Thrombectomy 24 to 72 hours after stroke improved outcomes in a China trial

Pulling a clot out of a blocked brain artery is now standard care for the most disabling kind of stroke — but only within a clock. Trials established the benefit first within six hours of onset, then out to 24 hours in carefully selected patients. What happens to someone who reaches hospital later than that, a day or two after their symptoms began, has been one of the open questions in stroke care [s1].

LATE-MT, published in the New England Journal of Medicine on 23 October, is a direct test of that question [s1]. It asked whether thrombectomy still helps patients who present 24 to 72 hours after a large-vessel stroke — provided imaging shows there is still brain tissue worth saving.

What the trial did

This was an open-label, randomised trial with blinded outcome assessment, run at comprehensive stroke centres in China [s1]. Patients had an acute ischaemic stroke caused by a large-vessel occlusion of the anterior circulation — the front part of the brain's blood supply — and, crucially, imaging that identified salvageable tissue in the window of 24 to 72 hours after they were last known to be well [s1].

They were assigned to one of two groups: endovascular therapy plus standard medical care, or standard medical care alone [s1]. The primary outcome was functional status at 90 days, measured on the modified Rankin scale — a 0-to-6 scale on which 0 means no symptoms, higher numbers mean greater disability, and 6 means death — analysed as the whole distribution of scores in each group rather than a single cut-point [s1].

A total of 336 patients were included in the intention-to-treat population, 168 assigned to each group [s1]. The median time from symptom onset, or from when the patient was last known to be well, to randomisation was 38.5 hours [s1] — well beyond the windows established by earlier thrombectomy trials. The study is registered as LATE-MT, NCT05326932, and is now completed [s1] [s2].

What happened

Endovascular therapy produced a more favourable distribution of modified Rankin scale scores at 90 days than standard medical care alone, with a common odds ratio for a better functional outcome of 1.74 (95% confidence interval, 1.19 to 2.55) [s1]. A common odds ratio above 1 means that, across the whole disability scale, patients who had the procedure tended to end up in a better category than those who did not — the benefit was a shift of the entire distribution, not a single headline percentage [s1].

The lower bound of that interval, 1.19, sits clearly above 1, so the result is statistically significant [s1]. But the interval is wide, reaching up to 2.55, which is what a trial of this size — a few hundred patients — can be expected to deliver [s1].

The harm side of the ledger

Opening a long-closed artery in already-injured brain carries a cost, and LATE-MT measured it plainly [s1]. Symptomatic intracranial haemorrhage — bleeding into the brain severe enough to cause symptoms — occurred in 4.8% of patients in the thrombectomy group and in none of the control group [s1]. The 90-day case fatality rate was 6.0% with thrombectomy and 1.8% with standard care alone [s1].

Those are small absolute numbers, but they move in the wrong direction, and both the bleeding and the deaths were concentrated in the group that had the procedure [s1]. The trial's own conclusion holds the two findings together: endovascular therapy resulted in better functional outcomes than standard care but led to a higher risk of intracranial haemorrhage and death [s1].

How far the result travels

Two features shape how widely LATE-MT can be read. First, it was conducted entirely in China, at comprehensive stroke centres with the imaging and interventional expertise to select patients by salvageable tissue [s1]. Whether the same selection — and the same result — can be reproduced in health systems with different imaging access is not something a single-country trial can answer [s1].

Second, the eligibility hinged on imaging, not on the clock alone [s1]. These were not simply "late" patients; they were late patients whose scans still showed brain worth saving [s1]. The trial does not say that everyone who arrives after 24 hours should have a thrombectomy — it says that, among a carefully imaged subset, the procedure shifted outcomes toward independence despite a real increase in bleeding and death [s1].

What to watch

LATE-MT extends a line of thrombectomy trials that have steadily pushed the treatable window outward, from six hours to 24 and now, in selected patients, to 72 [s1]. The immediate task for guideline writers is to weigh a distribution-wide functional benefit against a measurable excess of symptomatic haemorrhage and death — and to decide which imaging criteria reliably identify the patients in whom the trade favours treatment [s1]. Replication outside China, and a clearer read on which patients carry the highest bleeding risk, are the obvious next steps [s1].

This article describes trial results. It is not advice about any treatment, and nothing here should be used to start, stop or change care.