Author + information
- Received July 11, 2018
- Revision received November 25, 2018
- Accepted November 28, 2018
- Published online March 13, 2019.
- Aniela Petrescu, MDa,∗,
- Pedro Santos, PhDb,∗,
- Marta Orlowska, MScb,
- João Pedrosa, PhDb,
- Stéphanie Bézy, MSca,
- Bidisha Chakraborty, MScb,
- Marta Cvijic, MD, PhDa,
- Monica Dobrovie, MDa,
- Michel Delforge, MD, PhDc,
- Jan D’hooge, PhDb,† and
- Jens-Uwe Voigt, MD, PhDa,†∗ ()
- aDepartment of Cardiovascular Science, Division of Cardiology, University Hospital Leuven, University of Leuven, Leuven, Belgium
- bDepartment of Cardiovascular Science, Cardiovascular Imaging and Dynamics, University Hospital Leuven, University of Leuven, Leuven, Belgium
- cDepartment of Hematology, University Hospital Leuven, University of Leuven, Leuven, Belgium
- ↵∗Address for correspondence:
Prof. Dr. Jens-Uwe Voigt, University of Leuven and University Hospitals Leuven, Herestraat 49, 3000 Leuven, Belgium.
Objectives This study sought to evaluate whether velocity of naturally occurring myocardial shear waves (SW) could relate to myocardial stiffness (MS) in vivo.
Background Cardiac SW imaging has been proposed as a noninvasive tool to assess MS. SWs occur after mechanical excitation of the myocardium (e.g., mitral valve closure [MVC] and aortic valve closure [AVC]), and their propagation velocity is theoretically related to MS, thus providing an opportunity to assess stiffness at end-diastole (ED) and end-systole. However, given that SW propagation in vivo is complex, it remains unclear whether natural SW velocity effectively relates to MS.
Methods This study prospectively enrolled 50 healthy volunteers (HV) (43.7 ± 17.1 years of age) and 18 patients with cardiac amyloidosis (CA) (68.0 ± 9.8 years of age). HV were divided into 3 age groups: group I, 20 to 39 years of age (n = 24); group II, 40 to 59 years of age (n = 11); and group III, 60 to 80 years of age (n = 15). Parasternal long-axis views were acquired using an experimental scanner. Tissue (Doppler) acceleration maps were extracted from an anatomical M-mode along the midline of the left ventricular septum.
Results SW propagation velocity was significantly higher in CA patients than in HV after both MVC (3.54 ± 0.93 m/s vs. 6.33 ± 1.63 m/s, respectively; p < 0.001) and AVC (3.75 ± 0.76 m/s vs. 5.63 ± 1.13 m/s, respectively; p < 0.001). Similarly, SW propagation velocity differed significantly among age groups in HV, with a significantly higher value for group III than for group I, both occurring after MVC (p < 0.001) and AVC (p < 0.01). Moreover, SW propagation velocity after MVC was found to be significantly higher in patients with an increasing grade of diastolic dysfunction (p < 0.001). Finally, positive correlation was found between SW velocities after MVC and mitral inflow-to-mitral relaxation velocity ratio (E/E′) (r = 0.74; p = 0.002).
Conclusions End-diastole SW velocities were significantly higher in patients with CA, patients with a higher grade of diastolic dysfunction, and elderly volunteers. These findings thus suggest that the speed of naturally induced SWs may be related to MS.
Several noninvasive imaging modalities provide information for cardiac morphology and function. Conventional echocardiographic parameters allow quantification of systolic and diastolic functions based on the visualization and interpretation of changes in volumes and velocity measurements of the blood (1,2). Cardiac magnetic resonance (CMR) plays a complementary role in identifying underlying causes of cardiac diseases and represents the gold standard modality for the assessment of chamber volumes and ejection fraction (2,3). However, none of these modalities can directly measure cardiac relaxation and contractility. Left ventricular (LV) myocardial stiffness (MS) gives information about forces acting inside the myocardium, thus allowing a more direct assessment of its systolic and diastolic properties (4–6). To date, direct assessment of MS can only be provided invasively by pressure-volume loops (7). However, due to the limited feasibility of this method for routine assessment, clinicians still rely on the interpretation of surrogate parameters such as ejection fraction (2) or global longitudinal strain (8) for systolic function and complex decision making algorithms for diastolic function (1).
Recently, cardiac shear wave (SW) imaging was introduced as a noninvasive method for detecting changes in myocardial elasticity and contraction during the cardiac cycle (9,10). This technology requires an initial mechanical stimulus, which can be either external (e.g., acoustic radiation force  or harmonic vibration ) or internal (e.g., closure of the valves) (13,14). Local tissue perturbation then creates a SW that propagates throughout the myocardium with a speed that is dependent on the tissue stiffness and typically ranges from 1 to 10 m/s (10). Current clinical ultrasonography machines are still limited to a temporal resolution of typically ∼50 to 100 frames per second for conventional B-mode imaging, which is sufficient to assess morphology and some aspects of cardiac function but cannot resolve short-lived cardiac mechanical events such as SWs. Therefore, SW imaging requires high frame rate (HFR) imaging methodologies, typically available in research scanners only. Diverging wave sequences particularly use very broad transmission beams and thus allow reconstructing a frame from a few transmissions only, therefore offering up to a few thousand frames per second (15).
Although assessing velocities of externally induced SWs has recently been investigated in both human (16) and animal models (17), obtaining quantitative information about the absolute stiffness from the SW velocity alone remains challenging. First, the SW propagation speed depends not only on MS but may depend also on myocardial wall thickness, viscosity, or ventricular loading conditions (18). Second, SW wavelengths are relatively large compared to the septal dimensions, thereby possibly affecting the relationship between MS and SW speed as the SW cannot be considered to travel on an infinite medium but becomes a guided one instead (19). Third, because the SW velocity depends on local anisotropy (i.e., the wave propagates faster along the direction of the fiber than across it), the sampling location and direction can influence the resulting estimate (10). Altogether, the relationship between SW speed and MS may thus be altered in the clinical setting, complicating the interpretation of the SW speed measurement. The above-described behavior applies equally to SWs naturally induced in the septal wall due to mitral valve closure (MVC) at end-diastole (ED) and aortic valve closure (AVC) at end-systole (ES). To date, these natural SWs have been studied only in animal models and healthy volunteers (HV) (13,20,21), but the ability to detect changes in stiffness and, thus, to detect disease remains unclear.
The present study therefore aimed at evaluating whether the velocity of naturally occurring myocardial SWs relates to MS. Hereto, selected patient populations with plausibly increased MS were selected and compared to healthy controls. In particular, cardiac amyloidoses (CA) are a group of infiltrative disorders caused by deposition of amyloid fibrils in the myocardial interstitium (22). The storage of this highly rigid material in vivo is likely to stiffen cardiac tissue (23), leading to a rapid deterioration in tissue elasticity and resulting in increased LV stiffness (24). LV stiffness is also known to progressively increase with age due to changes in structure of the heart (25). Therefore, these groups of patients are ideal for the assessment of MS in vivo based on measurement of SW speed with HFR Imaging.
This study was prospectively conducted at the University Hospital Gasthuisberg, Leuven, Belgium. Eighteen patients with CA who had been diagnosed and followed-up in the Departments of Hematology and Cardiovascular Diseases from January 2018 until June 2018, were screened for enrollment in the study. Twelve of the 18 patients underwent right ventricular biopsy, which confirmed CA. The remaining patients showed a positive extracardiac biopsy result together with echocardiographic and/or CMR findings typical for CA. Subjects with a dual pacemaker, left bundle branch block, or atrioventricular block greater than first degree were excluded. To avoid potential wrong interpretations of echocardiographic estimates of filling pressure due to overfilling, care was taken to ensure that all patients were clinically compensated, without LV dilation and with not more than mild mitral regurgitation.
Furthermore, 50 HV were recruited who were divided into 3 age groups: group I was 20 to 39 years of age (n = 24); group II was 40 to 59 years of age (n = 11); and group III was 60 to 80 years of age (n = 15). HV-specific exclusion criteria were history of heart disease, systolic blood pressure >140 mm Hg, diastolic blood pressure >90 mm Hg, cardiac arrhythmia, more than mild valvular disease, reduced LV systolic function (ejection fraction <50%), and poor echocardiographic window.
This study was approved by the Ethics Committee of the University Hospital Leuven, and all participants gave written informed consent before inclusion.
Standard echocardiography: acquisition and analysis
In all participants, standard echocardiographic acquisitions were performed with a Vivid E9 system (GE Vingmed Ultrasound, Horten, Norway). These data were analyzed offline by using EchoPAC software version 202 (GE Vingmed Ultrasound, Horten, Norway). Left heart dimensions and function were assessed according to recent guidelines (2). Diastolic function was determined according to current guideline recommendations (normal grade 0 or abnormal grades 1 to 3) (1). Strain measurements were performed using speckle tracking echocardiography.
SW imaging: acquisition and analysis
SW imaging data were acquired using a fully programmable experimental scanner (High channel Density Programmable ULtrasound System [HD-PULSE]) equipped with a clinical phased array transducer (Medison model P2-5AC, Samsung, Suwon, Gyeonggi-do, South Korea) (26). The protocol started with conventional imaging (i.e., low frame rate) to maximize image quality and therefore to facilitate finding a good echocardiographic view. Upon obtaining a parasternal long-axis view of the heart, a fast imaging sequence was loaded, and raw channel data were streamed to a disk along with electrocardiographic data. HFR imaging (1,150 ± 245 frames per second) was achieved with diverging transmission waves, followed by coherent compounding of the echoes (27). Data reconstruction and further processing was performed using Matlab 2017a software (MathWorks Inc., Natick, Massachusetts).
Tissue velocity was estimated between every pair of frames by using a regular Doppler autocorrelation method, and tissue acceleration was subsequently obtained by temporal differentiation of the velocity data. Both tissue velocity and acceleration were thus provided at the same frame rate as the B-mode images. Anatomical acceleration M-mode maps were extracted along the midline of the LV septum, offering a spatiotemporal representation of the local tissue acceleration throughout the cardiac cycle.
The SWs created after the valve closure events could be perceived as a transient vibration starting at the valve insertion point and propagating toward the apex and are thus depicted on the anatomical color M-modes as tilted color bands. The waves of interest were therefore identified after AVC (ES) and MVC (ED) timing, and their slopes were measured semiautomatically (Figure 1, Supplemental Video 1). Data were analyzed by 2 observers who were blinded to the clinical information of the subjects. However, because the M-mode lines were drawn by each observer (and for each analysis), the B-mode images used for the analysis might have revealed typical findings for CA, such as myocardial hypertrophy, in some patients. Each measurement was repeated 3 times, and the average value was used to estimate SW propagation velocity. The methodology has been described previously (28) and is also detailed in the Supplemental Appendix.
The methodology described here was validated previously in phantoms (see Supplemental Appendix for further information).
Continuous data are displayed as mean ± SD. Categorical variables are presented as frequencies and percentages. Normality of data was tested using the Shapiro-Wilk test. Comparisons between 2 groups were performed using a 2-tailed Student’s t-test for continuous normally distributed data and the Mann-Whitney U test for continuous non-normally distributed variables. Comparisons of categorical variables were assessed by chi-square tests. When more than 2 groups were compared, continuous normally distributed variables were tested using 1-way analysis of variance (ANOVA); when the F test result was significant at the 0.05 level, pairwise comparisons were performed, which were adjusted by using the Tukey method to correct for multiple comparisons. Pearson’s correlation coefficient was used to evaluate the correlation between SW velocity at ED and mitral inflow-to-mitral relaxation velocity ratio (E/E′). Interobserver and intraobserver variability were evaluated in 22 HV from group I and 11 patients with CA using an intraclass correlation coefficient (2-way mixed model, absolute agreement between single measures) and Bland-Altman analysis. All statistical analyses were performed using SPSS version 25.0 software (SPSS, IBM, Armonk, New York). A 2-sided p value of 0.05 was considered statistically significant for all tests.
A total of 50 HV without a history of cardiovascular disease and 18 patients with CA were prospectively screened to be enrolled in the study. Four HV were excluded based on the above-describe exclusion criteria (systolic blood pressure >140 mm Hg). One patient with CA was excluded due to the presence of a dual-chamber pacemaker. Therefore, 46 HV (n = 24, 20 to 39 years of age; n = 11, 40 to 59 years of age; and n = 11, 60 to 80 years of age) and 17 patients with CA were included in the analysis (Figure 2). Patient characteristics, including echocardiographic results, are reported in Table 1.
Myocardial SW velocities in HV
SW starting immediately after the closure of the valves and propagating from the LV base to the apex were detected in 53 subjects (84%) after MVC and in 59 subjects (94%) after AVC. In group I, SW propagation velocities were higher after AVC than after MVC (3.48 ± 0.70 m/s vs. 3.07 ± 0.51 m/s, respectively; p = 0.002). The other 2 groups of volunteers showed comparable SW propagation velocities at MVC and AVC (3.75 ± 0.78 m/s vs. 3.84 ± 0.79 m/s, respectively, in group II and 4.50 ± 1.13 m/s vs. 4.33 m/s ± 0.61 m/s, respectively, in group III). SW propagation velocity differed significantly among age groups, with significant post hoc test results between groups I and III, both after MVC (p < 0.001) and AVC (p < 0.01) (Figure 3).
Myocardial SW velocities IN CA
The mean SW propagation velocity in patients with CA was 6.33 ± 1.63 m/s at MVC and 5.63 ± 1.13 m/s at AVC. Both values differed significantly from the ones obtained in HV after MVC (p < 0.001) and AVC (p < 0.001) (Figure 4).
There was a gradual change with age of the main echocardiographic parameters of the diastolic function in HV: early-to-late mitral inflow velocity ratio decreased from 1.78 ± 0.48 m/s in group I to 1.50 ± 0.35 m/s in group II and to 0.98 ± 0.46 m/s in group III (p < 0.001); E/E′ increased from 5.92 ± 1.28 m/s in group I to 5.96 ± 1.06 m/s in group II and to 7.44 ± 1.25 m/s in group III (p < 0.01). Similarly, SW velocities increased with age (see above).
SW propagation velocities after MVC were found to be significantly among groups with different diastolic dysfunction (ANOVA p < 0.001) (Figure 5A), with a significant post hoc test result when comparing the group with normal diastolic function with each of the groups with diastolic dysfunction (p < 0.05). A significant difference was also observed between the group with first-degree diastolic dysfunction and third-degree diastolic dysfunction (p < 0.001).
Positive correlation was found between SW velocities after MVC and E/E′ as conventional echocardiographic parameter of diastolic function (r = 0.74; p = 0.002) (Figure 5B).
Analysis of the intraobserver and interobserver variability showed a low intraobserver and interobserver variability by using both an intraclass correlation coefficient (intraclass correlation coefficient for SW at MVC was 0.96; 95% confidence interval [CI]: 0.91 to 0.98 and 0.92; and 95% CI: 0.84 to 0.96, respectively; intraclass correlation coefficient for SW at AVC was 0.96; 95% CI: 0.92 to 0.98 and 0.93; 95% CI: 0.86 to 0.97, respectively) and Bland-Altman plots (Figure 6).
Study data showed that myocardial SW velocities, measured with HFR echocardiography, are higher in patients with CA than in HV from different age groups. Moreover, SW velocities appear to increase significantly with age. In addition, SW propagation velocities at ED are significantly different among the group with normal diastolic function and the groups with different grades of diastolic dysfunction.
Over the past few years, cardiovascular SW imaging has developed considerably (29). The technique used most often to generate SWs in the experimental setting relies on the use of an acoustic radiation force (30). However, this approach requires dedicated scanning sequences with a push beam which is well aligned and focused on the cardiac wall. In contrast, myocardial SWs occurring naturally after closure of the valves have a consistent origin and pattern of propagation and can be detected with regular HFR scanning (18). They can be regularly captured after closure of the MV and AV. In the present study, natural analyzable SW were found in 94% of all measurements after AVC and 84% after MVC. This difference was interpreted as a result of the less forceful leaflet coaptation during MVC, which might result in less mechanical excitation of the myocardium. By using this technique, the stiffness of the myocardium can be repeatedly measured at time points close to ED and ES. Values in HV have been already reported in the published studies (14,18), but to date, the direct link between SW velocities and MS has not been demonstrated in a clinical setting.
In young volunteers, myocardial SW velocities, measured with HFR echocardiography, are higher at ES than at ED in HV. This finding is consistent with previous reports (13) and is in line with the fact that ES stiffness is known to be higher than ED stiffness as a result of myocardial contraction (5). The observed SW velocities increased significantly with age (Figure 3), which confirms the findings of Villemain et al. (16), who described a linear relationship between SW speed and age, using SW imaging at ED based on the remote generation of the SW by acoustic radiation force. Given that aging is known to be associated with a reduction in LV compliance (31) and, thus, increased MS, these findings suggest there is a relationship between the detected SW speed and MS.
ES and ED SW velocities were found to be significantly higher in patients with CA than in HV, which can be explained by the fact that this deposition disease leads to stiffening of the myocardium (24). Accordingly, by using CMR elastography, Arani et al. (32) demonstrated a 1.39-fold increase of MS in patients with CA than in HV (median: 11.4 kPa vs. 8.2 kPa, respectively). End-diastolic SW velocity in the present study was 1.78-fold increased, which is in the same order of magnitude as in the referenced study. However, to the best of the present authors’ knowledge, the increased speed of naturally induced SWs in these hearts has not previously been reported and is in line with the assumption that SW speed relates to MS.
In the present study, SW propagation velocity after MV was significantly different between the group with normal diastolic function and the groups with diastolic dysfunction Furthermore, E/E′ as well-defined standard echocardiographic parameter for predicting diastolic filling pressures (1) showed a good correlation with SW velocities measured at ED. Given that filling pressures are determined largely by relaxation and compliance of the LV (33), the present observations are again consistent with the hypothesis that SW speed is related to MS.
There are several limitations to this study. First, MS was not assessed by independent reference methods, and increased stiffness was implied from clinical profiles in combination with previous studies. As such, the relationship between natural SW velocity and MS is likely but not strictly proven. Although it would be better to estimate MS invasively, this was not possible because there was no clinical indication for invasive measurements in the subjects included in the study.
Second, the present measurements cannot be extrapolated to the whole ventricle as they were limited to the interventricular septal wall in the long-axis parasternal view after MVC and AVC. This approach was chosen because SWs cause a particle displacement oriented perpendicularly to their propagation and, with this, perpendicular to the septum. This also implies that the reconstructed velocity maps may have been influenced by the (longitudinal) cardiac motion. Although it was assumed that the motion in the time interval under investigation was small, a 2-dimensional motion estimator could be used for correction and to improve the accuracy of the method further.
Third, SWs caused by AVC and MVC occurred at the beginning of the respective isovolumetric period when MS is likely to undergo relevant changes. The present values, therefore, do not strictly represent ES or ED myocardial properties. It must be assumed, however, that changes of myocardial compliance, as they occur with aging and in deposition disease, are one component of MS which has an effect throughout the cardiac cycle.
Fourth, the reproducibility of the shear wave velocity measurements in the present study was moderate. SW velocities were obtained by measuring slopes in M-mode traces. For this, M-mode lines were manually drawn along the septum in the image which may account to some extend for the reported interobserver variability. This effect of the M-mode extraction may arise from local changes in wave propagation or, more likely, from the poor signal-to-noise ratio. A second factor may be temporal resolution. Along a 3-cm-long M-mode and a SW velocity of 3 m/s, a variation by 1 sample (1/1,200 frames/s = 0.833 ms) leads to an overestimation or underestimation of a 8% to 9%. Additional variance arises from the observer, as the limited image quality required manual selection of the region of interest and indication of the slope to be measured. An automatic segmentation approach would reduce user interaction and therefore improve precision.
Finally, although the present findings showed marked elevated natural SW velocity for CA patients, the sample size was relatively small. Therefore, it is not possible to define clear threshold values to differentiate between physiological and pathological SW velocities. Future studies will be necessary to determine the sensitivity of this method to detect patients with more subtle cardiac abnormalities leading to increased MS.
Natural SW velocity can be measured using high-frame rate ultrasonography and seems to be related to intrinsic MS. Cardiac natural SW imaging, thus, has the potential to become a powerful tool for the assessment of myocardial properties.
COMPETENCY IN MEDICAL KNOWLEDGE: To the best of the present authors’ knowledge, this is the first study where natural SW velocities were assessed in both HV and patients with CA by using cardiac ultrasonography. The study showed that this method was able to distinguish disease from normality in the study population. It is expected that natural SWs could become a new diagnostic tool for the assessment of MS in both systole and diastole in the clinic.
TRANSLATIONAL OUTLOOK: Current echocardiographic assessment of diastolic functions is complex, as there is no single noninvasive parameter that provides a direct measurement of relaxation, restoration of forces, compliance, or LV filling pressure. Still today, estimation of diastolic function is based on the combination of many parameters. Studies in larger population groups may clarify whether the measurement of MS during diastole could better define the LV diastolic function.
The authors thank Vangjush Komini, Jürgen Duchenne, and Ahmed S. Beela for their support.
↵∗ Drs. Petrescu and Santos contributed equally to this study.
↵† Drs. D’hooge and Voigt are senior authors.
Supported by European Research Council grants FP7/2007-2013 and ERC/281748 and Research Foundation-Flanders grants FWO/G002617N and FWO/G092318N. Dr. Petrescu is supported by a German Society of Cardiology research grant. Dr. Cvijic is supported by a European Association of Cardiovascular Imaging research grant. Dr. D'hooge performs research under contracts for GE Vingmed Healthcare and Cairdac. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- aortic valve closure
- cardiac amyloidosis
- cardiac magnetic resonance
- mitral inflow-to-mitral relaxation velocity ratio
- high frame rate
- healthy volunteers
- left ventricle
- myocardial stiffness
- mitral valve closure
- shear wave
- Received July 11, 2018.
- Revision received November 25, 2018.
- Accepted November 28, 2018.
- 2019 American College of Cardiology Foundation
- Nagueh S.F.,
- Smiseth O.A.,
- Appleton C.P.,
- et al.
- Hundley W.G.,
- Bluemke D.A.,
- et al.
- Kass D.A.,
- Bronzwaer J.G.F.,
- Paulus W.J.
- Pernot M.,
- Couade M.,
- Mateo P.,
- Crozatier B.,
- Fischmeister R.,
- Tanter M.
- Voigt J.-U.
- Vos H.J.,
- van Dalen B.M.,
- Heinonen I.,
- et al.
- Brekke B.,
- Nilsen L.C.L.,
- Lund J.,
- et al.
- Couade M.,
- Pernot M.,
- Tanter M.,
- et al.
- Villemain O.,
- Correia M.,
- Mousseaux E.,
- et al.
- Santos P.,
- Lovstakken L.,
- Samset E.,
- D’hooge J.
- Guan J.,
- Mishra S.,
- Falk R.H.,
- Liao R.
- Fitzpatrick A.W.P.,
- Park S.T.,
- Zewail A.H.
- Fikrle M.,
- Paleček T.Š.,
- Kuchynka P.,
- et al.
- ↵Ortega A, Lines D, Pedrosa J, et al. HD-PULSE: High channel density programmable ultrasound system based on consumer electronics. IEEE Int Ultrason Symp October 21–24, 2015, Taipei, Taiwan.
- Papadacci C.,
- Pernot M.,
- Couade M.,
- Fink M.,
- Tanter M.
- Santos P.,
- Petrescu A.,
- Pedrosa J.,
- et al.
- Lakatta E.G.,
- Levy D.
- Arani A.,
- Arunachalam S.P.,
- Chang I.C.Y.,
- et al.
- Flachskampf F.A.,
- Biering-Sørensen T.,
- Solomon S.D.,
- Duvernoy O.,
- Bjerner T.,
- Smiseth O.A.