Tranexamic acid to improve functional status in adults with spontaneous intracerebral haemorrhage: the TICH-2 RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 11/129/109. The contractual start date was in March 2013. The draft report began editorial review in September 2018 and was accepted for publication in March 2019. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Rustam Al-Shahi Salman is a member of the Efficacy and Mechanism Evaluation Funding Board panel. Lelia Duley reports grants from the Nottingham Clinical Trials Unit during the conduct of the study. Christian Ovesen reports grants from the Velux Foundation (Søborg, Denmark), the Hojmosegaard Grant/Danish Medical Association (Copenhagen, Denmark), the Axel Muusfeldt’s Foundation (Albertslund, Denmark), the University of Copenhagen (Copenhagen Denmark) and non-financial support from Merck Sharp & Dohme (MSD; Kenilworth, NJ, USA) outside the submitted work. Robert A Dineen reports grants from the National Institute for Health Research (NIHR) Health Technology Assessment (HTA) programme (project number 11/129/109) during the conduct of the study. Timothy J England reports grants from the NIHR HTA programme during the conduct of the study. Thompson G Robinson reports grants from the University of Leicester. Christine Roffe has been a member of the HTA General Board since 2017. David Werring reports personal fees from Bayer AG (Leverkusen, Germany) outside the submitted work. Philip M Bath reports grants from the British Heart Foundation and the NIHR HTA programme during the conduct of the study, others from Platelet Solutions Ltd (Nottingham, UK) and personal fees from Diamedica (UK) Ltd (Bratton Fleming, UK), Nestlé SA (Vevey, Switzerland), Phagenesis Ltd (Manchester, UK), ReNeuron Group plc (Bridgend, UK), Athersys Inc. (Cleveland, OH, USA) and Covidien (Dublin, Ireland) outside the submitted work.

Copyright statement

© Queen’s Printer and Controller of HMSO 2019. This work was produced by Sprigg et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2019 Queen’s Printer and Controller of HMSO

Scientific background

Rationale for trial

Haematoma expansion is a not uncommon complication of ICH and is associated with devastating consequences of early death and severe disability in those who survive. There is no effective drug treatment to prevent haematoma expansion. Tranexamic acid, an antifibrinolytic drug, is effective at reducing bleeding and death due to bleeding in other conditions. Tranexamic acid is widely available, affordable and appears to be safe in studies in other bleeding conditions conducted to date. The rationale for this study was to test the safety and efficacy of tranexamic acid in ICH, the hypothesis being that tranexamic acid will prevent haematoma expansion and reduce death and disability after ICH.

Objectives

Design

The Tranexamic acid in IntraCerebral Haemorrhage 2 (TICH-2) study was an international, double-blind, randomised, placebo-controlled, parallel-group, Phase III trial performed in two phases: an 18-month start-up phase (with an aim to activate 30 centres and recruit a minimum of 300 participants) and then a main phase (with 120 centres and to recruit to a total of 2000 participants). There was no break in recruitment as the trial proceeded from the start-up phase to the main phase because the stopping criteria were not met.

Participants were enrolled by investigators from acute stroke units at 124 hospital sites in 12 countries: Denmark, Georgia, Hungary, Ireland, Italy, Malaysia, Poland, Spain, Sweden, Switzerland, Turkey and the UK. Ethics approval was obtained in each site and country prior to commencement of the study. The trial was adopted in the UK by the National Institute for Health Research Clinical Research Network and registered with the International Standard Randomised Controlled Trial Number (ISRCTN) Registry as ISRCTN93732214. The full TICH-2 trial protocol41 and statistical plan42 have been published.

Study settings

The study was set in acute stroke units at 124 hospitals in 12 countries (as detailed in Design). A full listing of sites and investigators can be found in Report Supplementary Material 1.

Participants

Data collected at baseline

Investigators recorded the participants’ age, sex, ethnic group, medical history and whether or not they were taking antiplatelet agents, as well as their assessment of ICH location, IVH, spot sign and blood pressure. Investigators assessed prestroke dependence with the mRS, and stroke severity using the National Institutes of Health Stroke Scale (NIHSS) and GCS.

Interventions

The intervention, tranexamic acid, was given intravenously as a 1-g loading dose in 100 ml of 0.9% normal saline infused over 10 minutes, followed by another 1 g in 250 ml of 0.9% normal saline, which was infused over 8 hours. The comparator was a matching placebo (i.e. 0.9% normal saline), administered with an identical regimen.

The dosing regimen selected (1-g bolus and 1-g infusion) was chosen to achieve plasma concentrations sufficient to inhibit fibrinolysis. Studies in cardiac surgery have shown that use of a dose > 10 mg/kg bolus and 1 mg/kg infusion does not provide any additional haemostatic benefit. 43 In the emergency situation, administration of a fixed dose is more practicable and the fixed dose chosen is efficacious for patients weighing > 100 kg and it is safe for patients weighing < 50 kg. The short duration of treatment allows for the full effect of tranexamic acid on the immediate risk of haematoma expansion without extending too far into the acute phase response seen after stroke.

Randomisation

All participants eligible for inclusion were randomised centrally using a secure internet site in real time. Randomisation involved stratification by country and minimisation on key prognostic factors: age; sex; time since onset; SBP; stroke severity (as assessed via the NIHSS); presence of IVH; and known history of antiplatelet treatment used immediately prior to stroke onset. This approach ensured concealmentof allocation, minimises differences in key baseline prognostic variables, and slightly improves statistical power.

Randomisation allocated a number corresponding to a treatment pack and the participants received treatment from the allocated numbered pack.

In the event of computer failure (e.g. server failure), investigators would follow the working practice document for computer system disaster recovery, which will allow the participant to be randomised manually following a standardised operating procedure.

Blinding

Clinicians, patients and outcome assessors (i.e. research nurse and radiologist) were blind to treatment allocation. In general, there should be no need to unblind the allocated treatment. If some contraindication to antifibrinolytic therapy developed after randomisation (e.g. clinical evidence of thrombosis), the trial treatment would simply be stopped. Unblinding was to be done only in those rare cases when the doctor believed that clinical management depended importantly on the knowledge of whether or not the patient received antifibrinolytic or placebo. For those few patients for whom urgent unblinding was considered necessary, the emergency telephone number was telephoned, giving the name of the doctor authorising unblinding and the treatment pack number. The caller was then told whether the patient received antifibrinolytic or placebo. The rate of unblinding was monitored and audited.

In the event of breaking the treatment code, this was recorded as part of managing a serious adverse event (SAE; see Chapter 2, Methods, Secondary outcomes, Safety for more details) and such actions were reported in a timely manner. The chief investigator (delegated the sponsor’s responsibilities) was informed immediately (within 24 hours) of any SAEs and determined seriousness and causality in conjunction with any treating medical practitioners.

Adherence

Adherence was assessed by examining the participant’s drug chart and recording the trial treatment administered at day 2 (i.e. whether or not all treatment was given, the time and date of the two doses and any other comments). Adherence was verified by both the central review of the drug chart and pharmacies recording returns of residual or unused trial medications.

Assessments after randomisation

Participants were reviewed at days 2 and 7, and on the day of death or hospital discharge, whichever came first, to gather information on clinical assessment (i.e. the NIHSS score), the process-of-care measures (e.g. blood-pressure-lowering treatment, neurosurgical intervention) and discharge date and destination (e.g. home or institution). A second research CT scan was carried out after 24 hours of treatment to assess haematoma expansion.

The central assessors, who were trained and certified in administration of the mRS and masked to treatment allocation, carried out the final follow-up at 90 days by telephone from the co-ordinating centre in each country. If the participant or carer could not be contacted, they received a questionnaire covering the same outcome measures by post.

Primary efficacy outcome

Functional status, death or dependency (ordinal shift on the mRS) at day 90 was compared between tranexamic acid and placebo by intention to treat using ordinal logistic regression (OLR), with adjustment for stratification and minimisation factors. The assumption of proportional odds was tested using the likelihood ratio test.

Secondary outcomes

Study oversight

The trial was overseen by a Trial Steering Committee (TSC) and an International Advisory Committee comprising each national co-ordinator. A Trial Management Committee, based at the Stroke Trials Unit in Nottingham, UK, was responsible for the day-to-day conduct of the trial. Study data were collected, monitored and analysed in Nottingham. The trial was performed according to the principles of good clinical practice and the Declaration of Helsinki. 50

Sample size

The null hypothesis (H₀) is that tranexamic acid does not alter death or dependency in participants with acute primary ICH. The alternative hypothesis (H_A) is that death or dependency differs between those participants randomised to tranexamic acid and those randomised to saline. A total sample size of 2000 (1000 per group) participants with acute primary ICH is required, assuming:

• an overall significance of (alpha) = 0.05

• a power of (1 – beta) = 0.90

• a distribution in mRS score (mRS 0 = 4%, 1 = 17%, 2 = 16%, 3 = 19%, 4 = 24%, 5 = 7% and 6 (death) = 13%; based on data from participants with primary ICH in the Efficacy of Nitric Oxide in Stroke (ENOS) trial

• an ordinal OR of 0.79

• increases due to losses to follow-up of 5%

• a reduction of 20% for baseline covariate adjustment. 51

In summary, a trial of 2000 participants (1700 from the main phase and 300 from the start-up phase) will have 90% power to detect an ordinal shift of mRS outcome with an OR of 0.79.

Protecting against bias, including blinding

Numerous steps were taken to protect against bias in this double-blind RCT, with allocation concealed from all staff throughout the study.

Neuroimaging and scan adjudication

Brain imaging by CT was done as part of routine care before enrolment; a second research CT scan was done after 24 hours of treatment to assess haematoma expansion. When multiple scans were done, the scan closest to 24 hours after randomisation was used. Central independent expert assessors, who were masked to treatment assignment, assessed CT scans for the location of the ICH using a web-based adjudication system. Semi-automated segmentation of the ICH was done on Digital Imaging and Communications in Medicine [DICOM^®; Medical Imaging & Technology Alliance (MITA), Arlington, VA, USA]-compliant images to give ICH volumes. The user-guided three-dimensional active contour tool in the itk-SNAP software (version 3.6; www.itksnap.org/pmwiki/pmwiki.php, accessed 3 May 2019) was used for segmentation and, as required, one of the three assessors did manual editing. All assessments were masked to treatment assignment.

Haematoma expansion was defined as an absolute increase of > 6 ml or a relative growth of > 33%.

Sites, investigators and monitoring

Investigators were trained in trial procedures, via a site initiation teleconference, after obtaining all of the necessary regulatory approvals. New investigators who joined the study after the site initiation training were required to complete and pass an online training test covering the trial protocol. Monitoring of trial data included confirmation of informed consent in all participants; source data verification; data storage and data transfer procedures; local quality control checks and procedures, back-up and disaster recovery of any local databases; and validation of data manipulation. The trial co-ordinator carried out monitoring of trial data throughout the study; entries on case report forms (CRFs) were verified by inspection against the source data. A sample of CRFs (10% as per the trial risk assessment) were checked for verification of all entries made. In addition, the subsequent capture of the data on the trial database was checked. Where corrections were required, a full audit trail and justification was documented.

Protocol amendments

During the course of the trial there was a number of amendments to the study protocol, which are detailed in Chapter 2, Methods, Substudies. In summary, these amendments were to add a number of substudies [i.e. a day 365 follow-up, a magnetic resonance imaging (MRI) substudy and a biomarker substudy], make minor changes to trial documentation and allow co-enrolment to the ongoing RCT, RESTART. Full details of protocol amendments can be found in Appendix 2.

Substudies

Three substudies were performed and will be presented elsewhere. First, an additional follow-up at day 365 was performed and, second, a MRI substudy (funded by the British Heart Foundation) was performed, which will be presented separately. Finally, a single-centre plasma biomarker substudy examined the effect of tranexamic acid on markers of haemostasis and inflammation.

Meta-analysis

In due course, data from TICH-2 will be added to summary and individual patient data meta-analyses in acute stroke. This will include data being used in a Cochrane review25 and in an individual patient data meta-analysis of all tranexamic acid effects in ICH by the Antifibrinolytics Triallists Collaboration (ATC). 52

Recruitment

Recruitment started on 1 March 2013 and ended on 30 September 2017, after a 12-month extension was sought and approved to enable the trial to reach its target sample size (Table 1). This slower-than-planned recruitment was due to delays in opening trial sites outside the UK. Subsequently, recruitment increased and the TSC agreed that the study should exceed the target of 2000 and continue until the end of the extension. Therefore, a total of 2325 participants (1161 randomised to tranexamic acid and 1164 to placebo) were recruited from 124 sites in 12 countries over 55 months (Tables 1 and 2 and Figures 1–3).

FIGURE 1.

Recruitment graph.

FIGURE 2.

Recruitment by centre.

FIGURE 3.

Participant Consolidated Standards of Reporting Trials (CONSORT) flow diagram.

There was good data completion, with nine patients lost to follow-up or withdrawn in each of the treatment groups (see Figure 3).

Baseline data

The majority of participants were recruited in the UK (1910, 82.2%). The mean age of participants was 68.9 years (SD 13.8 years); 1301 participants (56%) were male. The median time from stroke onset to randomisation was 3.6 hours [IQR 2.6–5.0 hours] and one-third of participants (833, 36%) were recruited within 3 hours (Table 3). The mean baseline blood pressure was 173/93 mmHg. The majority of participants had a haematoma that was deep and supratentorial (1371, 59.0%), with 738 (31.7%) participants having a lobar supratentorial haematoma; one-third of participants had IVH. The mean HV was 24.0 ml (SD 27.2 ml) and the median HV was 14.1 ml (IQR 5.9–32.4 ml). Contrast-enhanced imaging, conducted as CTA, was performed in 249 (10.9%) of the participants. Twenty-four out of 121 (19.8%) participants in the tranexamic acid group and 32 out of 128 (25%) in the placebo group were spot positive. Treatment groups were well balanced at baseline.

Adherence/tranexamic acid treatment

Adherence to the allocated treatment was high, that is, 2207 (94.9%) participants received all of their randomised treatment, whereas only 15 (0.6%) participants received no treatment. There was no difference in adherence rates between the treatment groups. The median time from randomisation to treatment was 21 minutes (IQR 13–33 minutes) (Table 4).

Time to treatment

The median time for stroke onset to randomisation was 3.6 hours (IQR 2.6–5.0 hours) (Figure 4), with 827 (35.6%) participants enrolled within 3 hours and 1571 (67.6%) participants enrolled within 4.5 hours. The median time to a CT scan was 1.9 hours (IQR 1.4–2.9 hours); 77% of participants had a CT scan within 3 hours of symptom onset and 93% had a CT scan within 4.5 hours (Figure 5). The median time from a CT scan to randomisation was 1.3 hours (IQR 0.8–2.1 hours) (Figure 6), with 861 (37%) participants randomised within 1 hour of a CT scan and 1681 (72.3%) participants randomised within 2 hours of a CT scan. The majority of participants took more than 1 hour to enrol in the trial after a CT scan was performed. The median time to first dose was 4.1 hours after stroke onset (Figure 7), 25% of participants were treated within 3 hours of stroke onset and 58% within 4.5 hours.

FIGURE 4.

Time from stroke onset to randomisation.

FIGURE 5.

Time from stroke onset to CT scan.

FIGURE 6.

Time from CT scan to randomisation.

FIGURE 7.

Time from stroke onset to treatment.

Consent

The majority of participants were enrolled in the trial after consent by a relative (Table 5). Participants with more severe stroke (i.e. a higher NIHSS score) were more likely to use relative or doctor consent. Independent doctor consent had the shortest time from onset to randomisation (3.25 hours) compared with full informed consent from a relative, which took the longest (4.21 hours).

Primary efficacy outcome

The primary outcome of mRS score at day 90 was determined in 2307 (99.2%) participants; 9 (0.4%) were lost to follow-up and 9 (0.4%) withdrew from their day 90 follow-up (see Figure 3). There was no difference in the distribution (i.e. shift) in the mRS score at day 90 after adjustment for stratification and minimisation criteria [adjusted odds ratio (aOR) 0.88, 95% CI 0.76 to 1.03; p = 0.11] (Figure 8). A formal goodness-of-fit test gave no evidence that the proportional odds assumption was violated (p = 0.97). There was no difference between the tranexamic acid and placebo treatment groups in the proportion of participants who were dead or dependent (mRS score of > 3: aOR 0.82, 95% CI 0.65 to 1.03; p = 0.08).

FIGURE 8.

Primary outcome: shift plot of the day 90 mRS score (aOR 0.88, 95% CI 0.76 to 1.03; p-value = 0.11). mRS score level descriptions: 0 = no symptoms; 1 = no disability despite symptoms; 2 = slight disability, but able to look after own affairs; 3 = moderate disability, but able to walk without assistance; 4 = moderately severe disability – unable to walk or attend to own bodily needs; 5 = severely disabled – bedridden and requiring constant nursing care; and 6 = death. Reproduced from Sprigg et al. 1 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Reproduced from Sprigg et al. 1 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.

Subgroup analysis

When the primary outcome was assessed in prespecified subgroups (Figure 9), the only significant interaction was between mRS score and baseline SBP (interaction p = 0.016). Participants with a baseline SBP of < 170 mmHg had a favourable shift in mRS score with tranexamic acid compared with those participants with a SBP of ≥ 170 mmHg. There was no heterogeneity of treatment effect by time to enrolment (see Figure 9) whether dichotomised at < 3 hours versus ≥ 3 hours (interaction p = 0.89) or ≤ 4.5 hours versus > 4.5 hours (interaction p = 0.28); similarly, there was no interaction between treatment effect and time when analysed as a continuous variable (aOR 0.98, 95% CI 0.90 to 1.07; p = 0.69).

FIGURE 9.

Forest plot of the day 90 mRS scores in the subgroups. Haematoma volume was not prespecified and was an exploratory post hoc analysis.

In exploratory post hoc analysis, participants with a moderate haematoma size (i.e. 30–60 ml) have a favourable shift in mRS score with tranexamic acid treatment (aOR 0.66, 95% CI 0.44 to 0.98, p = 0.039), but the interaction between haematoma size and treatment was not statistically significant (p = 0.84). This difference was statistically significant when studying 1350 participants with a haematoma size of ≤ 60ml who were enrolled within 4.5 hours of onset (aOR 0.821, 95% CI 0.67 to 0.9; p = 0.031). However, the interaction between haematoma size and treatment in patients randomised within 4.5 hours was not statistically significant (p = 0.55).

Secondary outcomes

Process-of-care measures

The majority of participants received intravenous blood pressure-lowering treatment (i.e. 75.3% in the tranexamic acid group and 74.5% in the placebo group) on the first day. By day 7 nearly all participants had received blood pressure-lowering treatment (i.e. 81.6% in the tranexamic acid group and 83.5% in the placebo group).

The use of do not attempt resuscitation (DNAR) orders increased from 17% in the first 24 hours, to 22% by day 7, but there was no difference between the treatment groups (Table 8).

Only 5% of participants underwent neurosurgery, with similar rates for both treatment groups. A total of 10% of participants were transferred to intensive care units (ICUs), with 7% of participants undergoing ventilation. Rates of ICU transfer and ventilation did not differ between the treatment groups.

Safety outcomes

There were fewer deaths by day 7 in the tranexamic acid group (101, 8.7%) than in the placebo group (123, 10.6%; aOR 0.73, 95% CI 0.53 to 0.99; p = 0.041).

The number of deaths by day 90 did not differ between the tranexamic acid group (250/1161, 21.5%) and the placebo group (249/1164, 21.4%; see Table 7). There was no difference in survival between treatment groups over 90 days (adjusted hazard ratio 0.92, 95% CI 0.77 to 1.10; p = 0.37) (Figure 10).

FIGURE 10.

Plot of cumulative mortality. HR, hazard ratio.

Serious adverse events

A total of 1599 events were reported as SAEs, of which 1355 were SAEs and 200 serious adverse reactions (SARs) (Figure 11). There were no suspected unexpected SAR (Table 9).

FIGURE 11.

Serious adverse events reported by investigators. SUSAR, suspected unexpected serious adverse reaction.

Fewer participants experienced safety outcomes and SAEs at day 2 (32.6% vs. 35.8%), 7 (39.3% vs. 42.7%) and 90 days (44.9% vs. 47.8%) were significantly lower in the tranexamic acid group (see Table 9). There was no increase in the number of venous thromboembolic events (3.4% in the tranexamic acid group vs. 3.2% in the placebo group; p = 0.98), or arterial occlusions (MI, acute coronary syndrome or peripheral arterial occlusion) in the tranexamic acid group (see Table 9).

Seizure was the most common safety outcome, but there was no difference between the treatment groups [77 (7%) patients in the tranexamic acid group vs. 85 (7%) patients in the placebo group]. Nervous system disorders were the most common SAEs [149 (13%) in the tranexamic acid group vs. 163 (14%) in the placebo group], followed by infections [98 (8%) in the tranexamic acid group vs. 116 (10%) in the placebo group].

Protocol violations

A total of 331 protocol violations were reported by investigators, with 87 meeting the protocol definition of a protocol violation (Table 10). The commonest protocol violation was failure of a participant to receive all of the allocated treatment, which occurred in 1% of participants.

Interpretation

In this international multicentre RCT of tranexamic acid after acute ICH, there was no statistically significant difference in the primary outcome of functional status at day 90. Despite no significant improvement in primary outcome, there were statistically significant yet modest reductions in the prespecified secondary outcomes of early death, haematoma expansion and SAEs. There was no increase in thromboembolic events or seizures.

In keeping with an antifibrinolytic effect, tranexamic acid was associated with a modest yet statistically significant reduction in haematoma expansion and smaller growth in HV, key factors known to influence outcome after ICH. However, the small reduction in HV growth (i.e. 1.4 ml), and proportion of participants experiencing haematoma expansion (29% in the placebo group compared with 25% in the tranexamic acid group) may have been too small to translate into improved functional status in this population.

Therefore, although it is possible that tranexamic acid is not effective at improving outcome after ICH, an alternative explanation for the study’s findings is that the observed treatment effect was too modest. The study’s sample size calculation was based on an estimated effect size of an OR of 0.79, which the trial did not detect. The observed OR was 0.88 and a much larger sample size would be required to detect this. Indeed, previous RCTs of tranexamic acid in other settings have randomised more than 10-fold the number of participants to identify more modest effects on bleeding-related deaths after trauma (i.e. an OR of 0.85)23 and postpartum haemorrhage (i.e. an OR of 0.81). 13 Furthermore, an individual patient data meta-analysis of two RCTs (with > 40,000 participants), published after enrolment in TICH-2 had completed, has subsequently confirmed that commencing tranexamic acid within 3 hours of the start of bleeding is necessary to obtain the benefit of tranexamic acid after trauma or postpartum haemorrhage. 15 Despite repeated efforts to encourage investigators to reduce the time to enrolment, the majority of the study’s participants were enrolled after 3 hours. Recent meta-analysis of individual patient data in 5435 patients has demonstrated that the probability of haemorrhage growth declines mostly steeply 3 hours after symptom onset,53 highlighting the need for urgent treatment.

In the same meta-analysis, baseline HV was the strongest predictor of outcome after SICH. The probability of haematoma growth increased with baseline HV, but peaked at 75 ml. 53 In TICH-2, in an exploratory post hoc analysis, participants with a baseline HV of between 30 and 60 ml, who received tranexamic acid appeared to have better outcome (aOR 0.66, 95% CI 0.44 to 0.98; p = 0.039; see Figure 9), but the interaction between HV and treatment was not significant (p = 0.84). Although this could be due to chance and caution is needed when interpreting subgroups, it is also compatible with the hypothesis that patients with moderate-size haematomas may be more likely to benefit from haemostatic therapies and, hence, could be targeted for future studies, as has been postulated with rFVIIa. 23,24

The statistically significant interaction demonstrated in this study between baseline SBP and treatment suggests participants with lower blood pressure were more likely to benefit from tranexamic acid. This finding could have been confounded by stroke severity, as larger haematomas have increased blood pressure and worse outcomes. 22,26 Despite being the only intervention to date to improve functional outcome after ICH, early intensive blood pressure-lowering treatment3 remains controversial. 11 Although there were no significant effects on haematoma growth in INTERACT-2,3 secondary analysis suggested blood pressure lowering did attenuate bleeding in a dose-dependent manner. 25 The interaction demonstrated in this study between SBP and tranexamic acid suggests complementary approaches to reduce haematoma expansion, with haemostatic and haemodynamic therapies warranting further investigation.

Unlike other acute ICH studies with rFVIIa24 and aggressive lowering of blood pressure,10 the study found no evidence of adverse effects. Notably, there was no increase in VTE or seizures in this significantly older population with more comorbidities than in participants in previous studies of tranexamic acid. 13,23 It is therefore unlikely that any potential benefit of tranexamic acid was offset by harm, as has been suggested with rFVIIa. 24 In a Phase 3 trial, there was no evidence of clinical benefit of rFVIIa, with a reduction in haematoma expansion but an increased risk in arterial occlusive events. 24 Although tranexamic acid and rFVIIa are both haemostatic agents, tranexamic acid has antifibrinolytic mechanisms and rFVIIa is a procoagulant and, hence, they have different risk–benefit profiles. Recent studies with rFVIIa have attempted to use the spot sign to identify those patients at risk of haematoma expansion and, as such, most likely to benefit from rFVIIa. 26,27 However, the trials were stopped as a result of poor recruitment. 28 The study has also demonstrated that selecting large numbers of patients based on the spot sign is not practicable, as only 10% of participants had advanced imaging performed. Although the presence of the spot sign does improve prediction of ICH growth, time from onset, antithrombotic therapy and baseline HV remain the strongest predictors. 53

Generalisability

The study’s inclusion criteria were deliberately broad, with recruitment from international sites, across secondary and tertiary care, reflective of the clinical population. However, the study was unable to collect screening logs across such a large number of sites and, as such, was unable to determine the proportion of population who would be eligible for haemostatic therapy if it were effective.

Overall evidence

In the systematic review,25 two small RCTs of tranexamic acid with a total of 54 participants were found, with no clear evidence of benefit or harm associated with tranexamic acid. 38,39 Several smaller RCTs of tranexamic acid in ICH are ongoing,54–58 with a total enrolment of < 1000 participants across all the trials. A number of these ongoing trials are targeting participants likely to be at greater risk of haematoma expansion, based on an earlier time window,54,55 spot sign status55,58 and anticoagulation. 59 Although an individual patient data meta-analysis is planned,33 further large randomised trials are needed to confirm or refute a clinically significant treatment effect of tranexamic acid.

Strengths

The strengths of this study include double-blinding and strong allocation concealment, with low risk of bias, high adherence to treatment and high completion of follow-up, with very few missing primary outcomes. The use of an approved brief (see Report Supplementary Material 1) and proxy consent processes allowed rapid enrolment of patients without capacity, who are important in acute stroke studies. Independent doctor consent was the most rapid form of consent and may be appropriate in emergency studies, especially when relatives struggle to make informed decisions. Minimisation techniques were effective, and the treatment groups were well matched at baseline. The sample size was large, making this the largest haemostatic therapy RCT after ICH. Pre-study support from the Nottingham Clinical Trials Unit ensured that approvals were obtained rapidly once the grant was activated, allowing recruitment to begin in a timely manner. This rapid set-up process led to a large number of sites being activated ahead of schedule, with recruitment in the UK exceeding initial expectations, which led to early achievement of the stop–go criteria. The large number of sites participating in the UK would not have been possible without the support of the NIHR Clinical Research Network in England and adoption of TICH-2 on the study portfolio. The absence of an effective drug therapy for ICH, in combination with streamlined processes of acute stroke care to facilitate reperfusion therapies in ischaemic stroke and the absence of competing trials, led to TICH-2 fitting well into the acute stroke clinical pathway and ensured good engagement with clinicians. Finally, permission was sought and granted to allow co-enrolment into RESTART; an investigator-led, randomised trial comparing starting versus avoiding antiplatelet drugs for adults surviving antithrombotic-associated ICH. 60 The TSCs for both studies felt that co-enrolment did not affect the safety of participants or the quality of the scientific data and yet allowed for both important clinical questions to be answered, something that the patient and public involvement group strongly supported (20 participants were co-enrolled, 10 in the tranexamic acid group, 10 in the placebo group).

Limitations

The wide inclusion criteria led to a heterogeneous population of patients, with multiple comorbidities, more severe strokes, larger HVs and a greater proportion of participants with lobar haematomas and IVH as compared with other ICH trials. 3,22,24,27 This inclusion of older participants with more comorbidities and larger haematomas, in particular, could have diluted any treatment effect.

As already highlighted, the majority of participants were enrolled more than 3 hours after ICH onset, which could explain the absence of a significant interaction with time in the subgroup analysis. In 2016, the protocol for the ongoing RCT CRASH-3 (testing tranexamic acid in traumatic brain injury) was amended to reduce the time window for eligibility (originally within 8 hours of injury) to patients within 3 hours of injury. 61 The TSC for TICH-2 discussed on a number of occasions whether or not the inclusion criteria should be amended to limit recruitment to participants within 3 hours of symptom onset in line with other tranexamic acid studies. However, as the majority of recruitment had been completed, and there are important pathophysiological differences between ICH and traumatic brain injury, the decision was taken to leave the inclusion criteria unchanged but to focus on efforts to ensure rapid treatment. Had this been an adaptive trial design, enrichment could have allowed further recruitment within subgroups where tranexamic acid was most likely to be effective, such as those patients presenting < 3 or < 4.5 hours after the onset of symptoms, or those with moderate-sized haematomas. Furthermore, sample size reassessment to account for the fact that the observed effect size was lower than predicted in the original sample size calculation may have been beneficial. 62

The study did not seek to explore mechanistic approaches by which tranexamic acid could improve outcome. First, this could be done by addressing how tranexamic acid may reduce haematoma expansion, and it is thus unclear whether the reduction in number of participants experiencing haematoma expansion is related to attenuation of primary bleeding, secondary vessel rupture, both or neither. More research is needed to understand the time course of changes after acute ICH and differences between lobar and deep haematomas. Although the study collected data on haematoma location and local investigators assigned a final diagnosis, it is acknowledged that recognising cerebral amyloid angiopathy (CAA) is a challenge and many of the participants, particularly elderly participants, may have undiagnosed CAA. Further work is currently being undertaken to analyse participants’ brain images and will be presented at a later date. Second, beyond antifibrinolytic effects there is increasing recognition that because of its inhibition of plasmin, tranexamic acid has anti-inflammatory properties. This anti-inflammatory action has been speculated to be important for the efficacy of tranexamic acid. 63 Inflammation is a key contributor to secondary brain injury after ICH and the effects of tranexamic acid on this warrant further investigation.

Although the recruitment process was simple and enrolment could be done by a doctor, the majority of participants were enrolled during working hours, due in part to a reliance on clinical research network staff. Only a small number of hyperacute stroke research centres in the UK have coverage from clinical research network staff in the evenings and at weekends. It is therefore likely that a large number of potential participants were not considered for enrolment as they presented with stroke symptoms outside normal working hours.

Furthermore, requiring the NIHSS score to be measured before randomisation reduced the number of participants enrolled in the emergency department because of a lack of availability of trained staff, again, particularly out of hours. For this reason, the study did not collect HV at baseline, as measuring HV is not currently part of standard clinical care in the majority of hospitals. It was therefore not possible to minimise on the basis of HV.

Finally, although the study set up and recruitment in the UK exceeded expectations, set up in international sites took longer than expected, with governance issues contributing to significant delays. This was the main contributing factor that necessitated the 1-year contract extension in order to allow recruitment targets to be met. This delay in starting the study in international sites meant that the majority of participants were recruited from the UK, limiting international generalisability.

Public and patient involvement

Haemorrhagic stroke was highlighted as a research priority by the Stroke Association and stroke survivors have been involved throughout the trial, since the conception of the study. The Nottingham Stroke Research Partnership, comprising stroke survivors and carers, reviewed the proposed study, in an iterative process, commented on its design and conduct. In particular, the Nottingham Stroke Research Partnership influenced the approach to taking informed consent within the study. The group felt strongly that everything should be done to allow as many people suffering from stroke as possible to take part in the study. In particular, the group recognised that in the emergency situation of an acute stroke it is often impossible to take in information to allow people to fully decide whether or not they wish to take part in the study, as a result of speech and processing problems caused by the acute stroke. In this instance, it was believed that the person would want to be included in such a study, unless there was a medical reason not to or if they had pre-expressed a desire not to want to take part. The group also recognised that many people arrive in hospital alone after a stroke without family members present. It was for this reason that the study sought permission for enrolment after consultation with an independent physician when no family or friends are present.

Malcom Jarvis and Christine Knott, both stroke survivors, were members of the TSC throughout the duration of the trial. In addition to shaping the informed consent processes, stroke survivors (i.e. the Nottingham Stroke Research Partnership group) provided advice on how best to inform participants of the study results, which included developing a lay summary for the trial, which is available on the study website. Malcolm Jarvis and Christine Knott helped prepare the primary publication and this report.

Conclusions

The TICH-2 study is the first large multicentre international RCT of tranexamic acid, in acute SICH. There was no statistically significant improvement in functional status at 90 days. Significant yet modest reduction in the number of participants with haematoma expansion and fewer early deaths in those allocated tranexamic acid are consistent with tranexamic acid having an antifibrinolytic effect after ICH. Tranexamic acid was safe, with fewer SAEs and no increase in thromboembolic events or seizures. Prespecified subgroup analyses suggested that treatment for participants with lower blood pressure may be beneficial. Although the analysis did not demonstrate an interaction with time to enrolment, earlier treatment in those patients with a modest haematoma size (excluding those with very large haematomas, who may have already undergone haematoma expansion) may be optimal.

Implications for health care

Although there is insufficient evidence to support the routine use of tranexamic acid in clinical practice for SICH, the results do not exclude a modest but clinically important treatment effect.

Tranexamic acid is affordable, widely available, easy to administer and appears to be safe. Even a modest benefit could have a global impact on the excessive burden of mortality and morbidity after ICH.

Recommendations for research

Although a number of smaller randomised trials are ongoing,54–58,64 further large randomised trials are likely to be needed to confirm or refute a clinically significant treatment effect. Future research should investigate which subgroups of patients may benefit. Enriching future trial populations with participants most likely to benefit from haemostatic agents (e.g. those patients presenting early, with moderate haematomas, excluding participants with predestined poor outcome) may reduce sample size but should not be done at the expense of increasing the complexity of the study design. Focus must primarily be on enrolling participants as soon as possible after a diagnosis of ICH. One ongoing study is already seeking to enrol participants in the prehospital time window. 64 Trial design (including participant selection and consent processes) needs to be streamlined and as fast as possible in this emergency situation. Embedding trials into the clinical pathway is essential to ensure adequate recruitment and generalisability of results. Finally, harmonisation of approval processes could reduce set-up time and expense needed to complete large international studies, thereby reducing research waste. 65

Contributions of authors

Nikola Sprigg (Chief Investigator) conceived the trial, wrote the first draft of the manuscript, critically appraised the manuscript and approved the final version.

Katie Flaherty (Trial statistician) performed the statistical analysis, critically appraised the manuscript and approved the final version.

Jason P Appleton (Trial medic) was responsible for the day-to-day running of the trial, critically appraised the manuscript and approved the final version.

Rustam Al-Shahi Salman (Member of the Efficacy and Mechanism Evaluation funding committee) was a co-applicant and member of the TSC.

Daniel Bereczki (National Co-ordinator in Hungary) critically appraised the manuscript and approved the final version.

Maia Beridze (National Co-ordinator in Georgia) critically appraised the manuscript and approved the final version.

Alfonso Ciccone (National Co-ordinator in Italy) critically appraised the manuscript and approved the final version.

Ronan Collins (National Co-ordinator in Ireland) critically appraised the manuscript and approved the final version.

Robert A Dineen (Co-applicant and member of the TSC) led the adjudication of neuroimaging, critically appraised the manuscript and approved the final version.

Lelia Duley (Member of the TSC) critically appraised the manuscript and approved the final version.

Juan José Egea-Guerrero (National Co-ordinator in Spain) critically appraised the manuscript and approved the final version.

Timothy J England (Co-applicant and member of the TSC) led the adjudication of SAEs, critically appraised the manuscript and approved the final version.

Michal Karlinski (Lead Investigator in Poland) critically appraised the manuscript and approved the final version.

Kailash Krishnan performed radiological measurements (including HV), critically appraised the manuscript and approved the final version.

Ann Charlotte Laska (National Co-ordinator in Sweden) critically appraised the manuscript and approved the final version.

Zhe Kang Law (Trial medic) was responsible for the day-to-day running of the trial, critically appraised the manuscript and approved the final version.

Christian Ovesen led the spot sign subgroup analysis, critically appraised the manuscript and approved the final version.

Serefnur Ozturk (National Co-ordinator in Turkey) critically appraised the manuscript and approved the final version.

Stuart J Pocock (Lead Statistician) supervised the analysis performed by Katie Flaherty and Polly Scutt.

Ian Roberts (Co-applicant and member of TSC) critically appraised the manuscript and approved the final version.

Thompson G Robinson (Co-applicant and member of TSC) critically appraised the manuscript and approved the final version.

Christine Roffe (Co-applicant and member of TSC) critically appraised the manuscript and approved the final version.

Nils Peters (Lead Investigator in Switzerland) critically appraised the manuscript and approved the final version.

Polly Scutt (Statistician) performed the statistical analysis with supervision, critically appraised the manuscript and approved the final version.

Jegan Thanabalan (National Co-ordinator in Malaysia) critically appraised the manuscript and approved the final version.

David Werring (Co-applicant and member of TSC) critically appraised the manuscript and approved the final version.

David Whynes (Co-applicant and member of the TSC) critically appraised the manuscript and approved the final version.

Lisa Woodhouse (Statistician) performed some of the statistical analysis, critically appraised the manuscript and approved the final version.

Philip M Bath (Deputy Chief Investigator, co-conceived the trial) critically appraised the manuscript and approved the final version.

Publications

Data-sharing statement

Once completed, individual patient data from TICH-2 will be made available to the ‘Virtual International Stroke Trials Archive’ (VISTA) and, subsequently, over the web, as with the International Stroke Trial. Similarly, anonymised baseline and on-treatment neuroimaging data will be published. After publication of the planned primary and secondary analyses, the trial data may be shared, upon reasonable request to the corresponding author and the TSC.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.

Acute coronary syndrome

Ischaemic stroke

A clinical syndrome characterised by rapidly developing clinical symptoms and/or signs of focal (and at times global) loss of cerebral function, with symptoms lasting for > 24 hours or leading to death, with no apparent cause other than that of vascular origin. Classification as being ischaemic is based on results of a head CT/MRI imaging excluding haemorrhage.

Transient ischaemic attack

A sudden focal neurological deficit of the brain or eye presumed to be of vascular origin and lasts < 24 hours.

Peripheral arterial disease

A sudden blockage of a peripheral artery; the blockage may result from a blood clot, embolism, dissection or trauma. Symptoms usually start suddenly. Acute peripheral arterial limb occlusion includes the seven symptoms listed below:

1. severe pain

2. coldness

3. paraesthesia

4. loss of sensation

5. paleness in an extremity

6. lack of pulse in an extremity

7. blue skin in an affected limb.

Acute peripheral arterial limb occlusion can also affect the arteries that carry blood from the kidneys and the stomach.

Examples of evidence required:

• clinical

• radiological details, for example angiography.

Seizure/convulsions

Focal or generalised seizures, tonic–clonic seizures or partial seizures; diagnosed clinically after review by an appropriately trained physician.

Venous thromboembolism

This encompasses both deep-vein thrombosis (DVT) and pulmonary embolism (PE).

Examples of evidence required:

• clinical detail

• DVT –

  • ultrasound

  • venography

• PE –

  • VQ (i.e. a ventilation–perfusion) scan

  • CT pulmonary angiography scan.

3 December 2012

The protocol had to be amended following Research Ethics Committee (REC) review. The changes relate to participant identifiers and the consent process for potential participants who are not able to consent for themselves and do not have relatives present.

9 October 2013

Update to part B, section 3 (Radiation) on the REC and Research and Development Form.

28 February 2014

Five amendments were made to the protocol.

4 March 2015

The amendment was the addition of the MRI substudy.

24 July 2015

This amendment was to update MRI substudy consent forms to include a missing initial box and to remove the participant signature line on the personal legal representative consent form, as the participant would not sign this form.

9 November 2015

This was a minor amendment to the MRI substudy consent forms.

26 May 2016

Addition of plasma biomarkers substudy in one centre.

1 September 2016

To apply for a 12-month extension to recruitment, with recruitment planned to be completed by September 2017 and grant closure by February 2018.