Selecting Stroke Patients for Late Thrombectomy in Real World

March 22, 2018

Two new clinical trials — DAWN and DEFUSE 3 — showing large benefits of removing the clot with endovascular intervention in selected patients with ischemic stroke presenting after 6 hours look set to revolutionize the field, substantially expanding the pool of patients eligible for thrombectomy and allowing many more patients experiencing large strokes to achieve good outcomes.

But many logistical challenges need to be addressed for this to happen routinely. Medscape Medical News asked several key opinion leaders on their thoughts about how these latest data will become translated into clinical practice. 

Both the DAWN and the DEFUSE 3 trial showed similar results, with patients receiving endovascular treatment three times more likely to achieve a good functional outcome than controls.   

But the key to achieving these good outcomes in the later-presenting patients is by careful patient selection, which was achieved in the trials by sophisticated imaging and computer software programs. The issue now is how to translate these trials into the real world.

Until recently, endovascular removal of the clot in stroke patients was recommended only for those with a large-vessel occlusion in the anterior circulation in whom the procedure could be performed within 6 hours of symptom onset. The DAWN and DEFUSE 3 trials now extend this up to 24 hours, but only in certain patients with salvageable brain tissue.

The main questions appear to be exactly who these patients are and how best to identify them in clinical practice.

"We have known that some patients have salvageable brain tissue for quite some time, but we didn't know that intervening in these patients would be so beneficial," Alejandro Rabinstein, MD, Mayo Clinic, Rochester, Minnesota, vice chair of the new acute stroke guidelines committee, commented. "The extent of the benefit in these two trials was very surprising. I thought they would be positive, but no one expected the benefit to be so large. They both showed a rate of functional independence of around 45%. That's amazing for patients with a large stroke presenting after 6 hours."

But the key behind these results is the selection of the right patients to treat, and how they will be identified, he said. "This data is very exciting," Rabinstein said. "It will mean great benefits for patients, but big questions remain on how it is going to be delivered."

Jeffrey Saver, MD, University of California Los Angeles, senior author of the DAWN study, noted, "We expect these results will have substantial effect on how we think about stroke and how we triage and image stroke."

He estimates that about 10% of stroke patients are eligible to receive thrombectomy within 6 hours, and with the new extension out to 24 hours this could increase to about 15%. "However, these patients account for about one third to half of the burden of stroke disability as these are the patients having very large disabling strokes."

Saver added, though, that it is still important to act quickly to evaluate patients because there are "the unfortunate ones" in whom brain tissue is dying at an extremely fast rate.

"But we found that many patients are much more fortunate — their hourglass does not empty so quickly — so we have an opportunity to treat many of these patients in time windows we never thought would be possible before, and it is vital to identify these patients quickly and efficiently," he said. "We have a lot of work getting there, though. The system will have to reorganize further."

Which Patients?

To identify exactly which patients are now eligible for late thrombectomy, the patients enrolled into DAWN and DEFUSE 3 have to be considered. Both trials aimed to include patients with a "target mismatch," that is, a large area of brain tissue threatened by the stroke (the penumbra) but with only a small area that has actually died (the infarct core) at the time of presentation, but they had different entry criteria.    

Gregory Albers, MD, Stanford University Medical Center, California, lead investigator of DEFUSE 3, explained that their trial had broader entry criteria than DAWN and so extends late treatment to a larger population.

"DAWN was conducted in what might be considered the ideal patients — those with a very small infarct core and a high stroke severity," he said. "DEFUSE-3 had less stringent criteria. For example, we included patients with a larger core, up to 70 milliliters, and we allowed prestroke status of mRS [modified Rankin Scale] 2 or less, whereas DAWN was 1 or less."

DAWN was also stricter in stroke severity scores, with a National Institutes of Health Stroke Scale (NIHSS) score of 10 or more, "whereas we stipulated 6 or more."  

Still, he said, "our results in a broader population were just as good as in the more restricted DAWN eligible population."

The other major difference between the two trials was cutoff time for treatment. DEFUSE 3 recruited patients 6 to 16 hours after symptom onset or "last known well," whereas DAWN extended this to 24 hours.    

Commenting on this anomaly, William Powers, MD, University of North Carolina at Chapel Hill, chair of the new acute stroke guidelines committee, said that if a patient presents under 16 hours, either study entry criteria are applicable.

"DEFUSE 3 had broader inclusion criteria so picks up extra patients," Powers said. "So, strictly speaking, if under 16 hours DEFUSE criteria will identify more eligible patients, but if between 16 and 24 hours it's best to use DAWN."

Saver said he thought some centers may extrapolate the broader DEFUSE 3 population out to 24 hours, although this would be off guideline.

Rabinstein made the point that despite different entry criteria, patients enrolled in the two studies were actually very similar. 

"When we write protocols on how we are going to incorporate these results into clinical practice, we have to look at the actual patents recruited into the studies rather than the entry criteria — they are not always the same," he said.

The typical patient in both these studies had large NIHSS scores of 16 to 18 and a very small core size of about 10 mL, he noted. In DEFUSE 3, the only trial that measured penumbra, patients had an average 100 mL of salvageable brain tissue.

"So when considering core size, although the studies accepted up to 70 mL, actually the vast majority of patients had much smaller cores," Rabinstein said. "So how do we deal with that? My personal interpretation is that I would feel comfortable treating a patient with a core size of up to about 30mls but maybe not one with a core size of 60mls. And if in doubt I would go back to basics: I would think about the age of patient, potential for recovery, and the location of the core – not all the brain is the same. Is the core in an area key to paralysis and aphasia, and if the rest of the penumbra was saved would this likely prevent a significant proportion of the deficit?"

Imaging Technology

There were also some differences in imaging requirements for the two studies.

"DAWN used slightly less complicated technology," Albers noted. "Patients could be included with MRI scans — they did not necessarily have to have perfusion imaging, and they used the clinical stroke score to assess salvageable tissue, whereas DEFUSE 3 stipulated perfusion imaging. So if the patient came in with an MRI, then they would have to have both the standard diffusion to identify the core and the perfusion sequence to identify salvageable tissue. So in DAWN there was potentially one less injection of contrast."

Both studies also used a commercial software program (RAPID, developed by Stanford University) to quantify mismatch.

The major issue following these trials is how to implement this imaging requirement in order to select the correct patients for treatment. Most believe the imaging should take place at the initial hospital receiving the patient.

"It would be best to do this perfusion imaging at the primary center," Albers said. "If not, every patient within the 6- to 24-hour time window who might fit these criteria would have to be transferred. But many will not fit."

Only about 50% of patients screened for DEFUSE 3 met the entry criteria on imaging, he noted. "By doing the imaging at the primary stroke center, we would be saving a lot of money and effort transferring the patients who do not have the correct mismatch."

Saver believes that it is not so much the imaging procedure that will limit the dissemination of extended treatment but rather the process changes involved.

"It is having systems of care available to enable perfusion imaging in hospitals which have not previously offered this, and having someone to read the image in real time," he said. "There's going to be some process change needed. It might not reach small hospitals, but all secondary and tertiary hospitals should able to manage that quite easily."

Powers pointed out that telemetry systems could help with this. "It should not be necessary for someone at every individual frontline hospital to interpret these perfusion studies. They only need to perform the scan, then the results can be uploaded into electronic systems where an expert at the more specialist hospital can read the results and advise on what to do."

Albers said that was already happening in some places. "At Stanford, the outlying hospital does the scan and then I get an email on my phone in real time and if this looks like a good patient for endovascular therapy, then we will immediately send the helicopter for them. Efficient systems like this need to be rolled out."

Peter Panagos, MD, professor of emergency medicine at Washington University, St Louis, Missouri, who was not involved in the studies, said these results offer both a "tremendous" opportunity and challenges to implementation. 

"As of today, screening these patients is not something we are set up to do, so we are going to have to have innovative collaborations locally to share our expertise to design systems for this to happen," Panagos said. "We don't want every patient going to the large endovascular center — that will overburden those centers. We also want patients in smaller hospitals to get best care — they shouldn't be penalized for having a stroke in a local hospital."

Most hospital have CT scanners, and most can perform CT with contrast, "so the technology exists," he added. "The additional imaging selection using perfusion studies may be available but will not be used regularly in most hospitals, so that will be a steep learning curve. I think the science we heard today will put a lot of pressure on community hospitals to get on board with this. Ideally we want perfusion imaging done at the local hospital."

American Heart Association (AHA)/American Stroke Association (ASA) spokesperson Ralph Sacco, MD, Miller School of Medicine, Miami, Florida, also believes more primary stroke centers will embrace the extra imaging required, and many will morph into the new category of "thrombectomy ready" hospitals, recently announced by the AHA/ASA.

"The imaging is not that complicated and difficult to do, and it will become more widely available, so in some areas where there are not many endovascular centers, doing it at primary stroke centers will help with triage. In other areas where there are more endovascular centers this may not be necessary as emergency services are taking patients straight there. So it depends on the community what will happen.

 With these new results from DAWN and DEFUSE, the number of patients eligible for thrombectomy will increase, he added.

"That's why AHA/ASA has established a new 'thrombectomy ready' hospital classification," Sacco said. "These are intermediate between primary and comprehensive centers. They are capable of performing thrombectomy, but don't have to have neurosurgical backup. This is an opportunity for hospitals who are not ready to become comprehensive centers but have more capabilities than a primary. This is the next step. They get certification, and more patients get treated appropriately."

Commercial Pressures   

Rabinstein notes that some European countries are better set up to implement these findings. "Germany has great regionalization of stroke care, where the large endovascular center is linked up with all the smaller primary stroke centers in the region.'

 He says this is more difficult to achieve in the United States because of commercial pressures. 

"In the US, larger institutions are competing with each other and the smaller hospitals are considering getting their own technology and becoming endovascular capable so as not to lose business. It may be good to have more centers capable of performing endovascular treatment, but only if they can do it properly." 

Maintaining quality requires case volume and experienced operators, he noted. It may be better for the patient to be transferred to a nearby larger hospital with more experienced interventionalists. 

"But this may not happen, not because of scientific reasons or public health reasons, but for commercial reasons," Rabinstein said. "These interventions are well paid, and medium-sized centers will not want to lose their stroke patients to larger centers.

"We really need a national program to make a joined-up system of care," he added. "But I don't think anybody truly knows how many more patients there will now be eligible for endovascular therapy, so it makes planning very difficult." 

Rabinstein suggests that not all primary stroke centers need to have perfusion imaging and that patients can be triaged by using clinical skills.

"While the ideal way of defining the core size is to use MRI or CT perfusion imaging, I suggest that we don't necessarily need these sophisticated imaging techniques to identify suitable patents for late thrombectomy," he said. "We can use clinical expertise. Clinical symptoms show us someone is having a large stroke, so will have a large area of threatened brain tissue, and we can make a rough estimate of the core size from the normal CT scan which all stroke patients routinely have. 

"If we see a very large, dark area, then that is a large core and we know there is probably nothing that we can do," he added. "The larger the core, the easier it can be visualized on the initial CT scan, so we can make a good educated guess, and a CT angiogram, which visualizes the blood vessels, will give a lot more information."

"I personally don't think all small hospitals need perfusion imaging," Rabinstein said. "They just need a way of triaging these patients. If I have a video of a patient last seen well 8 hours ago and I can see the normal CT scan, which looks unremarkable — no evidence of a large infarction (core) — and the patient can't talk or is paralyzed on one side so has obviously had a large stroke, then I would advise that that patient is transferred to our center for possible intervention.  We would then do the perfusion imaging when they got here. I would think this is probably what is going to happen at a lot of smaller hospitals."

Many small hospitals are not going to want to purchase expensive CT perfusion machines and computer software to analyze the results, he noted. "If you are a small-volume hospital and would use it maybe once a month, it's probably not worth it."

He also pointed out that many large endovascular centers have CT perfusion capability but haven't yet purchased the quantification software, like the RAPID program. 

"We still use CT perfusion, and we interpret the results ourselves," he said. "It doesn't give us an exact computer readout but you need to estimate the core size, and anyone treating patients and doing CT perfusion regularly can tell the difference between a 20-mL and 80-mL core. And if I have a relatively young patient with a bad stroke, unless I see something terrible I'm probably going to intervene anyway up to 24 hours as there is good chance that it will be of some benefit."

Powers agrees. "We instituted a DAWN protocol at our institution without using the RAPID software. It's a little bit clutzy — it takes some operator interaction, and it's probably not quite as fast, but we do use CT perfusion imaging and plot out the volume of the perfusion deficit. It is possible to do."

Saver stressed that each individual institution will have to decide the right approach for them. "RAPID is a good software to analyze the results but others can be used too, and we will undoubtedly see new software programs developed. But it is the biology that has been established by these trials that is foremost.

"The important thing is to do something to identify these patients," he said. "Whether this is the full package or just a routine CT scan with ASPECTS score and looking at collaterals, there are lots of options and there won't be same answer for everyone." 

Rabinstein agrees. "The main message is that we can treat many more patients now and we need to reorganize our systems locally to get this done. It's a great time to be in stroke at the moment. Twenty years ago we couldn't really do anything. Now we can save many of the worst cases, and this has increased significantly with these two new trials."

The DEFUSE 3 trial was supported by grants from the National Institute of Neurological Disorders and Stroke. The RAPID software platform was provided to all sites by iSchemaView. Albers has equity in and consults for iSchemaView. The DAWN trial was supported by Stryker Neurovascular, the manufacturer of the Trevo stent retriever device. Saver has received consultancy fees from Stryker regarding trial design and conduct. He is also an employee of the University of California, which has patent rights in retrieval devices for stroke.

For more Medscape Neurology news, join us on Facebook and Twitter


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.