Real-time 3-dimensional echocardiographic quantification of left ventricular volumes. Multicenter study for validation with magnetic resonance imaging and investigation of sources of error

Objectives: We sought to study: 1) the accuracy and reproducibility of real-time 3-dimensional echocardiographic (RT3DE) analysis of left ventricular (LV) volumes in a multicenter setting, 2) interinstitutional differences in relationship with the investigators' specific experience, and 3) potential sources of volume underestimation. Background: Reproducibility and accuracy of RT3DE evaluation of LV volumes has not been validated in multicenter studies, and LV volumes have been reported to be underestimated compared to cardiac magnetic resonance (CMR) standard. Methods: A total of 92 patients with a wide range of ejection fractions underwent CMR and RT3DE imaging at 4 different institutions. Images were analyzed to obtain LV end-systolic volume (ESV) and end-diastolic volume (EDV). Reproducibility was assessed using repeated analyses. The investigation of potential sources of error included: phantom imaging, intermodality analysis-related differences, and differences in LV boundary identification, such as inclusion of endocardial trabeculae and mitral valve plane in the LV volume. Results: The RT3DE-derived LV volumes correlated highly with CMR values (EDV: r = 0.91; ESV: r = 0.93), but were 26% and 29% lower consistently across institutions, with the magnitude of the bias being inversely related to the level of experience. The RT3DE measurements were less reproducible (4% to 13%) than CMR measurements (4% to 7%). Minimal changes in endocardial surface position (1 mm) resulted in significant differences in measured volumes (11%). Exclusion of trabeculae and mitral valve plane from the CMR reference eliminated the intermodality bias. Conclusions: The RT3DE-derived LV volumes are underestimated in most patients because RT3DE imaging cannot differentiate between the myocardium and trabeculae. To minimize this difference, tracing the endocardium to include trabeculae in the LV cavity is recommended. With the understanding of these intermodality differences, RT3DE quantification of LV volume is a reliable tool that provides clinically useful information.