Long-term outcomes of stenting and endarterectomy for symptomatic carotid stenosis: a preplanned pooled analysis of individual patient data

Background The risk of periprocedural stroke or death is higher after carotid artery stenting (CAS) than carotid endarterectomy (CEA) for the treatment of symptomatic carotid stenosis. However, long-term outcomes have not been sufficiently assessed. We sought to combine individual patient-level data from the four major randomised controlled trials of CAS versus CEA for the treatment of symptomatic carotid stenosis to assess long-term outcomes. Methods We did a pooled analysis of individual patient-level data, acquired from the four largest randomised controlled trials assessing the relative efficacy of CAS and CEA for treatment of symptomatic carotid stenosis (Endarterectomy versus Angioplasty in Patients with Symptomatic Severe Carotid Stenosis trial, Stent-Protected Percutaneous Angioplasty of the Carotid Artery versus Endarterectomy trial, International Carotid Stenting Study, and Carotid Revascularization Endarterectomy versus Stenting Trial). The risk of ipsilateral stroke was assessed between 121 days and 1, 3, 5, 7, 9, and 10 years after randomisation. The primary outcome was the composite risk of stroke or death within 120 days after randomisation (periprocedural risk) or subsequent ipsilateral stroke up to 10 years after randomisation (postprocedural risk). Analyses were intention-to-treat, with the risk of events calculated using Kaplan-Meier methods and Cox proportional hazards analysis with adjustment for trial. Findings In the four trials included, 4775 patients were randomly assigned, of whom a total of 4754 (99·6%) patients were followed up for a maximum of 12·4 years. 21 (0·4%) patients immediately withdrew consent after randomisation and were excluded. Median length of follow-up across the studies ranged from 2·0 to 6·9 years. 129 periprocedural and 55 postprocedural outcome events occurred in patients allocated CEA, and 206 and 57 for those allocated CAS. After the periprocedural period, the annual rates of ipsilateral stroke per person-year were similar for the two treatments: 0·60% (95% CI 0·46–0·79) for CEA and 0·64% (0·49–0·83) for CAS. Nonetheless, the periprocedural and postprocedural risks combined favoured CEA, with treatment differences at 1, 3, 5, 7, and 9 years all ranging between 2·8% (1·1–4·4) and 4·1% (2·0–6·3). Interpretation Outcomes in the postprocedural period after CAS and CEA were similar, suggesting robust clinical durability for both treatments. Although long-term outcomes (periprocedural and postprocedural risks combined) continue to favour CEA, the similarity of the postprocedural rates suggest that improvements in the periprocedural safety of CAS could provide similar outcomes of the two procedures in the future.