Prediction of Left Atrial Appendage occluder size based on 3D printed models

Abstract

Abstract Funding Acknowledgements Type of funding sources: None. Background occlusion of the left atrial appendage (LAA) is an established method in the prevention of thromboembolic complications in patients with atrial fibrillation and high bleeding risk. There is a huge anatomical variability of LAA and choosing the right type and size of the occluder may be challenging and result in a time-consuming procedure with repeated repositioning of the selected device, or a need to use another device. These aspects may lead to a prolonged or more aggressive manipulation resulting in periprocedural complications and a suboptimal result. The current technology of 3D printing allows us to obtain a real model of the LAA and to simulate the course of implantation. Purpose in a post-hoc analysis to compare occluder sizes predicted as optimal by simulation on a 3D-printed LAA model to actually implanted LAA occluders, and analyse potential benefits of such simulations. Methods We analysed 32 consecutive patients after a TEE-guided implantation of the Amulet occluder in whom a CT cardiac scan had been performed prior to the procedure. Based on the CT, an LAA 3D model was printed out post-hoc for each patient. Two experienced operators blinded to the procedural results simulated the implantation ex-vivo and agreed on the most appropriate size of the occluder to fit in the LAA. If the simulation-based size matched the size of the primary operator`s choice and this occluder was actually implanted („intention-to-implant and implanted"), the match was considered positive; otherwise it was negative. The number of positive and negative matches and corresponding procedural characteristics were analysed. Results The positive match was found in 9 (28.1%) cases. Out of 23 (71.9%) negative matches, there were 11 (47.8%) implantations using a bigger-than-simulated and 12 (52.2%) smaller-than-simulated occluder; in 2 (8.7%) cases the actually implanted occluder matched the simulated size but was a 2nd -choice occluder. In positive-match cases compared to negative-match cases, there was a statistically significant less need for an occluder reposition (median 0.0 vs 1.0, p = 0,02), and a trend to a shorter procedural time (average 66.7+ 13.3 min vs 85.2 + 32.8 min, p = 0.112) and less contrast agent consumption (average 188.9 + 67.9ml vs 220.9 + 112.4ml, p = 0.43). Conclusion The preliminary results of our study suggest that a CT-based 3D modelling of the LAA shape and simulating the LAA occlusion procedure ex-vivo have a potential to become useful tools to reduce procedural time and complications. Our results might serve as a rational for a prospective study. Abstract Figure. Occluder in the 3D model