Patients with end-stage liver disease frequently incur large-volume blood loss during liver transplantation associated with mechanical factors, preexisting coagulopathy, and intraoperative fibrinolysis.
Between April 1992 and May 1994, the authors of this double-blind, randomized, placebo-controlled study examined the effect of high-dose tranexamic acid (maximum of 20 g) on blood loss and blood product requirements in patients undergoing primary isolated orthotopic liver transplantation. Primary outcome measures were volume of blood loss (intraoperative blood loss and postoperative drainage) and erythrocyte, plasma, platelet, and cryoprecipitate use during surgery and the first 24 h of intensive care unit stay.
Patients receiving tranexamic acid (n = 25) had less intraoperative blood loss (median, 4.3 l; interquartile range, 2.5 to 7.9; P = 0.006) compared with the placebo group (n = 20; median, 8 l; interquartile range, 5 to 15.8), and reduced intraoperative plasma, platelet, and cryoprecipitate requirements. Median perioperative erythrocyte use was 9 units (interquantile range, 4 to 14 units) in patients receiving tranexamic acid and 13 units (interquantile range, 7.5 to 31 units) in controls (P = 0.03). Total perioperative donor exposure was 20.5 units (interquantile range, 16 to 41 units) in patients receiving tranexamic acid and 43.5 units (interquantile range, 29.5 to 79 units) in controls (P = 0.003). Results for postoperative wound drainage were similar. Hospital stay and need for retransplantation were comparable in both groups. No patient in either group showed clinical evidence of hepatic artery or portal venous thrombosis within 1 month of transplantation.
High-dose tranexamic acid significantly reduces intraoperative blood loss and perioperative donor exposure in patients with end-stage parenchymal liver disease who are undergoing orthotopic liver transplantation, with marked reductions in platelet and cryoprecipitate requirements.
Key words: Liver: transplantation. Blood: antifibrinolytics; tranexamic acid; hemorrhage.
Orthotopic liver transplantation is routinely associated with large-volume blood loss. With increasing experience, blood product use has decreased; however, patients still frequently require ten units of erythrocytes. Perioperative transfusion requirements are not well predicted by preoperative clinical findings or laboratory assessment, and excessive blood loss is associated with increased postoperative morbidity, death, and intensive care unit stay. Accelerated fibrinolysis is one of several important factors associated with major blood loss. Antifibrinolytic agents are commonly administered during orthotopic liver transplantation, but studies of their effectiveness have been limited by small numbers, the absence of blinding or randomization, or the use of historical controls. [3-8]The objective of this study was to assess the efficacy of tranexamic (TA) acid in reducing blood loss and blood product requirements in patients undergoing liver transplantation.
Patients and Methods
After receiving local institutional review board approval, we enrolled in our study 45 patients undergoing primary, isolated orthotopic liver transplantation from April 1992 to May 1994. Participants were approached for recruitment during pretransplant counseling sessions, and written consent was obtained at admission for surgery. Patients thought to be at greater risk based on a diagnosis of necrotic liver disease (chronic active hepatitis, cirrhosis) were selected for recruitment. Patients were excluded if (1) they had diagnoses such as primary biliary cirrhosis or primary sclerosing cholangitis, which are associated with low or intermediate risk for major blood loss [9,10]; (2) their underlying diagnosis suggested predisposition to a thrombotic tendency (Budd-Chiari syndrome, for example); or (3) they presented in fulminant hepatic failure. We excluded patients in category 1, in whom preoperative coagulation screening has a lower incidence of abnormality, [9,10]to reduce variability in sample size calculations and enhance study power. Patients with fulminant hepatic failure were excluded because their small numbers precluded meaningful study.
Patients were randomized to receive a continuous infusion of TA in normal saline (40 mg [centered dot] kg sup -1 [centered dot] h sup -1 to a maximum dose of 20 g) or an equivalent volume of placebo (normal saline alone). Study agents were prepared by the hospital pharmacy using a randomization schedule provided in sealed envelopes; all other personnel were blinded to randomization status. The infusion was begun after induction of anesthesia and was discontinued when portal veins were unclamped. A solution of dipyridamole-heparin (5,000 heparin units, 150 mg dipyridamole in 250 ml D5W) was begun at 10 ml/h immediately after completion of the hepatic arterial anastomosis and was infused for 24 h.
General anesthesia was induced at the discretion of the attending anesthesiologist caring for the patient. After insertion of monitoring and vascular access catheters, a standard surgical technique was applied, with axilloportalfemoral venovenous bypass used when deemed necessary by the surgical team. Liver allografts were preserved using University of Wisconsin solution. Hourly complete blood count and coagulation factor data were obtained in all patients. Erythrocyte, plasma, platelet, and cryoprecipitate administration were according to a predetermined protocol. Packed erythrocytes were administered to maintain a hematocrit concentration of 0.25 to 0.30 throughout surgery. Plasma was administered in two-unit aliquots when prothrombin time exceeded 20 s or factor VII level decreased to less than 0.5 U/ml. Cryoprecipitate was administered in four-unit aliquots when the fibrinogen level decreased to less than 1.5 g/l or factor VIII level decreased to less than 1.5 U/ml. Platelets were administered in six-unit aliquots if the platelet count decreased to less than 50,000/micro liter. Autotransfusion was not used. A member of the research group was present throughout the procedure to collect data. Intraoperative blood loss was measured by adding suction volumes (less irrigation used) and sponge weights. Postoperative blood loss was measured by recording the output from surgical drains in the first 24 h after operation. The number of units of erythrocytes, plasma, platelets, and cryoprecipitate given were recorded during surgery and during the first 24 h after operation. Blood samples were routinely drawn for complete blood count prothrombin time and partial thromboplastin time when patients arrived in the surgical intensive care unit, and on the first and second postoperative days.
Hematocrit and coagulation management was continued in the surgical intensive care unit in the same manner as during operation for the first 24 h after surgery. Decisions concerning fluid, ventilation, and surgical management were the responsibility of the surgical intensive care unit and surgical teams. Data collected included standard demographics, medical diagnosis, use and duration of venovenous bypass, duration of stay in the surgical intensive care unit, and rates of 30-day mortality. Thrombotic complications leading to impaired hepatic function occurring within 30 days of surgery were also recorded. Screening techniques (e.g., Doppler) to identify these complications were not used.
We calculated sample size based on initial experience with TA, in which we found a 50% reduction in perioperative erythrocyte use. The trial was designed to detect a reduction of at least 40% in blood loss for the TA group, with a one-tailed alpha error of 0.05 and a beta error of 0.20, yielding a total sample size of 35 for parametric test purposes. Because of the reduced power of nonparametric statistical testing, total sample size was adjusted upward to 45. Demographic data were compared using Student's t test and chi squared or Fisher's exact test. Because ischemia times, duration of patient stay, blood loss, and blood product use data appeared asymmetrical, they were compared using the Mann-Whitney U test corrected for ties. Data are presented as means (SD) or medians (interquartile ranges [IR]) as appropriate. Differences with probability values of 0.05 or less were considered significant.
Demographic and Surgical Variables
Patient treatment groups were demographically similar with regard to preoperative diagnosis, medical status, and incidence of varices, encephalopathy, and ascites (Table 1). Use of venovenous bypass and graft cold ischemia times and rewarming times did not differ between groups. Early and late intraoperative coagulation factor levels were similar in both groups (Table 2). Hospital and median intensive care unit stay were similar. Three patients died at 30 days in the placebo group compared with none in the TA group (P = 0.08). No patient sustained clinical evidence of hepatic artery thrombosis or portal vein thrombosis within 30 days of surgery. Five study patients (TA = 2, control = 3) have since undergone retransplantation (median time to reoperation, 5 months).
Blood Loss and Transfusion
Blood loss and transfusion requirements were not normally distributed. Median intraoperative blood loss was significantly less in patients receiving 4.3 l TA (IR, 2.5 to 7.9) compared with 8 l placebo (IR, 5 to 15.8; P = 0.006). Median postoperative drainage was 0.8 l (IR, 0.5 to 2.1) in patients receiving TA compared with 1.2 l (IR, 0.9 to 1.8) in controls (P = 0.39). Patients receiving TA received less fresh frozen plasma (P = 0.04), cryoprecipitate (P = 0.01), and platelets (P = 0.02) during operation (Table 3). Median intraoperative donor exposure was 18 units (IR, 9.8 to 32.8 units) associated with TA and 35 units (IR, 21 to 62 units) in controls (P = 0.01). Although fresh frozen plasma use was reduced in the TA group after operation, reductions in erythrocyte and platelet use were not statistically significant. Postoperative cryoprecipitate use was comparable in both groups. Median postoperative donor exposure was 1 unit (IR, 0 to 6 units) in patients receiving TA and 7.5 units (IR, 3 to 14.5 units) in controls (P = 0.01). Overall median perioperative donor exposure was 20.5 units (IR, 16 to 41 units) in patients receiving TA and 43.5 units (IR, 29.5 to 79 units) in controls (P = 0.003).
Immediate postoperative hemoglobin, platelet count, prothrombin time, and partial thromboplastin time were comparable in the TA and control groups, as were hemoglobin, platelet count, and partial thromboplastin time on the first postoperative day (Table 4). Despite receiving significantly more plasma in the early postoperative period, prothrombin time was significantly longer in control patients compared with those receiving TA.
Prophylaxis with TA was associated with moderate decreases in intraoperative blood loss and fresh frozen plasma requirements and more marked decreases in platelet and cryoprecipitate use. Perioperative erythrocyte use was decreased compared with controls. Our data confirm previous findings documented in open or historical control studies [7,8]and suggest that in patients with the disease profile described, in whom transfusion requirements frequently exceed one blood volume, TA prophylaxis may reduce perioperative erythrocyte use and donor exposure by as much as 50%. Large blood product requirements during liver transplantation are associated with increased morbidity and reduced survival. [1,9,10]Pharmacologic techniques that may reduce blood loss thus are being explored. Antifibrinolytic prophylaxis in blood conservation has been examined because of reports that aprotinin, epsilon-aminocaproic acid, and TA reduce clinical coagulopathy and blood loss seen in cardiac surgery with cardiopulmonary bypass.
Increasing insights into the hemostatic management of liver transplantation has led to empirical use of such agents in this setting. Reduced procoagulant factors, thrombocytopenia, and increased fibrinolysis frequently occur in patients with end-stage liver disease. These are aggravated by intraoperative dilutional factors, with disproportionate decreases in fibrinogen and factors V and VIII suggesting an accelerated fibrinolytic process. [17-19]An association between severe fibrinolysis and increased blood product requirements has suggested a role for antifibrinolytic therapy. Preliminary work indicates that low-dose epsiolon-aminocaproic acid attenuated hyperfibrinolysis during graft reperfusion, but no blood product use data were presented. More recent reports [4,5]suggest that aprotinin reduces fibrinolysis and blood loss during transplantation, with a large open trial showing decreased erythrocyte requirements. We found similar reductions with TA in a pilot study. A recent open trial of low-dose TA has reported decreases in blood loss, although not in blood product use, but this report was limited by an open design, small patient numbers, and the recruitment of patients at relatively low risk. The present trial was undertaken because of concerns about uncritical use of such agents in a setting where clinical management is changing over time.
In this clinical trial, we focused on the clinical efficacy of TA and did not address potential mechanisms of action. Despite the association between excessive fibrinolysis and blood loss, the efficacy of TA does not imply that fibrinolysis is the sole (or even the main) mechanism of excessive blood loss. Reduced platelet aggregation after graft reperfusion may contribute to the coagulopathy seen at this time. In patients having cardiac surgery, TA administration attenuates plasmin-induced platelet dysfunction, another potential mechanism of action. In the previously mentioned open trial of TA in liver transplantation, Yassen and colleagues described multiple effects on thromboelastograph indices of hemostasis, which were not limited to clot lysis alone. Further work is necessary to delineate the relative importance of fibrinolysis and platelet dysfunction and the therapeutic role of TA.
Thrombosis due to antifibrinolytic therapy is the potential complication of greatest concern, and the risk-benefit relationship of these agents is still not described. Differential rates of synthesis occur for pro- and anticoagulant factors in newly transplanted livers, with most procoagulant activity returning to normal within 24 h, whereas anticoagulant factors such as antithrombin III and proteins C and S normalize more slowly. To minimize thrombotic risk, we maintained low erythrocyte volume values and used prophylactic dipyridamole-heparin, practices that predated this trial. Our zero incidence of thrombotic complications is not amenable to statistical analysis because the small denominators involved (n [nearly equal] 25) permit a 95% confidence interval for the range of possible risk of 0 to 12.5%. No thrombotic sequelae have been documented with the drug in patients having either liver transplantation or cardiac surgery; a single thrombotic event was recently reported in a patient having liver transplantation after aprotinin use. Despite the absence of documented adverse effects, perioperative thrombotic events might be exacerbated by impaired clot lysis.
Our data on retransplantation and mortality compare favorably with previous outcome data from our institution (8.8% incidence of retransplantation, 27% 1-y mortality rate). Because increased blood loss correlates with worse outcome, a reduction in blood loss may have a favorable effect on morbidity or mortality, but our study cannot address this issue given our limited sample size. A sample size of approximately 350 patients per group would be necessary to evaluate the effects of TA or other agents on infrequent events such as mortality, posttransplant graft failure, and so forth. Our study contains some limitations. We recruited a relatively small number of patients. However, our design reflects the fact that we expected large reductions in blood product use. However, study bias was minimized by randomization and the double-blind design.
In addition, blood products were given based on objective criteria outlined in Patients and Methods. The similarities in early postoperative hemoglobin and coagulation data confirm compliance with the factor-replacement protocol. Our surgical teams use venovenous bypass as they feel appropriate to minimize hemodynamic and metabolic disturbance during the anhepatic phase. The absence of a relationship between treatment status and bypass use suggests that bypass management did not affect our results.
Our drug dosage was higher than that previously used. No dose-response data exist for TA in liver transplantation, and limited data exists on its use in patients having cardiac surgery, so our dosage was based on pilot use and on investigational data from our institution's use of TA in cardiac surgery. [27,28]This latter report showed significantly better results with high-dose (10 g) rather than low-dose (6 g) TA.
In high-risk patients undergoing liver transplantation, high-dose TA prophylaxis is associated with appreciable reductions in blood loss and perioperative exposure to allogenic blood products. Tranexamic acid is an acceptable alternative to aprotinin therapy and is the only agent to have proved efficacy in reducing blood product requirements in this setting. The risk-benefit relationship for antifibrinolytic prophylaxis remains to be determined.