New England Journal of Medicine FEBRUARY 22, 2018

Thrombectomy for Stroke at 6 to 16 Hours with Selection by Perfusion Imaging

Gregory W. Albers, Michael P. Marks, Stephanie Kemp, et al.

Bottom Line

In patients with acute ischemic stroke caused by large vessel occlusion occurring 6 to 16 hours after being last known well, endovascular thrombectomy guided by perfusion imaging criteria for salvageable tissue significantly improved functional outcomes at 90 days compared to standard medical therapy alone.

Key Findings

1. The primary outcome of a favorable shift in the distribution of functional disability on the modified Rankin scale (mRS) at 90 days favored the endovascular group with an odds ratio of 2.77 (95% CI 1.63-4.70; P<0.001).

2. Functional independence (mRS score 0-2) at 90 days was achieved by 45% of patients in the endovascular group compared to 17% in the medical therapy group (P<0.001).

3. The 90-day mortality rate was lower in the endovascular group (14%) compared to the medical therapy group (26%; P=0.05).

4. Rates of symptomatic intracranial hemorrhage were not significantly different between the endovascular (7%) and medical therapy groups (4%; P=0.75).

Study Design

Design

RCT

Open-Label

Sample

182

Patients

Duration

90 days

Median

Setting

Multicenter, US

Population Patients with acute ischemic stroke due to occlusion of the internal carotid or proximal middle cerebral artery, presenting 6 to 16 hours after being last known well, with evidence of salvageable tissue on perfusion imaging.

Intervention Endovascular thrombectomy plus standard medical therapy

Comparator Standard medical therapy alone

Outcome Distribution of scores on the modified Rankin scale (mRS) at 90 days

Study Limitations

• The trial was terminated early based on a planned interim analysis, which may lead to an overestimation of the treatment effect size.

• The study size was relatively small (N=182), which limits the power to detect differences in less common adverse events.

• The generalizability of the findings is dependent on the availability of advanced perfusion imaging software and experienced endovascular teams.

• As an open-label trial, there is a theoretical potential for observer bias in the management of the control group, although the primary outcome was assessed in a blinded fashion.

Clinical Significance

The DEFUSE-3 trial provided pivotal evidence establishing Class I, Level A evidence for extending the window for mechanical thrombectomy in ischemic stroke to 16 hours, shifting clinical practice from strict time-based eligibility to personalized, physiology-based selection utilizing perfusion imaging to identify penumbral tissue.

Historical Context

Prior to DEFUSE-3 and the parallel DAWN trial, mechanical thrombectomy was largely restricted to a 6-hour window post-symptom onset. DEFUSE-3 represents the culmination of two decades of research into stroke pathophysiology, specifically the concept of the 'ischemic penumbra', proving that salvageable brain tissue can exist well beyond traditional time windows if selected by advanced neuroimaging.

Guided Discussion

High-yield insights from every perspective

Med Student

Medical Student

In the context of the DEFUSE-3 trial, explain the physiological concept of the 'ischemic penumbra' and why it is more critical than 'time from last known well' when deciding on thrombectomy in the 6 to 16-hour window.

Key Response

Traditional stroke management followed a strict time-is-brain model. However, DEFUSE-3 emphasizes 'physiology-is-brain' by identifying the ischemic penumbra—tissue that is hypoperfused but not yet infarcted. Using perfusion imaging, clinicians can identify patients with robust collateral circulation who still have salvageable tissue (mismatch) even after the standard 6-hour window has closed, whereas others may have completed their infarct much earlier.

Resident

A patient presents 10 hours after being last known well with an NIHSS of 15 and an occlusion of the M1 segment of the middle cerebral artery. According to the DEFUSE-3 selection criteria, what specific perfusion imaging parameters are required to proceed with endovascular thrombectomy?

Key Response

DEFUSE-3 utilized automated software (RAPID) to identify patients with an initial infarct core volume (Tmax > 10s or CBF < 30%) of less than 70 ml, an ischemic core-to-penumbra ratio of 1.8 or greater, and an absolute mismatch volume of at least 15 ml. Meeting these criteria demonstrated a significant functional benefit (mRS 0-2) compared to medical therapy.

Fellow

Contrast the patient selection strategy of DEFUSE-3 with that of the DAWN trial. How does the use of 'perfusion-core mismatch' in DEFUSE-3 differ from the 'clinical-core mismatch' used in DAWN, and how might this influence the number of patients eligible for treatment?

Key Response

DAWN relied on a mismatch between clinical severity (NIHSS) and infarct core size (e.g., high NIHSS but small core), which can capture patients up to 24 hours. DEFUSE-3 used an imaging-only mismatch (perfusion defect vs. core size). DEFUSE-3 criteria are often considered broader because they don't require a specific high NIHSS floor for smaller cores, but they are limited to a 16-hour window and require perfusion software availability, unlike the simpler core-volume limits in some DAWN subgroups.

Attending

DEFUSE-3 demonstrated a Number Needed to Treat (NNT) of 2 for functional independence (mRS 0-2). Given this profound effect size, how should the 'hub-and-spoke' stroke system of care be redesigned for community hospitals that lack advanced perfusion imaging software?

Key Response

The high efficacy in the extended window necessitates a shift in triage. Spoke hospitals must either implement automated perfusion software or have a lower threshold for transferring LVO suspects to 'thrombectomy-capable' or 'comprehensive' centers. The 'drip-and-ship' model must now incorporate the reality that a patient is not automatically 'out of window' at 6 hours, potentially requiring rapid transfer for advanced imaging that isn't available locally.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD

The DEFUSE-3 trial was terminated early for efficacy following the release of the DAWN trial results. What are the potential statistical implications of early termination on the reported treatment effect size, and how does this affect the precision of the safety outcome estimates such as symptomatic intracranial hemorrhage?

Key Response

Early termination for efficacy can lead to 'truncation bias,' where the treatment effect (odds ratio for functional improvement) is potentially overestimated because the trial stopped at a high point in the data oscillation. Furthermore, smaller sample sizes in truncated trials reduce the power to detect rare but critical safety events, such as symptomatic intracranial hemorrhage (sICH) or parenchymal hematoma, leading to wider confidence intervals for these complications.

Journal Editor

Considering the reliance on the RAPID software for inclusion in DEFUSE-3, how do you evaluate the internal validity versus the generalizability of the findings, especially regarding centers using alternative post-processing perfusion platforms?

Key Response

A major concern is 'software-dependency.' Different automated perfusion platforms (e.g., Olea, Viz.ai, or Vitrea) use different algorithms to calculate Tmax and CBF thresholds. If a center uses a different software that overestimates or underestimates the core compared to RAPID, the trial's results may not apply perfectly. As an editor, one would flag the potential for selection bias if the technology is proprietary and not standardized across the industry.

Guideline Committee

Based on the DEFUSE-3 results, how should the Level of Evidence and Class of Recommendation be updated for endovascular thrombectomy in the 6-16 hour window, and how does this specifically change the 2018/2019 AHA/ASA Stroke Guidelines?

Key Response

DEFUSE-3, alongside DAWN, provided the definitive evidence to move extended-window thrombectomy from an 'uncertain' status to a Class I, Level of Evidence A recommendation. The 2019 AHA/ASA focused update specifically incorporates these trials, stating that in patients with LVO in the carotid or M1 segments who meet DEFUSE-3 or DAWN criteria, mechanical thrombectomy is recommended in the 6-16 hour (for DEFUSE-3) or 6-24 hour (for DAWN) window.

Clinical Landscape

Noteworthy Related Trials

2015

MR CLEAN Trial

n = 500 · NEJM

Tested

Endovascular thrombectomy

Population

Patients with acute ischemic stroke due to large vessel occlusion

Comparator

Standard medical care including IV alteplase

Endpoint

Functional independence at 90 days (mRS 0-2)

Key result: Thrombectomy combined with standard care was significantly more effective than standard care alone for patients with large vessel occlusions.

2015

ESCAPE Trial

n = 316 · NEJM

Tested

Endovascular thrombectomy with imaging-based patient selection

Population

Patients with large-vessel occlusion stroke and good collateral circulation

Comparator

Standard medical care

Endpoint

Functional independence at 90 days (mRS 0-2)

Key result: Thrombectomy provided a significant functional benefit and lower mortality compared to standard care in patients with good collateral circulation.

2018

DAWN Trial

n = 206 · NEJM

Tested

Endovascular thrombectomy plus standard care

Population

Stroke patients 6-24 hours after last known well with clinical-core mismatch

Comparator

Standard medical care alone

Endpoint

Functional outcome measured by mRS at 90 days

Key result: Thrombectomy was associated with better functional outcomes at 90 days compared to medical management alone in a late time window.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis