Interventional therapy for stroke: the evidence

interventional therapy for stroke

In part 2 of our EM Cases Journal Jam, we explored the literature looking at interventional therapy for acute ischemic stroke. (Part 1 on the evidence for thrombolytics can be found here.) The studies of interventional therapy for stroke tend to get broken down into the early (negative) studies and the later (positive) studies. For consistency, I’ll use the same break down.

The early (negative) studies


Broderick JP, Palesch YY, Demchuk AM. Endovascular therapy after intravenous t-PA versus t-PA alone for stroke. The New England Journal of Medicine. 2013; 368(10):893-903. PMID: 23390923 [free full text]

This is a multicenter, randomized, open-label trial.

  • Patients: 656 adult (age 18-82) ischemic stroke patients who had received tPa within 3 hours of symptom onset and had a NIHSS score of 10 or more (trying to identify patients with a major arterial occlusion).
  • Intervention: tPa (⅔ usual dose) plus endovascular therapy (approach chosen by the interventionalist).
  • Comparison: Intravenous tPa alone (usual dose).
  • Primary outcome: Modified Rankin scale of 2 or less at 90 days.


  • There was no difference in the primary outcome (mRs<3): 40.8% with endovascular therapy and 38.7% with tPa alone (ARR 1.5%; 95% CI −6.1 to 9.1%).
  • No difference in mortality (19.1% vs 21.6%, p=0.52).
  • No difference in symptomatic ICH (6.2% vs 5.9%, p=0.83). There was an increase in asymptomatic ICH with endovascular treatment (27.4% vs 18.9%, p = 0.01)


  • The trial was stopped early for futility. They had planned to enroll 900 patients.
  • They started the trial using the NIHSS score to identify major arterial occlusion, although it is only 80% accurate. As the trial progressed, more centers starting using CT angiography.


Ciccone A, Valvassori L, Nichelatti M. Endovascular treatment for acute ischemic stroke. The New England Journal of Medicine. 2013; 368(10):904-13. PMID: 23387822 [free full text]

This is a pragmatic, multicenter, randomized, open-label trial.

  • Patients: 362 adult (age 18-80) ischemic stroke patients within 4.5 hours of symptom onset if getting tPa or 6 hours for interventions.
  • Intervention: Endovascular therapy (no intravenous tPa) (mostly intra arterial tPa here, about ⅓ had a device used).
  • Comparison: Intravenous tPa
  • Primary outcome: Disability free survival (modified Rankin scale 0-1) at 90 days.


  • The primary outcome (disability free survival, mRs 0-1, at 90 days) was the same in both groups: 30.4% with endovascular therapy and 34.8% with standard care: OR 0.71; 95% CI 0.44 to 1.14; P = 0.16.
  • No difference in mortality: 8% with endovascular therapy and 6% with IV tPa p=0.53.
  • No difference in symptomatic intracerebral hemorrhage: 6% vs 6% p = 0.99.


  • Endovascular treatment took about 1 hour longer to get started than tPa.
  • Accepting different times for the different groups might be problematic. If you believe time is brain, waiting longer is clearly worse. On the other hand, waiting longer might just result in different populations, as it gives more time for TIAs, rapidly resolving strokes, and stroke mimics to become obvious. These patients have better outcomes, and are over-represented in the earlier time frames.
  • Primary outcome collected by telephone.


Kidwell CS, Jahan R, Gornbein J. A trial of imaging selection and endovascular treatment for ischemic stroke. The New England Journal of Medicine. 2013; 368(10):914-23. PMID: 23394476 [free full text]

This is a phase 2, multicenter, randomized, open-label trial.

  • Patients: 118 adult (age 18-85) ischemic stroke patients within 8 hours of large vessel, anterior-circulation strokes. Patients were all treated with tPa and were eligible if there was no recanalization on a subsequent CT or MR angiogram.
  • Intervention: Mechanical embolectomy.
  • Comparison: Standard care.
  • Primary outcome: Modified Rankin scale at 90 days.


  • Same mean score on the mRs at 90 days (3.9 vs 3.9, p = 0.99). Unfortunately, they only present the more standard metrics (mRS 2 or less) stratified by CT findings, so it is hard to summarize. There are no differences reported.
  • 17 (24%) patients in the endovascular group had complications.
  • 90 days mortality was very high (21%) and not different between groups.


  • This is an exceptionally select subgroup. 127 patients from 22 centers over 7 years means each center contribute less than one patient per year. (This is the only study that gives us a patient flow diagram (figure 1), but they leave out the selection process.)

The later (positive) studies


Berkhemer OA, Fransen PS, Beumer D. A randomized trial of intraarterial treatment for acute ischemic stroke. The New England Journal of Medicine. 2015; 372(1):11-20. PMID: 25517348 [free full text]

This is a pragmatic, multicenter, randomized clinical trial with open-label treatment and blinded end-point evaluation.

  • Patients: 500 adult (18 years and older) ischemic stroke patients. Initiation of the intervention had to be possible within 6 hours of symptom onset. Had to be an anterior circulation stroke (distal internal carotid, middle cerebral artery, or anterior cerebral artery) based on CT or MR angio.
  • Intervention: Intra-arterial treatment at the discretion of the treating physician (mechanical, intra-arterial thrombolysis, or both).
  • Comparison: Usual care
  • Primary outcome: Modified Rankin score at 90 days (collected by telephone interview).


  • Improved excellent outcome (mRS 0-1): 11.6% vs 6.0%
  • Improved functional independence (mRS 0-2): 32.6% vs 19.1% (ARR 13.5%; 95% CI 95% CI, 5.9 to 21.2)
  • No change in 30 day mortality: 18.9% vs 18.4%
  • No change in symptomatic intracerebral hemorrhage: 7.7% vs 6.4%


  • There were 502 patients at 16 sites over more than 3 years. That is less than 10 patients a year per site, or less than 1 a month.
  • The outcomes here were distinctly bad. In NINDS 39% of the tPa cohort achieved an mRS of 0-1. Here, only 6% on the control group got to this level. The intervention group actually fared much worse than the NINDS treatment group.
  • 84% of the interventional group actually had some kind of intra-arterial therapy done. Most of the time it was a retrievable stent. 10% of the time it was intra-arterial lytics in addition to a stent. 1 patient had intra-arterial thrombolytics in isolation.
  • 30 patients in the interventional arm also had stenting of their internal carotid done at the time of the procedure.
  • 89% of patients received intravenous tPa.

Other FOAMed commentaries


Campbell BC, Mitchell PJ, Kleinig TJ. Endovascular therapy for ischemic stroke with perfusion-imaging selection. The New England Journal of Medicine. 2015; 372(11):1009-18. PMID: 25671797 [free full text]

This is a multicenter, randomized, open-label trial.

  • Patients: 70 adult patients with acute ischemic stroke who were receiving tPa less than 4.5 hours after symptom onset. All patients had to have occlusion of the internal carotid or middle cerebral artery and evidence of salvageable brain tissue on CT perfusion imaging.
  • Intervention: tPa plus endovascular therapy initiated within 6 hours.
  • Comparison: tPa alone
  • Primary outcome: Co-primary outcomes were reperfusion at 24 hours and early neurologic improvement (≥8-point reduction on the NIHSS or a score of 0 or 1 at day 3).


  • More reperfusion at 24 hours (100% vs. 37%; P<0.001).
  • Increased neurologic improvement at 3 days (80% vs. 37%, P = 0.002).
  • Improved functional independence (mRS 0-2) (71% vs. 40%; P = 0.01).
  • Non-significant increase in excellent outcome (mRS 0-1) (51% vs 29%; P = 0.05).
  • No statistical difference in mortality, although the endovascular group looks better (9% vs 20%; P = 0.18)
  • No difference in symptomatic intracerebral hemorrhage (2 vs 0 patients, P = 0.49).


  • This is a very select group of patients. At 14 centers over more than 2 years, they enrolled 70 patients. (2.5 patients per center per year!) Even if you discount the 4 centers that didn’t enroll a single patient, they only managed 3.5 patients per center per year!
  • Reperfusion is not a patient oriented outcome. It is not ideal as a primary outcome. There should never be more than one primary outcome.
  • Trial was stopped early (original target was 100 patients) based on the MR CLEAN results.


Goyal M, Demchuk AM, Menon BK. Randomized assessment of rapid endovascular treatment of ischemic stroke. The New England Journal of Medicine. 2015; 372(11):1019-30. PMID: 25671798 [free full text]

This is a multicenter, randomized, open-label trial.

  • Patients: 316 adult patients with disabling ischemic strokes (NIHSS>5) within 12 hours after symptom onset. Patients had to have a small infarct core (ASPECTS score 6-10) and a CTA had to demonstrate proximal MCA occlusion (with or without associated ICA occlusion), and good collateral circulation (> 50% filling of MCA territory).
  • Intervention: Endovascular treatment in addition to standard care (including tPa). (Retrievable stents were recommended).
  • Comparison: Standard care.
  • Primary outcome: Modified Rankin score at 90 days.


  • Increase in functional independence (mRs 0-2): 53.0%, vs. 29.3% in the control group; P<0.001.
  • Reduced mortality: 10.4%, vs. 19.0% in the control group; P = 0.04.
  • No change in symptomatic intracerebral hemorrhage: 3.6% vs 2.7% in the control group; P = 0.75.


  • The trial was stopped early. Originally planned to include 500 patients. They had crossed a prespecified boundary for stopping early.
  • Like all of these studies, this is a very select group of patients. They say that they enrolled 1.44 patients per center per month. However, their math doesn’t work. 316 patients at 22 centers over 19 months is about 0.76 patients a month per center.
  • They state that they purposefully did not require screening logs, meaning that they are unable to estimate how many patients they screened to get to the 316 they included.
  • Of the patients randomized to endovascular therapy, 91.5% actually had endovascular treatment.

Read more


Saver JL, Goyal M, Bonafe A. Stent-retriever thrombectomy after intravenous t-PA vs. t-PA alone in stroke. The New England Journal of Medicine. 2015; 372(24):2285-95. PMID: 25882376 [free full text]

This is a multicenter, randomized, open-label trial.

  • Patients: 196 adult ischemic stroke patients receiving tPa who could get endovascular therapy within 6 hours. Patients all had imagining confirmed occlusions in the proximal anterior intracranial circulation and lacked large ischemic cores.
  • Intervention: Endovascular treatment with a stent retriever in addition to tPa.
  • Comparison: tPa alone.
  • Primary outcome: Modified Rankin scale at 90 days.


  • Functional independence (mRs 0-2) was higher in the endovascular group: 60% vs 35%, p < 0.001.
  • There was no difference in mortality (9% vs 12%, p = 0.5).


  • The trial was stopped early for efficacy.
  • Another very select group of patients. Although there is no flow diagram, the trial ran for 2 years at 39 centers, meaning that each center only contributed 2.5 patients per year.
  • They changed their imaging criteria halfway through.


Jovin TG, Chamorro A, Cobo E. Thrombectomy within 8 hours after symptom onset in ischemic stroke. The New England Journal of Medicine. 2015; 372(24):2296-306. PMID: 25882510 [free full text]

This is a multicenter, randomized, open-label trial.

  • Patients: 206 adult ischemic stroke patients who could be treated within 8 hours. Patients had to have imaging confirming and anterior circulation occlusion and couldn’t have a large infarct size. Patients were either ineligible for tPa or had tPa, but did not recanalize within 30 minutes. (Unlike the other newer trials, they did not use perfusion imaging as an inclusion criteria here.)
  • Intervention: Endovascular therapy with the Solitaire stent retriever in addition to medical management (include tPa for eligible patients).
  • Comparison: Medical management alone.
  • Primary outcome: Modified Rankin scale at 90 days.


  • Higher rate of functional independence (mRs 0-2) with endovascular therapy: 43.7% vs. 28.2% (OR 2.0; 95% CI 1.1 to 3.5).
  • Mortality non statistically different (18.4% with endovascular vs 15.5%, p=0.6).


  • The trial was stopped early “because of a loss of equipoise” after other trials were published. They had originally planned to enroll 690 patients. At the time the trial was stopped, they had not crossed the pre-specified stopping boundaries.
  • As an interesting note on the bias you see in these trials, in order to be a study site in this trial you had to be performing more than 60 mechanical thrombectomy procedures a year. Remember, when this trial was started, all the evidence was negative. Why were they performing (at least) 60 procedures a year when all the trials were negative? This gives you a sense of the bias of the treating docs: they all really believed this stuff works. Unfortunately, these trials are open label and using a relatively subjective outcome measure as their primary outcome, which makes the data somewhat less reliable.


3 early trials that were clearly negative. 5 later trials that were remarkably positive. So does that mean that our newest endovascular devices are miraculous; that endovascular treatment is the way of the future? Maybe, but we need to acknowledge a few significant problems with this literature.

First, every single study was open label. Although the outcome assessors were blinded, the treating physicians, patients, and their families were not. This is a very big deal, because the primary outcome of these trials is incredibly subjective. (For further discussion on the subjectivity of the modified Rankin scale, see Part 1: thrombolytics for ischemic stroke.) The primary outcome was also frequently assessed by telephone follow-up. Consider how patients or their relatives might respond to questions about their current abilities if they are aware that they received the incredible, new, technologically advanced endovascular treatment; or if they knew they had been denied such an amazing therapy. I expect the treating physicians also had a strong bias towards treatment. Remember, in order to be part of some of these studies, you already had to be performing at least 60 procedures a year, despite prior negative trials.

Another problem when trying to interpret this literature is that 6 of the 10 studies were stopped early. As a researcher, if you look at your data early and find no benefit, the trial simply continues. However, if you look and find a benefit, you stop the trial and report the results. Because outcomes are random, and the differences between the two groups will randomly vary over time, stopping positive trials early tends to systematically over-estimate the reported benefit.

The data here is also quite heterogenous. The studies had different inclusion criteria, different imaging requirements, differents timeframes, and different treatments. As a result, they can’t easily be summarized in a meta-analysis, and the best treatment population is not yet clear.

The big problems: screening and indication creep

Very few patients were eligible for these trials. As I emphasized above, even at major tertiary stroke centers, the best trials were able to enroll about 1 patient a month, and most trials only enrolled a handful of patients every year. This leads to a number of questions.

What is the true net benefit for all stroke patients? Unfortunately, none of these trials had an appropriate patient flow diagram. Without this information, it is impossible to know exactly how many people were screened and why they were included or excluded. We don’t see the harms to the patients being screened. There is clearly a benefit among patients who were eligible for these trials, but very few patients were eligible. If hundreds of contrast CT angiograms need to be done in order to identify a single eligible patient, how do the harms of those many tests weigh against the potential benefit for that one individual?

Furthermore, with so few patients being eligible, but such dramatically positive results, we are going to be tempted to expand use of this therapy to patients who would not have been included in these trials. In fact, I think we are already seeing this. If you work at a stroke center and see more than 1 or 2 patients a month being offered endovascular therapy, that almost certainly represents indication creep. We don’t know if endovascular therapy will help patients who were excluded from these trials. The benefits are certain to be less than what is demonstrated in these carefully selected populations, and if applied too widely we could easily end up causing harm.

Is it worth it?

The data here is of poor methodologic quality, but the more recent studies are consistent. Endovascular therapy appears to provide benefit to some stroke patients. Unlike tPa, which I am still unsure about, if I had the right kind of stroke, I would want endovascular therapy for myself. The most important question, in my mind, is not whether this treatment is worthwhile for individual patients, but instead what are the costs and how we make it work for our medical system?

In these trials, eligible patients were extremely rare. In some trials, they were only able to recruit 2 or 3 patients a year! At most, it seems like a major stroke center might anticipate seeing 1 or 2 patients eligible for this therapy each month. However, because eligibility is based on advanced imaging techniques, many hundreds of patients will still have to be screened to determine who is eligible. The question is whether this system works. Is the massive expense of redesigning our stroke systems and diverting ambulances worth it for one patient a month? Could that money be better spent elsewhere? How do the harms of screening hundreds of patients with contrast studies balance against the benefit to that one patient? How do we keep interventional stroke teams adequately trained if they are only going to see 1 or 2 patients a month? How do we fund these teams, that will be needed 24/7, if only 1 or 2 patients are to be seen? These questions aren’t really for the individual physician. Honestly, they probably shouldn’t be left in the hands of physicians at all. These are societal questions that have to have tackled on a much higher level.


The studies here have to be rated as low quality. I would like to see at least 1 properly blinded study, and we clearly need more research to determine which patients are best managed with endovascular treatment, the harms of screening, and the societal impacts of this strategy. However, the recent studies are consistent and demonstrate an important benefit. Based on what we know today, I would want this therapy for myself.

Other FOAMed commentaries

EM literature of note

SGEM #137: A Foggy Day – Endovascular Treatment for Acute Ischemic Stroke

EM Nerd: The Adventure of the Cardboard Box Continues

EM Nerd: A Truncated Summation of the Adventure of the Cardboard Box

Cite this article as:
Morgenstern, J. Interventional therapy for stroke: the evidence, First10EM, July 11, 2017. Available at:

Leave a Reply

7 thoughts on “Interventional therapy for stroke: the evidence”

%d bloggers like this: