ABC | Volume 114, Nº4, April 2020

Original Article Güneş et al. Epicardial adipose tissue and insulin resistance Arq Bras Cardiol. 2020; 114(4):675-682 Methods Study Population For this prospective and cross-sectional study, 94 obese patients aged 8–18 years admitted to the Kahramanmaraş Sütçüimam University Pediatric Endocrinology outpatient clinic between August 2018 and February 2019 were included. An outpatient clinic nurse performed all anthropometric measurements, including weight and height with the patients wearing underwear only. Body mass index (BMI) was calculated by dividing body weight into kilograms by the square of height in meter. Obesity was defined as BMI greater than the 95 th percentile in an age- and sex-specific percentile curve. A value above the 99 th percentile was defined as morbid obesity. 19 All patients’ insulin resistance was calculated, and they were categorized on the basis of insulin resistance. Insulin resistance was calculated using the Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) (fasting plasma glucose (mmol/L) × fasting plasma insulin (mU/L)/22.5) and defined as HOMA-IR greater than the 90 th percentile in an age- and sex-specific percentile curve. 20 Patients with DM, Cushing Syndrome, known insulin resistance; those using drugs for insulin resistance; those having hypoglycemia; those with known metabolic cardiovascular and hepatic disorders; and those with poor acoustic windows for echocardiography were excluded. Demographic and laboratory data of the patients were recorded. All patients underwent standard transthoracic echocardiography including tissue Doppler examination and echocardiographic examinations were performed by an expert cardiologist. The same cardiologist evaluated pre-discharge TTE results of 20 randomly selected patients to assess the reproducibility of EAT thickness and tissue Doppler parameters for atrial electromechanical delay. Using the Bland-Altman method, the mean difference in terms of intra-observation was 3.8% (0.23 ± 0.54%), indicating good reproducibility. Echocardiography Transthoracic echocardiographic examinations were performed by expert echocardiographers who were blind to the patients’ clinical information with the Vivid 7 ® cardiac ultrasonography system (GE VingMed Ultrasound AS; Horten, Norway) using 2.5- to 5-MHz probes. The echocardiographic images were taken in left lateral and supine positions and 2D, M-mode, pulsed, and color flow Doppler echocardiography examinations were performed in every patient. Parasternal long and short axes, apical and subcostal windows were used to obtain Doppler tracings and two-dimensional images. Left and right atrial diameters left ventricular end-systolic and end‑diastolic diameters, as well as the posterior and septal wall thicknesses of the left ventricle at diastole, were quantified. Left atrial volumes were quantified with the disc method LV ejection fraction (EF) with the Simpson’s rule. LV diastolic function was quantified using mitral inflow velocities, i.e. peak E (early diastolic), peak A (late diastolic), E/A ratio as well as E-wave (DT) deceleration time and isovolumic relaxation time (IVRT). Echocardiographic assessment and quantification of epicardial fat were done by identifying the echo-free space between the outer lining of the myocardium and the visceral layer of pericardium. Its measurement was made perpendicular to the free wall of the right ventricle in the parasternal long-axis window. The measurement level was at the mid ventricle and the timing was set to end-diastole, with an average of 3 cardiac cycles being taken. To align the ultrasound beam perpendicular to the right ventricular free wall, aortic annulus was accepted as the anatomic landmark. 21 Tissue Doppler Echocardiography (TDE) The pulsed Doppler sample volume was placed at the level of LV lateral mitral annulus, septal mitral annulus, and RV tricuspid annulus from an apical four-chamber view. The time interval from the onset of the P-wave on surface ECG to the beginning of the late diastolic wave (Am), which is called PA, was taken from the lateral mitral annulus (lateral PA), septal mitral annulus (septal PA), and RV tricuspid annulus (tricuspid PA). The difference between septal PA and tricuspid PA (septal PA − tricuspid PA) was identified as an intra-atrial electromechanical delay while the difference between lateral PA and tricuspid PA (lateral PA − tricuspid PA) was identified as inter-atrial electromechanical delay. 16 Statistical Analysis All statistical analyses were performed using the SPSS version 14 (SPSS Inc., Chicago, IL, USA) software package. Statistical significance was set to a two-sided p-value < 0.05. Categorical variables were expressed as number and percentage while continuous variables as mean±standard deviation (SD) or median and interquartile ranges (IQR), depending on their normality of distribution. The normality assumption of the data was determined using the Kolmogorov Smirnov test. The independent sample t-test and Mann–Whitney U test were used to compare the groups’ means. The Chi‑square test was used to compare categorical data. Correlation analyses were performed using the Pearson correlation test for normally distributed variables and Spearman correlation test for non‑normally distributed variables. An optimal cut-off point was determined for EAT to predict IR using the receiver operator characteristic (ROC) curve analysis MedCalc (v12.7.8). This was accomplished by determining the area under the curve (AUC) with 95% confidence interval. The best cutoff value for EAT was determined by calculating the highest sum of sensitivity and specificity-1. The IR and available variables were analyzed for correlation using the univariate analysis. Variables with significant correlation in the univariate analysis were entered in the multivariate logistic regression model using the backward stepwise method along with other potential confounders to determine independent predictors of IR. Results Enrolled patients were divided into two groups based on the presence of insulin resistance. Forty patients had insulin resistance and 54 patients did not. Both groups had similar age and gender distribution. (p = 0.102, p = 0.069, respectively). Among the anthropometric measurements, weight, height, and 676