Clinical validation of predictive procedural planning software in a real-world TAVI population

Abstract Background Pre-procedural planning of transcatheter aortic valve implantation (TAVI), including prosthesis sizing and anticipation of possible complications, is crucial to improve procedural success and patient outcome. The General Electric (GE) Healthcare TAVI Analysis software and FEops HEARTguideTM software both allow a morphometric analysis of the aortic root. In addition, FEops HEARTguideTM enables patient-specific prediction of post-procedural complications. Purpose The study aimed (1) to compare morphometric aortic root measurements performed by GE software and FEops HEARTguideTM; and (2) to validate the accuracy of FEops HEARTguideTM in predicting post-TAVI paravalvular aortic regurgitation (AR) and new conduction abnormalities. Methods In 106 consecutive TAVI patients, aortic root dimensions essential for prosthesis sizing were measured using both GE software and FEops HEARTguideTM. FEops HEARTguideTM predictions were compared with the severity of post-TAVI AR as assessed by echocardiography, angiography, invasive hemodynamics, and the development of conduction abnormalities as assessed by serial electrocardiography. Results Bland-Altman analysis revealed smaller annulus perimeter, area and coronary artery ostia heights with FEops HEARTguideTM compared with GE software (Figure 1), leading to differences in prosthesis sizing in 23% of cases. The predicted AR was significantly higher in patients with moderate AR versus none/trace AR on echocardiography (14.3 [5.1 - 45.2] vs. 2.8 [0.4 - 5.2] ml/s, p = 0.006) (Figure 2) and angiography (23.6 [12.7 - 45.8] vs. 4.0 [1.1 - 9.7] ml/s, p = 0.009), though an overlap between adjacent AR grades was observed. The predicted AR did not significantly differ between patients with an AR index < 25 and those with an AR index ≥ 25 (10.2 [3.5 - 21.5] vs. 5.7 [0.2 - 10.7] ml/s, p = 0.079). In contrast, the contact pressure index, which indicates the likelihood of conduction abnormalities, was significantly higher in patients with new conduction abnormalities after TAVI compared with those without (17.5 [14.3 - 20.0] vs. 10.0 [2.0 - 14.0] %, p = 0.006), although overlap was observed at an individual patient level. Conclusions FEops HEARTguideTM generally underestimates aortic root dimensions compared with GE software, affecting prosthesis sizing in a substantial proportion of patients. While FEops HEARTguideTM has the ability to predict the risk of post-TAVI complications at a population level, overlap in predicted parameters is encountered at an individual patient level.