Abstract 11970: True Prediction Model f Clinical Outcome in Tetralogy of Fallot Using Contraction Fraction From the Single Center Cardiac Magnetic Resonance Outcomes Registry in Tetralogy of Fallot

Abstract

Introduction: There are no large scale true prediction models (as opposed to association) to determine predictors of clinical outcomes in repaired tetralogy of Fallot (rTOF) which is the purpose of this study. In addition, right ventricular (RV) contraction fraction (CF), a novel parameter indexing stroke volume (SV) by mass reflecting the amount of blood volume displaced for a given mass has not been studied. Methods: We retrospectively reviewed data of all rTOF pts undergoing cardiac magnetic resonance (CMR) from 2005-2020. Composite outcome 1 (CO1) included death, transplant, ventricular tachycardia and pacemaker placement & composite outcome 2 (CO2) added cardiovascular hospitalizations (CVH) to CO1. For CF, SV was indexed to BSA 1.4 & RV mass was indexed to height 2.7 . An elastic net was utilized to determine variables of significance to enter a Best Subsets logistic regression model. Data was split into training testing sets. Results: 761 rTOF pts were studied with a followup period of 5.1 + 4.2 years. 54 events occurred in total (table). Of all CMR variables considered for CO1, RV CF was the most significant predictor (AUC 0.65) yielding an odds ratio of 0.54 (95% CI 0.34-0.85), P=0.0063. Adding CVH for CO2, both RV CF and left ventricular (LV) mass were significant predictors (AUC 0.74) yielding adjusted odd ratios of 0.60 (95% CI 1.02-1.05), P=0.0063 & 1.03 (95% CI 0.41-0.86), P<0.0001 respectively. RV CF between those who met and didn’t meet CO1 and CO2 endpoints were significantly different (2.0 + 0.76 vs 2.48 + 1.1 cc*cm 2.7 /g*m 1.4 , P=0.006). This was similar to LV mass for CO2 (54.37 + 17.7 vs 69.56 + 26.9 g/m 2 , P=0.015). Conclusion: With this largest study of a true prediction model in rTOF, RV CF is an important novel predictor and should be incorporated in followup of rTOF patients. Adding CVH to CO1 adds LV mass as another predictor. A measure that incorporates all important RV remodeling parameters such as CF should be followed longitudinally to identify patients at risk.