Predictors of Paravalvular Leakage After Transcatheter Aortic Valve Replacement in Patients With BAV

Figures

• 

  Figure 1. Flowchart of the Bicuspid Aortic Valve Population AS = aortic stenosis; BAV = bicuspid aortic valve; PVL = paravalvular leakage; TAVR = transcatheter aortic valve replacement; THV = transcatheter heart valve.

• 

  Figure 2. Definitions of Bicuspid Aortic Valve Classifications and Measured Parameters in This Study (A) Bicuspid aortic valve classification. (B) Calcification volume was measured in the supra-annular, annular, and LVOT planes, respectively. (C) Leaflets, raphe, and annular surroundings of the calcification volume were measured, respectively. (D) Perimeter was measured in the supra-annular, annular, and left ventricular outflow tract planes, respectively. CV = calcification volume; LVOT = left ventricular outflow tract; P = perimeter; other abbreviation as in Figure 1.

• 

  Figure 3. Random Forest Model Performance (A) The calculated flowchart of random forest. When the 7 top-ranking risk factors were included (the dotted line), the chart shows that the specificity and sensitivity of the model represented the best result. (B) Receiver-operating characteristic curves for the random forest and logistic regression models. (C) The decision curve analysis of the random forest model. The red curve represents the predicted performance of the random model. The black line represents the hypothesis that all patients have paravalvular leakage, and the gray line represents the hypothesis that no paravalvular leakage occurs. ACC = accuracy; AUC = area under the curve; LR = logistic regression; RF = random forest.

• 

  Figure 4. Clinical Applicability and Net Benefit of the Logistic Regression Analysis (A) The decision curve analysis of the logistic regression model. The red curve represents the predicted performance of the variables screened by the logistic regression analysis, and the blue curve represents the predicted performance of the remaining variables. The gray line represents the hypothesis that all patients have paravalvular leakage, and the black line represents the hypothesis that no paravalvular leakage occurs. (B) The clinical impact curve of the logistic regression model. The red curve represents the number of patients classified as positive (high risk) by the random forest model at each threshold probability, and the blue curve is the number of true positives at each threshold probability.

• 

  Central Illustration. Predictors of Paravalvular Leakage After Transcatheter Aortic Valve Replacement in Patients With BAV: A Machine Learning Model The predictors were screened using the random forest algorithm. Seven predictors were selected according to their relevance ranking, including Corr-CVleaflet, Corr-CVsurrounding, Corr-CVLVOT, PLVOT/Psuper-annulus, horizontal aorta, annular ellipticity, and postdilation. The yellow lines represent the left ventricular outflow tract and the supra-annular plane, and the red lines represent the long and short diameters of the annulus. These 7 predictors collectively form a comprehensive framework for identifying the risk of paravalvular leakage after transcatheter aortic valve replacement in patients with bicuspid aortic valve, providing crucial guidance for preprocedural planning and risk stratification. Corr-CV = corrected calcification volume; other abbreviations as in Figures 1 and 2.