Hye, Robert J.

Background The Carotid Revascularization Endarterectomy versus Stenting Trial (CREST) enrolled 2502 patients between December of 2000 and mid-July of 2008. We analyzed temporal changes in outcomes for both CEA and CAS over the course of this study. Methods Enrollment was divided into 3 consecutive epochs (5 years, 14 months, and 16 months), each with approximately 834 patients ( Table ). Rates for the primary outcome of death, stroke, and myocardial infarction (DSMI) and for death and stroke (DS) during the periprocedural period were calculated. Poisson regression was used to adjust rates for age, sex, dyslipidemia, and symptomatic status, all of which were found to influence outcomes. Results For CAS, there was a 26% decline in DSMI (6.2% → 4.6%) and a 35% decline in DS (5.5% → 3.6%). For both composite endpoints in CEA there was no consistent pattern ( Table ). As CREST progressed, it enrolled younger patients, more men, patients more likely to be dyslipidemic and enrolled more asymptomatic patients (all p < 0.05). Adjustment for these changes mediated the improvement in event rates for the CAS patients, and had no consistent effect on CEA event rates ( Table ). Conclusion Periprocedural safety for CAS improved over time in CREST. Changes were inconsistent for CEA. Improvements for CAS in part appear to reflect changes in patient selection related to age, sex, and risk factors, as adjustment for those variables attenuated the decline in rates. Linking rates to operator experience was not feasible because of the large number of new interventionalists and surgeons entering CREST over time.

Background: Carotid revascularization was addressed by NASCET and ACAS. NASCET showed a 2-year stroke and death of 32.3% in medical patients and a 15.8% incidence in surgical patients (16.5% absolute/51% relative reduction). Similarly, ACAS revealed an 11% 5-year risk in medical patients and 5.1% in surgical patients (5.9% absolute reduction/53% relative reduction). Absolute risk (rather than relative risk) reduction is of prime interest to the patient. However, in these studies, treatment efficacy was assessed on a relative basis, and these procedures were equally efficacious (51% vs. 53% relative risk). The higher event rates in the symptomatic patients resulted in a larger absolute risk reduction in NASCET (16.5% vs. 5.9%). CREST is the only carotid revascularization study to include both symptomatic and asymptomatic patients undergoing CEA or carotid artery stenting (CAS), and we looked at absolute and relative risks associated with both procedures. Methods: Treatment differences in CREST by symptomatic status were reviewed and interaction terms analyzed. Results: Symptomatic periprocedural stroke and death rates were 6% for CAS vs. 3.2% for CEA (2.8% absolute/47% relative difference/p = 0.02), and were 2.5% for CAS vs. 1.4% for CEA (1.2% absolute/53% relative difference/p = 0.15) for asymptomatic. While treatment differences in periprocedural rates were significant for symptomatic but not asymptomatic patients, there was no difference in treatment efficacy as determined by relative risk reduction. Even the most prominent interaction terms were clearly not significant, p > 0.38. Discussion: There is no evidence of a CAS-CEA difference in treatment efficacy by symptomatic status in CREST. There are significant findings in symptomatic patients. This is likely a product of an increased background rate of events in this strata. While symptomatic status can identify a subgroup with an increased background rate, other factors (e.g. age or stenosis) could identify subgroups of asymptomatic patients with high event rates and large absolute treatment differences.