Author + information
- Received January 29, 2014
- Revision received April 18, 2014
- Accepted April 30, 2014
- Published online November 11, 2014.
- Richard Ro, MD∗,
- Dan Halpern, MD∗,
- David J. Sahn, MD†,
- Peter Homel, PhD∗,
- Milla Arabadjian, NP∗,
- Charles Lopresto, BA∗ and
- Mark V. Sherrid, MD∗∗ ()
- ∗Division of Cardiology, Mount Sinai Roosevelt and St. Luke's Hospitals, Icahn School of Medicine at Mount Sinai, New York, New York
- †Division of Pediatric Cardiology, Oregon Health and Science University, Portland, Oregon
- ↵∗Reprint requests and correspondence:
Dr. Mark V. Sherrid, Division of Cardiology, Mount Sinai Roosevelt Hospital, 1000 10th Avenue, 3B-30, New York, New York 10019.
Background The hydrodynamic cause of systolic anterior motion of the mitral valve (SAM) is unresolved.
Objectives This study hypothesized that echocardiographic vector flow mapping, a new echocardiographic technique, would provide insights into the cause of early SAM in obstructive hypertrophic cardiomyopathy (HCM).
Methods We analyzed the spatial relationship of left ventricular (LV) flow and the mitral valve leaflets (MVL) on 3-chamber vector flow mapping frames, and performed mitral valve measurements on 2-dimensional frames in patients with obstructive and nonobstructive HCM and in normal patients.
Results We compared 82 patients (22 obstructive HCM, 23 nonobstructive HCM, and 37 normal) by measuring 164 LV pre- and post-SAM velocity vector flow maps, 82 maximum isovolumic vortices, and 328 2-dimensional frames. We observed color flow and velocity vector flow posterior to the MVL impacting them in the early systolic frames of 95% of obstructive HCM, 22% of nonobstructive HCM, and 11% of normal patients (p < 0.001). In both pre- and post-SAM frames, we measured a high angle of attack >60° of local vector flow onto the posterior surface of the leaflets whether the flow was ejection (59%) or the early systolic isovolumic vortex (41%). Ricochet of vector flow, rebounding off the leaflet into the cul-de-sac, was noted in 82% of the obstructed HCM, 9% of nonobstructive HCM, and none (0%) of the control patients (p < 0.001). Flow velocities in the LV outflow tract on the pre-SAM frame 1 and 2 mm from the tip of the anterior leaflet were low: 39 and 43 cm/s, respectively.
Conclusions Early systolic flow impacts the posterior surfaces of protruding MVL initiating SAM in obstructive HCM.
- hypertrophic cardiomyopathy
- hypertrophic obstructive cardiomyopathy
- LVOT obstruction
- vector flow map
Systolic anterior motion of the mitral valve (SAM) was discovered in the late 1960s as the most common cause of obstruction in hypertrophic cardiomyopathy (HCM) (1). Although its hydrodynamic cause is controversial, there is agreement that SAM is caused by an interaction between left ventricular (LV) flow and the elongated mitral valve (MV). Investigators initially hypothesized that a Venturi effect, caused by narrowing of the outflow tract, and high flow velocities there might draw the MV anteriorly. However, others have shown that there are low velocities present in the outflow tract when SAM begins, precluding significant Venturi forces at this time (2,3) (Online Figure 1). This observation, combined with other observations regarding the geometry of the LV and MV in obstructive HCM, has led other investigators to aver that the pushing force, flow drag, sweeps the leaflets into the septum (2–5). Color flow strikes the posterior aspect of the mitral valve leaflets (MVL) with an increasing angle of attack as SAM progresses, consistent with a flow drag mechanism. Nevertheless, the controversy persists, with recent reference to the Venturi mechanism (6–8). Investigators observed that, in some patients, SAM may begin in isovolumic systole, before the aortic valve opens (2,3). This movement is too early to be explained by ejection flow or by Venturi forces. Because the mechanism of SAM has important therapeutic implications, we studied the interaction of early systolic LV flow with the MV using a new echocardiographic method, vector flow mapping (VFM) (9–15) (Online Refs. 7–12).
Between October 2011 and October 2012, echocardiograms in 3 patient groups referred for outpatient echocardiography for clinical indications were compared: obstructive HCM, nonobstructive hypertrophic cardiomyopathy (nonobHCM), and normal control patients. Patients were diagnosed and included in the HCM group if they had segmental hypertrophy ≥15 mm and a nondilated LV without a hemodynamic or clinical cause for the degree of thickening observed. Consecutive obstructive HCM patients were selected if they had systolic anterior motion (SAM) of the MV, mitral-septal contact, and a resting left ventricular outflow tract (LVOT) gradient ≥30 mm Hg. Five obstructed patients were excluded for technical reasons. NonobHCM patients had no SAM or resting gradients. Patients with completely normal echocardiographic examinations were recruited as a control group. Patients were excluded if they had mid-LV obstruction or papillary muscle apposition with the septum, poor windows, arrhythmia, or latent obstruction. Patients from all 3 groups signed informed consent approved by our institutional review board.
Image acquisition and measurements
Routine M-mode, 2-dimensional (2D), and Doppler imaging were performed on a Hitachi-Aloka Alpha 7 System (Wallingford, Connecticut) and recorded on the ProSolv (FujiFilm Medical Systems, Stamford, Connecticut) archiving system. Conventional measurements were performed as previously described (Online Ref. 4) and are shown in Online Figure 2.
Vector flow mapping
VFM is a novel method of processing Doppler information that demonstrates the vector of local blood flow velocity in intravascular structures (9–15) (Online Refs. 7–12). It differs from routine color Doppler: in VFM, a post-processing computational algorithm extracts information from the distribution of Doppler color flow in the beam direction, estimates the radial (perpendicular) component of the flow distribution, and displays it without angle dependency. It is able to demonstrate the direction and magnitude of blood flow velocity over 360° (Figures 1 and 2⇓⇓). To achieve a desirable frame rate for VFM imaging, color flow sector was reduced in size. 2D gain was set to ∼50 dB and flow gain to ∼60 and below artifact level. The color Nyquist limit was set to 40 to 50 cm/s to detect low velocities associated with mitral inflow and its vortices. Color Doppler loops were acquired in apical 3- and 5-chamber and parasternal views, primarily saving VFM loops of the apical 3-chamber view, as this was expected to be the most informative view.
Loop and image processing
See the Online Appendix.
In patients with obstructive HCM, in early systole, the frame before initial SAM of the MV was designated pre-SAM, and the frame after as post-SAM. The durations from the peak of the R-wave were calculated. For both moments, we saved reference 2D frames (without color Doppler) and frames with color Doppler both before and after aliasing correction. In the control patients and in patients with nonobHCM, frames at times corresponding to those of the obstructive patients were used for comparisons.
Ejection flow or isovolumic vortical flow
Velocity vector display was enabled for pre- and post-SAM frames. With this function, we determined the direction and velocity of the flow directly impacting the MVL tip at the moments before and after SAM began. We determined whether the flow striking the posterior surface of the MVL was ejection flow, isovolumic vortical flow, or a confluence of the 2 acting on the leaflet tip.
Color flow posterior to the leaflets
We used the alias-corrected color flow maps and the corresponding velocity vector map to assess the presence of flow posterior to the MVL. In ejection SAM cases, this is the flow width that overlapped the MV apparatus, thus appearing posterior to the protruding leaflets in the cul-de-sac between the posterior leaflet and the LV posterior wall. In SAM associated with the isovolumic vortex, it is the anteriorly-directed limb of the clockwise isovolumic vortex.
Diameter LVOT and flow width in the LVOT
In frames showing ejection SAM, a line was drawn from the interventricular septum to the tip of the anterior MV; this was the anatomic LVOT width. A parallel line was then drawn 0.5 cm above the first line, from the septum to where the velocity vectors approached 0; this is the width of the flow approaching and immediately upstream of the LVOT. Overlap of the vector velocity flow approaching the LVOT, and the LVOT itself, was the difference between these widths.
Angle of attack onto the leaflet
The velocity vector map was used to measure the angle of attack onto the posterior surface of the protruding MVL in pre- and post-SAM frames. To measure the angle, a line was drawn through the point of coaptation closest to the annulus and through the anterior leaflet tip. A second line was then drawn parallel to the velocity vector of flow acting locally, just upstream from the tip of the leaflet; the angle between the lines was measured. The technique was similar to that reported previously with color Doppler flow (4) (Central Illustration).
Ricochet vector flow was defined as the reflection of blood flow after impact with the protruding MV that rebounds posteriorly into the cul-de-sac behind the posterior leaflet. To be characterized as ricochet, vectors must have reciprocal reflection angles equal, by inspection, to the angle of attack described above.
Vortex quantitative measurements
For information regarding the vortex quantitative measurements, please see the Online Appendix.
Velocity in the LVOT
To assess the possible contribution of Venturi forces on the MV tip, we measured local velocities in the LVOT just anterior to the tip of the anterior leaflet at 1 and 2 mm anterior to the tip (13) (Online Ref. 12).
Continuous variables were described as mean ± SD; categorical variables were described by percentage. Tests of group differences of continuous data were performed using 1-way analysis of variance. Omnibus F-tests were used to detect differences between groups, whereas mean contrast tests using the error estimate from the analysis of variance were used for pairwise group differences in the case of any statistically significant F-test. Chi-square tests were used for omnibus differences between groups of categorical data, and pairwise group differences were tested for any significant chi-square test using logistic regression. A Bonferroni level of significance <0.017 was used for each pairwise group test to correct for multiple comparisons. The p values for pairwise comparisons are presented only for outcomes where the omnibus test was significant. Pearson correlations were calculated to determine relationships between maximal vortex variables and transmitral Doppler A waves. Intraobserver and interobserver agreement were assessed for the angle of attack of vector flow onto the MVL in pre- and post-SAM frames. A single physician repeating 10 angle measurements at 2 different times assessed intraobserver agreement; 2 physicians repeating 10 angles independently assessed interobserver agreement. Intraobserver and interobserver agreement were measured by calculating an intraclass correlation coefficient (ICC) for each assessment. On the basis of criteria established by Nunnally (16), an ICC >0.80 represents good agreement, whereas an ICC >0.90 represents excellent agreement. SAS version 9.1 (SAS Institute Inc., Cary, North Carolina) was used for statistical analyses.
We compared 82 patients (22 obstructive HCM, 23 nonobHCM, and 37 normal) by measuring 164 pre- and post-SAM velocity vector displays, 82 maximum isovolumic vortices, and 328 2D frames.
In the obstructive HCM patients, LV resting outflow gradients were 65 ± 28 mm Hg. Obstructive HCM patients had greater septal and maximal wall thickness and longer anterior and posterior leaflet lengths, protrusion height, and residual leaflets compared with nonobHCM and control patients (Table 1). Obstructive HCM patients had less distance from between the coapted MVL to the anterior septum, and greater distance from the short-axis papillary muscle plane to the posterior wall.
Velocity vector display maps
VFM frame rates were 38.5 ± 10/s and did not differ between the 3 groups. In obstructive HCM patients, the pre-SAM frame and post-SAM frames occurred 98.6 ± 22 ms and 125.4 ± 34 ms after the R-wave, respectively. By study design, there was no difference in the timing of the measured frames between the 3 patient groups. In control patients, LV ejection flow did not overlap the MV and, in an orderly fashion, flow remained in the outflow tract throughout systole (Figures 1 and 2, Table 2).
We observed color flow posterior to and impacting the MVL in the early systolic frames in 95% of obstructive HCM patients, 22% of the nonobstructive patients, and 11% of the control subjects (p < 0.001). In the obstructive HCM patients, early velocity vector flow overlapped the anatomic LVOT by 8 mm in the pre-SAM frame and by 12 mm in the post-SAM frame. The overlap between LV flow and the MV was of 2 types: it was ejection flow in 13 (59%) patients and isovolumic vortical flow in 9 (41%). In ejection SAM cases, the bulging septum initially baffles ejection flow posteriorly. As flow courses around the septum, it comes from a relatively posterior direction towards the outflow tract, impacting the MV on its posterior surface (Figures 3 and 4⇓). In vortical SAM patients, the coapted MVL protrude through the center of the clockwise isovolumic LV vortex and into the flow coursing from a posterior to anterior direction in that limb of the vortex (Figures 5 to 7⇓⇓). When either ejection or vortical flow initiates SAM, in both the pre- and post-SAM frames, vector flow impacts the posterior surface of the leaflets with a high angle of attack >60°. Angles of attack of anteriorly-directed flow onto the posterior MVL surface were 67° in the pre-SAM beat and 64° in the post-SAM beat. Angle of attack was not measured for the nonobHCM and normal patients because there was generally no flow posterior to the leaflets. Ricochet of the posterior flow off the leaflets and into the cul-de-sac was noted in obstructed patients on 54.5% of the pre-SAM beats and on 82% of the post-SAM beats. It was noted in 9% of the nonobHCM patients and in none of the controls (p < 0.001).
Flow velocity in the outflow tract at SAM onset
In the obstructive HCM patients, flow velocities in the LVOT on the pre-SAM frame 1 and 2 mm from the tip of the anterior leaflet were 39 and 43 cm/s, respectively.
Isovolumic systolic vortical flow
Vortex area was smaller in the HCM groups combined than in normal patients (p = 0.002) (Table 3). The longitudinal and anteroposterior position of the center of maximal and pre-SAM vortices in the LV did not differ between the 3 groups. There was correlation in the whole group of 82 patients between the transmitral Doppler A-wave velocity and the maximum vortex flow rate, r = 0.48 (p < 0.0001); there was no correlation between any vortex measurement and transmitral E-wave or anterior leaflet length.
For the angle of attack of velocity vector flow onto the posterior surface of the MVL, intraobserver variability was excellent (ICC coefficient of 0.94), and interobserver variability was good (ICC coefficient of 0.82).
In this study, we observed that when SAM begins in obstructive HCM, there is LV flow that impacts the posterior aspect of the MVL tip, causing the initial anterior motion. Early systolic velocity vector flow overlaps the anatomic LVOT by 8 mm in the pre-SAM frame and by 12 mm in the post-SAM frame. Overlap of early systolic flow and the MV was not observed in normal control subjects and rarely observed in nonobstructed patients. Moreover, the anteriorly-directed flow impacts the protruding leaflet tip with angles of attack of >60° before and after the beginning of SAM. Even in airfoils specifically designed for lift, lifting force declines after the angle of attack exceeds 15° (17). At the angles of attack in the frames before and after SAM onset, drag, the pushing force of flow, is the dominant hydrodynamic force on the leaflets (Central Illustration). In obstructed patients, there is an important contribution to drag from the abnormally elongated MV that protrudes higher into the LV compared with nonobstructed and control patients, with protrusion heights of 2.6, 1.9, and 1.3 cm, respectively (18) (Online Refs. 4,13), as well as from the previously described abnormal valve shape and slack (2,19,20).
Despite this similarity, there is otherwise heterogeneity in the nature of the flow that begins SAM. In 59% of patients, the early anterior motion is caused by ejection flow sweeping around the bulging septum, overlapping with and striking the MVL on their posterior surfaces. In the other 41%, we observed that SAM was initiated early in systole by the anteriorly-directed isovolumic vortex.
SAM beginning during ejection
The abnormal overlap flow was seen on the native color Doppler and velocity vector displays, appearing posterior to the leaflets in the cul-de-sac (Figures 3 and 4). In ejection SAM, the high angle of attack is set up by incompressible flow that first is deflected posteriorly by the septal bulge and then must course from a posterior to anterior direction. En route to the outflow tract, it catches the posterior surface of the anteriorly-positioned protruding MV, with a resulting high angle of attack. Although drag on the MV in obstructive HCM has been observed previously with conventional color Doppler (3,4) and in experimental preparations (20), velocity vector mapping provides greater resolution, clarity, and simultaneity. Ricochet of blood into the cul-de-sac, which rebounds with a reciprocal angle to the angle of attack, is consistent with the flow drag mechanism of SAM. It is not seen in normal control or nonobstructed patients; it is a novel observation of this work.
SAM beginning during isovolumic vortical flow
LV isovolumic vortical flow is a well-studied consequence of the interaction between late diastolic LV filling and the MVL (21,22). It has been analyzed with cardiac magnetic resonance (CMR) (14,15,21,22), particle imaging velocimetry (PIV) (23,24), and VFM (9–11,13). We previously noted that SAM may begin in isovolumic systole in close temporal relation with the “atrial reflected wave,” which is the component of the anteriorly-directed isovolumic vortex in the LVOT, shown in Online Figure 1 (3,25). Thus, we suspected that this flow might cause SAM in some patients. In the vortical SAM patients, we observed that the tip of the elongated MV protrudes into and through the center of the vortex, extending into the anteriorly-directed component of the vortical flow. The angle of attack was always high, implicating flow drag (Figures 5 to 7). In several patients, we observed a confluence of the vortical vectors with very early ejection vectors in the LV on the post-SAM frame. In these cases, vortical SAM pre-positions the leaflets into the ejection stream, where they are impacted by ejection flow (Figure 7).
Flow velocity in the outflow tract at SAM onset
In the obstructive HCM patients, flow velocities in the LVOT on the pre-SAM frame 1 and 2 mm from the anterior leaflet tip were low: 39 and 43 cm/s, respectively. Both lift and drag are generated whenever flow traverses a surface such as the MV. Pertinent to the issue here, the lift-to-drag ratio declines with decreasing velocity (17,26,27). Thus, the Venturi contribution to SAM is very small.
A cause of failed myectomy has been surgical resection restricted to the subaortic outflow tract, tailored to relieve supposed Venturi forces in the LVOT (28,29) (Online Figure 3). Recognition that drag is the dominant force for SAM has led to an altered surgical approach, with the excision extended further down into the LV cavity when needed because of midseptal thickening. This was termed an extended myectomy (30,31), designed to redirect flow away from the MV to reduce drag forces on the leaflets (28). In addition to myectomy, papillary muscle release allows the MV to drop down posteriorly into the cavity, explicitly separating the inflow and outflow portions of the LV (31). After alcohol ablation, investigators found that 78% of patients had persistent SAM and mitral regurgitation, even when gradient reduction was considered successful. Resistant SAM resulted because the MVL were anteriorly interposed into the ejection flow stream (32). DDD pacing with short AV delay for gradient reduction has unpredictable results (5). RV pacing alters the pattern of isovolumic flow by PIV; the effect of pacing on isovolumic vortical flow might offer insights to improve this therapy (33).
Velocity flow mapping
In VFM, the spatial distribution of Doppler velocities in the flow field allows calculation of the component of velocity orthogonal to the beam direction. The orthogonal velocities are calculated with a computational algorithm that employs the stream function of fluid mechanics, as described previously (Online Refs. 7–9,16). Previous work with VFM has validated good accuracy in comparison with computer-generated flow simulation (Online Ref. 9), in vivo animal experiments (13), and an ability to delineate the spatial features and temporal evolution of LV flow in patients with a wide variety of pathologies (9–12) (Online Refs. 10–12). Our prior work with conventional color Doppler (4) was limited in several ways that are improved by using VFM, including better resolution and spatial orientation, especially of low-velocity vectors. This allowed us to image vortical and ricochet flow and their relation to the valve. Moreover, VFM allows simultaneous imaging of the flow and the valve leaflets, which was previously difficult to achieve. In comparison to PIV, VFM has the advantage of not requiring intravenous contrast injection, where PIV microbubbles obscure the exact position of the MV tip, thus interfering with simultaneous assessment of flow and the valve. CMR flow visualization is generated from averaged values over many cardiac cycles, not from single beats, and has limited spatial resolution.
In normal patients, the anteriorly-directed isovolumic vortex is thought to preserve the momentum of late transmitral flow into the LVOT during the transition from diastole to systole (21). We found a strong correlation between the transmitral A-wave and the maximum vortex flow rate. Thus, when LV ejection begins, blood is already moving toward the aortic valve. This “running start” of LV ejection flow could offer efficiency, especially during exercise; it is thought to explain the adaptation of the d-loop, which is highly conserved in mammalian evolution. Although the system normally works well, it has a low tolerance for anatomic error; centimeter alterations in septal and MV anatomy can lead to overlap in the inflow and outflow portions of the LV, resulting in SAM, either from the vortical flow itself or from ejection.
A slack MV, caused by loss of restraint from anteriorly-positioned papillary muscles and elongated leaflets, permits SAM (2,18,20). In obstruction, we found increased MVL length, protrusion height, residual leaflet length, and distance from the posterior wall to the plane of the papillary muscles, similar to previous reports (2,18–20). In the present study, the relevance of these abnormalities to SAM is demonstrated by VFM; the elongated MV extends into anteriorly-directed flow. MV plication, performed at myectomy in selected patients, has been introduced to decrease the anterior leaflet’s length and stiffen it (31).
Later in systole
After SAM begins, the MV is swept toward the septum, which increases the angle of attack onto the leaflets (to 40° to 45° in a previous publication ) and increases form drag to such an extent that drag is unquestionably the dominant force later in systole. This is why we focused on the earliest SAM in the present study. After mitral-septal contact, the pressure gradient itself pushes the leaflet further into the septum (4).
VFM is a new technique. The potential artifacts and pitfalls of its use have not yet been fully elaborated. It is possible that our observations are susceptible to artifact. However, VFM has been validated with good accuracy (13) (Online Ref. 9), and it delineates the shape and temporal evolution of LV flow similar to observations with CMR and PIV (9–15) (Online Refs. 9,14,15). As such, we believe it unlikely that our observations about the relation of ejection, vortical, and ricochet flow to the MV are all artifactual. Besides, VFM offers the only credible explanation to date of the hitherto mysterious phenomenon of pre-ejection SAM (2,3) (Online Figure 1). In VFM frames, vortical flow may appear to traverse the protruding MVL. Direct observations in the operating room demonstrate that the protruding leaflet is usually only ≤1-cm wide for its entire length above the coaptation plane. Thus, we are imaging flow passing on either side of the thin protruding leaflet. We, therefore, believe that vortical flow has duration, despite its association with the MV. Isovolumic vortical flow is a 3-dimensional shape, likely toroidal; we selected the 3-chamber view because we thought it would provide the most information regarding interaction between the flow field and MV. Other observations could be made from orthogonal views, but were beyond the scope of this study, as were factors that differentiate ejection SAM from vortical SAM, and factors that determine maximal vortex velocity.
Using VFM, we observed that early systolic flow impacts the posterior surfaces of protruding MVL, causing SAM in obstructive HCM. Two kinds of flow interaction with the MV were observed to initiate the anterior motion: ejection flow and isovolumic vortical flow.
COMPETENCY IN MEDICAL KNOWLEDGE: VFM echocardiography demonstrates that the cause of systolic anterior motion in patients with obstructive HCM, rather than Venturi forces, is early systolic ejection flow (60% of patients) or isovolumetric vortical flow (40%) pushing the MVL toward the interventricular septum.
TRANSLATIONAL OUTLOOK: Prospective studies are needed to assess the clinical outcomes of surgical approaches that facilitate posterior leaflet descent, plicate and shorten the anterior leaflet, and reposition the papillary muscles alone and in combination with myectomy and to compare these to isolated myectomy or alcohol septal ablation procedures.
The authors appreciate the technical expertise of Karina Vaysfeld, RCDS, who acquired the images in this study.
Hitachi-Aloka, Ltd., provided the echocardiography machine used in this research. The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Steven Nissen, MD, served as the Guest Editor for this paper.
- Abbreviations and Acronyms
- cardiac magnetic resonance
- hypertrophic cardiomyopathy
- left ventricular
- left ventricular outflow tract
- mitral valve
- mitral valve leaflets
- nonobstructive hypertrophic cardiomyopathy
- particle imaging velocimetry
- systolic anterior motion of the mitral valve
- vector flow mapping
- Received January 29, 2014.
- Revision received April 18, 2014.
- Accepted April 30, 2014.
- American College of Cardiology Foundation
- Shah P.M.,
- Gramiak R.,
- Kramer D.H.
- Sherrid M.V.,
- Gunsburg D.Z.,
- Moldenhauer S.,
- et al.
- Sherrid M.V.,
- Chu C.K.,
- Delia E.,
- et al.
- Gersh B.J.,
- Maron B.J.,
- Bonow R.O.,
- et al.
- ↵McKenna WJ. Types and pathophysiology of obstructive hypertrophic cardiomyopathy. 2013. Available at: http://www.uptodate.com/contents/types-and-pathophysiology-of-obstructive-hypertrophic-cardiomyopathy. Accessed August 30, 2014.
- Ommen S.R.
- Nogami Y.,
- Ishizu T.,
- Atsumi A.,
- et al.
- Rodriguez Munoz D.,
- Markl M.,
- Moya Mur J.L.,
- et al.
- Nunnaly J.C.
- Fox R.,
- McDonald A.T.
- Klues H.G.,
- Proschan M.A.,
- Dollar A.L.,
- et al.
- Levine R.A.,
- Vlahakes G.J.,
- Lefebvre X.,
- et al.
- Kheradvar A.,
- Pedrizzetti G.
- Sengupta P.P.,
- Pedrizetti G.,
- Narula J.
- Pai R.G.,
- Suzuki M.,
- Heywood J.T.,
- et al.
- Vogel S.
- Vogel S.
- Sengupta P.P.,
- Khandheria B.K.,
- Korinek J.,
- et al.