Intraoperative hypercoagulability during liver transplantation as demonstrated by thromboelastography


Address reprint requests to Dominik Krzanicki, M.B.Ch.B., F.R.C.A., Department of Anaesthesia, Royal Free Hospital, Pond Street, London NW3 2QG, United Kingdom. Telephone: +44 (0) 20 7794 0500; E-mail:


Thrombotic complications are more common in liver disease than might be expected because of the coagulopathy described by conventional coagulation tests. Some of these complications may be life-threatening. The phenomenon of hypercoagulation is associated with complications in many populations, but the incidence in liver transplant recipients is unclear. We performed a retrospective database review of intraoperative thromboelastography (TEG) for 124 liver transplant recipients. We assessed the prevalence of hypercoagulation in this group and investigated the relative frequency of both shortened TEG reaction times (R times) and increased net clot strength (G) values. These findings were correlated with thrombotic complications. At the baseline, the prevalence of high G values was 15.53% on native TEG, and the prevalence of shortened R times was 6.80% on native-heparinase TEG. Patients with cholestatic pathologies had particularly high rates of hypercoagulation (42.9% with primary biliary cirrhosis and 85.7% with primary sclerosing cholangitis), but hypercoagulation was also common in patients with fulminant hepatic failure (50%) and nonalcoholic steatohepatitis (37.5%). There was a poor correlation between the TEG R time and the international normalized (INR), with 37.7% of TEG analyses demonstrating a short R time with an INR > 2. Six of the patients developed early hepatic artery thrombosis (5%); 3 of these patients had TEG evidence of high G values (P = 0.25), and 4 had short R times (not significant). In conclusion, intraoperative TEG evidence of high G values and short R times is relatively common in liver transplantation. It is unclear what bearing this condition has on thrombotic complications. Conventional coagulation tests have no ability to diagnose this condition. It is conceivable that such patients may come to harm if hypercoagulability is unrecognized and, therefore, inappropriately managed. Liver Transpl 19:852-861, 2013. © 2013 AASLD.

EPL, estimated percentage lysis; FFP

fresh frozen plasma


net clot strength


hepatic artery thrombosis


international normalized ratio; K, time from end of R until until amplitude of 20mm - speed of clot formation; LY30, percentage clot lysis 30 minutes following maximal amplitude


maximal amplitude


orthotopic liver transplantation


prothrombin time

R time

reaction time




measures the rapidity of fibrin build-up and cross-linking (clot strengthening)

Historically, orthotopic liver transplantation (OLT) was frequently complicated by massive blood transfusions. With developments in surgical and anesthetic techniques, the median blood loss associated with the procedure has fallen dramatically,[1] although there are still a number of patients who will require significant amounts of blood products intraoperatively.

There is emerging evidence that thrombotic complications are common in patients with both cirrhotic and noncirrhotic liver disease.[2, 3] Portal vein thrombosis is a common complication with an incidence of 10% to 15%, and although altered local flow dynamics probably play a large role, a relatively hypercoagulable state and a genetic prothrombotic predisposition may also be relevant.[4] The etiology of thrombosis in liver disease is multifactorial and includes flow obstruction and chronic inflammation, among other things.[5] Studies have reported an incidence of venous thromboembolism of 0.5% to 1.9% in patients with liver disease,[6] and this range represents a relative risk of 1.7 to 1.9.[7] Northup et al.[2] reported venous thromboembolism events in 0.5% of cirrhotic inpatients despite an elevated international normalized (INR). Coagulation changes in liver disease are complex. In liver disease, the hemostatic balance is precarious, and both endogenous and exogenous factors can readily tip the balance toward either a bleeding tendency or a prothrombotic state because these patients lack the buffering capacity of the large functional reserve with its associated regulatory mechanisms that is seen in healthy individuals.[8]

In the setting of liver disease and liver surgery, conventional coagulation tests give no information about where the balance of coagulation lies, and it has been shown that these tests can indicate hypocoagulability when, in contrast, global viscoelastic tests demonstrate an enhanced hemostatic capacity or hypercoagulability.[9] Coagulation monitoring using viscoelastic coagulation tests [eg, thromboelastography (TEG) and rotational thromboelastometry] has been used during liver transplantation for many years[10, 11]; however, less than 30% of centers use viscoelastic tests routinely.[12] Some patients with liver disease are at risk for hypercoagulability, and this may potentially be exacerbated during the transplant procedure.[13] Different aspects of the coagulation system have been implicated as responsible for this phenomenon; for instance, platelet hyper-reactivity has been found in many patients with cholestatic liver disease.[14] In cirrhosis, the ratio of the most powerful procoagulant and the most powerful anticoagulant in plasma (factor VIII and protein C, respectively) changes significantly, with the factor VIII level increasing and with the protein C level falling. This imbalance appears to be more significant with increasing disease severity.[15]

Thromboembolic events during OLT are associated with high morbidity and mortality rates[16] and have a multimodal etiology. A prospective study using transesophageal monitoring detected incidental intracardiac thromboemboli in 1.9% of OLT patients.[17] A review of case reports of cardiopulmonary emboli occurring during liver transplantation found that hypercoagulable viscoelastic coagulation tests were often temporally associated with these events.[18]

A demonstration of hypercoagulability on viscoelastic tests has shown to be associated with an increased risk of thromboembolic events, both arterial and venous, in general surgery patients[19] and also in critical care and trauma patients[20] and cardiac surgery patients.[21] However, the definition of hypercoagulability using viscoelastic parameters is not standard, and it has included a short reaction time (R time), an increased maximal amplitude (MA), a net clot strength (G) value, or a combination of these parameters.[20, 22]

The prevalence of preexisting hypercoagulability in patients presenting for liver transplantation is unclear; in addition, little is known about the de novo development of hypercoagulability during the procedure. We examined a cohort of patients undergoing liver transplantation to quantify and describe this issue. As a secondary aim, we examined the relationship between conventional coagulation tests and hypercoagulation and any relations with adverse thrombotic outcomes.


At our institution, intraoperative TEG (Haemonetics Corp., United States) is performed by dedicated trained operatives throughout liver transplantation according to a standard protocol.

Native and native-heparinase TEG panels were performed at the baseline and during the dissection, anhepatic, and reperfusion stages. Additional TEG panels were performed at the discretion of the attending clinician (eg, in cases of severe hemorrhaging). When more than 1 TEG panel was performed during a particular stage, only the first panel from that stage was used for the analysis. This was in conjunction with point-of-care INRs (Hemochron Signature Elite, ITC, United States), full blood counts (PocH-100i, Sysmex Europe), and arterial blood gas analysis (RapidLab 1265, Siemens AG, Germany). Standard postoperative thromboprophylaxis involved low-molecular-weight heparin administration once the INR was <1.5.

We performed a retrospective analysis of our database for an 18-month period from 2008 to 2010. All point-of-care and TEG data were analyzed and compared.

In advance of the analysis of these data, advice was sought from the local ethics committee, which advised that formal institutional approval was not required because these were anonymized data routinely collected in the liver transplant database and because all patients had consented a priori to data collection for research purposes when they had consented to liver transplantation.

TEG generates a number of variables from a sample of blood as it clots (Fig. 1). MA data from TEG are a reflection of platelet function, fibrinogen levels, and the interaction between platelets and the coagulation system. All TEG MA data were converted to their respective G values before the analysis (a mathematical transformation):

display math

where G is a unit of force. G allows a direct linear comparison of 2 different values, whereas MA does not. Therefore, a G value of 10,000 dyne·cm−2 would reflect a clot twice as strong as one with a G value of 5000 dyne·cm−2. A 2-fold increase in MA does not equate to a 2-fold increase in the clot strength because it does not represent a linear relationship. A clot with a high tensile strength is more resistant to both mechanical and enzymatic degradation. By investigating TEG R times across the database, we also analyzed plasmatic (enzymatic) coagulation. The R time represents the time to initial fibrin formation, and a short R time indicates an accelerated rate of fibrin formation.

Figure 1.

TEG tracing and reference ranges for TEG parameters. Image of the TEG® Thromboelastograph® Haemostasis Analyzer tracing used by permission of Haemonetics Corporation.

Because of the significant effects of endogenous heparin-like substances on native TEG, particularly at reperfusion, all calculations were performed for both native and native-heparinase TEG panels.[23] We identified all patients who met the high G value criteria (G > 7100 dyne·cm−2) and had shortened R times (<12 minutes), and we further characterized their underlying pathologies. TEG data were also compared to the results of conventional clotting tests and platelet counts. We reviewed patients' radiological case notes for the detection of thromboembolisms (deep venous, pulmonary, and hepatic arterial/venous) in the 30 days following transplantation.

Intraoperative transfusion data were available for all patients. At our institution, the transfusion of blood and blood products is based on an algorithm using TEG parameters and point-of-care full blood counts. Red cell transfusions are given at a hemoglobin level of 8 g/dL or less.

All results were handled and analyzed with Microsoft Excel 2008 for Mac; the statistical analysis was performed with free statistics software (version 1.1.23-r7) at


One hundred twenty-four consecutive liver transplant operations (between 2008 and 2010) were included in the study. This reflected 117 patients, including 7 (6%) who underwent retransplantation within the study period. The median Model for End-Stage Liver Disease score was 15 (range = 6-39) at the time of listing (not weighted for hepatocellular carcinoma).

Within the regraft group, 1 procedure was performed within 2 days of the original transplant for primary graft nonfunction, 4 were performed between 3 and 34 days for hepatic artery thrombosis (HAT), and 2 took place at various time points for graft rejection. In all, 784 separate TEG panels were identified in the database, and this number reflected a mean of 6.3 panels per transplant. One hundred eight of the 784 TEG panels were performed at the baseline, 258 were performed during the dissection phase, 130 were performed during the anhepatic period, and 288 were performed after reperfusion.

Underlying Disease

Indications for liver transplantation are outlined in Table 1. The most common etiologies were alcoholic liver disease and viral hepatitides. Cholestatic pathologies accounted for the largest number of the remainder of the transplants.

Table 1. Etiologies of Liver Disease in the Study Population
EtiologyFrequency [n (%)]Concurrent Hepatocellular Carcinoma [n (%)]a
  1. a

    The percentages are based on the frequency n values.

Alcoholic liver disease20 (16.1)1 (5)
Alcoholic liver disease plus hepatitis B or C16 (12.9)4 (25)
Amyloidosis3 (2.4)
Autoimmune hepatitis1 (0.8)
Epithelioid tumor1 (0.8)
Fulminant hepatic failure6 (4.8)
Hepatitis B5 (4.0)3 (60)
Hepatitis C26 (21.0)12 (46)
Hepatitis B plus hepatitis C2 (1.6)2 (100)
Nonalcoholic steatohepatitis8 (6.5)2 (25)
Primary biliary cirrhosis7 (5.6)
Polycystic kidney disease2 (1.6)
Cryptogenic2 (1.6)
Drug-induced1 (0.8)
Primary hepatocellular carcinoma1 (0.8)
Nodular regenerative hyperplasia1 (0.8)
Regraft for HAT4 (3.2)
Regraft-other3 (2.4)
Oxalosis1 (0.8)
Primary sclerosing cholangitis14 (11.3)2 (14)

A substantial proportion of OLT procedures (21%) were performed for patients with concurrent hepatocellular carcinoma.


High G Values

Overall, 11.2% of the native TEG G values and 13.1% of the native-heparinase TEG G values were above the upper reference range of G (7100 dyne·cm−2) at some stage of the procedure. This means that 27.4% (34/124) of the patients had a high G value on at least 1 native TEG trace during OLT, and 30.6% (38/124) had a high G value on at least 1 native-heparinase TEG trace during OLT.

Shortened R Times

Overall, 19.1% and 20.3% of the native and native-heparinase TEG R times, respectively, were below the lower reference range for R times (12 minutes) at some stage of the procedure: 59.7% of the patients had a shortened R time on at least 1 native TEG trace during OLT, and 61.3% of the patients had a shortened R time on at least 1 native-heparinase TEG trace during OLT.

As described previously, when more than 1 panel was performed during a particular stage for a particular patient, only the first panel was used for the analysis in this article (unless otherwise specified).



The mean values were compared for native and native-heparinase TEG tracings. The mean R times were 20.22 and 20.61 minutes for the native and native-heparinase tracings, respectively [P = 0.64 (Student t test)]. The mean G values were significantly higher for the native-heparinase TEG tracings (5322 dyne·cm−2) versus the native tracings [4613 dyne·cm−2, P = 0.001 (Student t test)]. This was reflected in the higher incidence of high G values on native-heparinase TEG (20.39%) versus native TEG (15.53%; Table 2). There was a 6.80% to 10.68% prevalence of short R times in patients presenting for OLT.

Table 2. Native and Native-Heparinase TEG Parameters at the Baseline
TEG TypeParameter%Median (Minutes)Range (Minutes)
  1. NOTE: Bolded data denote an enhanced hemostatic potential. Percentages reflect the number of patients with the characteristic.

NativeShort R time10.6810.28.6-11.9
Normal R time71.8417.112-25.8
Long R time17.4833.5527.3-58.8
Native-heparinaseShort R time6.8010.37.9-11.3
Normal R time77.671812.1-26
Long R time15.5329.4526.4-55.4
TEG TypeParameter%Median (dyne·cm−2)Range (dyne·cm−2)
NativeHigh G value15.538986.77165-16,097
Normal G value45.634505.73291-7048
Low G value38.832407.99122-3169
Native-heparinaseHigh G value20.399970.067106-19,752
Normal G value54.374363.333116-6904
Low G value25.242283.51117-3077


The prevalence of high G values peaked during dissection: 18.49% of native traces and 20.87% of native-heparinase traces demonstrated this characteristic.

There was a larger increase in the frequency of short R times in patients during dissection: 22.69% on native TEG and 17.39% on native-heparinase TEG (Table 3).

Table 3. Distributions of Normal and Abnormal TEG Parameters by the Stage
TEG TypeParameterDissection (%)Anhepatic (%)Reperfusion (%)
  1. NOTE: Bolded values are related to those reflecting an enhanced hemostatic potential. Percentages refer to patients with the characteristic.

NativeShort R time22.6929.463.31
Normal R time68.9166.9622.31
Long R time8.403.5774.38
Native-heparinaseShort R time17.3928.5717.50
Normal R time74.7867.8670.83
Long R time7.833.5711.67
NativeHigh G value18.498.044.96
Normal G value57.9851.7923.14
Low G value23.5340.1871.90
Native-heparinaseHigh G value20.878.049.17
Normal G value61.7456.2544.17
Low G value17.3935.7146.67


During the anhepatic stage, short R times peaked: 29.46% on native TEG and 28.57% on native-heparinase TEG. During this stage, the prevalence of high G values was low at 8.04% on both native and native-heparinase tracings (Table 3).


At reperfusion, the endogenous heparinoid effect was clearly visible, with only 3.31% of native traces showing a short time and with 74.38% of patients showing a significant heparin-like effect with prolonged R times reversed on the native-heparinase trace. The native-heparinase tracings possibly suggested an ongoing enhanced hemostatic potential, with 17.50% of patients displaying a short R time. The prevalence of high G values was lowest immediately after reperfusion (Fig. 2).

Figure 2 demonstrates the changes in TEG parameters by stage.

Figure 2.

Graph showing the change in the frequency of prothrombotic native and native-heparinase TEG characteristics by the stage of transplantation.

Laboratory and Transfusion Data

The baseline (preoperative) hematological and clotting data for the cohort are described in Table 4. The group had mild thrombocytopenia with a prolonged INR and anemia. The mean fibrinogen level was within the normal range.

Table 4. Hematological Parameters and Transfusion Requirements According to the G Values at the Baseline
Laboratory TestsAll PatientsNormal or Low G ValueHigh G ValueP Valuea
  1. NOTE: The data are presented as medians and interquartile ranges.

  2. a

    Mann-Whitney U test.

Hemoglobin (g/dL)9.8 (8.4-11.3)9.9 (8.8-11.6)8.9 (7.5-10.8)0.02
Platelets (×109/L)86 (54-128.8)71.6 (51.5-103)139.5 (99.5-181.5)<0.001
INR1.55 (1.3-1.8)1.60 (1.3-1.825)1.45 (1.3-1.625)0.52
Fibrinogen (g/L)2 (1.5-3)1.9 (1.5-2.7)3.2 (2.05-4.13)0.007
Transfused ProductsAll PatientsNormal or Low G ValueHigh G ValueP Valuea
Packed red cells (U)3 (0-6)3 (1-6)3 (0-8)0.9
FFP (U)3 (0-6)4 (2-6)1 (0-4.3)0.05
Platelets (U)1 (0-2)1 (0-2)0 (0-2.3)0.27

The median hemoglobin level was lower in the high G group (8.9 versus 9.9 g/dL, P = 0.02), whereas the platelet and fibrinogen levels were significantly higher.

There was no significant difference in the baseline INR.

Packed red cell transfusion volumes were equivalent for patients with and without high G values (on native-heparinase TEG). Fresh frozen plasma (FFP) transfusions showed a tendency toward lower volumes in the group with higher G values [P = 0.05 (Mann-Whitney U test)]. The median volumes of platelet transfusions were 0 U for the high G group and 1 U for the low/normal G group [P = 0.27 (Mann-Whitney U test)].

When we compared the likelihood of no transfusion versus any transfusion, there was no difference between the groups for red blood cell transfusions [75.5% for normal G values versus 65% for high G values, P = 0.33 (χ2)], but there was a lower chance for transfusions of FFP [75.5% for normal G values versus 50% for high G values, P = 0.02 (χ2)] and platelets [59.6% for normal G values versus 35% for high G values, P = 0.04 (χ2)].

Etiology of Liver Disease

The phenomenon of high G values was not evenly distributed by etiology. Table 5 shows patients who had prothrombotic TEG results on any tracing during the procedure according to etiology. Patients with cholestatic pathologies (primary sclerosing cholangitis and primary biliary sclerosis) had high rates of G values above the reference range (85% and 43%, respectively) and also shortened R times. The incidence was also high among patients with fulminant hepatic failure (50%) and patients undergoing regrafting for HAT (50%), although the numbers of these patients were small. Patients with viral or alcoholic liver disease had a different pattern: between 65 and 100% had short or hypercoagulable R times, but only 10–12% had an increased G value. Only 3 of the 26 patients who had concurrent hepatocellular carcinoma had high G values on TEG (median native G value = 4108 dyne·cm−2, native R time = 16 minutes). In contrast, all other etiologies had similar incidences of both shortened R times and increased G values.

Table 5. Prevalence of Intraoperative Evidence of an Enhanced Hemostatic Potential
DiagnosisCases (n)Cases With a Short R Time [n (%)]aMedian Baseline R Time†Cases With a High G Value [n (%)]aMedian Baseline G Value
  1. a

    Cases with an abnormal parameter on any intraoperative TEG examination are listed.

  2. b

    The data are presented as medians and interquartile ranges. Values are provided only when the number of cases was greater than 3.

  3. c

    The other category comprises of etiologies where only two or fewer cases were found in the series.

Primary sclerosing cholangitis1410 (71.4)17.8 (14-19.2)12 (85.7)8889 (5417-11234)
Primary biliary cirrhosis73 (42.9)16.8 (14.6-21.7)3 (42.9)6236 (3026-7658)
Fulminant hepatic failure63 (50)21 (18.4-26.8)3 (50)2631 (1174-5450)
Alcoholic liver disease2013 (65)16.8 (14.1-19.5)2 (10)
Hepatitis C2618 (69)17.7 (15.2-27.4)4 (15.4)3526 (2331-4704)
Alcoholic liver disease plus hepatitis B and/or C1611 (68.8)18.7 (13.8-23.3)2 (12.5)
Nonalcoholic steatohepatitis83 (37.5)17.3 (15.5-18.5)3 (37.5)4074 (3375-6069)
Regraft for HAT42 (50)2 (50)
Regraft-other32 (66.7)2 (66.7)
Amyloidosis31 (33.3)1 (33.3)
Hepatitis B55 (100)21 (18.4-26.8)1 (20)
Other124 (33)4 (33)
Total transplants12475 (60.5)39 (31.5)

Therefore, there appears to be a different distribution of the nature of hypercoagulability that is dependent on the etiology of liver disease.

Hypercoagulation by Conventional Clotting Tests

A comparison of paired G values with point-of-care INR tests showed that there was no significant correlation between the 2 parameters [r = −0.33, P = 0.001 (Spearman's rank correlation)]. Figure 3 shows a scatter plot of the 2 measurements and reveals that TEG traces that were hypercoagulable (ie, above the normal reference range) could be associated with an INR between 0.9 and 3.8.

Figure 3.

x-y scatter plot of all native-heparinase TEG G values against corresponding INR values.

G values were compared with platelet counts, and a moderate correlation was found [r = 0.62, P = 0.001 (Spearman's rank correlation)].

None of the patients with high G traces had platelet counts above the normal reference range.

There was no correlation between the R time and INR [r = 0.04, P = 0.27 (Spearman's rank correlation)].

One hundred sixty-two of the 784 TEG R time measurements were found to be shorter than the normal range (12-26 minutes) with a median time of 8.95 minutes. In this group, the median INR was 1.8 with a range of 1.1 to 10. On the basis of INR values, 128 TEG analyses (79.0%) showing a shortened R time would be described as coagulopathic (INR > 1.5), and 61 (37.7%) would be described as markedly so (INR > 2).

Perioperative Thrombotic Events

One patient had an intraoperative portal vein thrombosis that required on-table thrombectomy. This was associated with a grossly shortened R time (2.6 minutes) on both native TEG and native-heparinase TEG during early reperfusion. The concurrent G value was within the normal range at 3347 dyne·cm−2. There were no intraoperative pulmonary emboli identified by the attending clinicians.

The database review identified thrombotic complications within 30 days of transplantation.

Among the 117 primary transplants (ie, with the exclusion of the 7 regraft procedures), there were 6 cases (5%) of HAT. Four of the 6 required a regraft procedure as a result of HAT. Three of the 6 cases had high G traces during their initial transplants [P = 0.25 (χ2)], and 4 of the 6 cases demonstrated shortened R times [P = 0.79 (χ2)].

The underlying etiologies for those patients undergoing regrafting after HAT were primary sclerosing cholangitis,[2] alcoholic liver disease, and hepatitis B cirrhosis. The etiologies for the 2 patients who did not undergo regrafting were amyloidosis and hepatitis C virus. It is notable that the occurrence of hypercoagulable traces in patients who developed HAT was more frequent than that in the general population of patients undergoing liver transplantation.

There were only 2 postoperative pulmonary emboli in the primary transplant cohort (1.7%).


This study, which we believe is the first to examine such a large number of intraoperative data sets, demonstrates that a significant number of patients with end-stage liver disease undergoing liver transplantation present with or develop hypercoagulable thromboelastograms during the procedure. In this series, the incidence of patients with baseline G values greater than 7100 dyne·cm−2 was 20.39% on heparinase TEG and 15.53% on native TEG.

Patients with cholestatic diseases (primary sclerosing cholangitis and primary biliary cirrhosis) had a high incidence of high G values (85.8% and 42.9%, respectively). Ben-Ari et al.[24] also found a high prevalence of hypercoagulability in these patients (43% and 28%, respectively), although they used a composite definition of an increased MA, an increased α angle (a measure of the rapidity of fibrin build-up and cross-linking (clot strengthening)), and a short R time to define hypercoagulability. There was no correlation with either platelet counts or fibrinogen levels, which were all within the normal ranges. Notably in their series, only 5% of patients with noncholestatic cirrhosis met their definition of hypercoagulability, and no healthy volunteers were found to be hypercoagulable. Although there were only 6 patients with acute liver failure, 50% of these demonstrated high G values. This result may be surprising because all these patients had an INR of 2 or more, but it is compatible with similar findings reported recently.[25, 26] Nearly 40% of patients with nonalcoholic steatohepatitis had evidence of both short R times and excessive clot strength (increased G values). It has been found that these patients have an increased platelet contribution to clot strength and reduced inducible fibrinolysis.[27] In addition, the ongoing inflammatory state associated with nonalcoholic liver disease leads to low-level activation of the coagulation system.[28]

Many intraoperative factors may contribute to the persistence or de novo development of hypercoagulation. These include vascular stasis, coagulation activation by endotoxins (within the portal vein), and endothelial cell injury and local inflammation of the graft organ caused by ischemia/reperfusion injury. Models of endotoxemia have shown that this results in significant shortening of the R time or clotting time with accelerated initiation of coagulation,[29] but the results of conventional coagulation tests [prothrombin time (PT)/INR)] remain prolonged.[30] Ischemia/reperfusion injury leads to the activation of coagulation and also platelets,[31] and it is notable that reports of intracardiac thrombi are most common around the time of reperfusion. In a review of 27 case reports of thromboembolic events occurring during liver transplantation, TEG profiles were hypercoagulable in more than 70% of cases, whereas all conventional coagulation tests indicated hypocoagulability.[18] In addition, most patients undergoing liver transplantation do not routinely undergo thrombophilia screening, but a genetic component involving prothrombotic gene polymorphisms may be present in some.[32] No patient in our series had an intraoperative event suggestive of intracardiac thrombosis or pulmonary embolism.

In this series, the incidence of HAT was 5%. Although its etiology is known to be multifactorial and significantly related to difficulties associated with the arterial anastomosis and arterial reconstruction, it is notable that 4 of the 6 patients who developed HAT had TEG evidence of enhanced coagulability at some point during the procedure. This is a higher incidence than that in the general population of liver transplant patients, and there may be some causal association, especially in the context of a difficult reconstruction and altered blood flow. With the relatively low incidence of HAT and corresponding small number of cases, this study is not powered to detect a statistically meaningful relationship, and a type II error cannot be excluded. It would seem physiologically plausible that such a relationship could exist, but further research is required to demonstrate a conclusive link.

The clinical implications of hypercoagulability occurring during liver transplantation have yet to be evaluated. However, emerging evidence suggests that hypercoagulability detected by viscoelastic testing puts certain patients in an at-risk group for both venous and arterial thrombotic events.[19, 20] A recent systematic review of 10 studies in surgical patients showed an increased MA to be the most important parameter for predicting postoperative thrombo-embolic events. However, there was considerable variability in which parameters were used to define hypercoagulability, and no study was adequately powered. Nevertheless, the vast majority of patients who had a thrombo-embolic event were hypercoagulable according to 1 or more TEG parameters,[33] and future prospective studies are recommended. It has also been shown that TEG can display hypercoagulation in thrombosis-prone patients.[34]

We showed that patients with high G values on TEG, despite a baseline hemoglobin level that was 1 g/dL lower, received similar median red cell transfusions, and this suggests that intraoperative blood loss was lower in this group. These patients were also significantly less likely to receive an FFP transfusion even though there was no statistical difference in INR between the groups, presumably because FFP transfusions at our center are based on TEG.

The definition of hypercoagulation based on viscoelastic testing is not standard. Different researchers have chosen to use different parameters or a combination of more than 1 parameter. The G value is a function of MA and represents clot strength rather than initial clot formation kinetics; as such, it is used as the definition of hypercoagulation in this article. The ability to directly compare G values with one another linearly enables a better appreciation of the magnitude of the change between different results. Plasmatic (or enzymatic) coagulation is also likely to contribute to a prothrombotic state, by which initial clot formation is accelerated (as represented by a short R time).

Conventional coagulation tests do not provide information about the quality of the clot or the dynamics of its formation. Unless the platelet count or fibrinogen levels are elevated above normal values (which was not the case for any of our patients), conventional tests are unable to identify a hypercoagulable state. It is clear from our data that a high INR does not exclude the possibility that a patient has a prothrombotic tendency and TEG evidence of enhanced coagulability. It has consistently been shown that there is only a weak correlation between the R time/clotting time and PT/INR.[35, 36] Another problem is that the INR value varies between laboratories in patients with liver disease.[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 polymerization, which happens at very low levels of thrombin generation of approximately 10 to 20 nM (<5% of the total thrombin that can be generated).[38]

Wide derangements in INR may not represent a defect in coagulation as assessed by TEG criteria; indeed, this reflects the fact that INR has poor predictive ability for clinically important bleeding.[39] In liver disease, the endogenous anticoagulants as well as procoagulant factors are all reduced, and the balance of procoagulants to anticoagulants may be altered in favor of a prothrombotic state, but the INR reflects primarily procoagulant activity.[40] It is increasingly recognized that patients with cirrhosis are at risk of thrombotic complications as a result of low endogenous anticoagulant levels.[41] Viscoelastic monitoring adds valuable qualitative information to the management of these cases.[42]

What remains unclear is what action should be taken when hypercoagulability is demonstrated on TEG. When there is evidence of an excessively shortened R time and a normal or high MA or G value, then it may be reasonable to give a small intravenous dose (3000-5000 U) of heparin (Andre DeWolf MD, oral communication, May 2006). It would also seem prudent to avoid prohemostatic agents, including FFP, platelets, and antifibrinolytics, when there is TEG evidence of hypercoagulability.

The limitations of this study are its retrospective design and the fact that it is not adequately powered to detect thrombotic events such as HAT and venous thromboembolic events, which have reported incidences of 5% and 1.5%, respectively. In addition, the study did not extend into the postoperative period, and as such, it is difficult to comment on persistence of the phenomenon.

The definitions of hypercoagulability vary in different publications. We have defined hypercoagulability as TEG parameters outside the normal range, whereas others have used values of 2 or more standard deviations.[20]

Most publications that have looked at the phenomenon of hypercoagulability have done so in cardiac surgery patients, trauma patients, or general surgery patients, all of whom are known to have a higher incidence of both hypercoagulability and thrombotic events. Establishing a normal reference range for TEG values in patients with end-stage liver disease is problematic because there is significant variability according to the etiology of the underlying disease process and the stage of the disease.[36]

We used both native and native-heparinase TEG analysis during liver transplantation. Native TEG is extremely sensitive to the presence of heparin and heparin-like substances.[43] In recent years, there has been increasing interest in the detection and significance of endogenous heparins. Under conditions of endothelial stress such as surgery and sepsis, the endogenous release of very small quantities of glycosaminoglycans can be detected systemically.[44] When there is more significant damage to the vascular endothelium (eg, from ischemia/reperfusion injury), systemic activation of coagulation is promoted, and it is thought that this shedding of glycosaminoglycans into the circulation is an adaptive response to keep a progressively more procoagulant microvasculature open by means of endogenous heparinization.[45] When shed, the glycocalyx glycosaminoglycans retain their anticoagulant activity, and this is detectable by a prolonged R time on TEG analysis.[46] In this series, there were only minor differences in R times and G values between native and native-heparinase TEG traces with the exception of early reperfusion TEG, which showed a pronounced heparin-like effect with prolonged R times and significantly lower G values in the native TEG tracings.

We feel that on the basis of the presented results, more research should be aimed at investigating the phenomenon of hypercoagulability within liver transplantation and chronic liver disease. The need for further work in this area has been highlighted,[47] and this work contributes to that need. Crucially, an effort needs to be made to ascertain the presence or absence of a causal relationship between this state and thrombotic events and indeed ultimate posttransplant outcomes.

We suggest, however, that because hypercoagulability is not detected by conventional testing and can be present when these tests indicate hypocoagulation, viscoelastic testing should be used routinely for coagulation monitoring during liver transplantation. In addition to identifying coagulopathy early and allowing specific hemostatic therapy to be instituted, TEG may be crucially valuable in preventing unnecessary and potentially harmful transfusions of blood products into patients who are hypercoagulable on viscoelastic tests and at increased risk of developing thromboembolic complications.