• Open Access

Clinical utility of viscoelastic tests of coagulation in patients with liver disease

Authors


Abstract

The prothrombin time (PT) and international normalised ratio (INR) are used in scoring systems (Child-Pugh, MELD, UKELD) in chronic liver disease and as a prognostic tool and for dynamic monitoring of hepatic function in acute liver disease. These tests are known to be poor predictors of bleeding risk in liver disease; however, they continue to influence clinical management decisions. Recent work on coagulation in liver disease, in particular thrombin generation studies, has led to a paradigm shift in our understanding and it is now recognised that haemostasis is relatively well preserved. Whole blood global viscoelastic tests (TEG®/ROTEM®) produce a composite dynamic picture of the entire coagulation process and have the potential to provide more clinically relevant information in patients with liver disease. We performed a systematic review of all relevant studies that have used viscoelastic tests (VET) of coagulation in patients with liver disease. Although many studies are observational and small in size, it is clear that VET provide additional information that is in keeping with the new concepts of how coagulation is altered in these patients. This review provides the basis for large scale, prospective outcome studies to establish the clinical value of these tests.

Abbreviations
ALF

acute liver failure

ALI

acute liver injury

BCS

Budd Chiari syndrome

BT

bleeding time

CCT

conventional coagulation tests

CLD

chronic liver disease

DVT

deep vein thrombosis

ETP

endogenous thrombin potential

FFP

fresh frozen plasma

GAG

glycosaminoglycans

HLE

heparin like effect

INR

international normalised ratio

ISI

international sensitivity index

LMWH

low molecular weight heparin

MELD

model of end stage liver disease

MRTG

maximum rate of thrombin generation

NAFLD

non alcoholic fatty liver disease

PAI

plasminogen activator inhibitor

PBC

primary biliary cirrhosis

PSC

primary sclerosing cholangitis

PVT

portal vein thrombosis

ROTEM®

thromboelastometry

TF

tissue factor

TEG®

thromboelastography

TTG

total thrombus generation

TMRTG

time to maximum rate of thrombus generation

TIPS

transjugular intrahepatic portosystemic shunt

TRALI

transfusion related acute lung injury

UFH

unfractionated heparin

VET

viscoelastic test

VTE

venous thromboembolism

Conventional coagulation tests (CCT) are abnormal in acute and chronic liver disease (CLD) and are interpreted as demonstrating an underlying bleeding diathesis. However, standard coagulation tests do not predict bleeding, nor do they provide sufficient information to optimise the management of bleeding events [1, 2] Recently, a new paradigm has been described whereby thrombin generation in patients with liver disease is much better conserved than previously thought when the test conditions were adapted to reflect the contribution of the anticoagulant pathways [3] and it is now recognised that there is an increased risk of thromboembolism in CLD [4].

Thrombin generation tests (TGT) have revealed important new information on haemostasis in liver disease, but these tests are not readily available, and therefore have poor clinical applicability and furthermore, there are no studies comparing the TGT to a clinical endpoint. Whole blood global viscoelastic tests (VET) of coagulation are increasingly used for point of care (POC) analysis of the complex coagulopathies that can occur during cardiac surgery and following major trauma [5, 6]. They differ from CCT as they evaluate the kinetics of coagulation from initial clot formation to final clot strength. These dynamic tests provide a composite picture reflecting the interaction of plasma, blood cells and platelets, and more closely reflect the situation in vivo than do CCT, as these are performed solely in plasma and measure only isolated end points. In addition, VET provide valuable information on the presence and severity of fibrinolysis and also hypercoagulability [7]. Since the early 1980s, VET have been used for POC coagulation monitoring during orthotopic liver transplantation (OLT) [8]. The possibility that there may be more clinical benefit in using VET rather than CCT to assess and stratify bleeding or thrombotic risk in patients with liver disease is a tantalising idea that can only be answered by prospective clinical outcome studies [9] The purpose of this article is to review relevant published studies on VET and liver disease, in the context of the current understanding of the coagulopathy of liver disease, to establish evidence if VET could be used as routine coagulation tests in this setting.

Coagulopathy of liver disease

Coagulation and haemostasis is a dynamic process with interplay between primary haemostasis and platelet plug formation, secondary haemostasis and thrombin generation and fibrinolysis. The haemostatic changes that accompany liver disease affect all aspects of coagulation. An increased bleeding diathesis has been considered a traditional hallmark of acute and CLD [10], but it is now recognised that systemic hypercoagulability and thrombosis can also be present and these patients cannot be considered ‘auto-anticoagulated’ [11]. Stable patients with liver disease exhibit finely tuned ‘rebalancing’ of their haemostatic profile [12] and this is reflected in an increasing number of patients with chronic liver disease (CLD) who undergo major abdominal surgery, such as OLT without need for blood or blood products [13]. However, the haemostatic balance is precarious and both endogenous and exogenous factors can readily tip the balance towards either a bleeding tendency or a prothrombotic state, as these patients lack the buffering capacity of a large functional reserve with its associated regulatory mechanisms that is seen in health [14]. Quantifying this imbalance is the key to establishing a clinically useful paradigm for managing patients with liver disease [15].

Primary haemostasis

Platelets exert important haemostatic functions in vivo including primary platelet plug formation (adhesion/aggregation) and provide a membrane surface for the assembly of complexes necessary for thrombin generation. Platelet numbers decrease progressively because of portal hypertension and associated hypersplenism and impaired hepatic synthesis of thrombopoetin. In addition, there can be abnormalities in platelet function. However, increased levels of von Willebrand factor (vWF) and reduced activity of its cleaving enzyme ADAMTS- 13 compensate for some of these changes [16]. Platelet hyperactivity has been reported in patients with cholestatic liver disease [17, 18] A systematic review evaluating qualitative and quantitative aspects of platelet function [16] concluded that primary haemostasis is not normally defective in cirrhosis. Therefore, a low platelet count should not necessarily be considered as indicating an increased risk of bleeding, with the caveat that with severe thrombocytopenia, correction is advised if bleeding occurs, or prior to performing invasive procedures. There is consensus that platelet transfusion is indicated in cirrhotic patients with low-platelet counts (50 000 or less) during active bleeding [19]. The evidence for the commonly set lower cut off values for platelet count is sparse and limited by small sample size. A preprocedure platelet count of 50 000 is considered adequate [20] and this is reinforced by endogenous thrombin generation studies [21]. Giannini [22] studied 121 consecutive patients who were being evaluated for liver transplantation and were undergoing invasive procedures. Bleeding occurred in 31% with severe thrombocytopenia and in none of those with moderate thrombocytopenia [23]. Platelet function was traditionally assessed by bleeding time (BT). However in cirrhosis, there is a poor association between platelet count and BT and a prolonged BT can be seen in patients with platelet counts >100 000 and vice versa. [24] As platelet activation is not diminished but may be increased in some patients with cirrhosis, it is possible that BT prolongation is also a result of changes in vasoreactivity and/or arterial dysfunction [24].

Secondary haemostasis

In chronic liver disease (CLD), most procoagulant factors concentrations are decreased, except factor VIII, which is elevated. Decreased levels of procoagulants are accompanied by a concomitant decrease of the naturally occurring anticoagulants (antithrombin, protein C and S) [3]. In normal conditions, the coagulation system is balanced by the two opposing drivers and thrombin generation is no different or even increased in stable liver disease compared to healthy individuals when the test is modified to incorporate the natural anticoagulant pathways [3, 25]. This apparent paradox is explained by the fact that Protein C (PC) and antithrombin (AT) need to be activated to exert their full anticoagulant activity with thrombomodulin and with glycosaminoglycans (GAGs) [26, 27] which are located on the vascular endothelium. This aspect is not evaluated in the majority of coagulation analyses and this pitfall is particularly important in cirrhosis where both anticoagulants and procoagulants are reduced [28].

Fibrinolysis

Fibrinolysis is a complex physiological process involving the interaction and balance between a number of different activators and inhibitors. In liver disease, there is increased fibrinolytic activity and clot instability because of increased tissue plasminogen activator (tPA) with low levels of alpha 2 antiplasmin, factor XIII and reduced thrombin activated fibrinolysis inhibitor (TAFI). However, this is balanced by increased levels of the acute phase reactant plasminogen activator inhibitor (PAI-1). Levels of PAI-1 are particularly high in acute liver failure and in cholestatic liver disease, and significant fibrinolyisis is rare in these groups [29, 30].

Limitations of standard coagulation tests in patients with liver disease

The PT/INR was developed to monitor oral anticoagulant therapy and PTT to investigate the inheritable single factor deficiencies, for example, haemophilia, and to monitor heparin therapy. These tests were never intended to model in vivo haemostasis or to assess perioperative bleeding risk. Many patients with liver disease have a normal PTT, despite mild baseline deficiencies of multiple procoagulant factors. This may be because of the elevated levels of factor VIII which shorten PTT and compensate for the multiple procoagulant factor deficiencies [31]. The PT/INR is widely used to assess the risk of bleeding in patients with liver disease; however, the evidence from clinical practice and the literature is that it does not correlate with bleeding after liver biopsy or other procedures [32]. Despite this, transfusion of fresh frozen plasma (FFP) is often used in an attempt to correct the international normalised ratio (INR) [33, 34]. Epidemiological studies suggest that patients with CLD have the greatest individual risk of transfusion related acute lung injury compared to other populations [35]. Observational studies show that even major procedures, such as liver transplantation, can be performed without administration of FFP despite an increased INR [36]. Most importantly, the INR value varies between laboratories in patients with liver disease, so defining a set cut off value is problematic [37] Other limitations of PT/INR are that it is not possible to estimate the overall strength and stability of the clot because these tests are read at the initiation of fibrin polymerisation which happens at very low levels of thrombin generation of about 10–20 nM, which is less than 5% of the total thrombin that can be generated [38].

The INR threshold of 1.5 for bleeding risk is derived from studies that originally used a PT threshold and thromboplastin reagents which had an international sensitivity index (ISI) greater than or equal to 2. Whilst the calculated INR of 1.5 mathematically corresponds to a PT ratio of 1.5 for thromboplastin reagents with an ISI of 1.0 as used currently, this does not take into account the fact that many of the earlier studies on PT threshold were done with less sensitive thromboplastins and the corresponding INR would actually be 2.25 to 4.0 [31]. This, together with the fact that the INR does not reflect the concurrent reduction in anticoagulant levels in patients with liver disease, may explain why there is no consistent relationship between bleeding and a mild to moderate increase in INR in patients with CLD.

There is no good evidence for administering prophylactic FFP according to baseline INR or indeed to improve outcomes [39, 40]. This leads to unnecessary and wide variability in the use of FFP. Tripodi assessed the effects of in vitro addition of pooled normal plasma (PNP) to the plasma of 58 adult patients with advanced cirrhosis and showed that although the PT ratio shortened in many patients, there was no change in thrombin generation. These results cast doubt on the efficacy of FFP to reduce the bleeding risk in patients with liver disease who are undergoing invasive procedures [41] and is an area that needs urgent research.

Thrombin generation

Thrombin generation assays are global coagulation tests that measure the dynamics of thrombin production using small amounts of tissue factor (TF) as a trigger factor to achieve full interplay of all coagulation factors. TGT have been used to identify patients at increased risk of thrombosis [42-44], and a high endogenous thrombin potential (ETP) is associated with an increased risk of recurrent thrombosis. Conversely, reduced thrombin generation is documented in patients with a bleeding tendency [45, 46]. The normal, or even enhanced, thrombin generation in stable patients with CLD explains, at least in part, why many of these patients do not have a significant increased bleeding risk and may be at increased risk of thrombosis [44]. Following Tripodi's landmark paper, in which thrombin generation in cirrhosis was shown to be the same as in healthy people when thrombomodulin was added to activate protein C [21], further papers indicate that thrombin generation may actually be increased [47]. Gatt et al. [25] studied 73 adult patients with cirrhosis and also 38 healthy individuals. Thrombin generation was assessed using the calibrated automated thrombography (CAT) [48]. Rather than thrombomodulin, Protac® modified TG was used. (Protac is a snake venom extract that activates PC).This study showed a hypercoagulable TG profile in plasma in cirrhosis, with an increased velocity of TG and higher ETP ratios. This is the same profile as patients with protein C/protein S deficiency and factor V Leiden, in whom a greater risk of thrombosis is well documented [49]. Overall, the data on TG velocity and Protac resistance demonstrate a prothrombotic tendency in plasma of patients with cirrhosis. These findings are in keeping with reports that patients with liver disease are not protected against thrombosis despite a raised INR [4], [50] and have an increase thrombotic risk compared with age-matched controls.

Although thrombin generation studies have increased our understanding of the coagulopathy in liver disease, they are for the time being, mainly research tools that are laboratory based. CCT have been shown to be inadequate for the purposes of stratifying bleeding and thrombotic risk in patients with liver disease and this mandates a search for alternative means of assessment which better reflect functional changes in coagulation [51].

Viscoelastic tests of haemostasis: thromboelastography (TEG®) and thrombelastometry (ROTEM®)

The cell-based model of haemostasis [52], in contrast to the traditional description of intrinsic and extrinsic pathways, emphasises the role of platelets in thrombin generation, and highlights the importance of the dynamics of thrombin generation influencing the quality and stability of thrombus formed. For haemostasis to occur effectively, there must be sufficient thrombin generation (coagulation factors and platelets), adequate substrate (fibrinogen), and clot stability. VET measure changes in clot tensile strength over time and give information on the dynamics of clot formation (coagulation factor and anticoagulant activity), clot strength (platelets and fibrinogen), and clot stability (fibrinolysis and factor XIII). The use of goal-oriented algorithms based on VET facilitates targeted transfusion with specific haemostatic agents and avoids empirical administration of multiple components with potentially hazardous effects [53] A Cochrane systematic review has confirmed that VET reduce overall transfusion and associated costs, but the outcome benefit is unclear [54].

Thromboelastography (TEG)/thromboelastometry (ROTEM)

There are two commercially available devices, both based on Hartert's invention in 1948 [55], the TEG® (Hemonetics Corporation, Braintree, MA, USA) and the ROTEM® (TEM International GmbH, Munich, Germany). In this review, the term (VET) will be used to describe general principles of the common technology, but differences will be specified as TEG or ROTEM respectively.

Basic principles

The changes in viscoelastic or tensile strength that occur during coagulation are recorded by both TEG and ROTEM. The rate of fibrin polymerisation with incorporation of platelets and overall clot strength is displayed visually on the TEG/ROTEM trace (Fig 1) and also numerically, providing a complete picture of clot initiation, formation and stability [56].The TEG and ROTEM parameters have a different nomenclature, but refer to identical stages of clot formation. The R (reaction) time on the TEG is the same as the CT (clotting time) on the ROTEM and is the time to initial fibrin formation. The K and α angle on the TEG [represented by the clot formation time (CFT), and α angle on the ROTEM] are indicative of the kinetics of clot formation. The maximum amplitude (MA) on the TEG, or maximum clot firmness (MCF) on the ROTEM, corresponds to the maximal clot strength and is dependent on platelet count and function, fibrinogen and also factor XIII. The clot lysis index (CL 30,60 for TEG and LY30,LY60 for ROTEM) is the% reduction from MA or MCF at 30 or 60 minutes and is used to determine if fibrinolysis is present (Fig 1). Different haemostatic profiles give characteristic VET tracings (Fig 2). For both tests, different reagents and activators are available and increase the diagnostic capabilities (Table 1) Although both TEG and ROTEM give broadly similar information, differences in the disposable blood sampling cups, and in the nature of the activators used, means that results are not exactly comparable, and also algorithms developed for one technique are not directly transferable to the other [57]. For further information on general principles, there are many excellent reviews [58-60]. The TEG has two channels, whereas the ROTEM has 4, allowing more samples or diagnostics to be run simultaneously (Fig 3). The ROTEM is also less susceptible to movement and vibration artefact. The factors determining whether TEG or ROTEM was used in the studies covered in this review appear to have been influenced significantly by local issues, such as availability of the device and/or historical use.

Table 1. TEG/ROTEM Activators & Reagents
 Activator Clinical comments
TEG
 NativeVery sensitive to trace quantities of heparinUseful in detection of HLE and for monitoring heparin and LMWH
KaolinActivation of Intrinsic system

Less sensitive to heparin.

Standard activator for TEG analysis

HeparinaseReverses heparin and heparinoids

Detection of HLE and heparin

Assess efficacy of LMWH heparin

Tissue Factor

(rapid TEG)

Activation of extrinsic system

R time too short to give clinically useful information

Rapid acquisition of MA

Functional fibrinogenPlatelet contribution to clot removed with abciximab

Fibrinogen contribution to MA

Good correlation with Clauss fibrinogen

ROTEM
 INTEMIntrinsic system activated with ellagic acidDetects abnormalities in the intrinsic system
EXTEMExtrinsic system activated with tissue factorDetects abnormalities in the extrinsic system
FIBTEMCytochalasin D removes platelet component of MCF

Functional fibrinogen

Good correlation with Clauss fibrinogen

HEPTEMReverses heparin

Detection of heparin

Assess efficacy of protamine

APTEMAprotinin reverses any fibrinolysisConfirms fibrinolysis and allows assessment of coagulation status post antifibrinolytic therapy
Figure 1.

Schematic of TEG/ROTEM parameters.

Figure 2.

Examples of Different Haemostatic Profiles on TEG®.

Figure 3.

ROTEM® tests.

Correlation of CCT and VET

PT/INR

The plasma-based tests, PT and aPTT, reflect the lag time for the initiation of polymerised fibrin gel formation after activation with TF (extrinsic pathway) and contact activation with ellagic acid, kaolin and others respectively (intrinsic pathway). Correlation between reaction time and clotting time (R/CT) and PT/INR is weak (r = 0.24–0.37) [9, 57, 61, 62]. Several studies demonstrate that the R/CT is not sensitive to mild to moderate increases in INR ( = <1.6) [63, 64]. In models of dilutional coagulopathy, an increase in CT occurs only when clotting factor concentrations are reduced to levels of 30% or less [65]. In addition, in contrast to plasma based CCT, the inclusion of platelets (i.e. whole blood) in VET affects the onset (R/CT) and rate (K, CFT) of fibrin polymerisation caused by platelet-mediated procoagulant reactions, and platelet–fibrinogen interactions. In liver disease, wide derangements in INR are not often mirrored by similar changes in R/CT [66] and this is in keeping with the fact that the INR is a poor predictor of clinically important bleeding [32].

Platelets

Clot strength as assessed by the maximum amplitude (MA) or maximum clot firmness (MCF) is highly influenced by both fibrinogen levels and platelet count [67]. The minimal platelet count for normal clot formation on VET is markedly affected by the fibrinogen level. In a study in patients with idiopathic thrombocytopenic purpura (ITP), the critical cut-off for platelet count to affect MCF was 31 × 109and the critical fibrinogen level was 375 mg/dl [68]. Other than the platelet count the MCF was the most important parameter in predicting bleeding in patients with ITP. Others have found that MCF is greatly decreased when the platelet count falls below 50 000 × 109 [69]. In liver disease, where fibrinogen levels are usually within the normal range, platelet count may have a more significant impact on changes in MA/MCF. The combination of both a low platelet count and a low fibrinogen always results in a reduced MA/MCF and is strongly associated with an increased bleeding tendency [70].

Fibrinogen

Fibrinogen levels vary greatly among patients. The Clauss method is currently the gold standard for determination of fibrinogen and depends on thrombin induced fibrin formation. It is affected by multiple factors including the presence of colloidal solutions (starch and gelatins) and also direct thrombin inhibitors [71, 72]. In the ROTEM FIBTEM test, the addition of cytochalasin D inhibits GPIIb/IIIa interaction thereby removing the platelet contribution to MCF and has good correlation with plasma fibrinogen levels [73]. A functional fibrinogen assay (abcixmab is used to remove the platelet contribution) is also available for the TEG [74]. The two tests give slightly different values [75].

Fibrinolysis

Validation of VET to assess fibrinolytic activity and to determine individual susceptibility to fibrinolysis is an area of increasing interest [76, 77]. Clinically significant fibrinolysis is detected by the clot lysis index (CLI), when there is a rapid decline in MA/MCF over time. A CLI of >15% is considered hyperfibrinolysis [59]. The fibrin clot is more susceptible to fibrinolysis after massive haemodilution because of the progressive loss of endogenous fibrinolysis inhibitors [78]. The ROTEM assay, APTEM, which contains aprotinin, confirms the diagnosis and in addition, by reversing any fibrinolysis, allows pre-assessment of the coagulation profile after antifibrinolytic therapy, thus enabling earlier administration of other prohaemostatic therapy if necessary [60].

Thrombin generation

The rate and amount of thrombin generation are considered predictive for both thrombosis and haemorrhage. A thrombus velocity curve or V curve can be obtained from the TEG waveform using a software program. The V curve is plotted from the first derivative of changes in clot resistance expressed as a change in clot strength per of unit time (dynes/cm2/s) representing the maximum velocity of clot formation. Parameters obtained are total thrombus generation (TTG), maximum rate of thrombin generation and time to maximum rate of thrombus generation (TMRTG) (Fig 4). A small study in healthy volunteers demonstrated that thrombin–antithrombin (TAT) complexes, a surrogate marker for thrombin generation, correlated well with TMRTG and TTG [79]. However, the rate of clot formation on the TEG can only be assumed to be directly proportional to the rate of thrombus formation if the platelet count, fibrinogen and factor XIII levels are normal. This means that thrombin generation inferred from the V curve, in liver disease, should be interpreted with care as the platelet count affects not only final clot firmness (MA) but also the rate of clot propagation. The TEG has also been used to assess the rate of clot formation in haemophiliac patients [80, 81].

Figure 4.

Thrombin generation. The image “TEG® Velocity Parameters” is used by permission of Haemonetics Corporation. TEG® and Thrombelastograph® are registered trademarks of Haemonetics Corporation in the US, other countries or both.

Viscoelastic tests and chronic liver disease

Because TEG/ROTEM are global tests providing a composite analysis that reflect function of plasma, blood cells and platelets, they are increasingly viewed as an appropriate tool to investigate the coagulopathy of CLD. Tripodi et al. compared ROTEM parameters between 58 healthy volunteers and 51 adult patients with cirrhosis [9]. Abnormal ranges were defined as above the 95th percentile for CT and CFT or below the 5th percentile for MCF. ROC curves were constructed to identify patients with cirrhosis (true positives) from healthy individuals (true negatives). The CT did not distinguish between healthy and cirrhotic individuals and there was no correlation between PT and CT (r = −0.264) and only 27% of patients with cirrhosis had any prolongation of CT despite the fact that PT was prolonged. MCF was a good discriminator and 76% of patients with cirrhosis had an abnormal (low) value. The MCF also correlated well with model of end stage liver disease score. There was good correlation between platelet count and MCF (r = 0.691) and also CFT (r = 0.741). Clauss fibrinogen correlated reasonably well with MCF (r = 0.590). It was concluded that VET may be useful to assess the severity of CLD and can be used to distinguish between healthy and cirrhotic individuals.

Heparin like effect and CLD

The native TEG is extremely sensitive to the presence of heparin and heparin like substances. Coppell et al. investigated the effects of unfractionated heparin (UFH), low molecular weight heparin (LMWH) and danaparoid on native and heparinase TEGS. The difference between parameters in these two tests was able to differentiate between a range of low concentrations (0.005–0.05 U/ml) of these heparin like substances and demonstrated a clear dose response, and in the case of UFH there was greater sensitivity than with anti-Xa activity [82]. Although the standard assay for monitoring LMWH is by inhibition of factor Xa (anti-Xa activity), this test is not routinely available at all institutions, and there are some concerns relating to interassay variability. Whilst native TEG is undoubtedly the most sensitive method to detect low concentrations of heparin, kaolin-activated TEGS have also been found to be a useful method to monitor and guide LMWH therapy in sick hospitalised patients, where co-morbid conditions can impact on both the pharmacodynamics and pharmacokinetics of LMWH [83].

In recent years, there has been increasing interest in the detection of, and the significance of endogenous heparins. Under conditions of endothelial stress, such as surgery or sepsis, endogenous release of very small quantities of GAGS may be detected systemically [84]. Minor disturbances of the endothelial glycocalyx can lead to the selective cleavage of heparan and chondrotin sulphate side groups from the luminal layer of the glycocalyx. Where there is more significant damage to the vascular endothelium from ischaemia or sepsis, systemic activation of coagulation is promoted, and it is thought that this shedding of GAGS into the circulation is an adaptive response to keep a progressively more procoagulant microvasculature open by means of endogenous heparinisation [85]. When shed, the glycocalyx GAGs retain their anticoagulant activity and this is detectable by a prolonged r-value on TEG analysis [86]. These endogenous GAGS may represent an increase bleeding risk for some patients [87, 88] and demonstration of their presence may provide clinically useful information. In a prospective observational study in 30 patients with cirrhosis, Mancuso et al. demonstrated that citrated samples (allowing a delay in running the analysis) give comparable results to samples that are run immediately and facilitates the logistics of using TEG when it is not close to the patient [89]. Bacterial infection in cirrhosis induces a Heparin like effect (HLE) detected by TEG [88] and this reverses with antibiotics and resolution of the infection. HLE is associated with detectable anti-Xa activity [90, 91] and appears to differentiate patients at increased risk of variceal rebleeding [92]. A transient HLE in systemic venous blood after transjugular intrahepatic portosystemic shunt (TIPS) has been reported, suggesting a high concentration of heparinoids in the portal venous system prior to TIPS placement [93].

Hypercoagulability and CLD

Hypercoagulability may have an important role in many aspects of liver disease and intrahepatic microthrombi have been implicated in the progression of fibrosis [94]. Portal vein thrombosis (PVT) is a common complication of cirrhosis, with an incidence of 10–25%, with a greater tendency to thrombosis with more severe liver disease [95]. In cirrhosis, the ratio of the two most powerful pro and anticoagulants in the plasma, factor VIII and protein C, respectively, show a balance strongly in favour of factor VIII indicating hypercoagulability [96].

Ben-Ari et al. evaluated hypercoagulability in patients with primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC) using TEG. 28% of patients with PBC, and 43% of patients with PSC were found to be hypercoagulable compared to only 5% of non-cholestatic cirrhosis and none in healthy controls [17]. In a prospective, observational study in non-alcoholic fatty liver disease (NAFLD) using TEG, a significantly stronger clot development was found in patients compared to healthy controls (MA 58.3 ± 6.3 vs. 52 ± 10 mm P = 0.01,) and the platelet contribution to overall clot strength was higher in NAFLD patients with a trend to reduced inducible clot lysis (P = 0.03) [97]. In a prospective study in 23 patients with obstructive jaundice, 80% were found to be hypercoagulable on TEG analysis (increased MA) and this was independent of prolonged PT times. A repeat TEG 3 weeks after a biliary drainage procedure, showed all TEG parameters had returned to normal range [98].

The clinical implications of these findings have yet to be evaluated. However, emerging evidence suggests that hypercoagulability detected by VET puts patients in an ‘at risk’ group for both venous and arterial thrombotic events [7, 99]. A recent systematic review of 10 studies in surgical patients showed an increased MA to be the most important parameter to predict postoperative TE events. However, there was considerable variability as to which parameters were used to define hypercoagulability and no study was adequately powered. Nevertheless, the vast majority of patients who had a TE event were hypercoagulable on one or more TEG parameters [100] and future prospective studies are recommended.

Viscoelastic tests and acute liver disease

In over 1000 patients reviewed by The Acute Liver Failure Study Group (ALFSG), the mean INR was 3.8 [101] Patients with ALF are assumed to have a bleeding diathesis based on an elevated INR. However, clinically significant bleeding is rare. Although blood clot formation by TEG is generally preserved in stable patients with cirrhosis [88], patients with acute liver injury (ALI) and ALF have not been extensively studied.

As an ancillary project of the ALFSG, Stravitz prospectively studied 51 patients with ALI/ALF with kaolin initiated TEG [102]. Despite a mean INR of 3.4 (range 1.5–9.6) mean TEG parameters were within normal limits for the entire study population, and all 5 individual TEG parameters were completely normal in 63% of patients suggesting that the dynamics of clot formation are generally well preserved. Moreover, 8% of patients were hypercoagulable. The TEG was significantly more sensitive than INR for predicting bleeding and the INR was not significantly different in those who bled and those who did not. The MA was higher in ALF than ALI and correlated with increasing severity of liver injury. The preservation or even increase in MA in patients with ALI/ALF may be because of increased factor VIII levels, decreased ADAMTS13 activity, increased vWF and increased levels of fibrinogen and or platelets as acute phase reactants. This important study demonstrates TEG parameters in ALF/ALI are generally well preserved and potentially provides an explanation for why clinical bleeding is rare despite the elevated INR. The authors conclude that INR, although a valid indicator of prognosis, is not a good guide for administration of procoagulant therapy.

In a prospective study, in our own institution of 20 patients with ALF admitted to the intensive care unit, coagulation analysis was performed on admission and at 48 h. [104]. CCT suggested a markedly hypocoagulable state with a significantly raised INR (mean 4.3); however, TEG values were hypocoagulable in only 20% of patients, whilst 45% had normal and 35% had hypercoagulable profiles All patients with hypocoagulable TEGs had platelet counts <100 000.

HLE & ALF

A HLE is commonly seen in ALF [104, 105] This HLE is thought to be because of the release of endogenous heparinoids and reflects the vascular endothelial injury inherent with ALI. In ALF, the R time is significantly increased in the presence of infection, renal failure and in those with bleeding complications [102]. In an observational study comparing TEG parameters in ALF to those in cirrhosis, Senzolo et al. found that R and K time and alpha angle on native TEGs were significantly more hypocoagulable in ALF patients undergoing OLT compared to control stable patients with cirrhosis. These TEG changes were ascribed to endogenous heparinoids as heparinase reversed these differences [106]. Using the TEG ‘V’ curve as a surrogate for thrombin generation, TTG was generally found to be similar to normal controls. Therefore, although endogenous heparinoids slow the velocity of initial clot formation, they did not ultimately affect final clot strength. Heparinase-modified TEG should be considered a useful adjunct in the assessment of coagulopathy in ALF.

Anticoagulation and liver disease

Patients with liver failure have traditionally been managed with no or minimal anticoagulation because the abnormal clotting tests are perceived to reflect an increased bleeding risk. However, patients with cirrhosis can develop deep vein thrombosis (DVT) despite a prolonged INR and can do so even when receiving antithrombotic prophylaxis [107]. In addition, as many as 5–10% of patients with advanced liver disease will develop PVT each year [108] Anticoagulant drugs should be administered with caution in patients with liver disease. The bioavailability of heparin and LMWH cannot be assumed to be stable in patients with liver disease as this will be affected by fluctuations in liver synthetic function and also changes in hepatic clearance and renal function [109]. It is known that patients with cirrhosis show an increased response to LMWH and this correlates with the severity of liver disease [10].

LMWH are routinely prescribed for venous thromboembolism prophylaxis in general medical and surgical patients in a standardised dose and monitoring is generally thought to be unnecessary. However, in patients with liver disease, effective and safe dosing is more problematic. Antifactor Xa levels are the gold standard for monitoring LMWH activity, but these tests are not routinely available, and they are expensive and standardisation between different laboratories can be a problem [111]. In addition, monitoring anti-Xa levels in cirrhosis is unreliable because of the low levels of antithrombin [112]. An increasing number of published papers suggest that VET may be a useful way to assess the efficacy of LMWH therapy in general surgical patients with various co-morbidities that will affect the pharmacokinetics of these drugs. Van et al. measured antifactor Xa levels and performed simultaneous kaolin and heparinase TEGS in 61 surgical ICU patients (261 time points) all receiving prophylactic therapy with Enoxaparin: 17 patients developed a DVT. Overall, there was a mean increase in TEG R value in the kaolin trace compared to the heparinase trace, demonstrating that TEG is able to quantify functional anticoagulation. In the group that developed a DVT, there was no significant difference between R values of kaolin TEG and heparinase TEG, suggesting that these patients were not receiving adequate thromboprophlyaxis [113]. Performing simultaneous kaolin and heparinase TEGs appears to be a sensitive methodology for detecting evidence of anticoagulation with LMWH, and at the low doses used for prophylaxis is a better differentiator than antifactor Xa.

A recent prospective randomised control study of fixed dose prophylactic LMWH vs. no therapy administered for 1 year in 70 patients with advanced cirrhosis demonstrated that no patients in the enoxaparin group developed PVT compared with 17% in the control group. In addition, the incidence of documented bacterial infections was significantly lower in the enoxaparin group (8.8% vs. 33.3%). Surprisingly, there were no reports of haemorrhagic complications in the treated group [114]. This study raises interesting hypothesis as to whether LMWH act by improving intestinal microcirculation and thereby reduce the frequency of portal endotoxaemia. In addition, it is possible that systemic anticoagulation reduces the formation of intrahepatic microthrombi which are implicated in the progression of portal hypertension and parenchymal extinction over time [115]. The ability to monitor the efficacy and safety of anticoagulant therapy in patients with liver disease is becoming a real clinical dilemma, a challenge that could be met, in part, by using point of care viscoelastic tests of coagulation.

It is known that following liver transplantation, there can be a temporary hypercoagulable state, because of the imbalance between pro and anticoagulant systems and the post operative fibrinolytic shutdown [116]. It has been suggested that these haemostatic changes, as well as technical and surgical factors, may have a role in the early development of hepatic artery thrombosis (HAT) [117]. In an observational study of 298 liver transplant patients, high fibrinogen levels and low protein C levels were significantly associated with post-transplant thrombotic events [118]. The optimal anticoagulant regime in these patients is still an open question, and in the first week post transplantation using heparin with monitoring based on CCT still leads to significant bleeding complications in certain patients [108, 119]. Large scale prospective outcome studies are necessary to evaluate the impact of VET in managing thromboprophylaxis in these groups of patients. Anticoagulant therapy in patients with Budd Chiari syndrome (BCS) is also challenging and major bleeding, especially during invasive procedures, is common [120]. A recent case report of a complex patient with BCS and a TIPS occluded with thrombus, describes the use of TEG to guide the successful management of anticoagulant therapy and resultant recanalisation of the stent [121].

Conclusion

The complex haemostatic changes that occur in liver disease are difficult to assess using CCT. These tests are known to be poor predictors of bleeding risk and also, importantly, thrombosis. Consequently, the routine use of CCT to assess coagulation in patients with liver disease needs to be re-assessed. VET have been used for coagulation monitoring and to guide haemostatic therapy in liver transplantation for many years, but to date they have not been used to any great extent in hospitalised patients with liver disease. The summative information provided by these tests has the potential to be used in future clinical studies to determine a means of stratifying bleeding and thrombotic risk in these patients. It is clear that haemostasis is critically dependent on platelet number and function and fibrin clot formation, which are not evaluated by CCT. The current lack of randomised controlled trials of coagulopathy in liver disease is largely because of the inability to develop satisfactory surrogate end points in measuring coagulation. Global coagulation tests such as TEG/ROTEM could provide the basis on which to develop such criteria. The purpose of this review is to stimulate interest in producing the large prospective outcome studies that are needed to establish the clinical utility of VET in liver disease and to determine threshold values of VET that predict bleeding or thrombosis and thus optimise haemostatic and antithrombotic interventions.

Ancillary