Key Points
Question
Does tranexamic acid reduce red blood cell transfusion within 7 days of liver resection?
Findings
In this randomized clinical trial, red blood cell transfusion within 7 days of surgery occurred in 16.3% of participants receiving tranexamic acid and 14.5% receiving placebo. Participants receiving tranexamic acid experienced significantly more complications perioperatively compared with placebo (odds ratio, 1.28; 95% CI, 1.02-1.60; P = .03).
Meaning
Among patients undergoing liver resection for a cancer-related indication, tranexamic acid did not reduce bleeding or blood transfusion but increased perioperative complications.
Abstract
Importance
Tranexamic acid reduces bleeding and blood transfusion in many types of surgery, but its effect in patients undergoing liver resection for a cancer-related indication remains unclear.
Objective
To determine whether tranexamic acid reduces red blood cell transfusion within 7 days of liver resection.
Design, Setting, and Participants
Multicenter randomized clinical trial of tranexamic acid vs placebo conducted from December 1, 2014, to November 8, 2022, at 10 hepatopancreaticobiliary sites in Canada and 1 site in the United States, with 90-day follow-up. Participants, clinicians, and data collectors were blinded to allocation. A volunteer sample of 1384 patients undergoing liver resection for a cancer-related indication met eligibility criteria and consented to randomization.
Interventions
Tranexamic acid (1-g bolus followed by 1-g infusion over 8 hours; n = 619) or matching placebo (n = 626) beginning at induction of anesthesia.
Main Outcomes and Measures
The primary outcome was receipt of red blood cell transfusion within 7 days of surgery.
Results
The primary analysis included 1245 participants (mean age, 63.2 years; 39.8% female; 56.1% with a diagnosis of colorectal liver metastases). Perioperative characteristics were similar between groups. Red blood cell transfusion occurred in 16.3% of participants (n = 101) in the tranexamic acid group and 14.5% (n = 91) in the placebo group (odds ratio, 1.15 [95% CI, 0.84-1.56]; P = .38; absolute difference, 2% [95% CI, −2% to 6%]). Measured intraoperative blood loss (tranexamic acid, 817.3 mL; placebo, 836.7 mL; P = .75) and total estimated blood loss over 7 days (tranexamic acid, 1504.0 mL; placebo, 1551.2 mL; P = .38) were similar between groups. Participants receiving tranexamic acid experienced significantly more complications compared with placebo (odds ratio, 1.28 [95% CI, 1.02-1.60]; P = .03), with no significant difference in venous thromboembolism (odds ratio, 1.68 [95% CI, 0.95-3.07]; P = .08).
Conclusions and Relevance
Among patients undergoing liver resection for a cancer-related indication, tranexamic acid did not reduce bleeding or blood transfusion but increased perioperative complications.
Trial Registration
ClinicalTrials.gov Identifier: NCT02261415
This randomized clinical trial assesses the effect of tranexamic acid vs placebo on the need for blood transfusion among patients undergoing cancer-related liver resection.
Introduction
Liver resection, the optimal treatment for patients with primary or metastatic liver malignancies, is a major abdominal operation, with 15% to 25% of patients receiving perioperative blood transfusions.1,2 Intraoperative blood loss and transfusion are major risk factors for postoperative morbidity and mortality and are strongly associated with long-term cancer recurrence and lower overall survival.1,3,4,5 This relationship is likely multifactorial, partly related to patients with more advanced disease having a higher incidence of bleeding and blood transfusion, but also potentially directly related through host immunosuppression caused by allogenic blood transfusion.6,7,8,9
Tranexamic acid reduces the probability of receiving a blood transfusion by a third (risk ratio, 0.62 [95% CI, 0.58-0.65]) in a variety of surgical procedures.10 However, evidence supporting tranexamic acid is largely derived from cardiac and orthopedic surgery, in which bleeding often occurs as oozing from small vessels in raw surfaces, such as the pericardium or bones. The mechanism of bleeding differs in major abdominal surgery, such as liver resection, in which bleeding often occurs from large blood vessels. Furthermore, although existing evidence suggests no increase in thrombotic events, very little evidence supports the use of tranexamic acid in patients having surgery for cancer, where the risk of venous thromboembolism (VTE) is higher and may be propagated by tranexamic acid. As a result of these limitations, results from prior trials have not changed practice in North America, and most patients do not receive tranexamic acid prior to liver resection.11
To assess the effect of tranexamic acid compared with placebo on receipt of red blood cell transfusion and perioperative complications in patients undergoing liver resection for malignancy, the Hemorrhage During Liver Resection: Tranexamic Acid (HeLiX) trial was conducted.
Methods
Trial Design
The HeLiX trial is a multicenter, placebo-controlled, parallel, pragmatic randomized clinical trial to evaluate the effect of tranexamic acid on bleeding, blood transfusion, and perioperative complications in patients undergoing liver resection. The full trial protocol has been previously published12 and is provided in Supplement 1. Before recruitment began, the local research ethics boards at 11 participating tertiary care centers with specialty in hepatopancreaticobiliary surgery and Health Canada approved the trial. There were no major changes to the protocol during the course of the trial. Reporting follows the Consolidated Standards of Reporting Trials (CONSORT) guideline.13
Trial Oversight
The Applied Health Research Centre at the Li Ka Shing Knowledge Institute of St Michael’s Hospital and the University of Toronto was responsible for the trial coordination, site training, site start-up and activation, essential document management, database development, data management, and statistical analysis. Each site entered study data using Research Electronic Data Capture (REDCap) hosted at St Michael’s Hospital. REDCap is a secure, web-based software platform designed to support data capture for research studies.14,15 The Applied Health Research Centre generated data queries sent to site research staff for resolution.
Participants
Eligible patients were 18 years of age or older and scheduled to undergo liver resection (open or minimally invasive) for a cancer-related indication (ie, precancer, suspicion of cancer, or definite cancer). Patients with severe anemia (hemoglobin levels <9 g/dL), arterial or venous thrombosis within the prior 3 months, active treatment with anticoagulants, disseminated intravascular coagulation, creatinine clearance of less than 30 mL/min, history of seizure disorder, inability to receive blood products, or previous enrollment in the trial were excluded. eTable 1 in Supplement 2 provides complete eligibility criteria.
Procedures
After research staff obtained written informed consent and baseline demographic variables and medical history were collected, participants were randomized to 1 of 2 groups in a 1-to-1 ratio: experimental (tranexamic acid) or placebo (normal saline). The randomization code was generated in random permuted blocks of 2 and 4, stratified by site, using a computer-based randomization program. The research team communicated the randomization code to the institutional research pharmacy that prepared the blinded intervention for each participant, thus ensuring concealed randomization. Participants, clinicians (surgeons, anesthesiologists, and nurses), data collectors, outcome adjudicators, and data analysts were unaware of group allocation. Although the success of blinding was not formally assessed, both tranexamic acid and placebo are transparent, nonviscous solutions; therefore, the ability to distinguish between interventions is very low.
Following induction of anesthesia and prior to surgical incision, an anesthesiologist administered a bolus dose of either 1 g of tranexamic acid in 10 mL of normal saline or matching placebo intravenously. Following administration of the loading bolus, the anesthesiologist began the intravenous maintenance infusion, 1 g of tranexamic acid added to a 250-mL bag of normal saline or matching placebo, administered at a rate of 35 mL/h, until the complete dose was given (approximate duration of infusion, 8 hours).
All other aspects of the operation, including use of topical hemostatic agents other than topical tranexamic acid, were left to the surgeons’ discretion, including the specific techniques used for liver dissection and parenchymal transection. Epidural catheters and/or regional blocks for pain management were at the discretion of the treating teams. Laboratory investigations and VTE prophylaxis were managed per institutional guidelines, wherein the majority of patients receive VTE prophylaxis in the hospital and/or after discharge.
eTable 2 in Supplement 2 shows transfusion guidelines provided to participating sites, outlining indications for red blood cell, platelet, frozen plasma, and cryoprecipitate transfusion, reflecting current best practices.16,17 However, the decision to transfuse was left at the discretion of the medical teams.
Outcomes
The primary outcome was receipt of red blood cell transfusion from start of surgery until postoperative day 7, reported as the proportion of participants receiving transfusion.
Prespecified secondary outcomes included intraoperative blood loss, rigorously measured by adding the net weight of sponges and fluid suction (minus irrigation and intraoperative bile or other fluids in suction/sponge). The surgical team estimated the amount of bile spillage if applicable. Total blood loss (postoperative days 0 to 7) was assessed by the Gross formula,18 which uses the maximum postoperative decrease in the level of hemoglobin adjusted for the weight and height of a participant. Total number of red blood cell units transfused and receipt of other blood products was measured. Study staff at each site assessed postoperative complications within 90 days of surgery using the Clavien-Dindo classification, defining a major complication as grade III or greater.19 Local study staff documented symptomatic VTE confirmed with either computed tomography angiogram (for pulmonary embolism) or venous Doppler ultrasound (for deep venous thrombosis) within 90 days of surgery. Diagnosis of complications (including VTE) was determined by site investigator clinical judgment. Quality of life was measured at baseline, 30 days after surgery, and 90 days after surgery with the validated European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire (QLQ) C30 and the QLQ Liver Metastases Colorectal 21.20 Perioperative mortality (between postoperative days 0 and 7) was assessed. All outcome assessors were blinded to group allocation.
Additional prespecified secondary outcomes will be reported in future publications (recurrence-free survival within 5 years of surgery, overall survival within 5 years of surgery, and economic analysis).
Sample Size Calculation
The trial sample size was calculated based on the primary objective of red blood cell transfusion, an estimated transfusion incidence of 20%. A total sample size of 1230 participants (615 in each group) enabled us to detect a relative risk reduction in blood transfusion of 30% or greater using a 2-sided, 2-sample test of proportions at 80% power and α = .05. Accrual was planned to continue until the sample size of 1230 included participants was achieved, and therefore anticipated recruitment of approximately 1400 participants to allow for postrandomization ineligibility (eg, in cases of enrolled participants not receiving liver resection or study drug) and withdrawals.
Statistical Analysis
All analyses were performed using R version 4.1.1.21 All tests were 2-sided, and P < .05 was considered statistically significant. The statistical analysis plan is available in Supplement 3.
Primary Outcome
The primary efficacy analysis included enrolled participants who proceeded with liver resection and received study drug (tranexamic acid or placebo). Participants who were randomized but did not proceed with liver resection (generally due to medical reasons or a finding of unresectable disease intraoperatively) and participants who did not receive any study drug (due to practical challenges, particularly during the COVID-19 pandemic) were excluded from the primary analysis. Participants were analyzed in the group to which they were randomized.
Receipt of red blood cell transfusion was summarized using proportions and conducted comparisons between groups using χ2 analyses. A logistic regression model was used to estimate the effect of treatment on the odds of red blood cell transfusion. There were no missing data for the primary outcome. The analysis of all randomized participants, which included 1384 participants, is shown in eTable 5 in Supplement 2.
Secondary Outcomes
The effect of treatment on the prespecified secondary outcome measures of intraoperative blood loss and total blood loss was estimated using unadjusted linear regression models. The numbers of units transfused across groups were compared using a binomial distribution to model the zero counts, while a truncated negative binomial distribution was used for the nonzero counts. Complications were reviewed for both the primary analysis cohort and the safety cohort, which included all participants who received study intervention, irrespective of whether liver resection occurred. The effect of treatment on the incidence of symptomatic VTE and composite complications was estimated using logistic regression models. The EORTC QLQ-C30 scores and subscores at day 90 were calculated using the PROscorer R package, version 0.0.4.22 Scores were calculated as long as at least half of the items on the given scale were valid, nonmissing item responses. The QLQ-C30 summary score for a respondent was calculated only if all 13 scales contributing to that score did not have missing data. Scores calculated in the presence of missing items were prorated so that their theoretical minimum and maximum values were identical to those from scores calculated from complete data. Analysis of covariance was used to estimate treatment effect on the total score at 90 days adjusted for the preoperative total score. Perioperative mortality up to postoperative day 7 was assessed and compared between groups using logistic regression. Analysis of all randomized participants is shown in eTables 5 and 6 in Supplement 2.
If there were missing data for secondary outcomes, complete cases were analyzed. For secondary outcomes, the percentage of missing outcome data was at most 4%. Follow-up was complete for perioperative outcomes. For outcomes measured within 90 days of surgery, missing data due to loss to follow-up accounted for less than 1% of the data.
Potential subgroup effects were explored using logistic regression models with treatment by covariate interactions. The following prespecified subgroup effects were explored: sex, extent of liver resection (major vs minor), diagnosis (colorectal liver metastases vs other), presence of cirrhosis, surgical approach (minimal invasive vs open), and use of inflow occlusion (Pringle maneuver). Each interaction was explored using a separate model that included the treatment, the covariate of interest, and an interaction term between the treatment and the covariate of interest. There was a minimal amount of missing data among covariates used for adjusted analyses of the primary outcome (<0.2%).
Interim analysis for efficacy was not conducted. The data and safety monitoring board performed 2 interim safety analyses when 30% (n = 369) and 60% (n = 738) of the study sample had been enrolled. At each review, the data and safety monitoring board examined the following adverse events that occurred within 7 days of administration of the study intervention by assessing the 99% CI based solely on risk difference: (1) serious adverse events with a Common Terminology Criteria for Adverse Events grade of 3 (severe) or greater; (2) VTE and/or seizure graded as 2 (moderate) or greater; and (3) any serious adverse event graded as possibly, probably, or definitely related to tranexamic acid. Stopping rules determined a priori indicated that if the boundary of the 99% CI did not exceed a 1% increase in serious adverse events graded as 3 or greater, the study would continue. Stopping criteria were not met at either review; therefore, the trial continued until the planned sample size was met.
Results
Patients, Follow-Up, and Adherence
To establish feasibility, the trial began with a vanguard phase at 5 representative sites in November 2014, then expanded to a total of 11 sites in Canada and the United States in 2020 and continued until achievement of the intended sample size in August 2022. A total of 3768 patients were assessed for eligibility, of whom 1384 were eligible and were randomized to receive tranexamic acid (n = 694) or placebo (n = 690) (Figure 1). Overall, 47% (1218/2602) of potentially eligible patients declined to participate. Of 139 participants excluded, 11 withdrew consent, 35 did not receive study drug, and 93 did not undergo liver resection, leaving 1245 participants included in the primary analysis. No participants were missing data for the primary outcome, and 90-day outcome data were complete for 1233 participants (99.0%). All participants received the complete bolus of study drug, and the infusion was completed in 1226 participants (98.5%).
Figure 1. Flow of Participants Through the HeLiX Trial.
aOther reasons included participation in another study and/or scheduling problems.
bParticipants may have undergone laparotomy with or without biopsy and/or resection of a different organ.
The baseline and perioperative characteristics of participants were similar between the 2 groups (Table 1). The most common indications for surgery were colorectal liver metastases (56.1%), hepatocellular carcinoma (13.7%), and cholangiocarcinoma (10.0%). Surgeons used minimally invasive techniques in 527 cases (42.4%). Major liver resection (≥4 segments) was performed in 467 (37.5%). Fifty participants (4.0%) required biliary reconstruction and 37 (3.0%) required vascular reconstruction. Surgeons applied inflow occlusion in 481 participants (38.6%).
Table 1. Baseline and Perioperative Participant Characteristics.
| Characteristics | Tranexamic acid (n = 619) | Placebo (n = 626) |
|---|---|---|
| Age, mean (SD), y | 63.1 (11.5) | 63.4 (11.4) |
| Sex, No. (%) | ||
| Female | 241 (38.9) | 254 (40.6) |
| Male | 378 (61.1) | 372 (59.4) |
| Body mass index, mean (SD)a | 28.2 (6.3) | 28.4 (6.3) |
| Medical history, No. (%) | ||
| Hypertension | 290 (46.8) | 308 (49.2) |
| Hypercholesterolemia | 134 (21.6) | 171 (27.3) |
| Diabetes | 110 (17.8) | 140 (22.4) |
| Cirrhosis | 35 (5.7) | 26 (4.2) |
| Prior liver resection | 34 (5.5) | 20 (3.2) |
| Alcohol consumption >2 drinks/d | 30 (4.8) | 36 (5.8) |
| Prior myocardial infarction | 24 (3.9) | 40 (6.4) |
| Other thrombosis | 12 (1.9) | 11 (1.8) |
| Stroke | 6 (1.0) | 15 (2.4) |
| Pulmonary embolism | 5 (0.8) | 7 (1.1) |
| Preoperative laboratory values | ||
| Hemoglobin, mean (SD), g/dL | 13.5 (1.58) | 13.52 (1.52) |
| Hemoglobin <10 g/dL, No. (%) | 13 (2.1) | 8 (1.3) |
| Platelet count, mean (SD), ×103/μL | 226.0 (79.6) | 224.1 (71.9) |
| Platelet count <150 × 103/μL, No. (%) | 94 (15.2) | 84 (13.5) |
| Creatinine, mean (SD), mg/dL | 1.03 (0.25) | 1.03 (0.27) |
| Creatinine >1.36 mg/dL, No. (%) | 19 (3.1) | 27 (4.3) |
| Bilirubin, mean (SD), mg/dL | 0.61 (0.59) | 0.59 (0.54) |
| Bilirubin >1.17 mg/dL, No. (%) | 32 (5.3) | 26 (4.2) |
| International normalized ratio >1.5, No. (%) | 0 | 1 (0.2) |
| Diagnosis, No. (%) | ||
| Colorectal liver metastasis | 353 (57.0) | 346 (55.3) |
| Hepatocellular carcinoma | 88 (14.2) | 82 (13.1) |
| Cholangiocarcinoma | 59 (9.5) | 66 (10.5) |
| Benign tumorb | 38 (6.1) | 42 (6.7) |
| Neuroendocrine metastasis | 33 (5.3) | 41 (6.5) |
| Metastasis from other primary site | 26 (4.2) | 21 (3.4) |
| Gallbladder cancer | 10 (1.6) | 19 (3.0) |
| Other malignant tumorc | 12 (1.9) | 9 (1.4) |
| American Society of Anesthesiologists score, No. (%) | ||
| 1 (Healthy, fit patient) | 1 (0.2) | 3 (0.5) |
| 2 (Mild systemic disease) | 64 (10.3) | 56 (8.9) |
| 3 (Severe systemic disease) | 347 (56.1) | 354 (56.5) |
| 4 (Incapacitating disease that is a constant threat to life) | 207 (33.4) | 213 (34.0) |
| Minimally invasive approach, No. (%)d | 269 (43.5) | 258 (41.3) |
| Major resection (≥4 segments), No. (%) | 229 (37.0) | 238 (38.0) |
| Other viscera resected, No. (%) | ||
| Colorectal | 50 (8.1) | 50 (8.0) |
| Bile duct | 24 (3.9) | 18 (2.9) |
| Small bowel | 17 (2.7) | 24 (3.8) |
| Diaphragm | 14 (2.4) | 12 (2.1) |
| Pancreas | 6 (1.0) | 2 (0.3) |
| Kidney | 4 (0.6) | 1 (0.2) |
| Stomach | 3 (0.5) | 2 (0.3) |
| Other viscerae | 20 (3.5) | 13 (2.2) |
| Biliary duct reconstruction, No. (%) | 26 (4.2) | 24 (3.8) |
| Vascular reconstruction, No. (%)f | 21 (3.4) | 16 (2.6) |
| Pringle maneuver used, No. (%) | 244 (39.4) | 237 (37.9) |
| Total duration of Pringle maneuver, median (IQR), min | 30 (18-45) | 29 (17-44) |
| Duration of procedure, median (IQR), min | 233 (169-313) | 230 (168-299) |
SI conversion factors: To convert creatinine to μmol/L, multiply by 88.4; to convert bilirubin to μmol/L, multiply by 17.104.
Calculated as weight in kilograms divided by height in meters squared.
The category of benign tumors includes participants who had premalignant conditions (eg, adenomas) and participants who were thought to have malignancy based on preoperative imaging but were determined on final pathology to have a benign tumor (eg, hemangiomas).
Other malignant tumors are uncommon primary liver tumors (eg, angiosarcoma).
Minimally invasive approach includes all procedures intended to be performed laparoscopically and robotically.
Other viscera includes other organs resected, such as adrenal organs, appendix, or reproductive organs.
Vascular reconstruction is resection and reconstruction of hepatic artery, portal vein, or hepatic vein.
Primary Outcome
The primary outcome of red blood cell transfusion within 7 days of surgery occurred in 101 participants (16.3%) in the tranexamic acid group and 91 participants (14.5%) in the placebo group (odds ratio, 1.15 [95% CI, 0.84-1.56]; P = .38; absolute difference, 2% [95% CI, −2% to 6%]). There were no appreciable differences in transfusions administered intraoperatively or postoperatively (Table 2).
Table 2. Effects of Tranexamic Acid on Bleeding and Blood Transfusion Within 7 Days.
| Outcomes | Tranexamic acid (n = 619) | Placebo (n = 626) | Absolute difference (95% CI)a | Odds ratio (95% CI) | P value |
|---|---|---|---|---|---|
| Primary outcome: red blood cell transfusion within 7 d of surgery, No. (%) | 101 (16.3) | 91 (14.5) | 0.02 (−0.02 to 0.06) | 1.15 (0.84-1.56) | .38 |
| No. of red blood cell units transfused, median (IQR)b | 2.0 (1.0-4.0) | 2.0 (1.0-3.0) | |||
| Intraoperative red blood cell transfusion, No. (%) | 57 (9.2) | 64 (10.2) | −0.01 (−0.04 to 0.02) | 0.89 (0.61-1.30) | .55 |
| No. of intraoperative red blood cell units transfused per participant, median (IQR)b | 2.0 (1.0-4.0) | 2.0 (1.0-3.0) | |||
| Postoperative red blood cell transfusion, No. (%) | 61 (9.9) | 48 (7.7) | 0.02 (−0.01 to 0.05) | 1.32 (0.89-1.96) | .17 |
| No. of postoperative red blood cell units transfused per participant, median (IQR)a | 2.0 (1.0-2.0) | 1.0 (1.0-2.0) | |||
| Intraoperative transfusion of other blood products, No. (%) | |||||
| Platelets | 11 (1.8) | 7 (1.1) | |||
| Frozen plasma | 17 (2.7) | 12 (1.9) | |||
| Otherc | 45 (7.3) | 52 (8.3) | |||
| Postoperative transfusion of other blood products within 7 d of surgery, No. (%) | |||||
| Platelets | 7 (1.1) | 1 (0.2) | |||
| Frozen plasma | 13 (2.1) | 9 (1.4) | |||
| Otherc | 38 (6.1) | 28 (4.5) | |||
| Intraoperative blood loss, estimated mean (95% CI), mL | 817.3 (733.4-901.2) | 836.7 (753.5-919.9) | −19.4 (−137.5 to 98.8) | .75 | |
| Total intraoperative plus postoperative blood loss, estimated mean (95% CI), mLd | 1504 (1429.5-1578.5) | 1551.2 (1477.3-1625.0) | −47.2 (−152.1 to 57.8) | .38 |
Differences for percentages are expressed in proportions.
Median values and their IQRs are for units received among those who received any red blood cell transfusion.
Includes albumin, cryoprecipitate, and/or fibrinogen.
Calculated using the Gross formula: VL = EBV × (HO, HF/HAV), where VL is allowable blood loss, EBV is patient’s estimated blood volume, HO is patient’s initial hematocrit (or hemoglobin concentration), HF is patient’s minimum allowable hematocrit (or hemoglobin concentration), and HAV is the mean of the initial and minimum allowable hemoglobin concentrations or hematocrit measurements.22
Secondary Outcomes
Measured intraoperative blood loss was similar between groups (tranexamic acid, 817.3 mL; placebo, 836.7 mL; absolute difference, −19.4 mL [95% CI, −137.5 to 98.8 mL]; P = .75), as did total estimated blood loss over 7 days (tranexamic acid, 1504.0 mL; placebo, 1551.2 mL; absolute difference, −47.2 mL [95% CI, −152.1 to 57.8 mL]; P = .38).
The transfusion of other blood products also was similar between groups, both intraoperatively and postoperatively, as was the median number of red blood cells transfused in participants who received transfusion (2 in each treatment group; Table 2).
Participants in the tranexamic acid group were more likely to have postoperative complications (271 [43.8%] vs 237 [37.9%] in the placebo group; odds ratio, 1.28 [95% CI, 1.02-1.60]; P = .03; risk difference, 0.06 [95% CI, 0-0.11]) (Table 3). No significant differences were found in the safety cohort that included all participants who received study intervention irrespective of whether liver resection occurred (eTable 3 in Supplement 2). There were 18 instances of reoperation in the tranexamic acid group and 17 in the placebo group. Further details on types of complications are given in eTable 4 in Supplement 2.
Table 3. Effects of Tranexamic Acid on 90-Day Outcomes.
| Outcomes | No. (%) | Risk difference (95% CI)a | Odds ratio (95% CI) | P value | |
|---|---|---|---|---|---|
| Tranexamic acid (n = 619) | Placebo (n = 626) | ||||
| Any surgical complication | 271 (43.8) | 237 (37.9) | 0.06 (0 to 0.11) | 1.28 (1.02-1.60) | .03 |
| Major complication (Clavien-Dindo grade ≥III19) | 104 (16.8) | 78 (12.5) | 0.04 (0 to 0.08) | 1.42 (1.03-1.95) | .03 |
| Highest Clavien-Dindo grade19 | |||||
| None | 347 (56.1) | 389 (62.1) | |||
| I (No pharmacological or other treatments required other than antiemetics, antipyretics, analgesics, diuretics, electrolytes, or physiotherapy) | 55 (8.9) | 61 (9.7) | |||
| II (Pharmacological treatment other than for grade I) | 113 (18.3) | 98 (15.7) | |||
| IIIa (Treatment requiring local anesthesia) | 43 (6.9) | 32 (5.1) | |||
| IIIb (Treatment requiring general or epidural anesthesia) | 27 (4.4) | 19 (3.0) | |||
| IVa (Single organ dysfunction) | 11 (1.8) | 8 (1.3) | |||
| IVb (Multiorgan dysfunction) | 5 (0.8) | 4 (0.6) | |||
| V (Death) | 18 (2.9) | 15 (2.4) | 0.01 (−0.01 to 0.02) | 1.22 (0.61-2.48) | .58 |
| Any venous thromboembolism | 31 (5.0) | 19 (3.0) | 0.02 (0 to 0.04) | 1.68 (0.95-3.07) | .08 |
| Deep vein thrombosis | 21 (3.4) | 12 (1.9) | |||
| Pulmonary embolism | 11 (1.8) | 7 (1.1) | |||
Differences for percentages are expressed in proportions.
There was no significant increase in VTEs in participants who received tranexamic acid (31 [5.0%] vs 19 [3.0%]; odds ratio, 1.68 [95% CI, 0.95-3.07]; P = .08; risk difference, 0.02 [95% CI, 0-0.04]). In the tranexamic acid group, no adverse events (assessed within 7 days of administration of study intervention) were graded as probably or definitely related to tranexamic acid. In the analysis of all randomized participants (eTable 7 in Supplement 2), additional VTEs were reported in participants who received tranexamic acid (34 [5.0%]). The estimated mean difference in total score for the QLQ-C30 between the tranexamic acid and placebo groups was −0.35 (95% CI, −2.3 to 1.6; P = .73) (eTable 8 in Supplement 2). Overall perioperative (up to postoperative day 7) mortality was 1.8%, with no apparent difference between groups (1.9% with tranexamic acid vs 1.6% with placebo; odds ratio, 1.22 [95% CI, 0.52-2.91]; P = .64; risk difference, 0.003 [95% CI, −0.011 to 0.018]).
The prespecified subgroup analyses demonstrated consistency of findings, with no compelling suggestion of interactions based on sex, underlying diagnosis, presence of cirrhosis, extent of liver resection, use of minimally invasive techniques, or inflow occlusion (P values for interaction ranging between .13 and .89) (Figure 2).
Figure 2. Subgroup-Specific Odds Ratios of Treatment Effects.
Risk differences are expressed in proportions.
Post Hoc Analyses
Restriction to major complications (Clavien-Dindo grade III or greater) yielded similar results (tranexamic acid: 104 [16.8%]; placebo: 78 [12.5%]; odds ratio, 1.42 [95% CI, 1.03-1.95]; P = .03; risk difference, 0.04 [95% CI, 0-0.08]) (eTable 5 in Supplement 2). Overall 90-day mortality was 2.7%, with no apparent difference between groups (tranexamic acid: 2.9%; placebo: 2.4%; odds ratio, 1.22 [95% CI, 0.61-2.48]; P = .58; risk difference, 0.01 [95% CI, −0.01 to 0.02]).
To determine if there was an effect at different centers, an unadjusted logistic regression model of the primary outcome was conducted using study center as a random effect. The results were similar (with center effect: odds ratio, 1.15 [95% CI, 0.84-1.57]; P = .40; without center effect: odds ratio, 1.15 [95% CI, 0.84-1.56]; P = .38), and the interclass correlation coefficient was 0.12. The risk difference with center effect was 0.02 (95% CI, −0.02 to 0.05) and without center effect was 0.02 (95% CI, −0.02 to 0.06).
Discussion
In this trial of patients undergoing liver resection for a cancer-related indication, tranexamic acid did not reduce blood transfusion within 7 days of surgery, intraoperative bleeding, or total blood loss over 7 days. Perioperative complications and major complications were more common in participants who received tranexamic acid.
These results contrast with recent trials in traumatic brain injury,23 cardiac surgery,24 and, most recently, a range of noncardiac surgeries25 (POISE-3). In these large trials, tranexamic acid significantly reduced bleeding, blood transfusion, or disability due to bleeding. The HeLiX trial results also differ from a meta-analysis including many smaller trials of surgery, which demonstrated a reduction in blood transfusion by almost 40% (risk ratio, 0.62 [95% CI, 0.58-0.65]; P < .01).10
Although the meta-analysis and recent POISE-3 trial included a wide range of surgical procedures, only a small proportion of patients had major cancer surgery, in which the mechanism of bleeding and risks of complications differ considerably. When bleeding occurs in liver resection, it tends to be from major blood vessels, rather than the typical ooze that occurs from many other surgical sites such as soft tissue, bone, pericardium, and brain. It is plausible that tranexamic acid is effective in areas of microvascular bleeding or general ooze but ineffective in areas in which bleeding occurs from major blood vessels.
Other trials in which tranexamic acid was ineffective support this hypothesis. The HALT-IT trial randomized 12 009 patients with acute gastrointestinal bleeding to 4 g of tranexamic acid vs placebo over 24 hours.26 Tranexamic acid had no effect on the primary outcome of death due to bleeding (risk ratio, 0.99 [95% CI, 0.82-1.18]). In a recent trial of prophylactic tranexamic acid to prevent obstetrical hemorrhage after cesarean delivery, administration of tranexamic acid did not clearly lower the risk of maternal death or blood transfusion (adjusted relative risk, 0.89 [95% CI, 0.74-1.07]).27
The increase in complications observed in the current trial warrants a cautious approach to adoption of tranexamic acid in surgical procedures in which it has not clearly been shown to be effective. In a systematic review including 197 trials comparing systemic tranexamic acid with placebo or no study intervention, the risk of any adverse event (relative risk, 1.05 [95% CI, 0.99-1.12]) and VTE in particular (relative risk, 0.95 [95% CI, 0.78-1.15]) were similar.28 The POISE-3 trial specifically examined a composite cardiovascular outcome including VTE and, while it failed to meet the prespecified noninferiority margin, the results suggested that any possible difference was small (hazard ratio, 1.02 [95% CI, 0.92-1.14]). In POISE-3, the VTE event rate was very low due to the patient population included, resulting in a 95% CI that includes important differences (0.7% vs 0.6%; hazard ratio, 1.15 [95% CI, 0.69-1.91]). It is possible that advanced cancer, a major risk factor for VTE, increases the baseline risk sufficiently to detect an increased risk with tranexamic acid administration. There may also be a different relative effect of tranexamic acid on VTE in this trial due to the importance of bleeding and blood transfusions on the development of complications.29 If tranexamic acid were effective at reducing bleeding or blood transfusion, the drug might lower the risk of other complications that are also associated with the development of VTEs. The increased risk of VTEs seen in other trials where tranexamic acid was ineffective, such as HALT-IT (risk ratio, 1.85 [95% CI, 1.15-2.98]), support this hypothesis.26
Differences between the HeLiX trial results and those of prior studies, in particular POISE-3, may be at least in part due to chance. This remains possible: the 95% CI around the primary outcome includes a relative odds reduction of 15%. This is, however, appreciably less than benefits seen in prior studies, and there are other reasons to think the difference is real. There is a compelling biological rationale for the difference—the nature of the bleeding that occurs in different surgical conditions. Moreover, the benefits of tranexamic acid are not consistent across surgical procedures, and the other surgical procedures in which randomized clinical trials have failed to show benefit are consistent with this hypothesis.
Strengths of the trial include the pragmatic design, with broad eligibility criteria and patient-important outcomes, ensuring that the results of this trial are applicable to most patients undergoing liver resection. Adherence to study medication was excellent, with the complete bolus of study drug administered in all participants and the infusion completed in 98.5% of participants. Reasons for nonadherence included scheduling changes wherein the intervention was unable to be prepared or provided prior to surgical incision. There were 11 participants who received open-label tranexamic acid postoperatively, 45% (5/11) in the tranexamic acid group and 55% (6/11) in the placebo group. The process of data collection was rigorous and complete, including systematic assessment of blood loss intraoperatively with measurement of suction volumes and sponge weights. No participants were missing data for the primary outcome, and 90-day outcome data were complete in 99.0% of participants.
Limitations
The trial is limited first in its ability to determine differences between specific complications postoperatively due to their heterogeneous nature, without a clear common underlying pathophysiology. Second, the trial lacked power to detect a difference in VTE rates; thus, there is residual uncertainty regarding the impact on tranexamic acid administration on VTE in this population. Third, we did not collect data regarding VTE pharmacoprophylaxis during the trial. Fourth, the study was not powered to detect differences in subgroups; it is possible that there was a subgroup of participants included in the trial who may have benefited from tranexamic acid use.
Conclusions
This trial demonstrated no improvement in blood transfusion, bleeding, or other perioperative outcomes in patients undergoing liver resection for a cancer-related indication treated with tranexamic acid vs placebo. Participants treated with tranexamic acid had increased overall complications and major complications in particular. These considerations should give pause to routine use of tranexamic acid in liver resection. The HeLiX trial results further suggest avenues for future research. In particular, the effect of tranexamic acid on bleeding and postoperative complications in patients undergoing other operations such as gastrectomy, pancreatectomy, sarcoma resection, and other major cancer surgery requires investigation.
Trial Protocol
eTable 1. Detailed eligibility criteria for the Hemorrhage During Liver Resection: Tranexamic Acid (HeLiX) trial
eTable 2. Blood product transfusion guidelines
eTable 3. Incidence of all complications up to post-operative day 90 in safety cohort
eTable 4. Incidence of all complications up to post-operative day 90
eTable 5. Incidence of major (Clavien-Dindo ≥ grade III) complications up to post-operative day 90
eTable 6. Effects of tranexamic acid on bleeding and blood transfusion within seven days in all randomized participants (n=1384)
eTable 7. Effects of tranexamic acid on 90-day outcomes in all randomized participants (n=1384)
eTable 8. European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 results
Statistical Analysis Plan
Data Sharing Statement
References
- 1.Dhar VK, Wima K, Lee TC, et al. Perioperative blood transfusions following hepatic lobectomy: a national analysis of academic medical centers in the modern era. HPB (Oxford). 2019;21(6):748-756. doi: 10.1016/j.hpb.2018.10.022 [DOI] [PubMed] [Google Scholar]
- 2.Kim S, Jung YK, Lee KG, et al. A systematic review and meta-analysis of blood transfusion rates during liver resection by country. Ann Surg Treat Res. 2023;105(6):404-416. doi: 10.4174/astr.2023.105.6.404 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bui LL, Smith AJ, Bercovici M, Szalai JP, Hanna SS. Minimising blood loss and transfusion requirements in hepatic resection. HPB (Oxford). 2002;4(1):5-10. doi: 10.1080/136518202753598672 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Katz SC, Shia J, Liau KH, et al. Operative blood loss independently predicts recurrence and survival after resection of hepatocellular carcinoma. Ann Surg. 2009;249(4):617-623. doi: 10.1097/SLA.0b013e31819ed22f [DOI] [PubMed] [Google Scholar]
- 5.Kusano T, Sasaki A, Kai S, et al. Predictors and prognostic significance of operative complications in patients with hepatocellular carcinoma who underwent hepatic resection. Eur J Surg Oncol. 2009;35(11):1179-1185. doi: 10.1016/j.ejso.2009.04.008 [DOI] [PubMed] [Google Scholar]
- 6.Acheson AG, Brookes MJ, Spahn DR. Effects of allogeneic red blood cell transfusions on clinical outcomes in patients undergoing colorectal cancer surgery: a systematic review and meta-analysis. Ann Surg. 2012;256(2):235-244. doi: 10.1097/SLA.0b013e31825b35d5 [DOI] [PubMed] [Google Scholar]
- 7.Heiss MM, Mempel W, Delanoff C, et al. Blood transfusion-modulated tumor recurrence: first results of a randomized study of autologous versus allogeneic blood transfusion in colorectal cancer surgery. J Clin Oncol. 1994;12(9):1859-1867. doi: 10.1200/JCO.1994.12.9.1859 [DOI] [PubMed] [Google Scholar]
- 8.Kooby DA, Stockman J, Ben-Porat L, et al. Influence of transfusions on perioperative and long-term outcome in patients following hepatic resection for colorectal metastases. Ann Surg. 2003;237(6):860-869. doi: 10.1097/01.SLA.0000072371.95588.DA [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Zakaria S, Donohue JH, Que FG, et al. Hepatic resection for colorectal metastases: value for risk scoring systems? Ann Surg. 2007;246(2):183-191. doi: 10.1097/SLA.0b013e3180603039 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Ker K, Edwards P, Perel P, Shakur H, Roberts I. Effect of tranexamic acid on surgical bleeding: systematic review and cumulative meta-analysis. BMJ. 2012;344(344):e3054. doi: 10.1136/bmj.e3054 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Truong JL, Cyr DP, Lam-McCulloch J, Cleary SP, Karanicolas PJ. Consensus and controversy in hepatic surgery: a survey of Canadian surgeons. J Surg Oncol. 2014;110(8):947-951. doi: 10.1002/jso.23748 [DOI] [PubMed] [Google Scholar]
- 12.Karanicolas PJ, Lin Y, McCluskey S, et al. ; Hepatopancreaticobiliary Community of Surgical Oncologists: Clinical, Evaluative, and Prospective Trials (HPB CONCEPT) Team . Tranexamic acid versus placebo to reduce perioperative blood transfusion in patients undergoing liver resection: protocol for the Haemorrhage During Liver Resection: Tranexamic Acid (HeLiX) randomised controlled trial. BMJ Open. 2022;12(2):e058850. doi: 10.1136/bmjopen-2021-058850 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Schulz KF, Altman DG, Moher D; CONSORT Group . CONSORT 2010 statement: updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340:c332. doi: 10.1136/bmj.c332 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Harris PA, Taylor R, Minor BL, et al. ; REDCap Consortium . The REDCap Consortium: building an international community of software platform partners. J Biomed Inform. 2019;95:103208. doi: 10.1016/j.jbi.2019.103208 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377-381. doi: 10.1016/j.jbi.2008.08.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Carson JL, Stanworth SJ, Guyatt G, et al. Red blood cell transfusion: 2023 AABB international guidelines. JAMA. 2023;330(19):1892-1902. doi: 10.1001/jama.2023.12914 [DOI] [PubMed] [Google Scholar]
- 17.Raasveld SJ, de Bruin S, Reuland MC, et al. ; InPUT Study Group . Red blood cell transfusion in the intensive care unit. JAMA. 2023;330(19):1852-1861. doi: 10.1001/jama.2023.20737 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Gross JB. Estimating allowable blood loss: corrected for dilution. Anesthesiology. 1983;58(3):277-280. doi: 10.1097/00000542-198303000-00016 [DOI] [PubMed] [Google Scholar]
- 19.Clavien PA, Barkun J, de Oliveira ML, et al. The Clavien-Dindo classification of surgical complications: five-year experience. Ann Surg. 2009;250(2):187-196. doi: 10.1097/SLA.0b013e3181b13ca2 [DOI] [PubMed] [Google Scholar]
- 20.European Organisation for Research and Treatment of Cancer . EORTC Quality of Life. https://qol.eortc.org/
- 21.R Core Team . R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing; 2021. https://www.R-project.org/
- 22.Baser R. PROscorer: Functions to Score Commonly-Used Patient-Reported Outcome (PRO) Measures and Other Psychometric Instruments. Comprehensive R Archive Network; 2023. https://cran.r-project.org/web/packages/PROscorer/index.html
- 23.CRASH-3 Trial Collaborators . Effects of tranexamic acid on death, disability, vascular occlusive events and other morbidities in patients with acute traumatic brain injury (CRASH-3): a randomised, placebo-controlled trial. Lancet. 2019;394(10210):1713-1723. doi: 10.1016/S0140-6736(19)32233-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Myles PS, Smith JA, Forbes A, et al. ; ATACAS Investigators of the ANZCA Clinical Trials Network . Tranexamic acid in patients undergoing coronary-artery surgery. N Engl J Med. 2017;376(2):136-148. doi: 10.1056/NEJMoa1606424 [DOI] [PubMed] [Google Scholar]
- 25.Devereaux PJ, Marcucci M, Painter TW, et al. ; POISE-3 Investigators . Tranexamic acid in patients undergoing noncardiac surgery. N Engl J Med. 2022;386(21):1986-1997. doi: 10.1056/NEJMoa2201171 [DOI] [PubMed] [Google Scholar]
- 26.HALT-IT Trial Collaborators . Effects of a high-dose 24-h infusion of tranexamic acid on death and thromboembolic events in patients with acute gastrointestinal bleeding (HALT-IT): an international randomised, double-blind, placebo-controlled trial. Lancet. 2020;395(10241):1927-1936. doi: 10.1016/S0140-6736(20)30848-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Pacheco LD, Clifton RG, Saade GR, et al. ; Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network . Tranexamic acid to prevent obstetrical hemorrhage after cesarean delivery. N Engl J Med. 2023;388(15):1365-1375. doi: 10.1056/NEJMoa2207419 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Khair S, Perelman I, Yates J, et al. Exclusion criteria and adverse events in perioperative trials of tranexamic acid in cardiac surgery: a systematic review and meta-analysis. Can J Anaesth. 2019;66(10):1240-1250. doi: 10.1007/s12630-019-01393-w [DOI] [PubMed] [Google Scholar]
- 29.Hallet J, Mahar AL, Nathens AB, et al. The impact of perioperative blood transfusions on short-term outcomes following hepatectomy. Hepatobiliary Surg Nutr. 2018;7(1):1-10. doi: 10.21037/hbsn.2017.05.07 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial Protocol
eTable 1. Detailed eligibility criteria for the Hemorrhage During Liver Resection: Tranexamic Acid (HeLiX) trial
eTable 2. Blood product transfusion guidelines
eTable 3. Incidence of all complications up to post-operative day 90 in safety cohort
eTable 4. Incidence of all complications up to post-operative day 90
eTable 5. Incidence of major (Clavien-Dindo ≥ grade III) complications up to post-operative day 90
eTable 6. Effects of tranexamic acid on bleeding and blood transfusion within seven days in all randomized participants (n=1384)
eTable 7. Effects of tranexamic acid on 90-day outcomes in all randomized participants (n=1384)
eTable 8. European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 results
Statistical Analysis Plan
Data Sharing Statement