To the Editor:
The results of the prematurely terminated Stenting and Aggressive Medical Management for Preventing Recurrent Stroke in Intracranial Stenosis (SAMMPRIS) trial because of the excessively high rate of stroke and death in patients randomized to intracranial stent placement is expected to affect the practice of endovascular therapy for intracranial atherosclerotic disease. A critical review of the patient population included in SAMMPRIS trial is provided with emphasis on generalizability of results and bias attributed to the cherry-picking phenomenon. The technical aspects of endovascular treatment protocol consisting of intracranial angioplasty and stent placement (IAS) using the Gateway balloon and Wingspan self-expanding nitinol stent and credentialing criteria interventionalists are reviewed. The influence of each component is estimated based on previous literature including multicenter clinical trials reporting on intracranial angioplasty and stent placement.
A National Institutes of Health (NIH)–funded randomized phase III clinical trial, (SAMMPRIS) (NCT00576693), was initiated to determine whether (IAS) and intensive medical therapy are superior to intensive medical therapy alone1 in reducing the primary endpoint (any stroke or death within 30 days after enrollment or stroke in the territory of the symptomatic intracranial artery beyond 30 days) by 35% (α = .05, 80% power, 2% loss to follow-up, and 5% crossover) during a mean follow-up of 2 years in high-risk patients with symptomatic stenosis of a major intracranial artery (http://www.clinicaltrials.gov/ct2/show/NCT00576693). Patients with transient ischemic attack (TIA) or nondisabling stroke within 30 days before enrollment attributed to 70% to 99% stenosis of a major intracranial artery were randomized (1:1) to intensive medical therapy alone or IAS plus intensive medical therapy. Intensive medical therapy in both arms of the study consisted of aspirin 325 mg/d for entire follow-up, clopidogrel 75 mg/d for 90 days after enrollment, and aggressive risk factor management primarily targeting blood pressure less than 130/80 mm Hg and low-density cholesterol concentration less than 70 mg/dL. Recruitment began in November 2008 and halted on April 11, 2011, NIH based on the recommendation of the Data Safety Monitoring Board after 451 (59%) of the planned 764 patients had been enrolled at 50 participating sites in the United States because 14% of patients treated with IAS experienced a stroke or died within the first 30 days after enrollment compared with 5.8% of patients treated with medical therapy alone, a highly significant difference http://www.nlm.nih.gov/databases/alerts/intracranial_arterial_stenosis.html. The 30-day rate of stroke and/or death in the IAS patients was substantially higher than the estimated rate of 5.2% to 9.6% based on registry data. There were 5 stroke-related deaths in the IAS arm and one non stroke-related death in the medical arm within 30 days after enrollment.
THE PRINCIPLES OF INTERPRETATION
The follow-up of currently enrolled patients and comprehensive analysis of the total trial data set will be important in the final interpretation of this study. However, in the interim, neurointerventionalists, neurologists, neurosurgeons, and comprehensive stroke centers as an organization must decide whether the procedure should be offered to patients with symptoms related to high-grade intracranial stenosis with or without local Institutional Review Board oversight. One of the major concerns is that the results of the prematurely concluded trial are overinterpreted and institutions choose policies that are applied to patient populations that were not included in the trial. Patients recruited in clinical trials may have more favorable baseline clinical and procedural characteristics (cherry-picking phenomenon) compared with those treated in the general population. Such bias impairs the generalization of results and reduces the ability to detect meaningful differences in effectiveness trials because of the low risk of recurrent events in the medical treatment group.2
SAMMPRIS TRIAL PATIENT POPULATION
The study included patients aged 30 years and older and 80 years and younger. Patients 30 to 49 years of age required at least 1 additional criterion to confirm atherosclerotic etiology. Similar exclusions of patients 80 years and older in Stenting of Symptomatic Atherosclerotic Lesions in the Vertebral or Intracranial Arteries (SSYLVIA) study,3 Wingspan study,4 Vitesse Intracranial Stent Study for Ischemic Therapy (VISSIT), and Apollo Stent for Symptomatic Atherosclerotic Intracranial Stenosis (ASSIST) study5 are expected to reduce the impact of the threefold higher rate of periprocedural stroke and death observed among stent-treated patients aged 80 years and older.6
TIA vs Ischemic Stroke as an Index Event
Patients with TIA or nonsevere stroke within 30 days of enrollment (defined using clinical and radiological criteria) were included. The qualifying event (TIA vs ischemic stroke) can influence the rate of periprocedural stroke and death as in the NIH-Intracranial Stent Registry7 in which patients in whom the qualifying event was ischemic stroke had a higher rate of stroke and death after 1 month compared with those with TIA (5% vs 16%). This finding could not be reproduced in the European INTRASTENT registry8 and another multicenter study.9
Interval Between the Index Event and Procedure
The time of interval to stroke or death in the SAMMPRIS trial was shorter than the 90 days used in the ASSIST5 study. The current data do not demonstrate a consistent relationship of periprocedural stroke and/or death with the interval between the index event and procedure. In 1 analysis, the rate of periprocedural stroke and death was similar among patients treated in less than 7 days from the qualifying ischemic event and those treated 7 days or longer after the qualifying event (4 of 33 vs 5 of 59),9 but in an analysis of NIH-ISR,7,10 a higher rate of stroke and death at 1 month was seen in patients in whom stent placement was performed within 10 days after the qualifying ischemic event (8% vs 17%).
Patients with progressive neurological signs within 24 hours before enrollment or thrombolytic therapy within 24 hours before enrollment were excluded from the SAMMPRIS trial. Such exclusion eliminates the 50% rate of periprocedural stroke and death with IAS in neurologically unstable patients11 and the 5% rate of symptomatic intracranial hemorrhages in acute stroke patients.12
Angiographic Severity of Stenosis
The SAMMPRIS trial required target stenosis of 70% to 99%, unlike in the SSYLVIA study,3 Wingspan study,4 and ASSIST study5 (50%-99%). The 1-month rates of stroke and death were 6.6%, 4.5%, and 6.5% in the SSYLVIA study,3 Wingspan study,4 and ASSIST study,5 respectively. However, 4 observational studies found no difference in the periprocedural stroke and/or death rate among patients with 50% to 69% stenosis compared with those with stenosis of 70% or more.7–9,13
Anterior vs Posterior Circulation
The SAMMPRIS trial, similar to the SSYLVIA study,3 Wingspan study,4 VISSIT, and ASSIST study,5 included both patients with anterior and posterior circulation stenosis. An analysis of NIH-ISR7 and another systematic review of 31 studies (1177 procedures)14 found that patients undergoing posterior circulation IAS had a 12% to 20% rate of stroke and death at 1 month compared with a 6% rate for the anterior circulation.14 Other observational studies have not found an association between the posterior circulation location of a lesion and the rate of periprocedural stroke or death.8,9,15
The SAMMPRIS trial required that the target artery had a normal diameter of 2.0 to 4.5 mm with lesion length of 14 mm or less. There has been variability in lesion criteria in SSYLVIA study3 (diameter 2.5-4.5 mm and length ⩽5 mm), VISSIT study (diameter 2.0-5.0 mm and length 16-31 mm), and ASSIST trial5 (diameter 2.0-4.0 mm and length <20 mm). The European INTRASTENT registry8 reported a lower rate of stroke and death with a target lesion shorter than 5 mm than lesions 5 mm and longer (6% vs 13.6%). Another multicenter study found no difference in periprocedural stroke or death among patients with lesions treated that were shorter than 7 mm and 7 mm and longer.9 The ASSIST5 study found that stent delivery failure was more frequent in lesions longer than 10 mm compared with those shorter than 10 mm (25% vs 3%), although no relationship could be demonstrated with periprocedural stroke and death.
Overall, the characteristics of the patients included do not possess any unique features that would increase the rate of periprocedural stroke or death compared with those treated outside the trial.
PROCEDURAL ASPECTS OF THE SAMMPRIS TRIAL
The primary intervention of the SAMMPRIS trial was IAS using the Gateway balloon and Wingspan stent (or any future US Food and Drug Administration–approved iterations of the balloon, stent, or delivery systems) to reduce the severity of stenosis to less than 50%. Patients, if randomized to stent placement, underwent IAS as soon as possible after enrollment (optimally within 1-3 days). All interventions were performed through a 6-French access system. Heparinization was instituted to a targeted activated coagulation time (ACT) of 250 to 300 seconds. Submaximal angioplasty (≈80% of target artery diameter) was performed with Gateway balloon (to a pressure of between 6 and 12 atm for ≈120 seconds). Subsequently, the self-expanding nitinol Wingspan stent delivery system (0.5-1.0 mm larger than the parent artery) was advanced over the exchange wire across the target lesion and deployed. The definition of technical success was based on postprocedure residual stenosis of less than 50%, consistent with reporting standards for IAS16 compared with other studies in which it has varied from 20% or less17 to 30% or less,18 to 50% or less.4
Wingspan Stent vs Balloon-Expandable Stents
The US Food and Drug Administration had approved the Gateway balloon/Wingspan stent system (Boston Scientific, Natick, Massachusetts) under a Humanitarian Device Exemption4 based on a safety study of 45 patients with intracranial stenosis of 50% or more4 that found a procedural success rate of 98% and 30-day and 6-month rates of death or ipsilateral stroke of 4% and 7% post-procedure. Subsequent, the postmarketing US Wingspan registry19 and NIH-ISR10 reported successful treatment in up to 99% of 78 and 129 patients, respectively. There were 5 strokes and/or deaths (6%) within 30 days in the US Wingspan registry,19 and the rate of stroke or death within 30 days or ipsilateral stroke within 6 months was 14% in the NIH-ISR.10 The high rate (25%-30%) of 50% or more angiographic restenosis in both registries10,19 and the high rates of stroke and/or death at some sites in the NIH-ISR raised early concerns about self-expanding stent performance.10 There are 3 balloon-expandable 316L stainless steel stents for intracranial stenosis: Neurolink stent (Guidant, Advanced Cardiovascular Inc., Indianapolis, Indiana), Pharos Vitesse intracranial stent (Micrus Endovascular, San Jose, California), and Apollo stent (MicroPort Medical). The periprocedural stroke and death rates were approximately 7% with use of balloon-expandable stents in the SSYLVIA study3 and ASSIST5 study. The current data suggest (not consistently) the possible superiority of self-expanding stents over balloon-expandable stents,9,20 although a review of 31 studies14 found similar periprocedural stroke and/or death rates in patients treated with balloon-expandable or self-expandable stent, respectively (9.5% vs 7.7%).
Restrictive Policy Toward Primary Angioplasty
Primary angioplasty as the sole modality or an acceptable option in selected patients was discouraged in SAMMPRIS trial. In general, primary angioplasty is preferred for complex long lesions that would require multiple stents (>12 mm), tortuous proximal vessels (≥2 acute curves requiring traversing), or relatively unstable guide catheter. The question remains whether periprocedural complication rates may be reduced by providing a more feasible option of angioplasty in technically challenging lesions. The approach is supported by a multicenter review of 190 patients21; the adjusted 30-day rate of stroke and/or death was 8.4% and 9.2% in the angioplasty- and stent-treated groups, respectively. In a subsequent systematic review of 69 studies, Siddiq et al22 found a rate of stroke and/or death in 8.9% and 8.1% of the angioplasty-treated and stent-treated groups, respectively. A recent multicenter study23 found a 30-day stroke/death rate of 5% with the new generation of angioplasty balloon catheters.
Periprocedural Antiplatelet and Anticoagulant Treatment
In the SAMMPRIS trial, aspirin 325 mg/d (for entire follow-up) and clopidogrel 75 mg/d (for 90 days) was initiated after enrollment. The duration of clopidogrel use was longer than the 30-day period of administration post-procedure used in the SSYLVIA study3 and the Wingspan study.4 However, longer pre- (at least 7 days) and posttreatment (6 months) aspirin plus clopidogrel daily was used in the ASSIST trial.5 Whether a longer duration of pretreatment with dual antiplatelet agents or additional use of probucol, an antioxidation agent (500 mg twice daily) resulted in lower rates of periprocedural stroke and/or death is not known.
The SAMMPRIS trial used broad principles for monitoring the intensity of anticoagulation without adjustment for various instruments (ACT target of 250-300 seconds) similar to the Wingspan study4 (ACT at 2-3 times baseline) and SSYLVIA study3 (ACT of 200-300 seconds). Based on variations in instruments, the percutaneous transluminal coronary angioplasty guidelines24–26 recommend a target ACT of 250 to 300 seconds with the HemoTec device (Medtronic ACT Plus, Minneapolis, Minnesota), 300 to 350 seconds with Hemochron device, more than 250 seconds with the Medtronic ACTII or more than 300 seconds with the i-STAT device (Abbott Laboratories, Abbott Park, Illinois).27 The ASSIST trial5 used the ACT target of 160 to 220 seconds based on the results of a previous study15 that found similar rates of intracranial hemorrhage or stent thrombosis compared with patients with ACT targets of 250 to 300 seconds. Unlike the SAMMPRIS trial, postprocedural heparin infusion was continued in the Wingspan study4 for 24 hours, and postprocedure subcutaneous low molecular weight heparin every 12 hours for 3 days was used in the ASSIST trial.5
Efforts before initiation of the trial to determine the optimal intraprocedural anticoagulation regimen and target ACT adjusting for the instrument may have affected the results favorably and should be considered in future trials.
CREDENTIALING CRITERIA FOR INTERVENTIONALISTS
The candidate interventionalists for the SAMMPRIS trial submitted procedure notes and documentation of the outcomes from past 20 consecutive IAS cases for independent review. The procedures acceptable in order of preference were Wingspan stent, Neuroform stent for atherosclerosis, coronary stents for atherosclerosis, angioplasty for atherosclerosis, and Neuroform stent for aneurysms. Different operator experience for carotid artery stent (CAS) placement and carotid endarterectomy in the Endarterectomy Versus Angioplasty in Patients with Symptomatic Severe Carotid Stenosis (EVA-3S) trial28 (12 CAS placements or 35 stent procedures in the supra-aortic trunks with 5 CAS placements) and Carotid Revascularization Endarterectomy Versus Stenting Trial (CREST)29 (10-30 CAS placements previously performed and up to 20 lead-in cases)30 possibly contributed to a twofold higher 1-month stroke and death rate associated with CAS placement in EVA-3S trial. An analysis from the NIH-ISR found that the rates of stroke and death at 1 month were 23% and 9% among low enrollment (<10 patients) and high enrollment (>10 patients) sites, respectively. Based on such data, the American Society of Neuroimaging31 recommended performing at least 25 endovascular procedures for intracranial stenosis as an appropriate qualification concurrent with Accreditation Council for Graduate Medical Education recommendation.32 Further analysis of the SAMMPRIS trial results according to strata based on procedure volumes may help provide a better understanding of this issue.
IMPACT ON RECOMMENDATIONS FROM PROFESSIONAL ORGANIZATIONS
The guidelines that consider a role for such procedures in general practice such as the American Society of Interventional and Therapeutic Neuroradiology, Society of Interventional Radiology, and American Society of Neuroradiology,33 American Heart Association Council on Cardiovascular Radiology and Intervention (class IIb, level of evidence C),34, and the consensus conferences on intracranial atherosclerotic disease35 may have to adopt 1 of 3 options: (1) demonstrate that the results of the trial do not apply to all settings and provide a clear description of such settings; (2) recommend continuation of current practice but with local institutional oversight and stringent ascertainment of periprocedural stroke and death; or (3) consider this procedure as experimental and recommend performance only in clinical trials.
The American Heart Association Stroke Council in 201036 and the National Institute for Health and Clinical Excellence37 considered IAS investigational and the American Heart Association rated it as class IIb (usefulness/efficacy is less well established by evidence or opinion) and level of evidence C (consensus opinion of experts). There appear to be 2 options: (1) the same class and level of evidence would continue until the detailed results of the SAMMPRIS trial are made available or (2) the recommendations may be changed to class III (conditions for which there is evidence and/or general agreement that the procedure or treatment is not useful/effective and in some cases may be harmful) and level of evidence B (data derived from a single randomized trial or nonrandomized studies).
Although the US Food and Drug Administration approved the intracranial stent in August 3, 2005,38 Centers for Medicare and Medicaid Services,39 considered IAS a promising but unproven therapy and provided coverage as part of investigational device exemption clinical trials only. However, the Centers for Medicare and Medicaid Services and American Medical Association issued relative value units and specific CPT codes for IAS procedures.40 The results of SAMMPRIS trial are expected to reduce the chances of reimbursement for IAS by other health insurers using a variety of mechanisms including per diems, diagnosis-related groups, case rates, or percentage of billed charges.
CONSIDERATIONS AFTER THE SAMMPRIS TRIAL
The need to develop new and effective treatments for patients with symptomatic intracranial stenosis cannot be undermined. The 1-year risk of stroke in the symptomatic intracranial stenosis territory was 19% for stenosis 70% or greater in patients in Warfarin-Aspirin Symptomatic Intracranial Disease study,41 and the combined stroke and TIA rate was 61% in patients with hemodynamic symptoms in the Groupe d'Etude des Stenoses Intra-Craniennes Atheromateuses symptomatiques study.42
The prerandomized trial period require modeling similar to safety, efficacy, and learning assessment of percutaneous transluminal coronary angioplasty using National Heart, Lung, and Blood Institute registry43,44 and balloon-expandable stent registries using standardized data collection and independent data monitoring45–47 before randomized trials. The SAMMPRIS trial relied on data derived from the NIH-ISR, which lacked the quality parameters for data ascertainment and collection seen in percutaneous transluminal coronary angioplasty and coronary stent studies.
It is likely that IAS will be offered at most institutions in some limited capacity. However, it would be of paramount importance to ensure that outcomes are tracked and reliably ascertained through either local or national registries.48 Participation in such a registry also reduces medicolegal liability for the operator and ensures support from peers. The American Heart Association Stroke Council49 metrics for measuring quality of care in comprehensive stroke centers require ascertainment of any stroke or death within 30 days of IAS. A phase II single-arm futility design single-arm intervention to prespecified stopping criteria maybe indicated to test primary angioplasty or another intervention to accrue appropriate data before proceeding to a larger scale trial.50