Tranexamic acid in non-traumatic intracranial bleeding: a systematic review and meta-analysis

Non-traumatic intracranial bleeding (NTIB), comprising subarachnoid hemorrhage (SAH) and intra-cranial bleeding (ICH) is a significant public health concern. Tranexamic acid (TXA) is a promising treatment with benefits yet to be fully demonstrated. We conducted a systematic review and meta-analysis on the impact of TXA on mortality in NTIB. We searched the PubMed, Cochrane Library, Google Scholar and ScienceDirect databases for studies reporting mortality data following the use of TXA in NTIB for comparisons with a control group. We computed random-effect meta-analysis on estimates of risk and sensitivity analyses. We computed meta-regression to examine the putative effects of the severity of NTIB, sociodemographic data (age, sex), and publication date. Among potentially 10,008 articles, we included 15 studies representing a total of 4883 patients: 2455 receiving TXA and 2428 controls; 1110 died (23%) during the follow-up. The meta-analysis demonstrated a potential of 22% decrease in mortality for patients treated by TXA (RR = 0.78, 95%CI 0.58–0.98, p = 0.002). Meta-regression did not demonstrate any influence of the severity of NTIB, age, sex, length of treatment or date of publication. Sensitivity analyses confirmed benefits of TXA on mortality. TXA appears to be a therapeutic option to reduce non-traumatic intracranial bleeding mortality, particularly in patients with SAH.


Scientific Reports
| (2021) 11:15275 | https://doi.org/10.1038/s41598-021-94727-y www.nature.com/scientificreports/ medicines for bleeding trauma. However, only a limited number of studies have provided statistical evidence for the benefits of tranexamic acid on mortality among patients suffering from non-traumatic intracranial bleeding. As a result, we decided to conduct a systematic review and meta-analysis on the impact of TXA on mortality in NTIB patients.

Methods
Search strategy and selection criteria. We conducted a systematic search on PubMed, Cochrane Library, Google Scholar, and ScienceDirect to identify all prospective studies that reported a comparison of tranexamic acid versus placebo injection or control care. The research was not limited to specific years, and no language restriction was applied. The following relevant MESH terms/keywords were used until January 27th, 2021: ("Tranexamic Acid" OR "TXA") AND ("subarachnoid hemorrhage" OR "SAH" OR "intracerebral hemorrhage" OR "ICH"). Reference lists of all publications meeting the inclusion criteria were manually searched to identify any further studies that were not found using the electronic search, as well as reference lists from reviews. Three authors (JBBM, CC, and NP) conducted the same research strategy separately. Databases, titles, and abstracts were screened to decide which articles could be included. Two other authors (FD and JS) were asked to review the articles when the first researchers disagreed on suitability for inclusion. Studies had to be prospective clinical trials that reported to control the conditions, placebo injection and the efficacity of injection of tranexamic acid in the case of SAH or ICH. Furthermore, they had to study mortality during a followup period. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Table 1 summarizes important data. Data collection included as much information as possible, first author's name, year of publication, journal of publication, country of the first author, study design (randomized-controlled-trial or not), type of bleed (SAH or ICH), diagnostic criteria of bleeding (acute headache, meningeal irritation, blood-stained cerebrospinal fluid not due to trauma, angiography, CT-scan), number and name of groups, number of men in each group, number of women in each group, time of mortality reported, www.nature.com/scientificreports/ duration of treatment, posology of therapy, number of patients in each group, number of dead patients in each group, neurologic status (Botterell classification, Hunt and Hess classification 6 , World Federation of Neurologic Surgeon-WFNS classification 7 , Glasgow Coma Scale 8 , National Institute of Health Stroke Score-NIHSS 9 ) and mean age of groups. When the mean age was not described but the population was distributed, we used the following formula to calculate mean age: S[(n i x (min i + max i )/2] where n is the number of people, i the occurrence, min the minimal age of the event and max the maximal age of the occurrence. Considering we have an event (dead or not dead) in two different populations (exposed to tranexamic acid or not), we decided to calculate all risk ratio (with 95% interval) for all studies to homogenize results.

Quality of articles.
We used the Scottish Intercollegiate Guidelines Network (SIGN) criteria to check the quality of randomized clinical trials comprising 10 items 10 . Three authors (JBBM, CC, and NP) evaluated the methodological quality of the studies. Items were assessed for leading causes of bias. We calculated an overall quality score by quoting each item (yes-1 point; no, can't say or not applicable-0 point). Each study had a score between 0 and 10.
Data analysis. Statistical analysis was conducted using Stata software © (v16, StataCorp, College Station, US). Baseline characteristics were summarized for each study sample and reported as means ± standard deviations and number (%) for continuous and categorical variables, respectively. Effect sizes were defined as relative risk. Random effects meta-analyses (DerSimonian and Laird approach) were conducted when data could be pooled. P values less than 0.05 were considered statistically significant. We conducted a meta-analysis on the impact (risk ratio) of TXA on mortality in NTIB. Risk ratio < 1 with a 95%CI not containing 1 reflected a decreased risk of mortality with TXA, and a risk ratio > 1 denoted an increased risk of mortality. We assessed statistical heterogeneity between results by examining forest plots, confidence intervals, and using I 2 , which is the most common metric for measuring the magnitude of between-study heterogeneity and is easily interpretable. I 2 is a percentage and is typically considered as low if heterogeneity scores < 25%, modest if between 25 and 50%, and high if > 50% 11 . For rigor, we used a funnel plot of the meta-analysis to search for potential publication bias. We used visual inspection of funnel plot asymmetry to assess publication bias sensitivity analysis, excluding studies not evenly distributed around the funnel base. We secondly used Egger's test to determine other biases.
To verify the strength of the results' , a further meta-analysis was then conducted, excluding studies that were not evenly distributed outside of the metafunnel. We also further computed another sensitivity meta-analysis by considering only randomized controlled trials. Furthermore, we proposed stratification by the type of bleeding (SAH or ICH) and meta-regressions to study the relationship between the use of TXA, and age, sex, year of publication, duration of treatment, or severity of the intracranial bleeding. We expressed the results of the metaregression as regression coefficients at 95% CI.
TXA protocol. TXA or placebo was injected intravenously in all studies except one that administered it by mouth three times a day 16 . TXA injection frequency ranged between 1 g every 8 h 20,22,23 to 1.5 g every 4 h 24 . Placebo was injected at the same frequency as TXA and was NaCl 0.9% in all studies except one which used Epsilon Amino Caproic Acid (EACA) 16 . All studies specified that patients of both groups received the same regime of good nursing and medical care with attention to antihypertensive therapy and adequate analgesia and sedation. Only four studies 13,18,19,24 defined the cause of death between "total death" and "death secondary to rebleeding".  (Fig. 4), and with a 22% decrease in mortality when considering only RCT (0.78, 0.58 to 0.99, p < 0.001) (Fig. 5).

Discussion
To our knowledge, this is the first systematic review and meta-analysis studying the impact of TXA on mortality among patients suffering from NTIB (SAH and ICH). Among the 15 included studies, ten had an RR < 1 which was non-significant. Our meta-analysis demonstrated a possible significant benefit of TXA on mortality in NTIB patients. Furthermore, we failed to show a significant impact of age, sex, date of publication, duration of treatment or severity of initial bleeding on mortality after meta-regression. This result suggests that TXA can be largely prescribed to treat non-traumatic intracranial bleeding. We chose to use the "random-effect model", which is a more realistic assumption of results. Indeed, patients characteristics, treatment administration and outcomes are not the same in all studies (and relate to patient's care) 28 .

The challenge of TXA in SAH and ICH treatment. Spontaneous-non-traumatic-intracranial
bleeding-i.e., non-traumatic-is composed of two main entities with different anatomic and pathophysiologic causes. The first one is spontaneous subarachnoid hemorrhage (SAH). It is a severe stroke type mainly caused by the rupture of an aneurysm in the subarachnoid space. The leading cause of premature death in SAH is rebleeding 21 . Rebleeding must be prevented as early as possible by endovascular or neurosurgical obliteration of the ruptured aneurysm. The challenge of TXA treatment is to give time for the obliteration of the aneurysm. The second type of hemorrahage is intracerebral hemorrhage (ICH). Unfortunately, this type is mainly caused by the rupture of very-small vessels in the brain parenchyma. Therefore, vessels responsible for an ICH are not accessible to endovascular or neurosurgical obliteration. The only treatment is lowering blood pressure if needed 23 . Monitoring the condition is still a challenge in neurocritical care 29,30 . In this context, TXA has good potential to decrease mortality in ICH. If our meta-analysis demonstrated evidence-based benefits of TXA in SAH, the Keywords used for search strategy: "Tranexamic acid" OR "TXA" AND "Sub-arachnoid haemorrhage" OR "Intracerebral haemorrhage" Potential eligible articles n = 10 008 www.nature.com/scientificreports/ discarded. Lumbar puncture is limited to cases with a severe headache and clear CT-scan 31 . Care standards for patients suffering from NTIB have also changed over the last 40 years. Prior to the 1990s, aneurysms' late treatment led researchers to develop antifibrinolytics therapies such as aminocaproic acid and tranexamic acid to prevent rebleeding. However, a 2003 Cochrane Review demonstrated an increase in cerebral ischemic events after using antifibrinolytics drugs and recommended not to use antifibrinolytic medicines to treat patients with SAH 32 . Those results seem to reduce the impact of antifibrinolytic therapies because of an increase cerebral ischemia and poor neurological outcomes. However, another meta-analysis 33 discussed the impact of antifibrinolytic therapies in cerebral ischemia after a subarachnoid hemorrhage. Indeed, antifibrinolytic therapies were administrated during long periods before the years 2000s (up to 3 months 15 ) and were not associated with neuroprotective medications such as calcium channel blockers, and used no associated hypervolemic therapies to prevent ischemic insult. They concluded that short-term use of antifibrinolytic therapy (before aneurysm exclusion, always less than 3 days) associated with calcium channel blocker and hypervolemic therapies in the management of aneurysmal SAH diminishes rebleeding rate, and does not augment the risk of cerebral infarction. Since the late 1990s, it is now highly recommended to surgically repair aneurysms as soon as possible. A recent meta-analysis proved the benefit of early and mini-invasive procedures for supra-tentorial spontaneous ICH 34 . The development of intensive care units specialized in neurology with continuous monitoring of intracranial pressure, use of calcium channel blockers to protect the neurological status, and decrease the risk of delayed ischemic stroke have greatly improved SAH's care. Recent developments in endovascular coiling for cerebral aneurysms will probably further reduce post-operative complications. This will also allow elderly patients who are currently deprived of open neurosurgery to benefit from this technique. This will probably become the gold standard in the next decade.
Other indications of tranexamic acid. When TXA was discovered in the 1960s in Japan, its discoverers   This study concluded that TXA might decrease hematoma expansion on subsequent imaging and not increase the risk of adverse events 38 . However, a further study found that prehospital tranexamic acid administration was associated with increased mortality in patients with isolated severe TBI 39 . The HALT-IT randomized, doubleblind, placebo-controlled trial 40 , also published in 2020, did not benefit patients suffering from acute gastrointestinal bleeding on 5 days mortality (RR 0.99, 95CI 0.82 to 1.18). TXA is also used in emergent hemoptysis or epistaxis and in planned orthopedic or cardiac surgery, which were not studied because of their weak mortality.
Limitations. Our study has some limitations. Firstly, we decided to pool both SAH and ICH although those two pathologies have different populations, etiology and some difference in standard of care. To decrease the impact of this choice, we stratified our meta-analysis by type of bleeding (SAH versus ICH). Also, the main interventions for the care of intracranial bleeding are surgery and critical care. This is an important bias for the generalization of our results. Indeed, a hospital without neurosurgeons or interventional radiologists and critical care physicians cannot focus on the care associated with TXA in isolation. However, all patients included in all studies had the same access to neurosurgeons, interventional radiologists, and critical care physicians. Secondly, the difference in diagnosis and care between the 1970s and the 2020s induced a bias. However, no difference was found in mortality based on publication date. Thirdly, all trials' primary outcomes, treatment protocols, and duration of follow-ups were not identical. We also included both RCT and non-RCT trial which can decrease the power of generalization of our results. Fourthly, the duration of follow-up is heterogenic from 21 days through to 3 months duration between all studies. This could be an important bias for generalization. Fifthly, the inherent limitations of meta-analysis, such as publication bias, cannot be ignored. Lastly, the severity classifications were not similar. However, classification grades have been clustered in the meta-regression that did not find any effect. We did not study the adverse effect of TXA because many studies recently published proved its safety under many conditions, such as melasma in 2017 41

Conclusion
Tranexamic acid, a lysin analog, appears to be a therapeutic option to reduce non-traumatic intracranial bleeding mortality, particularly in patients with SAH. Its effects on mortality on ICH are yet unclear and require further investigation.

Data availability
All data are available on the different articles.