Effect of intravenous urokinase vs best medicine treatment on functional outcome for patients with acute minor stroke (TRUST): a randomized controlled trial

Clinical Question (PICO)

In adults aged 18–80 years with acute ischemic stroke, NIHSS 0–5, and treatment possible within six hours of onset, does intravenous urokinase 1,000,000 units improve excellent functional outcome at 90 days compared with investigator-selected, guideline-based antithrombotic treatment?

The trial admitted both disabling and nondisabling deficits, but excluded patients already intended for standard alteplase or endovascular treatment. Prespecified subgroup variables included age, sex, NIHSS category, treatment timing, IIb/IIIa inhibitor use, hypertension, diabetes, dyslipidemia, and previous antithrombotic use. The later disabling/nondisabling comparison is treated here as an additional exploratory analysis: its prespecification and operational disability definition are not established in the reviewed materials.

Bottom Line

One million units of intravenous urokinase did not demonstrate improved 90-day excellent functional outcome: mRS 0–1 occurred in 427/503 patients (84.9%) assigned urokinase and 425/496 (85.7%) assigned medical treatment; adjusted RR 1.00, 95% CI 0.96–1.05, P=0.87. Symptomatic intracranial hemorrhage was uncommon, but four events and a very wide confidence interval do not establish equivalent safety, despite the abstract’s reassuring characterization. TRUST, Table 2.

Design

Trial type: Multicenter, randomized, open-label, blinded-endpoint superiority trial; not an equivalence or noninferiority trial.
N: 999 randomized after 1,005 were assessed.
- Urokinase assignment: 503.
- Best medical treatment assignment: 496.
Randomization: Central web-based allocation, 1:1, permuted blocks of four, stratified by center; sequence generated by an independent statistician.
Setting: 25 hospitals in China.
Enrollment: October 4, 2020–November 11, 2022; final follow-up February 20, 2023. The abstract’s October 2020–February 2023 interval combines recruitment and trial completion.
Mean follow-up: Not reported; the scheduled primary assessment was at 90 days.
Analysis: Intention-to-treat by randomized assignment, with log-binomial and logistic regression. The publication reports adjustment for age, sex, baseline NIHSS, systolic blood pressure, and myocardial infarction. Supportive per-protocol, as-treated, and instrumental-variable analyses addressed treatment switching.
Primary outcome: Excellent functional outcome, mRS 0–1 at 90 days, assessed with blinded endpoint evaluation.
Power assumption: 81.5% excellent outcome with control versus 88% with urokinase, a 6.5-percentage-point improvement; target 1,002 patients for 80% power, allowing 5% attrition.

Population

Inclusion Criteria

Age 18–80 years.
Acute ischemic stroke with NIHSS ≤5, including NIHSS 0.
Treatment initiation within six hours of onset.
First stroke, or previous stroke with premorbid mRS ≤1.
Brain CT or MRI and informed consent.

Exclusion Criteria

Selected exclusions from the article and supplementary Table S2:

Intracranial hemorrhage or an infarct exceeding one-third of the middle cerebral artery territory.
Other thrombolysis contraindications, including recent stroke, relevant bleeding risks, platelet count ≤100×10⁹/L, warfarin-associated INR >1.5, recent heparin with elevated aPTT, or persistently uncontrolled blood pressure ≥185/110 mm Hg despite treatment.
Prior stroke together with diabetes; pregnancy or lactation; urokinase allergy.
Severe illness with expected survival under three months, inability to complete follow-up, or conflicting trial participation.
An existing intention to receive standard alteplase thrombolysis or endovascular therapy. This is a major applicability boundary, not a trivial administrative exclusion.

Baseline Characteristics (overall or representative arm)

Median age 64 years overall; mean 62.7 years with urokinase versus 63.1 with control.
Women: 186/503 (37.0%) versus 185/496 (37.3%). These total 371/999, or 37.1% calculated from Table 1; the abstract prints an inconsistent 36.9%.
Median NIHSS: 3 (IQR 2–4) versus 2 (IQR 1–4).
Hypertension: 54.1% versus 51.0%; diabetes: 14.9% versus 14.5%; previous stroke: 30.0% versus 27.6%.
Mean systolic blood pressure: 152.7 versus 149.9 mm Hg.
Treatment before 4.5 hours: 70.0% versus 67.9%; the table and subgroup figure use slightly different notation at the 4.5-hour boundary.
Supplementary Table S9 classifies 163/999 patients (16.3%, calculated) as disabling and 836 as nondisabling; this is not a predominantly disabling-stroke population.

Interventions

Urokinase: 1,000,000 units in 100 mL saline, continuous intravenous infusion over 30 minutes, begun within six hours. The statistical analysis plan specifies guideline-based antiplatelet treatment beginning 24 hours after thrombolysis.
Best medical treatment: Guideline-based antithrombotic therapy selected and started immediately by the local investigator; the discussion says dual antiplatelet therapy predominated. A single uniform drug, loading dose, or duration was not imposed across the control arm.
Monitoring: Follow-up CT or MRI at 22–36 hours and blinded central adjudication of efficacy and safety endpoints.
Treatment switching: 114/496 controls (23.0%) crossed to urokinase; 63/503 urokinase-assigned patients (12.5%) crossed to medical treatment. The 177 switchers constitute 17.7% of all randomized patients, calculated from the reported counts.

Outcomes

Primary Outcome — mRS 0–1 at 90 days:

Urokinase: 427/503 (84.9%).
Control: 425/496 (85.7%).
Adjusted RR 1.00 (95% CI 0.96–1.05), P=0.87; adjusted OR 1.06 (0.73–1.53), P=0.76.
Calculated unadjusted absolute difference: −0.79 percentage points for urokinase. This is not the adjusted effect estimate.
Per-protocol primary outcome: 85.8% versus 86.3%, adjusted RR 1.01 (0.96–1.06), P=0.78; as-treated: 85.8% versus 86.0%, adjusted RR 1.01 (0.96–1.06), P=0.69. These supportive comparisons do not preserve randomization in the way ITT does.

Prespecified subgroup results:

The following are the adjusted odds ratios printed in Figure 3, not risk ratios. Each P value tests interaction across the two levels of the subgroup, not treatment efficacy within either level. Separate within-subgroup P values are not supplied in that figure. Prespecification does not remove multiplicity; the authors regard these analyses as exploratory.

Subgroup	First level: adjusted OR (95% CI)	Second level: adjusted OR (95% CI)	P for interaction
Age	<65 years: 0.72 (0.38–1.37)	≥65 years: 1.25 (0.79–1.99)	0.02
Sex	Men: 0.85 (0.54–1.35)	Women: 1.63 (0.87–3.03)	0.06
NIHSS	0–3: 1.57 (0.92–2.68)	4–5: 0.74 (0.42–1.29)	0.18
Onset to treatment	≤4.5 hours: 1.39 (0.89–2.16)	>4.5 hours: 0.64 (0.32–1.26)	0.16
IIb/IIIa inhibitor prescription	No: 1.03 (0.71–1.49)	Yes: unavailable in figure	0.39
Hypertension	No: 1.07 (0.60–1.94)	Yes: 1.11 (0.69–1.77)	0.21
Diabetes	No: 1.37 (0.91–2.07)	Yes: 0.41 (0.17–0.98)	<0.001
Dyslipidemia	No: 1.05 (0.72–1.53)	Yes: unavailable in figure	0.17
Prior antithrombotics	No: 1.00 (0.64–1.56)	Yes: 1.23 (0.64–2.37)	0.03

The diabetes subgroup had excellent outcome in 53/75 urokinase patients (70.7%) versus 62/72 controls (86.1%). This signal warrants scrutiny and replication; it is not a validated treatment-selection rule. Figure 3’s NIHSS-category denominators omit 10 urokinase-assigned and seven control-assigned participants; Figure 1 reports those numbers as having NIHSS >5 after randomization.

Additional disabling/nondisabling analysis:

Supplementary Table S9 reports, in ITT, 64/89 (71.9%) versus 54/74 (73.0%) for disabling deficits, OR 0.85 (0.42–1.72); and 363/414 (87.7%) versus 371/422 (87.9%) for nondisabling deficits, OR 1.20 (0.78–1.85). P for interaction is 0.61; within-stratum P values are not given. The table labels these estimates unadjusted but provides an adjustment footnote, so adjustment status is internally inconsistent. Its non-ITT population labels and totals also do not align with the main paper’s per-protocol/as-treated definitions. Do not use this exploratory table as proof that thrombolysis cannot help disabling minor stroke.

Secondary Outcomes:

New vascular events at 90 days: 33/503 (6.6%) versus 37/496 (7.5%); adjusted RR 0.91 (0.58–1.44), P=0.68.
Ordinal mRS shift: adjusted common OR 1.22 (0.95–1.55), P=0.12; mRS 5 and 6 were combined for the modeled ordinal endpoint.
Barthel Index ≥95: 56.2% versus 57.4%; adjusted RR 1.01 (0.94–1.09), P=0.72. These percentages use available outcome data rather than both full randomized denominators.
EQ-5D score, presented on a 0–100 scale in Table 2: 85.5 versus 86.4; adjusted mean difference −0.20 (−2.90 to 2.49), P=0.88.
Exploratory ordinal shifts were nominally favorable in the per-protocol analysis, common OR 1.33 (1.01–1.75), P=0.04, and as-treated analysis, 1.32 (1.03–1.70), P=0.03. Neither rescues the negative ITT primary outcome; both are vulnerable to selection and multiple testing. TRUST, supplementary Tables S5–S6.

Adverse Events / Safety:

Symptomatic intracranial hemorrhage within 36 hours, using the trial’s NINDS-based definition: 3/503 (0.6%) versus 1/496 (0.2%); adjusted RR 1.83 (0.16–20.27), P=0.62.
All-cause mortality at 90 days: 7/503 (1.4%) versus 13/496 (2.6%); adjusted RR 0.52 (0.21–1.33), P=0.18.
Any adverse event: 13.9% versus 14.3%; adjusted RR 0.97 (0.71–1.32), P=0.84.
Serious adverse events: 3.6% versus 5.6%; adjusted RR 0.61 (0.34–1.10), P=0.10.
Interpretation: Low observed bleeding rates do not equal a demonstrated absence of excess bleeding. The hemorrhage confidence interval is too imprecise for that claim.

Figures

Original, unaltered article images are embedded below; the explanatory captions are paraphrased. Article license: CC BY-NC-ND 4.0. Images remain attributable to Tao and colleagues and the original publication.

TRUST eligibility, allocation, treatment switching, and analysis populations — *Figure 1.* Participant flow through screening, randomization, assigned treatment, crossover, and the ITT, per-protocol, and as-treated analyses. NIHSS denotes the National Institutes of Health Stroke Scale.

Ninety-day modified Rankin Scale distribution by randomized group — *Figure 2.* Ninety-day mRS distributions in the randomized population. The scale spans no residual symptoms at 0, symptoms without important disability at 1, increasing disability at 2–5, and death at 6.

Prespecified subgroup odds ratios and treatment interaction tests — *Figure 3.* Excellent functional outcome by prespecified subgroup in the ITT population, with adjusted odds ratios, confidence intervals, and interaction tests. These exploratory comparisons should not be read as independently confirmed treatment indications.

Criticisms

Substantial contamination of the treatment contrast: 177 participants crossed treatment groups. ITT estimates the effect of assignment under this degree of switching, not perfectly adherent exposure. The discussion’s 23.8% crossover figure does not reconcile with 177/999=17.7%; the direction-specific counts are more transparent.
Negative superiority is not equivalence: A trial designed around a 6.5-point gain can reject that optimistic expectation without ruling out smaller benefits or harms. No equivalence margin was tested.
Disability is a crucial unresolved boundary: Only 163 participants were classified as disabling, the classification’s operational criteria are unclear, and patients intended for alteplase or endovascular therapy were excluded. The exploratory disability results cannot support withholding established reperfusion from otherwise eligible patients.
Open-label treatment and nonuniform comparator: Blinded endpoints reduce detection bias, but clinicians could influence crossover and cointerventions. “Best medical treatment” was not one standardized regimen.
Analysis documentation: The published five-variable adjustment set differs from the covariate list in the supplied statistical analysis plan. The latter specifies simple random imputation of missing baseline covariates and exclusion of patients without primary outcome data. Table 2 nevertheless uses all 999 randomized patients for its primary denominators; it does not establish complete data for every secondary endpoint. These distinctions should not be collapsed into a claim of universal complete follow-up.
Exploratory signals and reporting inconsistencies: Multiple subgroup tests were not multiplicity-adjusted. Supplementary Table S9’s adjustment and non-ITT population labels conflict internally; nominal ordinal-shift benefits in selected populations require caution, not promotion to primary evidence.
Sparse safety evidence: Four sICH events cannot establish safety equivalence, and nonsignificant mortality differences do not demonstrate a survival benefit.
External validity and dose: This was a Chinese cohort treated with a fixed 1,000,000-unit dose. It does not test larger or weight-adjusted urokinase doses, other thrombolytic drugs, patients over 80, or a US acute-stroke pathway.

Funding

National Key R&D Program of China, grant 2017YFC1308202. The funding statement does not specify the funder’s role in design, analysis, or publication; independence should not be assumed from that omission. The authors declared no competing interests. Trial registration: NCT04420351.

The paper

Authors. Yongli Tao et al., for the TRUST Trial Investigators.
Title. Effect of intravenous urokinase vs best medicine treatment on functional outcome for patients with acute minor stroke (TRUST): a randomized controlled trial.
Journal. BMC Medicine, 23:6.
Year. 2025; published January 6, 2025. This is a historical look-back, not a new October 2026 trial.
DOI. 10.1186/s12916-024-03820-2.
PMCID. PMC11702182.
Reviewed material. Complete Europe PMC JATS article, all three original figures, and both publisher-hosted supplementary documents, including the protocol/statistical analysis plan and Tables S1–S9.

Deep Dive — click to expand

What this is

Published in January 2025, TRUST extended the minor-stroke thrombolysis debate to a drug used in China: low-dose intravenous urokinase. Its 999-patient comparison found almost identical 90-day excellent-outcome rates, 84.9% versus 85.7%, with adjusted RR 1.00. The useful question is not whether this proves that every mild stroke should avoid thrombolysis. It is whether this particular treatment strategy earns a routine place in a selected, mostly nondisabling population—and how much uncertainty remains when treatment switching, dose choice, and outcome definitions are examined. TRUST.

1. Shadow Audit

The most revealing omission from a simple “negative but safe” summary is the difference between no measured benefit and adequately measured safety. A hemorrhage RR of 1.83 with CI 0.16–20.27 is not reassurance that harm has been excluded. The counts are small, not the uncertainty.

A second shadow is a genuine signal hidden by the broadly negative narrative: per-protocol and as-treated ordinal mRS analyses were nominally favorable, with common ORs 1.33 and 1.32 and P values 0.04 and 0.03. These do not overturn ITT—selection after randomization and multiple comparisons make them fragile—but “every analysis was negative” is inaccurate. The fair summary is that the randomized primary endpoint was negative, while exploratory selected-population shift analyses deserve neither erasure nor clinical elevation. Supplementary Tables S5–S6.

Finally, the treatment contrast was diluted by 114 switches toward urokinase and 63 away from it. The disability comparison contained only 163 patients and excluded people already intended for established reperfusion. TRUST did not randomize every patient with a disabling NIHSS 3 deficit who arrives in a US emergency department.

2. Inversion Engine

For the original optimistic conclusion to hold, urokinase would need to produce a 6.5-point absolute increase in excellent outcomes. At the observed control rate of 85.7%, that would mean roughly 92.2%, corresponding to RR about 1.076—not the observed adjusted RR of 1.00, whose upper confidence limit is 1.05. These are illustrative calculations using the control event rate, not a substitute for an adjusted absolute-effect model.

A more modest opposite conclusion remains possible: a carefully selected subgroup or a different dose might benefit. To establish it, investigators would need a reproducible treatment-by-selection interaction, clinically important absolute benefit, preserved randomized separation, and a bleeding estimate precise enough to assess net benefit. The diabetes signal points toward worse outcomes with urokinase in that subgroup, while the apparently favorable estimates for older patients or women have confidence intervals spanning no effect. Picking the most attractive forest-plot row is not inversion; it is selection after seeing the data.

3. Second-Order Catalyst

The first plausible implementation target is a Chinese hospital using urokinase as a routine option for low-NIHSS presentations—not a US hospital deciding between guideline-supported alteplase and tenecteplase. A local pathway review could remove automatic low-dose urokinase merely because a patient arrives within six hours, and require explicit disability assessment, appropriate etiologic treatment, and a documented reason for any exception. This is an implementation proposal, not an outcome tested by TRUST.

For US practice, the safeguard is already explicit in the AHA/ASA 2026 guideline summary, updated January 26, 2026: eligible disabling deficits warrant rapid thrombolytic treatment within 4.5 hours regardless of NIHSS score; nondisabling deficits in that window favor dual antiplatelet therapy. TRUST reinforces the need to separate these branches, but does not itself determine the drug choice or validate a six-hour US thrombolysis protocol.

4. Asymmetric Leverage

The asymmetric opportunity is better selection across a common presentation, rather than extracting a benefit from an unsuccessful drug strategy. The control group’s 85.7% excellent-outcome rate leaves limited room for a large average gain, yet the disabling subgroup’s control rate was only 73.0%. That difference identifies a higher-risk population needing reliable evidence; it does not identify urokinase as its solution.

Small process improvements applied to many nondisabling presentations may avoid unnecessary infusion and monitoring burdens. Conversely, missing a genuinely disabling deficit because the NIHSS is low can be consequential. Neither economic savings nor a net resource benefit was established by this trial. The leverage is to spend clinical attention on disability, contraindications, vascular assessment, and the right treatment branch—not to infer cost-effectiveness from a neutral efficacy endpoint.

5. Paradigm Destroyer

The reflex this paper challenges is “minor stroke plus an available thrombolytic equals a justified infusion.” The opposite reflex—“NIHSS ≤5 means never thrombolyse”—is equally unsupported.

A two-sentence protocol update consistent with the retrieved 2026 AHA/ASA summary would be: “Document whether the presenting deficit is disabling rather than using NIHSS alone to select treatment. Preserve rapid guideline-concordant reperfusion for eligible disabling deficits, while routing nondisabling presentations to the appropriate medical-treatment pathway; TRUST does not justify routinely adding low-dose urokinase.” This is an educational protocol principle, not a patient-specific prescription.

MVP — Minimum Viable Proof

For this drug, dose, window, and recruited population, the minimum defensible statement is: 427/503 versus 425/496 achieved mRS 0–1 at 90 days, adjusted RR 1.00 (0.96–1.05); routine incremental benefit was not demonstrated. That supports not introducing low-dose urokinase solely on the basis of minor stroke. It does not establish equivalence, erase the sparse-safety uncertainty, or prove that established thrombolysis is futile for disabling deficits. TRUST, Table 2.

Best Combination

Read TRUST alongside three distinct randomized questions rather than as a pooled vote against reperfusion:

PRISMS, 2018: alteplase versus aspirin in minor nondisabling stroke within three hours; no demonstrated functional benefit, but early termination limited certainty.
ARAMIS, 2023: dual antiplatelet therapy was noninferior to alteplase for excellent 90-day outcome in minor nondisabling stroke within 4.5 hours. Noninferiority of a specific comparator is not proof that all thrombolytic regimens are interchangeable.
TEMPO-2, 2024: tenecteplase in minor stroke with intracranial occlusion or focal perfusion abnormality within 12 hours; no functional benefit and signals of harm. Imaging selection and a different endpoint distinguish it from TRUST.

The synthesis is a case against indiscriminate thrombolysis in low-NIHSS presentations, especially nondisabling ones—not against timely treatment of an eligible disabling deficit. TRUST adds urokinase-specific evidence, with a six-hour window, a nonuniform medical comparator, and substantial treatment switching. Current guidance must remain separate from this historical trial’s own conclusions.

Overvalue Warning

Do not turn nonsignificant bleeding into “proven safe.” Three versus one sICH events, CI 0.16–20.27, cannot establish the absence of clinically meaningful excess hemorrhage; favorable mortality counts are similarly inconclusive.
Do not turn a neutral average into a universal rule or a subgroup prescription. The low-dose urokinase result does not generalize to every thrombolytic, disabling deficit, or treatment window. Diabetes, age, sex, and prior-antithrombotic interactions are exploratory; the inconsistencies in supplementary disability reporting add a reason to demand replication rather than change eligibility from a forest plot.

Educational use notice. The Time Is Brain Review is dedicated to medical education. Figures from published articles are embedded for teaching purposes, with attribution to the original publication. We do not claim copyright over any embedded figure; all rights remain with the original publisher and authors under the article's published license. If you are a rights holder and would like a figure removed, please contact the editorial team. See our editorial notice for the full policy.