← Back
NeuroTrials.ai
Neurology Clinical Trial Database

SELECT2

Trial of Endovascular Thrombectomy for Large Ischemic Strokes

Share Share Copied! Compare

Year of Publication: 2023

Authors: Amrou Sarraj, Ameer E Hassan, Michael G Abraham, ..., Gregory W Albers

Journal: New England Journal of Medicine

Citation: Sarraj A, Hassan AE, Abraham MG, et al. Trial of Endovascular Thrombectomy for Large Ischemic Strokes. N Engl J Med. 2023;388(14):1259-1271.

Link: https://doi.org/10.1056/NEJMoa2214403

Clinical Question

Does endovascular thrombectomy within 24 hours after stroke onset lead to better functional outcomes than standard medical care alone in patients with acute ischemic stroke due to large vessel occlusion with large ischemic-core volume?

Bottom Line

Among patients with large ischemic strokes (ASPECTS 3-5 or core ≥50mL), endovascular thrombectomy resulted in significantly better functional outcomes than medical care alone at 90 days. Thrombectomy nearly tripled the rate of functional independence (20% vs 7%) with similar mortality, though 18.5% of thrombectomy patients experienced procedural vascular complications.

Major Points

  • First major international trial enrolling patients based on either low ASPECTS (3-5) OR large perfusion/diffusion core volume (≥50mL)
  • Trial stopped early for efficacy at second interim analysis after 352 patients enrolled
  • Primary outcome: 90-day mRS distribution significantly favored thrombectomy (gOR 1.51, p<0.001)
  • Functional independence (mRS 0-2): 20.3% thrombectomy vs 7.0% medical care (RR 2.97)
  • Independent ambulation (mRS 0-3): 37.9% vs 18.7% (RR 2.06)
  • Mortality similar: 38.4% thrombectomy vs 41.5% medical care (RR 0.91, NS)
  • Symptomatic ICH very low in both groups: 0.6% vs 1.1%
  • Procedural complications occurred in 18.5% of thrombectomy patients (dissection 5.6%, perforation 3.9%, vasospasm 6.2%)
  • Early neurologic worsening more common with thrombectomy (24.7% vs 15.5%)
  • 85% of patients had ASPECTS ≤5; 75% had core volume ≥50mL; 67% met both criteria
  • Successful reperfusion (mTICI 2b-3) achieved in 79.8% of thrombectomy patients
  • Quality of life scores (Neuro-QoL) better in thrombectomy group across all domains

Design

Study Type: Phase 3, international, multicentre, randomised, open-label, adaptive enrichment trial with blinded endpoint assessment

Randomization: 1

Blinding: Open-label with blinded outcome assessment. Trained, certified assessors unaware of trial-group assignments and imaging results collected 30-day and 90-day outcomes.

Enrollment Period: September 2019 to September 2022

Follow-up Duration: 90 days

Centers: 31

Countries: USA, Canada, Spain, Switzerland, Australia, New Zealand

Sample Size: 352

Analysis: Intention-to-treat. Two-sided Wilcoxon-Mann-Whitney test for superiority. Effect size determined with Wilcoxon-Mann-Whitney probability of superiority and assumption-free generalized odds ratios with 95% CI. mRS scores 5 and 6 merged for analysis. Multiple imputation under missing-at-random assumption. Overall two-sided type I error rate controlled at 0.05, accounting for cumulative alpha of 0.018 spent at interim analyses. Originally designed for 560 patients with 90% power at alpha 0.05 to detect standardized difference of 0.34.

Inclusion Criteria

  • Age 18-85 years
  • Acute ischemic stroke due to occlusion of internal carotid artery (cervical or intracranial) or M1 segment of middle cerebral artery
  • Large ischemic core defined as: ASPECTS 3-5 on non-contrast CT, OR ischemic core volume ≥50mL on CT perfusion (rCBF <30%), OR ischemic core ≥50mL on diffusion-weighted MRI (ADC <620×10⁻⁶ mm²/s)
  • No upper limit for ischemic-core volume
  • Within 24 hours of last known well
  • Pre-stroke modified Rankin Scale score 0-1
  • No evidence of intracranial hemorrhage on neuroimaging

Exclusion Criteria

  • Pre-stroke disability (mRS >1)
  • Intracranial hemorrhage on baseline imaging
  • Multiple or ineligible clot locations
  • More than 24 hours since stroke onset

Arms

FieldEndovascular thrombectomy plus medical careControl
InterventionEndovascular thrombectomy performed with stent retrievers, aspiration devices, or both (various manufacturers, depending on trial site), plus standard medical care according to institutional protocols in accordance with AHA/ASA, ESO, and Stroke Foundation guidelines. Approximately 60% of procedures performed under general anesthesia. Tandem occlusions and isolated cervical ICA occlusions permitted.Standard medical care according to institutional protocols in accordance with AHA/ASA, ESO, and Stroke Foundation guidelines, including blood-pressure management, critical care, and in-hospital and outpatient rehabilitation. Intravenous thrombolytics administered to eligible patients within 4.5 hours. Decompressive hemicraniectomy permitted according to local practices.
DurationAcute intervention within 24 hours of last known well, with 90-day follow-up90-day follow-up

Outcomes

OutcomeTypeControlInterventionHR / OR / RRP-value
Ordinal score on modified Rankin Scale at 90 days (range 0-6, with scores 5 and 6 merged for analysis)PrimaryMedian mRS 5 (IQR 4-6)Median mRS 4 (IQR 3-6)<0.001
Functional independence (mRS 0-2) at 90 daysSecondary12/171 (7.0%)36/177 (20.3%)RR 2.97Not adjusted
Independent ambulation (mRS 0-3) at 90 daysSecondary32/171 (18.7%)67/177 (37.9%)RR 2.06Not adjusted
Successful reperfusion (mTICI 2b-3)SecondaryN/A142/178 (79.8%)
Early neurologic improvement (≥8-point NIHSS reduction or NIHSS 0-1 at 24h)Secondary13/172 (7.6%)20/174 (11.5%)RR 1.47Not significant
Neuro-QOL Mobility domain (higher = better)Secondary25.1 (16.5-33.0)35.2 (23.9-43.9)Difference 10.10Not adjusted
Neuro-QOL Depression domain (higher = worse)Secondary53.6 (46.8-57.4)47.9 (43.1-54.3)Difference -5.70Not adjusted
Neuro-QOL Social domain (higher = better)Secondary33.5 (27.7-37.8)37.1 (32.7-42.0)Difference 3.60Not adjusted
Neuro-QOL Cognitive domain (higher = better)Secondary37.9 (30.9-42.9)41.9 (35.0-49.6)Difference 4.00Not adjusted
Discharge location - HomeSecondary10 (5.7%)19 (10.7%)
Discharge location - Inpatient rehabilitationSecondary65 (37.4%)72 (40.4%)
Discharge location - HospiceSecondary19 (10.9%)11 (6.2%)
In-hospital deathSecondary44 (25.3%)42 (23.6%)
mRS distribution at 90 daysSecondarymRS 0: 0%, mRS 1: 1.7%, mRS 2: 5.2%, mRS 3: 11.5%, mRS 4: 15.2%, mRS 5: 8.4%, mRS 6: 40.8%mRS 0: 1.1%, mRS 1: 5.1%, mRS 2: 14.0%, mRS 3: 17.4%, mRS 4: 20.7%, mRS 5: 18.4%, mRS 6: 38.2%
Symptomatic intracranial hemorrhage within 24hAdverse2 (1.1%)1 (0.6%)RR 0.49Not significant
Parenchymal hematomaAdverse3 (1.7%)5 (2.8%)RR 1.63Not significant
Early neurologic worsening (≥4-point NIHSS increase within 24h)Adverse27 (15.5%)44 (24.7%)RR 1.59Significant
Death from any cause within 90 daysAdverse71/171 (41.5%)68/177 (38.4%)RR 0.91Not significant
Arterial access-site occlusionAdverseN/A3 (1.7%)
Arterial access-site hematomaAdverseN/A1 (0.6%)
Arterial access-site infectionAdverseN/A1 (0.6%)
Vascular dissectionAdverseN/A10 (5.6%)
Vascular perforationAdverseN/A7 (3.9%)
Intraprocedural vasospasmAdverseN/A11 (6.2%)
Any procedural complicationAdverseN/A33 (18.5%)

Subgroup Analysis

Treatment effect consistent across prespecified subgroups: age (<70 vs ≥70: gOR 1.66 vs 1.36), NIHSS (<20 vs ≥20: gOR 1.53 vs 1.52), occlusion location (ICA vs MCA: gOR 1.31 vs 1.68), time to randomization (<12h vs ≥12h: gOR 1.48 vs 1.58; <6h vs ≥6h: gOR 1.63 vs 1.49), ischemic core volume (<70mL vs ≥70mL: gOR 1.39 vs 1.62; <100mL vs ≥100mL: gOR 1.57 vs 1.55; <150mL vs ≥150mL: gOR 1.51 vs 1.73), ASPECTS (0-2: gOR 1.40; 3-5: gOR 1.61; 6-10: gOR 1.24), mismatch ≥1.8/≥15mL (yes vs no: gOR 1.36 vs 1.83), affected hemisphere (left vs right: gOR 1.42 vs 1.60), geographic region (US vs non-US: gOR 1.63 vs 1.13). Benefit persisted in patients with ASPECTS ≤5 AND core ≥70mL (gOR 1.58), core >100mL, and core >150mL, though functional independence rates were lower with larger cores. Trial not powered for subgroup comparisons.

Criticisms

  • Trial stopped early for efficacy, which may cause treatment effect to be overestimated
  • Smaller sample size than anticipated (352 vs planned 560), underpowered for subgroup analyses
  • Open-label treatment (though outcome assessment was blinded)
  • Only ~20% of patients received IV thrombolytics before randomization
  • Some patients enrolled with low ASPECTS had smaller ischemic-core volumes than intended (11 patients with ASPECTS >5 and core <50mL per core lab)
  • In 48 of 352 patients, planimetric measurement of CT hypodensity substituted for prespecified perfusion method
  • Early neurologic worsening more common with thrombectomy (24.7% vs 15.5%), possibly related to reperfusion edema
  • 18.5% procedural complication rate (dissection 5.6%, perforation 3.9%)
  • No adjustment for multiplicity in secondary outcome comparisons
  • Limited geographic diversity (primarily Western countries)
  • General anesthesia used in ~60% of thrombectomy procedures, which may affect outcomes

Funding

Investigator-initiated grant from Stryker Neurovascular to University Hospitals Cleveland Medical Center and University of Texas McGovern Medical School. Stryker did not provide trial equipment, had no role in design/execution/analysis/writing, and use of Stryker products was not mandated.

Based on: SELECT2 (New England Journal of Medicine, 2023)

Authors: Amrou Sarraj, Ameer E Hassan, Michael G Abraham, ..., Gregory W Albers

Citation: Sarraj A, Hassan AE, Abraham MG, et al. Trial of Endovascular Thrombectomy for Large Ischemic Strokes. N Engl J Med. 2023;388(14):1259-1271.

Content summarized and formatted by NeuroTrials.ai.