← Back

Extended-Window IVT Meta-Analysis

IV Thrombolysis in the Extended Time Window for Acute Ischemic Stroke: An Updated Systematic Review and Meta-Analysis

Share Share Copied! Compare

Year of Publication: 2026

Authors: Palaiodimou L, Papageorgiou NM, Romoli M, et al; Tsivgoulis G (senior author)

Journal: Neurology

Citation: Neurology 2026;107(3):e218294. DOI: 10.1212/WNL.0000000000218294

Link: https://doi.org/10.1212/WNL.0000000000218294

PDF: https://www.neurology.org/doi/pdf/10.121...000000000218294

Clinical Question

In adults with acute ischemic stroke presenting >4.5 hours from last known well, does IV thrombolysis added to best medical therapy improve 90-day functional outcomes and safety compared with best medical therapy alone?

Bottom Line

Across 12 RCTs and 1 IPDM (4,867 patients), IVT beyond 4.5 hours significantly improved excellent functional outcome at 90 days (mRS 0–1: 40.2% vs 32.5%; RR 1.23, 95% CI 1.15–1.33; NNT=13) with a consistent shift toward less disability, at the cost of a higher symptomatic ICH rate (3.1% vs 1.2%; RR 2.11; NNH=75) but no increase in 90-day mortality. Benefit was consistent across thrombolytic agents (alteplase and tenecteplase), time windows (4.5–9 h, 4.5–24 h, wake-up), imaging strategies, vascular territories, and with or without EVT.

        Major Points
        13 studies pooled: 12 RCTs plus 1 individual patient data meta-analysis (Campbell 2019 pooling EXTEND, ECASS4-EXTEND, EPITHET); IVT arm n=2,456 (mean age 68.2, 37% female, mean NIHSS 11.5) vs BMT n=2,411 (68.6 y, 38% female, NIHSS 11.0); mean time from last known well ~10.7–11.1 hours; 16% received EVT in each arm.
Included trials span alteplase (0.6–0.9 mg/kg) and tenecteplase (0.25 mg/kg): ATTENTION LATE, Campbell IPDM (EXTEND / ECASS4-EXTEND / EPITHET), EXPECTS, HOPE, OPTION, ROSE-TNK, THAWS, TIMELESS, TNK PLUS, TRACE III, TRACE 5, TWIST, WAKE-UP.
Primary efficacy — excellent functional outcome (mRS 0–1 at 90 d): 40.2% (980/2,439) IVT vs 32.5% (780/2,400) BMT; RR 1.23 (95% CI 1.15–1.33; PI 1.14–1.34; I²=0%; p<0.01); NNT=13 (95% CI 9–21); no subgroup differences by agent, time window, imaging, territory, or EVT.
Secondary efficacy — mRS 0–2: 54.7% vs 48.0%; RR 1.15 (1.05–1.25; I²=59%; p=0.002). Disability shift (≥1-point mRS reduction): common OR 1.26 (1.13–1.39; I²=0%; p<0.001).
Primary safety — sICH: 3.1% (76/2,429) IVT vs 1.2% (29/2,367) BMT; RR 2.11 (1.36–3.28; I²=0%; p<0.001); NNH=75 (95% CI 37–231); sICH RR 1.52 (0.75–3.08) in EVT trials vs 5.03 (2.35–10.77) in non-EVT trials (subgroup p=0.02).
Secondary safety — any ICH: 12.5% (105/841) vs 9.4% (79/839); RR 1.32 (1.00–1.74; p=0.051; 4 studies). All-cause mortality at 90 d: 13.2% (321/2,438) vs 12.6% (300/2,387); RR 1.03 (0.89–1.18; p=0.72).
Post hoc subgroup within advanced imaging (perfusion CTP/MRP vs DWI-FLAIR mismatch): no significant differences across any outcome, with consistent direction of effect.
Trial sequential analysis: cumulative Z-curve crossed both the conventional (Z=1.96) and O'Brien-Fleming sequential monitoring boundaries for the primary outcome — evidence deemed conclusive for the assumed 3% absolute risk difference.
GRADE certainty: Moderate for mRS 0–1, mRS 0–2, and disability shift; High for sICH (upgraded 1 grade for RR>2); Low for any ICH and 90-day mortality (imprecision).
Small study effects: Egger p=0.01 for sICH suggested some small-study bias; no other outcomes showed asymmetry; leave-one-out sensitivity confirmed robustness of all outcomes except any ICH (borderline).

      

Design

Study Type: Aggregate-data systematic review and meta-analysis of RCTs (PRISMA 2020) with trial sequential analysis and GRADE evidence rating

Randomization: 1

Blinding: Included trials: WAKE-UP, TIMELESS, and the 3 IPDM component trials were placebo-controlled; others were open-label or open-label with blinded outcome adjudication

Enrollment Period: Included trials span April 2001 (EPITHET) through August 2025 (OPTION)

Follow-up Duration: 90 days for all primary and secondary outcomes

Centers: 0

Countries: China, USA, Australia, New Zealand, Canada, Japan, Belgium, UK, Germany, France, Taiwan, Finland, Norway, Sweden, Denmark, Italy, Greece, Multiple international sites

Sample Size: 4867

Power Calculation: Trial sequential analysis: type I error 5%, type II error 10%, anticipated intervention effect RD=3% for mRS 0–1; heterogeneity variance from Cochrane empirical distribution; continuity correction 0.5 for zero-event trials; RTSA v0.2.2 package.

Analysis: DerSimonian-Laird random-effects pairwise meta-analysis; RRs (double-arcsine variance stabilization) for dichotomous outcomes and generic inverse-variance common ORs for mRS shift; I² and Cochran Q for heterogeneity (thresholds 25%/50%); prediction intervals reported; leave-one-out sensitivity; funnel plots and Egger regression (when ≥4 studies); prespecified subgroups by agent, time window (4.5–9 h, 4.5–24 h, >4.5 h/wake-up), imaging (advanced vs standard), circulation, EVT; post hoc perfusion vs DWI-FLAIR subgroup; GRADE for certainty; intention-to-treat data; R v3.5.0 (meta, RTSA).

Inclusion Criteria

Randomized controlled trials or individual patient-data meta-analyses of RCTs (PICO population)
Adult patients with acute ischemic stroke
Presentation >4.5 hours after last known well (including wake-up and unknown-onset strokes)
Intervention: IV thrombolysis (alteplase or tenecteplase) plus best medical therapy
Comparator: best medical therapy alone, with or without placebo
Concomitant endovascular treatment permitted in either arm
Reported prespecified functional (mRS) or safety (ICH, mortality) outcomes
No language restrictions; database inception through March 3, 2026

Exclusion Criteria

Observational cohort studies, noncontrolled studies, case series, case reports
Trials directly comparing thrombolytic agents without a nonthrombolysis control arm
Editorials, narrative reviews, commentaries
Design/protocol papers without outcome data

Baseline Characteristics

Characteristic	IVT + BMT (n=2,456)	BMT alone (n=2,411)
Mean age, y	68.2	68.6
Female n (%)	~37%	~38%
Mean baseline NIHSS	11.5	11.0
Mean time from last known well, h	10.7	11.1
Received EVT n (%)	~16%	~16%
Studies contributing	13 (12 RCTs + Campbell IPDM of EXTEND/ECASS4-EXTEND/EPITHET)	13

Arms

Field	IV Thrombolysis + BMT	Control
Intervention	Alteplase 0.6 or 0.9 mg/kg (max 90 mg) OR tenecteplase 0.25 mg/kg (max 25 mg) plus guideline-directed best medical therapy; concomitant EVT permitted	Guideline-directed best medical therapy alone (with or without matching placebo); concomitant EVT permitted
N	2456	2411

Outcomes

Outcome	Type	Control	Intervention	HR / OR / RR	P-value
Excellent functional outcome at 90 days, defined as modified Rankin Scale score 0–1	Primary	780/2,400 (32.5%)	980/2,439 (40.2%)	1.23	<0.01
Good functional outcome (mRS 0-2) at 90 d	Secondary	1,151/2,400 (48.0%)	1,334/2,439 (54.7%)	1.15	0.002
Reduced disability (≥1-point shift across mRS) at 90 d	Secondary				<0.001
Symptomatic intracranial hemorrhage (trial-defined)	Safety	29/2,367 (1.2%)	76/2,429 (3.1%)	2.11	<0.001
Any intracranial hemorrhage	Safety	79/839 (9.4%)	105/841 (12.5%)	1.32	0.051
All-cause mortality at 90 d	Safety	300/2,387 (12.6%)	321/2,438 (13.2%)	1.03	0.72

Subgroup Analysis

Excellent functional outcome — no significant subgroup interactions: thrombolytic agent (alteplase vs tenecteplase) p=0.36; time window (4.5–9 h vs 4.5–24 h vs >4.5 h/wake-up) p=0.91; neuroimaging (advanced vs standard) p=0.78; affected circulation (anterior vs posterior vs both) p=0.98; EVT administration p=0.26. sICH — significant subgroup interaction only for EVT stratification (p=0.02): RR 1.52 (0.75–3.08) with EVT vs RR 5.03 (2.35–10.77) without EVT; all other subgroup interactions non-significant (agent p=0.18; time window p=0.42; imaging p=0.11; circulation p=0.17; sICH definition p=0.31). Post hoc: perfusion-based vs DWI-FLAIR mismatch imaging — no significant differences across any outcome. Leave-one-out sensitivity: robust for all outcomes except any ICH (borderline). Trial sequential analysis: cumulative Z-curve crossed both conventional (Z=±1.96) and O'Brien-Fleming sequential monitoring boundaries — evidence conclusive for the assumed 3% absolute risk difference.

Criticisms

Aggregate study-level data only — no individual patient data; cannot precisely define optimal patient selection subgroups.
Substantial clinical heterogeneity across trials in inclusion criteria (time window 4.5–9 h to 24 h to wake-up), imaging paradigms (CT alone vs perfusion vs DWI-FLAIR), core-volume/mismatch thresholds, and vascular territory.
Egger regression p=0.01 for sICH suggests small-study effects for the primary safety outcome.
sICH definitions differ across trials; a formal harmonized definition was not applied.
Standard-imaging trials disproportionately enrolled posterior circulation strokes and cannot be directly generalized to unselected anterior circulation patients.
3 trials (ATTENTION LATE, TNK PLUS, TRACE-5) were only presented at the 2026 International Stroke Conference — not yet peer-reviewed at time of publication.
Any-ICH outcome relied on only 4 trials, was borderline significant, and was sensitive to leave-one-out.
Concomitant EVT in ~16% of patients complicates isolation of IVT's independent pharmacologic effect.

Funding

No targeted funding reported by the authors.

Based on: Extended-Window IVT Meta-Analysis (Neurology, 2026)