Incremental prognostic value of intensity-weighted regional calcification scoring using contrast CT imaging in TAVR

Abstract Aims Aortic valve calcification scoring plays an important role in predicting outcomes of transcatheter aortic valve replacement (TAVR). However, the impact of relative calcific density and its causal effect on peri-procedural complications due to sub-optimal valve expansion remains limited. This study aims to investigate the prognostic power of quantifying regional calcification in the device landing zone in the context of peri-procedural events and post-procedural complications based on pre-operative contrast computed tomography angiography (CCTA) images. Assess the effect of calcification on post-procedural device expansion and final configuration. Methods and results We introduce a novel patient invariant topographic scheme for quantifying the location and relative density of landing zone calcification. The calcification was detected on CCTA images based on a recently developed method using automatic minimization of the false positive rate between aortic lumen and calcific segments. Multinomial logistic regression model evaluation and ROC curve analysis showed excellent classification power for predicting paravalvular leakage [area under the curve (AUC) = 0.8; P < 0.001] and balloon pre-dilation (AUC = 0.907; P < 0.001). The model exhibited an acceptable classification ability for left bundle branch block (AUC = 0.748; P < 0.001) and balloon post-dilation (AUC = 0.75; P < 0.001). Notably, all evaluated models were significantly superior to alternative models that did not include intensity-weighted regional volume scoring. Conclusions TAVR planning based on contrast computed tomography images can benefit from detailed location, quantity, and density contribution of calcific deposits in the device landing zone. Those parameters could be employed to stratify patients who need a more personalized approach during TAVR planning, predict peri-procedural complications, and indicate patients for follow-up monitoring.