New Insight Into the Evaluation of Abnormal Left Ventricular Wall Motion

Evaluation of mechanical dyssynchrony using echocardiography has failed to improve refractory heart failure in patients treated with cardiac resynchronization therapy. Previous predictors may not accurately reect cardiac dyssynchrony. It was hypothesized that the spatially and temporary continuous information of the whole endocardium is required when the mechanical dyssynchrony is assessed using echocardiography. This study aimed to examine differences in the locus of the centroid of the left ventricle between abnormal and normal wall motion.


Results
The locus of the centroid of the left ventricle in the normal wall motion group showed a horizontally inverted β shape, whereas this shape was absent in the other groups. When corrected by left ventricular end-systolic volume, the total and each directional length of the locus of the centroid of the left ventricle in the abnormal wall motion groups were clearly reduced compared with those recorded in the normal wall motion group. The acceleration of the centroid was also reduced in the abnormal wall motion groups. Multiple regression analysis with a stepwise method revealed a corrected antero-posterior shift of the centroid of left ventricle by left ventricular end-systolic volume and N-terminal pro-brain natriuretic peptide, which strongly correlated with the LVEF (adjusted R 2 : 0.6818, p≤2.2e-16).

Conclusion
Use of the locus of the centroid of the left ventricle provides novel insight into the evaluation of abnormal left ventricular contractions.
Trial registration retrospectively registered Background In the last century, echocardiography emerged in the clinical eld as a visualization tool for the evaluation of cardiac anatomical and pathological abnormalities, as well as the ow dynamics of heart diseases [1].
With the technological development, this diagnostic tool has contributed to the evaluation of cardiac diseases, such as cardiomyopathies, ischemic heart diseases, valvular heart diseases, etc. However, despite the application of several techniques, the objective judgement of the left ventricular (LV) wall motion remains a challenge.
For example, it is recognized that cardiac resynchronization therapy is a therapeutic strategy for patients with medical resistant refractory heart failure in whom LV systolic function is severely reduced. The European Society of Cardiology guidelines established in 2016 and 2019 stated that patients with a symptomatic sinus rhythm, reduced LV ejection fraction (LVEF), prolonged conduction time which met the QRS duration ≥ 130 ms, and a left bundle branch block shape on an electrocardiogram could be responders to cardiac resynchronization therapy [2,3]. As demonstrated in the PROSPECT study [4] and ECHO-CRT trial [5], echocardiography is unable to identify responders to this novel therapeutic strategy. Previous echocardiographic assessment of mechanical dyssynchrony was limited to regional LV wall information utilizing a tissue Doppler technique or M-mode calculation, although the heart is a threedimensional moving muscle. Blood ow was also assessed using a pulse Doppler technique, which estimated the LV dysfunction using the time difference obtained from the blood stream at the LV and right ventricle in ow/out ow. Obviously, this was indirect information and did not critically re ect the whole LV wall motion. Therefore, the present author presumed that the accurate evaluation of LV dyssynchrony requires information for each wall of the heart during a consecutive cardiac cycle, particularly to determine the suitability of resynchronization therapy.
In the present study, it was hypothesized that the centroid of the LV re ects a cardiac wall motion because this technique requires the spatial coordinates of each endocardial position on the LV wall with time information during a cardiac cycle. Therefore, the locus of the centroid could include information on each regional area, such as the presence of interstitial/replacement brosis which limits LV contraction. This study investigated the role of the locus of the centroid in the assessment of abnormal LV wall motion as a novel approach differentiated from previous methods.

Study population
Continuous digital videos of 633 patients, which were obtained from the clinical echocardiography laboratory from September 2016 to August 2017, were utilized in this study. Cases of dilated cardiomyopathy (DCM) and old myocardial infarction (OMI) with aneurysm were evaluated. Cases of valvular heart disease, OMI without aneurysm, congenital structure cardiac disease, pulmonary hypertension, and those with images of poor quality were excluded from the study. The remaining videos that showed normal wall motion (NWM) of the LV were also used in this study as control.
The protocol of this retrospective study was approved by the SOYOKAZE CVD ethics committee (soyokaze-cvd, 2018-03). The purpose of this study was conveyed to the patients on the information board and the website homepage of the clinic. An opt-out option was provided to patients who did not wish to participate in the study. Two-dimensional tissue tracking system of the LV Audio/video interleave les, obtained during one cardiac cycle using a trans-thoracic echocardiography equipped with a high-resolution sector probe (AVIUS; Hitachi Ltd., Tokyo, Japan), were evaluated. The twodimensional speckle tracking algorithm (Hitachi Ltd.) is a pattern-matching method which forwards dozens of pixels into the region of interest (1 cm 2 ) through an off-line system using an application termed '%WT' programmed in e-Tool viewer (Hitachi Ltd.) [6,7]. In this off-line system, approximately 50 points were automatically allocated on the manually traced line as the border of the endocardium. The coordinates of each point, which were followed frame by frame during one cardiac cycle, were saved as a comma-separated value le.
Locus of the centroid of the LV The locus of the centroid of the LV was subsequently calculated through each frame image using an original application. The centroid of the three-dimensional LV was identi ed as the middle point between the centroid of the four-chamber image and that of the two-chamber image. The locus of the centroid of the three-dimensional LV was also shown on the same sheet.

Statistical analysis
The normally distributed continuous data are shown as mean values ± standard deviation. Non-normally distributed data are shown as medians with the rst and third quartile values. Data with exponential distribution were logarithmically converted to obtain a normal distribution. Quantitative data were evaluated using one-way analysis of variance, with post-hoc analysis utilizing Tukey's method. Qualitative data were assessed through Fisher's exact test. Non-parametric data for each group were processed by Kruskal-Wallis analysis with a post-hoc method. A univariate linear regression analysis and a stepwise multiple linear regression analysis were applied to identify the independent predictors of the LVEF. A pvalue < 0.05 denoted a statistically signi cant association in the nal model.
All statistical analyses were performed with EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria). More precisely, it is a modi ed version of R commander designed to include additional statistical functions frequently used in biostatistics [8].

Results
A total of 260 patients were evaluated in this study. Patient background information is summarized in Table 1. The ndings of physiological examination and a biomarker analysis for each group are summarized in Table 2.  Data are presented as the mean ± standard deviation. In the disease groups, the actual locus length of the centroid and the box volume were not signi cantly different among the groups (Figs. 2a, 2b). These were calculated from the x-, y-, z-transferred distance which the centroid of the three-dimensional LV had moved toward each direction during one cardiac cycle.
However, when corrected by the LV end-systolic volume (LVESV), the length of the locus of the centroid of  (Fig. 2d).
In the DCM group, the locus length of the centroid for the lateral and antero-posterior directions was longer than that of the NWM group (Figs. 3a, 3b); nevertheless, the length for the longitudinal direction was shorter than that of the NWM group (Fig. 3c). When these data were corrected by LVESV, the lengths were signi cantly reduced compared with those obtained from the NWM group (Figs. 3d, 3e). The corrected longitudinal length of the abnormal wall motion groups was markedly reduced compared with that of the NWM group (Fig. 3f).  . 4b). When the acceleration values were corrected by LVESV, those of the abnormal wall motion groups were more markedly reduced compared with those of the NWM group (Figs. 4c, 4d).
In the univariate analysis, the transfer distance for the lateral direction, total distance of centroid movement during one cardiac cycle, maximum and minimum accelerations of the LV centroid, QRS duration, and value of the N-terminal pro-brain natriuretic peptide were associated with LVEF (Table 3).
However, when the data concerning the distance, volume, velocity and acceleration of the centroid were corrected by LVESV, all data included in Table 3 were related to LVEF. The QRS duration, total distance of centroid movement, maximum velocity, and mean acceleration of the LV centroid were excluded from the multivariate analysis because these data were considered variance in ation factors. Finally, a multiple linear regression analysis with stepwise methods revealed that the N-terminal pro-brain natriuretic peptide and an antero-posterior shift of the LV centroid were strong predictors of the LVEF (multiple R 2 : 0.6882; adjusted R 2 : 0.6818; p ≤ 2.2e-16) ( Table 4). A-P, antero-posterior; CI, con dence interval; LCG, locus of the center of gravity; Log; logarithmic; LVEF, left ventricular ejection fraction; LVESV, left ventricular end-systolic volume; NT-proBNP, N-terminal probrain natriuretic peptide; SD, standard error A-P, antero-posterior, CI, con dence interval; Log, logarithmic; LVEF; left ventricular ejection fraction, LVESV; left ventricular end-systolic volume, NT-proBNP; N-terminal pro-brain natriuretic peptide, SE, standard error

Discussion
The present study demonstrated that the LV centroid of the NWM group had moved like a mirror image of a β shape in a counterclockwise direction. In case of abnormal wall motion (e.g., dyssynchrony associated with DCM or LV aneurysm caused by an OMI), the corrected distance for any direction of the locus of the centroid by LVESV was signi cantly reduced compared with that noted for the NWM group, especially in the longitudinal direction. The LV centroid had to shift toward the anterior direction in an early phase of LV contraction to increase the LVEF.
In patients with an extended LV chamber, the length which the LV centroid had moved during one cardiac cycle was limited to the enlarged cavity. This is because the cardiac muscle is con ned in the cardiac sac, which is constructed by a tight brous membrane. To obtain su cient stroke volume from the LV cavity, the LV centroid had to shift to the frontal position into the LV during the early contraction phase. This is because the blood stream had to be directed toward a LV out ow rather than a LV in ow positioned in a mitral ring. After moving toward the apical direction, it returned to the original position with two loose loops, which might have been caused by the cardiac translation of a diastolic phase and an atrial kick. Furthermore, the box volume calculated from the locus of the LV centroid to the LVESV was reversely correlated with the LVEF. This indicated that the oating motion of the LV centroid had not contributed to an increase in the LVEF.
The evaluation of LV wall motion (e.g., asynergy or dyssynchrony) using echocardiography is often subjective and depends on the experience of the physicians and expert technicians. It has been reported that strain is a good evaluator of the regional asynergy of ischemic heart disease [9,10,11]. In contrast, there are no standard markers available for the evaluation of mechanical dyssynchrony. Several studies assessed mechanical dyssynchrony using a tissue Doppler method [12,13,14], the time difference between the opposite sites using a M-mode technique [15], and the time discrepancy between the preejection time of both ventricles or the LV lling time for a R-R interval [16]. However, the PROSPECT trial demonstrated that the aforementioned echocardiographic parameters failed to identify the responders to cardiac re-synchronization therapy [4] because those evaluations were restricted to local or indirect information of the LV wall. The results of this study proposed the LV centroid as a novel approach, which requires information on each point of the LV endocardium and each time point during a cardiac cycle.
This method depends on the spatial and temporal information of the whole LV endocardium of the moving heart rather than selected information from a part of the LV and a limited time. Based on these reasons, this novel approach may contribute to the selection of patients with mechanical dyssynchrony who could respond to cardiac resynchronization therapy.
In conclusion, the locus of the centroid of the LV associated with abnormal contraction of the heart did not resemble the mirror image of a β shape. Furthermore, the corrected antero-posterior shift of the LV centroid by LVESV was a strong predictor of the LVEF. The present evidence demonstrated that these are important factors for maintaining su cient systolic volume.

Limitations
In this study, the LV centroid calculated from two orthogonal views using a two-dimensional tracking method was utilized as a three-dimensional LV gravity. If a real-time three-dimensional speckle tracking system was available, the association with the LV centroid could have been examined in more detail.