Assessment of Left Atrial Volume in Dogs: comparisons of Two-Dimensional and Real-Time Three-Dimensional Echocardiography with ECG-gated Multidetector Computed Tomography Angiography

Left atrial size (LAS) aids clinical decision-making in left-sided heart disease. Left atrial to-aortic-root ratio on short axis view is commonly used to assess LAS. This unidirectional measurement might not represent changes in other dimensions. Therefore, measurement of LA volume (LAV) is recommended for evaluation of LAS. Two-dimensional methods such as biplane area length method (ALM) and biplane method of disks (MOD) can estimate LAV, assuming the LA to be a fixed shape. Alternatively, real-time three-dimensional echocardiography (RT-3DE) measures LAV without geometric assumptions. In veterinary medicine none of these have been compared with volumetric gold standard like multi-detector computed tomography angiography (MCTA), which correlates well with cardiac magnetic resonance imaging. We hypothesised that RT-3DE estimation of LAV is more accurate than ALM and MOD and correlates more strongly, using ECG-gated MCTA as gold standard. To investigate this hypothesis, we compared maximum and minimum LAV (LAVmax and LAVmin respectively), identified as the frame immediately prior to mitral valve opening and immediatly after mitral valve closure respectively. ALM, MOD and RT-3DE were compared with 64-slice ECG-gated MCTA in 20 client-owned dogs (9.1-44.5kg; 24.9±9.3). Levels of agreement and intra- and inter-observer variability were calculated. This single centre, prospective, cross-sectional study was approved by the local Veterinary Ethical Review Committee. Dogs of various breeds undergoing anaesthetised thoracic MCTA for various diagnostic purposes were enrolled. Eligible dogs underwent a physical examination including blood pressure measurement prior to anaesthesia and cardiac auscultation to ensure that no heart murmur was present. Echocardiography immediately followed ECG-gated MCTA. The agreement between ALM, MOD, RT-3DE and MCTA were evaluated using Bland-Altman analysis to express bias and 95% limits of agreement (95%LoA). Correlations were calculated along with coefficients of variation (CVs) to quantify intra- and inter-observer variability. ALM (r=0.79 and 0.72), MOD(r=0.81 and 0.70) and RT-3DE (r=0.94 and 0.82) correlated with MCTA for LAVmax and LAVmin respectively (all p<0.05). Biases for LAVmax (-0.96mL, (95%LoA: [-5.6–3.7]) and LAVmin (-0.67mL, (95%LoA: [-5.4–4.1]) were minimal with 3D-RTE, reflecting a systematic underestimation. Conversely, MOD (LAVmaxbias=3.19mL, 95% LoA: [-5.7–12.1]; LAVmin=1.96mL, 95%LoA: [-4.6–8.5]) and ALM (LAVmaxbias=4.05, 95%LoA: [-5.7–13.8]; LAVmin=2.80mL, 95%LoA: [-3.9–9.5]) systematically overestimated LAV. Intra- and inter-observer variability for LAVmax and LAVmin were adequate using ALM, MOD, RT-3DE and MCTA (CVs<15%). In conclusion, assessment of LAV with RT-3DE is a simple, non-invasive, accurate and feasible method for measuring LAS that has superior accuracy to ALM and MOD in dogs without cardiac disease. Objectives The primary goal of this study was to compare left atrial (LA) volume and functional indices in a population of normal dogs using transthoracic one-dimensional M-mode, two-dimensional (2D), and three-dimensional (3D) echocardiography. Animals Forty clinically normal dogs of various breeds were stratified into four groups based on weight. Methods Left atrial volume (LAV) was estimated by converting single linear measurements derived from M-mode and 2D images into volumes via the cube and sphere formulas. 2D echocardiography was employed to estimate LAV using linear measures and rectangular and ellipsoid formulas, as well as area-length and the monoplane method of disks. LAV was estimated using the biplane area-length method on 4- and 2-chamber 2D images. LAV was directly measured using 3D volume data obtained over four consecutive cardiac cycles. LAV estimates were compared amongst methods, correlated to body size, and evaluated for reproducibility. Functional indices derived from these various echocardiographic methods were compared to identify any systematic differences based on imaging modality. LA functional indices included LA ejection fraction (LAEF), total LA emptying volume, passive LA emptying volume/fraction, active LA emptying volume/fraction, and LA expansion index. Results The M-mode derived estimates of LAV were significantly smaller than all 2D- and 3D-derived LAV (p<0.0001). The left-sided 3D LAV had the highest correlation with body weight (r2??=??0.8806) for all phases of the LA cycle with acceptable interobserver variability (CV 11%). Conclusions 3D echocardiography is a feasible, non-invasive method to measure LAV in a population of normal dogs.