A Review of Key Thrombectomy, ICH Trials From 2020

In this podcast, Neurology Learning Network's Stroke & Vascular Section Editor Amrou Sarraj, MD, discusses important thrombectomy and intracerebral hemorrhage trial data that surfaced in 2020, including the BASILAR trial, RESILIENT trial, and more. A full transcript is provided below.

Discover more insights from your peers in our Stroke & Vascular Excellence Forum.

Amrou Sarraj, MD, is an associate professor of neurology at McGovern Medical School at UTHealth in Houston, Texas.

Liked This Podcast? See Also:


  1. Liu X, Dai Q, Ye R, et al. Endovascular treatment versus standard medical treatment for vertebrobasilar artery occlusion (BEST): an open-label, randomised controlled trial. Lancet Neurol. 2019;19(2):115-122. doi:10.1016/S1474-4422(19)30395-3
  2. Llorente IL. American Skip study clinical trial details. American Heart Association. Accessed January 15, 2021. https://professional.heart.org/en/science-news/skip-study-clinical-trial-details
  3. Martins SO, Mont'Alverne F, Rebello LC, et al. Thrombectomy for stroke in the public health care system of Brazil. N Eng J Med. 2020;382:2316-2326. doi:10.1056/NEJMoa2000120
  4. Writing group for the BASILAR Group. Assessment of endovascular treatment for acute basilar artery occlusion via a nationwide prospective registry. JAMA Neurol. 2020;77(5):561-573. doi:10.1001/jamaneurol.2020.0156
  5. Yang P, Zhang Y, Zhang L, et al. Endovascular thrombectomy with or without intravenous alteplase in acute stroke. N Eng J Med. 2020;382:1981-1993. doi:10.1056/NEJMoa2001123


Dr Amrou Sarraj: Greetings. This is Dr Amrou Sarraj. I'm an associate professor of neurology and a vascular neurologist at UT McGovern in Houston.

In this podcast, we will do a year‑end review, a summary of the main clinical trials and major papers that were published within the last year or so, focusing on the different main categories of our daily stroke neurology practice.

We will go over the different recent trials and major papers that were published around thrombectomy with its different subcategories, and we'll conclude with the data of the management of hypertensive intracerebral hemorrhage. This will be an overall review of the data.

We'll go over the question of the trial or the paper, the main results, and the conclusion, and how this applies to the practice. There will be a summary document available with this podcast for people who would like to see the details of the numbers, with references to all of the papers and trials that I will be going over.

We all know that there has been a revolution update in the last few years on intravascular thrombectomy that initially showed efficacy and safety based on 7 randomized clinical trials in the early treatment window. That's defined 0 to 6 hours last known well to groin puncture.

Patients with anterior circulation large‑vessel occlusion with specific imaging profile, selected patients, I should say, those who had minimal ischemic changes on CT or profusion.

Then, the donor in DEFUSE 3 trial showed benefit beyond the 6 hours, up to 24 hours in, again, selected patients with anterior circulation occlusions in the internal carotid artery and the middle cerebral artery first segment MCA M1.

With this revolution, remains a few questions. Those trials that I just mentioned showed that intravascular thrombectomy has adjunctive benefit to best medical management, including IV tPA, when patients present in the first 4.5 hours from last known well.

The question now is reversed. Should we give IV tPA prior to thrombectomy if patients present within the first 4.5 hours? There are arguments and counter‑arguments to that.

There are safety concerns with giving IV thrombolysis prior to thrombectomy and increase the risk of symptomatic hemorrhage, increase the risk of mortality, hypothetically, at least. There is the economic concern which added cost with IV thrombolysis, especially if it doesn't improve the outcome.

There are the time concerns that stopping in the emergency room and giving IV thrombolysis, delay going to the Angio suite, performing in the vascular thrombectomy and achieving great perfusion. The counter argument is there is a large population of patients that may not have access to endovascular thrombectomy.

IV thrombolysis will be the first option to them that can be delivered as soon as possible, with not enough endovascular thrombectomy centers available to deliver the treatment in a timely manner. Now, it's a reverse question, should we go directly to the Angio suite? Should we stop in the emergency room and give IV thrombolysis? This is the current practice.

There were two randomized trials that the data was available within the last year. The first one is the DIRECT‑MT trial. This trial was already published in the New England Journal of Medicine from 41 Chinese stroke centers that looked at the question that I just mentioned.

Bridging therapy with IV alteplase then thrombectomy, versus directly going to a thrombectomy with patients with large vessel occlusion in the ICA, or on one or both. Patients presented within the first 4.5 hours in this trial. The trial enrolled 656 patients.

This was a non‑inferiority trial that attempted to show that direct thrombectomy is non‑inferior to bridging therapy with IV alteplase, and it showed that. It showed that going directly to the Angio was not inferior to giving IV alteplase and then going to the Angio.

There are specifics about the trial, there are specifics about the design of this trial, the SKIP trial, the other trial that we'll be discussing that I will go over in a minute. The overall functional independence was not lower with a direct thrombectomy as compared to bridging therapy.

There were numerically a lower rate with successful reperfusion in patients going directly to thrombectomy without IV Thrombolysis, similar safety profiles with death and symptomatic hemorrhage.

Now, what are those specifics that make the data not completely satisfying to answer the question? Not to criticize the trial, but important to know those for how you adjust your practice. This wasn't an inferiority trial that set up a margin of 20% non‑inferiority of direct thrombectomy, as compared to bridging therapy.

This is a very generous margin in non‑inferiority design. It's unclear how this number was came up with. The literature numbers are not necessarily at this wide margin. All of these patients presented directly to the Thrombectomy Center. There were no transfer patients. To digitalize the ability of the drip and ship, a model that's currently in practice, may not apply here.

Also, when you present directly to the Thrombectomy Center, receive TPA and go to the Angio Suite, there is a very short period of time and not enough time for IV Thrombolysis to work. In this trial specifically, there were 86.5% of the patient that completed IV Thrombolysis while they are receiving endovascular thrombectomy.

You ask, how would the treatment work if it was not even completely delivered? This is a very fair question. There was also a significant number of crossovers in the number of patients who did not receive endovascular thrombectomy as well.

I would say that this is data from a randomized trial on patients presenting directly to the Thrombectomy Center. You may say, "If I'm in the Thrombectomy Center, why are you stopping in giving TPA?" I would say that there has to be more data, and there are ongoing clinical trials on that. I will mention it, and we will be discussing it in the future.

The second trial is the SKIP Trial that was done in Japan. As of the time of recording this podcast, it's not published. It was presented in the International Stroke Conference in Los Angeles.

It asked the same question, IV alteplase plus Thrombectomy Bridging Therapy versus directly going to thrombectomy without bridging therapy in patients with large vessel occlusion, and then traditional collision at the ICA, and then one with good imaging profile on CT or MRI, only 204 patients.

It was a non‑inferiority design similar to the DIRECT‑MT trial. It showed the same that direct thrombectomy without bridging therapy was non‑inferior to bridging therapy with fewer numbers of symptomatic intracerebral hemorrhage.

Again, same concerns. Very generous non‑inferiority margin at 25%, even higher than the DIRECT‑MT limited number of patients, only 200 patients. The use of low‑dose IV alteplase in Japan. The dose was 0.6 milligram per kilogram, as compared to 0.9 milligram per kilogram in the United States should open the rest of the world.

I would say, not enough data to uphold IV Thrombolysis at this point. Not a generalizable data, because it's only patients who presented directly to the Thrombectomy Center did not give IV Thrombolysis the time to work, which potentially can happen in patients who are transferred.

There are ongoing trials that we will go in details more in the Future Horizons podcasts. These are the twist‑direct that's being done in the EU in Europe, and Mr. Clean‑No‑IV as well. We'll discuss those in more detail.

Within the same subject of the thrombectomy and its access, as I mentioned, this is a highly effective treatment. However, patient access to thrombectomy is not ideal. Improving the access, we can get the current access is an important subject. We, myself, as the first author, in a paper, looked at endovascular thrombectomy access and deaths in the United States, using geo‑mapping.

We found that only one‑third of the United States population has direct access to endovascular thrombectomy, within 30 minutes. This is a limited number of patients. We proposed optimization methodology with increasing the number of thrombectomy centers, flipping the current stroke treating centers from non‑thrombectomy to thrombectomy, or bypassing the non‑thrombectomy centers to going directly to thrombectomy centers.

This was a hypothetical methodology without looking at specific patients, proposing a potential solution. As you may imagine, bypassing resulted in large areas a number of patients are covered state by state. This has to be tailored at state level and local level. In some cases maybe makes more sense to establish a thrombectomy center in a certain population area. In general, bypassing has resulted in more number of patients being covered. This was published in "Stroke" earlier this year.

On that subject, yes. Bypassing would result in more patients covered, but does it result in better clinical outcomes? Logically, it should, because you're taking the patients with large vessel inclusions directly to the thrombectomy centers and performing thrombectomy earlier with reperfusion. This needed to be proven in a randomized clinical trial.

This was the REVASCAT's Trial. The REVASCATs, was a trial that was conducted in Catalonia, in Spain, using the RACE score. RACE is the Rapid Arterial Occlusion Evaluation, which is an assessment based on the stroke severity based on the facial droop, the arm weakness, the gait, aphasia. It's a 10‑point score, 0 to 9. Usually, when you are more than 4, you're more likely to have a large vessel occlusion.

There are other scores that paramedics and clinicians use to identify large vessel occlusions. The REVASAT Trial used the RACE score. It randomized patients in the field to either the traditional methodology, which was taking them to the closest center, if it's a non‑thrombectomy center, then they may receive IV thrombolysis and get transferred to the thrombectomy center, versus bypassing the non‑thrombectomy center and going directly to the thrombectomy center continues to show faster reperfusion, and that may reflect a better clinical outcome.

The trial was a big trial of 1,401 patients in 27 hospitals, in Catalonia. It showed faster reperfusion. As you may expect, these patients arrived faster to the thrombectomy centers. However, and the data is not published yet, but was presented in the European Stroke Conference. The primary outcome with improved functional outcome was not achieved.

Bypassing the non‑thrombectomy center and going directly to the thrombectomy center did not result in better functional outcome. It resulted in faster treatment. It resulted in more thrombectomies done without increasing the rate of complications during bypassing. There was only one of the patients who bypassed the non‑thrombectomy center versus 0.5% that went directly to the local center.

I may not say, surprisingly, but it did not show improved end of clinical outcome. Of course, there are many details that can play into that. There's the distance of the hospital and how much you would bypass to show benefit because of the balance of how long you drive, versus how fast you perform thrombectomy. Other details about the stroke severity also is important.

The other point that's important to mention is, one‑quarter, 25% of the patients in this trial ended up having intracerebral hemorrhage. Of course, that did not affect the main outcome, but one would say, "So what? Our patients with intracerebral hemorrhages are probably better off going to the large hospitals."

True, but it's as important to mention that this may potentially happen that you may take a patient, which you may assume has a large vessel occlusion to the thrombectomy center and then ended up having intracerebral hemorrhage.

The next trial was the Brazilian Trial, which was in Brazil. The trial aimed to show endovascular thrombectomy efficacy and safety in patients with large vessel occlusion and anterior circulation, in low‑income countries. You may say, how could they randomize patients to thrombectomy, versus medical management when there is already evidence from the prior trials that thrombectomy is efficacious and safe in this population?

The circumstances where the prior trials were done are not the same as in low‑income countries. The treatment was not reimbursed in Brazil, and probably not in many low‑income countries, so this trial was important to show that.

Indeed, it did show improved clinical outcomes, better outcomes, better functional independence, and a shift towards better outcomes in patients receiving endovascular thrombectomy as compared to best medical management without increasing symptomatic hemorrhage rates, in the first 8 hours.

This trial had a lot of impact. It showed that clinical trials at this level can be conducted in low‑income countries. It showed that thrombectomy can be feasible, efficacious, and safe in low‑income countries. That would help millions of patients across the world to have talked to the investigators of this trial.

Within thrombectomy, there remains a population that was not included in clinical trials. That's not a whole gut in population, but a population that was excluded, which is the basilar occlusion. The clinical issue with this population and why they were not included, different presentations, different imaging profiles, anterior circulation, and definitely different expected outcomes if patients do not get reperfused.

Patients with basilar occlusions may have very ominous outcomes if not reperfused. That's why it is extremely difficult to randomize a patient to thrombectomy, versus medical management. This is why, probably, this population was largely excluded. This is why we still do not have enough data from randomized trials, and who knows if we will ever have, but there were 2 pieces of data that were published within the last year.

The first one is the basilar prospective cohort study. This was not a randomized study, rather a prospective cohort study of 829 patients in China as they were assigned, where the treating physician without randomization, about three‑quarters of the patients ended up in thrombectomy, one‑quarter in medical management.

Again, non‑randomized fashion up to 24 hours, it showed better functional outcomes with thrombectomy. However, it was not randomized. There was an increase in the symptomatic hemorrhage rate, but significant reduction in mortality and better functional outcomes, as I mentioned.

The randomized trial in basilar occlusions is the best trial that was published in Lancet Neurology, and again was done in China, but the same difficulties with randomizing patients. There were only 131 patients in this trial with significant crossover both ways, but specifically to the thrombectomy arm.

It did not show statistical significance in improving clinical outcomes, because of the low number, the difficulty of randomizing these patients, the results definitely confounded by the crossover, but again, speaking to the same issue with the difficulty of randomizing this sub‑population to thrombectomy versus medical management.

Those were the two pieces of evidence in 2019 that people may apply to their practice. I believe people will continue to do thrombectomy in these patients, if presented without complete infarction seen on CT or MRI within the first 24 hours until further treatment options, or a basis of evidence become available, if they do.

The next important question in thrombectomy is imaging selection. This is a major question, because there are different camps that believe in different imaging profiles, there is a camp that believe in the advanced perfusion images, a camp that believes in simple not going to try CT. There are pros and cons to each definitely.

CT is more widely available, faster, cheaper, but it requires certain skills in reading it. There's inter‑rater reliability issues with difference in readings between different people. Perfusion can provide a systematic assessment, but there are times when it fails in patients with low ejection fractions, with motion artefact, with prior iodine contrast administration.

The question also stemmed further from the fact that the clinical trial, especially in the early ones done, 0 to 6 hours window for thrombectomy used different imaging methodologies, mostly CT, but they to EXTEND‑IA trial and a large part of this was prime trial used advanced perfusion images, so which one to use.

Does the CT perfusion add value to the imaging, selection in paradigms, prior to performing thrombectomy? We looked into that in the SELECT trial in a prospective manner.

The data on that has always been retrospective analysis without the systematic approach in this trial in a prospective cohort study, all patients received unified imaging profiles with CT/CTA, and perfusion images with mismatch assessment by the Rabid Software.

We had three specified imaging profiles, which spike are favorable and what is unfavorable. We looked at the agreement between CT and perfusion, how often they agree. We looked at the correlation between the favorable CT and the favorable perfusion images with the clinical outcomes after thrombectomy.

We looked at the different imaging profiles when the two agree or disagree, and how they correlate with the outcomes. After enrolling 361 patients, 285 went to thrombectomy and 76 went to medical management, prospective cohort non‑randomized fashion based on the treating physician decision.

We found high agreement between CT and perfusion. 81% of the patients had agreement between CT and perfusion, and had good imaging profile. About 20% had disagreement between CT and perfusion in images, which is an important population given the fact that one imaging is telling you to treat, versus the other telling you not to treat. What do you do with the patient?

In those who had favorable imaging profile in both CT and perfusion, which is the population that probably most of the clinical trials enrolled, not surprisingly was associated with the higher likelihood of good outcome. Patients with discordant profiles still had good rates of functional outcomes with thrombectomy and better than medical management, again not in a randomized fashion.

We are looking into in the SELECT2 trial in a randomized fashion, but there was signal for improved outcomes with thrombectomy as compared to medical management in discordant profiles when seen, when one imaging profile is favorable and the other one is not.

However, in patients who had unfavorable profile on perfusion images, even if the CT looked good, there were higher rates of symptomatic hemorrhage and mortality. The CT perfusion may add some prognostication value.

Again, we're looking to the heterogeneity of the treatment effect and how that changes based on the perfusion findings. How the thrombectomy treatment effect changed on the perfusion findings in patients with certain CT imaging profiles in the SELECT2 trial.

Until then, there is still no randomized evidence of CT, versus perfusion in which imaging modality you would choose this data provided the best available evidence at this point until we have the assessment of the heterogeneity of treatment effect in SELECT2, which we will talk about further in the future horizons.

The other important assessment from the SELECT trial was the large core sub‑population, which we published in [inaudible 24:28] , looking at patients with large core CT or perfusion. These patients were largely excluded from prior clinical trials, not sure of the efficacy and safety of thrombectomy.

It is a major subject of question and frequently presented population in the daily stroke practice. There are different arguments both ways. Patients with large ischemia already may not benefit from thrombectomy, higher risk of symptomatic hemorrhage. However, if you don't treat them, they will continue to infarct. They will end up having hemicraniectomy and high mortality rates, but is thrombectomy effective in those, we do not know?

In this pre‑specified secondary analysis from the SELECT trial, we looked at this population, there was a signal towards better outcomes with thrombectomy without increasing the safety concerns. This population is currently being randomized in the SELECT2 trial, which again, we will discussing further in the Future Horizons podcast.

The last subcategory, we would not finish this without discussing management of intracerebral hemorrhage and specifically blood pressure and intracerebral hemorrhage. To go to the end of 2019, to have the data on this, a paper in Lancet by Malowi Atole, that was an individual level meta‑analysis of the INTERACT2 and the ATAK2 trial.

These trials assist blood pressure management in patients with acute intracerebral hemorrhage. It is always a question, should you drop the blood pressure quickly, is it safe, does it result in improved functional outcomes, what are the parameters?

Does the variability of the blood pressure affect the clinical outcome? How do you apply this to your practice? The evidence on this so far has not been showing that, the latest was the ATAK2 trial not shown there, improved functional outcomes with rapid reduction of the blood pressure.

This individual level meta‑analysis from the INTERACT2 and ATAK2 indeed showed that lower mean systolic blood pressure was associated with better functional and safety outcomes, that higher variability in the blood pressure was associated with worst outcomes.

What are the best parameters to use, how to reduce the blood pressure, how fast, definitely highly advice that this is individualized by your patients. You check the patient overall personally. When there is no risk of intracerebral ischemia, if the patient has severe intracranial stenosis or internal carotid artery stenosis, I tend to drop the blood pressure quickly lower than 140 or 150 mmHg, and this data supports that.

I'm sure we will have more on intracerebral hemorrhage management in the future, but this was a good piece of data supporting lower and faster blood pressure control in this population.

Thank you so much. I hope that this was a helpful summary that can be applicable to your practice. As I mentioned, a detailed document, higher-level summary, let's say, with the references, and the main results, and the conclusions of these papers, and trials will be available with this podcast.