A comparison of left ventricular mass between two-dimensional echocardiography, using fundamental and tissue harmonic imaging, and cardiac MRI in patients with hypertension

Purpose: To compare left ventricular mass (LVM) as measured by two-dimensional (21)) echocardiography using two different calculation methods: truncated ellipse (TE) and area length (AL), in both fundamental and tissue harmonic imaging frequencies, to LVM as measured by, the cur-rent gold standard, cardiac magnetic resonance imaging (MRI). Turbo gradient echo (TGE) pulse sequence was utilized for MRI. Materials and methods: Thirty-two subjects with history of hypertension were recruited. The images were acquired, contours were traced and the LVM was calculated for all four different echocardiography methods as well as for the cardiac MRI method. The intra-observer variabilities were calculated. The four different echocardiography methods were compared to cardiac MRI using the method described by Bland and Altman. Results: Twenty-five subjects had adequate paired data sets. The mean LVM as measured by cardiac MRI was 162 +/- 55 g and for the four different echocardiography methods were: fundamental AL 165 +/- 55 g, harmonic AL 168 +/- 53 g, fundamental TE 148 +/- 50 g, harmonic TE 149 +/- 45 g. The intra-observer variability for cardiac MRI method, expressed as bias I standard deviation of the difference (S.D.D.), was 2.3 +/- 9.2 g and for the four different echocardiography methods were: fundamental TE 0.4 +/- 26.8 g, fundamental AL 0.6 +/- 27.0 g, harmonic TE 6.7 +/- 21.8 g, harmonic AL 6.4 +/- 22.9 g. The mean LVM for the AL method was closest to the cardiac MRI technique, while TE underestimated LVM. The 95% limits of agreement were consistently wide for all the 2D echocardiography modalities when compared with the cardiac MRI technique. Conclusion: The intra-observer variability in measurements of 2D echocardiographic LVM, together with the wide limits of agreement when compared to the gold standard (cardiac MRI) are sufficiently large to make serial estimates of LVM, of single patients or small groups of subjects, by 2D echocardiography, unreliable. (C) 2003 Elsevier Ireland Ltd. All rights reserved.