ABC | Volume 114, Nº2, February 2020

Original Article Ribeiro et al. Simplified method to assess diastolic function by CMR Arq Bras Cardiol. 2020; 114(2):284-292 Figure 1 – Longitudinal displacement of the atrioventricular junction (AVJ). The same four- chamber slice is shown in three different cardiac phases during AVJ rapid movement: A and B, C and D, E and F. On the left column the arrowhead represents the reference used by Saba et al. 9 and the thin arrow shows the anatomical reference used in the present study. Note that when the cardiac motion is faster (small arrow in C and D), we could not precisely identify the site of the mitral valve insertion; however, the adjacent coronary sinus wall is still well defined. On the right column, we showed the lines used for the LV longitudinal measurements on this study (thinner line) and by Saba et al. 9 (thicker line) AVJ motion analysis Means and standard deviations were calculated for each of the AVJ motion variables (MD, MVED, VDS, VDS/MVED) of AVD patients and normal control subjects. AVJ data were compared between patients and controls. We found statistically significant differences in MD and the three CMR correlates of diastolic LV function (MVED, VDS, VDS/MVED) in patients with AVD compared to normal controls, as noted in Table 3 and Figure 3. Patients with AVD showed significantly lower normalized MD at the AVJ compared to healthy volunteers. AVJ of patients with AVD recoiled at significantly slower normalized maximum velocities (s -1 ) in early diastole compared with healthy volunteers. Conversely, during diastasis, AVJ motion occurred at significantly faster normalized velocities in patients with AVD. We found a 17-fold higher VDS/MVED ratio in AVD compared with healthy volunteers (Figures 3 and 4, Table 3). The Bland-Altman analysis (Figure 5) for MD revealed a bias of -2.81 and 95% CI (confidence interval) of (-3.66 to -1.95) for normal controls (p < 0.001) and a bias of -2.97, 95% CI of (-4.11 to -1.83) for AVD patients with p < 0.001. Comparison of diastolic function based on AVJ parameters between patients with predominant stenosis and predominant regurgitation did show significant differences in all diastolic function at CMR (Table 4). Impairment of diastolic function was higher in patients with aortic regurgitation compared to stenosis. Diastolic function, LV structure and clinical parameters Results of univariate and multiple linear regression analysis including LV mass, volumes and function as well as patient characteristics such as age, gender, heart rate and blood pressure are shown in Table 5. In a univariate analysis, MD and MVED correlated significantly with LV volume, left ventricular ejection fraction (LVEF) and LV mass. MVED also correlated to systolic blood pressure (SBP). VDS/MVED correlated with LV mass, LVEF and heart rate (HR). VDS showed correlation only with LV mass and HR (Table 5). In a multivariate linear regression model, MD and MVED were predicted only by LV mass. All other parameters of LV structure, volume and function were not predictive of MD and MVED in this multivariate approach. Gender and LV mass independently predicted VDS, while VDS/VMED ratio was predicted by LV mass and LVEF. In summary, these results indicate that LV mass maintains a significant correlation across the four measured diastolic parameters in the univariate and forward stepwisemultiple linear regression. Additionally, gendermaintained a significant correlation with VDS and LVEF with VDS/MVED ratio. The remaining variableswere not independently correlatedwithdiastolic function parameters derived from linear measurements. Thus, body size (body mass index), HR and blood pressure did not influence significantly linear diastolic parameters measured by CMR. 286