One dimensional M-mode echocardiography is the method most commonly used for assessing in vivo cardiac function in small animals. Recent advances in ultrasound technology permit acquisition of high quality, temporally and spatially resolved 2D images even in mice. Despite this, most studies still report cardiac morphology measured using M-mode. Here, we determined the echo acquisition method that provides the most accurate, reproducible and sensitive measures of mouse heart morphology and function.
Accuracy of geometric equations frequently employed to calculate left ventricular (LV) volumes from echo acquisitions (Simpsons biplane & multi-slice, area length, hemisphere cylinder, cubed & Teichholz formula) was tested by comparison with 3D virtual casts of excised infarcted hearts (n = 11), constructed using ex vivo MRI (TE/TR = 1.8/15 ms, voxel size 31 x 31 x 62 μm). From appropriate slices and orientations, diameters and areas needed for calculating LV volume were measured and compared with the true volume. Linear regression and Bland Altman analyses showed that more complex equations (particularly Simpsons 4 slice - error 7%, R2 = 0.99, P < 0.0001) were more accurate than simple 1D methods (Teichholz - error 27%, R2 = 0.47, P = 0.04).
In vivo comparison of 1D and 2D echo with "gold standard" cardiac MRI (stack of true short axis 1 mm thick cineFLASH images, TE/TR = 1.1/5 ms) performed on the same 16 mice, demonstrated the greater accuracy of 2D echo with Simpsons 4 slice analysis (end diastolic volume; error 18%, R2 = 0.58, P < 0.001) compared to M-mode echo with Teichholz analysis (error 28%, R2 = 0.09, P = 0.25). Reproducibility of the echo methods, assessed by repeated measures on the same 16 mice, illustrated the greater precision of the 2D methods (Simpsons 4 slice = 5% error) compared with M-mode (Teichholz = 13% error). The increased sensitivity of 2D echo to alterations in LV function facilitated identification of a 20 ± 7% increase in end systolic volume and a 5 ± 2% decrease in ejection fraction 2 weeks after trans aortic constriction (n = 17, P < 0.05) which was missed when M-mode was used (ΔESV -3 ± 10%, ΔEF 0 ± 5%).
These results indicate that, as in the clinical setting, 2D echo should supersede 1D M-mode for analysis of mice, as it measures heart morphology more accurately and precisely.