An echocardiographic model predicting severity of aortic regurgitation in congenital heart disease

Abstract

Background- Multiple echocardiographic parameters have been identified to predict the severity of aortic regurgitation (AR) with variable reliability. This study was performed to identify which echocardiographic parameters best predict the severity of AR in a cohort of patients with congenital heart disease, using cardiovascular MRI quantification as a reference standard. Methods and Results- The study involved 2 phases. In phase 1, predictive models were developed on the basis of multivariable analysis of various morphometric and Doppler variables obtained from 174 echocardiograms that best predicted the severity of AR as defined by paired cardiovascular MRI examinations. A nonlinear estimate of regurgitation fraction, using the variables parasternal vena contracta-derived area divided by body surface area and abdominal aorta Doppler retrograde velocity-time integral divided by antegrade velocity-time integral, was identified through multivariable analysis as the best predictive model for AR fraction. In phase 2, the predictive models were prospectively tested on 43 echocardiographic examinations for which a paired cardiovascular MRI was performed. The agreement between the observed and predicted AR fraction was assessed using Bland-Altman analysis. For the 30 studies of the validation data set that had adequate quality images of both the parasternal vena contracta width and the abdominal aorta flow profile, the predicted AR values had a mean bias±SD of 0.4±7.3% (P=0.80). Conclusions- A model using the 2 variables parasternal vena contracta-derived area divided by body surface area and abdominal aorta Doppler retrograde velocity-time integral divided by antegrade velocity-time integral can predict AR severity in patients with a wide variety of congenital heart disease.