Endovascular thrombectomy for ischemic stroke beyond 6 hours from onset of symptoms

3-4 for functional independence at 90 days

Benefits in NNT

3-4 for functional independence at 90 days
28-33% for functional independence at 90 days

Harms in NNT

35 for symptomatic intracranial hemorrhage
2.8% for symptomatic intracranial hemorrhage
View As:


Nogueira RG, Jadhav AP, Haussen DC, Bonafe A, Budzik RF, Bhuva P, et al; DAWN Trial Investigators. Thrombectomy 6 to 24 Hours after Stroke with a Mismatch between Deficit and Infarct. N Engl J Med. 2018;378:11-21.

Study Population: 388 patients with large vessel (internal carotid artery or proximal middle cerebral artery) occlusion, mismatch between infarct size and either clinical deficit or ischemic penumbra, and presentation more than 6 hours (up to 16 or 24 hours) after onset of stroke symptoms

Efficacy Endpoints

Disability score at day 90; functional independence (defined as a score of 0-2 on the Modified Rankin Scale at day 90)

Harm Endpoints

Death (90-day mortality); symptomatic intracranial hemorrhage; serious adverse events


In ischemic stroke, there are two types of affected brain tissue. The center of the lesion is composed of dead or non-salvageable brain tissue. This core is surrounded by a region of ischemic tissue called the penumbra. The penumbra is the area that theoretically might be salvaged with reperfusion therapy.1,2

Endovascular thrombectomy has been shown to reverse the ischemia in this penumbra in certain stroke patients with large vessel occlusion if it is performed within 6 hours of onset of symptoms. However, as the time from symptom onset increases, the chance of an improved neurologic outcome declines.3 The time window for saving ischemic tissue may be different in different patients, suggesting that time may not be the only factor determines the likelihood of benefit from reperfusion. Some investigators believe that in a carefully selected group of patients a substantial amount of brain tissue remains salvageable beyond the 6-hour treatment window.

Recently, two randomized trials1,2 suggested that removing a large vessel clot can improve outcomes even 16-24 hours after the onset of symptoms. One study targeted patients with a clinical deficit that was disproportionately severe relative to the imaged infarct volume, which suggested that symptoms were due to a large area of hypoperfused but not yet infarcted tissue. The other study targeted patients with a large neuroimaging mismatch between ischemic tissue and infarcted tissue. These trials used CT perfusion imaging or MRI diffusion and perfusion scanning paired with automated image post-processing systems to map the affected area of the brain. Both trials were terminated early due to efficacy.

DEFUSE 31 was a multicenter, randomized, open-label trial that was conducted at 38 centers in the United States (Table 1). In this trial, thrombectomy plus medical therapy, as compared with medical therapy alone, was associated with a favorable shift in the distribution of functional outcomes on the Modified Rankin Scale (mRS) at 90 days (odds ratio: 2.77; 95%CI, 1.63-4.70, P<0.001), as well as a higher percentage of patients who were functionally independent, defined as a mRS score of 0-2 (45% vs. 17%, P<0.001, Absolute Risk Difference [ARD]: 28%, NNT:4).

The 90-day mortality rate was 14% in the thrombectomy group and 26% in the medical-therapy group (ARD:12%, Odds ratio:0.55; 95%CI, 0.30–1.02, p=0.05), and there was no statistically significant difference in the frequency of symptomatic intracranial hemorrhage (7% and 4%, respectively; odd ratio:1.47, 95%CI, 0.40–6.55; P=0.75) or of serious adverse events (43% and 53%, respectively; P=0.18) between the groups. Thrombectomy-related complications occurred in two patients (one vessel perforation resulting in subarachnoid hemorrhage and one device-related vasospasm).

DAWN2 was a multicenter, randomized, open-label trial with a Bayesian adaptive–enrichment design and with blinded assessment of end points conducted at 26 centers in the United States, Canada, Europe, and Australia (Table 1). This trial was sponsored by Stryker Neurovascular, which provided funding, thrombectomy devices, performed regulatory monitoring at each site, and provided central database maintenance.

The mean score on the Utility-Weighted Modified Rankin Scale (which weights each category according to patient perception of health-related quality of life) at 90 days was 5.5 in the thrombectomy group as compared with 3.4 in the control group (adjusted difference [Bayesian analysis], 2.0 points; 95% CI, 1.1-3.0). The rate of functional independence at 90 days was 49% in the thrombectomy group, as compared with 13% in the control group (ARD:33%; 95% CI, 24-44; NNT:3). The rate of symptomatic intracranial hemorrhage did not differ significantly between the two groups (6% in the thrombectomy and 3% in the control group, P=0.50), nor did 90-day mortality (19% and 18%, respectively; P=1.00).

The rates of safety end points and serious adverse events — including stroke-related death at 90 days, death from any cause at 90 days, and symptomatic intracerebral hemorrhage — did not differ significantly between the two treatment groups. The rate of neurologic deterioration was lower in the thrombectomy group than in the control group (14% vs. 26%; absolute difference, −12 percentage points; 95% CI, −23 to −1; P = 0.04). Seven patients in the thrombectomy group (7%) experienced procedure-related complications (intramural arterial dissection, arterial perforation, and access-site complications leading to intervention).

The results of these two trials indicate that in patients with good pre-stroke baseline functional status who had large vessel occlusion and a mismatch between infarct size and either clinical deficit or ischemic penumbra and an onset more than 6 hours (up to 16 or 24 hours) before presentation, neurologic outcomes at 90 days could be better with thrombectomy. While time is still the most important element in treatment success, these trials highlight the importance of adding tissue-based criteria in selecting patients for revascularization. The most important harm associated with this strategy is the increased risk of symptomatic intracranial hemorrhage. The combined number of patients with symptomatic intracranial hemorrhage in both trials was 19/306 (6.2%) in intervention groups vs. 10/296 (3.4%) in control groups, ARD: 2.8% (95% CI, 0.7-6.0%), resulting in a number-needed-to-harm (NNH) of 35 (95% CI, 15-149).


DAWN and DEFUSE 3 are important trials with significant clinical implications for patients, physicians, and hospitals. These trials employed a new approach to identify potentially salvageable brain tissue (penumbra) and to select stroke patients for revascularization.

These findings have potentially serious service implications for stroke systems of care. At the prehospital level, EMS providers would be required to screen suspected stroke patients for possible large vessel occlusion up to 24 hours from last-seen-well (LSW) time, and preferentially transport those meeting eligibility criteria to a center capable of rapid advanced imaging and endovascular therapy.

At the hospital level, stroke centers would need to perform CT perfusion or MRI rapidly in patients up to 24 hours after the onset of symptoms. This might require not just increased equipment, but also advanced software, increased staffing by CT and MRI technologists, and enhanced neuroradiology availability to rapidly interpret these images. It is not clear how many hospitals or even primary stroke centers could feasibly offer these services. In addition, hospitals without endovascular teams would need to put in place systems to rapidly evaluate and transport patients to endovascular-capable hospitals, as necessary.

Such a change in current clinical practice has been estimated to potentially increase the number of patients eligible for thrombectomy by only 2-5%. However, many patients would need to be screened for eligibility (including with advanced imaging). At many hospitals, the costs of maintaining the necessary 24/7 advanced imaging might not be feasible. The cost effectiveness of this approach, as well as scalability outside of academic medical centers, is thus not yet clear. In addition, recent publication of the use of intravenous thrombolysis in patients selected by similar criteria raises the concern for even more complex decision-making in the acute phase.4

It must also be emphasized that the criteria for enrollment in DAWN and DEFUSE 3 were relatively strict and were limited to patients with good baseline functional status, large vessel ischemic strokes limited to proximal carotid artery or middle cerebral artery, and large enough mismatches between the size of ischemic vs infarcted tissue. The enrolled patients had relatively small core infarctions. It is likely that most stroke patients would not meet these criteria.

A major limitation in generalizing the findings of these trials is the method of determining infarct size. These trials employed sophisticated software (RAPID, iSchemaView) with DWI or CTP imaging. It is not clear that centers without this software, or rapid MRI/CTP, could achieve similar results. The more widely available method of determination of perfusion mismatch (volumetric or qualitative) should be evaluated in identifying similar patients and producing similar outcomes.

While we have assigned a GREEN color recommendation, we caution the readers about patient selection. The inclusion criteria for this treatment do not apply to majority of stroke patients. In addition, the complexity of the technology and resource demands may limit which centers can safely and effectively employ this approach.


Shahriar Zehtabchi, MD; Joshua N. Goldstein, MD, PhD


September 25, 2018