Article Text

Download PDFPDF

A conceptual and practical approach to haemostasis in paediatric liver disease
  1. Maria Magnusson1,2,3,4,
  2. Vera Ignjatovic3,4,
  3. Winita Hardikar3,5,
  4. Paul Monagle3,4,6
  1. 1CLINTEC, Division of Pediatrics, Karolinska Institutet, Astrid Lindgren Children's Hospital, Karolinska University Hospital, Stockholm, Sweden
  2. 2MMK, Clinical Chemistry and Blood Coagulation Research, Karolinska Institutet, Stockholm, Sweden
  3. 3Department of Paediatrics, University of Melbourne, Melbourne, Australia
  4. 4Haematology Research, Murdoch Childrens Research Institute, Melbourne, Australia
  5. 5Department of Gastroenterology, Royal Children's Hospital, Melbourne, Australia
  6. 6Department of Clinical Haematology, Royal Children's Hospital, Melbourne, Australia
  1. Correspondence to Dr Maria Magnusson, CLINTEC, Division of Pediatrics, MMK, Clinical Chemistry and Blood Coagulation Research, Karolinska Institutet, Astrid Lindgren Children's Hospital, Karolinska University Hospital, Huddinge, Stockholm SE-141 86, Sweden; maria.magnusson{at}


Children with liver disease can develop severe bleeding episodes and thrombosis. Liver failure usually results in decreased levels of procoagulant and anticoagulant factors. Additional risk factors, including changes in vascular flow and endothelial function, are of importance for the development of bleeding or thrombosis in individual vascular beds. Detailed studies of haemostatic disturbances in the setting of paediatric liver disease are sparse and extrapolation from adult studies is common. The spectrum of liver diseases and the haemostatic system differs between children and adults. Specific paediatric liver diseases are reported to have more distinctive effects on haemostasis and the risk of bleeding and/or thrombosis. Conclusion: we propose a model regarding haemostasis in paediatric liver disease, taking into account a number of specific variables and mechanisms, as well as the type of liver disease, which will provide a framework for clinical decision-making in these complex patients.

  • Haematology
  • Hepatology

Statistics from


Haemorrhage in the context of liver disease may be a life-threatening event. However, in addition to the risk of mucocutaneous bleeding, bleeding from gastrointestinal varices and cerebral bleeding; children with liver disease may develop thrombosis especially portal vein thrombosis. Haemostatic proteins, both procoagulant and anticoagulant, synthesised by the liver, are used as prognostic markers and to evaluate the risk of bleeding both in adults and children.1

Numerous studies have explored the effect of liver disease on coagulation, fibrinolysis and platelets in both acute and chronic adult liver disease.2 ,3 ,4 ,5 ,6 ,7 ,8 ,9 ,10 ,11 The concept of a ‘rebalanced’ coagulation has been developed to explain the changes in the counteractive procoagulant and anticoagulant mechanisms that occur in liver disease.12 This has been important for the understanding that both coagulation proteins and inhibitors have an impact on the patient's capacity to form blood clots. The extensively used haemostatic assay international normalised ratio (INR), unlike in the setting of anticoagulant therapy is not predictive of bleeding risk in patients with liver disease. Consequently, a patient with liver failure and an increased INR cannot per se be regarded as anticoagulated. In the early publications within this field, it was highlighted that patients with liver disease have a very sensitive balance that can easily tip towards both bleeding and thrombosis. However, later studies describe a prothrombotic profile in patients with cirrhosis or acute liver failure.13 These later observations are largely based on thrombin generation assays modified with thrombomodulin and not accounting for the effect of platelets. In addition, most of the patients with cirrhosis in these studies were above 50 years of age and the aetiologies of liver disease were mainly alcohol cirrhosis, hepatitis B or hepatitis C.2 ,3 ,4 ,5 ,9 The bleeding problems observed in these patients are now primarily attributed to pressure effects instead of a specific coagulation dysfunction.14

Detailed studies of haemostatic disturbances in the setting of paediatric liver disease are sparse.15 ,16 ,17 Extrapolation from adult studies is not appropriate as both the spectrum of liver diseases, as well as the haemostatic system itself differs between children and adults. While the simplicity of the rebalance model is appealing, it is not sufficient to describe the loss of function in the haemostatic system and the specific factors (eg, blood flow, endothelial damage, type of liver disease) that may affect haemostasis in children with liver disease. We believe that a model, inspired by Virchow's triad, taking into account a number of specific variables and mechanisms, as well as the type of liver disease, is more appropriate in the setting of paediatric hepatology.18 Here, we propose a new model to consider haemostasis in paediatric liver disease. We also discuss the paediatric-specific elements of haemostasis that this model takes into account, as well as current data focusing on haemostasis in liver disease. We believe that this model represents a tool that will significantly assist clinicians in decision-making in the setting of paediatric hepatology.

Haemostasis in healthy children

The concentration of coagulation proteins and inhibitors differs from adults, especially in the neonatal period and infancy. Neonates have higher levels of von Willebrand factor (vWF), factor VIII (FVIII) and lower levels of vitamin K-dependent factors and inhibitors.19 Vitamin K supplementation is used worldwide in breastfed infants to prevent vitamin K deficiency bleeding.20 The coagulation system continues to evolve during childhood before the coagulation proteins reach adult levels.19 ,21 The levels of antithrombin are low in neonates, which might be of importance considering its antiangiogenic properties. The levels of another thrombin inhibitor Α-2 macroglobulin are increased in neonates and remain high during childhood and adolescence.22 Little is yet known regarding functional differences of the haemostatic system during childhood. However, the endogenous thrombin potential of the thrombin generation assay has been shown to increase with age.19 ,23 ,24 The liver itself does not seem to regulate the developmental changes of the haemostatic system. Specifically, a liver graft from an adult donor produces coagulation proteins according to the age-specific pattern, when transplanted to a child.25 How these developmental changes are regulated remains unknown, however, microRNAs have been suggested as potential modulators.26

The haemostatic system in healthy children is thought to contribute to decreased incidence of spontaneous bleeding and clotting events in infants and children. The risk of thrombosis is significantly lower in children compared with adults (5.3/10 000 compared with 2.5–5/100 in hospitalised adults).27 Thromboprophylaxis is rarely used in children.28 However, when bleeding or thrombosis occurs in this young population, they are associated with a higher morbidity and mortality.29 ,30 Bleeding symptoms may develop secondary to loss of reserve in the procoagulant system (eg, hereditary bleeding disorders), increased blood flow (eg, systemic hypertension, portal hypertension) and endothelial dysfunction/damage (eg, trauma, surgery, aneurysms, necrotising enterocolitis). When thrombosis occurs, mainly in infancy and teenagers, it is often associated with a combination of several risk factors including loss of reserve in the anticoagulant system (eg, hereditary prothrombotic disorders), reduced blood flow (eg, immobilisation, central venous catheters (CVCs), dehydration), endothelial dysfunction/damage (eg, trauma/surgery, CVC) and severe underlying conditions (eg, malignancy, infections, systemic lupus erythematosus).31 ,32 The combination of a prothrombotic condition with stasis and endothelial damage is consistent with the triad attributed to Virchow that explains the pathogenesis of thrombosis.18

Haemostasis in adult and paediatric liver disease

Loss of reserve

Both adult and paediatric studies have shown that a diminished synthetic capacity of hepatocytes in liver failure results in decreased levels of fibrinogen, FII, FV, FVII, FIX, FX and FXI.1 ,16 ,33 Prothrombin time, reported as INR, is often increased in liver disease compared with the healthy population. Activated partial thromboplastin time (APTT) can be normal or elevated.15 ,34 Natural anticoagulant proteins are, however, also synthesised in the liver and the levels of antithrombin, protein S and protein C decline with impaired liver function.16 ,33 ,35 Vitamin K deficiency secondary to cholestasis is well described in paediatric liver disease, especially in infants.36 ,37 This condition leads to impaired γ-carboxylation of the vitamin K-dependent coagulation proteins and inhibitors (including FII, FVII, FIX, FX, protein S and C) and can cause severe spontaneous bleeding including intracerebral haemorrhage.35 ,38

The fibrinolytic system may also be affected by the loss of reserve. Specifically reduced levels of thrombin-activatable fibrinolysis inhibitor, α2-antiplasmin, FXIII and elevated tissue plasminogen activator, pointing towards hyperfibrinolysis, are described in liver cirrhosis.39 In contrast, changes associated with hypofibrinolysis including decreased levels of plasminogen in liver disease and increased levels of plasminogen activator inhibitor 1 in acute liver failure are also described.7 These findings as a whole are based on adult studies. Data regarding fibrinolysis in paediatric liver disease are very limited, although hyperfibrinolysis has been described.15

We hypothesise that the loss of reserve in the procoagulant and anticoagulant protein systems, and probably also in the fibrinolytic system, in liver disease narrows the distance between the clinical end points of bleeding and that of thrombosis (figure 1). Whether the patient actually develops bleeding or thrombosis depends on the additional risk factors in individual patients and is specific to the vascular bed.

Figure 1

A new model to describe mechanisms involved in haemostasis in liver disease; loss of reserve, vascular flow, endothelial damage and special circumstances in specific paediatric liver diseases with distinct impact on haemostasis. PAI-1, plasminogen-activator-inhibitor 1; TAFI, thrombin-activatable fibrinolysis inhibitor; tPA, tissue-plasminogen activator; vWF, von Willebrand factor.

Vascular flow and endothelial damage

Increased vascular resistance, due to structural changes of the liver in combination with a reduced capacity to mediate the vascular tone, is of importance for the decreased portal flow that may develop secondary to cirrhosis.40 Such reduction in portal flow increases the risk of portal vein thrombosis.41 Portal vein thrombosis has been described in 3.7%–10% paediatric patients prior to liver transplantation and in 15% of the subset of patients with end-stage biliary atresia.42 ,43 ,44

The decrease in portal flow also increases the risk of the development of portal hypertension. The resulting increased blood flow in the splanchnic and systemic circulation then leads to secondary splenomegaly, portal hypertensive gastropathy and colopathy, oesophageal varices and rectal varices. Rupture of such varices (ie, endothelial damage) results in significant bleeding. The type of vascular flow affects the structure of the fibrin network during clot formation and therefore has implications on clot strength and stability.45 Most clinical data regarding the effects of portal hypertension in children arise from studies in patients with biliary atresia. Approximately 20% of children with biliary atresia experience oesophageal variceal haemorrhage.46 Recurrent variceal bleeding is associated with a high mortality.47 The importance of correcting the portal pressure to treat this type of bleeding is supported by the fact that pharmacological reduction of the portal pressure with octreotide is effective in achieving bleeding control in approximately 70% of paediatric patients with varicose bleeding secondary to liver disease.48 Bleeding related to portal hypertension from liver disease is associated with a higher mortality compared with portal hypertension secondary to other conditions, supporting the hypothesis that the vascular flow conditions are not the only factor of importance in these patients.49 Consequently, the risk of bleeding and thrombosis differs when comparing different vascular beds within the same patient at the same time point (figure 1). Thus, in a patient with cirrhosis the impedance to flow in the liver may lead to thrombosis of the portal vein at the same time as rupture and bleeding of oesophageal varices occur.

vWF and FVIII are expressed in sinusoidal endothelial cells and are often increased in both adult and paediatric liver disease,16 ,50 a finding that has been linked to endothelial dysfunction induced by endotoxemia.51 The activity of ADAMTS-13 is commonly decreased in liver disease, which also impairs the clearance of vWF.52 ,53 Increase in vWF has been shown to correlate with portal pressure and as recently demonstrated, the stage of fibrosis in adult liver disease.54 ,55 Adult patients with viral hepatitis in combination with an inherited prothrombotic condition have been shown to have an increased progression to cirrhosis.56 This suggests that endothelial dysfunction in combination with loss of reserve may lead to increased development of fibrosis. In children, the impact of endothelial damage and increased levels of vWF are especially highlighted in the pathogenesis of veno-occlusive disease (VOD, also referred to as sinusoidal obstruction syndrome). This condition most commonly occurs following haematopoietic stem cell transplantation or alkaloid intoxication in children and includes disruption of the endothelial lining, fibrin deposition within sinusoids, microthrombosis and development of fibrosis.57

Thrombocytopenia is common in portal hypertension due to sequestration or destruction of platelets in the spleen. Other mechanisms related to liver disease are platelet-associated antibodies, endotoxemia or reduced levels of the platelet growth factor thrombopoetin.58 ,59 ,60 Both platelet dysfunction and platelet hyperactivity have been described in adult liver disease.59 ,61 ,62 The increased levels of vWF and FVIII in liver disease may compensate for thrombocytopenia and platelet dysfunction in this setting.1 ,53 ,63

Special circumstances in specific paediatric liver diseases with distinct impact on haemostasis

Paediatric liver disease encompasses a large spectrum of diseases and the presentation may be acute, chronic or acute-on-chronic. Chronic liver diseases may progress to cirrhosis and portal hypertension.64 Although coagulopathy, bleeding and thrombosis may occur in any type of liver failure, some specific diagnoses are more often reported to be associated with coagulopathy and/or specific increased risk of bleeding and/or thrombosis (figure 1). Coagulopathy predominates in the setting of specific inborn errors of metabolism.65 ,66 Bleeding occurs in vitamin K deficiency secondary to cholestasis, for example, in biliary atresia, Alagille syndrome or α1-antitrypsin deficiency.38 ,67 ,68 Thrombosis is more commonly described in liver abscess and cystic fibrosis.69 ,70 Underlying mechanisms are characterised for some diagnoses including vessel abnormalities in Alagille syndrome, endothelial dysfunctions in veno-occlusive disease and acquired von Willebrands disease in Gaucher disease.57 ,71 ,72 A summary of data related to a number of specific paediatric liver diseases, extracted from mainly small paediatric studies and case reports is given in online supplementary table S1. In several of these conditions, stabilisation of the underlying condition is crucial for management of the haemostatic situation.

Other common complicating factors in liver disease

Both acute liver failure and chronic end-stage liver disease are associated with secondary extrahepatic organ dysfunction.73 ,74 Renal failure is often associated with platelet defects and a bleeding tendency.75 However, if renal failure is a part of systemic inflammatory response syndrome in acute liver failure, a thrombotic picture may dominate.76 Inflammation may lead to increased levels of the acute phase protein fibrinogen promoting a procoagulant state.77 ,78 Sepsis infers a risk of disseminated intravascular coagulation and formation of fibrin thrombi in the sinusoids.79 ,80 However, infection leads also to an increased risk of bleeding from varicose veins due to a heparin-like effect.81 Again data specific to children are very limited and the effect of these complicating factors can therefore be hard to predict. Concurrent diseases may be of importance. A high risk of severe complications has been observed after liver biopsy in patients with previous or ongoing haemato-oncological disease, especially after haematopoietic stem cell transplantation, and during sickle cell crisis.82 Pharmacological treatments may also have an impact on haemostasis.83 ,84 ,85

Laboratory tests specific to haemostasis in paediatric liver disease

Loss of reserve

The loss of function in the procoagulant system is often screened with INR, APTT and fibrinogen.82 For the evaluation of loss of reserve in natural anticoagulants, protein S, protein C or antithrombin may be used, however, the general availability of these tests in different laboratories is poor. D-dimer is the most commonly used laboratory assay used to indicate hyperfibrinolysis. However, none of these tests have been demonstrated to predict the risk of bleeding or thrombosis. Hence, a tool that provides a global overview of haemostasis, including the effect of procoagulant and anticoagulant proteins, platelets and the fibrinolytic system is warranted. Thrombelastography (TEG, Haemonetics Corporation, Braintree, Massachusetts, USA), thromboelastometry (ROTEM, Tem Innovations GmbH, Munich, Germany), thrombin generation (eg, the calibrated automated thrombogram) and sonorheometry are referred to as global coagulation assays and are under evaluation in liver disease.33 ,86 ,87 However, these tests cannot yet fulfil all required clinical needs and have not been shown to correlate with bleeding or thrombosis.86 Thrombelastography and thromboelastometry have, however, been used for several years to guide transfusion therapy in the setting of liver transplant.88 ,89 Modifications of the global coagulation assays are probably necessary to allow their use in children, specifically accounting for the age-specific differences in the coagulation system. This includes for example adjustments for the age-specific differences in α2-macroglobulin.90 Standardised terminology should be used in future studies when reporting how the levels of coagulation factors were measured; that is, the use of antigenic or functional measurements of the haemostatic proteins should be specified.91

Vascular flow and endothelial function

Ultrasound and MRI are used to measure portal blood flow. Preoperative Doppler ultrasound in children with biliary atresia has shown that hepatofugal portal vein flow combined with high hepatic artery resistance index (≥1) are strong risk factors for portal vein thrombosis after liver transplantation.92 Another study investigating post-transplantation Doppler ultrasound showed that absent or low-velocity portal venous flow <30 cm/s or low-velocity hepatic venous flow <25 cm/s was significantly associated with vascular complications.93 The use of hepatic venous gradient measurement (HVPG) in specialised clinical practice is debated. The normal reference range for HVPG measurement in adults is 1–5 mm Hg. Complications of portal hypertension are seen in adults when HVPG measurement is above 10 mm Hg. Small studies indicate similar pressure thresholds in children, however, this approach is currently used as a research tool.94 ,95 ,96 Surrogate markers such as spleen size, thrombocytopenia and the development of varicose veins are routinely used to evaluate the degree of portal hypertension. A clinical prediction rule including platelet count, spleen size and albumin has been developed for the prediction of oesophageal varices in children.97 VWF has been suggested as a marker of both endothelial dysfunction and portal hypertension.51 ,54

Currently, no single test can predict the risk of bleeding or thrombosis in patients with liver disease because the risk may be different in individual vascular beds based on the local impact of abnormalities in flow or endothelial function (table 1).

Table 1

Laboratory tests used to evaluate haemostasis in paediatric liver disease, in clinical practice or as research tools

Further focused research effort is required to understand how laboratory assays can best measure loss of liver functional reserve and hence the degree of additional insult that will cause bleeding or clotting in this population. A framework to superimpose the specific coagulation impact of particular types of liver disease is then required.


We hypothesise that bleeding and thrombosis in paediatric liver disease most often occur secondary to a combination of loss of reserve in the haemostatic components, changes in blood flow and endothelial damage or dysfunction, the type of liver disease and the wider clinical context. Conceptualising the bleeding or clotting risks in this way enables a more practical approach to the holistic care of the patient. Consideration of the impact of interventions, such as CVCs that cause endothelial or flow problems and the risks of complicating features such as varices, in the context of the status of the haemostatic system provides an improved framework for clinical decision-making.


View Abstract


  • Contributors All members analysed the data and developed the new model. MM drafted and revised the manuscript. VI, WH and PM revised the draft manuscript. All authors approved the final manuscript prior to submission.

  • Funding Samariten foundation, the Sven Jerring Foundation, the Swedish Order of Freemasons, the HRH Crown princess Lovisa Foundation, Swedish Society of Medicine, Karolinska Institutet foundation.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.