Thrombolysis for minor nondisabling stroke: the evidence
What the PRISMS, ARAMIS and TEMPO-2 trials found on clot-dissolving treatment for mild, nondisabling stroke, and why 'minor' is hard to define.
By Manouchehr Hessabi, MD, MPH
Three randomized trials have asked whether clot-dissolving treatment helps people whose ischemic stroke is mild and not clearly disabling. None of them found that it improved recovery at 90 days, and two of them, PRISMS and TEMPO-2, reported more symptomatic bleeding or more deaths among treated patients. Each trial used a different design, so the more useful question is what each one could and could not conclude.
Thrombolysis means giving a drug that breaks down the clot blocking a brain artery. It is widely used in acute ischemic stroke, the kind caused by a blocked vessel rather than a bleed. The question here is narrower: does it still help when the deficits are mild?
What does "minor nondisabling stroke" mean?
Stroke severity in trials is usually scored with the National Institutes of Health Stroke Scale (NIHSS), a structured bedside exam that runs from 0 to 42, where higher scores mean more severe deficits. Trials of minor stroke typically enroll patients scoring 0 to 5. The authors of the PRISMS trial noted that more than half of patients with acute ischemic stroke present with scores in that range, yet the earlier landmark thrombolysis trials enrolled few patients whose deficits were not clearly disabling.
"Nondisabling" is a second layer, and it is partly a clinical judgment. The same low total can come from deficits that matter very differently in daily life, which is why trials add a judgment about whether the deficit is disabling. PRISMS enrolled patients whose deficits were "judged not clearly disabling." ARAMIS made the definition more explicit by also requiring no more than 1 point on several key individual items of the scale. The same total score can therefore describe different patients from trial to trial.
Outcomes were measured with the modified Rankin Scale (mRS), which grades disability from 0 (no symptoms) to 6 (death). A score of 0 or 1 is often called an excellent outcome. More on how this scale is used is in the explainer on the modified Rankin Scale.
What did PRISMS find, and why is it inconclusive?
PRISMS, published in JAMA in 2018, was a double-blind, double-placebo trial run in the United States. Each patient received either intravenous alteplase with an oral placebo or oral aspirin with an intravenous placebo, so neither the patient nor the treating team knew which was active. Patients had NIHSS scores of 0 to 5, deficits judged not clearly disabling, and treatment within 3 hours of onset.
The trial was designed for 948 patients but stopped early at 313. At 90 days, 78.2% of the alteplase group and 81.5% of the aspirin group had a favorable outcome (mRS 0 or 1), an adjusted risk difference of -1.1% with a 95% confidence interval from -9.4% to 7.3%. Symptomatic intracranial hemorrhage occurred in 5 alteplase patients (3.2%) and none in the aspirin group.
The width of that confidence interval is the key teaching point. A confidence interval shows the range of effects compatible with the data. An interval running from about 9 points of harm to 7 points of benefit cannot rule out a meaningful effect in either direction. That is why the investigators wrote that the very early termination "precludes any definitive conclusions." A trial that stops at one third of its planned size loses much of its ability to answer its own question.
What does a non-inferiority trial like ARAMIS actually test?
ARAMIS, published in JAMA in 2023, asked a different kind of question. A superiority trial tries to show that one treatment is better than another. A non-inferiority trial tries to show that a new or simpler approach is not worse than an established one by more than a prespecified amount, called the non-inferiority margin.
ARAMIS randomized 760 patients at 38 hospitals in China, within 4.5 hours of onset, to either dual antiplatelet therapy, a short course of clopidogrel plus aspirin, or intravenous alteplase. Excellent outcome at 90 days occurred in 93.8% of the dual antiplatelet group and 91.4% of the alteplase group, a risk difference of 2.3% (95% CI, -1.5% to 6.2%). The margin was -4.5%. Because the lower end of the interval (-1.5%) stayed above it, the trial met its non-inferiority criterion. Symptomatic intracerebral hemorrhage occurred in 0.3% and 0.9% of the two groups.
Three limits belong next to that result. The trial was open-label, meaning patients and clinicians knew the assignment, although outcomes were assessed by blinded raters. It was conducted in one country. And a non-inferiority result answers only the question it was built for: the simpler approach was not meaningfully worse. It does not by itself show that either treatment was better.
Why did TEMPO-2 test patients with a visible blocked artery?
TEMPO-2, published in The Lancet in 2024, started from a different hypothesis. A minor stroke with a visible intracranial occlusion, a blocked artery inside the skull seen on imaging, carries a higher risk of worsening, so these patients might be the ones most likely to benefit. The trial enrolled patients with NIHSS 0 to 5 and either an occlusion or a focal perfusion abnormality, within 12 hours of onset.
Patients were randomized to intravenous tenecteplase, a related clot-dissolving drug, or to non-thrombolytic standard care, across 48 hospitals in 10 countries. The trial was stopped early for futility after 886 patients. Return to baseline function at 90 days occurred in 75% of the control group and 72% of the tenecteplase group (risk ratio 0.96; 95% CI, 0.88 to 1.04). Twenty patients died in the tenecteplase group (5%) compared with five (1%) in the control group, an adjusted hazard ratio of 3.8 (95% CI, 1.4 to 10.2). Symptomatic intracranial hemorrhage occurred in 2% versus under 1%.
The investigators concluded there was "no benefit and possible harm." Like ARAMIS, TEMPO-2 was open-label with blinded outcome assessment. Its population also differs from the other two trials: these patients were selected because imaging showed a blocked vessel, and the drug was tenecteplase rather than alteplase. A separate explainer covers the comparison of tenecteplase versus alteplase in broader stroke populations.
How do the trials read together?
A 2025 meta-analysis in Neurology by Doheim and colleagues pooled randomized trials of intravenous thrombolysis in minor stroke. A concise evidence summary of that review, published by Long and Gottlieb in Academic Emergency Medicine, focused on 4 trials that enrolled only minor-stroke patients, 3,364 participants in total, defining minor stroke as NIHSS 5 or less whether or not the symptoms were disabling.
In that summary, thrombolysis did not improve excellent recovery (odds ratio 0.85; 95% CI, 0.7 to 1.03). It was associated with lower odds of 90-day functional independence, mRS 0 to 2 (odds ratio 0.7; absolute risk difference 2.5%; number needed to harm 40), and higher mortality (odds ratio 2.4; absolute risk difference 1.5%; number needed to harm 66).
These figures need some translation. An odds ratio below 1 means the outcome was less common in the treated group. The absolute risk difference is the plain percentage-point gap between groups. The number needed to harm turns that gap into a count: about 1 in 40 treated patients in these trials lost functional independence who otherwise would not have, and about 1 in 66 died who otherwise would not have. Absolute measures are usually easier to weigh than relative ones.
The summary's authors also listed real limitations. Heterogeneity was high: the trials used different definitions of disabling symptoms, different drugs (alteplase, tenecteplase and pro-urokinase), different time windows and different background care. Pooling such trials answers a broader question than any one of them asked. Patients whose only symptoms are poorly captured by the NIHSS, such as isolated slurred speech (dysarthria), unsteadiness (ataxia) or sensory symptoms, were not enrolled.
What does the evidence not settle?
First, PRISMS and ARAMIS deliberately excluded minor deficits that were judged disabling. Long and Gottlieb note that guidelines suggest thrombolysis for minor stroke with disabling symptoms, which is a different population from the one these two trials studied.
Second, the bedside judgment of "nondisabling" is not perfectly reproducible. Two clinicians can score the same patient identically and disagree about whether the deficit is disabling for that person's life and work. Trials can tighten definitions, as ARAMIS did, but they cannot remove judgment entirely.
Third, trial results describe trial populations. Each estimate is an average across the patients who were eligible, consented and were randomized. Reading a result well also means knowing which patients were analyzed, a point covered in the explainer on intention-to-treat analysis.
Frequently asked questions
What is the NIHSS? A standardized neurological exam scored from 0 to 42. Higher scores reflect more severe stroke deficits.
Why do trials stop early? Reasons include slow enrollment, a planned interim analysis showing the question is unlikely to be answered (futility), or clear benefit or harm. PRISMS was terminated before any interim analysis or unblinding; TEMPO-2 was stopped for futility.
What is a non-inferiority margin? The largest loss of benefit, set in advance, that investigators are willing to accept in exchange for a simpler or safer approach. A result is non-inferior if the confidence interval does not cross that line.
Does a mild stroke mean low risk? Not necessarily. The TEMPO-2 investigators described minor stroke with an intracranial occlusion as carrying an increased risk of poor outcomes, and Long and Gottlieb cite studies reporting that up to one third of minor-stroke patients have functional impairment within 90 days. These are statements about groups of patients, not predictions for an individual.
Across three designs, the randomized evidence points in one direction for mild, nondisabling stroke: no measurable gain in recovery, with signals of harm in some trials. How stroke outcomes are measured and studied is described further on the stroke outcomes research page, and the peer-reviewed publications are the place to continue into the primary literature.