What is bright is not always gold.

To the Editor: The recent paper from Myerson et al1 on the “reliability” of formulae for calculation of echocardiographic left ventricular (LV) mass needs to be carefully considered. Quantitative echocardiography presents problems of reproducibility, but less than expected from initial studies performed without 2D guidance. There is evidence that the test-retest “error” of echocardiography is similar to the “normal” error affecting many biological determinations.2,3 In particular, in the Reliability of M-mode Echocardiographic Studies (RES) trial,2 using M-mode tracings from several echocardiography laboratories of different experience, testretest variability was in the range of 10% to 15% (90% confidence interval). Most error is related to image quality and temporal drift, but, as previously reported,2,3 error can be minimized by both appropriate standardization of the examination and in-laboratory self-learning procedures. Reliability can be high, with a coefficient of test-retest concordance of 87% for identification of prognostically adverse LV hypertrophy (LVH).2 Thus, together with accuracy (proved by necropsy validation4–6), reproducibility of echocardiographic LV mass2,3 is better than pessimistically stated by Myerson et al.1,7 Moreover, Daniels et al5 showed that the geometric formula used for echocardiographic LV mass determination predicts necropsy LV weight accurately from linear LV wall thickness and chamber dimension measurements. Myerson et al1 tried to demonstrate the limited “reliability” of echocardiographic LV mass using a magnetic resonance imaging (MRI) reference standard that had several limitations, including use of a machine with low magnetic field (0.5 Tesla), with low spatial resolution and 1 cm slice thickness, whereas use of higher magnetic fields (1.5 Tesla on average, but up to 3 Tesla) are strongly recommended for cardiac cine-MRI.8 In Myerson’s study,1 measurements were taken “just basal to the papillary muscles,” which, with a 1 cm slice thickness, may not have been standardized exactly at the papillary muscle tips. These pitfalls might be especially aggravated by the intrinsic difficulties of MRI in detecting (and reproducing) the largest LV minor axis, because of great slice thickness and difficulty in aligning acquisitions in near-perfect orthogonal planes with LV structures. These limitations can be offset by echocardiography, using orthogonal images or, especially, long axis with precise anatomic markers. If these limitations are taken into account, the poor result shown in Figure 3 of the paper1 can be explained by higher tool-related technical variability in obtaining a single tomographic slice at the LV minor axis, as compared with using Simpson’s rule. The test-retest variability reported by Myerson et al1 should in fact be compared with analogous tests performed in Prospective Randomized Enalapril Study Evaluating Regression of Ventricular Enlargement (PRESERVE; mostly 2D) and RES (only M-mode). In the RES, for the same 95% confidence interval, test-retest variability was 35 g from echocardiograms performed in 16 different centers, compared with the higher values ( 45 to 55 g) from applying echocardiographic formulae to single-slice MRI data. It should be emphasized also that what is clinically important is the ability to detect prognostically adverse values and significant changes,9 both of which are accomplished by echocardiography.2,3 Technical variability is 15% at the 90% confidence limit, similar to other biological determinations. Of fundamental importance, we have performed a necropsy comparison study that showed no systematic overor underestimation of LV mass,4 and numerous investigators have demonstrated the prognostic relevance of echocardiographic LV mass. Based on results of the PRESERVE study, groups of 41 patients per treatment arm are sufficient to provide statistical power of 90% at an error level of 1% to detect a between-group difference of at least 10 g/m in LV mass. Thus, echocardiographic LV mass remains an excellent bioassay for clinical trials. Although higher magnetic fields and temporal resolution are improving the performance of cardiac MRI, advances in harmonic imaging and in 3D ultrasound are producing parallel improvements in the ability of echocardiography to image the LV and calculate its mass.