Tranexamic Acid in Gastrointestinal Bleeding A.35
Tranexamic Acid in Gastrointestinal Bleeding A.35
Tranexamic Acid in Gastrointestinal Bleeding A.35
G
astrointestinal bleeding (GIB) can occur anywhere along the gastro-
intestinal tract and is traditionally classified by location, upper GIB
or lower GIB, or by etiology, nonvariceal or variceal. GIB is associated
with significant morbidity and mortality. Although advancement in manage- Copyright © 2021 by the Society of
ment of GIB outcomes has improved (1), the mortality associated with GIB Critical Care Medicine and Wolters
remains at 10–11% (1, 2). Kluwer Health, Inc. All Rights
Optimal management of GIB requires a multidisciplinary and multimodal Reserved.
approach that includes hemodynamic resuscitation, the use of blood transfusions, DOI: 10.1097/CCM.0000000000005362
and outcome measurements. We evaluated statistical Cochrane Risk of Bias Tool (S.J.W.O., J.C.D.) (10). We
heterogeneity using the I2 statistic, considering values assessed each study based upon randomization, allo-
of ~30–60% moderate heterogeneity and values greater cation concealment, blinding of participants and per-
than ~60% substantial heterogeneity (10). Where sta- sonnel, loss to follow-up, selective outcome reporting,
tistical heterogeneity existed, we conducted sensi- whether or not analysis was by intention-to-treat, and
tivity analyses of fixed versus random effects models any other identifiable bias. The overall certainty of ev-
Downloaded from http://journals.lww.com/ccmjournal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
to assess the impact of heterogeneity upon effect esti- idence for each outcome was assessed using Grading
mates, recognizing that random-effects estimates may of Recommendations Assessment, Development and
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 11/15/2023
produce exaggerated effects, especially in the presence Evaluation (GRADE) methodology (12). GRADE con-
of small studies (11). siders multiple factors, including study design, risk of
Study-level risk of bias assessment was performed bias, inconsistency, indirectness, imprecision, as well
by two independent reviewers using a modified as risk of publication bias.
Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.
reviewed using the full study text. We identified 12 did not meet the optimal information size, assuming a
eligible studies for inclusion in the review, including true RR of 0.5 with a control group rate of 16%.
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 11/15/2023
CI, 0.99–1.01; p = 0.48 and ARD, 0%; 95% CI, –0.9 to for several reasons. First, these results are driven by
0.9; high certainty) or low-dose IV/enteral form (RR, smaller, single-center studies, and although statistically
1.03; 95% CI, 0.93–1.13; p = 0.61 and ARD, 2.1%; 95% significant, the results barely meet optimal informa-
CI, –4.1 to 9.0; moderate certainty) (Supplement 2H, tion size and are thus imprecise. As demonstrated by
http://links.lww.com/CCM/G911). We had lower cer- HALT-IT, promising pooled effect estimates of small
tainty in the low-dose IV/enteral TXA arm due to single-center trials are often smaller than those demon-
Downloaded from http://journals.lww.com/ccmjournal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
imprecision as the 95% CI did not exclude clinically strated in larger, more rigorous clinical trials due to both
significant harm nor benefit. HALT-IT also reported increased precision and stability of estimates, as well as
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 11/15/2023
rates of plasma transfusion, platelet transfusion, and associated methodological rigor which usually comes
the mean volumes of transfusions and did not find any with larger trials. Second, studies of low-dose IV or en-
differences between the two arms (7). teral TXA did not report adverse events, such as arte-
rial or venous thromboembolism or seizure. Although
the risks of these adverse events may be lower with low-
DISCUSSION
dose, shorter IV administration and enteral administra-
In summary, we found high certainty evidence tion of TXA, there is little evidence to support this claim
(Supplement 3, http://links.lww.com/CCM/G912) from the GIB literature. However, use of similar doses in
that high-dose IV TXA given over 24 hours does not hundreds of RCTS in trauma, postpartum hemorrhage,
reduce mortality, rebleeding, or surgical interventions and prevention of bleeding in surgery has not found
in patients with GIB but does result in small increases any evidence of increased risk of venous thromboem-
in seizures and venous thromboembolism. Although bolism (24). Third, these trials have been performed
the absolute risk of these events is very small (< 1%), before the publications supporting a restrictive RBC
these risks appear to outweigh any measurable benefit transfusion policy for GIB, which appears to reduce
of high-dose IV TXA. the number of rebleeds (25). It is unknown whether
Despite the substantial statistical heterogeneity in low-dose IV or enteral-only TXA would still be useful
studies reporting mortality, we have high confidence in the context of modern restrictive transfusion prac-
in these results. The statistical heterogeneity is driven tices and endoscopic management including dual mo-
by the differences between the very precise results of dality interventions to control bleeding (e.g., injecting
the HALT-IT trial and those of earlier trials. The high with epinephrine and applying endoclips to the bleed-
precision of HALT-IT will increase statistical hetero- ing lesion) that would lead to a reduction in rebleeding.
geneity which may not be clinically meaningful. With Fourth, the use of enteral administration of TXA may
the exception of the study by Barer et al (14), in the lead to a bleeding reduction by topical application to the
remaining three studies, the 95% CIs overlap the point bleeding lesion, which may explain the differences seen
estimate of HALT-IT. Second, the rigorously con- between IV and enteral dosing of the medication. The
ducted HALT-IT trial is most likely to have an unbi- association between local hyperfibrinolysis and peptic
ased estimate of effect. It is reassuring that sensitivity ulcer disease has been well recognized for decades and
analyses using a random-effects model, which weigh is the rationale for the use of antifibrinolytic agents such
the smaller studies more heavily, similarly do not dem- as TXA (27). It has been suggested that variceal bleed-
onstrate statistically significant reductions in mor- ing in patients with cirrhosis is also associated with local
tality with TXA. Last, it is implausible that high-dose fibrinolysis (28), and it is quite clear at this point that
extended-use IV TXA would result in a reduction in systemic fibrinolysis occurs in cirrhosis and is associ-
mortality without significantly impacting rebleeding, ated with hemorrhage (29, 30). Thus, although there is
need for surgery, or transfusion. a strong theoretical basis for topical TXA to counteract
In contrast, although studies evaluating low-dose IV local fibrinolysis in variceal hemorrhage, there is to date
or enteral-only TXA did not demonstrate reduction little direct evidence on the topic. Further evidence is
in mortality, they did show a reduction in rebleeding, necessary before low-dose IV or enteral TXA should
need for surgical intervention, and volume of blood be considered in widespread clinical use. There may re-
transfused. We encourage cautious interpretation of main a role for enteral TXA, if future multicenter trials
these results, despite moderate certainty in GRADE, demonstrate its safety and effectiveness.
Other areas of uncertainty include the use of TXA 5 Guy’s & St Thomas’ Thrombosis & Haemophilia Centre,
in subtypes of GIB, including lower GIB and variceal London, United Kingdom.
bleeding. Only one small study (23) specifically evalu- 6 Department of Anesthesiology Emergency and Intensive
Care Medicine, Fondazione Policlinico Universitario
ated enteral TXA in lower GIB, and HALT-IT only A.Gemelli IRCCS, Rome, Italy.
included 11% of patients with suspected lower GIB; 7 Istituto di Anestesiologia e Rianimazione, Università
thus, there remains uncertainty as to whether or not Cattolica del Sacro Cuore, Rome, Italy.
Downloaded from http://journals.lww.com/ccmjournal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
TXA may play a role in this subset of patients whose 8 Department of Anesthesiology, Amsterdam UMC, location
bleeding may often be less brisk but still cause signifi- AMC, Amsterdam, The Netherlands.
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 11/15/2023
cant morbidity. Nearly half of patients in HALT-IT had 9 Department of Intensive Care, Amsterdam UMC, Amsterdam,
The Netherlands.
suspected variceal bleeding, and this group accounted
Members of the ESICM Transfusion Taskforce are: Alexander
for the majority of deaths and a higher rate of venous
Vlaar, Joanna C. Dionne, Sanne de Bruin, Marije Wijnberge, S.
thromboemobolism (VTE). Jorinde Raasveld, Frank E.H.P. van Baarle, Massimo Antonelli,
Last, it is worth considering this study in the con- Cecile Aubron, Jacques Duranteau, Nicole P. Juffermans, Jens
text of other large trials evaluating the efficacy of Meier, Gavin J. Murphy, Riccardo Abbasciano, Marcella Müller,
Marcus Lance, Nathan D. Nielsen, Herbert Schöchl, Beverly J.
TXA in acutely bleeding patients. Although Clinical Hunt, Maurizio Cecconi, and Simon Oczkowski.
Randomisation of an Antifibrinolytic in Significant Members of the GUIDE Group are: Joanna Dionne, Simon
Haemorrhage 2 (CRASH-2) demonstrated a small Oczkowski, Waleed Alhazzani, Emilie Belley-Cote, Erick Duan,
but significant reduction in all-cause mortality Bram Rochwerg, John Centofanti, Mats Junek, Lawrence
(3), other trials including CRASH-3, Tranexamic Mbuagbaw, Mohammed Alsharani, Fayez Alshamsi, Wojciech
Szczeklik, Roman Jaeschke, Karin Dearness, Morten Hylander,
acid for hyperacute primary Intracerebral hem- Ekkehard Kasper, Mark Soth, Dan Peri, Lehana Thabane, and
orrhage (TICH-2), and the World Maternal Gordon Guyatt.
Antifibrinolytic Trial (WOMAN) have demon- Supplemental digital content is available for this article. Direct
strated much more modest effects (4–6). However, URL citations appear in the printed text and are provided in the
HTML and PDF versions of this article on the journal’s website
hundreds of RCTs using either IV or topical TXA
(http://journals.lww.com/ccmjournal).
in the prevention of bleeding during and after sur-
Drs. Dionne and Oczkowski are cofirst authors.
gery have consistently shown efficacy and reduc-
Dr. Oczkowski received payment as a methodology consultant
tion of bleeding by about one third of expected (26). for the American Thoracic Society. The remaining authors have
Our data suggest that the initial enthusiasm of TXA as disclosed that they do not have any potential conflicts of interest.
an effective and safe treatment for managing all types For information regarding this article, E-mail: dionnejc@mcmas-
of established bleeding needs reconsideration; it may ter.ca
be effective only in certain clinical scenarios of bleed-
ing. More importantly, we feel that high-dose TXA
over 24 hours has an unacceptable risk of seizure and
REFERENCES
VTE as demonstrated in HALT-IT. Furthermore, rec-
ognizing that higher dose does not increase reduction 1. Barkun AN, Almadi M, Kuipers EJ, et al: Management of non-
variceal upper gastrointestinal bleeding: Guideline recommen-
in bleeding in surgical patients but is associated with dations from the International Consensus Group. Ann Intern
seizure; we recommend caution in using high dose in Med 2019; 171:805–822
any clinical scenario (26). Although TXA may help 2. Stanley AJ, Laine L: Management of acute upper gastrointes-
to mediate the balance between bleeding and throm- tinal bleeding. BMJ 2019; 364:l536
bosis, the importance of dose and length of time it is 3. Shakur H, Roberts I, Bautista R, et al; CRASH-2 trial collabo-
rators: Effects of tranexamic acid on death, vascular occlusive
administered will impact upon patient outcomes. events, and blood transfusion in trauma patients with signifi-
cant haemorrhage (CRASH-2): A randomised, placebo-con-
trolled trial. Lancet 2010; 376:23–32
1 Department of Medicine, McMaster University, Hamilton,
ON, Canada. 4. Shakur H, Roberts I, Fawole B, et al: Effect of early tranexamic
acid administration on mortality, hysterectomy, and other mor-
2 Department of Health Research Methods, Evidence, and
bidities in women with post-partum haemorrhage (WOMAN):
Impact, McMaster University, Hamilton, ON, Canada.
An international, randomised, double-blind, placebo-controlled
3 Guidelines in Intensive Care, Development, and Evaluation trial. Lancet 2017; 389:2105–2116
(GUIDE), Hamilton, ON, Canada. 5. Sprigg N, Flaherty K, Appleton JP, et al; TICH-2 Investigators:
4 King’s Healthcare Partners, London, United Kingdom. Tranexamic acid for hyperacute primary intracerebral
haemorrhage (TICH-2): An international randomised, pla- 17. Biggs JC, Hugh TB, Dodds AJ: Tranexamic acid and upper
cebo-controlled, phase 3 superiority trial. Lancet 2018; gastrointestinal haemorrhage–a double-blind trial. Gut 1976;
391:2107–2115 17:729–734
6. Crash T: Effects of tranexamic acid on death, disability, vas- 18. Bagnenko SF, Verbitskii VG: Antifibrinolitic therapy for the
cular occlusive events and other morbidities in patients with treatment of massive ulcerative gastro-intestinal bleedings.
acute traumatic brain injury (CRASH-3): A randomised, pla- Khirurgiia (Mosk) 2011; (4):42–46
cebo-controlled trial. Lancet 2019; 394:1713–1723 19. Cormack F, Chakrabarti RR, Jouhar AJ, et al: Tranexamic
Downloaded from http://journals.lww.com/ccmjournal by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCy
7. Roberts I, Shakur-Still H, Afolabi A, et al: Effects of a high- acid in upper gastrointestinal haemorrhage. Lancet 1973;
dose 24-h infusion of tranexamic acid on death and thrombo- 1:1207–1208
wCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8KKGKV0Ymy+78= on 11/15/2023
embolic events in patients with acute gastrointestinal bleeding 20. Bergqvist D, Dahlgren S, Hessman Y: Local inhibition of the
(HALT-IT): An international randomised, double-blind, placebo- fibrinolytic system in patients with massive upper gastrointes-
controlled trial. Lancet 2020; 395:1927–1936 tinal hemorrhage. Ups J Med Sci 1980; 85:173–178
8. Vlaar AP, Oczkowski S, de Bruin S, et al: Transfusion strategies 21. Hawkey GM, Cole AT, McIntyre AS, et al: Drug treatments in
in non-bleeding critically ill adults: A clinical practice guide- upper gastrointestinal bleeding: Value of endoscopic findings
line from the European Society of Intensive Care Medicine. as surrogate end points. Gut 2001; 49:372–379
Intensive Care Med 2020: 46;673–696
22. Saidi HS, Ghavami S, Mirafzal A, et al: Role of intra-gastric
9. Bennett C, Klingenberg SL, Langholz E, et al: Tranexamic acid
tranexamic acid in management of acute upper gastrointes-
for upper gastrointestinal bleeding. Cochrane Database Syst
tinal bleeding. IIOAB J 2017; 8:76–81
Rev 2014; 11:CD006640
23. Smith SR, Murray D, Pockney PG, et al: Tranexamic acid for
10. Higgins JP, Green S: Cochrane Handbook for Systematic
lower GI hemorrhage: A randomized placebo-controlled clin-
Reviews of Interventions. Chichester, England, Hoboken,
ical trial. Dis Colon Rectum 2018; 61:99–106
2008
11. Schroll JB, Moustgaard R, Gøtzsche PC: Dealing with sub- 24. Hunt BJ: The current place of tranexamic acid in the manage-
stantial heterogeneity in Cochrane reviews. Cross-sectional ment of bleeding. Anaesthesia 2015; 70(Suppl 1):50–53, e18
study. BMC Med Res Methodol 2011; 11:22 25. Villanueva C, Colomo A, Bosch A, et al: Transfusion strategies
12. Guyatt GH, Oxman AD, Vist GE, et al; GRADE Working Group: for acute upper gastrointestinal bleeding. N Engl J Med 2013;
GRADE: An emerging consensus on rating quality of evidence 368:11–21
and strength of recommendations. BMJ 2008; 336:924–926 26. Ker K, Edwards P, Perel P, et al: Effect of tranexamic acid on
13. Engqvist A, Broström O, von Feilitzen F, et al: Tranexamic acid in surgical bleeding: Systematic review and cumulative meta-
massive haemorrhage from the upper gastrointestinal tract: A analysis. BMJ 2012; 344:e3054
double-blind study. Scand J Gastroenterol 1979; 14:839–844 27. Stenberg B, Bylock A, Risberg B: Local fibrinolytic activity and
14. Barer D, Ogilvie A, Henry D, et al: Cimetidine and tranexamic its structural basis in experimental gastric ulcers. Eur J Clin
acid in the treatment of acute upper-gastrointestinal-tract Invest 1983; 13:297–300
bleeding. N Engl J Med 1983; 308:1571–1575 28. Oka K, Tanaka K: Local fibrinolysis of esophagus and stomach
15. von Holstein CC, Eriksson SB, Källén R: Tranexamic acid as an as a cause of hemorrhage in liver cirrhosis. Thromb Res 1979;
aid to reducing blood transfusion requirements in gastric and 14:837–844
duodenal bleeding. Br Med J (Clin Res Ed) 1987; 294:7–10 29. Piscaglia F, Siringo S, Hermida RC, et al: Diurnal changes of
16. Tavakoli N, Mokhtare M, Agah S, et al: Comparison of the ef- fibrinolysis in patients with liver cirrhosis and esophageal var-
ficacy of intravenous tranexamic acid with and without top- ices. Hepatology 2000; 31:349–357
ical administration versus placebo in urgent endoscopy rate 30. Violl F, Basili S, Ferro D, et al: Association between high values
for acute gastrointestinal bleeding: A double-blind random- of d-dimer and tissue plasminogen activator activity and
ized controlled trial. United European Gastroenterol J 2018; first gastrointestinal bleeding in cirrhotic patients. Thrombo
6:46–54 Haemost 1996; 76:177–183