Portal de Periódicos da CAPES

Overview of the 2011 Food and Drug Administration Circulatory System Devices Panel Meeting on the ACCULINK and ACCUNET Carotid Artery Stent System

2012; Lippincott Williams & Wilkins; Volume: 125; Issue: 18 Linguagem: Inglês

10.1161/circulationaha.111.073486

1524-4539

William A. Gray, Charles A. Simonton, Patrick Verta,

Cardiovascular Health and Disease Prevention

HomeCirculationVol. 125, No. 18Overview of the 2011 Food and Drug Administration Circulatory System Devices Panel Meeting on the ACCULINK and ACCUNET Carotid Artery Stent System Free AccessResearch ArticlePDF/EPUBAboutView PDFView EPUBSections ToolsAdd to favoritesDownload citationsTrack citationsPermissionsDownload Articles + Supplements ShareShare onFacebookTwitterLinked InMendeleyReddit Jump toSupplemental MaterialFree AccessResearch ArticlePDF/EPUBOverview of the 2011 Food and Drug Administration Circulatory System Devices Panel Meeting on the ACCULINK and ACCUNET Carotid Artery Stent System William A. Gray, MD, Charles A. Simonton, MD and Patrick Verta, MS Stat, DVM, MD William A. GrayWilliam A. Gray From the Center for Interventional Vascular Therapy, Columbia University, New York, NY (W.A.G.); and Abbott Vascular, Endovascular Global Clinical Science, Santa Clara, CA (C.A.S., P.V.). , Charles A. SimontonCharles A. Simonton From the Center for Interventional Vascular Therapy, Columbia University, New York, NY (W.A.G.); and Abbott Vascular, Endovascular Global Clinical Science, Santa Clara, CA (C.A.S., P.V.). and Patrick VertaPatrick Verta From the Center for Interventional Vascular Therapy, Columbia University, New York, NY (W.A.G.); and Abbott Vascular, Endovascular Global Clinical Science, Santa Clara, CA (C.A.S., P.V.). Originally published8 May 2012https://doi.org/10.1161/CIRCULATIONAHA.111.073486Circulation. 2012;125:2256–2264IntroductionThe Circulatory System Devices Panel of the Medical Devices Advisory Committee to the US Food and Drug Administration (FDA) was convened on January 26, 2011, to review the application by Abbott Vascular (Santa Clara, CA) for a post–market approval supplement (PMA-s) to extend the indications for use of their ACCULINK Carotid Stent System. The stent system had received FDA approval in 2004 for patients at high risk for adverse events after carotid endarterectomy (CEA). Abbott Vascular was seeking to extend its approval to patients at standard risk for CEA on the basis of the data and outcomes derived from the Carotid Revascularization Endarterectomy versus Stenting Trial (CREST; ClinicalTrials.gov Identifier NCT00004732). Herein are presented the essential elements of that day-long meeting.CREST BackgroundStudy Conception and TimelineThe organizational work on the CREST project was initiated in 1996 by the project's executive committee, which first published on its concept, rationale, and design in 1997.1 The experiential basis for the trial was chiefly 2 small stenting registries in high-surgical-risk, symptomatic patients totaling ≈200 procedures and reporting mortality rates of up to 3% and stroke rates of between 3% and 6%.2,3 The CREST executive committee recognized the novel nature and early stage of the technique, and in accordance with the tenets of clinical equipoise, as well as to provide reassurance to randomizing physicians that a reasonable safety of carotid artery stenting (CAS) had been established at an operator level and across sites, a rigorous credentialing phase was detailed for all participating centers. One of the stated secondary goals of CREST, to describe this experience in the lead-in phase of CREST, has been completed and published elsewhere.4 Although a 3-year enrollment period was originally anticipated for the main randomized trial, based on 50 sites recruiting 20 symptomatic patients per year, CREST ultimately required nearly 8 years and 120 centers to complete enrollment. Some of this delay can be ascribed to the pace of center certification given the aforementioned CAS credentialing requirements and the lack of qualified operators early in the course of the trial; although there were 47 centers selected to participate in CREST as of June 2001 (6 months after the initial patient was randomized), only 8 centers were approved to enroll lead-in CAS patients, and only a single center was qualified to randomize patients.5National Institutes of Health Analysis FrameworkCREST received its operational funding from the National Institutes of Health, National Institute of Neurological Disorders and Stroke (NIH-NINDS) in January 1999 through an investigator-originated (R01) grant (NS 38384), which was subsequently converted to a cooperative agreement (U01) for the duration of the trial. Device support was provided by Guidant (now Abbott Vascular, Santa Clara, CA). Before trial initiation, the Healthcare Financing Administration was required to modify a longstanding national noncoverage policy for carotid angioplasty to allow reimbursement for the hospital and physician costs of the trial for participating Medicare and Medicaid beneficiaries, which became effective in July 2001. The primary NIH analysis has been published,6 along with several of the secondary analyses.7–9 Objectives, end points, and analyses for the NIH evaluation are compared below to those specified by the FDA for the post–market approval (PMA) analysis (see Methodological Differences Between NIH and PMA Analyses).FDA Binding Agreement HistoryIn 1997, during the initial CREST application review, NIH-NINDS recommended the use of a single stent device. In May 1999, Guidant Corporation (now Abbott Vascular), the manufacturer of the ACCULINK Carotid Stent System, agreed to participate in CREST as the sole supplier of the study device and initiated formal discussions with the FDA to support an indication for patients at standard risk for adverse events from CEA. These efforts resulted in a binding agreement between FDA and Guidant in July 1999. In this FDA binding agreement, the primary end point for device analysis was defined as stroke, myocardial infarction (MI), or death within 30 days, plus ipsilateral stroke to 1 year, which differed from the NIH-NINDS primary end-point follow-up of 4 years. To support product approval, the FDA requested a hypothesis test of noninferiority between the 2 procedures for the primary end-point analysis, with a margin of 2.6%. Importantly, this analysis was predefined as distinct from the NIH-NINDS superiority analysis and was in keeping with predicate and subsequent carotid stent device trials in terms of noninferiority hypothesis testing formulation, safety and effectiveness end-point definitions, and temporal duration of assessment.10Although the potential for disparate outcomes between the NIH and FDA analyses was contemplated by the CREST executive committee, this probability was calculated to be 85 countries, with no evidence of safety concerns. On the basis of the results of the ACCULINK for Revascularization of Carotids in High-Risk Patients (ARCHeR) pivotal trial,13 Guidant previously submitted the PMA to the FDA for the device system tested, which consisted of the ACCULINK carotid stent and the ACCUNET embolic filter. In August 2004, the FDA approved the ACCULINK Carotid Stent System for CAS in patients with severe carotid stenosis at high risk for adverse events from CEA, both with and without prior neurological symptoms. The CREST panel deliberations did not include this high-risk population; therefore, the outcome of the panel vote would not affect the existing indications for use14 in high-risk patients.As a condition of the initial 2004 approval, the FDA required Guidant to conduct a postmarket study (ie, CAPTURE [Carotid ACCULINK/ACCUNET Post Approval Trial to Uncover Rare Events]) intended to assess both the incidence of rare and unanticipated events not captured in the pivotal study (ARCHeR, enrollment completed in 2002) and the adequacy of technology transfer from the trial setting to the clinical environment via physician training programs. The outcomes of CAPTURE (enrollment completed in 2006) and the subsequent study, CAPTURE 2 (performed to provide Centers for Medicare and Medicaid Services coverage with evidence development; enrollment completed in 2010) have been extensively published elsewhere.12,15–18 Since conducting ARCHeR, rates of 30-day DSMI with the ACCULINK/ACCUNET system have been noticeably reduced; specifically, 30-day DSMI rates for ARCHeR, CAPTURE, and CAPTURE 2 were 8.3%, 6.1%, and 3.5%, respectively. These studies, representing >180 sites and 450 operators in the United States, established the safety and effectiveness of the device for patients at high surgical risk in nontrial clinical settings. The clinical outcomes achieved in these postmarket studies of CAS in patients at high surgical risk met the American Heart Association guidelines threshold of adverse events set for CEA in patients at standard surgical risk; these AHA guideline thresholds were derived from outcomes of prior CEA trials in nonoctogenarian, standard-risk surgical patients.Current Application for a PMA-sIn submitting the CREST PMA-s, the funder sought an extension to the current approved use of the ACCULINK Carotid Stent System (ie, in patients with carotid stenosis considered at high risk for adverse events from CEA) to include patients with carotid stenosis considered at standard risk, regardless of symptomatic status. The proposed changes to the indications for use are consistent with the population studied in CREST, specifically symptomatic standard-risk patients with carotid artery stenosis >70% stenosis by ultrasound or >50% stenosis by angiogram and asymptomatic standard-risk patients with >70% stenosis by ultrasound or >60% stenosis by angiogram.Methodological Differences Between NIH and PMA AnalysesThe results of the NIH analysis of the CREST data were published in the New England Journal of Medicine in 2010.6 Although different in objectives and methodologies, these 2 analyses were nevertheless shown to be consistent and complementary (Table 2).Table 2. Differences Between CREST NIH-NINDS and PMA AnalysesTrial componentNIH-NINDSPMAHypothesisSuperiorityNon-inferiorityAnalysis populationIntent-to-treat*Per-protocol (primary analysis)Additional populations: Adjusted per-protocolIntent-to-treat†As-treatedModified as-treatedPrimary end pointComposite of periprocedural death, stroke, or MI, plus ipsilateral stroke to 4 yComposite of periprocedural death, stroke, or MI, plus ipsilateral stroke to 1 yDefinition of "periprocedural"From day of randomization to 30 d post-procedure. When no procedure was done, or if the non-assigned procedure was performed or if procedure was performed more than 30 d after randomization, from day of randomization to Day 36From day of procedure (Day 0) to Day 30Start of adverse event countDay of randomizationDay of procedureAdverse events included in analysisMI was defined as CK-MB or troponin elevation to a value of ≥2 times the upper limit of normal for individual clinical center's laboratory plus either chest pain or ECG evidence of ischemiaMI was defined as all adjudicated MIs determined by an independent Myocardial Infarction Adjudication Committee to be "possible" or "definite"*All patients as randomized.†All patients as randomized minus those with a primary end point event before the procedure.The objectives of the PMA analyses are device specific and meant to support FDA assessment and possible approval of the device, whereas the objectives of the NIH analyses are to provide academic and scientific evaluation of 2 carotid revascularization strategies. Because the mandate of the FDA is to provide reasonable assurance of safety and effectiveness of the device, these principles guide the statistical methodologies and analytic plan for the PMA. The PMA analyses used a noninferiority hypothesis test performed on a primary end point at 1 year, using the per-protocol population, whereas the NIH analyses used a superiority hypothesis test performed on a primary end point at 4 years, using the intent-to-treat population. The PMA analyses included an adverse event count that began on the day of procedure, whereas the NIH analyses tallied adverse events beginning on the day of randomization. Another key difference was that all adjudicated primary end points were used in the PMA analyses; for example, the PMA analyses included all MIs adjudicated as "possible" and "definite." This strict adjudication ensured the most conservative approach, in keeping with the FDA precedent for determining device safety. In contrast, the NIH analysis used the MI protocol definition from amendment V, in which biomarker changes were required to define an MI in addition to either clinical symptoms or ECG changes consistent with ischemia. As a result, there were 20 fewer MIs reported in the NIH analyses than in the PMA analyses (12 for CEA and 8 for CAS). Despite all the differences outlined above, the 2 analyses yielded similar results. The noninferiority hypothesis test was performed post hoc to the NIH primary end-point outcomes, which yielded results that were fundamentally the same as those achieved in the NIH analysis, with the primary end point meeting the noninferiority criteria.For a more detailed description of the distinct objectives of the 2 parallel primary end-point analyses (ie, NIH and PMA) and respective secondary and post hoc analyses, as well as the associated differences in end-point definitions, analysis populations, and statistical methodologies, please refer to the online-only Data Supplement Addendum to this panel report.PMA Analysis ResultsPrespecified Primary End-Point AnalysesFigure 1 is the graphical representation of the primary end point of CREST PMA analyses (ie, composite of all death, any stroke, or MI to 30 days plus ipsilateral stroke from 31–365 days) in the per-protocol analysis population. There is an absolute observed difference of 0.5% between the therapies, which, along with the accompanying 1-sided 95% confidence limit of 2.26%, is within the 2.6% margin of noninferiority (Figure 1A). Thus, CAS was demonstrated to be not inferior to CEA in standard-risk patients. In addition, every other prespecified analysis performed on the analysis populations consistently satisfied the noninferiority criteria (Figure 1B) for the stent system, therefore further supporting the robustness of the conclusion and interpretation.Download figureDownload PowerPointFigure 1. Comparison of carotid artery stenting (CAS) and carotid endarterectomy (CEA) in the Carotid Revascularization Endarterectomy versus Stenting Trial for the primary end-point outcomes in the primary per-protocol population (A) and in additional prespecified analysis populations (B). 95% CL indicates 95% confidence limit; pNI, probability of noninferiority; PP, per protocol; Adj, adjusted; ITT, intent-to-treat; AT, as treated; and MAT, modified intent-to-treat.Periprocedural Primary End-Point Outcome MeasuresRates of the periprocedural component of the composite end point (DSMI at 30 days) did not differ significantly for patients undergoing CAS (5.8%; 95% CI, 4.5% to 7.3%) or CEA (5.1%; 95% CI, 3.9% to 6.5%) and met the prespecified noninferiority hypothesis demonstrating that CAS was noninferior to CEA (Table 3). Rates of death and major stroke were very low compared with results from prior large randomized trials and were not significantly different between therapies. Minor strokes occurred more frequently in the CAS arm (3.2%; 95% CI, 2.2% to 4.4%) than in the CEA arm (1.5%; 95% CI, 0.9% to 2.4%). MI occurred more frequently in the CEA group (3.4%; 95% CI, 1.2% to 2.9%) than in the CAS arm (2.0%; 95% CI, 2.4% to 4.6%) and generally balanced the outcome event totals, which resulted in the equivalent composite end-point outcomes between groups for DSMI. The probability values listed in Table 3 are for descriptive purposes only, because CREST was neither designed nor powered to draw statistical inference in comparisons of components of the primary end point. No adjustments were made for multiple comparisons.Table 3. Periprocedural Outcomes for the PMA AnalysisPer ProtocolCAS N=1,131CEA N=1,176DifferenceUnadjusted P-value*All Death, Stroke or MI5.8% (65)5.1% (60)0.7%0.5200Death0.53% (6)0.26% (3)0.27%0.3335Any Stroke4.1% (46)1.9% (22)2.2%0.0019Major Stroke0.9% (10)0.4% (5)0.5%0.2005Minor Stroke3.2% (36)1.5% (18)1.7%0.0088MI2.0% (22)3.4% (40)−1.5%0.0387*Fisher's exact P-values were not adjusted for multiple comparisons; P-values for descriptive purposes only.Post Hoc Analysis of the Temporal Trends of End-Point OutcomesCREST enrolled subjects over an 8-year period, from 2000 to 2008. In 2004, the first FDA approval for CAS was granted (ACCULINK Stent System for high-surgical-risk patients based on the results of the ARCHeR studies13), and in 2005, the Centers for Medicare and Medicaid Services authorized reimbursement of the ACCULINK Stent System in high-risk surgical patients who were either symptomatic or participating in approved studies. As a condition of approval, the FDA mandated a postmarket study. During the same time frame, other devices were approved for the same indication, and in every case, a postmarket study was mandated; those studies (CAPTURE, CAPTURE 2, Emboshield and Xact Post-Approval Carotid Stent Trial [EXACT], Carotid Artery Stenting With Emboli Protection Surveillance Study [CASES]) enrolled >8000 patients by August 2006. As a result, a great deal of experience, previously not available, was obtained by operators outside of CREST, roughly coincident with the addition of asymptomatic patients to CREST enrollment. In addition, the rates of 30-day DSMI in both FDA device approval and postmarket studies were observed to be declining at a significant pace.With the non-CREST CAS case load increasing in the United States and the documented rates of complications associated with CAS decreasing, it was reasonable to query the CREST outcomes to assess any in-trial effects of these external factors; an analysis by year was therefore conducted. Given the slow pace of enrollment early in the trial, the first 4 years were grouped together in a single category (ie, 2000–2004), which led to 5 relatively balanced (numerically) groups: 2000 to 2004 (n=160), 2005 (n=201), 2006 (n=308), 2007 (n=298), and 2008 (n=164). A post hoc analysis of the composite of death and any stroke showed an initial increase in event rates from 4.4% in the first period (2000–2004) to 7.0% in 2005, then a steady decline thereafter to 4.6% in 2006, 3.4% in 2007, and 1.8% in 2008 (Figure 2A). The initial bump in event rates in 2005 may have been related to the sharp increase in new sites and CAS operators added to an already highly experienced initial group. The subsequent decline in rates may be the result of experience acquired by these newer operators (largely outside of CREST) or to the influence of more appropriate case selection, which for CAS was still evolving during the decade of CREST.Download figureDownload PowerPointFigure 2. Death and stroke rates over the period of enrollment in the Carotid Revascularization Endarterectomy versus Stenting Trial for patients undergoing carotid artery stenting. Death and stroke rates for all patients (A), death and major stroke rates for all patients (B), death and stroke rates for the symptomatic subset (C), and death and major stroke for the symptomatic subset (D).The symptomatic subgroup was likewise subjected to this post hoc analysis to determine whether the temporal effect on the composite of death and any stroke over time could be observed independent of the potential impact of asymptomatic patients on lowering event rates (Figure 2B). This analysis yielded different event rates but a similar trend; after an initial increase in event rates from 4.4% in the first period to 9.0% in 2005, there was a steady decline from 8.5% in 2006 to 4.2% in 2007 and finally to 2.6% in 2008. These findings confirm that the decline in event rates observed between 2006 and 2008 in CREST was not caused by the addition of asymptomatic patients in 2005 but was more likely related to the overall gain of experience in CAS both within and outside the trial after device approval in this country.Another post hoc analysis conducted to determine the change in rate of the composite of death or major stroke yielded similar results. After an initial plateau in event rates at 2.5% in the first period and in 2005, there was a steady decline to 0.7% in 2006, 0% in 2007, and 0.6% in 2008 (Figure 2C). In the symptomatic subgroup, the effect was even more pronounced (Figure 2D). The initial increase in event rates from 2.5% in the first period to 3.6% in 2005 was followed by a sharp decline to 0.8% in 2006 and 0% in 2007 and 2008. Of note, there was not a single death or major stroke in symptomatic patients undergoing CAS in the second half of CREST. No change in event rates over time was observed in a similar temporal analysis of CEA outcomes. A comparison of baseline demographics between CAS and CEA in the second half of CREST did not show any major imbalance between arms. The decreasing rates of stroke and death for CAS over time and the similar rates of death and major stroke for CAS and CEA in the second half of the study were among the key findings that led to a conclusion statement by Abbott Vascular that CAS and CEA had balanced outcomes.Other Prespecified Secondary End PointsOutcomes by Symptomatic Status and Age According to Octogenarian StatusAlthough the primary composite end-point rates were higher in symptomatic and octogenarian patients than in their asymptomatic and nonoctogenarian counterparts, there was no evidence of a statistically significant difference between CEA and CAS in either subgroup, as depicted in Figure 3. Two noninferiority hypotheses were also prespecified in the 2005 binding agreement, 1 for symptomatic patients, and 1 for asymptomatic patients. In addition, a noninferiority hypothesis for nonoctogenarian patients was added to the Statistical and Analytic Plan, well in advance of data unblinding. For each of these subgroups (symptomatic, asymptomatic, and nonoctogenarian), CAS demonstrated noninferiority to CEA in outcome rates for the composite end point of periprocedural DSMI plus ipsilateral stroke to 1 year.Download figureDownload PowerPointFigure 3. Primary composite end point in the Carotid Revascularization Endarterectomy versus Stenting Trial by symptomatic or octogenarian status at 1 year. CAS indicates carotid artery stenting; CEA, carotid endarterectomy; and CL, confidence limits.Cranial Nerve InjuryCranial nerve injury is a complication of CEA that tends to resolve in the first few months after surgery.19,20 The most common cranial nerves affected are V, VII, IX, X, and XII. The symptoms of cranial nerve injury can include lower jaw numbness, lower lip weakness, difficulty swallowing, hoarseness caused by vocal cord paralysis, and difficulty with speech because of tongue deviation. There were no cranial nerve injuries experienced in the CAS arm of CREST in the per-protocol population (online-only Data Supplement Table I). For the CEA arm, there was a 5.3% postoperative incidence of cranial nerve injury, which was reduced to 2.1% at 6 months. Of note, the majority of these (80%) involved a motor deficit.Access Site Complications Requiring TreatmentAccess site complications arising from either CEA or CAS are important secondary outcomes. For CAS, a bleeding complication can result in a retroperitoneal accumulation of blood and is potentially life-threatening. For CEA, bleeding at the surgical site is also potentially serious and may require reoperation to prevent or treat airway obstruction. In CREST, access site complications were significantly more frequent in the CEA group, with ≈3 times the occurrence as with CAS (3.7% versus 1.1%, P=0.0001). Bleeding that required reoperation was >8 times more frequent after CEA (n=17) than after CAS (n=2; online-only Data Supplement Figure II).Long-Term Outcomes and Predictors of MortalityCAS and CEA did not differ in terms of long-term ipsilateral stroke prevention or requirement for target-lesion revascularization over the 4-year period (online-only Data Supplement Figures IIIA and IIIB). All-cause mortality was also comparable in the 2 treatment groups, as shown in the Kaplan-Meier 4-year plot (online-only Data Supplement Figure IIIC).For periprocedural (ie, <30 days of procedure) stroke or MI, the presence of tobacco use, diabetes, ischemic heart disease/congestive heart failure, male sex, and advanced age were all associated with an increased risk of postprocedure death. The results of the Cox regression model for prediction of long-term mortality are shown in online-only Data Supplement Table II. Importantly, the 2 strongest predictors of mortality were stroke (hazard ratio [HR], 2.49; P=0.0011) and MI (HR, 2.14; P=0.0079). These results were consistent with the survival analysis that compared patients with periprocedural stroke or patients with periprocedural MI to patients without either periprocedural stroke or MI, defined as the control group (Figure 4A); however, a survival analysis comparing minor stroke with the 2 other subgroups showed that there was no impact of minor stroke on