Prediction of future redo-TAVI feasibility by preoperative computed tomography

Background Future redo transcatheter aortic valve implantation (redo-TAVI) has gained increasing importance for younger patients with longer life expectancy following initial TAVI procedures. Purpose This study aimed to evaluate the predictive capability of pre-operative computed tomography (CT) in determining the feasibility of redo-TAVI. Methods We retrospectively reviewed 115 consecutive patients who underwent TAVI with a balloon-expandable valve (BEV) and post-operative contrast-enhanced CT between November 2019 and December 2024. All procedures were performed using the lucent line technique. Pre-operative CT images were analyzed to predict the feasibility of future redo-TAVI, with cases classified as "feasible" if: (1) the lowest coronary ostia (coronary risk plane [CRP]) was located above the virtual BEV commissures as indicated by the top of the tabs (neo-skirt plane [NSP]), or (2) when the NSP was above the CRP, virtual valve-to-aorta (VTA) distance measured in the zone from the NSP to the CRP was >2 mm. VTA measurements included distances between the virtual valve and the sinus of Valsalva or sino-tubular junction. The virtual BEV size was based on the actual implanted size for each patient. Feasibility was predicted at three virtual BEV implantation depths. Post-operative CT was used to confirm the true feasibility of redo-TAVI. Results The mean age of patients was 82.8 ± 4.2 years, with 68.7% female. The 23-mm BEV was used in 62 patients (53.9%), and the 26-mm BEV in 35 patients (30.4%). Post-operative CT showed that 73 patients (63.5%) were feasible for redo-TAVI, while pre-operative CT predicted feasibility in 37.4% of cases at a very high implant depth (100:0), 65.2% at a high implant depth (90:0), and 88.7% at a regular implant depth (80:0) as shown in Figure. The accuracy of predictions was 67.0%, 73.9%, and 69.6%, respectively (P=0.523). Among the pre-operative CT factors, only greater oversizing of the BEV relative to the native annulus was significantly associated with unfeasible redo-TAVI on post-operative CT, despite feasible redo-TAVI prediction by pre-operative CT (P=0.042). Conclusion Pre-operative CT can predict redo-TAVI feasibility with an approximate error of 30%, regardless of the virtual implantation depth. A higher BEV oversizing may lead to under-expansion, causing insufficient shortening and a higher NSP on post-operative CT, which may impact redo-TAVI feasibility.