Mechanical thrombectomy is the most powerful intervention in acute stroke care, with a number needed to treat for reduced disability of roughly 2.6 in the pooled early-window trials.1 But the benefit is concentrated in patients who still have brain worth saving. The practical task at the bedside is not “is there a clot?” — it is matching occlusion site, time, and tissue status to the evidence. This page walks the three eligibility regimes most fellows are tested and paged on: the early window, the late window, and the expanding large-core frontier, plus the posterior circulation.

Early window: 0–6 hours, the 2015 cohort

Five randomized trials reported in 2015 — MR CLEAN, ESCAPE, REVASCAT, SWIFT PRIME, and EXTEND-IA — established thrombectomy for proximal anterior-circulation occlusion. Their individual patient data were pooled in the HERMES collaboration, which confirmed benefit across age, sex, occlusion site, and stroke severity.1 The composite early-window profile drawn from these trials is:

  • NIHSS ≥6 (clinically meaningful deficit).
  • ASPECTS ≥6 on non-contrast CT (limited established core).
  • Anterior large-vessel occlusion: ICA terminus, M1, or proximal M2.
  • Pre-stroke modified Rankin Scale (mRS) ≤1 (good baseline function).
  • Age ≥18.

In this window the imaging is deliberately simple: non-contrast CT plus CT angiography is enough to select most patients. Perfusion imaging is not required to act on a clear ASPECTS ≥6 anterior occlusion presenting early.

Bedside pearl

In the early window, do not wait for perfusion software to confirm what plain CT and CTA already show. A clear M1 occlusion with ASPECTS ≥6 at three hours is a transfer-and-treat decision, not an imaging puzzle.

Late window: 6–24 hours, selection by mismatch

Beyond six hours, time-based criteria give way to tissue-based ones. Two trials extended the window using mismatch — a small completed core with a large volume of threatened but not-yet-infarcted brain. DAWN selected patients 6–24 hours from last-known-well using a clinical-imaging mismatch: a substantial NIHSS deficit paired with a small core on CT perfusion or diffusion MRI.2 DEFUSE-3 selected patients 6–16 hours out using a perfusion-imaging mismatch — an initial infarct under 70 mL with a penumbra-to-core ratio of at least 1.8.3 Both showed a large treatment effect, and both depend on advanced imaging (CT perfusion or MRI) to identify the favorable mismatch profile.

The conceptual point worth internalizing: late-window selection identifies slow progressors — patients whose collaterals have kept the penumbra alive longer than average. It does not slow infarct growth in any individual patient. Earlier reperfusion remains better, so a 24-hour window is a safety net, not permission to move slowly.

Pitfall

Do not treat the late window as a license to delay. The patient in front of you is a snapshot, not a guarantee — a favorable mismatch can collapse within hours. Faster reperfusion yields better outcomes at every point on the curve.

Large-core expansion: ASPECTS 3–5

The historical ASPECTS ≥6 floor excluded patients with large established infarcts, on the assumption there was too little salvageable tissue. Several 2022–2023 trials challenged that — SELECT2, RESCUE-Japan LIMIT, ANGEL-ASPECT, and TENSION — and demonstrated benefit from thrombectomy in selected patients with large cores, including ASPECTS in the 3–5 range.4,5,6,7 SELECT2, for example, randomized patients with large ischemic strokes and found a shift toward better functional outcomes with thrombectomy plus medical care versus medical care alone.4

The clinical reading is nuanced. Benefit is real but the absolute outcomes remain modest, more patients survive with significant disability, and selection still matters — this is not a green light to treat every large core. Specific large-core thresholds, the role of perfusion versus plain ASPECTS, and age cutoffs vary across trials and across institutional protocols, so anchor decisions to your own center’s pathway and to current guidelines.

Posterior circulation: basilar occlusion

Basilar-artery occlusion sits apart from the anterior-circulation trials and was historically high-mortality and under-evidenced. Two randomized trials — BAOCHE and ATTENTION — now support thrombectomy for acute basilar occlusion in selected patients, with BAOCHE extending treatment out to 24 hours from onset.8,9 The posterior circulation tolerates a wider time window than the anterior circulation in part, but the same principle holds: confirm the occlusion, assess the extent of established brainstem and posterior injury, and weigh baseline function before proceeding.

Putting it together at the bedside

A workable mental algorithm: confirm a large-vessel occlusion on CTA; if within 6 hours with ASPECTS ≥6, proceed on CT/CTA alone; if 6–24 hours, obtain perfusion or MRI and look for mismatch (DAWN/DEFUSE-3 logic); if the core is large (ASPECTS 3–5), thrombectomy may still help selected patients, but counsel realistically; for basilar occlusion, apply BAOCHE/ATTENTION criteria with a wider window. Throughout, pre-stroke mRS, age, and goals of care frame every decision.

This page is for clinician education and does not constitute individualized treatment advice. Eligibility thresholds, imaging requirements, and large-core practice differ between guidelines and institutions and continue to evolve — always confirm against your current local protocol and the most recent guideline edition before acting on any individual patient.