Echocardiographic Assessment of Right Ventricular Function by Two-Dimensional Strain In Patients with Left-Sided Valvular Heart Disease : Comparison with Three-Dimensional Echocardiography

Mailing Address: Alex dos Santos Felix Rua das Laranjeiras, 374. Postal Code: 22240-004, Laranjeiras, Rio de Janeiro, RJ Brazil. E-mail: alex.felix.ext@dasamed.com.br, alexsfelix@gmail.com.br Echocardiographic Assessment of Right Ventricular Function by Two-Dimensional Strain In Patients with Left-Sided Valvular Heart Disease: Comparison with ThreeDimensional Echocardiography Alex dos Santos Felix,1 Ana Paula dos Reis Velloso Siciliano,1 Luciano Herman Juacaba Belém,1 Fabiula Schwartz de Azevedo,1 Sergio Salles Xavier,2 Andrea Rocha De Lorenzo,1 Clerio Francisco de Azevedo Filho1 Instituto Nacional de Cardiologia (INC),1 RJ Brazil Fundação Oswaldo Cruz (FIOCRUZ),2 RJ Brazil


Introduction
2][3][4] The assessment of RV function by conventional echocardiographic parameters has major challenges, and its accuracy is limited by the irregular geometry of the RV chamber, the distinct pattern of contractility (mostly based on longitudinal deformation), the trabeculated inner contour of the cavity with poor endocardial border definition, separate RV function by 2D strain in left-sided valve disease Int J Cardiovasc Sci.2018;31 (6)630-642

Original Article
inflow and outflow chambers which may be adequately visualized only from separate views, load-dependency, and influence of ventricular interdependency. 5,6 some pathologic conditions where changes in preload (e.g.severe tricuspid regurgitation) and afterload (e.g.pulmonary hypertension) are seen, the evaluation of RV function is particularly difficult.In patients with severe VHD, RV function assessment is challenging, not only because of the hemodynamic alterations frequently seen in these patients, but also because its primary etiopathogenic process itself may impair RV function, as we often see in rheumatic heart disease. 7 the last few years, studies have shown the applicability and clinical value of three-dimensional echocardiography (3DE) and two-dimensional strain (2DS) techniques in the evaluation of RV systolic function, with good accuracy for detecting RV dysfunction, and additional prognostic value in various diseases. 80][11][12] Real-time 3DE is a well-established echocardiographic technique that has the great advantage of displaying the entire right ventricle in a single dataset, despite its irregular shape.[15][16][17]

Objectives
The aim of this study was to analyze RV systolic function in patients with severe left-sided VHD using conventional echocardiography and 2DS techniques, testing the correlation of these techniques with RV ejection fraction (RVEF) measured by 3DE, and to evaluate the accuracy of these three techniques for the detection of RV dysfunction (RVEF < 44%).

Methods
This prospective observational cross-sectional study was approved by the local ethics committee.

Study population
From May 2013 to May 2014, in a tertiary cardiology hospital, we recruited consecutive adult patients with diagnosis of severe mitral and/or aortic valve disease 18 referred for preoperative evaluation for cardiac surgery (valve replacement, repair or both).We only included patients with no previous history of cardiac surgery, to avoid the influence of pericardiotomy on the accuracy of echocardiographic parameters of RV systolic function, and patients with no history of coronary artery disease to avoid confounding factors in determining the cause and severity of pulmonary hypertension and RV disease.We excluded patients with poor echocardiographic window for analysis of RV systolic function (either conventional parameters or 2DS), patients with severe tricuspid regurgitation (TR) and those who refused to participate in the study.

Conventional echocardiography
Echocardiography was performed using standard views, with the patient in the left lateral decubitus position, using a commercially available ultrasound machine (Vivid E9, GE Healthcare, Horten, Norway).Conventional echocardiographic images and cine loops of all patients were obtained by a single experienced examiner using a M5S transducer.Left ventricular (LV) EF was calculated using the biplane method of discs, and all the Doppler parameters necessary to quantitate the severity of valvular lesions and pulmonary artery systolic pressure (PASP) were obtained and analyzed in accordance with the criteria defined on the EAE/ASE/ EACVI guidelines. 19,20RV diastolic and systolic areas were measured to calculate RV fractional area change (FAC).With the pulsed-wave Doppler sample volume positioned at the lateral tricuspid annulus in the RV focused apical 4-chamber view, the peak systolic velocity (PSV) by tissue Doppler was obtained.We also measured the tricuspid annular plane systolic excursion (TAPSE), placing the M-mode cursor through the base of the lateral tricuspid annulus, quantitating its longitudinal motion at peak systole.

Speckle-tracking echocardiography
For STE analysis, digital loops of the right ventricle were obtained from apical 4-chamber and/or right ventricle-focused apical 4-chamber views.Three cardiac cycles were acquired from each view at a frame rate of 40-80 frames/sec in patients in sinus rhythm and five consecutive cycles in patients with atrial fibrillation (AF).The data were exported at the end of the test to a workstation (EchoPac BT12, GE Vingmed, Horten, Norway) for further offline analysis.
Preliminary analysis was performed online in the ultrasound machine was performed online in the ultrasound machine to check if the image quality of the loops was good enough to permit adequate tracking of the acoustic markers (speckles) of the myocardium during the entire cardiac cycle.STE analysis was performed semi-automatically by the system, after manual setting of 3 points on the endocardial border of the right ventricle by the operator (2 basal and one at the apex).When the region of interest (ROI) included the whole thickness of the right ventricle and excluded other structures such as trabeculae, moderator band and valvular tissue, the processing was started, and analysis proceeded on a frameto-frame basis using an automatic tracking system (Figure 1).If the tracking was poor, the operator could repeat the the acquisition of loops, readjusting the endocardial tracing (editing) or change software parameters such as ROI width, frame rate or gain, until an adequate tracking of the entire myocardium was achieved.
The ROI generated by the software included basal, mid and apical segments of RV free wall and septum, dividing it into 6 segments (Figure 1).Longitudinal peak strain values were measured for each segment, and the RV free wall longitudinal strain (RVFWS) and the RV global longitudinal strain (RVGS), analyzed by 2DS, were calculated by averaging the values from the three segments of the RV free wall and all the six segments along the entire right ventricle, respectively.These initial results were blinded to the investigators until the offline analysis of the remaining parameters was performed.

Three-dimensional echocardiography
3DE was performed in all subjects immediately after the two-dimensional echocardiographic examination using the same ultrasound machine, equipped with a 4V probe.RV three-dimensional (3D) images were obtained in a full-volume dataset from the apical fourchamber view, optimized for analysis of RV function.Multi-beat (3-6 beats) data were obtained during apnea, on the multislice (short axis) visualization mode, to make sure that the right ventricle was entirely included in the dataset (Figure 2).
All the measurements of RV volumes and EF were made off-line, using a dedicated software (TomTec Imaging Systems GmbH, Munich, Germany).Semiautomatic analysis was performed, with manual tracing of the endocardial borders in end-systolic and end-diastolic frames in the sagittal, four-chamber and coronal views, obtained from the full-volume dataset.In addition, end-diastolic volume (EDV), end-systolic volume (ESV), stroke volume and EF were calculated using the software (Figure 2).

Assessment of reproducibility
Evaluation of inter-and intraobserver reproducibility were performed.Fourteen patients, chosen by simple random selection, were assessed by two observers for analysis of interobserver variability, and for intraobserver variability, the second analysis was performed with a minimum interval of two weeks from the first analysis.The readers were blinded to previous measurements.Interobserver and intraobserver variability were assessed using the intraclass correlation coefficient and Bland-Altman analysis. 21

Statistical analysis
Demographic data are presented as mean ± standard deviation (SD) and categorical data are presented as frequencies.Normality of the distribution of numerical variables was tested by the Kolmogorov-Smirnov test; normally distributed variables were expressed as mean ± SD and variables with abnormal distribution as median with interquartile range.We compared all RV function parameters between subgroups of patients according to their predominant valvular lesion.To reduce the occurrence of alpha error, we used one-way ANOVA test with Bonferroni post-hoc correction.Conventional echocardiographic and 2DS parameters were compared with real-time 3DE RVEF using Spearman correlation test.For comparisons between groups of patients (A and B), based on the presence of RV dysfunction defined as RVEF (3DE) < 44%, we used nonparametric Mann-Whitney U test.ROC (receiver operating characteristic) curve analysis was used to assess the clinical utility of all RV function variables in defining RV dysfunction.P values < 0.05 were considered statistically significant.Statistical analyses were performed using SPSS version 13 (SPSS Inc, Chicago, IL).

Patients' characteristics
A total of 57 consecutive patients with severe VHD were enrolled in this study.Of these, two had severe TR and other two refused to participate, thus the final study group was comprised of 53 patients (31 women; mean age, 52,4 ± 15,9 years).Most patients were symptomatic, with 50.9% classified as New York Heart Association (NYHA) functional class II and 43.4% as NYHA III (Table 1).The predominant etiology of valve diseases was rheumatic valve disease (53.6%), myxomatous valve disease (18.9%), degenerative valve disease (13.2% ) and congenital valve disease (11.3%).All patients were submitted to conventional echocardiography, RV 2DS and 3DE.Most patients were in sinus rhythm; 14 patients (26.4%) with permanent AF were not excluded because we were able to analyze all echocardiographic parameters despite the presence of arrhythmia.

Echocardiographic parameters
Technically adequate measurements of TAPSE, PSV, FAC and 2DS parameters were obtained in all patients.
Real-time 3DE images of the RV were successfully analyzed in 51 of the 53 patients evaluated (96.2%).Image quality was considered inadequate for analysis in two patients, due to unsatisfactory echocardiographic window (missing the anterior wall of the RV).
Considering the entire study population, mean values of LV chamber dimensions were increased, despite normal LV systolic function.Overall, RV dimensions and function were normal, as summarized in Table 2.
The patients were classified into subgroups according to their predominant valve lesion as follows: (1) mitral stenosis (n = 11; 20.8%), (2) mitral regurgitation (n = 21; 39.6%), (3) aortic stenosis (n = 8; 15.1%), (4) aortic regurgitation (n = 9; 17%), and (5) combined lesions (n = 4; 7.5%).We defined combined lesions as the presence of two or more severe mitral and/or aortic valve lesions in the same patient.Of the four patients with combined lesions, two had severe mitral stenosis and regurgitation, one had severe aortic stenosis and regurgitation and the other one had severe mitral regurgitation and aortic regurgitation.The echocardiographic variables were compared between these subgroups (ANOVA for multiple comparisons), and we found a significant difference between the groups in all the parameters, except for PSV (Figure 4).Patients with stenotic valve lesions had lower values of PSV compared with patients with regurgitant lesions.Patients with combined lesions had lower values of all conventional RV function parameters compared with the other groups.We observed lower absolute values of RVFWS, RVGS (less deformation) and higher PASP, and lower FAC and RVEF 3D in patients with mitral stenosis and patients with combined lesions.

Intra-and interobserver variability analysis
Reproducibility analysis showed excellent accordance between repeated measurements for the RV 2DS parameters by Bland-Altman analysis (Figure 6).Both RVFWS and RVGS showed high intraclass correlation coefficient (range, 0.97 -0.98) with narrow confidence intervals (Table 4).

Discussion
There are few studies in literature focusing on RV function in left-sided VHD, most of which had limited number of patients and analyzed only conventional RV function parameters, without using 2DS parameters or 3DE.The great challenge in VHD is to detect early alterations in RV function, when subclinical disease may point to a worse clinical prognosis and contribute A population of severe mitral and/or aortic valve disease patients was enrolled in this study, predominantly rheumatic in etiology, unlike other studies from Europe and North America that also evaluated RV function in VHD patients, in which the predominant etiology was degenerative valvular disease. 22,23In Brazil, rheumatic fever is still a prevalent cause of VHD, and almost 60% of the patients that undergo cardiac surgery for valvular repair or replacement have rheumatic etiology. 24These patients frequently present with disease in more than one valve, and therefore, we decided to include in the present study patients with combined mitral and/ or aortic lesions, to better represent the entire clinical spectrum of the disease.
Considering RVEF measured by 3DE as an established reference standard for evaluation of RV function, most patients in our study had normal RV function (RVEF 3D ≥ 44%), although the number of patients with RV dysfunction was significant (n = 14; 27.4%).The mean  , secondary to the advanced stage of the disease in these patients.We excluded patients with severe TR, a condition that may affect the accuracy of RV functional assessment by alterations in RV preload. 25 obtained acceptable 3DE images for RVEF analysis in 51 patients (96,2%), showing good feasibility of the technique, as previously shown by other authors (Kong et al., 26 -97%, Niemann et al., 27 -100%).The mean values of EDV, ESF and EF were normal in the overall population.
Analyzing the patients according to their predominant valve lesions using multivariate analysis, we observed significant differences between the groups for RVGS, RVFWS, TAPSE, FAC, and RVEF 3D, showing a tendency towards lower absolute values of RVGS, RVFWS (less deformation) and lower levels of FAC and RVEF 3D in patients with mitral stenosis and combined lesions.These findings are probably related to higher levels of pulmonary capillary pressure and RV pressure overload in mitral stenosis and combined lesions than in regurgitant lesions and isolated aortic stenosis.Furthermore, all patients with combined lesions were rheumatic, pointing to the possibility of a concurrent  demonstrated the occurrence of alterations in RV systolic function in patients with pulmonary artery hypertension, proportional to the severity of the disease, which could help explain some of our findings.
We compared the parameters of RV systolic function with RVEF 3D, and found a moderate, negative correlation between RVEF 3D and RVGS, RVEF 3D and RVFWS, and a moderate positive correlation between RFVE 3D and FAC, with weaker correlations for TAPSE and PSV.These findings are in accordance with previous studies, showing good correlation between RV 2DS parameters and RVEF measured by CMR 30 and FAC with RVEF measured by CMR. 31 When the population was divided into two categories, according to the absence of RV dysfunction (group A, RVEF ≥ 44% by 3DE) and the presence of RV dysfunction (group B, RVEF < 44% by 3DE) we found a significant difference in all parameters of RV function between the groups.ROC curve analysis was performed to test the diagnostic performance of these variables to detect RV dysfunction.The best AUC was obtained for FAC (0.932) followed by RVGS (0.872) and RVFWS (0.851), showing the clinical utility of these parameters in detecting RV dysfunction.Among all, FAC had the best performance, and this may be explained by the fact that this is the only parameter directly related to RV ejective function, while all others are closely related to longitudinal systolic function.
We performed intraobserver and interobserver analysis for RVGS and RVFWS and found good reproducibility for both parameters, making these measurements more robust and reliable, confirming previous data. 4,32,33ttal et al., 34 did not find any correlation between RV systolic parameters and PASP in 22 mitral stenosis patients, attributing RV myocardial dysfunction to inflammatory damage caused by the rheumatic disease.Ozdemir et al., 35 demonstrated that patients with mild-to-moderate mitral stenosis had altered values of longitudinal RV 2DS compared to controls, probably unrelated to pulmonary hypertension, since they found only a mild elevation of PASP in these patients (39 ± 14 mmHg).Tanboga et al., 32 studied patients with mild-to-moderate mitral stenosis and also found altered values of longitudinal RV 2DS compared to controls, but did not find any correlation of these values with PASP.Castro et al., 36 studied 46 patients with isolated severe mitral stenosis, showing reduced longitudinal RV 2DS compared to controls, and a weak correlation between 2DS and PASP.Galli et al., 37 studying 200 patients with degenerative aortic valve stenosis, demonstrated RV dysfunction in 24% of these patients, and established concomitant LV and RV dysfunction as the major predictor of mortality in 16 months.Le Tourneau et al., 38 evaluating RV systolic function in 208 patients with severe organic mitral regurgitation found severe RV dysfunction (RVEF ≤ 35% measured by radionuclide angiography) in 63 patients (30%).The authors showed a weak correlation between PASP and RVEF and suggested a direct relation of RV dysfunction with septal function alteration and of LV enlargement with remodeling (ventricular interdependence).Mitral valve disease typically causes greater overload in the right chambers than aortic valve diseases, 39 probably due to an exceptional elevation of capillary pulmonary pressure, either by volume overload in mitral regurgitation or pressure overload in mitral stenosis.
Our findings suggest that RVGS and RVFWS may be reliable markers of RV dysfunction in VHD patients, with good accuracy and the potential advantage of early detection of alterations in myocardial function that precede alterations in the RV ejective function.

Limitations
Our findings must be validated in other studies involving a larger number of subjects.Our small sample size does not allow us to extrapolate these results to other populations.This was a clinical, uncontrolled study, which included consecutive patients with leftsided VHD from different etiologies and mechanisms of valvular dysfunction, reflecting the population of patients currently treated in our clinical practice.

Conclusion
In left-sided VHD patients, RVGS and RVFWS showed good correlation when compared with RVEF 3DE and good accuracy in detecting RV dysfunction.2DS might be a useful tool for the early detection of changes in RV function in VHD patients.

Figure 1 -
Figure 1 -Speckle-tracking analysis of right ventricular (RV) systolic function using the automated function imaging technique.(Top) Semi-automated delineation of endocardial borders and definition of the region of interest (ROI).Mid and lower images illustrate two-dimensional strain analysis: (middle-left) shows global longitudinal RV strain (-25.6%),(middle-right) two-dimensional strain (2DS)-time curves, (bottom-left) peak 2DS values per segment, and (bottom-right) 2DS-time parametric analysis.

Figure 2 -
Figure 2 -(a,b) Right ventricular (RV) multi-beat, full-volume acquisition by three-dimensional echocardiography.(c,d) RV volumes and ejection fraction analysis using TomTec software.Rendering of RV chamber on the top (c), and volume-time curve and quantitative measurements on the bottom (d)

Table 1 -Clinical profile and comorbidities of the enrolled subjects Clinical variable Number of patients (% total)
NYHA: New York Heart Association functional class; DM: diabetes mellitus.

Table 4 -Reproducibility analysis for retest measurements of RVFWS and RVGS
29 function by 2D strain in left-sided valve disease Int J Cardiovasc Sci.2018;31(6)630-642Original Article primary involvement of the myocardium, due to the inflammatory and fibrotic processes inherent to this disease.Using 2DS, Pirat et al.,28and Ikeda et al.,29