Portal de Periódicos da CAPES

Regarding “Protected carotid stenting in high-surgical-risk patients: The ARCHeR results”

2007; Elsevier BV; Volume: 45; Issue: 1 Linguagem: Inglês

10.1016/j.jvs.2006.08.089

1097-6809

A.R. Naylor,

Acute Ischemic Stroke Management

Although the authors are to be commended upon being the only one of the 10 or so "high-risk" stent registries to publish their results in a peer-reviewed journal,1Gray W.A. Hopkins L.N. Yadav S. Davis T. Wholey M. Atkinson R. et al.Protected carotid stenting in high-surgical-risk patients: the ARCHeR results.J Vasc Surg. 2006; 44: 258-269Abstract Full Text Full Text PDF PubMed Scopus (367) Google Scholar ARCHeR (as with SAPPHIRE2Yadav J.S. Wholey M.H. Kuntz R.E. Fayad P. Katzen B. Mishkel M.R. et al.SAPPHIRE InvestigatorsProtected carotid artery stenting versus endarterectomy in high-risk patients.N Engl J Med. 2004; 351: 1493-1501Crossref PubMed Scopus (2548) Google Scholar) raises more questions than it answers. Two important issues warrant attention. The first is the statistical appropriateness of performing a noninferiority analysis on nonrandomized data. Noninferiority studies (eg, is angioplasty as good as endarterectomy) are relatively uncommon in contemporary surgical practice, unlike equivalence or superiority studies. Trial methodology requires the investigator to specify a noninferiority margin before the trial starts (in this case, 4%). The two treatment modalities are then (usually) compared in a prospective randomized trial so that variables are matched and there is less chance for unwanted bias. In ARCHeR, historic surgical outcome data were used, and this will inevitably raise concerns about bias and generalizability, as was noted by Bill Mackey in his commentary. Moreover, although the authors apparently defined their noninferiority margin before ARCHeR 1 started recruiting, it is hard to see how this could remain statistically valid, for a number of reasons. First, the "hypothesis (page 259) was constructed statistically to assess whether the 1-year composite primary endpoint (anticipated at 10% for stenting) was within an acceptable margin of the historical control rate (14.4%)." This implies that the historic control of 14.4% was defined from the outset (ie, good trial practice). It is, therefore, hard to reconcile this statement with the subsequent observation (page 260) that "at the completion of the study, the proportion of actual enrollment in these two broad categories was calculated and the final 1-year endpoint comparator was estimated at 14.4%." Second, three sequential studies were undertaken (one of which did not complete enrollment), each with a different power (ARCHeR 1, 80%; ARCHeR 2, 90%; and ARCHeR 3, 85%) and each with a very heterogeneous cohort of symptomatic and asymptomatic patients. To this reader, it seemed inappropriate that the noninferiority margin of 4% was applied equally to both symptomatic and asymptomatic patients. The second area of concern is how the conclusions should be applied to mainstream clinical practice. The Figure presents a reanalysis of how the 30-day operative risk influences stroke prevention. These data were derived from a reanalysis of outcomes from the Carotid Endarterectomy Trialists Collaboration, which combined data from 6000 recently symptomatic patients with 50% to 99% stenoses randomized into the European Carotid Surgey Trial (ECST), the North American Symptomatic Carotid Endarterectomy Trial (NASCET) and the Veterans Administration (VA) Study studies and in whom all the prerandomization angiograms were remeasured by using the NASCET method.3Rothwell P.M. Eliasziw M. Gutnikov Warlow C.P. Barnett H.J.M. Carotid Endarterectomy Trialists CollaborationEndarterectomy for symptomatic carotid stenosis in relation to clinical subgroups and timing of surgery.Lancet. 2004; 363: 915-924Abstract Full Text Full Text PDF PubMed Scopus (1220) Google Scholar Parallel data for asymptomatic patients were derived from a reworking of the Asymptomatic Carotid Atherosclerosis Study results.4Executive Committee for the Asymptomatic Carotid Atherosclerosis StudyEndarterectomy for asymptomatic carotid artery stenosis.JAMA. 1995; 273: 1421-1428Crossref PubMed Scopus (5145) Google Scholar In asymptomatic patients (76% of the ARCHeR cohort), the presence of recurrent laryngeal nerve palsy, unstable angina, and pulmonary disease did not increase the long-term risk of stroke.5Naylor A.R. Golledge J. High risk plaque, high risk patient or high risk procedure?.Eur J Vasc Endovasc Surg. 2006; 32: 557-560Abstract Full Text Full Text PDF PubMed Scopus (13) Google Scholar Accordingly, there is no evidence that morbidity/mortality thresholds should be changed from the recommended 3% level. The 30-day death/stroke risk in asymptomatic patients in ARCHeR was 5.4%—well in excess of accepted guidelines. At this level of risk, only 29 strokes will be prevented at 5 years by performing 1000 angioplasty procedures (ie, there will be 971 unnecessary interventions). Accordingly, the recommendation should probably remain that provided that the procedural risk is 3% or less, angioplasty does have a role in high-risk asymptomatic patients. Conversely, most surgeons would accept that symptomatic patients face a much higher long-term stroke risk (ie, it would be reasonable to increase the threshold in selected high-risk individuals). However, few would thereafter accept that a greater than 11% 30-day risk justifies recommending angioplasty in all symptomatic patients who are otherwise considered high-risk for surgery. ARCHeR concluded that "carotid stenting is a safe, durable and effective alternative in high-surgical risk patients." To qualify as being high-risk, the patient really does have to be symptomatic, and it is debatable whether a procedural risk of more than 11% in symptomatic patients justifies this conclusion. Similarly, with a more than 5% procedural risk in asymptomatic patients, it is hard not to conclude from ARCHeR (as with SAPPHIRE) that few (if any) of these high-risk asymptomatic individuals required any intervention at all (Fig 1). ReplyJournal of Vascular SurgeryVol. 45Issue 1PreviewWe appreciate the commentary provided by Dr Mackey and Professor Naylor on the ARCHeR publication and the opportunity to respond. Full-Text PDF Open Archive