P3693Self-expandable transcatheter aortic valve implantation is associated with frequent periprocedural stroke detected by diffusion-weighted magnetic resonance imaging -Insight from propensity score match

Abstract

Abstract Background Little evidence is available regarding the risk of peri-procedural stroke detected by diffusion-weighted magnetic resonance imaging (DW-MRI) after transcatheter aortic valve implantation (TAVI). Our purpose was to evaluate stroke risk after TAVI using DW-MRI by enrolling consecutive patients who underwent transfemoral TAVI and post-procedural DW-MRI. Methods We prospectively enrolled 113 consecutive patients who underwent transfemoral TAVI and post-procedural DW-MRI. We used balloon-expandable valves as first-line therapy and selected self-expandable valves only for patients with narrow sinotubular junctions or annuli. We set the primary endpoint as the number of high intensity areas (HIA) detected by DW-MRI regardless of the size of the area. To evaluate the risks of the primary endpoint, we employed a multivariable linear regression model, setting the primary endpoint as an objective variable and patient and clinical backgrounds as explanatory variables. In addition, the relationship between valve type and the number of HIAs on DW-MRI was also confirmed by the propensity score matching analysis to evaluate the robustness of the result, using a multivariable linear regression model with the protocol described in the previous manuscript. Shortly, the propensity score was calculated with a logistic regression model by setting the treatment as the response variable and baseline characteristics and procedural information that were significantly different between 2 groups (balloon expandable and self-expandable) as explanatory variables, which included age, estimated glomerular filtration rate, oversizing rate, and BAV before THV deployment. Results Median patient age was 84 years, and 36.3% were men. Ninety-three patients underwent balloon-expandable TAVI and 20 underwent self-expandable TAVI. Symptomatic stroke occurred in 6 (5.3%) whereas asymptomatic stroke occurred in 59 (52.2%) patients. The incidence of symptomatic and total stroke was higher in patients who underwent self-expandable TAVI than those who underwent balloon-expandable TAVI (30.0% vs 0.0%, p<0.001 and 90.0% vs 50.5%, p=0.001, respectively). A multivariable linear regression model demonstrated an increased primary endpoint when self-expandable TAVI was performed (p<0.001). The other covariates had no significant relationship to the primary endpoint. Akaike information criterion-based stepwise statistical model selection revealed that valve type was the only explanatory variable for the best predictive model. This result was also confirmed with the propensity score matching analysis (estimate, 2.359; 95% CI, 0.426–4.292; p=0.019) after adjustments of propensity score, in which 28 patients were matched (n=14 in each group). Conclusions Self-expandable valves were associated with increased numbers of HIA on DW-MRI after TAVI in patients with severe aortic stenosis.