Association between tirofiban administration and sex in outcomes following endovascular treatment for large vessel occlusion stroke — prognosis is not treatment interaction

Clinical Question (PICO)

In adults with anterior-circulation large-vessel occlusion undergoing endovascular thrombectomy within 24 hours of last known well, does self-reported sex modify the effect of intravenous tirofiban before EVT versus placebo on 90-day functional independence (modified Rankin Scale [mRS] 0–2), intracranial hemorrhage, or mortality?

This is a post hoc sex-based reanalysis, not a prospectively powered trial of different treatments for women and men. Additional exploratory strata were age, NIHSS, ASPECTS, stroke etiology, occlusion site, and onset-to-randomization time.

Bottom Line

The authors found no statistically significant sex-by-treatment interaction for 90-day independence (interaction P = 0.103), and no demonstrated tirofiban benefit within either sex. Women in the tirofiban arm were less often independent than men—83/200 (41.5%) versus 145/263 (55.1%); adjusted OR 0.57, 95% CI 0.36–0.90—but that comparison is a prognostic association within one arm, not proof that tirofiban harms women or benefits men.

Design

Trial type: Exploratory secondary analysis of RESCUE BT, an investigator-initiated, multicenter, randomized, double-blind, placebo-controlled trial. This reanalysis was published September 9, 2026; the underlying trial was published in 2022.
N: 948 participants in this analysis: 391 women and 557 men.
- Tirofiban: 463, including 200 women and 263 men.
- Placebo: 485, including 191 women and 294 men.
- The original report records 950 randomized, with two immediate consent withdrawals excluded from all analyses; 948 remained in the reported trial population.
Randomization: 1:1, fixed blocks of four, stratified by participating center, baseline NIHSS ≤17 versus >17, and intracranial internal carotid artery occlusion versus other sites. Sex was not randomized.
Setting: 55 comprehensive stroke centers in China.
Enrollment: October 10, 2018–October 31, 2021; final follow-up January 15, 2022.
Follow-up: Outcomes assessed at 90 days; a mean follow-up duration is not reported.
Analysis: Binary logistic regression for dichotomous outcomes and ordinal logistic regression for mRS shift. Adjustments included atrial fibrillation, NIHSS, ASPECTS, last-known-well-to-puncture time, occlusion site, smoking, and stroke etiology; age was added in sensitivity analyses. Sex-by-treatment interaction was tested directly. There was no multiplicity correction.
Primary outcome for this reanalysis: mRS 0–2 at 90 days. The parent trial's primary outcome was the full ordinal mRS distribution, not this dichotomy. The parent report had complete 90-day primary outcome follow-up; its analysis followed assigned treatment among the 948 retained participants.

Allocation, dosing, rescue-treatment details, and the distinction between 950 randomized and 948 analyzed are verified against the original RESCUE BT report.

Population

Inclusion Criteria

Acute ischemic stroke with proximal intracranial large-vessel occlusion, presenting within 24 hours of last known well.
NIHSS ≤30 and ASPECTS ≥6.
Intracranial internal carotid artery or M1/M2 middle cerebral artery occlusion confirmed by CTA, MRA, or angiography.
Planned EVT and informed consent from the patient or representative.

Exclusion Criteria

Intravenous thrombolysis after stroke onset.
Dual antiplatelet treatment within the preceding week.
These are the principal exclusions reproduced in this paper, not an exhaustive eligibility checklist. Large-core populations with ASPECTS <6 and patients receiving bridging thrombolysis are outside this cohort.

Baseline Characteristics

Women: median age 71 years (IQR 63–77), versus 64 (55–72) in men.
Baseline NIHSS: 17 (13–20) versus 15 (10–19); ASPECTS: 8 (7–9) in both groups.
Atrial fibrillation: 46.5% versus 23.5%; cardioembolism: 58.1% versus 32.1%.
Large-artery atherosclerosis: 31.7% versus 55.8%; smoking history: 0.5% versus 39.5%.
Last known well to puncture: median 350 versus 439 minutes; women did not have longer treatment delays on this measure.
Prestroke mRS 0–1: 97.2% versus 97.7%.

Interventions

Tirofiban plus EVT: Intravenous bolus 10 μg/kg, followed by 0.15 μg/kg/min for up to 24 hours; study drug began within five minutes of randomization, before EVT.
Placebo plus EVT: Matching saline on the same schedule.
Shared procedural care: Thrombectomy, with angioplasty/stenting when required. Under the original protocol, rescue medication was allowed if flow could not be maintained after angioplasty/stenting: placebo-assigned patients could receive rescue tirofiban. Aspirin and/or clopidogrel began at hour 20; dual therapy was used after angioplasty/stenting and otherwise single antiplatelet therapy. These protocol details matter: the contrast is assignment to routine pretreatment, not absolutely no exposure to tirofiban in controls.

These are descriptions of the research regimen, not a current bedside dosing recommendation.

Outcomes

Primary Outcome: mRS 0–2 at 90 days

The table below reports women versus men within each treatment arm. These ORs must not be relabeled as tirofiban-versus-placebo effects.

Treatment arm	Women independent	Men independent	Adjusted OR, women vs men (95% CI)	P
Tirofiban	83/200 (41.5%)	145/263 (55.1%)	0.57 (0.36–0.90)	0.02
Placebo	81/191 (42.4%)	138/294 (46.9%)	0.95 (0.60–1.49)	0.82

Sex-by-treatment interaction: P = 0.103; with additional age adjustment, P = 0.083. Neither establishes effect modification, and neither establishes equivalence.
Treatment comparisons within sex: The paper reports P = 0.70 for women and P = 0.06 for men for tirofiban versus placebo, referring to Supplementary Table 7 for complete adjusted estimates. The main-text counts imply unadjusted absolute differences of −0.9 percentage points in women and +8.2 points in men, calculated here as tirofiban minus placebo. These descriptive differences are not adjusted causal estimates, and the men's P value is not evidence of a proven subgroup benefit.
Other post hoc strata: No significant heterogeneity was reported across age ≤67/>67 years, NIHSS ≤16/>16, ASPECTS ≤7/>7, large-artery atherosclerosis/cardioembolism/other etiology, ICA/M1/M2 site, or onset-to-randomization <400/≥400 minutes in the tirofiban-arm sex comparison (all interaction P >0.05). These are exploratory within-arm sex analyses, not validated treatment-selection rules.

Secondary Outcomes

All effect estimates below again compare women with men within the stated arm:

Ordinal mRS improvement: Tirofiban adjusted common OR 0.72 (95% CI 0.49–1.05), P = 0.09; placebo 0.92 (0.64–1.34), P = 0.69; sex-by-treatment interaction P = 0.465.
mRS 0–1: Tirofiban 29.0% versus 41.8%, adjusted OR 0.55 (0.34–0.89), P = 0.01; placebo 28.8% versus 34.7%, adjusted OR 0.85 (0.52–1.38), P = 0.50; interaction P = 0.219.
mRS 0–3: Tirofiban 56.5% versus 68.4%, adjusted OR 0.67 (0.42–1.06), P = 0.09; placebo 56.0% versus 65.3%, adjusted OR 0.89 (0.57–1.40), P = 0.62; interaction P = 0.403.

Adverse Events / Safety

Outcome	Tirofiban: women / men	Placebo: women / men	Sex-by-treatment interaction P
sICH within 48 hours	20/200 (10.0%) / 25/263 (9.5%)	13/191 (6.8%) / 18/294 (6.1%)	0.923
Mortality at 90 days	40/200 (20.0%) / 44/263 (16.7%)	36/191 (18.8%) / 46/294 (15.6%)	0.811
Any ICH, reported percentages	39.0% / 31.7%	31.6% / 25.6%	0.696

sICH used the Heidelberg Bleeding Classification. The adjusted women-versus-men OR for sICH was 0.78 (95% CI 0.38–1.59), P = 0.49, with tirofiban, and 0.80 (0.34–1.91), P = 0.62, with placebo. For mortality, corresponding ORs were 0.91 (0.51–1.61), P = 0.74, and 0.90 (0.52–1.56), P = 0.71. Lack of a sex interaction does not prove absence of a drug-related bleeding risk; the original trial reported overall sICH of 9.7% versus 6.4%, an absolute difference of 3.3 points (95% CI −0.2 to 6.8).

Figures

Ninety-day disability by sex and treatment arm — *Figure 1.* Distribution of modified Rankin Scale scores at 90 days. (A) Distribution of modified Rankin Scale scores at 90 days, stratified by sex. The distribution of modified Rankin Scale scores at 90 days among male and female patients. Clinical outcomes at the 90-day follow-up are shown for male and female patients. (B) Distribution of modified Rankin Scale scores at 90 days by sex, stratified by treatment arm. Scores on the modified Rankin Scale range from 0 to 6, with 0 indicating no symptoms; 1, symptoms without clinical disability; 2, slight disability; 3, moderate disability; 4, moderately severe disability; 5, severe disability; and 6, death.

Ninety-day disability by treatment within each sex — *Figure 2.* Distribution of modified Rankin Scale scores at 90 days by treatment arm, stratified by sex. Scores on the modified Rankin Scale range from 0 to 6, with 0 indicating no symptoms; 1, symptoms without clinical disability; 2, slight disability; 3, moderate disability; 4, moderately severe disability; 5, severe disability; and 6, death.

Exploratory within-tirofiban sex comparisons across clinical subgroups — *Figure 3.* Subgroup analysis. The forest plot presents adjusted ORs and 95% CIs for achieving an mRS score of 0–2 at 90 days among women compared to men in the tirofiban arm. The numbers and percentages represent the patients who achieved an mRS score of 0–2 within each sex and subgroup. Adjusted ORs were estimated using multivariable logistic regression models adjusted for atrial fibrillation, baseline NIHSS score, baseline ASPECTS, the time from last known well to puncture, occlusion site, smoking status, and stroke etiology. Squares indicate point estimates, horizontal lines indicate 95% CIs, and the vertical line indicates an OR of 1. p-values for interaction were derived from the interaction term between sex and the corresponding subgroup variable. Abbreviations: ASPECTS, Alberta Stroke Program Early Computed Tomography Score; CE, cardioembolism; CI, confidence interval; ICA, internal carotid artery; LAA, large artery atherosclerosis; MCA, middle cerebral artery; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio.

Criticisms

The central comparison is easily misread. A significant women-versus-men association in one arm and a nonsignificant association in the other do not constitute a significant treatment interaction. The relevant interaction P was 0.103.
Age is conspicuously absent from the main adjustment set. Women were seven years older by the group medians. Adding age left the interaction nonsignificant (P = 0.083), but this does not eliminate residual prognostic or social confounding of sex comparisons.
Multiplicity and endpoint choice: Numerous dichotomies, interactions, and marginal-effect analyses were uncorrected. The signal at mRS 0–2 was not accompanied by a significant adjusted ordinal-mRS sex comparison. An isolated sex-by-ASPECTS mortality interaction in the tirofiban arm (P = 0.020) was not consistently replicated and should not drive treatment.
No mechanistic proof: Thrombus histology, platelet composition, and relevant sociobehavioral variables were unavailable. More cardioembolism among women does not establish the mechanism behind the observed association.
Power and applicability: Only 391 women participated; subgroup precision is limited. All centers were in China, IV thrombolysis was excluded, and ASPECTS was ≥6. Findings cannot be assumed to apply to bridging thrombolysis, large-core EVT, or distinct rescue-stenting indications.
Reporting boundary: The main text does not reproduce the complete adjusted treatment-effect estimates within sex from Supplementary Table 7. They are not reconstructed here. The parent trial reported no missing primary outcome data; the reanalysis does not give a comparably detailed new missing-covariate workflow, and any-ICH percentages use available observations.

Funding

This secondary analysis reports support from the National Natural Science Foundation of China Young Scientists Fund Project (Category C), grant 82501597. Its funding statement does not specify a separate funder role; the authors declare no commercial or financial conflicts.

For context, the original RESCUE BT trial was supported by Lunan Pharmaceutical Group, National Natural Science Foundation of China grants 82090040, 81901236, 82071323, and 81801157, and Army Medical University awards 2019XLC2008 and 2019XLC3016. Lunan manufactured and supplied study medication and placebo. The original report states that funders had no role in design, conduct, data management or analysis, manuscript preparation, or the publication decision. Registration: ChiCTR-IOR-17014167, not an NCT identifier.

The paper

Authors. Sheng Zhou, Yongjiao Dai, Lingyu Zhang, et al.
Title. Association between tirofiban administration and sex in outcomes following endovascular treatment for large vessel occlusion stroke.
Journal. Frontiers in Neurology, volume 17, article 1850978.
Year. 2026; published September 9.
DOI. 10.3389/fneur.2026.1850978
PMCID. PMC13597796
PMID. 42780889
Article and figure license. Creative Commons Attribution 4.0; original figures are reproduced with attribution, without modification.

Deep Dive — click to expand

What this is

A September 2026 reanalysis revisits a neutral 2022 randomized trial of tirofiban before thrombectomy. Its useful lesson is not a new sex-specific prescription, but the difference between who fares worse and whose treatment effect differs. Women receiving tirofiban had lower independence than men receiving tirofiban, yet the randomized treatment-by-sex interaction remained unproven. The commentary below is an interpretation of these data, not an assertion about current professional guidelines.

1. Shadow Audit

The visually striking finding is 41.5% independence in women versus 55.1% in men receiving tirofiban. The less conspicuous question is what happens when each sex is compared with its own randomized control: women were at 42.4% with placebo, and men at 46.9%. Women in this cohort were older, had higher NIHSS scores, and had nearly twice the prevalence of atrial fibrillation. Sex is therefore a bundle of prognostic differences here, not an assigned exposure.

The shadow finding is also statistical: the main adjusted model omitted age, and the adjusted ordinal sex comparison in the tirofiban arm crossed the null, despite the dichotomized mRS 0–2 association. That does not invalidate the signal; it narrows it. The defensible statement is an exploratory association at a particular disability threshold, with no established sex-specific treatment response.

2. Inversion Engine

For the opposite conclusion—"sex should determine tirofiban pretreatment"—to become persuasive, a randomized treatment contrast would need to differ reproducibly by sex, not merely yield two different within-arm P values. Here the primary interaction P is 0.103 and remains 0.083 with age adjustment, rather than demonstrating the claimed heterogeneity.

Calculated directly from the reported counts, the unadjusted treatment differences are −0.9 percentage points for women and +8.2 for men, a descriptive separation of 9.1 points. That is a signal worth testing prospectively, not a threshold already crossed. An adequately powered confirmation should prespecify the interaction scale and clinically important difference, show uncertainty around both treatment effects, and assess hemorrhage jointly. Conversely, a nonsignificant interaction here cannot establish that the two effects are identical.

3. Second-Order Catalyst

The first useful change belongs in a comprehensive stroke center's evidence-review process, not in its sex-based eligibility checklist. A thrombectomy quality committee can immediately require every proposed subgroup rule to show the randomized within-subgroup contrast and interaction estimate alongside the more memorable prognostic comparison.

For research teams, the next move is a prospectively specified sex-interaction analysis with adequate representation of women and better capture of baseline function, age, stroke mechanism, and social recovery determinants. Those steps could resolve uncertainty; a new "men only" tirofiban order would merely operationalize it.

4. Asymmetric Leverage

The asymmetric payoff is avoiding a large treatment-selection error from a small interpretive shortcut. Among the 391 women, independence was nearly the same across assigned treatments—83/200 versus 81/191—even though the within-tirofiban comparison with men looked unfavorable. Mistaking that prognostic contrast for drug harm could propagate a sex-specific exclusion unsupported by the randomized comparison.

There is a second asymmetry: an inexpensive statistical reporting requirement—put the interaction beside the subgroup headline—can prevent an expensive protocol change. This analysis does not supply a reliable number needed to treat or harm for either sex, and none should be manufactured from its exploratory absolute differences.

5. Paradigm Destroyer

The habit this paper should retire is "significant in one group, nonsignificant in another, therefore different effects." It should not retire thrombectomy eligibility for women, nor establish tirofiban pretreatment for men.

A proposed two-sentence local evidence-review update would read: "Do not add a sex-specific tirofiban pretreatment rule on the basis of this post hoc analysis. Any proposed change must specify the clinical setting, randomized treatment effects within each subgroup, the treatment-by-subgroup interaction, and bleeding uncertainty, and undergo review against the full contemporary evidence."

MVP — Minimum Viable Proof

The minimum proof needed for a sex-specific change is a replicated, prospectively specified, clinically meaningful difference in randomized benefit versus harm, not the observed women-versus-men adjusted OR of 0.57 inside a single arm. This paper does not provide that proof: its interaction P = 0.103 and imprecise safety comparisons leave the decision unresolved. Its immediate contribution is to identify the right confirmatory question.

Best Combination

Read the reanalysis alongside the 2022 RESCUE BT randomized report, whose primary ordinal-mRS result was neutral: adjusted common OR 1.08 (95% CI 0.86–1.36), P = 0.50. The parent trial also retained uncertainty about symptomatic bleeding, with sICH 9.7% versus 6.4%. The later sex analysis refines questions about heterogeneity; it does not replace that overall randomized result.

The practical synthesis is design-specific: routine pre-EVT treatment in a cohort excluding IV thrombolysis should not be conflated with tirofiban after lysis, without LVO, or as rescue therapy for procedural failure. Evidence from those settings must be appraised on its own eligibility, timing, and outcomes rather than imported as an answer to this trial.

Overvalue Warning

Do not turn a prognostic OR into a treatment OR. The 0.57 estimate compares women with men receiving tirofiban, not tirofiban with placebo in women. It cannot establish that the drug caused poorer outcomes in women.
Do not convert "no significant interaction" into equivalence or safety. Limited power can miss clinically important heterogeneity. Likewise, no detected sex difference in sICH does not remove the parent trial's uncertainty about an overall bleeding increase or establish safety in excluded populations.

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.