Thursday Embargoed Media Briefing

Audio begins at 1:59

Kyra:Good morning, everyone. Welcome to the Embargoed Briefing for the Late-Breaking Science presentations that will be featured in today's Plenary Session II. Remember that the embargo on these presentations will lift at 11am Pacific Time. I also want to let you know that Dr. Mark Alberts, who's sitting over here on this side, is the American Stroke Association spokesperson. He'll be available to provide commentary after the presentations and the briefings. What we'll do is, like yesterday, we'll hear all the presentations, then we'll open things up for discussion. First, Dr. Thomas Brott from the Mayo Clinic in Jacksonville, Florida, will present the long-term results of the CREST study.

Dr. Brott:Good morning. It's a delight to be here to talk about the long-term results of the CREST trial. The background is that previously reported, now it's in 2010, that from 2,502 patients with greater than 70% carotid stenosis, randomized to stenting or endarterectomy, that after two and a half years of follow-up there was no difference in the occurrence of stroke, myocardial infarction, or death at 30 days, or subsequent ipsilateral stroke. We, of course, knew, all of us know, how life expectancy is increasing. You can see, there, that the life expectancy for a Medicare-aged woman is 20, and 18 for a Medicare-aged man. Whether or not two and a half years of follow-up is sufficient our patients could certainly debate, so we extended the results out to 10 years.

The study design, we won't go into in great detail. It'll be in the paper, today. You may have seen it already. 117 centers. Symptomatic and asymptomatic patients, which makes it different from the European trials. Intention to treat. The protocol was amended and the patients were re-consented. The primary long-term outcome was to look at durability, to answer questions for patients and their doctors who are going to be living a long period of time. We had the usual statistical analyses. In our study population we followed everybody after the four-year follow-up, where we re-consented patients to be followed out to 10 years. As you can see, there were 1,607 patients consented.

They differed a little bit. This is where a pointer would help. If we look at the second row, you can see that the patients who were consented were more often asymptomatic. Some of that had to do with the fact that we didn't begin studying asymptomatic patients until halfway through the trial. You can also see in the third row that patients that did not consent were more often diabetic. If we go down to the fifth row, you can see that they were more often current smokers. What were these long-term results? For long-term stroke, which was the endpoint, you can see that there were only 83 strokes. The percentages broke down as 6.9% for stenting, and 5.6% for surgery. That's over 10 years. There was no significant difference.

If we take a look by symptomatic status, the asymptomatic patients and the symptomatic patients, first of all, if we look at five years you can see the rates. If we compare that to what we have from the NASCET trial, the classic trial from the 1990s that looked at symptomatic patients, after those patients had their operation ... In the medical group, in that study, the medical group, their rate of stroke over five years was almost 10 times higher than those rates that you see. If we look at the next iteration, the asymptomatic patients, their rate over 10 years, and we compare that to the medical group in the asymptomatic trial reported in 2004, the Asymptomatic Carotid Surgery Trial, the rates in CREST were half what they were in the ACST trial. Very low rates.

If we look at all of the events for the patients, broken down by whether they were symptomatic or asymptomatic, you can see that they were the same regardless of treatment. Very interesting to us. For the first time, we can see that patients who are symptomatic, once they have their procedure, they're no more likely to have a stroke than patients who are asymptomatic. This hasn't been found in prior trials. The long-term composite results, that adds everything together, the results from the surgery or the stenting, the early results where there's a risk, and compares things over 10 years. You can see the percentages are still low when divided by 10. There's really no significant difference between stenting or surgery for that long-term endpoint, which is stroke, MI, or death in the first 30 days, and then subsequent stroke out to 10 years. That's reflected in this Kaplan–Meier curve, which you can see. You can see when we blow it up a little bit that the action, if there's any difference, is right at the beginning, and then the two procedures are the same.

The composite endpoint, all the different subgroups, is there are no significant differences. There's a little bit of a pattern for age, in that there's a greater separation of the two procedures, stenting not doing as well with the elderly patients as surgery. Those are in the fourth, fifth, and sixth rows. The p-value there means it's not significant. If we look at the composite of stroke and death, though, we do see a persisting difference similar to the difference we found in the first report. Stroke and death is 37% more likely in the stenting group. That's what that hazard ratio means of 1.37. That was statistically significant.

Again, with the Kaplan–Meier curves, when you stretch it out to 100 you don't see much. When you magnify it up, you can see it again. All the action is in the first few days. The separation in these procedures, with regard to stroke and death, happens within several days, during that time when the procedures are being done and the complications are most likely to occur. You can see the patterns again for the subgroups. Important subgroups, like men and women, symptomatic and asymptomatic, none of these show what we call an interaction, except age. You can see, again, it's getting close to significance, but not quite there.

Restenosis, revascularization, used to be a big problem in the '80s and '90s. We found no difference between stenting and surgery out to 10 years. Again, the rates are very low, only about 1% per year. That's reflected in the Kaplan-Meier curve, as you can see. We also found that our patients are living longer relative to the general population than probably was true before, although we don't have published data. You can see that for our women they only lived one year shorter than the general population, and for men they only lived two years less. I think another positive for the way we treat risk factors today.

In conclusion, our post-procedural rates for stroke, for stenting and surgery, were similar, and less than 0.7% annually for symptomatic and asymptomatic patients. That number is about the same for the general, walking-on-the street population age 65 to 75. In contrast to previous trials, symptomatic status was not a predictor of outcome. The composite, everything added together, were similar for stenting and surgery. This time, though, over a time horizon appropriate for elderly patients who have extended life expectancies, the difference in stroke and death favoring surgery at 10 years was driven by a greater number of periprocedural strokes following stenting. Thanks very much.

Kyra:Thank you, Tom. Our next presenter will be Dr. Christian Stapf from the University of Montreal who will present the five-year outcomes of the ARUBA study.

Dr. Stapf:Thank you very much. It's my privilege being here with you this morning. Thank you for coming. Here are the slides. I'm particularly grateful to our statistical team, Dr. Michael Parides, who is in the room, and to Dr. Overbey, who worked on the data that I'll be able to show. I have no disclosures. AVMs constitute a rare disease not everybody is familiar with. They are tangles of abnormal brain vessels with falsely-linked arteries and veins in the brain. They may be asymptomatic or present with headaches, deficits, seizures, or, worst-case scenario, stroke, hemorrhage. It's a rare disease, but nonetheless in North America there are more or less 5,000 new cases every year, usually young adults, and 3,000 will be diagnosed with an unruptured AVM. These are the patients we are talking about in the ARUBA trial.

ARUBA has been an NIH-funded trial, international, including all types of therapy for AVM treatment. We were planning to enroll 400 patients to test the effect on stroke prevention between best possible AVM eradication versus medical management. Standard of care was considered to be the eradication. The experiment in this trial was not to touch the AVM. It was a pragmatic trial, enrolling all MRI-confirmed AVMs. There should not have been any sign of hemorrhage or any prior treatment before enrollment. It's a stroke prevention trial, so we've been counting death and symptomatic stroke and, of course, as a secondary endpoint, the neurological status five years after randomization.

This trial has been funded since 2007. As in most trials, after 50% of the cohort had been enrolled there was an interim analysis. This interim analysis in 2013 had stopped enrollment because of a significant difference between the two arms. Follow-up at this time was relatively short. NIH funding stopped in July 2014, but all sites were eager to proceed to follow-up until July last year to get the five-year results. These are the results we will be presenting here.

The ARUBA cohort, just to give you a little glimpse as compared to population-based data, is highly representative of what AVMs look like in the general population. I will just draw your attention to the lower left cell, modified ranking scores at entry, only zero and one, because none of these patients had a stroke. Then, when you look at the AVM sizes, they are overall smaller in the trial because we excluded the untreatable AVMs. That's an artifact. It's in the nature of the selection. The so-called Spetzler-Martin grading scores for treatment, we have more of the easily-treatable AVMs in this trial as compared to the general population where there are also untreatable AVMs.

The primary outcome as randomized for stroke and death shows that the risk of stroke and death in the medical group, red line, is significantly lower as compared to the interventional group, blue line, with a hazard ratio of 0.31. If you still consider treatment, despite these data, the numbers needed to harm by treatment is five patients over five years. These results were based on 41 primary endpoint events in the interventional arm versus 15 events in the medical management arm. In the lower part of the table, if you analyze it per protocol, there are a few crossovers. Eight patients crossed over from medical management to IT, and 15 in IT never received IT. That's an artifact due to the fact that the interim results basically stopped the interventional treatment plans in a couple of the patients. When you look at the data as treated, the effect is even stronger with an almost 80% risk reduction for death and stroke when you do not touch the AVM. The number needed to harm on treatment is three patients.

The neurological outcome at five years goes into the same direction. Remember, all patients we're ranking zero and one at entry. Among the 96 patients who reached the five-year outcome point, there's a significant difference, with 38% disability in the interventional group, 18% in the medical management group, mainly due to stroke, and a relative risk of 0.46 favoring non-interventional follow-up. This trend is even stronger when analyzed as treated. There was serious harm associated with treatments with overall 52 strokes, also secondary and third strokes in the interventional arm, significantly more focal deficits independently of stroke, significantly more epileptic seizures, and no difference seen for headache.

The question is, are there subgroups that would behave differently? The sensitivity analysis show a very robust effect that's not dependent on age, on sex, on symptomatic status, on the AVM size, venous drainage, and no difference between world regions. There is one subgroup, that's the Spetzler-Martin I group, that shows a non-significant trend favoring intervention, but that effect disappears in the as-treated analysis.

The treatment itself usually was carried out two months after randomization. If you look at patients undergoing treatment who had a primary endpoint event, the median time between treatment and the event was one day, suggesting a relationship. If you look at the treatment interventions in the treated group, among patients receiving any endovascular treatment during the course, 66 patients, 50% had a primary endpoint. Among the 22 patients undergoing surgery at any moment, 41% reached a primary endpoint. For radiotherapy, 36%. These effects remained the same if patients were treated only by one modality or by multimodal treatment.

To summarize, this is the first pragmatic management trial in unruptured AVM patients. It's a representative study cohort. The bias in the cohort, if any, is favoring intervention. Lessons learned is, the downside is living with an AVM is a risk factor for stroke. The spontaneous hemorrhage rate was 2.1% per year. This risk goes up by the factor of 4.5 if you start touching the AVM, and the risk of having a functional deficit multiplies by 2.5, and there's no benefit for epilepsy and headaches. This risk exists across all treatment modalities and AVM subgroups.

ARUBA is a stroke prevention trial that has shown the largest effect ever seen in such a trial, with an 80% risk reduction for death and stroke when you don't touch the disease. The treatment choice is available worldwide and at low cost. The trial results should be systematically disclosed to patients. Based on current knowledge, we have to assume that interventions may be dangerous, numbers needed to harm, five, cannot be safely recommended and should only be offered as part of another controlled clinical study. I think the subgroup analysis may lead the path to more hypotheses. Thank you very much.

Kyra:Thank you, Christian. Our next speaker will be Dr. Daniel Hanley from Johns Hopkins Hospital. He'll present the primary results of the CLEAR-III study.

Dr. Hanley:Thanks, Kyra. Thank you for this opportunity. Can I have the pointer? The phantom pointer. The background on this is, the best estimates are there are 5,000,000 brain hemorrhages in the world every year. This is a major problem in the United States, where, if you count first and second strokes and you say there are 20% of those are hemorrhages, there are about 200,000 hemorrhages here. The worldwide number, I think, grabs everybody's attention. A second issue is that the estimates of disability, because we have so many treatments for ischemic stroke now, the estimates of disability are that those 5,000,000 hemorrhages cause more disability than all the ischemic strokes, in terms of disability, lives per year. Hemorrhage has been ignored because it's the smaller brother or sister to ischemic stroke with that 20/80 paradigm.

We've been fortunate that the last background information is that both the peer review environment, several of those people are actually in this room now, I don't know specifically who they are, but I think several of those people are in this room, the peer review environment has recently been supporting and helping the NIH fund trials in this area. CLEAR is the first big phase III trial that NIH has funded. The subgroup that CLEAR-III identified is a small subgroup. Just like the carotid T occlusions in ischemic stroke, it's a group where, at least hypothetically, there was a potential for benefiting that subgroup of patients. It represents a model of how ICH might be attacked. That's the background. These are our disclosures.

The overall hypothesis was, does IVH reduction, so IVH is when the blood spreads into the ventricular compartment from a bleeding focus within the brain, does IVH reduction, if you use a routinely-placed extraventricular drain, EVD, with the addition of ultra low-dose Alteplase, which I've presented here several times, does it alter functional outcomes in patients presenting with a large IVH and a small ICH? We had to use pragmatic definitions for those two things, large IVH and small ICH, since we were in a data-free vacuum at the time. I'm going to skip over a lot of the slides. We have a standardized protocol that we've done in the four prior studies.

A couple novel things that our reviewers helped us do, besides double-blinding, we used so called adaptive randomization, where in real time we looked at every single patient, in terms of the severity factors, and let a computer balance them by enrollment so that the main determinants of severity could be balanced throughout the entire trial. We did team training. We actually started it before the interventional trial started. The same principles, where the stroke physician, the surgery physician, and our surgery physician's on my left, IssamAwad, and the ICU physicians, and the nursing team all got training as part of investigator training. Then, they got up to monthly renewals. We validated how well or how poorly they complied with the training.