HALT-IT: Has our hero, Tranexamic Acid (TXA), finally met its match?

Episode recap:

When last we left our hero, known throughout the globe as Clotting Wonder, it was after a grueling battle with traumatic intracranial hemorrhage. Having been put to the test by post-partum hemorrhage in the WOMAN trial, our hero vanquished the evil villain of traumatic hemorrhage (in the exhilarating CRASH-2 study), achieving a 1.5% reduction in all-cause mortality [1,2]. Some have come to expect much from our caped crusader. Who could forget episode CRASH-3, in which traumatic intracranial hemorrhage proved itself a worthy beast [3]?

Fresh from this, its closest and most grueling of all battles, TXA found itself facing down one of its most diabolic nemesis yet: the dreaded, foulest of foul, gastrointestinal bleeding (GIB)! Our hero steeled itself for a pitched battle, and now, as the dust settles from the barrage of blows, let us see if we can suss out who has triumphed in this, the HALT-IT trial [4].

Study design

Patients

HALT-IT was a large, multi-center, placebo-controlled trial that enrolled around 12,000 patients in 164 hospitals among 15 countries (of note, did not include the USA) [4]. Patients were included in the study if they were adults (>16 or >18 years old depending on the country) and the responsible clinician was “substantially uncertain” whether to use TXA. GI bleeding also had to be “significant”, determined clinically as a risk of bleeding to death and included: patients with hypotension, tachycardia, or signs of shock, or those likely to need transfusion or urgent endoscopy or surgery.

Intervention

Patients were randomized to receive one gram TXA bolus followed by three grams of TXA in one liter of isotonic fluid provided at a rate of 125 ml/hr for 24 hours. Allocation was concealed from patients, caregivers, and those assessing outcomes, and patients were block-randomized in groups of eight.

Comparison

Placebo

Outcomes

-Primary outcome: death due to bleeding within five days 

-Secondary outcomes: death due to bleeding within 24 hours and 28 days; rebleeding within 24 hours, 5 days, and 28 days; surgery or radiological intervention; blood product transfusion, thromboembolic events, seizures, other complications, days in an intensive care unit, and functional status.

Pros

This study addressed an important and clinically relevant question that inquiring minds want to know. The study population was very large and drop out was minimal; only 20 patients (about 0.3%) were not included in the analysis in each study arm. The study also boasted a primary outcome that was relatively straightforward and sort of patient-centered, with many, many secondary outcomes of… lesser patient-centeredness. The intervention and placebo groups were very well-matched in regards to baseline characteristics with virtually no differences between age, sex, suspected location of bleeding, presence of hematemsis, fresh blood per rectum/melena, initial systolic blood pressure and heart rate, comorbid conditions, etc. Finally, study authors had no reported conflicts of interest, and study funding had no role in study design, data collection, analysis, interpretation, or writing.

Limitations

One glaring limitation of the study is the fact that the authors initially chose all-cause mortality as their primary outcome, but changed it to disease-specific mortality secondary to bleeding. Over five days, approximately 450 patients died from bleeding, while 325 patients died of other causes. This raises a lot of questions. Determining the cause of death can be hairy, and it’s hard to know in which way this could skew the data. This is one of the problems with disease-specific mortality as an outcome in general, and it is a recurring theme when reviewing other papers concerning TXA (including the WOMAN and CRASH-3 trials) [1,3]. Disease-specific mortality is a completely different, and far less patient-centered outcome than all-cause mortality. I don’t know many people who care much if they die from bleeding vs. anything else, do you? I think it’s more the whole dying thing itself. Although to be fair, given the choice between bleeding to death and being crushed to death by a plate glass window falling 40 stories in a Final Destination-type scenario, I can see where those distinctions can become significant.

Results

Ultimately this was, through and through, a negative study. None of the primary or secondary outcomes surrounding death or rebleeding showed statistically significant differences between treatment and placebo groups. Subgroup analysis did not reveal many hints of possible benefit either; patients with Rockall scores of 1-2 did show a reduction in risk of death due to bleeding within 5 days (0.64), but the 95% confidence interval is wide and crosses the null value (0.35 to 1.18) suggesting uncertainty in the point estimate. Finally, while the occurrences of deep vein thrombosis (DVT) or pulmonary embolism (PE) between groups did not reach statistical significance individually, when taken together, a statistically significant increase in venous thromboembolism was observed among the treatment group (0.8%) vs. the placebo group (0.4%).

Conclusions: IS THIS THE END FOR OUR HERO??

It is always difficult to see a hero falter, and yet I think in HALT-IT, it cannot be avoided. This was a large, well-executed trial with virtually no important differences in baseline characteristics between the study arms. None of the primary or secondary outcomes revealed any statistically or clinically significant differences, and unfortunately, it looks as if GI bleeding may have won the day.

Many factors could have shaped the results of this trial, and while the overall outcome is inarguably negative for all comers with GIB, it is still possible that select patients may benefit from TXA. Maybe the patients were not sick enough to see benefit. A more direct outcome comparison is not available from the two articles, but the overall mortality in the placebo arm of HALT-IT at 28-days was 9.2%, and the overall mortality rate in the Rockall score validation study during hospitalization was 14% [5].

Patients in both groups received an average of three units of whole blood or red cells. While this is not a small amount, is it possible that benefit would be seen in those patients with higher transfusion requirements? A similar argument has been made about the absence of benefit for TXA in the WOMAN trial, in which more than half of the included patients had less than one liter of blood loss, and mortality was similarly low (2-3%) in both treatment and placebo groups [1]. 

While it does seem that TXA may have once again met a bleeding villain it could not best, one thing should be said. CRASH-3, an equally humongous study that looked at the use of TXA in the setting of traumatic intracranial hemorrhage, used a protocol of one gram TXA bolus followed by one gram of TXA infused over eight hours and showed NO difference in the occurrence of DVT or PE between study and placebo groups [3]. The HALT-IT authors provide two hypotheses for the statistically significant increase in venous thromboembolism in their study: 

Firstly, the increased rate of venous thromboembolism may have been secondary to the prolonged maintenance infusion of three grams over 24 hours. This protocol was chosen because rebleeding within 24 hours is a significant problem in patients with GIB, and as TXA has a short half-life, a 24-hour infusion would be appropriate and a similar protocol was used in prior studies of TXA in GIB [6]. 

Secondly, the authors also hypothesize that given the large representation of liver disease patients in their trial (~45% of each study arm was composed of patients with suspected variceal bleeding), a predisposition to TXA-related venous thromboembolism in this population may also have played some role in the difference. The data to support this hypothesis is not presented in the article. However, the authors also point out that chirrotics with variceal bleeding have a mixed fibrinolytic phenotype, with the sickest patients trending towards hypofibronlysis. While the authors speculate that the increased rate of venous thromboembolism may be due to the effects of TXA in those chirrotic patients with variceal bleeding with this hypofibrinolytic phenotype, these may be the exact patients TXA could benefit the most. It would be interesting to see a trial specifically targeting variceal bleed patients using the one gram bolus, one gram maintenance infusion protocol popularized in the CRASH studies. However as it stands, I don’t think TXA has a role in the typical patient with GIB.

And so, as GI Bleeding guffaws and Ceftriaxone chuckles to himself, TXA, defeated, trods softly back to the hero’s headquarters, where Pantoprazole sits manning the monitors. In the pantheon of heroes who have gone up against GI bleeding, few have escaped unscathed, but TXA holds it’s head high and proudly tends to its wounds... it knows it is needed elsewhere.

To be continued….

References

1. 1. Woman trial collaborators. Effect of early tranexamic acid administration on mortality, hysterectomy, and other morbidities in women with post-partum haemorrhage (WOMAN): an international, randomised, double-blind, placebo-controlled trial. The lancet. 2017; 389: 2105–16. DOI: https://doi.org/10.1016/S0140-6736(17)30638-4.
2. 2. Crash-2 trial collaborators. Effects of tranexamic acid on death, vascular occlusive events, and blood transfusion in trauma patients with significant haemorrhage (CRASH-2): a randomised, placebo-controlled trial. The lancet. 2010; 376: 23–32. DOI: https://doi.org/10.1016/S0140-6736(10)60835-5.
3. 3. 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. The lancet. 2019; 394: 1713–23. DOI: https://doi.org/10.1016/S0140-6736(19)32233-0.
4. 4. 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. The lancet. 2020; 395: 1927–36. DOI: https://doi.org/10.1016/S0140-6736(20)30848-5.
5. 5. Vreeburg et al. Validation of the Rockall risk scoring system in upper gastrointestinal bleeding. Gut. 1999; 44: 331–335. DOI: 10.1136/gut.44.3.331.
6. 6. Bennett et al. Tranexamic acid for upper gastrointestinal bleeding (review). Cochrane database of systematic reviews. 2014; Issue 11. DOI: https://doi.org/10.1002/14651858.CD006640.pub3.

Edited by: Robby Allen, PGY-3