Absence of Consistent Sex Differences in Outcomes From Symptomatic Carotid Endarterectomy and Stenting Randomized
Virginia J Howard1, Ale Algra2, George Howard3
1Department of Epidemiology (V.J.H.), School of Public Health, University of Alabama at Birmingham.
Background And Purpose:
CREST (Carotid Revascularization Endarterectomy Versus Stenting Trial) reported a higher periprocedural risk for any stroke, death, or myocardial infarction for women randomized to carotid artery stenting (CAS) compared with women randomized to carotid endarterectomy (CEA). No difference in risk by treatment was detected for women relative to men in the 4-year primary outcome. We aimed to conduct a pooled analysis among symptomatic patients in large randomized trials to provide more precise estimates of sex differences in the CAS-to-CEA risk for any stroke or death during the 120-day periprocedural period and ipsilateral stroke thereafter.
Methods:
Data from the Carotid Stenosis Trialists' Collaboration included outcomes from symptomatic patients in EVA-3S (Endarterectomy Versus Angioplasty in Patients With Symptomatic Severe Carotid Stenosis), SPACE (Stent-Protected Angioplasty Versus Carotid Endarterectomy in Symptomatic Patients), ICSS (International Carotid Stenting Study), and CREST. The primary outcome was any stroke or death within 120 days after randomization and ipsilateral stroke thereafter. Event rates and relative risks were estimated using Poisson regression; effect modification by sex was assessed with a sex-by-treatment-by-trial interaction term, with significant interaction defined a priori as P≤0.10.
Results:
Over a median 2.7 years of follow-up, 433 outcomes occurred in 3317 men and 1437 women. The CAS-to-CEA relative risk of the primary outcome was significantly lower for women compared with men in 1 trial, nominally lower in another, and nominally higher in the other two. The sex-by-treatment-by-trial interaction term was significant (P=0.065), indicating heterogeneity among trials. Contributors to this heterogeneity are primarily differences in periprocedural period. When the trials are nevertheless pooled, there were no significant sex differences in risk in any follow-up period.
Conclusions:
There were significant differences between trials in the magnitude of sex differences in treatment effect (CAS-to-CEA relative risk), indicating pooling data from these trials to estimate sex differences might not be valid. Whether sex is acting as an effect modifier of the CAS-to-CEA treatment effect in symptomatic patients remains uncertain. Registration: URL: https://www.clinicaltrials.gov; Unique identifier: NCT00190398 (EVA-3S) and NCT00004732 (CREST). URL: https://www.isrctn.com; Unique identifier: ISRCTN57874028 (SPACE) and ISRCTN25337470 (ICSS).
