November 2013
Stroke Treatment Dispute — SIR Position Paper Responds to Perceived Threat Against Intra-Arterial Ischemic Stroke Revascularization
By Dan Harvey
Radiology Today
Vol. 14 No. 11 P. 28
Researchers conduct trials to improve patient care and add to the growing body of medical knowledge. But their good intentions sometimes can create confusion and even consternation among their peers. That’s the case with the Society of Interventional Radiology (SIR) and its reaction to three recent studies related to treatment for stroke victims.
The studies—Interventional Management of Stroke (IMS) III, Local Versus Systemic Thrombolysis for Acute Ischemic Stroke (SYNTHESIS Expansion), and Mechanical Retrieval and Recanalization of Stroke Clots Using Embolectomy (MR RESCUE)—all reached a similar conclusion: intra-arterial (IA) ischemic stroke revascularization provides no significant benefit over IV thrombolysis with tissue plasminogen activator (tPA). The findings—first revealed at the 2013 International Stroke Conference and subsequently reported in The New England Journal of Medicine—made headlines.
But SIR strongly disagrees with this conclusion, criticizing both the methodology and the implications of the studies’ findings. The society addressed its objections in a position statement published in the September issue of the Journal of Vascular and Interventional Radiology.
“Recent literature can leave patients, health care providers, and insurers uncertain about interventional treatments for stroke,” says SIR President Scott C. Goodwin, MD, FSIR, who is the Hasso Brothers Professor and chair of the department of radiological sciences at the University of California, Irvine. “SIR’s analysis explains why the minimally invasive endovascular treatments that interventional radiologists provide must remain available to certain individuals suffering a stroke. [Other] studies report good clinical outcomes in up to 60% of patients treated with minimally invasive, targeted techniques.”
“SIR is a stakeholder in providing endovascular stroke care, and these articles appeared simultaneously in a prestigious medical journal. Even though the studies had significant flaws, they came as a one-two-three punch to the face,” notes David Sacks, MD, FSIR, past president of SIR. “We felt it necessary to place the articles in perspective and in context. A first read of these three articles is extremely negative. Some prying beneath the surface is necessary to better appreciate the limitations, which are substantial.”
Carl Black, MD, chair of SIR’s Interventional Neuroradiology Service Line, says the society published its position statement mainly because the three studies do not reflect what most consider state-of-the-art patient selection and treatment of large-vessel ischemic stroke, which make the blanket findings “somewhat misleading.”
Specific Study Concerns
SIR believes the controversy regarding these three studies requires a close look at each one to detail its concerns.
IMS III
This multiyear study, sponsored by the National Institutes of Health (NIH), included patients who had suffered an ischemic stroke considered at least an NIH Stroke Score (NIHSS) of 10, indicating a moderately severe stroke. In about 80% of these cases, a large-vessel occlusion in the internal carotid, vertebrobasilar, or proximal intracranial arterial circulation causes the stroke.
IMS III randomized patients to receive IV tPA alone or as a bridge to IA treatment, with 434 patients receiving IA therapy but only 334 receiving the actual IA treatment plan. As it was a multiyear analysis, IA treatment was first represented as IA thrombolysis. Later, as reported, the researchers deployed mechanical thrombectomy using first- and second-generation devices (eg, stentrievers, large aspiration thrombectomy devices).
Most of the patients treated with IA thrombolysis with only tPA were those receiving endovascular treatment. Researchers concluded that IA therapy provided no additional benefit over IV therapy alone.
However, SIR cites these limitations of the study:
• Most patients in the IV-plus-IA treatment arm did not receive FDA-approved IV tPA treatment. Rather, they underwent a nonstandard, two-thirds IV dose infused for a 40-minute duration (as opposed to the standard 60 minutes), which could have distorted the findings of the IA treatment group.
• The IA dose of tPA could be as much as the remaining one-third of the standard systemic IV tPA dose (ie, 30 mg or less of IA tPA). SIR points out that there are no dose-ranging studies for catheter-directed IA tPA, but ultra-high catheter-directed doses of tPA have been shown to be less effective than lower doses. It is possible that the IMS III protocol therefore would be less effective than a much lower local dose for IA therapy.
• Approximately two-thirds of the patients in the endovascular group were treated with tPA alone. Only five patients receiving IA therapy were treated with a stentriever. SIR contends that IMS III is predominantly a trial of endovascular thrombolysis and not mechanical thrombectomy.
• Time to IA treatment in IMS III was 32 minutes longer than in IMS I and II. This could be a result of the increased use of a drip-and-ship transfer of patients to centers that provided IA therapy. IMS I/II and IMS III data indicate there is a 10% absolute reduction in good outcomes for every 30-minute delay in endovascular treatment. In IMS III, there was no clinical benefit of IA treatment following IV treatment when IA reperfusion started more than 51/2 hours after symptom onset. With patients treated within three hours of symptom onset, good outcomes were obtained in nearly 70% of patients in the IA treatment group vs. 39% of those in the IV treatment group in the IMS III trial.
• Patients without a large-vessel occlusion as the cause of ischemic stroke experienced much better outcomes compared with those with the occlusion. Patients without a large-vessel occlusion included those with peripheral branch occlusions or lacunar infarcts. They tended to do well with IV tPA alone.
As SIR points out, by definition, thrombectomy cannot be performed unless there is a large-vessel thrombus. “Current clinical practice reserves endovascular thrombectomy for patients who have a large vessel occlusion typically confirmed by preprocedural computed tomography (CT) or MR angiography,” according to the society’s position statement.
• Although IMS III did not require the use of an inline arterial air filter in endovascular cases, the filter’s use for intracerebral infusion is not uncommon. The actual incidence of using such filters on the tubing used to infuse IA tPA is not recorded. Such filters recently have been found to absorb 99.9% of the tPA.
Most patients in the IMS III trial received IA tPA, but patients treated with filters would not have received any therapeutic medication. In the absence of data on the filter’s use, their effect on this study’s outcomes is not known, but the effect could only be to reduce the likelihood of additional benefits from IA tPA treatment.
• Only 40% of patients in the endovascular arm showed thrombolysis in cerebral infarction (TICI) 2b or 3 revascularization (ie, more than 50% of affected territory), which are the categories with the highest rates of good outcomes (48% and 78%, respectively). These revascularization rates in the IMS III trial are lower than those achieved in current routine clinical practice using the same contemporary technologies utilized in IMS III.
In addition to its position statement, SIR added the following statement to its criticism of the IMS III: “The IMS III trial selected patients by NIHSS values, as opposed to the presence or absence of large vessel occlusion. Including patients who were likely to do well regardless of treatment and who would never be considered for IA therapy dilutes the results for both arms of IMS III and obscures the potential benefit for patients who receive arterial revascularization. This is analogous to having a trial of the effectiveness of penicillin when the selection criterion is ‘fever’ rather than ‘bacterial infection.’ The significance of this selection bias is confirmed in an analysis of the subset of patients who had a CT angiogram that confirmed the presence of a large vessel occlusion before randomization. When comparing those patients who would actually be treated with endovascular revascularization (i.e., those who had a large vessel occlusion), IA treatment provided patients with statistically significant benefit over IV tPA alone. Specifically, clinical outcomes in patients with an mRS [modified Rankin scale] score of 0-1 (almost complete neurologic recovery) were 35% for IA treatment and 19.8% for IV treatment. This is a complete reversal of the conclusion of IMS III that IA therapy provides no benefit over IV therapy and highlights the key point that IA therapy is effective in properly selected patients.”
SYNTHESIS Expansion
In this Italy-based trial, randomized patients received a systemic dose of IV tPA or IA catheter-directed tPA, with additional devices used in some patients. IA therapy was assigned to 181 patients but provided no significant benefit over IV therapy. Excellent clinical outcomes (ie, mRS score of 0 to 1 at 90 days) occurred in 35% of the IV therapy group and 30% of the IA therapy group. Good clinical outcomes (ie, mRS score of 0 to 2 at 90 days) occurred in 46% of the IV therapy group and 42% of the IA therapy group.
SIR contends the study has the following limitations:
• The trial used an unusual design of IV tPA selection criteria and doses for IA tPA administration and obtained results similar to those of IV tPA. There was no confirmation of large-vessel clots in the IV therapy group. If there was no large-vessel clot seen on cerebral angiography in the IA group, IA tPA was given anyway. As SIR points out, this is not the standard of care in the United States. The study authors did not report how often they did not find a large-vessel clot during angiography.
• Only one-third of patients were treated with a mechanical device, and only one-sixth were treated with a stentriever. “Therefore, this was really a trial of systemic-dose tPA administered IV versus IA,” SIR concluded in its position statement.
• Ultra-high, catheter-directed doses of tPA have been shown to be less effective than lower doses, and it is possible that the SYNTHESIS IA protocol therefore would be less effective than a much smaller IA dose or even the same systemic dose given via IV administration.
• There was no exclusion based on a low NIHSS score. Patients with a lower NIHSS score were expected to do well regardless of the type of treatment and tended to do well with IV tPA alone or even placebo, as demonstrated in the European Cooperative Acute Stroke Study III, in which 62% of placebo-treated patients with an average NIHSS score of 11 had an mRS of 0 to 2 at 90 days. Many patients with a low NIHSS score may never be more disabled than an mRS of 1 or 2. SYNTHESIS did not report outcomes stratified by an NIHSS threshold, such as a score of 10 as used in the IMS III trial.
• Reported outcomes for the endovascular treatment arm include 15 of 181 patients who did not receive IA treatment. The investigators provided no data on whether these patients did well or poorly, but their inclusion is not appropriate in an analysis of outcomes from IA therapy. An analysis of patients actually treated as intended is needed.
• TICI revascularization scores were not reported. Patients who do not undergo revascularization do not have good outcomes, but such critical data were not provided.
MR RESCUE
In the NIH-sponsored MR RESCUE trial, MRI identified patients with and without a significant ischemic penumbra. Patients were randomized to receive the standard of care (IV tPA if eligible or conservative treatment if not) or IA embolectomy with the use of first-generation devices. Sixty-four patients received IA treatment, 28 of whom (44%) received IV tPA.
As SIR reported, not only was there a lack of benefit of IA compared with IV therapy, but there was no benefit in selecting patients for IA therapy based on the presence of ischemic penumbra. Good outcomes (ie, mRS score of 0 to 2 at 90 days) occurred in 21% of the penumbra/embolectomy treatment group, 26% of the penumbra/standard of care group, 17% of the nonpenumbra/embolectomy group, and 10% of the nonpenumbra/standard of care group. The results of all study arms were similar to or worse than those for placebo reported in other IA stroke trials, SIR points out.
Limitations of this trial, according to SIR, include the following:
• The trial compared embolectomy and standard of care, with outcomes stratified by the presence of penumbra. Penumbra was defined as an infarct core less than 90 cm3 and core/total ischemic tissue volume less than 70% (ie, mismatch of 30% or higher). The trial failed to distinguish groups with or without penumbra. There was no separation based on the presence or absence of penumbra; the groups actually were separated on the basis of core infarct size.
• Revascularization was assessed at seven days, a clinically irrelevant time frame to assess revascularization, which must occur within minutes or hours to provide benefit.
• Angiographic revascularization was poor. TICI 2a flow was achieved in 40% of cases, but only 27% reached TICI 2b or 3 flow. These outcomes are even worse than those of IMS III and far worse than clinical practice documented in Interventional Stroke Therapy Outcomes Registry data with the use of similar techniques. Poor outcomes support a possible lack of operator skills or patient selection bias.
• Good outcomes were reported in only 21% of patients treated with penumbra and embolectomy. This is one-half the expected rate based on comparisons with other IA studies, including the IMS III trial and SYNTHESIS Expansion. This likely reflects low technical success of catheter-based revascularization, according to SIR.
• In the penumbra groups, patients treated with embolectomy had more infarct growth than seen with standard of care (27 cm3 vs. 7 cm3). For patients receiving the standard of care, there was almost no infarct growth in the penumbra group (7 cm3) but extensive infarct growth in the no-penumbra group (84 cm3). These results are counterintuitive and suggest that the conclusion of no embolectomy benefit is unfounded.
Potential Impact
The California Technology Assessment Forum concluded that mechanical thrombectomy lacked proven benefits based on the results of these three trials. SIR contends that these studies actually were not trials of mechanical thrombectomy.
SIR adds that the IMS III and SYNTHESIS Expansion trials used IA infusion of tPA at unproven doses; although the MR RESCUE trial included the use of thrombectomy devices, the conclusions are suspect as a result of limitations of the first-generation devices or operator failure. The society also contends that the studies do confirm that good outcomes require both rapid treatment and near-complete revascularization; when either is missing, outcomes are not satisfactory.
SIR concedes that the three studies with their negative findings—however limited—still have academic credibility, “even though the flaws are significant enough to mask beneficial effects of endovascular treatment.” But an obvious question arises: What will be the impact of these studies on how patients are treated and how they may wish to be treated?
Black indicates that some impact already is being felt. “As I talk to colleagues throughout the country, and even in our own practice, these papers have raised additional questions about utility of IA therapy,” he says. “However, for institutions that screen patients for IA treatment based on the presence or absence of large-vessel occlusion, the impact has been much less. The biggest potential impact could involve the institutions that are currently considering expanding their stroke therapy to include IA.”
Sacks says he isn’t quite sure at this point what the impact will be. “But I think it could be to some extent local—that is, within each hospital,” he says. “Facilities that have not yet offered endovascular stroke therapy might be dissuaded because it is an expensive therapy. If you were once on the fence and then you read these articles, which appeared in a prestigious periodical, you might not choose to offer it.”
Like Black, Sacks has spoken to colleagues about the studies. “The people I talked with that offer IA revascularization intend to keep offering it. On the other hand, it may be taken out of our hands due to reimbursement issues. It could become impossible to offer. But that hasn’t happened yet; reimbursement changes slowly.” The loss of insurance reimbursement could stymie what interventional radiologists believe remains a valuable procedure for some patients.
“Another reason why SIR wrote this commentary is that when insurance companies conduct analyses around reimbursement for stroke therapy, they want to have very thorough analyses of limitations, which takes in the latest studies and alternate interpretations,” Sacks says.
He hopes there isn’t a rush to judgment. “That appears to have happened in some places, and that will continue until contrary information is available,” he says. “That’s one of the reasons why SIR offered its commentary: to provide the alternative interpretation.”
Further Exploration
Black has a clear idea of what he believes needs to be done with further research. “Research must continue on multiple levels,” he says. “A well-designed registry, while not providing level 1 data, would still be important. This could be accomplished by modifying the recommendations for primary stroke centers that treat strokes using endovascular techniques and also for comprehensive stroke centers. In the standards now published, there needs to be a revision that includes greater emphasis on tracking outcomes at 90 days. In the standards now published, there needs to be greater emphasis on tracking clinical outcomes at 90 days. It is amazing how so few programs actually assess 90-day modified Rankin scores. There should also be additional controlled randomized trials, if possible, that directly look at the two therapies for similar treatment arms using latest-generation patient selection criteria and mechanical devices.”
Black cites a paper that comes closest to this: “Endovascular Therapy Yields Significantly Superior Outcomes for Large Vessel Occlusions Compared With Intravenous Thrombolysis: Is It Time to Randomize?” published in the September issue of the Journal of Neurointerventional Surgery. “It comes close to answering a lot of questions,” he says. “We do need to do a better job of tracking outcomes. At a minimum, we need to track current experience and learn from it, in the form of a registry.”
— Dan Harvey is a freelance writer based in Wilmington, Delaware.
Six Key Points From SIR’s Position Paper
The Society of Interventional Radiology’s position statement on intra-arterial (IA) ischemic stroke revascularization, as published in the Journal of Vascular and Interventional Radiology, is best summed up with six points that underscore how its position differs from the California Technology Assessment Forum:
• IA stroke revascularization is beneficial to patients in whom IV tissue plasminogen activator (tPA) fails or who are not eligible for IV tPA.
• Patients with a large-vessel occlusion who are treated rapidly (even with first-generation techniques) have improved outcomes compared with those treated with IV tPA alone.
• Second-generation mechanical thrombectomy devices are the most effective therapy for large-vessel occlusions.
• Randomized trials of second-generation mechanical thrombectomy devices compared with IV tPA alone need to be performed.
• Participation in research is critically important, but reimbursement for IA stroke revascularization should not be restricted to clinical trials.
• All IA cases should be contributed to a trial or national registry, including 90-day clinical outcomes. Reimbursement should be restricted to those facilities that participate in a trial or submit data to a national registry.
