Dissecting Myocardial Mechanics in Patients With Severe Aortic Stenosis: 2-Dimensional vs 3-Dimensional–Speckle Tracking Echocardiography

Background: Aortic valve stenosis (AS) commonly causes left ventricular (LV) pressure overload; thus, identifying patients with adverse remodeling/early LV dysfunction is critical. We compared 2-dimensional (2D) to 3-dimensional (3D) echocardiographic measures of LV myocardial deformation in patients with severe AS and studied the relation of LV preload and afterload (Zva) to myocardial deformation. Methods: We prospectively included 168 symptomatic patients (72±12 years) with severe AS and ejection fractions ≥50%. Strain parameters from those patients were compared with normal values found in the literature. 3D full-volume and 2D images were analyzed for global longitudinal strain (GLS), global radial strain (GRS), global circumferential strain (GCS), systolic strain rate (SRs), basal rotation (Rotmax-B), apical rotation (Rotmax-A), and peak systolic twist (Twistmax). Results: 2D–GLS and 2D–GCS decreased significantly compared with normal values (P˂.001 and P=.02, respectively); 2D Rotmax-B and Twistmax increased (P˂.001 vs normal values). Agreement between 2D–GLS and 3D–GLS by concordance correlation coefficient was 0.49 (95% CI, 0.39-0.57) in patients with AS. Both 2D– and 3D–GLS correlated with valvulo-arterial impedance (Zva) (r=0.34, P<.001; and r=0.23, P=.003, respectively). Conclusion: In patients with severe AS, GLS and GCS decreased, and basal rotation and twist increased to maintain LV ejection fraction. 2D– and 3D–GLS had a relatively fair agreement. Both 2D– and 3D–GLS correlated modestly with Zva. strain


4
Aortic valve stenosis (AS), which is the most common native valve disease, is characterized by left ventricular (LV) pressure overload. In patients with AS, the left ventricle often faces 2 afterloads: valvular and arterial (1,2). The increasing afterload can lead to LV remodeling and a change in coronary flow reserve. These alterations can cause subendocardial ischemia and fibrosis and may gradually affect LV systolic function (3)(4)(5). LV ejection fraction (LVEF) is the most important conventional parameter used to assess LV myocardial function. However, it is well known that LVEF is an index reflecting LV chamber function. A decrease in LVEF usually occurs at an end stage of severe AS (6,7). Global longitudinal strain by 2dimensional (2D) echocardiography can be used to detect early systolic dysfunction and has been proposed as an important marker of LV function in patients with AS (8)(9)(10)(11). However, the process of ventricular contraction is very complex because relaxation occurs in 3 dimensions and cannot be comprehensively quantified by 2D echocardiography (12,13). For this reason, 3-dimensional (3D) echocardiography and speckle tracking strain imaging are promising techniques that may provide a more complete picture of myocardial deformation (14,15). New insights into LV remodeling and myocardial deformation could potentially improve our ability to identify patients at high risk for adverse remodeling or early LV dysfunction.
Currently published studies are small and provide only limited and conflicting data on how myocardial deformation is affected by aortic stenosis, as measured by 3Dspeckle-tracking echocardiography (3D-STE) (3,4,16,17). Therefore, the purpose of this prospective study was 1) to determine whether and how myocardial deformation in 2D and 3D is affected in patients with severe AS and normal LVEF vs normal values found in the literature; 2) to compare 2D to 3D echocardiographic measures of LV myocardial deformation in patients with severe AS; and 3) to characterize the relationship of LV preload, afterload, and valvulo-arterial impedance (Z va ) with myocardial deformation, as measured by 2D and 3D echocardiography.

Study Population
We prospectively recruited 178 symptomatic patients who had been evaluated for severe AS, defined as a mean gradient of at least 40 mm Hg or an aortic valve area less than 1.0 cm 2 , and who had a normal ejection fraction (EF), defined as greater than 50%, on conventional transthoracic echocardiography (TTE

Image Acquisition and Analysis
Each patient underwent a standard 2D-TTE and a real-time 3D echocardiogram in the left lateral decubitus position, using commercially available equipment (IE33 and EPIQ7, Philips Medical Systems, Andover, Massachusetts) with a fully sampled matrix-array transducer (X5-1). Studies were performed by an experienced 6 cardiologist (X.B.). 3D full volume was acquired from the apical window with a highvolume rate (average, ≥30 volumes/s) and 6-beat acquisition, allowing for full coverage of the entire left ventricle by the pyramidal volume. Patients were told to hold their breath during image acquisition. Images were optimized for endocardial border visualization before acquisitions; overall gain was modified and time gain and compression adjusted, as needed. The acquired 2D images and 3D full-volume images were analyzed offline with TomTec 4D Echo software, version 4.6 (TomTec Imaging Systems, Image Arena, Unterschleissheim, Germany).
For 2D echocardiography, the standard 2D, M-mode, and Doppler measurements were obtained in accordance with guidelines from the American Society of Echocardiography. LV end-diastolic volume (LVEDV), end-systolic volume (LVESV), and ejection fraction (LVEF) were measured manually by using the biplane Simpson method. Three standard apical views (4-chamber, long-axis, and 2chamber) were obtained for the assessment of global longitudinal strain (GLS) and global longitudinal systolic strain rate (GLSR), and 3 parasternal short-axis views (basal, mid, and apical levels) were obtained for the assessment of global radial strain (GRS) and global radial systolic strain rate (GRSR), global circumferential strain (GCS) and global circumferential systolic strain rate (GCSR), and LV apical peak systolic rotation (Rot m ax -A), LV basal peak systolic rotation (Rot m ax -B), and peak systolic twist (Twist m ax ).
For 3D echocardiography, 3 standard apical views were automatically extracted from the 3D full-volume data sets. The mitral annulus and the LV apex were manually selected as the landmarks to initialize the LV boundaries. Then, the 3D endocardial surface was automatically reconstructed at end diastole and end systole. The endocardial surface reconstruction was manually adjusted, as 7 necessary, and the papillary muscles were included as part of the LV cavity.
Subsequently, 3D-speckle tracking was automatically characterized. The software provided anatomical longitudinal, radial, circumferential, and principal tangential strain/time curves for the 16 segments and peak global strain, as well as averaged peak strain at 3 LV levels (basal, mid-ventricular, and apical) ( Figure 1).
To measure afterload, total arterial stiffness (TAS) was measured by the formula: TAS = pulse pressure/stroke volume (SV); total arterial compliance (TAC) was measured by the formula: TAC = SV/pulse pressure; effective arterial elastance Z va = (SAP + MG net )/SVI, where SAP is the systolic arterial pressure, MG net is the mean net pressure gradient transvalvular pressure, and SVI is the stroke volume index. Therefore, Z va represents the valvular and arterial factors that oppose ventricular systole by absorbing the mechanical energy developed by the left ventricle (19).

Measurements
For reproducibility of 2D-and 3D-speckle-tracking echocardiographic measurements of deformation parameters, 20 patients were randomly selected and reanalyzed by the same observer to determine the intraobserver agreement and by 8 a second experienced echocardiographer, who was blinded to the initial results, to determine the interobserver agreement. Both measurements were obtained with the intraclass correlation coefficient (ICC).

Statistical Analysis
Data were presented as the mean±SD for continuous variables and as percentages for categorical variables. Agreement between parameters in 2D and 3D were assessed using the concordance correlation coefficient (CCC) with 95% CI.
Associations between 2 continuous variables were measured using the Pearson or Spearmen correlation (P) coefficient. Variability between the 2 sets of measurements was reported as the mean difference ±SD and the ICC with 95% CI.
Means were compared using a z test or t test when no SD was available for the normal-value data. Data were analyzed with JMP 10.0 software (SAS Institute Inc, Cary, North Carolina) and MedCalc statistical software, version 11.4.1.0 (MedCalc Software, Ostend, Belgium). All probability values were 2-sided, and a P value <.05 was considered statistically significant.

Parameters of 2D and 3D Echocardiography
2D and 3D echocardiographic parameters are shown in Table 2. Data from patients with AS was compared with normal values found in the literature.
Wall thickness and LV mass were greater in the study patients compared with normal values (P˂.001). The mitral inflow, tissue velocity, and early mitral inflow velocity/early diastolic mitral annular tissue velocity (E/e') ratio were significantly different between cases and normal values (P˂.001). The LVEDV, LV end-diastolic diameter (LVEDD), interventricular septal thickness in diastole (IVSD), LV posterior 9 wall diameter (LVPWD), LVESV, and the SV on 2D echocardiography, were significantly different between the 2 groups (P˂.0001). When we focused on the differences between 2D and 3D in the AS study patients, only LVEF was significantly lower when measured by 3D (P˂.001); however, it remained within normal clinical limits. The calculated indexes of LV afterload in the study patients with AS were as follows: total arterial stiffness, 0.8±0.3 mm Hg/mL; total arterial compliance,

Parameters of Speckle-Tracking Strain Imaging in 2D and 3D
The parameters of 2D-and 3D-speckle-tracking strain for both groups are shown in Table 3. For the 2D images, GLS and GCS were significantly lower for

Comparison of 3D Imaging With 2D Echocardiographic Data
The agreement of echocardiographic data between 3D and 2D images is shown in Figures 2 and 3

Imaging With Preload and Afterload
Both 2D-and 3D-GLS correlated with Z va (r=0. 34 Table 4 shows the results of the intraobserver and interobserver variability for 2D-STE (2D-speckle-tracking echocardiography) and 3D-STE measurements. Our results showed excellent correlation, with values ranging from 0.84 to 0.95 and a mean of 0.90.

Discussion
The main finding from this study was an increased basal rotation (Rot m ax -B) as well as twist (Twist m ax ) that compensates for the reduction in LV longitudinal and circumferential deformation in patients with severe AS, thus allowing the ventricle to maintain LVEF. The counter-coiled helix, which is composed of subepicardial and subendocardial fibers, generates an LV twist that has been proven to be fundamental to LV systole and, therefore, EF (22,23). The direction of LV twist is governed by the larger radial fibers at the subepicardium. Several previous studies have reported good correlation between LV twist derived from 2D-speckle-tracking echocardiography and magnetic resonance imaging (24,25). In our study, Rot m ax -B and Twist m ax in 2D images dramatically increased, a finding consistent with other reports (21,26). Possibly, subendocardial ischemia leads to a reduction of the opposing rotational forces of the subendocardial fibers, which would increase the difference in radius between the subepicardium and subendocardium. Such alterations would increase the arm of movement governed by the fibers of the subepicardium. In addition, LV hypertrophy might increase the arm force. More importantly, increased rotation and Twist m ax may be compensating for the reduction of LV deformation in the other directions in patients with AS (26). All of these potential mechanisms may theoretically also lead to increased Rot m ax -B.
Furthermore, twist was also significantly increased in AS study patients compared with normal values. Our results confirm that LV twist has an important role in LV ejection, which could explain why LVEF and cardiac output are preserved in patients with severe AS, but LV systolic function is impaired. A recent study by Musa et al (27) showed that transcatheter aortic valve implantation and surgical aortic valve replacement procedures were associated with comparable declines in rotational LV mechanics.
It is widely acknowledged that myocardial deformation on echocardiography can be described by 3 directions: longitudinal (LS), circumferential (CS), and radial (RS). LS denotes contraction of the longitudinally arranged endocardial fibers; CS denotes contraction of the circumferentially arranged mid-layer fibers; RS is defined as contraction of all the wall thickness (8,28). In patients with severe AS, the increasing afterload may lead to hypertrophy, decreased coronary perfusion, myocardial ischemia, and fibrosis. The endocardium is usually the most vulnerable to increased wall stress and stress-induced ischemia with LV pressure overload (29).
As a result, impairment of GLS usually occurs first among other strains.
The finding of decreased GLS in 2D and 3D is consistent with other reports (4,5,30). We also found that GCS on 3D echocardiography was not significantly different when compared with 2D measurements (3). In our study, 2D echocardiography-derived GCS decreased significantly, and GRS had no significant change. Delgado et al (17)  Moreover, GRS does not represent a specific set of muscle fibers, and the variation for this parameter is always greater (31). Our results confirm that GLS is consistently impaired when compared with other parameters of deformation and is compensated by an increase in twist.
In patients with degenerative AS, arterial compliance is frequently reduced, which contributes to increased afterload and decreased LV function. Hence, the left ventricle is often subjected to a double afterload-from valvular obstruction and from the systemic arterial system (1,2,18,32). Z va , a simple index proposed by Briand et al (1), provides an estimate of the global hemodynamic load imposed on the left ventricle and is an important index of AS severity and a predictor of LV dysfunction. We found that Z va was moderately elevated in patients with severe AS (33). A study by Pagel et al (34) showed that aortic valve replacement affected only the valvular component of afterload and had no effect on arterial compliance in elderly patients, which suggests that other comorbidities, such as hypertension and 13 atherosclerosis, may have an important impact on this parameter.
Our study showed a correlation between 2D-and 3D-GLS and Z va but no significant relationship between Z va and other deformation parameters (GRS, GCS).
We also found that impaired GLS, both in 2D and 3D, correlated with increasing LVEDV and E/e¢ ratio. Sato et al (35) had similar findings; diastolic dysfunction was present in their patients, although their study patients had low-flow, low-gradient severe AS. The increase in Z va , combined with the increased LVEDV, reflected a large global hemodynamic overload. According to the Laplace law, patients with severe AS are likely to have markedly increased wall stress, which may lead to depressed myocardial contractility. Maréchaux et al (19) reported similar results, confirming that LV longitudinal contraction is, in large part, determined by LV preload and afterload.
Previous studies have reported conflicting results of comparisons of 2D-and 3D-STE measurements, possibly because of major differences in the study populations (sample sizes and the severity of AS) and methodology (ie, software) (3,15,36,37). In a study by Altman et al (16), 3D-STE was not shown to be superior to 2D-STE for any of the 3 components of LV deformation. In our study, we found a modest agreement between 2D-GLS and 3D-GLS assessed with CCC; agreement of other 2D and 3D strain parameters was poor. We also observed a similar correlation between 2D-and 3D-GLS and Z va . Theoretically, 3D-STE should be more accurate than 2D-STE because 3D-STE can overcome well-known limitations of 2D-STE: 3D images can avoid foreshortening of apical views, and 3D images are able to give a more complete picture of myocardial deformation in 3 dimensions and, therefore, eliminate, to an extent, the problem of out-of-plane motion, which may affect the 14 accuracy of LV strain and twist measurements with 2D echocardiography. However, the lower temporal and spatial resolutions of 3D images are potential limitations and could adversely affect the accuracy of 3D-STE measurements in patients with 3D images at the lower frame rates (36,37).

Conclusions
In patients with severe AS, GLS is consistently compromised, and LVEF and cardiac output are maintained by increased basal rotation and twist. 3D-STE is comparable to 2D-STE. 2D-GLS correlated modestly with Z va . No significant correlation between Z va and other deformation parameters (i.e., GRS, and GCS) was observed.

Limitations
This was a single-center, single-ethnicity, observational study. We also did not validate deformation measurements against reference standards, including tagged magnetic resonance imaging or sonomicrometry. The relatively low frame rate of real-time 3D echocardiographic imaging could potentially lead to underestimating strain values. In addition, the relatively high body mass index in our cohort may have led to poor image quality. It is widely acknowledged that the accuracy of both

Consent for publication
17 not applicable.

Availability of data and materials
The datasets generated and/or analyzed during the current study are not publicly available because the information and data of the study population were extracted from Hospital Information System and were recorded manually in EXCEL to form our private database. But the data are available from the corresponding author on reasonable request.

9.
Fine  Abbreviations: AS, aortic valve stenosis; NT-proBNP, N-terminal fragment of the prohormone brain natriuretic peptide; NYHA, New York Heart Association.
a Data are expressed as mean±SD or as percentages unless otherwise indicated.     Agreement Between Parameters of Conventional Echocardiography in 2-dimensional (2D) and