Outcomes and predictors of stroke after transcatheter aortic valve replacement in the cerebral protection device era: a single-centre TVT registry analysis

Background/Introduction Stroke remains a feared complication of transcatheter aortic valve replacement (TAVR). Administrative databases have characterized stroke epidemiology, but lack clinical granularity including stroke subtype, precise timing, hemodynamic data, and cerebral protection device (CPD) details. Purpose To characterize the incidence, predictors, and outcomes of stroke/TIA after TAVR using granular clinical data from a single-centre TVT Registry, and to evaluate the real-world impact of Sentinel CPD use. Methods We analysed 1,039 consecutive TAVR patients (January 2021–June 2025) from the TVT/NCDR Registry. Stroke/TIA was identified through standardised follow-up and classified by subtype (ischaemic, haemorrhagic, undetermined, TIA) and timing (≤30 vs >30 days). Univariable and multivariable logistic regression identified predictors. Model discrimination was assessed by C-statistic. Sentinel CPD effect was evaluated after multivariable adjustment. Results Over median 354-day follow-up (746.7 person-years), 17 patients (1.6%) experienced stroke/TIA (incidence 2.28/100 person-years): 9 ischaemic strokes (52.9%), 5 TIAs (29.4%), 2 undetermined (11.8%), and 1 haemorrhagic (5.9%). Sixty-five percent occurred within 30 days, with nearly all ischaemic strokes within 18 days. On multivariable analysis (C-statistic 0.741), atrial fibrillation was the only independent predictor (aOR 2.86, 95% CI 1.03–7.93, p=0.043). Univariable signals included STS-PROM (OR 1.07/1%, p=0.012) and low stroke volume index <35 mL/m² (OR 2.26, p=0.043). Sentinel CPD (used in 17.9%) showed no stroke reduction (1.9% vs 1.9%, aOR 1.12, p=0.865). Stroke/TIA patients had markedly higher readmission rates (64.7% vs 27.1%, p<0.001) and numerically higher mortality (17.6% vs 7.8%, p=0.148). Conclusion(s) In this single-centre analysis with granular clinical data, stroke/TIA after TAVR was predominantly ischaemic and periprocedural. Atrial fibrillation was the sole independent predictor. Sentinel CPD showed no real-world stroke reduction. These findings reinforce AF management as the primary target for post-TAVR stroke prevention.