2D or Not 2D: Methods for Prediction of LVOT Obstruction Following Transcatheter Mitral Valve Replacement

Transcatheter mitral valve replacement (TMVR) is an emerging yet challenging technique for the treatment of mitral valve disease. Cardiac computed tomography angiography (CCTA) is useful for prediction of complications after TMVR, including risk of left ventricular outflow tract obstruction (LVOTO). LVOTO as predicted by CCTA-modeled neo-left ventricular outflow tract (LVOT) area, is associated with mortality.1Yoon S.H. Bleiziffer S. Latib A. et al.Predictors of left ventricular outflow tract obstruction after transcatheter mitral valve replacement.J Am Coll Cardiol Intv. 2019; 12: 182-193https://doi.org/10.1016/j.jcin.2018.12.001Crossref Scopus (151) Google Scholar Existing literature has been sparse on details of the methodology for neo-LVOT area assessment. The objective of this study is to compare methods for measurement of neo-LVOT area. Patients who underwent TMVR using a transcatheter heart valve (THV) into a failed surgical mitral valve replacement (SMVR) at Saint Francis Hospital and Heart Center between August 2016 and May 2022 were retrospectively enrolled via institutional review board waiver. Cardiac computed tomography angiographies were analyzed using 3mensio software in midsystole. The mitral annulus was reconstructed using semiautomated cubic-spline interpolation, and a SAPIEN 3 THV virtual implant size chosen as recommended by the MViV app. A THV was implanted virtually at 2 positions: (1) atrial alignment of THV and SMVR (ie, ventricular position) and (2) ventricular alignment of THV and SMVR (ie, atrial position). Two centerline paths were used for this study: (1) straight line dividing the LVOT and (2) curved line following the curvature of the interventricular septum. With a perpendicular axis to the centerline at the narrowest neo-LVOT, a measurement was performed at each virtual valve position in a 2 dimensional (D) plane or directly on a 3D volume rendering. The above resulted in 4 different methods for neo-LVOT measurement: (1) SA: straight centerline, THV atrial alignment; (2) SV: straight centerline, THV ventricular alignment; (3) CA: curved centerline, THV atrial alignment; and (4) CV: curved centerline, THV ventricular alignment (Figure 1). Each of these 4 approaches was performed with a 2D and a direct 3D measurement (Figure 1). neo-LVOT area/LVOT area ratios were calculated. The 16 measurements or calculations were compared to each other and to post-TMVR LVOT gradients. Thirty-two patients were enrolled in the present analysis. Mean age was 78 ± 8 years, and 47% were female. The median STS score was 8.7% (5.4-13.9). Baseline echocardiographic characteristics showed mean left ventricular ejection fraction of 51%, mean gradient (MG) across the mitral valve 10.8 mm Hg, and mitral regurgitation grade 3 or 4 in 50% of cases. The 2D planar measurements and their corresponding direct 3D measurements correlated strongly to each other regardless of the position of the virtual valve and the course of the centerline used (r ≥ 0.94; P < .001 for all). Direct 3D measurements resulted in underestimation of the neo-LVOT area as assessed by 2D planar measurements (P < .05). There was a modest and significant correlation between neo-LVOT area, and neo-LVOT/LVOT ratio, respectively, with LVOT systolic MG regardless of the used method, (r = 0.38-0.54; P < .05 for all). Only 2D neo-LVOT/LVOT ratios using a curved centerline with either ventricular or atrial positioned THVs had a correlation coefficient >0.50 with LVOT MG, and direct 3D measures did not perform better than 2D measures. The accurate measurement of neo-LVOT area is crucial for prediction of LVOTO and thus patient anatomical selection. However, to date, there is no standardized method to measure neo-LVOT area. In the most cited reviews on valve-in-valve TMVR, there is no clear indication of how centerlines are oriented in relation to the LVOT nor standardization on THV position.2Reid A. Ben Zekry S. Turaga M. et al.Neo-LVOT and transcatheter mitral valve replacement: expert recommendations.J Am Coll Cardiol Img. 2021; 14: 854-866https://doi.org/10.1016/j.jcmg.2020.09.027Crossref Scopus (39) Google Scholar,3Blanke P. Naoum C. Dvir D. et al.Predicting LVOT obstruction in transcatheter mitral valve implantation: concept of the neo-LVOT.J Am Coll Cardiol Img. 2017; 10: 482-485https://doi.org/10.1016/j.jcmg.2016.01.005Crossref Scopus (193) Google Scholar In the present study, we compared approaches to predict neo-LVOT area by modifying LVOT centerlines and the position of the virtual THV. Our results showed that 2D measurement-based neo-LVOT/LVOT ratios correlate better with metrics of LVOTO than direct 3D measurements or ratios performed on a 3D volume rendering. Although all correlations between our measurements and post-TMVR LVOT MG were significant, r-values >0.50 were only achieved using a curved centerline method without an impact of THV position. The current data apply only to valve-in-valve TMVR cases; given the variability of various TMVR procedures, they are likely not generalizable to the other types. The actual position of the THV was not modeled due to lack of fluoroscopic data; however, given the independence of postprocedural gradients to modeled valve position, it may be less relevant. This small cohort suggests a 2D slice-based neo-LVOT/LVOT ratio using a curved centerline is a superior method for LVOTO assessment. Based on these first-of-their-kind limited data, this method can be used without regard to THV modeling position. Larger data sets should be analyzed to further clarify the methodology for LVOTO prediction.