Skip to main content

Foundational Trials

NINDS rt-PA Stroke Study

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Tissue plasminogen activator for acute ischemic stroke
  • Acronym: NINDS rt-PA Stroke Study
  • Year: 1995
  • Journal published in: New England Journal of Medicine
  • Citation: National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med. 1995;333(24):1581-1587.

Context & Rationale

  • Background
    • Acute ischaemic stroke was (and remains) a leading cause of death and long-term disability, with limited acute disease-modifying therapies available at the time.
    • Reperfusion with intravenous thrombolysis was biologically plausible but carried a feared trade-off: intracranial haemorrhage and potential early harm.
    • Early pilot work suggested that very-early administration of intravenous alteplase (rt-PA) could produce neurologic improvement, motivating a definitive, time-critical trial within 3 hours of symptom onset.
  • Research Question/Hypothesis
    • In adults with acute ischaemic stroke treated within 3 hours, does intravenous alteplase improve neurologic and functional outcomes versus placebo, with an acceptable safety profile?
    • Two-part programme: Part 1 focused on early neurologic improvement at 24 hours; Part 2 focused on disability outcomes at 3 months.
  • Why This Matters
    • A positive result would justify major system-level redesign: rapid recognition, prehospital activation, emergent CT imaging, and treatment in minutes rather than hours.
    • The clinical endpoint focus (disability at 3 months) directly addressed patient-centred outcomes, not just early neurological change.
    • Quantifying haemorrhagic risk was essential to support regulatory approval, guideline adoption, and safe implementation.

Design & Methods

  • Research Question: Among patients with acute ischaemic stroke treated within 3 hours of symptom onset, does intravenous alteplase (rt-PA) improve early neurologic recovery and 3-month functional outcomes compared with placebo?
  • Study Type: Randomised, multicentre, double-blind, placebo-controlled trial; stratified by clinical centre and time from onset to start of treatment (0–90 vs 91–180 minutes); two sequential parts with different primary endpoints.
  • Population:
    • Setting: Hospital-based acute stroke care (time-critical emergency evaluation with CT and laboratory testing).
    • Key inclusion: Acute ischaemic stroke with clearly defined time of onset; measurable neurologic deficit on the NIH Stroke Scale; baseline CT without intracranial haemorrhage; treatment start within 3 hours (0–180 minutes).
    • Key exclusions (selected, as reported): Stroke or serious head trauma in prior 3 months; major surgery in prior 14 days; any history of intracranial haemorrhage; rapidly improving or minor symptoms; symptoms suggestive of subarachnoid haemorrhage; GI or urinary tract haemorrhage in prior 21 days; arterial puncture at noncompressible site in prior 7 days; seizure at onset; systolic BP >185 mmHg or diastolic BP >110 mmHg (including if aggressive lowering required); anticoagulant use or heparin in prior 48 hours with elevated aPTT; prothrombin time >15 seconds; platelets <100,000/mm3; glucose <50 mg/dL or >400 mg/dL.
  • Intervention:
    • Intravenous alteplase (rt-PA; Activase) 0.9 mg/kg (maximum 90 mg).
    • Administration: 10% as an initial bolus, with the remainder infused over 60 minutes.
    • Key co-intervention constraints: no anticoagulants or antiplatelet agents for 24 hours after treatment; blood pressure maintained within prespecified limits.
  • Comparison:
    • Matching placebo administered using the same bolus/infusion schedule and the same post-treatment co-intervention constraints (including antithrombotic avoidance for 24 hours and blood pressure targets).
  • Blinding: Double-blind (patients, treating teams, and outcome assessors); CT scans were interpreted centrally by a radiologist blinded to clinical information including treatment assignment.
  • Statistics: Part 2 planned 320 patients (160 per group) to detect a 20 percentage point difference in a single outcome measure with 95% power at the 5% significance level; primary efficacy analysis in Part 2 used a prespecified global test across four 3-month outcomes; analyses were intention-to-treat.
  • Follow-Up Period: Primary assessments at 24 hours and 3 months; safety monitoring included CT imaging at 24 hours and 7–10 days and repeat CT for clinical deterioration suggesting haemorrhage.

Key Results

This trial was not stopped early. Interim analyses were performed once during Part 1 and once during Part 2, with continuation to completion.

Outcome Alteplase (rt-PA) Placebo Effect p value / 95% CI Notes
Primary efficacy (Part 2, 0–180 min): Global statistic across 4 disability measures at 3 months N=168 N=165 OR 1.7 95% CI 1.2 to 2.6; P=0.008 Global test jointly assessed Barthel Index (95–100), mRS (0–1), Glasgow Outcome Scale (1), NIHSS (0–1).
Barthel Index 95–100 at 3 months (Part 2, 0–180 min) 84/168 (50%) 63/165 (38%) RR 1.3 95% CI 1.0 to 1.7; P=0.026 Also reported: OR 1.6; 95% CI 1.1 to 2.5.
Modified Rankin Scale 0–1 at 3 months (Part 2, 0–180 min) 65/168 (39%) 43/165 (26%) RR 1.5 95% CI 1.1 to 2.0; P=0.019 Also reported: OR 1.7; 95% CI 1.1 to 2.6.
Glasgow Outcome Scale 1 at 3 months (Part 2, 0–180 min) 74/168 (44%) 53/165 (32%) RR 1.4 95% CI 1.0 to 1.8; P=0.025 Also reported: OR 1.6; 95% CI 1.1 to 2.5.
NIHSS 0–1 at 3 months (Part 2, 0–180 min) 52/168 (31%) 33/165 (20%) RR 1.5 95% CI 1.0 to 2.2; P=0.033 Also reported: OR 1.7; 95% CI 1.0 to 2.8.
Efficacy subgroup: Global statistic at 3 months (combined results, 0–90 min) N=157 N=145 OR 1.9 95% CI 1.2 to 2.9; P=0.005 Stratified by onset-to-treatment time (0–90 minutes).
Efficacy subgroup: Global statistic at 3 months (combined results, 91–180 min) N=155 N=167 OR 1.9 95% CI 1.3 to 2.9; P=0.002 Stratified by onset-to-treatment time (91–180 minutes).
Early neurologic improvement at 24 hours (combined results, 0–90 min) 87/157 (55%) 61/145 (42%) RR 1.3 95% CI 1.0 to 1.7; P=0.02 Improvement defined as ≥4-point NIHSS improvement from baseline or complete resolution.
Symptomatic intracerebral haemorrhage within 36 hours 6.4% 0.6% Not reported P<0.001 Abstract (combined). Table 6: Part 1 symptomatic ICH 6% vs 0%; Part 2 symptomatic ICH 7% vs 1%.
All-cause mortality at 90 days 54/312 (17%) 64/312 (21%) Not reported P=0.30 No statistically significant difference in mortality at 3 months.
Minor external bleeding within 10 days 23% 3% Not reported Not reported Most frequent non-intracranial bleeding difference reported between groups.
New ischaemic stroke during follow-up Part 1: 8%
Part 2: 4%		 Part 1: 7%
Part 2: 4%		 Not reported Not reported Rates were similar between groups in both parts.
  • Across multiple disability scales at 3 months, alteplase increased the proportion of patients with minimal/no disability (e.g., mRS 0–1: 39% with alteplase vs 26% with placebo in Part 2).
  • Symptomatic intracerebral haemorrhage within 36 hours was higher with alteplase (6.4% vs 0.6%; P<0.001), while 90-day mortality was not statistically different (17% vs 21%; P=0.30).
  • Efficacy was observed in both time strata (global OR 1.9 in both 0–90 and 91–180 minute strata), supporting benefit across the full 0–3 hour window tested.

Internal Validity

  • Randomisation and allocation: Permuted-block randomisation (blocks of varying size) stratified by centre and time-to-treatment (0–90 vs 91–180 minutes); allocation concealment method was not detailed in the report, but blinding and identical placebo reduced selection/performance risks.
  • Dropout / exclusions: Minimal missing primary outcome data; in Part 1, 1 of 291 patients had missing 3-month data; in Part 2, 7 of 1332 primary outcome measures were missing across 4 patients, with prespecified handling (later measurements or worst-score assignment if unavailable).
  • Performance/detection bias: Double-blinding; clinical outcome assessments performed by certified examiners who were not involved in initial treatment; CTs read centrally by a radiologist blinded to clinical data including treatment group.
  • Protocol adherence: Dose delivery was high: Part 1 full dose (±5%) in 90% (alteplase) vs 92% (placebo); Part 2 full dose in 93% vs 93%.
  • Baseline characteristics: Overall similar stroke severity (median NIHSS 14 vs 14 in Part 1; 14 vs 15 in Part 2); notable imbalances reported included weight in Part 1 (76 ± 15 kg vs 80 ± 18 kg) and age in Part 2 (69 ± 12 years vs 66 ± 13 years), with additional baseline covariate adjustment performed when variables differed at P<0.05.
  • Heterogeneity: Two-part design with distinct primary endpoints introduces structural complexity; however, Part 2 used a prespecified global disability endpoint, and treatment effects were directionally consistent across four functional measures and both time strata.
  • Timing: Treatment initiation was within 0–180 minutes by design; the trial demonstrated feasibility of delivering full imaging/laboratory evaluation, consent, and randomisation within 90 minutes in a large subgroup (N=302).
  • Dose: Alteplase dose and administration were fixed (0.9 mg/kg; max 90 mg; 10% bolus, remainder over 60 minutes), supporting internal consistency and limiting dose-related confounding.
  • Separation of the variable of interest: Clear pharmacologic separation (alteplase vs placebo) with protocolised avoidance of antiplatelets/anticoagulants for 24 hours and prespecified BP targets in both groups.
  • Outcome assessment: Use of established stroke outcome instruments; deaths before 3 months were assigned the worst possible score on all outcomes in the global test framework.
  • Statistical rigour: Part 2 primary analysis used a global test with planned gatekeeping (univariate testing only after global significance at 0.05); interim analyses occurred once per part with adjusted critical values; primary inference was intention-to-treat.

Conclusion on Internal Validity: Overall, internal validity appears strong given double-blinding, stratified randomisation, high protocol adherence, and minimal missing outcomes, although the two-part structure and modest baseline imbalances warrant attention when interpreting effect size and clinical implementation.

External Validity

  • Population representativeness: Typical moderate-severity ischaemic strokes were enrolled (median NIHSS 14–15), but only patients presenting and treatable within 3 hours were eligible, and “minor/rapidly improving” strokes, recent bleeding risks, and severe uncontrolled hypertension were excluded.
  • Applicability: Findings translate best to health systems capable of rapid prehospital recognition and streamlined in-hospital workflows (urgent CT, immediate stroke expertise, and blood pressure/laboratory control); applicability is limited where onset-to-needle time routinely exceeds 3 hours or where imaging/laboratory infrastructure is constrained.
  • Contemporary relevance: The core time-dependent principle remains applicable, but modern practice adds advanced imaging selection, endovascular therapy for large-vessel occlusion, and refined haemorrhage risk stratification.

Conclusion on External Validity: External validity is moderate: results are highly applicable to patients meeting the strict early-presenter criteria in stroke-capable systems, but less generalisable to the broader stroke population who present later or fall outside trial eligibility constraints.

Strengths & Limitations

  • Strengths: Rigorous randomised double-blind design; multicentre enrolment; explicit time-to-treatment stratification; standardised alteplase dosing; central blinded CT interpretation; clinically meaningful 3-month disability outcomes using multiple validated instruments; minimal missing outcome data.
  • Limitations: Two-part design with different primary endpoints; modest sample size by modern standards; strict eligibility and very-early treatment window limit generalisability; increased symptomatic intracerebral haemorrhage risk requires careful translation into real-world systems and governance.

Interpretation & Why It Matters

  • Disability-focused benefit
    Alteplase improved 3-month functional outcomes across multiple disability scales (e.g., Part 2 mRS 0–1: 39% vs 26%), establishing thrombolysis as a disability-reducing intervention rather than a mortality-reducing therapy.
  • Accepting a quantified harm
    The principal cost of benefit was increased early symptomatic intracerebral haemorrhage (6.4% vs 0.6%), demanding strict eligibility, haemorrhage surveillance, and disciplined post-lysis BP/antithrombotic management.
  • System-of-care transformation
    The trial’s feasibility and efficacy within 3 hours effectively created a new “emergency” phenotype for stroke care, driving modern stroke pathways (rapid imaging, protocolised triage, door-to-needle metrics, and specialist stroke-team activation).

Controversies & Subsequent Evidence

  • The accompanying editorial framed thrombolysis as crossing a clinical threshold while emphasising the necessity of system-level readiness and careful patient selection in view of haemorrhagic risk.1
  • Debate focused on whether chance baseline imbalances and the two-part structure could inflate apparent benefit; a formal reanalysis of the NINDS dataset reported that baseline imbalance did not account for the observed treatment effect and supported the trial’s primary conclusions.2
  • Subsequent pooled and individual-patient meta-analytic evidence reinforced a strong time-to-treatment relationship for intravenous alteplase, with the greatest net benefit early and attenuation with delay—contextualising the NINDS “within 3 hours” window as the high-yield part of a broader time-dependent effect spectrum.34
  • Implementation studies demonstrated that thrombolysis could be delivered in routine practice with acceptable haemorrhage rates, but also highlighted frequent protocol deviations and the importance of governance, audit, and time-based quality targets for safe scale-up.5
  • Major contemporary guidelines incorporate the NINDS paradigm: intravenous alteplase is recommended for eligible patients treated within 3 hours, with additional selection criteria used for extended windows and special populations in later evidence syntheses.67

Summary

  • In adults with acute ischaemic stroke treated within 3 hours, intravenous alteplase improved 3-month disability outcomes across multiple validated scales.
  • The primary Part 2 global disability endpoint showed higher odds of favourable outcome with alteplase (OR 1.7; 95% CI 1.2 to 2.6; P=0.008).
  • Symptomatic intracerebral haemorrhage within 36 hours was increased (6.4% vs 0.6%; P<0.001), while 90-day mortality was not statistically different (17% vs 21%; P=0.30).
  • Benefit was present across both 0–90 and 91–180 minute strata, supporting the entire 0–3 hour window tested.
  • The trial reshaped acute stroke systems by making time-critical imaging and treatment pathways central to stroke care delivery.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • The NINDS trial’s eligibility and workflow requirements (CT-first, strict BP and laboratory thresholds, antithrombotic delay) became foundational implementation constraints for intravenous alteplase programmes.

Overall Takeaway

The NINDS rt-PA Stroke Study established that intravenous alteplase administered within 3 hours of acute ischaemic stroke onset improves disability outcomes at 3 months, at the cost of increased early symptomatic intracerebral haemorrhage but without a demonstrable mortality penalty. It is “landmark” because it converted stroke into a time-critical, pathway-driven emergency condition and defined the modern paradigm of reperfusion-oriented acute stroke care.

Overall Summary

  • Alteplase within 3 hours improved 3-month functional outcomes (Part 2 global OR 1.7) while increasing early symptomatic ICH (6.4% vs 0.6%).
  • Mortality at 90 days was not statistically different (17% vs 21%), reinforcing disability (not survival) as the primary net benefit.
  • The trial’s operational achievement (treating large numbers within 90 minutes) helped create modern stroke systems of care.

Bibliography