Begin typing your search above and press return to search.
EPUB Before Print


The Impact of Preoperative Variables on Intraoperative Blood Loss and Transfusion Requirements During Orthotopic Liver Transplant

Objectives: Liver transplant traditionally and potentially is associated with the risk of massive blood loss and transfusion, which can adversely affect transplant outcomes. Many variables influence the amount of bleeding, and these can be categorized as patient related, surgery related, and graft related. We aimed to assess the effects of these variables on the amount of bleeding and transfusion during liver transplant; predicting the risk of massive blood loss can help transplant teams to select and manage patients more effectively.

Materials and Methods: We retrospectively studied 754 patients who underwent liver transplant from 2013 to 2016 and analyzed more than 20 variables that could influence the volume of blood loss and packed cell transfusion.

Results: We found that at least 4 variables are strongly and independently correlated with blood loss volume: age, Model for End-Stage Liver Disease score, warm ischemia time, and total bilirubin. Furthermore, intraoperative blood loss had a weak but clinically important correlation with the underlying disease (ie, the cause of liver cirrhosis). Some variables, including international normalized ratio, platelet count, albumin, serum urea nitrogen, creatinine level, sodium level, and the amount of ascites, could be considered as “dependent” and weak predictors of massive blood loss. Sex of patient, cold ischemia time, surgery technique, and history of previous abdominal surgery were not correlated with the amount of bleeding.

Conclusions: With the use of the variables identified, we can properly select patients and surgical teams and promptly use modalities for decreasing and managing blood loss.

Key words : Intraoperative bleeding, Liver cirrhosis, Orthotopic liver transplant, Packed cell, Warm ischemia time


Intraoperative blood loss is an important com-plication during orthotopic liver transplantation, although significant improvements in surgical techniques have been made in the past decade, with some authors reporting that more than 70% of liver transplants have no need for transfusion.1 However, massive intraoperative bleeding can adversely affect liver transplant outcomes. Therefore, predicting patients who are at risk for massive bleeding and massive transfusion may help anesthesiologists to manage the patients more effectively and prevent adverse complications of bleeding and transfusion.2

The liver is the main site for production of coagulation factors, except factor VIII and XIII, as well as anticoagulants like protein C, protein S, and antithrombin. It also synthesizes some proteins of the fibrinolytic system, such as plasminogen and α2-antiplasmin.3 Patients with cirrhosis are in a rebalanced hemostatic state; that is, they are at risk for both bleeding and thrombosis, and routine laboratory tests cannot accurately predict the risk of bleeding, especially during the perioperative period.4

Up to now, many studies have been designed to determine variables that influence the volume of intraoperative blood loss during liver transplant. These studies assessed some preoperative or intraoperative factors, and the results were different and even contradictory.

In 2005, Cywinski and associates5 retrospectively analyzed data from 804 patients and found that Model for End-Stage Liver Disease (MELD) score (and its components) and platelet count meaningfully correlated with massive bleeding and transfusion; however, this strong correlation had no predictive value for the individual patient. They could not design a model for predicting the risk of massive bleeding in a given patient. They also revealed that massive bleeding and transfusion led to poor survival.

De Santis and associates6 studied 166 orthotopic liver transplants with piggyback technique from 2001 to 2011. The group concluded that Child-Turcotte-Pugh score (and not MELD), hemoglobin level, international normalized ratio (INR), and graft ischemia time were factors that strongly correlated with the amount of transfusion during liver transplant. They also assessed creatinine, bilirubin, and albumin as factors in their study.

Other studies include Alonso and associates,7 who evaluated 888 patients and found that MELD is a useful predictor of both intraoperative transfusion need and postoperative survival. In addition, Huang and colleagues8 studied 198 pediatric patients who underwent liver transplant; their results revealed that the only factor that could reliably predict the risk of massive transfusion was INR. Chidananda Swamy9 reported that the amount of bleeding during liver transplant was correlated with many factors, including hemostatic abnormalities, portal hypertension, numerous collaterals, portal vein thrombosis, previous abdominal surgery, poorly functioning new liver, severity of liver disease, surgeon experience, and duration of anhepatic phase. These variables can be placed into three categories: (1) recipient related, (2) surgery related, and (3) graft related.3

In this study, we conducted an analysis focused on variables of recipients from our center (Namazi Hospital, Shiraz, Iran), who were seen from 2013 to 2016. However, some surgery- or graft-related variables were included because of their profound confounding effects in statistical analyses. The goal was to distinguish those patients who are at risk of massive bleeding, so that they could be directed toward better surgical and anesthetic care. The advantage of our study was the increasing experience of our surgical teams and the dramatically diminished blood loss and transfusion; therefore, the confounding effects of these factors may be decreased. The other strength was the number of cases analyzed, making it one of the largest studies recently performed.

Materials and Methods

A sample size of as many as 754 patients was ascertained after consultation with statisticians. This sample size allowed calculation of 80% power and 5% type 1 error. All adult (18-65 y) patients who underwent orthotopic liver transplant from 2013 to 2016 at the Shiraz Namazi Hospital were included. All patients in this study were new cases, with those who required redo transplants excluded. Although this was a retrospective study, patients were followed for evaluation of survival. Data were gathered by 2 anesthesia nurses using a prepared questionnaire, with data then imported into SPSS software (SPSS: An IBM Company, version 24, IBM Corporation, Armonk, NY, USA) for statistical analyses. These data included age, sex, weight, MELD score, preoperative laboratory findings, history of abdominal surgery, the underlying disease, the particular surgeon, technique of surgery, the volume of ascites, warm and cold ischemia times, and finally the volume of blood loss and packed red blood cell (RBC) transfusion during operation. Data were analyzed using t test, Mann-Whitney U test, chi-square test, Pearson correlation coefficient, and multiple linear regression. We also used Kolmogorov-Smirnov test for assessment of normality.

Anesthesia technique
In our center, anesthesia for liver transplant is accomplished via a standard protocol; all patients were monitored first by electrocardiogram, noninvasive blood pressure monitoring, and pulse oximetry; anesthesia was then induced using midazolam (0.03 mg/kg), fentanyl (2-3 μg/kg), sodium thiopental (3-5 mg/kg), and pancuronium (0.2 mg/kg). If the patient had tense ascites, we used a rapid sequence technique using succinylcholine (1.5-2 mg/kg) rather than pancuronium and used cricoid pressure. We then inserted an arterial catheter in the radial artery and a large bore (No. 12) central venous pressure (CVP) catheter in the right internal jugular vein with the guide of sonography. Intravenous fluid included 1% to 2% albumin in normal saline, and the rate of infusion was regulated according to CVP, urine output, and the volume of blood loss.

Packed RBC transfusion started when the hemoglobin concentration was below 9 g/dL or the patient had hypotension without response to crystalloids/colloids. Excessive metabolic acidosis (base excess < -6.0) was treated with sodium bicarbonate. We used norepinephrine, epinephrine, and vasopressin as the main inotropic agents. We also applied infusion of calcium gluconate for preventing hypocalcemia induced by hemodilution or citrate intoxication.

Measurement of blood loss
The volume of blood loss was calculated by the sum of blood in suction reservoirs plus blood in sponges minus the volume of irrigation fluid and recorded in milliliters. All data had been actively measured and precisely recorded by the anesthesia team in our center.


We calculated the amount of blood loss and packed RBC transfusion in all 754 patients and assumed these were the main dependent variables so that we could analyze the impact of these versus all other variables (Table 1).

Table 2 shows the correlation between 19 quantitative variables and total bleeding during liver transplant. Our statistical analyses showed that age, weight, MELD score, INR, albumin, total bilirubin, serum urea nitrogen, creatinine, volume of ascites, and warm ischemia time had weak correlations with total bleeding. Platelet count, serum sodium level, serum albumin, bicarbonate, partial pressure of arterial carbon dioxide, and hemoglobin level were inversely correlated with total bleeding (negative correlation coefficient). Potassium, blood pH, base excess, and cold ischemia time were not meaningfully correlated with the amount of bleeding.

Figure 1 reveals the mean volume of blood loss on the basis of underlying disease (ie, the cause of liver cirrhosis). Maximum bleeding occurred in patients with nonalcoholic steatohepatitis (NASH) and Budd-Chiari disease. The patients with Wilson disease, primary sclerosing cholangitis, and primary biliary cirrhosis had the least intraoperative bleeding.

Figure 2 compares the mean intraoperative bleeding on the basis of the surgeon. It is clear that surgeon experience may have dramatically influenced the amount of bleeding. However, these differences may have been due to factors other than surgical experience. Therefore, we eliminated the effects of these confounding factors when we employed the multiple linear regression test. That is, we viewed the surgical team as not an independent predictor of blood loss (see Table 6).

Surgical technique was a variable that we viewed as having a dramatic influence on the amount of blood loss; however, when we compared mean blood loss in the 2 main surgery techniques, no significant differences were shown (Table 3).

The amount of intraoperative bleeding was not different in male versus female patients (Table 4). One factor traditionally considered as a main cause of bleeding is history of abdominal surgery; however, our study showed no difference between these patients and those who had no previous abdominal surgery (Table 5).

We also analyzed these factors with transfusion rate instead of amount of bleeding. Our results were nearly similar; therefore, for brevity, we excluded transfusion rate discussion.

To eliminate the effects of confounding factors, we used multiple linear regression analyses and found 4 factors that independently correlated with the amount of surgical bleeding during liver transplant: MELD, age, total bilirubin, and warm ischemia time (Table 6).


It should be noted that our findings may be different from other centers. In our center, most liver donations are from deceased donors. In contrast, in some other centers, especially in Asian countries, liver transplants are often from living donors. Each center should keep this in mind when interpreting our results for applicability.

Another point worth noting is whether to compare the amount of blood loss in 2 groups, which must be matched with regard to other effective variables, or to determine blood loss with a large enough sample size to eliminate confounding effects. We selected the second choice.

According to our results, the amount of bleeding during liver transplant was independently correlated with 4 factors (Table 6): age, MELD score, total bilirubin, and warm ischemia time. Bleeding was increased 36 mL per 1 year increase in age, 73 mL per 1 unit increment in MELD score, and 46 mL per 1 minute increase in warm ischemia time. Surprisingly bleeding was decreased 36 mL per 1 mg/dL increment in total bilirubin level (ie, the total bilirubin level was inversely correlated with the amount of bleeding during liver transplant).

Nevertheless, we cannot discard the results of our primary correlation tests (Tables 2-5 and Figures 1 and 2). Although the findings of the multiple linear regression test were useful in causal relations, the results of crude correlation tests and analyses of variances may have clues that are clinically important. For example in Table 2, we see that total bilirubin was directly correlated (r = 0.142) with volume of blood loss, and this correlation was statistically significant (P < .001); this is in conflict with regression test results that revealed inverse correlations of total bilirubin and blood loss volume. One explanation is that high bilirubin levels are usually seen in primary sclerosing cholangitis and primary biliary cirrhosis; we observed that these patients had less blood loss than those with other causes of liver cirrhosis during liver transplant (Figure 1). That is, an increase in total bilirubin was an index of more severe disease, and it is predictable that patients with higher bilirubin have higher MELD score, INR, and other factors. Therefore, our correlation tests (Table 2) showed that total bilirubin was directly correlated with the volume of bleeding during liver transplant. In general, the results of correlation tests may not have shown significant causality, but they have clinical importance and application.

Another important finding was that warm ischemia time in both correlation and regression tests was strongly correlated with the amount of bleeding, whereas cold ischemia time showed no correlation.

In a previous study, De Santis and colleagues6 found no correlation between MELD and bleeding, but they found a strong correlation between INR and total bleeding. Conversely, Cywinski and associates5 revealed that platelet count and MELD had strong correlations with massive bleeding; these differences may be due to statistical methods of analysis as previously mentioned. We saw in our results that platelet count, in correlation tests, was inversely correlated with bleeding; however, it was not independently correlated, as the regression test showed.

It may be that there are other factors and interventions that can influence the amount of blood loss during orthotopic liver transplant; we know that hyperfibrinolysis may aggravate coagulopathy during the anhepatic phase of liver transplant. This phenomenon is caused by decreased clearance of tissue plasminogen activator by the liver during the anhepatic phase of liver transplant.10 Boylan and associates11 assessed the effects of tranexamic acid on intraoperative blood loss. They found that tranexamic acid diminishes blood loss and the need for packed RBCs and/or clotting factors. Dalmau and colleagues12 revealed that 10 mg/kg tranexamic acid significantly reduced blood loss and fibrinolysis compared with aminocaproic acid and placebo. In our center, we use tranexamic acid when the rotational thromboelastometry (ROTEM) reveals fibrinolysis and the surgeon notes oozing in the surgical field.

As mentioned previously by our group,4 preoperative coagulation tests are not a reliable guide for administrating fresh frozen plasma, platelet concentrate, cryoprecipitate, or other procoagulant factors; therefore, we used these agents with the help of ROTEM and assessment of the surgical field by the surgeon. In our practice, coagulopathy and related bleeding are mainly seen as a result of massive transfusion, massive crystalloid and colloid admin-istration, and prolonged anhepatic phase but not with an abnormal preoperative coagulation profile alone. Our center presently performs many liver transplants without any transfusion of blood or coagulation factors, even with prolonged pre-operative INR and low platelet count.

The low CVP technique can effectively reduce bleeding during liver transplant without any adverse effects on morbidity, mortality, or outcome.13 It can be used if the predicted risk of massive bleeding is high according to preoperative findings. In our center, we keep CVP between 5 and 10 mm Hg using 2.5% albumin, packed RBCs if indicated, and inotropes like norepinephrine and vasopressin. Indeed, we abstain from the liberal use of crystalloids or colloids.

Surgery-related variables, including techniques of venous anastomoses (piggyback or standard methods), and surgeon experience are important variables; however, we found no meaningful differences between them (Tables 3 and 6). These variables were not the main targets of our study. Therefore, we only considered differences because they were confounding factors in evaluating preoperative variables.

There are some new modern laboratory analyses that may allow stronger correlations to be shown regarding bleeding and transfusion needs versus routine laboratory tests. Fayed and associates14 demonstrated that most ROTEM variables can accurately predict packed RBCs, fresh frozen plasma, cryoprecipitates, and platelet transfusion needs; however, this test is expensive and is not routinely performed before liver transplant. More economic models for prediction of blood loss risk are needed as they are more practical and cost effective. Indeed, application of ROTEM even during liver transplant was not associated with decreased packed RBC transfusion according to Roullet and colleagues.15

We can apply the variables identified here in deciding about surgery technique, adequate blood and factor reservation, proper selection of surgeons, and preparing for rapid infusion system or cell saver use.


Our study showed that age, MELD score, warm ischemia time, and total bilirubin were independent factors with strong influence on blood loss during liver transplant; in other words, an older patient with high MELD score and low bilirubin level will have higher blood loss and transfusion requirements during liver transplant regardless of graft- or surgery-related variables. Graft- and surgery-related factors may have independent effects on blood loss, but they were not the focus of our study. We can use other correlated factors such as underlying liver disease and INR in clinical practice for prediction of massive blood loss. If risk of bleeding is shown to be high, we may consider the low CVP technique, administration of tranexamic acid and other coagulation factors, and application of ROTEM, cell saver, the rapid infusion system, and/or other interventions for control and reduction of bleeding during liver transplant.


  1. Solves P, Carpio N, Moscardo F, et al. Transfusion management and immunohematologic complications in liver transplantation: experience of a single institution. Transfus Med Hemother. 2015;42(1):8-14.
  2. 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(11):1075-1079.
    CrossRef - PubMed
  3. Starczewska MH, Giercuszkiewicz D, Niewinski G, Kanski A. Perioperative bleeding in patients undergoing liver transplantation. Anaesthesiol Intensive Ther. 2016;48(1):34-40.
    CrossRef - PubMed
  4. Weeder PD, Porte RJ, Lisman T. Hemostasis in liver disease: implications of new concepts for perioperative management. Transfus Med Rev. 2014;28(3):107-113.
    CrossRef - PubMed
  5. Cywinski JB, Alster JM, Miller C, Vogt DP, Parker BM. Prediction of intraoperative transfusion requirements during orthotopic liver transplantation and the influence on postoperative patient survival. Anesth Analg. 2014;118(2):428-437.
    CrossRef - PubMed
  6. De Santis GC, Brunetta DM, Nardo M, et al. Preoperative variables associated with transfusion requirements in orthotopic liver transplantation. Transfus Apher Sci. 2014;50(1):99-105.
    CrossRef - PubMed
  7. Alonso IJ, Maestro OC, Gomez NF, et al. Mind the model for end-stage liver disease: model for end-stage liver disease score as an indicator of hemoderivate transfusion and survival in liver transplantation. Transplant Proc. 2015;47(1):97-99.
    CrossRef - PubMed
  8. Huang CJ, Cheng KW, Chen CL, et al. Predictive factors for pediatric patients requiring massive blood transfusion during living donor liver transplantation. Ann Transplant. 2013;18:443-447.
    CrossRef - PubMed
  9. Chidananda Swamy M. Blood transfusion practices in liver transplantation. Indian J Anaesth. 2014;58(5):647-651.
    CrossRef - PubMed
  10. Poon KS, Chen CC, Thorat A, et al. Fibrinolysis after reperfusion of liver graft. Acta Anaesthesiol Taiwan. 2015;53(1):41-43.
    CrossRef - PubMed
  11. Boylan JF, Klinck JR, Sandler AN, et al. Tranexamic acid reduces blood loss, transfusion requirements, and coagulation factor use in primary orthotopic liver transplantation. Anesthesiology. 1996;85(5):1043-1048; discussion 1030A-1031A.
  12. Dalmau A, Sabate A, Acosta F, et al. Tranexamic acid reduces red cell transfusion better than epsilon-aminocaproic acid or placebo in liver transplantation. Anesth Analg. 2000;91(1):29-34.
    CrossRef - PubMed
  13. Pandey CK, Singh A, Kajal K, et al. Intraoperative blood loss in orthotopic liver transplantation: The predictive factors. World J Gastrointest Surg. 2015;7(6):86-93.
    CrossRef - PubMed
  14. Fayed N, Mourad W, Yassen K, Gorlinger K. Preoperative Thromboelastometry as a predictor of transfusion requirements during adult living donor liver transplantation. Transfus Med Hemother. 2015;42(2):99-108.
    CrossRef - PubMed
  15. Roullet S, Freyburger G, Cruc M, et al. Management of bleeding and transfusion during liver transplantation before and after the introduction of a rotational thromboelastometry-based algorithm. Liver Transpl. 2015;21(2):169-179.
    CrossRef - PubMed

DOI : 10.6002/ect.2016.0325

PDF VIEW [230] KB.

From the 1Department of Anesthesiology, School of Medicine, Shiraz University of Medical Sciences, Shiraz, Iran; the 2Department of Epidemiology, Torbat Heydarieh University of Medical Sciences, Torbat heydarieh, Iran; and the 3Shiraz Organ Transplantation Center, Shiraz, Iran
Acknowledgements: The authors have no sources of funding for this study and have no conflicts of interest to declare.
Corresponding author: Kazem Samadi, Department of Anesthesiology, School of Medicine, Shiraz University of Medical Sciences, Shiraz, Iran
Phone: +98 9173033632