112 Reproducibility of left atrial assessment using cardiac magnetic resonance imaging

Abstract

<h3>Background</h3> Left atrial (LA) volume and strain are recognised as predictors of adverse outcomes in cardiovascular disease. Test-retest reproducibility assesses the whole of the data acquisition and analysis procedure, and good test-retest reproducibility of a technique is vital when monitoring treatment effect or disease progression in longitudinal studies. The test-retest reproducibility of LA strain has not yet been reported in any patient groups. We aimed to determine the test-retest reproducibility and inter-/intra-observer variability of LA assessment derived from cardiac MRI in subjects with and without cardiovascular disease. <h3>Methods</h3> 38 participants had two MRI scans a week apart, consisting of: aortic stenosis (n=16); type 2 diabetes (n=16) and healthy volunteers (n=6). Images were analysed by a single trained observer using Medis v3.1, medical imaging system, Leiden, Netherlands. LA strain and strain rate were assessed using Feature Tracking (QStrain v2.0), corresponding to LA reservoir, conduit, and contractile function. LA emptying fraction (LAEF) was calculated using biplane area-length method (QMass v8.1). Both were assessed on 4- and 2-chamber long-axis standard steady-state free precession cine images, and average values calculated. Intra- and inter-observer variability were assessed on 10 randomly selected participants. <h3>Results</h3> As shown in table 1 and figure 1, the test-retest reproducibility for strain and strain rate was moderate for reservoir phase (CoV 25-28%, ICC ∼0.70) and poor for conduit and contraction phases (CoV&gt;40%). The test-retest repeatability for LA volumes and LAEF was good (CoV 16-29%, ICC ≥0.70). The inter- (Table 2) and intra-observer variability of LA volumes and EF were superior to that of strain parameters. Moderate correlation was present between strain at reservoir phase and LAEF (r= 0.58 p-value &lt;0.001, Figure-2). <h3>Conclusion</h3> LA strain assessment, using feature tracking, is a poorly reproducible technique on cardiac MRI, and call into question its utility in monitoring disease progression in longitudinal studies. However, LA volumes and EF have good test-retest reproducibility. <h3>Conflict of Interest</h3> No Conflict of interest