Abstract TP190: SHOULD PATIENTS WITH ATRIAL FIBRILLATION BE INCLUDED IN STENT TRIALS: POST HOC ANALYSIS FROM RANDOMIZED CONTROLLED TRIALS

Background: Most trials evaluating the therapeutic benefit of stent placement have excluded patients with atrial fibrillation. We performed this analysis to determine the effect of the inclusion of patients with atrial fibrillation in stent trials. Methods: We performed pooled analysis of patients with vertebral artery stenosis (measuring ≥50% in severity) from two randomized controlled trials evaluating stent placement versus medical treatment alone. We compared the risk of vertebrobasilar arterial distribution ischemic stroke and/or death (and other endpoints) post-randomization in patients with and without atrial fibrillation between the two groups using Cox proportional hazards models and tested the interaction between atrial fibrillation and stent placement in the Cox proportional hazards models. Results: A total of 21 patients (12 randomized to stent placement and 9 to best medical treatment only) among 244 randomized patients (follow up period 37.8 ± 19.7 months) had atrial fibrillation at baseline. Patients with atrial fibrillation were older (mean age 74.0 versus 66.2 years), had a higher proportion of patients with hypertension (76.2% versus 68.2%), and lower proportion of cigarette smokers (4.8% versus 30.0%). Patients with atrial fibrillation were more likely to be on oral anticoagulants (47.6% versus 2.2%). The risks of vertebrobasilar arterial distribution ischemic stroke and/or death (14.3% versus 5.4%, hazard ratio HR 1.4, 95% confidence interval CI 0.4 - 4.8, p= 0.54), any stroke (9.5% versus 9.4%, HR 1.1, 95% confidence interval (CI) 0.3 - 4.7, p= 0.90), ischemic stroke (4.7% versus 9.4%, HR 0.6, 95% CI 0.1 – 4.1, p= 0.56), and any stroke and/or death (23.8% versus 13.9%, HR 1.9, 95% 0.7- 4.9, p= 0.19) were higher in patients with atrial fibrillation. The interaction terms between atrial fibrillation and allocated treatment (best medical treatment versus stent placement) in the Cox proportional hazards models (included age and location of stenosis) were not significant for vertebrobasilar arterial distribution ischemic stroke and/or death or any stroke and/or death. Conclusions: Our analysis identifies challenges in inclusion of patients with atrial fibrillation due to differences in baseline characteristics and associated outcomes. However, the treatment effect of stent placement was not modified by atrial fibrillation suggesting stratified randomization may be acceptable in randomized controlled trials .