Methods
Repaired TOF patients with native outflow tracts or RVPA conduits were referred for TPVR at a single institution between 2008-2019 were included in the retrospective analysis. All patients had at least moderate PR by CMR (pulmonary regurgitant fraction >20%). Pre-implantation CMR data was included for volumetric data analysis: EF, end-diastolic volume indexed (EDVI), end-systolic volume indexed (ESVI), and PR. RV mass/volume ratio was also calculated on pre-procedural CMR images available for review. CMR RV cine images were reanalyzed for RV mass/volume ratio using available software (MedisSuite MR, Medis Medical Imaging Systems BV, the Netherlands). Pre-implantation hemodynamic catheterization was included for analysis of degree of RVSP and pulmonic stenosis. RV end-diastolic pressure (RVEDP) was also included for analysis.
Myocardial Strain by Speckle Tracking Echocardiography
Serial echocardiography data was used for analysis: (a) prior to valve implantation, (b) at hospital discharge, and (c) within 18 months post valve implantation. An internal validation scale was used for assessing quality of GLS images: 0 = more than 2 segments not tracking; 1 = less than 1 segment not tracking; 2 = all segments tracking. Patients with poor quality imaging, assigned a 0, were excluded. GLS analysis using STE was obtained by tracing the RV and LV endocardium from the apical 4 chamber view using TomTec Imaging Systems (GmBH, Munich, Germany) (Figure 1). A less negative number is lower strain or abnormal strain magnitude, and a more negative number is higher strain magnitude or normal strain. A cut off of -17 is accepted as normal strain magnitude, previously validated in CMR controls with as the gold standard [12, 19-21].
Statistical Analysis
To account for the multiple strain measurements per patient, the univariable and multivariable associations with RV or LV GLS were modeled using linear mixed models with random intercepts for patients [22, 23]. The confidence intervals for these associations used profile likelihoods. Associations between other metrics, such as RV mass, RVSP pre-catheterization, were computed using linear models and confidence intervals were computed with the classical formula.
The associations between RV GLS and RVSP pre-implantation, LV GLS and RVSP pre- implantation, RV mass and RVSP pre- implantation, RV strain and RV end diastolic pressure (RVEDP), and RV GLS and RV mass, were adjusted for type of outflow. Adjusting for the type of outflow tract did not change the significance of any of the analyses. Other associations, including RV GLS with RVSP post- implantation, LV GLS with RVSP post- implantation, RV GLS with age, LV GLS with age, RV GLS with RV EF, LV GLS with RV EF, and LV GLS with LV end diastolic pressure (LVEDP), were unadjusted.
Additionally, GLS, grouped by right and left, over time were plotted using box plots. RV GLS, grouped by RVSP pre-catheterization, was also plotted over time using boxplots. Similar plots were made for LV GLS grouped by RVSP pre-catheterization. All analysis was conducted in R version 4 [24]. Plots were created using ggplot2 [25].