In CHD with RV volumetric or barometric overload, LV shape is altered. Little is known about the accuracy in this population of left ventricular (LV) volumes and ejection fraction (EF) measurement by standard echocardiographic methods using geometrical assumption: Teicholz, biplane Simpson and real-time 3D echocardiography (RT-3DE). Three-dimensional knowledge-based reconstruction (3D-KR) derived from two-dimensional echocardiographic imaging is a novel technique that has no geometrical assumption. The aim of our study is to assess the accuracy of several echocardiographic methods for measuring LV volumes and EF compared to cardiac magnetic resonance (CMR) measurements (gold standard) in this population.

68 patients (mean age 13,3±4years) with Congenital Heart Diseases (CHD) involving the RV and referred for cardiac MRI, were included. Among them, 13 patients had barometric overload, 37 patients had volumetric overload, and 18 patients had mixed overload. Echocardiographic images acquisition was performed using a standard ultrasound scanner linked to a Ventripoint Medical Systems unit. Analyzed parameters were end-diastolic volume (EDV), end-systolic volume (ESV), and LVEF measured by Teicholz, Simpson's modified formula and RT-3DE (GE Vingmed QVG Auto-4D). The method of disks was used for CMR LV volumes. Intra-observer, inter-observer, and inter-technique variability was assessed using intraclass correlation coefficient (CC), Pearson's CC, coefficients of variation, and Bland-Altman analysis.

Feasibility was 98% for Teicholz Method and 3D-KR, 71% for Simpson's modified formula, and 73% for RT-3DE. We found globally poor correlation and agreement for volumes and EF between the different methods and MRI. The agreement was better when RV volumes were lower than 120mL/m² (and thus, LV shape was less altered).

LV volumes and EFs cannot be measured accurately by echocardiography in patients with dilated right ventricles.