TICH-2
Tranexamic acid for hyperacute primary IntraCerebral Haemorrhage (TICH-2): an international randomised, placebo-controlled, phase 3 superiority trial
Year of Publication: 2018
Authors: Nikola Sprigg, Katie Flaherty, Jason P Appleton, ..., Philip M Bath
Journal: The Lancet
Citation: Lancet 2018; 391:2107-15
Link: http://dx.doi.org/10.1016/S0140-6736(18)31033-X
PDF: https://www.thelancet.com/action/showPdf...%2818%2931033-X
PDF: Clinical Question
Does intravenous tranexamic acid reduce haematoma expansion and improve functional outcome in adults with stroke due to intracerebral haemorrhage when administered within 8 hours of symptom onset?
Study Overview
Objective
Evaluate whether IV tranexamic acid reduces hematoma expansion and improves functional outcomes in patients with spontaneous ICH when administered within 8 hours of onset.
Study Summary
In patients with spontaneous ICH treated within 8 hours, tranexamic acid did not significantly improve 90-day functional outcomes compared to placebo. However, it reduced early deaths, hematoma expansion, and serious adverse events without increasing thromboembolic complications.
Intervention
- Multicenter, international, double-blind, placebo-controlled phase 3 trial. N=2325 adults with spontaneous ICH randomized 1:1 to receive:
- Tranexamic acid: 1 g IV bolus + 1 g IV infusion over 8 hrs
- Placebo: matching saline regimen Primary outcome: mRS at 90 days (ordinal analysis). Secondary outcomes included hematoma expansion, early mortality, adverse events, and quality of life measures.
Bottom Line
Tranexamic acid did not significantly improve functional status at 90 days in patients with acute intracerebral haemorrhage, despite reducing early deaths and haematoma expansion. While safe and inexpensive, larger trials are needed to confirm a clinically significant effect and identify benefiting subgroups.
Major Points
- TICH-2 is the largest randomized trial of tranexamic acid (TXA) for acute spontaneous intracerebral hemorrhage — 2,325 patients across 124 centers in 12 countries.
- Primary outcome (ordinal mRS shift at 90 days) was NOT significantly improved with TXA (aOR 0.88, 95% CI 0.76–1.03, p=0.11), making this a formally negative trial.
- However, TXA significantly reduced hematoma expansion at 24h (25% vs 29%, aOR 0.80, p=0.03) and early deaths by day 7 (9% vs 11%, aOR 0.73, p=0.04).
- The disconnect between reduced expansion/early death and unchanged 90-day mRS suggests hematoma expansion is necessary but not sufficient for poor outcome — other mechanisms (perihematomal edema, IVH, medical complications) also drive disability.
- TXA was remarkably safe — no increase in venous thromboembolic events (3% vs 3%) or arterial occlusive events, and actually FEWER serious adverse events at all time points.
- Informed by CRASH-2 (trauma) and WOMAN (postpartum hemorrhage), TICH-2 used the same dosing regimen: 1g IV bolus + 1g over 8h infusion.
- The only significant subgroup interaction was baseline SBP ≤170 mmHg (interaction p=0.019), suggesting TXA may be most effective when hypertensive hematoma expansion is less of a competing mechanism.
- Enrolled within 8h of onset — broader than many ICH trials. Median time to randomization was 3.6 h (IQR 2.6–5.0), with 36% enrolled within 3 h.
- The authors concluded there is insufficient evidence to support the routine use of tranexamic acid in clinical practice for spontaneous ICH, though results do not exclude a small effect.
- Several smaller ongoing RCTs of TXA in ICH were noted (STOP-MSU, STOP-AUST, TRAIGE, TRANSACT, TICH-NOAC; combined enrolment <1000), with a planned individual patient data meta-analysis to detect smaller treatment effects or identify subgroups most likely to benefit.
Design
Study Type: International, randomised, placebo-controlled, double-blind, parallel group, phase 3 superiority trial
Randomization: 1
Blinding: Concealed from patients, outcome assessors, and all other health-care workers involved in the trial. Central independent expert assessors for CT scans and serious adverse events were masked to treatment assignment.
Enrollment Period: March 1, 2013, and Sept 30, 2017
Follow-up Duration: 90 days (primary outcome)
Centers: 124
Countries: Denmark, Georgia, Hungary, Ireland, Italy, Malaysia, Poland, Spain, Sweden, Switzerland, Turkey, UK
Sample Size: 2325
Analysis: Intention-to-treat; ordinal logistic regression for primary outcome adjusted for stratification and minimisation criteria; sensitivity analyses for primary outcome; multiple linear regression for continuous secondary outcomes; binary logistic regression for binary secondary outcomes; Cox proportional hazards regression for time-to-event data. No adjustment for multiplicity.
Inclusion Criteria
- Adults (≥18 years) with acute spontaneous (non-traumatic) intracerebral hemorrhage confirmed on CT.
- Symptom onset (or last seen well) within 8 hours of randomization.
- No minimum NIHSS or hematoma volume specified (broad severity range allowed above the GCS ≥5 threshold defined by exclusion criteria).
- Clinical diagnosis of stroke with ICH.
Exclusion Criteria
- ICH secondary to anticoagulation (warfarin, DOACs) — these patients have a different pathophysiology requiring reversal agents.
- ICH secondary to thrombolysis (sICH after tPA/TNK).
- ICH due to trauma (traumatic contusions, epidural/subdural hematomas).
- Known underlying structural cause: AVM, aneurysm, tumor, cavernoma.
- Tranexamic acid contraindicated (known allergy, active thromboembolic disease, seizure history attributed to TXA).
- Prestroke dependency (mRS >4) — to ensure meaningful functional outcome assessment.
- GCS <5 — moribund patients unlikely to benefit from intervention.
- Life expectancy <3 months from comorbid illness.
- Pregnancy or breastfeeding.
- Participation in another interventional clinical trial.
Baseline Characteristics
| Characteristic | Control | Active |
|---|---|---|
| Age, years | 68.7 (13.9) [20-101] | 69.1 (13.7) [29-97] |
| Age >70 | 580 (50%) | 584 (50%) |
| Sex, male | 659 (57%) | 642 (55%) |
| Ethnic origin - White | 992 (85%) | 986 (85%) |
| Ethnic origin - Other | 172 (15%) | 174 (15%) |
| Onset to randomisation, h | 3.7 (2.6-5.0) [0.8-8.0] | 3.6 (2.6-5.1) [1.0-20.8] |
| Onset to randomisation ≤3h | 412 (35%) | 421 (36%) |
| Onset to randomisation ≤4.5h | 796 (68%) | 779 (67%) |
| History - Previous antiplatelet therapy | 295 (25%) | 316 (27%) |
| History - Statin use prior to admission | 303 (26%) | 319 (28%) |
| History - Previous stroke or transient ischaemic attack | 156 (14%) | 173 (15%) |
| History - Ischaemic heart disease | 92 (8%) | 110 (10%) |
| Prestroke mRS | 0 (0-1) [0-4] | 0 (0-1) [0-4] |
| Glasgow Coma Scale | 14 (2.1) [5.0-15.0] | 13 (2.2) [5.0-15.0] |
| NIHSS score | 13 (7.5) [0.0-42.0] | 13 (7.5) [0.0-41.0] |
| Systolic blood pressure, mm Hg | 174 (26.8) [99.0-265] | 172 (27.5) [98.0-265] |
| Diastolic blood pressure, mm Hg | 94 (17.8) [35.5-162] | 93 (18.4) [46.0-179] |
| Haematoma location - Supratentorial lobar | 359 (31%) | 379 (33%) |
| Haematoma location - Supratentorial deep | 696 (60%) | 675 (58%) |
| Haematoma location - Infratentorial | 76 (7%) | 73 (6%) |
| Haematoma location - Combination | 33 (3%) | 34 (3%) |
| Intracerebral haematoma volume (mL) | 12.5 (5.1-31.9) [0.0-163] | 14.1 (5.9-32.4) [0.0-207] |
| Intraventricular haemorrhage | 363 (31%) | 382 (33%) |
| CT angiography done | 128 (11%) | 121 (11%) |
| Spot positive | 32 (25%) | 24 (20%) |
| Spot negative | 96 (75%) | 97 (80%) |
Arms
| Field | Tranexamic acid | Control |
|---|---|---|
| Intervention | 1 g intravenous tranexamic acid bolus infused over 10 min, followed by an 8 h infusion of 1 g tranexamic acid. | Matching placebo (normal saline 0.9%), administered with an identical regimen. |
| Duration | 8 hours | 8 hours |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Functional status at day 90, measured by shift in the modified Rankin Scale (mRS), using ordinal logistic regression. | Primary | 0.88 | 0.11 | ||
| Haematoma expansion (absolute increase >6 mL or relative growth >33%) at 24 hours | Secondary | 304 (29%) of 1058 participants | 265 (25%) of 1054 participants | 0.8 | 0.0300 |
| Change in haematoma volume from baseline to 24 hours (mL) | Secondary | 4.90 (16.0) | 3.72 (15.9) | -1.37 | 0.0432 |
| Death by day 7 | Secondary | 123 (11%) of 1164 participants | 101 (9%) of 1161 participants | 0.73 | 0.0406 |
| NIHSS day 7 (mean difference) | Secondary | 10.29 (8.3) | 10.13 (8.3) | -0.43 | 0.10 |
| Death by day 90 | Secondary | 249 (21%) of 1155 participants | 250 (22%) of 1152 participants | 0.92 | 0.37 |
| mRS >3 at day 90 (dead or dependent) | Secondary | 826 (72%) | 814 (71%) | 0.82 | 0.08 |
| Length of stay in hospital, days | Secondary | 63.73 (48.1) | 63.12 (47.1) | 1.09 | 0.16 |
| Disposition at discharge - Home | Secondary | 453 (39%) | 465 (40%) | 1.14 | 0.20 |
| Disposition at discharge - Institution | Secondary | 506 (43%) | 505 (43%) | 0.99 | 0.90 |
| Disposition at discharge - Died by discharge | Secondary | 205 (18%) | 190 (16%) | 0.83 | 0.15 |
| EQ-5D health utility status at day 90 (out of 1) | Secondary | 0.34 (0.4) | 0.34 (0.4) | 0.01 | 0.30 |
| EQ-VAS at day 90 (out of 100) | Secondary | 48.34 (33.1) | 48.81 (33.8) | 1.75 | 0.13 |
| Barthel index at day 90 (out of 100) | Secondary | 53.21 (43.7) | 52.92 (44.0) | 1.7 | 0.20 |
| TICS-M cognition at day 90 (out of 39) | Secondary | 13.94 (12.8) | 13.57 (12.5) | -0.19 | 0.69 |
| Zung Depression Scale at day 90 (out of 100) | Secondary | 67.29 (29.9) | 67.28 (29.5) | -0.35 | 0.76 |
| Global analysis (Wei-Lachin test) at day 90 | Secondary | 0.85 | |||
| If discharged, days well at home | Secondary | 72.15 (29.1) | 69.94 (28.6) | -0.72 | 0.64 |
| _Note | Adverse | Safety analysed in intention-to-treat population: tranexamic acid n=1161, placebo n=1164. | |||
| Serious adverse events by day 2 | Adverse | 417 of 1164 (36%) | 379 of 1161 (33%) | 0.0272 | |
| Serious adverse events by day 7 | Adverse | 497 of 1164 (43%) | 456 of 1161 (39%) | 0.0200 | |
| Serious adverse events by day 90 | Adverse | 556 of 1164 (48%) | 521 of 1161 (45%) | 0.0393 | |
| Venous thromboembolic events | Adverse | 37 of 1164 (3%) | 39 of 1161 (3%) | 0.98 | |
| Arterial occlusions (myocardial infarction, acute coronary syndrome, or peripheral arterial occlusion) | Adverse | Note: No increase in arterial occlusive events in the tranexamic acid group compared with placebo (reported qualitatively; exact counts in appendix). | |||
| Seizure | Adverse | 85 of 1164 (7%) | 77 of 1161 (7%) | ||
| Nervous system disorders (most common serious adverse event) | Adverse | 163 of 1164 (14%) | 149 of 1161 (13%) | ||
| Infections (second most common serious adverse event) | Adverse | 116 of 1164 (10%) | 98 of 1161 (8%) | ||
Criticisms
- Formally negative trial — the primary outcome (ordinal mRS shift at 90 days) did not reach significance (p=0.11). The biologically plausible positive signals (reduced expansion, fewer early deaths) may reflect chance or underpowering.
- Wide inclusion criteria (any spontaneous ICH within 8h, no minimum volume) created a heterogeneous population — many patients with small hematomas that wouldn't expand regardless, diluting treatment effect.
- Median time to randomization was 3.6 h — most patients treated >3h post-onset. TXA may be most effective ultra-early when active bleeding is ongoing, and no significant time-by-treatment interaction was detected.
- No screening logs collected — unable to assess selection bias or determine how enrolled patients differed from those screened but excluded.
- The disconnect between reduced day-7 mortality and unchanged day-90 mortality raises concerns about delayed deaths (infections, withdrawal of care) not being prevented by TXA.
- Effect size smaller than expected (aOR 0.88 vs anticipated 0.79) — trial may have been underpowered for the true effect size. Original sample size calculation assumed a larger treatment effect.
- Only 11% of patients had CTA performed — the spot sign (a marker of active extravasation) could identify the ideal TXA target population, but this subgroup was drastically underpowered.
- No direct comparison with other hemostatic strategies (e.g., platelet transfusion for antiplatelet-associated ICH, rFVIIa). TICH-2 only tested TXA vs placebo.
- The 8-hour time window may have been too broad — biological plausibility for antifibrinolytic benefit is strongest within the first 1-3 hours when hematoma expansion is most active.
Subgroup Analysis
Array
Funding
National Institute of Health Research Health Technology Assessment Programme and Swiss Heart Foundation.
Based on: TICH-2 (The Lancet, 2018)
Authors: Nikola Sprigg, Katie Flaherty, Jason P Appleton, ..., Philip M Bath
Citation: Lancet 2018; 391:2107-15
Content summarized and formatted by NeuroTrials.ai.