Echocardiography and magnetic resonance imaging based strain analysis of functional single ventricles: a study of intra- and inter-modality reproducibility.

Abstract

The purpose of this study was to measure reproducibility in strain measurements by echocardiography and cardiac magnetic resonance imaging (CMR) in functional single ventricles in the Fontan circulation. Fontan echocardiograms and CMRs between 1/2005 and 6/2013 were analyzed retrospectively. Patients were included if studies occurred <1 year apart and without interim catheter-based or surgical intervention. Global peak systolic circumferential (GCS) and longitudinal (GLS) strain, and strain rates (GCSR and GLSR) were measured by speckle or feature tracking (echocardiogram vs. CMR) using commercial software. 134 study pairs were identified; 74 with dominant left and 60 with dominant right ventricles. Agreement for strain and strain rate between echocardiogram and CMR was modest for circumferential measurements, concordance correlation coefficient (Rc) = 0.56 for GCS, 0.44 for GCSR; and low for longitudinal measurements, Rc = 0.34 for GLS, and 0.27 for GLSR. There was good inter-observer agreement (n = 30) for strain measurements by echocardiogram and CMR (echocardiogram GCS Rc = 0.84, echocardiogram GLS Rc = 0.79, CMR GCS Rc = 0.94, CMR GLS Rc = 0.75). Inter-observer agreement for strain rate was slightly lower (echocardiogram GCSR Rc = 0.79, echocardiogram GLSR Rc = 0.73, CMR GCSR Rc = 0.93, CMR GLSR Rc = 0.54). Coefficients for intra-observer agreement for these variables were similar to inter-observer agreement. In conclusion, intra-modality reproducibility for strain assessment in functional single ventricles in the Fontan circulation is high, measured both by echocardiography and CMR; inter-modality reproducibility, however, is modest. These findings caution against mixing values from multiple modalities in the long-term assessment of patient health in this population.