Thrombectomy for Large Core Stroke: What the 2023 Trials Changed

The short answer

For years, patients with a large area of already-dead brain tissue after a stroke were usually excluded from mechanical thrombectomy, the procedure that pulls a clot out of a blocked artery. Two randomized trials published in early 2023, SELECT2 and ANGEL-ASPECT, changed that. Both showed that opening the artery in these large-core strokes shifted 90-day disability outcomes in a favorable direction compared with medical care alone, and neither raised death rates. The catch worth holding onto: most treated patients still ended up with meaningful disability, and the benefit is a shift in odds, not a cure.

This is educational content and not medical advice.

Why large cores were the last frontier

Thrombectomy became standard care around 2015 for strokes caused by a large-vessel occlusion, but the pivotal trials of that era mostly enrolled patients with a small infarct core, meaning a small volume of brain that imaging suggested was already irreversibly damaged. The reasoning was mechanistic: if most of the territory is dead, reopening the vessel might reperfuse tissue that cannot recover, and could instead bleed into the damaged area. Guidelines reflected that caution, and patients with low ASPECTS scores were frequently turned away.

ASPECTS (the Alberta Stroke Program Early CT Score) is a 10-point scale where lower numbers mean more visible early damage on a CT scan. A score of 3 to 5 signals a large core. The open question by 2022 was whether these patients were being denied a procedure that could still help, and the only honest way to answer it was a randomized trial.

What SELECT2 tested

SELECT2 was an international, open-label trial that randomized 352 patients with occlusion of the internal carotid artery or the first segment of the middle cerebral artery, plus a large core defined as an ASPECTS of 3 to 5 or an ischemic core of at least 50 mL on perfusion imaging. Patients could be enrolled up to 24 hours from when they were last known well. Half received thrombectomy plus medical care, half received medical care alone.

The primary outcome was the modified Rankin scale (mRS) at 90 days, a 0-to-6 measure of disability where 0 is no symptoms and 6 is death. Rather than a single cutoff, the analysis looked at the whole distribution. The generalized odds ratio for a shift toward better outcomes was 1.51 (95% confidence interval 1.20 to 1.89), reported in the New England Journal of Medicine. Functional independence, an mRS of 0 to 2, reached 20 percent in the thrombectomy group versus 7 percent with medical care. Mortality was similar, roughly 38 versus 42 percent, and symptomatic intracranial hemorrhage was low in both arms. The trial was stopped early for efficacy.

What ANGEL-ASPECT added

ANGEL-ASPECT, conducted across 46 centers in China, randomized 456 patients with anterior-circulation large-vessel occlusion and a large infarct, again treating within 24 hours. Its primary outcome was also a shift in the 90-day mRS distribution, with a generalized odds ratio of 1.37 (95 percent CI 1.11 to 1.69). The proportion of patients who could at least walk unassisted, an mRS of 0 to 3, was 47.0 percent with thrombectomy versus 33.3 percent with medical management. Like SELECT2, it was halted at a planned interim analysis after crossing its efficacy boundary.

The safety picture in ANGEL-ASPECT is where a careful reader should slow down. Symptomatic intracranial hemorrhage within 48 hours occurred in 6.1 percent of the thrombectomy group versus 2.7 percent of the medical group. That difference did not reach statistical significance in this trial, but the direction is real and consistent with the biology of reperfusing damaged tissue. Ninety-day mortality was essentially equal, about 21.7 versus 20.0 percent.

How to appraise these results

A few habits help when reading trials like these.

Read the outcome that was actually measured

Both trials used a shift analysis across the full mRS distribution, not a single "success or failure" line. A generalized odds ratio near 1.4 to 1.5 means the odds of landing in a better disability category improve, but it does not translate into most patients walking away well. In SELECT2, four out of five treated patients still had an mRS above 2 at 90 days. The honest framing is a better chance at a better outcome, distributed across a spectrum.

Weigh benefit against the hemorrhage signal

Reopening an artery into a large, already-injured territory carries a real bleeding risk. The trials suggest the net effect still favors treatment because functional gains outweighed bleeding-related harm, and mortality did not rise. But an individual patient's balance depends on core size, age, and how the vessel and tissue actually look on imaging, which is exactly why these decisions are made case by case rather than by a single score.

Notice what early stopping does to precision

Both trials stopped early for benefit. Stopping at an interim analysis, once a boundary is crossed, tends to produce a larger observed effect than a trial run to completion would, and it limits data on rarer harms. That does not invalidate the findings; convergent results across multiple trials, including RESCUE-Japan LIMIT before them and later trials such as TENSION and LASTE, strengthen confidence. It is a reason to read the effect size as directionally solid but not a fixed number.

Ask about generalizability

SELECT2 was international and ANGEL-ASPECT was conducted entirely in China, with differences in imaging selection, health systems, and populations. That the two, along with related trials, pointed the same way is reassuring. It also means the exact numbers travel less well than the overall conclusion that a large core, on its own, is no longer an automatic reason to withhold thrombectomy.

What changed in practice

The practical shift is a widening of eligibility. Selection now leans on the details of each case rather than a hard ASPECTS cutoff, and stroke systems have updated pathways so that patients who would previously have been excluded are at least considered. The evidence supports offering the procedure to carefully selected large-core patients, while being candid that the most common result is still meaningful disability, improved but not erased.