Rotation thromboelastometry velocity curve predicts blood loss during liver transplantation

Transcription

1 British Journal of Anaesthesia, 117 (6): (2016) doi: /bja/aew344 Cardiovascular Rotation thromboelastometry velocity curve predicts blood loss during liver transplantation L. A. Tafur 1, P. Taura 1, *, A. Blasi 1,2, J. Beltran 1, G. Martinez-Palli 1,2, J. Balust 1, and J. C. Garcia-Valdecasas 2 1 Department of Anaesthesiology, Liver Transplant Unit, Hospital Clinic, Barcelona, Spain, 2 Institut d Investigacions Biomédiques Agustı Pi i Sunyer and 3 Department of Surgery, Liver Transplant Unit, Hospital Clinic, Barcelona, Spain *Corresponding author. ptaura@clinic.ub.es Abstract Background. Patients undergoing liver transplantation (LT) have a high risk of bleeding. The goal of this study was to assess whether the first derivative of the velocity waveform (V-curve) generated by whole blood rotation thromboelastometry (ROTEM VR ) can predict blood loss during LT. Methods. Preoperative V-curve parameters were retrospectively evaluated in 198 patients. Patients were divided into quartiles based on blood loss: low (LBL) in the first quartile and high (HBL) in the higher quartiles. A subgroup analysis was performed with patients stratified according to cirrhosis aetiology. A logistic regression model and receiver operator characteristics (ROC) curve were used to test the capacity of the V-curve, to discriminate between LBL and HBL. Results. In the HBL group, the V-curve showed a lower maximum velocity of clot generation (MaxVel), a lower area under maximum velocity curve (AUC), and a higher time-to-maximum velocity (t-maxvel) than in the LBL group. t-maxvel was the only parameter showing a capacity to discriminate between the two groups, with a ROC area of 0.69 (95% CI; ). The ROC area was 0.78 (95% CI; -0.83) for the 148 patients with cirrhosis, 0.73 ( ) for patients with viral hepatitis and 0.83 ( ) for patients with alcoholic hepatitis, the group that showed the best discriminative capacity. Moderate but significant correlations were found between all parameters of V-curve and BL. Conclusions. Pre-transplant V-curve obtained from ROTEM is a promising tool for predicting BL risk during LT, particularly in patients with cirrhosis. Key words: clot formation; liver transplantation; thrombin generation; viscoelastic assays Liver transplantation (LT) is associated with blood loss as a result of pre-existing anatomical and physiological abnormalities, particularly in the haemostatic system of patients with end-stage liver disease. Extensive research has been done on the capacity of routine plasma-based tests to predict blood product requirements and bleeding during LT but conclusions are conflicting. 1 3 These tests inform on the initiation phase of clotting but not the haemostatic capacity, because plasma clots as soon as 5% of thrombin is activated, so they do not reflect the complexity of the haemostatic state. Limitations of these tests for predicting blood loss and transfusion make it advisable to use global haemostasis assays, that can better reflect compensatory effects within the system of patients undergoing LT. 4 Haemostasis is a complex process affected by many factors including both cellular and plasma components. The evaluation of plasma thrombin generation (TG) in platelet poor plasma Editorial decision August 28, 2016; Accepted: September 19, 2016 VC The Author Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please journals.permissions@oup.com 741

2 742 Tafur et al. Editor s key points Rotational thromboelastometry is a viscoelastic whole blood coagulation monitor that might be useful in predicting blood loss. ROTEM VR velocity waveforms were retrospectively compared with blood loss in 198 patients undering liver transplantation in a single centre. Preoperative velocity waveform parameters, which reflect the kinetics of fibrin clot formation in whole blood, could distinguish between patients with low or high risk for blood loss. (PPP) or platelet rich plasma (PRP), has become the preferred approach for assessing coagulation and thus evaluating the haemostatic capacity of the system. 5 6 Recent studies have reported that TG assays can predict bleeding risk and transfusion in major surgery and trauma patients, 7 8 but they are timeconsuming, making this method unsuitable for routine analysis of haemostasis. Alternatively, thromboelastography-based evaluation of haemostasis is a simple procedure, that records physical viscoelastic characteristics of whole blood coagulation and can provide information on initiation, development and final clot strength and fibrinolysis, generating a graphic representation of the fibrin polymerization process. Thromboelastography data, processed using commercially available software, provide derived parameters that allow the rate of development of tensile strength to be measured. The first derivative of the clot growth velocity curve (V-curve), based on changes in maximum amplitude (clot strength), was first described by Sorensen and showed a "profile similarity" to TG curves. 9 Unlike plasma based coagulation assays, viscoelastic assays provide information of fibrin polymerization under oscillatory shear stress, and though fibrin-based clot strength depends on several factors (platelets, fibrinogen, FXIII and TG), under oscillation the major signal comes from platelet contractile forces. 10 As a result, the kinetics of clotting is highly dependent on platelet concentration. In healthy individuals fibrin clot formation, structure, rate, stability, and strength depend on the pattern, the rate and the peak of TG, but in patients with cirrhosis, given the nature of the coagulation disturbances, many questions remain as potential compensatory mechanisms exist (rebalanced haemostasis). 15 As a result of a close relationship between TG and fibrin formation, the results are often considered interchangeable in research/clinical settings but they do not always go hand-inhand and both assays should be viewed as complementary. Given that the fibrinogen/fibrin cross-link plays a key role in forming a stable clot, which is essential to achieve effective haemostasis, the first objective of this retrospective study was to assess whether the kinetics of fibrin formation, the first derivative of the waveform (V-curve) generated by whole blood clotting in rotationed thromboelastometry (ROTEM VR ), could predict bleeding risk during LT. The second aim was to analyse the capacity of V- curve tracking for predicting bleeding for different cirrhosis aetiologies. Methods Study population The Institutional Review Board approved this study. The medical records of all consecutive patients who underwent LT at the Hospital Clinic, Barcelona between January 2012 and December 2014 were analysed. Exclusion criteria were repeat transplants, multiorgan transplantation, anticoagulation treatment and incomplete patient data. The same standardized surgical and anaesthetic protocols were used to manage all patients. Haemodynamic and blood coagulation monitoring and transfusion policy followed a previously described protocol. 16 Briefly, maintenance of a low central venous pressure (CVP) by restrictive fluid therapy, and coagulation parameters were not corrected preoperatively or intraoperatively in the absence of uncontrollable bleeding. Fibrinogen was administered when concentrations were 1g l 1,andfreshfrozenplasmaifmicrovascular bleeding was detected, provided that fibrinogen concentration was >1g l 1 and platelets were > l 1,but we did not attempt to correct the international normalized ratio (INR). The final haemoglobin threshold was 8 g dl 1. Solutions of 5% or 20% albumin balanced with crystalloids were administered during LT. Preservation of the inferior vena cava (piggyback technique) with a temporary porto-cava shunt was performed in all patients. Rotation thromboelastometry Point-of-care ROTEM VR (TEM International GmbH, Munich, Germany) was used to monitor coagulation. In this study, the coagulation process was assessed using the extrinsically activated assay with recombinant tissue factor (EXTEM) according to the manufacturers instructions. All measurements were handled in the same manner: blood (280 ml) was transferred to a polypropylene tube containing 0.5 ml of 3.2% sodium citrate; storage time was 15 min, and the sample was kept prewarmed (37 C). Assessment of whole blood clot formation was based on standard measurements that had been prospectively recorded in a computerized database, and among the parameters that were recorded we report on the following: clotting time (CT), the time from the start until initiation of clotting, and maximum clot firmness (MCF), the maximal amplitude of the tracing. ROTEM raw data were processed using CalcuRo software (Pentapharm GmbH, Munich, Germany) to provide derived parameters of dynamic velocity of clot fibrin formation. First derivative waveform (V-curve) imported from ROTEM Parameters that were assumed to concur with the propagation of whole blood clot formation, such as maximum velocity (MaxVel, 100*mm s 1 ) and time to reach maximum velocity (t-maxvel, s), were obtained (Fig. 1). The area under the velocity curve (AUC) was also recorded. On the basis of acceleration law 17 (reaction rates increase when reactant concentrations are increased), we calculated MaxVel/t-MaxVel ratio (mv/t-m) expressed in mm s 2, and hypothesized that if there is a high rate, the clot firmness will be greater, so this variable could be better correlated with bleeding. Blood loss monitoring The volume of blood loss was obtained from the surgical records, which had been prospectively recorded in a computerized database. To avoid underestimating the blood loss, nurses carefully quantified blood loss by weighing sponges and collecting suctioned blood in a volumetric container. Considering the possible influence of graft reperfusion on blood loss as the result of the ischaemic insult (metabolic events and the inflammatory response displaying

3 ROTEM velocity-curve predicts blood loss t-maxvel (s) 8 MaxVel (mm/s) CT AUC Fig 1 Principal characteristics of thromboelastometry parameters and first derivative parameters of the waveform (V-curve) generated by clotting of whole blood clot formation. CT (s) is the clotting time, MCF (mm) is the maximum amplitude of clot formation, MaxVel (mm s 1 ) is the maximum velocity of clot formation, t-maxvel (s) is the time to reach the maximum velocity of clot formation, and AUC represents the area under the velocity curve and indicates the maximum clot formation, an indirect measurement of clot strength. hypocoagulable features) we circumscribed this analysis at two fixed times: a) from start of surgery to just before graft reperfusion (P1), and b) from start to end of surgery, or total blood loss (P2). Clinical and underlying cirrhosis aetiologies and donor characteristics were recorded. ROTEM analysis just before surgery to assess the patients haemostatic profile was also prospectively recorded. Statistical analysis Categorical data are expressed as numbers and percentage and quantitative variables as median and interquartile range (IQR). The study population was divided into quartiles. As in previous studies, we defined 2 groups of subjects based on high vs low blood loss before graft reperfusion: the first quartile value of blood loss volume was used as the cut-off value. The v 2 test was used to compare qualitative variables and Student s t-test or Mann-Whitney U-test for independent samples to compare median values. Receiver operating characteristic (ROC) curves were constructed and the area under the ROC curve was calculated to assess the predictability of the V-curve parameters to discriminate between the two groups. A subgroup analysis based on cirrhosis aetiology, alcohol vs virus, was performed in patients with cirrhosis because of a different behaviour on primary haemostasis. 18 The Spearman rank correlation was calculated in order to show the association between blood loss as a continuous variable and V-curve parameters. Statistical analysis was performed with SPSS for Windows 16.0 (SPSS, Inc., Chicago, IL, USA). P 0.05 was considered to indicate statistical significance. Results MCF A total of 198 patients were eligible for our analysis (144 men and 54 women, age 34 to 68 yr). Preoperative clinical and laboratory data for all patients are reported in Table 1. The correlation of the Child-Pugh-score, taken as index of illness severity, vs MaxVel or t-maxvel was (P < 1) and (P < 1), respectively. Platelet count and fibrinogen correlated with MaxVel (8, P < 1 and 0.485, P < 1, respectively) and MCF (0.562, P < 1 and 4, P < 1, respectively), showing a weak correlation with t-maxvel ( and , respectively). Linear regression with the logarithm of MaxVel to assess the magnitude of change according to platelet count and fibrinogen, showed that platelet count could determine a greater extent of MaxVel values, with a significant impairment for platelets < 82x10 9 l 1 and fibrinogen < 2.1 g l 1. None of the standard coagulation parameters correlated with CT, AUC or the mv/t-m ratio. We divided the study population into quartiles according to the amount of bleeding; Q1 [661 (205) ml]; Q2 [1201 (164) ml], Q3 [1889 (219) ml], and Q4 [3087 (720) ml]. Low blood loss (LBL) was defined as a value in the first quartile (961 ml), and high blood loss (HBL) as a value above the first quartile (>961 ml); the mean blood loss was 661 (205) ml in LBL group and 2008 (874) ml in the HBL group. Based on this sample size, the power to detect a clinically relevant increase in mean blood loss of at least 50% (from 661 ml to 991 ml) was higher than 90%. There were no significant differences between the two groups in patient characteristics composition, aetiology, previous abdominal surgery, or donor characteristics (Table 1). However, illness scores were significantly different between the

4 744 Tafur et al. Table 1 Clinical and donor characteristics of all study patients and according to blood loss. Data are presented as number and percentage, median (IQR), and mean (SD). Definition of abbreviations: BDD, brain dead donor; FFP, fresh frozen plasma; LBL, low blood loss; HBL, high blood loss; LDLT, living donor liver transplantation; RBC, red blood cell. P, statistical difference between patients with low and high blood loss Total, n¼198 LBL, n¼51 HBL, n¼147 P Clinical characteristics Gender M/F 144/54 34/17 110/ Age (yr) 57 (49 62) 58 (50 62) 56 (48 62) BMI (kg m 2 ) 25.4 (23 28) 25.3 ( ) 25.7 ( ) Child-Pugh score (A/BC) 40/97/61 8/19/24 32/78/37 0 Lab MELD 18 (11 23) 10 (8 19) 19 (14 25) 0 Creatinin (mg dl 1 ) 0.9 ( ) 0.8 ( ) 1( ) 2 Bilirrubin (mg dl 1 ) 2.4 ( ) 1.1 ( ) 3.0 ( ) 1 Albumin (g dl 1 ) 3.3 ( ) 3.7 ( ) 3.1 ( ) 0 Haematocrit (%) 36.4 (2.8) 37.1 (3.3) 36.7 (4.6) Platelet-count (10 9 l 1 ) 78 (49 114) 99 (72 143) 69 (41 109) 2 Fibrinogen (g l 1 ) 2.1 ( ) 2.6 ( ) 2.0 ( ) 1 INR 1.5 ( ) 1.2 ( ) 1.6 ( ) 1 Aetiology of liver disease % Viral hepatitis 93 (47.0) 29 (56.9) 64 (43.5) Alcoholic hepatitis 43 (21.7) 7 (13.7) 36 (24.5) Acute liver failure 16 (8,1) 4 (7.8) 12 (8.2) Cholestatic cirrhosis 12 (6.0) 5 (9.8) 7 (4.8) Others 34 (17.2) 6 (11.8) 28 (19.0) Previous abdominal surgery % 67 (33.8) 10 (19.6) 57 (38.7) 0.09 Total blood loss (ml) 2925 (589) 1676 (276) 4165 (674) 0 RBC units 2 (0-3) 0 (0-2) 2 (1-4) 1 FFP ml 0 (0-0) 0 (0-0) 0 (0-480) 0 Donor characteristics % Age>65 yr 118 (59.5) 28 (54.9) 89 (60.5) BDD 161 (81.4) 39 (76.4) 122 (82.9) LDLT 37 (18.6) 12 (23.5) 25 (17) two groups. Unexpectedly, in the alcohol-related cirrhosis group the incidence of patients with HBL was significantly higher than that of patients with LBL [36 patients (84%) vs 7 patients (16%) respectively; P ¼ 0.032]. The thromboelastographic profile V-curve analysis related to bleeding risk in P1 A significant difference between the two groups was observed for all data representing clot initiation, propagation and firmness (Table 2). In the HBL group the CT and t-maxvel were longer, with a lower MaxVel and AUC than in the LBL group. The velocity/time ratio and MCF were also significantly lower. To investigate the ability of each V-curve variable to discriminate between the two groups, ROC curves were plotted. The only ROC curve with some discriminative capacity was t-maxvel, which had an area value of 0.69 (95% CI; ), as summarized in Figure 2A (left). In order to assess the relationship between V-curve parameters and blood loss, Spearmen correlation analysis was performed (Table 4). All parameters showed significant -though weak- correlation, indicating that a decrease in maximum velocity and prolonged time, together with a decline in clot firmness generation, correlates with higher bleeding. Patients with cirrhosis We independently analysed all 148 cirrhotic patients. The V- curve variables display similar ROC curve values to those observed for the whole patient group, but t-maxvel predicted bleeding risk with higher accuracy [area value of 0.78 (95% CI; -0.83)] (Table 3). In Spearmen correlation analysis, all V- curve variables showed a moderate correlation with blood loss (Table 4). In the subgroup analysis based on cirrhosis aetiology (alcohol vs virus hepatitis) no significant differences between the two groups were observed for all V-curve variables (data not shown). However, in patients with alcohol hepatitis the ROC curve of t-maxvel displayed the best discriminatory capacity of blood loss Figure 2C (left panel), with an area value of 0.83 (95% CI; ). Therefore, t-maxvel might less reliably discriminate between HBL and LBL in patients with viral hepatitis [area value of 0.73 (95% CI; )], as shown in Figure 2B (left panel). V-curve analysis related to bleeding risk in P2 The ROC curves (Fig. 2, right panel) and the correlations (Table 4) of all the V-curve variables were similar to those obtained in P1 but had a lower discriminatory capacity. Discussion Thromboelastographic evaluation of clot growth kinetics suggests that the time to reach maximum velocity of fibrin clot formation can discriminate bleeding risk, particularly in patients with cirrhosis aetiology undergoing LT.

5 ROTEM velocity-curve predicts blood loss 745 A 1.00 AUC (mm*100) MaxVel (mm*100 s 1 ) t, MaxVel (s) Reference 1.00 MaxVel/t,Maxvel (mm*100 s 2 ) 1.00 B C Fig 2 Receiver operating characteristic (ROC) curves of the V-curve parameters used for assessing the ability to discriminate between patients with high and low blood loss. Area under the ROC curve 0.5, prediction of blood loss not better than chance; area under the ROC curve ¼ 1.0, perfect prediction. P1(left panel). In A, ROC curves including all patients (n¼198): t-maxvel, area value of 0.69 (95% CI; ). In B, patients with viral hepatitis (n¼93): t-maxvel, area value of 0.73 ( ). In C, patients with alcohol hepatitis (n¼43): t-maxvel, area value of 0.83 ( ). P2 (right panel). In A, ROC curves including all patients: t-maxvel, area value of 0.62 (95% CI; ). In B, patients with viral hepatitis: t-maxvel, area value of 0.64 (95% CI; ). In C, patients with alcohol hepatitis: t-maxvel, area value of 0.73 (95% CI; ). In all scenarios, the ROC curves of MaxVel, AUC, and mv/t-m showed values<0.5. Historically, LT has been associated with massive blood loss and substantial transfusion requirements, but significant progress has led to marked change in transfusion practice over time. 19 Nevertheless, perioperative bleeding and transfusion requirements remain as potent predictors of postoperative complications, mortality and graft survival. 20 Since 1948, when Hartert presented the first viscoelastic test, the use of thromboelastography in the operating room has increased, particularly in LT and cardiac surgery, but also in trauma patients with life-threatening bleeding. Thromboelastometry has shown promise in directing blood product use in the perioperative LT period, and also in predicting transfusion requirements. 21 Recently, special interest in the relationship between the TG thrombogram and the V-curve of thromboelastography has emerged. Coagulation kinetics analysis performed in whole blood of healthy individual is quite different from that performed in plasma, mainly because of the presence of platelets. When TG is assayed in non-defibrinated platelet rich

6 746 Tafur et al. Table 2 Kinetics of clot growth generated by whole blood clotting rotational thromboelastometry (initiation, propagation and firmness) of all patients and patients according to blood loss. Data are presented as median (RIQ). Definition of abbreviations: AUC¼ area under curve of maximum velocity; CT¼ clotting time; MaxVel¼ maximum velocity; t-maxvel¼ time to reach maximum velocity; mv/t-m: time/velocity ratio. MCF¼ maximum clot firmness. P statistical results of Mann-Whitney U-test between patients with low and high blood loss Total, n¼198 LBL, n¼51 HBL, n¼147 P CT (s) 75.8 ( ) 76.3 ( ) 89.7 ( ) MaxVel (mm s 1 ) 10 ( ) 13 ( ) 10 ( ) 1 t-maxvel (s) 119 ( ) 79 ( ) 134 ( ) 1 mv/t-m (mm s 2 ) 10.8 ( ) 16.4 ( ) 7.7 ( ) 1 AUC 5027 ( ) 5519 ( ) 4730 ( ) 1 MCF EXTEM (mm) 51.2 ( ) 56.8 ( ) 44.1 ( ) 5 Table 3 ROC curves for estimating the probability of blood loss >961ml relative to t-maxvel parameter. Definition of abbreviations: t-maxvel, time to reach maximum velocity All study patients (n 5 198) ROC area 95% CI Proposed threshold sensitivity Specificity t-maxvel (s) > % 68.6% Total patients with cirrhosis aetiology (n ¼ 148) t-maxvel (s) > % 80.1% Alcohol hepatitis (n ¼ 43) t-maxvel (s) > % 82.4% Viral hepatitis (n ¼ 93) t-maxvel (s) > % 78.6% Table 4 Spearman correlations of all patients and patients with cirrhosis between V-curve parameters and blood loss in P1 and P2. Definition of abbreviations: AUC¼ area under curve of maximum velocity; MaxVel¼ maximum velocity (mm s 1 ); mv/t-m, time/velocity ratio (mm s 2 ); t-maxvel¼ time to reach maximum velocity (s) All patients (n¼198) plasma, a clear correlation with clot growth kinetics was observed. 22 The pattern produced from the V-curve software display a considerable degree of "profile similarity" to the TG thrombogram pattern: the speed of the thrombin burst [time to peak (ttp) t-maxvel]; its intensity [peak of TG (TP) MaxVel] and the total amount of thrombin generated [endogenous thrombin potential (ETP) AUC]. However, these assays should be viewed as complementary, with each providing different information. Care should be taken in linking data, mainly with platelet counts <60x10 9 l To our knowledge, there are no prospective studies correlating TG or fibrin formation with bleeding during LT surgery. In the present study, all V-curve variables correlated with blood loss, although t-maxvel was the only variable able to discriminate between low and high bleeding risk, particularly in patients with alcohol-related cirrhosis. A plausible explanation Patients with cirrhosis (n¼148) P1 P P2 P P1 P P2 P r r r r MaxVel < < <.001 t-maxvel < < < <.001 mv/t-m <.01 8 < < <.001 AUC < < < <.001 for this could be the fact that haemostatic capacity can be attributed to a combination of the rate of propagation and final strength of the clot, and the fact that fibrin clot structure, stability and strength are influenced by the kinetics, rather than total amount of thrombin generated. 23 Therefore, the final network structure and stability of fibrin clots plays an essential role in haemostasis, as both the primary end-product of the coagulation cascade and the ultimate substrate for fibrinolysis. Fibrin is cross-linked at lysine residues by factor XIIIa and forms fibrillar aggregates, which, together with platelets, red blood cells, and the complex interplay of many factors (ph, calcium and thrombin concentration), provide structural integrity to the growing thrombus (viscoelastic properties and resistance to fibrinolysis). For example, in patients with impaired clot kinetics with poor fibrin polymerization, the resulting thrombi are porous, composed of thick fibrin fibres and more

7 ROTEM velocity-curve predicts blood loss 747 susceptible to fibrinolysis. Interestingly, in a recent study Hugenholtz and colleagues 24 demonstrated that patients with cirrhosis show prolonged clotting times, as a result of delayed fibrin polymerization, that causes changes in fibrin clot structure resulting in dense fibrin networks composed of thin fibres resistant to lysis. The assumption that time has a close inverse correlation with the reactant concentrations, could explain why in our study t-maxvel was the variable with best capacity to discriminate between HBL and LBL and also the best correlation with blood loss. Curiously, the variable mv/t-m, based on our hypothesis that the faster the velocity (acceleration) of the fibrin polymerization, the more influence it could have on the fibrin clot structure and hence on bleeding, was no better at discriminating bleeding risk. In the setting of LT surgery, excessive blood loss is a consequence of mixed factors interrelated with the severity of the recipient s liver disease, such as the complexity of pretransplant coagulation conditions (derangements of the primary and secondary coagulation system), combined with surgical difficulties. One of the recognized contributors to blood loss in the LT surgery is the presence of portal hypertension (PH). Curiously, up to 80% of our patients with alcohol hepatitis were in the HBL group, nevertheless the rate of pre-transplant diagnosis of PH and oesophageal varices (grade II and III) were similar to patients with virus hepatitis. Recent clinical studies have shown that high concentrations of the platelet adhesion protein von Willebrand factor, with low concentrations of its regulator ADAMTS13, together with an increased potential to generate thrombin, mean that primary and secondary haemostasis are integrated physiologically to facilitate TG and fibrin formation throughout the surgical procedure. These findings underline the belief that, in patients with end-stage liver disease, the haemostatic system by itself is not as clinically relevant in determining perioperative bleeding, and that other factors might be involved. Anaesthetic management in particular could contribute to unwarranted blood loss during LT surgery. 27 There is evidence that liberal fluid therapy can lead to both increased PH (and hence, increased bleeding) and dilutional coagulopathy, which independently reduce TG and fibrin polymerization rate entailing changes in final network structure and stability of fibrin clot. 28 The results of the present study suggest that patients who preoperatively form a clot slowly are at increased risk of bleeding. Bearing in mind that fibrin(ogen) and platelets are the primary contributors to absolute clot strength, and also that fibrinogen concentration influences mechanical and functional properties of the clot growth kinetics, patients with deteriorated clot propagation and low functional fibrinogen might benefit from replacement therapy with fibrinogen-containing products. 29 Our suggestion is that standard, together with V- curve ROTEM parameters be considered to take a prophylactic rather than therapeutic approach, to help clinical decisions in selected cases of very high risk patients [e.g. patients with MCF FIFTEM <9 mm and MCF EXTEM at 10 min < 35 mm, 16,30 and high t-maxvel values (>115 s)], should be considered for fibrinogen-containing products. Although platelets and clotting factors, work in concert to allow formation of the fibrin network, reduced prophylactic use of platelet concentrates should be conceivable, as they are potentially harmful for the graft 31 and overall survival. 20 Some limitations of this study should be considered, including the retrospective design. First, we did not monitor the TG curve, and it would have been interesting to make a concordance between the thrombogram and the first derivative parameters and thus establish the accuracy of the V-curve for predicting bleeding risk in LT. Second, the dynamics of clot formation have been shown to be sensitive to storage time after blood collection. According with our protocol, storage time was 15 min and it is well known that there is a pronounced variance in the thrombelastographic parameters and particularly in the MaxVel parameter 9 within the first 30 min after venipuncture, presumably reflecting platelet activation. This effect might partially explain the variable reproducibility observed between studies carried out by viscoelastic systems. As thrombin is the central enzyme in the coagulation cascade, with the resulting thrombin burst crucial for the formation of a stable fibrin clot, improved management of patients undergoing LT requires more knowledge on the TG capacity and the effect on fibrin clot structure. In summary, we showed that the t-maxvel variable obtained from the ROTEM waveform, is a useful parameter to be considered as a measure of bleeding risk during LT, mainly in patients with a cirrhosis aetiology and might be considered as additional information (together with standard parameters) to aid assessment of pretransplant disturbances in global haemostatic capacity. If prospective studies confirm these findings, identifying the bleeding risk of patients undergoing LT will become easier, given that thromboelastography is a quick bedside test that is becoming increasingly available. Authors contributions Study design/planning: T.L., T.P., B.A., B.J., G.J.C. Study conduct: T.L., B.J., M.G. Data analysis: T.L., T.P., B.A., B.J., M.G. Writing paper: T.P. Revising paper: all authors Acknowledgements The authors gratefully acknowledge the help of Erika Cantor, of the Department of Bio-Statistics of Centro Medico Imbanaco (Calli, Colombia) for consulting on the statistical work. Declaration of interest None declared. Funding References 1. Massicotte L, Beaulieu D, Thibeault L, et al. Coagulation defects do not predict blood product requirements during liver transplantation. Transplantation 2008; 85: Ramos E, Dalmau A, Sabate A, et al. Intraoperative red blood cell transfusion in liver transplantation: influence on patient outcome, prediction of requirements, and measures to reduce them. Liver Transpl 2003; 9: Steib A, Freys G, Lehmann C, Meyer C, Mahoudeau G. Intraoperative blood losses and transfusion requirements during adult liver transplantation remain difficult to predict. Can J Anaesth 2001; 48:

8 748 Tafur et al. 4. Warnaar N, Lisman T, Porte RJ. The two tales of coagulation in liver transplantation. Curr Opin Organ Transplant 2008; 13: Hemker HC, Giesen P, Al Dieri R, et al. Calibrated automated thrombin generation measurement in clotting plasma. Pathophysiol Haemost Thromb 2003; 33: Tripodi A, Mannucci PM. The coagulopathy of chronic liver disease. N Engl J Med 2011; 365: Bosch YP, Al Dieri R, Ten Cate H, et al. Preoperative thrombin generation is predictive for the risk of blood loss after cardiac surgery: a research article. J Cardiothorac Surg 2013; 8: Cardenas JC, Rahbar E, Pommerening MJ, et al. Measuring thrombin generation as a tool for predicting hemostatic potential and transfusion requirements following trauma. J Trauma Acute Care Surg 2014; 77: Sørensen B, Johansen P, Christiansen K, Woelke M, Ingerslev J. Whole blood coagulation thrombelastographic profiles employing minimal tissue factor activation. J Thromb Haemost 2003; 1: Lang T, Toller W, Gütl M, et al. Different effects of abciximab and cytochalasin D on clot strength in thromboelastography. J Thromb Haemost 2004; 2: Wolberg AS. Thrombin generation and fibrin clot structure. Blood Rev 2007; 21: Kumar R, Béguin S, Hemker HC. The influence of fibrinogen and fibrin on thrombin generation evidence for feedback activation of the clotting system by clot bound thrombin. Thromb Haemost 1994; 72: Tripodi A, Primignani M, Chantarangkul V, et al. Global hemostasis tests in patients with cirrhosis before and after prophylactic platelet transfusion. Liver Int 2013; 33: Kleinegris MC, Bos MH, Roest M, et al. Cirrhosis patients have a coagulopathy that is associated with decreased clot formation capacity. J Thromb Haemost 2014; 12: Lisman T, Porte RJ. Rebalanced hemostasis in patients with liver disease: evidence and clinical consequences. Blood 2010; 6: Blasi A, Beltran J, Pereira A, et al. An assessment of thromboelastometry to monitor blood coagulation and guide transfusion support in liver transplantation. Transfusion 2012; 52: Jespersen ND, Kinetics, 4th Ed. In: ND Jespersen, eds. Barron s AP chemistry. Hauppauge, NY: Barron s Educational Series, 2007; Witters P, Freson K, Verslype C, et al. Review article: blood platelet number and function in chronic liver disease and cirrhosis. Aliment Pharmacol Ther 2008; 27: Ozier Y, Tsou MY. Changing trends in transfusion practice in liver transplantation. Curr Opin Organ Transplant 2008; 13: De Boer MT, Christensen MC, Asmussen M, et al. The impact of intraoperative transfusion of platelets and red blood cells on survival after liver transplantation. Anesth Analg 2008; 106: Wang SC, Shieh JF, Chang KY, et al. Thromboelastographyguide transfusion decreases intraoperative blood transfusion during orthotopic liver transplantation: randomized clinical trial. Transplant Proc 2010; 42: Sorensen B, Ingerslev J. Tailoring haemostatic treatment to patient requirements an update on monitoring haemostatic response using thromboelastography. Haemophilia 2005; 11(Suppl. 1): Wolberg AS, Campbell RA. Thrombin generation, fibrin clot formation and hemostasis. Transfus Apher Sci 2008; 38: Hugenholtz GC, Macrae F, Adelmeijer J, et al. Procoagulant changes in fibrin clot structure in patients with cirrhosis are associated with oxidative modifications of fibrinogen. J Thromb Haemost 2016; 14: Pereboom ITA, Adelmeijer J, van Leeuwen Y, Hendriks HGD, Porte RJ, Lisman T. Development of a severe von Willebrand factor/adamts13 dysbalance during orthotopic liver transplantation. Am J Transplant 2009; 9: Lisman T, Bakhtiari K, Pereboom ITA, Hendriks HGD, Meijers JCM, Porte RJ. Normal to increased thrombin generation in patients undergoing liver transplantation despite prolonged conventional coagulation tests. J Hepatol 2010; 52: de Boer MT, Molenaar IQ, Hendriks HG, Slooff MJ, Porte RJ. Minimizing blood loss in liver transplantation: progress through research and evolution of techniques. Dig Surg 2005; 22: Schols SE, Lancé MD, Feijge MA, et al. Impaired thrombin generation and fibrin clot formation in patients with dilutional coagulopathy during major surgery. Thromb Haemost 2010; 103: Lang T, Johanning K, Metzler H, et al. The effects of fibrinogen levels on thromboelastometric variables in the presence of thrombocytopenia. Anesth Analg 2009; 108: Lang T, Bauters A, Braun SL, et al. Multi-centre investigation on reference ranges for ROTEM thromboelastometry. Blood Coagul Fibrinolysis 2005; 16: Sindram D, Porte RJ, Hoffman MR, Bentley RC, Clavien PA. Platelets induce sinusoidal endothelial cell apoptosis upon reperfusion of the cold ischemic rat liver. Gastroenterology 2000; 118: Handling editor: Hugh Hemmings