Multi-modality comparison of volumes and ejection fraction by echocardiography, cardiac magnetic resonance and cardiac computed tomography in various left ventricular geometries

Abstract

Abstract Funding Acknowledgements Type of funding sources: Public hospital(s). Main funding source(s): The research fund of the Heart Center, Rigshospitalet, Denmark Background Assessment of left ventricular (LV) volumes and function is crucial in managing patients. New imaging modalities are becoming more common. It is therefore important to compare them with the standard echocardiographic method that most treatments rely on and to determine if they are suitable for all LV geometries. Purpose The purpose was to compare end-diastolic volume (EDV), end-systolic volume (ESV) and LV ejection fraction (LVEF) for the three most common imaging modalities; echocardiography, cardiac magnetic resonance (CMR) and cardiac computed tomography (CCT). Methods We included 85 subjects with various LV geometries; no cardiac disease (n = 44) and various cardiac disorders (n = 41). Cardiac assessment was performed using echocardiography followed immediately by CMR; re-examination after median 6 days, interquartile range 3-18 days using echocardiography followed immediately by CCT. We compared EDV, ESV and LVEF by three-dimensional echocardiography (echo-3D), CMR and CCT to echocardiographic biplane method of discs (echo-BP). The population was divided in four LV geometry profiles (normal, dilatation, hypertrophy, dilatation and hypertrophy) according to gender, age and indexed CMR-values of EDV and LV mass. We calculated inter-modality-ratios by dividing the values from echo-3D, CMR and CCT with echo-BP, to evaluate variances between the LV geometries. Results The figure demonstrates the agreement to echo-BP divided by geometry. Echo-3D had overall best agreement to EDV, ESV and LVEF. CMR overestimated both EDV and ESV. CCT overestimated EDV but not ESV. CCT overestimated LVEF by 4-16% in absolute values, whereas CMR and echo-3D had better agreement for LVEF. The correlation between echo-BP and echo-3D, CMR, and CCT, respectively was; EDV 0.91, 0.94, 0.90, ESV 0.86, 0.86, 0.79, and LVEF 0.40, 0.46, 0.38, all p < 0.001. CMR especially overestimated EDV and ESV in "hypertrophy and dilatation" whereas CCT especially underestimated EDV and ESV in solely "hypertrophy", with larger overestimation of LVEF. ANOVA-analysis of inter-modality-ratios between LV geometries indicated significant variation for EDV but not ESV by echo-3D (F = 2.9, p < 0.05 and F = 1.6, NS), no significant variation for EDV or ESV by CMR (F = 0.01 and 2.4, both NS), and significant variation for both EDV and ESV by CCT (F = 5.4, p < 0.01 and 7.2, p < 0.001). No significant variation for LVEF by echo-3D (F = 1.0, NS), but significant variation for CMR and CCT (CMR: F = 4.5, p < 0.01 and CCT: F = 8.6, p < 0.001) with slightly higher variation for CCT. Conclusions Echo-3D had the overall best agreement of volumes and LVEF, compared to echo-BP as a reference. CMR overestimated EDV and ESV whereas CCT overestimated EDV but not ESV, resulting in overestimation of LVEF by CCT but not CMR. In hypertrophic non-dilated LVs; CCT underestimated both EDV and especially ESV, with larger overestimation of LVEF. In general, CMR appears to be less dependent on LV geometry compared to echo-3D and CCT. Abstract Figure.