Acute severe gastrointestinal bleeds (GIBs) are a common cause of death worldwide, with a majority of hemorrhages being attributed to the upper gastrointestinal tract. Patients are often hemodynamically unstable and need urgent resuscitation with blood product transfusion and medical, surgical, or endoscopic treatment. A tremendous effort has been made to use medical therapy to reduce the burden of, and/or eliminate, GIBs while avoiding invasive endoscopy. One potential therapy is to minimize fibrinolysis with tranexamic acid (TXA).
In the widely publicized CRASH-2 trial and recent CRASH-3 trial, the trial collaborators showed that early use of TXA reduces surgical bleeding and death related to bleeding from traumatic injury and has reduced the burden of intracranial bleeding after head injury (CRASH-2 trial collaborators et al. Lancet. 2010;376:23-32; CRASH-3 trial collaborators et al. Lancet. 2019;394:1713-1723). Additionally, Bennett et al (Cochrane Database Syst Rev. 2014;11:CD006640) demonstrated a reduction in all-cause mortality for 1654 patients when TXA was used for GIBs (RR 0.61; p = 0.01). Based on all these findings, the HALT-IT trial (HALT-IT Trial Collaborators. Lancet. 2020;395:1927-1936) investigated whether early use of TXA would reduce overall mortality in the GIB population while limiting the burden of thromboembolic events, a known complication of TXA.
The HALT-IT trial was an international randomized double-blind placebo-controlled trial that included 12,009 adult patients with upper or lower GIBs and a clinical diagnosis of significant bleeding (hypotension, tachycardia, shock, with a risk of death) from 164 hospitals in 15 countries. Consent was optional, based on patient mental state and urgency of care. Randomization was achieved by a third party and involved intention to treat with either a standard TXA dose of a 1-g bolus over 10 minutes with an additional 3-g infusion at 125 mL/hr for 24 hours versus placebo. Outcome was collected at death, hospital discharge, or 28 days from randomization (whichever came first) with the primary outcome being death due to bleeding within 5 days of randomization. Secondary outcomes were a multitude of events, including all-cause mortality, thromboembolic events, and rebleeding.
Of the 5994 (49.9%) patients who received TXA and 6015 (50.1%) who received placebo, 99.5% received the first dose of allocated treatment. Death attributed to bleeding occurred in 222 (3.7%) in the TXA group versus 226 (3.8%) in the placebo group (RR 0.99; CI, 0.82-1.18). When adjusting for covariates and removing open-label TXA use, there was no difference in outcome. When comparing death due to bleeding within 24 hours (2.1% for TXA vs. 2.0% for placebo), death within 28 days (4.2% for TXA vs. 4.4% for placebo), and death from all causes (9.5% for TXA vs. 9.2% for placebo), there existed no statistical significant difference. The risk for thromboembolic events was overall not significant (0.4% for TXA vs. 0.3% for placebo); however, the risk of venous thromboembolic events was higher in the TXA group (0.8% vs. 0.4%; RR, 1.85; CI, 1.15 to 2.98). Additionally, the risk of seizures was higher in the TXA group (0.6% vs. 0.4%; RR, 1.73; CI, 1.03 to 2.93).
The HALT-IT trail was a large double-blind randomized controlled trial that demonstrated that TXA is no better than placebo in the prevention of death or rebleeding from a significant GIB. The study also demonstrated that the risk for venous thromboembolic events and seizures was increased with TXA use in this population. One potential explanation for these findings is that, when used within 3 hours of trauma or postpartum hemorrhage, there is a time-limited benefit with the use of TXA. The GIB population often presents with an unknown time of onset for the GIB, so the contribution of fibrinolysis to this hemorrhage is unknown. Additionally, nearly half of the GIBs in this study were due to variceal bleeds, a population with a mixed fibrinolytic picture.
Limitations of the study include the clinician-specific definitions of a significant GIB and the clinician-specific determination that GIB was the cause of death. The dosing of TXA for GIBs is unknown, and the dosing chosen in the study is higher than that used in previous studies in traumatic or surgical bleeding; thus, the risk for embolic events is expected to increase and the overall relation to mortality is unknown. Furthermore, the authors amended their primary outcome during the trial, citing the low likelihood that TXA would reduce mortality due to bleeding beyond 5 days of randomization. This required an increased sample size and research study adjustment. More research is needed to determine whether the use of TXA to prevent death from GIBs is beneficial.
Jeff Pepin, MD, is an assistant professor of emergency medicine at the University of Minnesota, Minneapolis, Minnesota, USA.
James H. Lantry III, MD, is an adjunct assistant professor of emergency and critical care medicine at the University of Maryland Medical Center in Baltimore, Maryland, USA. Dr. Lantry is an editor of Concise Critical Appraisal.