6 Comparison between CMR and a fully automated CT technique in the assessment of epicardial adipose tissue in type 2 diabetes

Introduction Dysregulated epicardial adipose tissue (EAT) may play an important role in the development of heart failure in Type 2 Diabetes (T2D). EAT can be measured using computed tomography (CT) but there are limited data directly comparing this with CMR-measured EAT, which offers unique advantages. This study aimed to assess CT- and CMR-measured EAT in people with T2D. Materials and Methods In this single-centre, prospective study of people with and without T2D, participants underwent phenotyping including an ECG-gated non-contrast CT and a multiparametric stress CMR at 3T. CT total EAT volume was determined using a fully automated deep learning algorithm incorporating the whole heart. CMR EAT area was quantified using a 4-chamber long axis cine image at end-systole obtained using a steady-state free precession technique. Contouring was performed by a single observer using two methods to measure EAT area: 1) Contour around the right ventricle starting from the atrioventricular groove up to the left ventricular apex (RV-EAT), and 2) contour around both ventricles (BV-EAT). Values were indexed to height. Twenty scans underwent repeat assessment by a second observer for inter-observer variability. Results 93 people with T2D (age 64.0±6.5, males 60.2%, HbA1c 7.0 [6.5–7.9]%) and 32 controls (age 59.0±7.5, males 65.6%, HbA1c 5.5 [5.3–5.7]%) were included in this analysis. The T2D group had significantly higher body mass index compared to controls and greater proportion of people with hypertension and hypercholesterolaemia. T2D participants had greater indexed EAT compared to controls as measured by CT (81.3±32.7 vs. 48.4±27.4 mL/m, P<0.001) and CMR (RV-EAT: 3.28±1.52 vs. 2.65±1.44 cm²/m, P=0.042; BV-EAT: 4.97±1.99 vs. 4.07±1.92 cm²/m, P=0.027). CT-EAT positively correlated with CMR measured RV-EAT (Whole cohort: r=0.599, P<0.001; T2D: r=0.552, P<0.001; Controls: r=0.712, P<0.001) and BV-EAT (Whole cohort: r=0.688, P<0.001; T2D: r=0.649 P<0.001; Controls r=0.769, P<0.001). Inter-observer variability analysis demonstrated moderate reliability for RV-EAT (ICC: 0.636, Bias: 1.153, LLOA: -4.197, ULOA: 6.503) and BV-EAT (ICC: 0.567, Bias: 1.522, LLOA: -4.292, ULOA: 7.336). Discussion Conclusion CMR-measured RV-EAT and BV-EAT have moderate correlation with CT-measured EAT and have moderate inter-observer reliability. Further work needs to assess association between CMR-measured EAT with cardiac structure, function and outcomes. Acknowledgements