DAWN Trial: Extended Window Thrombectomy in Acute Stroke

The DAWN trial fundamentally changed how doctors treat stroke by proving that a clot-removal procedure called mechanical thrombectomy could help patients up to 24 hours after symptom onset, far beyond the previous six-hour cutoff. Published in the New England Journal of Medicine in 2018, the trial found that roughly half of patients who received thrombectomy in the 6-to-24-hour window achieved functional independence at 90 days, compared with just 13 percent of those given standard medical care alone. That finding reshaped emergency stroke guidelines worldwide and opened the door for thousands of patients who previously would have been told it was too late.

What the Trial Tested and Why It Mattered

Before DAWN, the medical consensus held that removing a blood clot from a major brain artery was only worthwhile in the first six hours after a stroke began. After that window closed, the thinking went, the brain tissue starved of blood would be irreversibly damaged, and the risks of an invasive procedure would outweigh any benefit. DAWN challenged that assumption by asking a different question: instead of focusing on the clock, what if you focused on how much brain tissue was still alive?

The trial enrolled patients who had a large-vessel occlusion, meaning a clot was blocking one of the major arteries feeding the brain, and who were last known to be well between 6 and 24 hours earlier. The critical twist was a selection method based on what researchers called “clinical-core mismatch.” Patients had to show a significant gap between the severity of their neurological symptoms and the size of the dead brain tissue visible on imaging. In plain terms, if someone’s deficits looked severe but the actual area of irreversible damage was still small, that meant a large portion of their brain was struggling but salvageable, making clot removal potentially transformative.

The study used the Trevo stent retriever to physically pull the clot out of the artery. In the broader population of patients treated with this device, successful restoration of blood flow occurred in about 80 percent of cases.¹ The Trevo device also showed an advantage over some alternatives in achieving full blood-flow restoration.²

The Results That Stopped the Trial Early

DAWN’s findings were so striking that the trial was stopped ahead of schedule on ethical grounds: it would have been wrong to keep assigning patients to standard medical therapy alone when the clot-removal group was doing dramatically better. At 90 days, about 49 percent of thrombectomy patients had achieved functional independence compared with 13 percent of those who received only standard care, a 33-percentage-point difference. The overall disability score also strongly favored the thrombectomy group.³

A result that large in a stroke trial is rare. For context, most successful stroke treatments show modest improvements in the range of a few percentage points. A 33-point gap in functional independence is the kind of effect that makes clinicians rethink an entire field’s assumptions.

Safety in the Extended Window

One of the biggest concerns about treating stroke patients later was the risk of bleeding in the brain after restoring blood flow. In DAWN, the rate of symptomatic intracranial hemorrhage was 6 percent in the thrombectomy group and 3 percent in the control group, a difference that was not statistically meaningful. Mortality at 90 days was virtually identical between the two groups, around 18 to 19 percent.⁴ That mortality rate reflects how serious large-vessel strokes are regardless of treatment, not an added danger from the procedure itself.

Subsequent analyses confirmed this safety profile held across different patient subgroups. Whether patients presented as “wake-up strokes” (discovered upon waking), witnessed-onset strokes, or unwitnessed strokes, the rates of functional independence and bleeding complications in the thrombectomy arm were statistically similar across all subtypes.⁵ The consistency across different presentation types mattered because wake-up strokes, where nobody knows exactly when the clot formed, represent a large share of real-world cases.

Does Waiting Longer Within the Window Make Things Worse?

You might assume that even within the 6-to-24-hour window, earlier treatment would still be better. In practice, the data from DAWN tell a more nuanced story. A secondary analysis of the trial found no meaningful relationship between how long a patient waited before being randomized and their odds of functional independence or survival, as long as they met the imaging criteria. Each additional hour of delay did not significantly change outcomes in the thrombectomy group.⁶

A meta-analysis pooling data from extended-window trials reinforced this pattern. Among patients who received thrombectomy, longer onset-to-treatment times were not linked to worse disability or lower rates of functional independence. But among patients in the control groups who received medical therapy only, outcomes deteriorated as more time passed. The net effect was that the treatment benefit actually appeared to grow at later time points, because untreated patients kept getting worse while treated patients held steady.⁷

This does not mean there is no urgency. Faster treatment is always preferable when possible, and the imaging-based selection process itself takes time. But the finding reassures clinicians that patients presenting at 18 or 22 hours, if they meet the mismatch criteria, can still benefit enormously.

DAWN and DEFUSE-3 Together

DAWN did not work alone in reshaping stroke care. A companion trial called DEFUSE-3, published around the same time, asked a related but slightly different question. While DAWN used clinical-core mismatch (the gap between symptom severity and infarct size), DEFUSE-3 used perfusion imaging to identify patients with salvageable tissue in a 6-to-16-hour window. DEFUSE-3 enrolled patients with a proximal artery occlusion, a small initial infarct, and a ratio of at-risk tissue to dead tissue of at least 1.8.⁸ It too was stopped early for overwhelming benefit.

Together, the two trials established that the biology of each patient’s stroke matters more than the clock. Some people’s brain tissue dies quickly; others maintain viable tissue for many hours thanks to robust collateral blood flow. By identifying the patients whose brains are still fighting, doctors can intervene even when a purely time-based approach would have written them off.

A real-world comparison of patients who met DAWN or DEFUSE-3 criteria against those who did not found similar rates of symptomatic bleeding between the two groups. Patients who fell outside both trials’ strict criteria and still received thrombectomy did not show significantly different mortality or functional outcomes compared with those who met the criteria, though the evidence here is less certain.⁹

How Guidelines Changed

The strength of DAWN and DEFUSE-3’s results translated quickly into clinical practice. Professional societies incorporated the findings into updated guidelines, giving the extended-window thrombectomy their highest recommendation (Class 1A) for appropriately selected patients.¹⁰ That classification signals the strongest possible level of evidence and consensus. For stroke neurologists, the message was clear: the six-hour cutoff was no longer a hard wall.

The guideline changes triggered a cascade of logistical changes. Hospitals needed to ensure round-the-clock access to advanced brain imaging capable of identifying salvageable tissue. Emergency medical systems had to rethink transport protocols. Rural and community hospitals that lacked neurointerventional capabilities had to develop pathways for transferring late-presenting stroke patients to comprehensive stroke centers.

Who Actually Qualifies in the Real World

One of the most important and underappreciated realities of DAWN is that its strict selection criteria mean only a small fraction of stroke patients are eligible. In one stroke-center analysis, just 2.5 percent of acute ischemic stroke patients met the DAWN criteria, and about 5 percent met DEFUSE-3 criteria. When more liberal selection rules were applied, eligibility rose to around 11 percent.¹¹

A larger multicenter analysis found that only a minority of patients treated in routine practice met all inclusion criteria from any of the major thrombectomy trials, ranging from 3 percent for DEFUSE-3 to about 35 percent for the more permissive MR CLEAN trial criteria. Patients who did meet trial criteria had higher rates of good outcomes, as expected, but even those who fell outside the criteria still achieved meaningful rates of functional recovery.¹² This gap between trial eligibility and real-world practice remains one of the most active debates in stroke medicine: should doctors stick to strict criteria, or expand access to patients who look similar but don’t check every box?

The Transfer Problem

Most hospitals are not equipped to perform mechanical thrombectomy. The procedure requires a neurointerventionalist, a specialized angiography suite, and support staff, resources concentrated at comprehensive stroke centers. For patients who first arrive at a smaller community hospital, the question becomes whether the benefit of late-window thrombectomy justifies the added time and risk of a transfer.

The evidence strongly supports transferring eligible patients. An analysis of DAWN data on transferred versus direct-arrival patients found that late-window thrombectomy remained associated with substantial clinical benefit even when patients had to be moved from one hospital to another.¹³ The longer time window helps here: unlike the early window where every minute of transfer delay costs brain tissue, the extended window gives more room for the logistics of inter-hospital transport.

Still, the transfer process introduces its own challenges. Identifying the right patients at community hospitals requires CT angiography at minimum, and ideally CT perfusion imaging. In rural areas with limited imaging capabilities and long transport times, selecting appropriate candidates without overburdening the stroke system or initiating futile transfers is a practical challenge.¹⁴

Does Perfusion Imaging Matter for Selection?

DAWN used a specific kind of automated imaging analysis to identify eligible patients. But in everyday practice, not every hospital has access to CT perfusion software. This has led to a practical question: can you select late-window patients using simpler, more widely available imaging?

The evidence here is encouraging for hospitals with limited resources. A systematic review and meta-analysis found that functional independence rates after late-window thrombectomy were not significantly different between patients selected with CT perfusion and those selected with noncontrast CT alone. However, patients selected using perfusion imaging did have lower mortality rates.¹⁵ A separate study similarly found no difference in 90-day functional independence between patients selected by noncontrast CT and those selected by perfusion or diffusion-weighted imaging, after adjusting for confounders.¹⁶

The practical takeaway is that while perfusion imaging adds useful information and may identify the safest candidates, a lack of perfusion capability should not automatically disqualify a patient from being considered for transfer and treatment. This is especially relevant in low-resource settings where the choice is often between simpler imaging or no late-window treatment at all.

Cost-Effectiveness

Thrombectomy is an expensive procedure, requiring specialized staff, equipment, and often an ICU stay afterward. But when you factor in the alternative, which for many patients is severe long-term disability requiring years of nursing care, the economics shift decisively. An analysis based on DAWN’s results estimated that late-window thrombectomy cost roughly $662 per quality-adjusted life year gained, an extraordinarily favorable ratio by any country’s health-economic standards. Using DEFUSE-3 data, the figure was higher at around $13,877 per quality-adjusted life year, still well under the $100,000 threshold typically considered cost-effective in the United States. In both cases, the probability of thrombectomy being cost-effective exceeded 99.9 percent.¹⁷

A UK-focused analysis found similar results, with the cost per quality-adjusted life year remaining favorable whether the procedure was performed at 12, 16, or 24 hours after onset.¹⁸ Real-world data from a prospective cohort confirmed that the cost-effectiveness seen in the trials translated to practice, with late-window thrombectomy for DAWN-eligible patients yielding an incremental cost-effectiveness ratio of about $34,416 per quality-adjusted life year.¹⁹

The reason the economics work so strongly in favor of treatment is that avoiding severe disability saves enormous downstream costs. A patient who walks out of the hospital functionally independent does not need round-the-clock care, does not need years in a skilled nursing facility, and may return to work. The upfront procedural cost is significant, but the lifetime savings from avoiding dependency dwarf it.

Expanding Beyond DAWN’s Boundaries

DAWN enrolled patients with relatively small infarct cores, meaning the amount of already-dead brain tissue had to be limited. But many stroke patients present with larger areas of damage. A natural question arose after DAWN: could thrombectomy help patients with bigger strokes too?

Several recent trials have addressed this. The SELECT-2, ANGEL-ASPECTS, and RESCUE-JAPAN LIMIT trials collectively showed that thrombectomy performed within 24 hours in patients with large ischemic strokes, identified by imaging, was safe and associated with better functional outcomes than medical therapy alone. The treatment benefit persisted across all subgroups.²⁰ This represents a significant extension of DAWN’s legacy: not only was the time window pushed outward, but the definition of who qualifies is broadening to include patients DAWN would have excluded for having too much damage.

The Consent Dilemma in Acute Stroke Research

One aspect of DAWN that generated discussion beyond its clinical findings was how consent was handled. Stroke patients are, by definition, experiencing a brain emergency. Many are unable to speak, understand language, or make decisions. Their family members may not be reachable in the narrow window when treatment decisions must be made. This creates a genuine ethical tension between the need for informed consent and the need to study treatments that can only be tested on people in the middle of a medical crisis.

U.S. regulations allow for an exception from informed consent requirements in emergency research under specific conditions, including community consultation and independent review. Stroke researchers have long argued that this exception is sometimes the only way to conduct rigorous trials in an emergency setting where delays for consent could make enrollment impossible or even harm the patient.²¹ DAWN itself navigated this challenge by requiring consent from patients or surrogates, but the broader question of how to ethically study time-sensitive emergency procedures remains relevant as the field continues to push into new territory with larger strokes and more complex patient populations.