Episode Details
Back to EpisodesSGEM#524: High, High HOPE for Recovery – Alteplase between 4.5 and 24 hours
Published 4 weeks ago
Description
Date: October 6, 2026
Guest Skeptic: Dr. Ryan Stanton, creator of the EM podcast Front Line and the new President of the American College of Emergency Physicians (ACEP).
Reference: Zhou et al. HOPE investigators. Alteplase for Acute Ischemic Stroke at 4.5 to 24 Hours: The HOPE Randomized Clinical Trial. JAMA. 2025
Case: Mr. Wei Zhang is a 73-year-old retired teacher with hypertension and atrial fibrillation (not on anticoagulation), who is brought to the emergency department (ED) at 09:30 after being found by his wife with slurred speech and weakness of the right arm and leg. He was last known well the previous evening at 23:00, when he went to bed. On arrival, he is alert and oriented. The NIHSS score is 12 (moderate stroke; facial droop, dysarthria, right arm and leg weakness). Blood glucose is normal, CT head shows no hemorrhage, CT angiography shows no large-vessel occlusion, and the CT perfusion scan shows an ischemic core 30 mL, penumbra 60 mL (mismatch ratio = 2.0, mismatch volume = 30 mL). He is 10½ hours from last-known-well and not a candidate for thrombectomy .
Background: The use of thrombolytics for acute ischemic stroke (AIS) has been debated in emergency medicine for about three decades. It started with the 1995 NINDS trial of IV alteplase within three hours of symptom onset [1]. That trial became the foundation for thrombolysis in AIS and helped drive the concept that “time is brain.” Alteplase subsequently became incorporated into stroke systems of care and clinical practice guidelines.
That doesn’t mean everyone has agreed on the strength of the evidence. Emergency medicine has had a long and sometimes spirited debate about the benefits and harms of thrombolysis for stroke. Even the catchy phrase “time is brain” deserves some nuance [2]. Dr. Camilo Gómez, who coined the phrase, later argued that collateral circulation and other patient factors mean that ischemic injury does not progress at the same rate in everyone. In other words, time still matters, but the clock doesn’t tell us everything about the brain (Stroke Theory of Relativity).
Then came ECASS III in 2008, extending the alteplase window from three to 4.5 hours [3]. The trial reported improved functional outcomes but also more symptomatic intracranial hemorrhage (sICH), and guidelines subsequently adopted the expanded window. However, ECASS III has also been challenged. Brian Alper and colleagues reanalyzed the patient-level data and raised concerns about the original adjusted analysis and baseline imbalances [4]. Their reanalysis could not confirm a statistically significant efficacy benefit after adjustment, while the increased risk of sICH remained.
The next move was from a clock-based paradigm toward a tissue-based paradigm. WAKE-UP used MRI DWI-FLAIR mismatch to select patients with an unknown time of onset, while EXTEND used perfusion imaging to identify patients with potentially salvageable brain up to nine hours after onset or with wake-up stroke [5, 6]. Both suggested that selected patients might benefit beyond the traditional treatment window, although EXTEND was small, stopped early, and thrombolysis continued to carry an increased risk of sICH.
Things get even more interesting when we stretch the clock to 24 hours. TIMELESS tested tenecteplase (TNK) from 4.5 to 24 hours in perfusion-selected patients, most of whom subsequently underwent thrombectomy, and did not demonstrate improvement in its primary functional outcome [7]. Current guidelines therefore don’t simply say, see salvageable tissue, give thrombolytics. Patient selection, timing, imaging, and whether thrombectomy is planned all matter.
And this is where things get a little more skeptical. Garg and Brophy recently looked at six randomized trials of imaging-selected extended-window thrombolysis [8]. They found a high probability of improving the traditional dichotomized mRS outcome, but only a 1% probability of a clinically important improvement using a utility-weighted mRS. They