the infusion of 2000 IU PCC can be accomplished within 10-20 minutes. Substitution of PCC 25 IU/kg led to complete correction of the coagulation tests, whereas therapy with 12 ml FFP/kg body weight resulted in their improvement but not normalization (28). In order to avoid thrombotic complications the coagulation parameters, including AT-III activity should be monitored.

PCC substitution in bleeding cirrhotic patients has not been generally accepted, as use of this preparation is associated with increased risk of thrombogenicity including deep vein thrombosis, pulmonary embolism and systemic DIC. The use of PCC cannot be recommended in cirrhotic patients until randomized clinical studies confirm its safety in this clinical setting (tab. 2).

Exchange plasmapheresis is based on isovolumetric removal and transfusion of fresh plasma. The procedure reduces the risk of volume overload and corrects prolonged PT. Exchange plasmapheresis is useful in selected patients with hepatic failure and severe bleeding in whom FFP and CRYO failed to correct coagulopathy or in presence of severe fluid overload (29).

Normally, platelets are circulating at the periphery of the blood stream being closer to potential binding sites in the event of vascular breach. This pattern of platelet flow is adversely affected with hematocrit levels below 25% (30, 31). It is important therefore to maintain haematocrit level>25% for maintaining marginal platelet flow.

Despite the evidence that there are no threshold abnormalities in coagulation tests predicting bleeding during laparoscopic liver biopsy, measurement of INR and platelet count is essential to assess the bleeding risk of percutaneous liver biopsy. An audit from British Society of Gastroenterology (BSG; 1991) showed a doubling of the risk of bleeding in patients with INR>1.4, however, only 7.1% of the haemorrhagic events occurred with INR>1.5 and 90% of them with INR <1.3. Thus, normal INR clearly does not exclude the risk of bleeding. BSG guidelines require a platelet count above 60.000/μl³ to perform percutaneous liver biopsy without support of platelet transfusion (32).

Other minor procedures such as thoracocentesis, paracentesis or lumbar punctures performed in 608 patients with liver disease and mild coagulation abnormalities were accompanied by excessive bleeding requiring transfusion in only 0.2% of cases with 0.02% mortality. The bleeding risk was not correlated with PT, PTT and platelet count (33). Absolute contraindications to these procedures are symptoms suggestive of DIC or fibrinolysis.

1. FFP is first-line treatment in symptomatic coagulopathy secondary to liver disease.

2. CRYO is an excellent fibrinogen source in hypofibrinogenemia

3. Plasma products should not be prescribed for clinically stable patients without bleeding, as their use raises concern regarding fluid overload transfusion related lung injury.

4. rFVIIa can be used to prevent or control bleeding during placement of intracranial pressure sensor.

5. Use of PCC may be clinically effective, however, due to significant thrombotic risk this drug in some countries is not licensed for liver disease coagulopathies. During PCC infusion the AT-III activity should be monitored.

6. Antifibrinolytics are restricted to hypofibrinogenemia and dysfibrinogenemia occurring in bleeding patients with excessive fibrinolysis.

EACA and TA are fairly effective at lower doses, but may cause thrombotic complications when are prescribed at higher doses. They are contraindicated in DIC syndrome.