CHOICE2
Adjunctive Intra-Arterial Alteplase After Successful Thrombectomy for Acute Ischemic Stroke: The CHOICE-2 Randomized Clinical Trial
Clinical Question
Among patients with acute ischemic stroke due to large vessel occlusion who had successful recanalization after endovascular thrombectomy, does adjunctive intra-arterial alteplase improve functional outcomes compared with thrombectomy alone?
Bottom Line
In this multicenter Spanish RCT (n=440) of anterior-circulation LVO stroke with successful thrombectomy (eTICI 2b50-3), adjunctive intra-arterial alteplase (0.225 mg/kg, max 20 mg) increased 90-day mRS 0-1 from 42.5% to 57.5% (adjusted RD 15.0%, 95% CI 5.7-24.3; P=.002; NNT ≈ 7). Residual hypoperfusion at 36 h was halved (28.6% vs 50.5%). Symptomatic ICH was not increased (1.4% vs 0.5%; P=.33), but 90-day mortality was higher (12.1% vs 6.4%; adjusted RD 5.9%, 95% CI 0.5-11.3; P=.03), warranting further study.
Major Points
- Investigator-initiated, multicenter, open-label, blinded-endpoint, phase 3 RCT at 14 Spanish stroke centers (Dec 11, 2023 – Nov 26, 2025); 440 randomized (221 vs 219); 433 in modified full analysis set (214 vs 219).
- Population: anterior-circulation LVO (ICA, MCA M1/proximal M2, or ACA) treated with thrombectomy within 24 h and achieving expanded TICI 2b50-3; median NIHSS 15, median ASPECTS 10, prior IV thrombolysis in 274/433 (63%; 192 alteplase, 82 tenecteplase).
- Primary outcome — mRS 0-1 at 90 d: 57.5% (123/214) IA-alteplase vs 42.5% (93/219) control; adjusted risk difference 15.0% (95% CI, 5.7-24.3); P=.002; consistent across prespecified subgroups (age, sex, prior IV lytic, eTICI, time-to-reperfusion, glucose).
- Ordinal (shift) analysis of 90-d mRS: no significant between-group difference (P=.08); proportional-odds assumption was violated, so the common OR is not interpretable.
- Secondary imaging — residual hypoperfusion on 36-h CT perfusion: 28.6% (55/192) vs 50.5% (96/190); adjusted RD −22.0% (95% CI, −31.5 to −12.4); P<.001. Even in patients with complete angiographic reperfusion (eTICI 3), residual hypoperfusion was 14.6% vs 25.8% (post hoc).
- Volume of hypoperfusion (5.0 vs 6.1 mL) and infarct expansion ratio (56.6 vs 65.4) did not differ significantly.
- Patient-reported EQ-5D-5L median 0.9 vs 0.8; adjusted median difference 0.07 (95% CI, 0.01-0.12); P=.02. Barthel Index 95-100: 63.6% vs 58.4% (RD 5.1, 95% CI −4.1 to 14.3; P=.28).
- Safety — symptomatic ICH at 36 h: 1.4% (3/214) vs 0.5% (1/219); adjusted OR 3.10 (95% CI, 0.32-30.0); P=.33. Parenchymal hematoma type II 1.9% vs 1.8%; any asymptomatic ICH 24.3% vs 20.1%.
- Mortality at 90 d: 12.1% (26/214) vs 6.4% (14/219); adjusted RD 5.9% (95% CI, 0.5-11.3); P=.03. Only 1 death was adjudicated as related to the study medication; most deaths occurred beyond the first week.
- A recent meta-analysis of 6 RCTs (n=1971) showed no difference in 90-d mortality (18.3% vs 18.3%; RR 1.00) or sICH (RR 1.14) with adjunctive IA thrombolysis, suggesting the CHOICE-2 mortality signal may reflect random variation.
Design
Study Type: Investigator-initiated, multicenter, open-label, blinded-outcome-assessment, phase 3 randomized clinical trial
Randomization: 1
Blinding: Open-label to treating clinicians and patients regarding group assignment; outcome assessors, imaging core-laboratory readers, and statisticians remained blinded to allocation throughout; mRS assessed by certified central assessors via telephone/video.
Enrollment Period: Dec 11, 2023 - Nov 26, 2025
Follow-up Duration: 90 days (primary endpoint)
Centers: 14
Countries: Spain
Sample Size: 440
Power Calculation: Powered to detect an absolute difference of 14 percentage points in mRS 0-1 at 90 days, assuming a control event rate of 40% and 2-sided alpha 0.05; total 440 patients (220 per arm) accounting for a 5% dropout rate; also provided ≥95% power to detect an 18 percentage-point difference in residual hypoperfusion assuming a control rate of 58%.
Analysis: Primary analysis in modified full analysis set (all randomized who initiated assigned treatment; identical to safety population). Per-protocol population excluded major deviations; exploratory full analysis set included all randomized. Primary outcome: adjusted risk difference from binomial regression with identity link, adjusted for prior IV thrombolysis. Sensitivity analyses: log-binomial models, additional baseline covariates, treating center as random effect, as-randomized, as-treated, and treated per-protocol populations. Prespecified subgroups: age, sex, prior IV thrombolysis, eTICI, time to reperfusion, baseline glucose. Ordinal (shift) mRS by proportional-odds model (with van Elteren sensitivity). Hypoperfusion volume and infarct expansion ratio by generalized linear mixed models (γ distribution, log link). Missing 90-day mRS handled by hierarchical imputation (central blinded → local → mRS 4 if alive at 90±14 d); multiple imputation as sensitivity (100 datasets, Rubin's rules). Two-sided P<.05; no adjustment for multiplicity of secondary outcomes. Software: SAS 9.4M7 (TS1M7).
Inclusion Criteria
- Age ≥18 years
- Acute ischemic stroke due to isolated occlusion of internal carotid artery, MCA M1 or proximal M2, or anterior cerebral artery
- Prestroke mRS ≤1 (no substantial prestroke disability)
- Endovascular thrombectomy achieving expanded TICI score of 2b50 to 3 (≥50% reperfusion), as adjudicated by local investigators on cerebral angiography
- Enrollment within 4.5 h of last known well with ASPECTS ≥6 on NCCT (1 point subtracted per defined region with early ischemic change)
- OR enrollment within 24 h if advanced imaging demonstrated salvageable ischemic tissue per institutional protocols
Exclusion Criteria
- Contraindications to intravenous thrombolysis
- Known coagulopathy
- Intracranial hemorrhage on imaging
- Stenting requiring dual antiplatelet therapy
- Large established infarction on baseline imaging
- Admission NIHSS score >25
- Thrombectomy passes per vessel >3 (or >5 total per patient) — protocol limited procedure to ≤3 passes/vessel or ≤5/patient
Arms
| Field | Thrombectomy + intra-arterial alteplase | Control |
|---|---|---|
| Intervention | Standard endovascular thrombectomy (stent retriever, aspiration, or combined; ≤3 passes/vessel or ≤5/patient) followed immediately by intra-arterial alteplase 0.225 mg/kg (maximum 20 mg) infused continuously over 15 min via microcatheter positioned in the terminal ipsilateral internal carotid artery, after confirmation of successful reperfusion (eTICI 2b50-3) | Standard endovascular thrombectomy per operator practice (stent retriever, aspiration, or combined; ≤3 passes/vessel or ≤5/patient); no adjunctive intra-arterial thrombolytic therapy |
| N | 221 | 219 |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Excellent functional outcome — modified Rankin Scale score of 0 or 1 at 90 days, adjudicated by central blinded assessors | Primary | 93/219 (42.5%) | 123/214 (57.5%) | ≈ 7 | .002 |
| Residual hypoperfusion on 36-h CT perfusion (yes/no) | Secondary | 96/190 (50.5%) | 55/192 (28.6%) | <.001 | |
| Volume of hypoperfusion at 36 h, adjusted mean, mL | Secondary | 6.1 (95% CI, 4.5-8.2) | 5.0 (95% CI, 3.7-6.6) | .34 | |
| Infarct expansion ratio (final infarct / initial ischemic tissue), adjusted mean | Secondary | 65.4 (95% CI, 49.6-86.2) | 56.6 (95% CI, 42.9-74.6) | .47 | |
| Barthel Index 95-100 at 90 d (functional independence in ADLs) | Secondary | 128/219 (58.4%) | 136/214 (63.6%) | .28 | |
| EQ-5D-5L score at 90 d, median (Q1-Q3) [IQR] | Secondary | 0.8 (0.7-1.0) [0.3] | 0.9 (0.7-1.0) [0.3] | .02 | |
| Modified Rankin Scale score at 90 d — ordinal (shift) analysis by proportional-odds model | Secondary | distribution — see figure | distribution — see figure | .08 | |
| Symptomatic intracranial hemorrhage at 36 h (ECASS III criteria) | Safety | 1/219 (0.5%) | 3/214 (1.4%) | .33 | |
| All-cause mortality at 90 days | Safety | 14/219 (6.4%) | 26/214 (12.1%) | .03 | |
| Any serious adverse event at 90 d | Safety | 55/219 (25.1%) | 61/214 (28.5%) | ||
| Parenchymal hematoma type I at 36 h | Safety | 9/219 (4.1%) | 13/214 (6.1%) | ||
| Parenchymal hematoma type II at 36 h | Safety | 4/219 (1.8%) | 4/214 (1.9%) | ||
| Subarachnoid hemorrhage at 36 h | Safety | 13/219 (5.9%) | 13/214 (6.1%) | ||
| Hemorrhagic infarction type I at 36 h | Safety | 9/219 (4.1%) | 7/214 (3.3%) | ||
| Hemorrhagic infarction type II at 36 h | Safety | 9/219 (4.1%) | 15/214 (7.0%) | ||
| Any asymptomatic intracranial hemorrhage at 36 h | Safety | 44/219 (20.1%) | 52/214 (24.3%) | ||
| Vessel reocclusion | Safety | 2/219 | 5/214 |
Subgroup Analysis
Prespecified subgroup analyses (age, sex, prior IV thrombolysis, expanded TICI score, time to reperfusion, baseline glucose) showed the treatment effect on the primary outcome was directionally consistent across all subgroups with no significant treatment × baseline characteristic interactions. Notably, benefit was preserved in patients with complete angiographic reperfusion (eTICI 3) and among those who had received prior IV thrombolysis, with no evidence of heterogeneity by sex. Post hoc: residual hypoperfusion was associated with a higher risk of unfavorable outcome (mRS 2-6) — adjusted RD 11.0% (95% CI, 1.0-21.0%).
Criticisms
- Open-label design (placebo not feasible due to continent-wide alteplase shortage prioritizing active-drug production); mitigated by central blinded outcome assessment.
- Higher 90-day mortality in the intervention arm (12.1% vs 6.4%; RD 5.9%, 95% CI 0.5-11.3; P=.03) — although only 1 death was adjudicated related to study medication and most deaths were late, requires further investigation.
- Ordinal (shift) analysis of mRS was not significant (P=.08) and proportional-odds assumption was violated, so the binary excellent-outcome effect may not translate into whole-scale improvement across the disability spectrum.
- Noncontrast CT limits sensitivity for volumetric infarct measurements and detection of subtle infarct progression — possibly explaining the dissociation between improved perfusion and unchanged infarct expansion ratio.
- Incomplete documentation of screening exclusions (443 patients: 'reason not available') may have introduced selection bias.
- Trial conducted almost exclusively in Spain (~92%; 35/433 patients [8%] were from 19 different countries outside Spain) within a single national health system with relatively uniform care — limits generalizability, particularly to racially/ethnically diverse populations underrepresented here.
- Residual baseline imbalances possible (median age 2 y higher in intervention arm), though post hoc adjustments did not materially alter results.
- Some secondary outcomes had missing CT-perfusion data (22 IA-alteplase and 29 control patients missing follow-up perfusion, 10-13%).
Funding
Instituto de Salud Carlos III and grant ICI22/00064 from the European Union Next Generation EU/Mecanismo Para la Recuperación y la Resiliencia/Plan de Recuperación, Transformación y Resiliencia (Spain). Funders/sponsors had no role in design, conduct, analysis, interpretation, preparation, review, or approval of the manuscript, or decision to submit.
Based on: CHOICE2 (JAMA, 2026)
Authors: Renú A, Fernández-Couto MD, de la Riva P, et al; CHOICE-2 Investigators
Citation: JAMA. 2026;335(21):1859-1869. DOI: 10.1001/jama.2026.5164
Content summarized and formatted by NeuroTrials.ai.