PDF: IST-3
(2012)Objective
To assess the benefits and risks of tPA (0.9mg/kg) given within 6 hours from stroke onset in a wider range of patients.
Study Summary
• IST-3 did not show a significant benefit in the primary endpoint of functional independence at 6 months, but ordinal analysis favored tPA despite increased risk of symptomatic ICH.
Intervention
tPA administered at 0.9 mg/kg, up to 90 mg, within 6 hours of stroke onset.
Inclusion Criteria
Patients with acute ischemic stroke, treatable within 6 hours of stroke onset.
Study Design
Arms: tPA vs. Control
Patients per Arm: tPA: 1515, Control: 1515
Outcome
• QHS 0-1 at 6 months was 24% in tPA vs 21%, P=0.018 (after adjustment for risk factors), any ICH: 7% with tPA, 1% in placebo..
Bottom Line
Despite early hazards including increased intracranial hemorrhage and 7-day mortality, thrombolysis within 6 hours improved functional outcome at 6 months. Benefit did not seem to be diminished in elderly patients, with greater benefit observed in patients older than 80 years.
Major Points
- IST-3 was the largest individual randomized trial of stroke thrombolysis (3,035 patients, 156 centers, 12 countries), designed to answer whether tPA benefits a WIDER range of patients than the restrictive European license criteria — specifically elderly patients (>80 years) and those treated in the 3–6 hour window.
- 53% of patients were >80 years old — deliberately targeting the population EXCLUDED from NINDS, ECASS III, and all prior tPA trials. This was the first randomized evidence for thrombolysis in the elderly, an age group comprising ~30% of all stroke patients.
- 95% of enrolled patients did NOT meet the 2002 EU license criteria for tPA — making this a trial of an expanded indication, not a confirmatory trial. This deliberate design choice tested whether the 'uncertainty principle' could be used to study off-label populations.
- Primary outcome (alive and independent OHS 0–2 at 6 months): 36.6% tPA vs 35.1% control (OR 1.13, 95% CI 0.95–1.35, p=0.181) — NOT significant by conventional dichotomous analysis, a disappointing headline result.
- Pre-specified ordinal analysis SIGNIFICANT: shift across the entire OHS distribution favoring tPA (OR 1.27, 95% CI 1.10–1.47, p=0.001) — a more sensitive and informative analysis that captured benefit across the disability spectrum.
- Alive and favorable outcome (OHS 0–1) significantly favored tPA: 24% vs 21% (OR 1.26, p=0.018) — a 3% absolute increase in excellent outcomes. This secondary endpoint drove the ordinal shift.
- Symptomatic ICH dramatically increased: 7% tPA vs 1% control (OR 6.94, p<0.0001). Fatal ICH: 4% vs <1%. This was the highest sICH rate in any major tPA trial — reflecting the older, sicker population and the extended time window.
- Age interaction FAVORED the elderly: patients >80 years showed at least as much benefit as younger patients (interaction p=0.027, suggesting MORE benefit in elderly). This counterintuitive finding challenged the assumption that elderly patients have too much bleeding risk for thrombolysis.
- Time interaction: benefit greatest within 3 hours (consistent with all prior trials), but insufficient power to definitively exclude benefit at 3–6 hours. The 3–6 hour window remained uncertain until ECASS III and subsequently imaging-guided approaches (EXTEND, WAKE-UP).
- Key contribution to the Cochrane individual patient data meta-analysis of all tPA trials (2014): IST-3 data, combined with NINDS, ECASS, ATLANTIS, and EPITHET, provided the foundation for recommending tPA in patients >80 years — ultimately adopted by 2015 AHA/ASA guidelines.
Study Design
- Study Type
- International, multicenter, randomised, open-treatment trial (initially double-blind pilot phase)
- Randomization
- Yes
- Blinding
- Open-treatment for most patients (276 patients in initial double-blind phase)
- Sample Size
- 3035
- Follow-up
- 6 months primary analysis, 18 months extended follow-up in some countries
- Centers
- 156
- Countries
- 12 countries including UK, Norway, Sweden, Italy, Poland, Australia, others
Primary Outcome
Definition: Proportion of patients alive and independent (Oxford Handicap Score 0-2) at 6 months
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| 534/1520 (35.1%) | 554/1515 (36.6%) | - (0.95-1.35) | 0.181 |
Limitations & Criticisms
- Severely underpowered: recruited only 3,035 of the planned 6,000 patients — the failed primary outcome may simply reflect insufficient statistical power rather than absence of treatment effect. The ordinal analysis (more powerful) was significant.
- Predominantly open-label design (only 276 patients in initial blinded phase) — knowledge of treatment assignment could bias clinical management and outcome assessment. PROBE design mitigated but did not eliminate this.
- 11-year enrollment period (2000–2011) — background stroke care evolved dramatically during enrollment. Early-enrolled patients received different standard care than late-enrolled patients, introducing temporal confounding.
- Highest sICH rate of any major tPA trial (7%) — reflecting the high-risk population (elderly, extended window). Critics argued this made the risk-benefit ratio unfavorable, though 6-month mortality was identical (27% in both groups).
- Used the 'uncertainty principle' for enrollment — only patients where the physician was genuinely uncertain were randomized. This created selection bias: patients clearly expected to benefit or be harmed were not enrolled, potentially narrowing the study population to the therapeutic 'gray zone.'
- Regulatory and ethical challenges delayed recruitment — some countries withdrew or paused enrollment due to evolving guidelines, creating geographic bias and timeline inconsistencies.
- Open-label trial contamination: some control patients received tPA off-protocol, while some tPA-assigned patients had delayed or incomplete treatment — diluting the between-group difference.
- Oxford Handicap Score (used as primary outcome) differs slightly from the standard modified Rankin Scale — though they are highly correlated, this adds complexity to cross-trial comparisons.
- Could not definitively establish the 4.5–6 hour window — this question was later addressed by EXTEND (perfusion-guided) and WAKE-UP (MRI-guided), which used imaging selection rather than time alone.
Citation
The IST-3 collaborative group. The benefits and harms of intravenous thrombolysis with recombinant tissue plasminogen activator within 6 h of acute ischaemic stroke (the third international stroke trial [IST-3]): a randomised controlled trial. Lancet. 2012;379(9834):2352-2363.