Myocardial Stiffness Evaluation Using Noninvasive Shear Wave Imaging in Healthy and Hypertrophic Cardiomyopathic Adults

1Myocardial stiffness (MS) is known to play a key role in diastolic left ventricular (LV) function. Abnormalities in LV relaxation and MS are one of the key pathophysiological mechanisms in heart failure patients with preserved ejection fraction (HFpEF). Hypertrophic cardiomyopathy (HCM) is also associated to severe diastolic dysfunction mainly due to fibrosis and fiber disarray. Moreover, MS is also affected by aging due to progressive physiological changes and cellular and extracellular matrix alterations. However, as the clinical assessment of MS and of the diastolic function is still challenging, the study of MS remained limited to invasive explorations.

2In a general view, the assessments of diastolic function can be divided into those that reflect the process of active/auxotonic relaxation (depending on filling load and afterload) and those that reflect passive stiffness (independent of load conditions). In clinical practice, biological parameters are correlated with ventricular filling pressures (e.g., brain natriuretic protein [BNP]), echocardiographic parameters are identified to assess the auxotonic relaxation and/or the filling pressure, and cardiac magnetic resonance (CMR) imaging offers tools to evaluate myocardial fibrosis (late enhancement gadolinium [LGE]), or the collagen volume fraction (pre-post contrast T1 mapping or extracellular volume fraction [ECV]),. However, noninvasive estimation of passive stiffness remains challenging. To date, cardiac catheterization is the only validated option to assess the passive stiffness clinically, through the compliance estimation thanks to the pressure-volume loops. But the risks for the patients, the necessary equipment, and the costs of these interventions make this examination unfeasible in daily clinical practice.

3Shear wave imaging (SWI) is an ultrasound-based technique for quantitative, local, and noninvasive mapping of soft tissue's stiffness. The clinical impact of SWI has been shown during the last decade in the field of breast lesions and liver imaging. Quantification of MS using SWI has also been investigated extensively on animal models in previous studies. SWI was compared to invasive gold standard parameters derived from pressure-volume loops and was shown to quantify the end-diastolic MS (i.e., passive stiffness) accurately. More recently, the clinical feasibility and reproducibility of transthoracic SWI was shown on a small group of healthy volunteers (HVs) and on pediatric patients. The next step is to show the clinical interest and contribution of this technology for the assessment of diastolic MS in adults and its impact on diastolic LV function. Unlike other imaging techniques, echocardiography is inexpensive, widely available, and can be performed in real-time at the patient bedside allowing monitoring of the heart structure and function.

4In this study, we aimed to perform the first clinical proof of concept of noninvasive MS evaluation on normal and pathological patients. More specifically, the goals of our study were: 1) to quantify MS noninvasively with SWI in a healthy adult population to establish values of MS and its dependence with age; 2) to compare it to severely altered MS in HCM patients with HFpEF; and 3) to investigate the correlation of MS with conventional echocardiography and CMR index of diastolic function.

5This was a prospective study conducted at the Hôpital Européen Georges Pompidou, Paris, France. A population of HVs was contacted and recruited by the Clinical Investigation Center. HV-specific inclusion criteria were: no history of heart failure symptoms, LV ejection fraction (EF) >50%, early diastolic peak (E)/early diastolic mitral annular velocity (e′) <13, as well as normal values of BNP (<35 pg/ml). Three age groups were composed within the recruited HVs: 20- to 39-year-old patients, 40- to 59-year-old patients, and 60- to 79-year-old patients. Exclusion criteria included systolic blood pressure (SBP) ≥140 mm Hg or/and diastolic blood pressure (DBP) ≥90 mm Hg, any persistent cardiac arrhythmia, more than moderate valvular disease, any relevant coronary artery diseases, any contraindication to CMR, and an echogenicity.

6Patients with clinical, genetic, and echocardiographic evidence for sarcomeric HCM with HFpEF (HCM-HFpEF group) were included. HCM-HFpEF patients were identified according to the consensus of the European Society of Cardiology,, using specific inclusion criteria: wall thickness ≥15 mm in 1 or more LV myocardial segments; sarcomeric protein gene mutation identified; LVEF >50%; New York Heart Association (NYHA) functional class ≥II; at least 1 hospitalization for acute heart failure; and E/e′ ≥13 or E/e′ 8 to 13 combined with elevated BNP (>35 pg/ml).

7All subjects included in the study underwent clinical explorations, biological explorations (hematocrit, C-reactive protein, BNP), an echocardiography, a CMR and a cardiac SWI. All explorations were performed on the same day. Three months later and if no clinical event was noted (symptoms of heart failure, hospitalization for cardiac cause, modification of weight or blood pressure), a second echocardiography, and cardiac SWI were realized to estimate the reproducibility on 5 patients per HV subgroup, randomly selected (n = 15).

8The study was approved by the local ethics committee, and all patients gave written informed consent (Non-Invasive Evaluation of Myocardial Stiffness by Elastography [Elasto-Cardio]; NCT02537041).

9Echocardiographic explorations were performed on a Vivid 9 system (General Electric Healthcare, Chalfont St. Giles, Great Britain). Mitral valve inflow pattern (E and A velocity), E-wave deceleration time (E-wave DT), isovolumic relaxation time (IVRT), transmitral flow propagation velocity (Vp), septal mitral valve annular velocities (e′ and a′), as well as pulmonary veins S-wave on D-wave ratio (PV S/D ratio) were recorded in an apical 4-chamber view, to assess the markers of diastolic function according to American Society of Echocardiography guidelines. Global and septal longitudinal strain was also performed by the Speckle Tracking 2D Strain software of General Electrics, directly on the Vivid 9 system. Data were analyzed from stored images by experienced operators (O.V. and A.H.) who were unaware of other test results. Measurements were made over 3 cardiac cycles; the average was used for statistical analysis.

10The CMR protocol consisted of cine-sequences, T1-weighted spin-echo, and 2-dimensional inversion recovery gradient echo sequences for late enhancement assessment after gadobutrol administration (LGE). Post contrast T1 times (T1 mapping) was performed with a modified Look-Locker inversion recovery sequence with a 3(3)5 scheme before and 15 min after contrast application. Mapping was performed over all available short-axis slices. Extracellular volume (ECV) fraction was calculated on the basis of the combination of pre- and post-contrast T1 mapping data according to the approach proposed by Rommel et al.. All acquisitions were consistent with the Society for Cardiovascular Magnetic Resonance published guidelines. Data were interpreted by 2 experienced readers (E.M. and G.S.) who were unaware of the subjects’ clinical information and the results of other diagnostic tests.

11SWI is based on the remote generation of shear waves in soft tissue by acoustic radiation force combined with ultrasonic ultrafast imaging of the shear wave propagation (5,000 images/s), using the same ultrasonic transducer (Figure 1, Video 1). This modality has already been described in previous works,, and is also described in more details in the Supplemental Appendix. In this study, a phased array probe (2.75-MHz central frequency, SuperSonic Imagine, Aix-en-Provence, France) connected to an ultrafast ultrasound scanner (Aixplorer, SuperSonic Imagine) was used. A conventional real-time echocardiographic image was used to position the probe. The focus of the acoustic radiation force generation was adjusted manually by the operator on the myocardial wall. The operator than launched the SWI acquisition that lasted approximately 1 s.

12The explored myocardial segment was the anteroseptal basal segment (ASB segment). It was evaluated in 2 orthogonal axes (short- and long-axis views) (Figure 1). Short-axis measurements were used to derive the shear modulus, whereas long-axis and short-axis values were used to compute the fractional anisotropy. All acquisitions were performed at end-diastole and triggered by an electrocardiogram. The 30 frames recorded after the push were post-processed to visualize the shear wave and compute the speed.

13Data were interpreted off-line by 1 experienced reader (O.V.) who was unaware of the subjects’ clinical information and the results of other diagnostic tests.

14Similar to any fiber-composed muscular tissue, the myocardium presents a significant anisotropy of its elastic properties. Consequently, MS is expected to be higher when measured along the fibers, which are mainly oriented along the circumferential direction in the mid-wall layer. To evaluate the degree of anisotropy in the myocardium, the fractional anisotropy (FA) was computed. FA was defined using two shear wave speed measurements performed in orthogonal propagation directions (long-axis and short-axis views) using the formula published by Lee et al.. More details on the method are given in the Supplemental Appendix.

15Data for continuous variables are presented as mean ± SD, if normally distributed, or as median and interquartile range if non-normally distributed. Categorical variables are presented as frequencies and percentages. Comparisons between groups were made using chi-square tests for categorical variables. Continuous variables were compared with unpaired Student’s t tests or the nonparametric Mann-Whitney U test where appropriate. Univariate and stepwise multivariate linear regression analyses were performed to identify predictors of r (standardized coefficient of linear regression). Receiver-operating characteristic (ROC) curves and area under the curve (AUC) were computed to assess the effectiveness of MS to predict healthy or pathologic subjects. Reproducibility of MS estimation (3 months after the first estimation) was tested by the Bland-Altman limits of agreement. The reproducibility coefficient was calculated as 1.96 × the SD of the differences, as proposed by Bland and Altman. All the analyses were conducted using Medcalc (MedCalc Software, Mariakerke, Belgium).

16A total of 93 subjects (69 HV and 24 HCM-HFpEF) were prospectively screened for inclusion into the study (Figure 2). Eight patients from the HV group were excluded based on the exclusion criteria (1 congenital heart disease and 1 valvulopathy on echocardiography, 1 doubt on infract scar on CMR, 6 anechoic). Two patients of the HCM-HFpEF group were excluded based on the exclusion criteria (2 infarct scars seen on CMR, 2 anechoic). Finally, 80 subjects were included: HCM-HFpEF group (n = 20), HV 20-year-old to 39-year-old group (n = 20), HV 40- to 59-year-old group (n = 20), and HV 60- to 79-year-old group (n = 20) (Figure 2).

17The molecular genetic causes for the HCM-HFpEF group were: 8 mutations of MYH7, 6 mutations of MYBPC3, 2 mutations of TNNT2, 1 mutation of TPM1, 1 mutation of TNNI3, 1 mutation of MYL3, and 1 mutation of MYL2.

18The subjects’ baseline characteristics including clinical characteristics, laboratory data, echocardiographic results, and CMR results are shown in Table 1.

19There is no statistical difference between the HV group and the HCM-HFpEF group in terms of age (p = 0.22), body mass index (BMI) (p = 0.74), and BP (SBP, p = 0.41; DBP, p = 0.47). In both groups, there was no diabetic patient.

20Concerning the HCM-HFpEF group, 6 patients (30%) had a New York Heart Association (NYHA) functional class ≥III (4 patients were class III, 2 patients were class IV). Twenty patients (100%) had a BNP >35 pg/ml. Regarding echocardiographic results, LV mass index was significantly higher than the HV group (125 ± 34 g/m2 vs. 70.6 ± 20 g/m2; p < 0.01), the ASB segment was significantly thicker than the HV group (20.8 ± 5.1 mm vs. 5.9 ± 1.4 mm; p < 0.01), with a segment strain lower than the HV group (−6.4 ± 3.6% vs. −16.9 ± 2.2%; p < 0.01). All the main diastolic function parameters were significantly different (p < 0.01) than those of the HV group (E/A; e′, E/e′, E-wave DT, IVRT, Vp, E/Vp, PV S/D ratio). Regarding CMR results, 16 of 20 (80%) had an LGE on the ASB segment (p < 0.01). There was a difference between HV and HCM-HFpEF groups concerning myocardial T1 post-contrast (p = 0.02) and ECV (p < 0.01).

21The mean MS for the HV group was 4.47 ± 1.68 kPa (Figure 3). No patient from the HV group had an MS larger than 8 kPa. The mean MS was 2.59 ± 0.58 kPa for the 20- to 39-year-old HV group, 4.70 ± 0.88 kPa for the 40- to 59-year-old HV group, and 6.08 ± 1.06 kPa for the 60- to 79-year-old HV group. There was a statistical significant difference between all age groups (p < 0.01).

22A strong increase in MS with age was found (Figure 3). The correlation between age (x) and MS (y) values was robust (y = 0.087x + 0.1248; r2 = 0.77; p < 0.01). A multivariate linear regression analysis (including sex, age, BMI, hazard ratio SBP, and DBP) showed that age was the only clinical parameter correlated with MS (age, p < 0.01; sex, p = 0.77; BMI, p = 0.98; HR, p = 0.88; SBP, p = 0.33; DBP, p = 0.63). The correlation of echocardiographic parameters and age was lower: E/A, r2 = 0.30; E/e′, r2 = 0.23; E/Vp r2 = 0.01 (Supplemental Figure 1). In univariate analysis for the HV group, there was no correlation between LV mass and MS (r = 0.21, p = 0.44).

23The mean MS for the HCM-HFpEF group was 12.68 ± 2.91 kPa. Only 2 patients had an MS value <8 kPa (6.46 kPa and 7.97 kPa). The correlation between age and MS values for this pathological group was low (r2 = 0.14, r = 0.37, p < 0.01). We found no difference between MYH7 and MYBPC3 mutation subgroups (p = 0.34).

24There was a significant statistical difference between the MS healthy group and the MS HCM-HFpEF group (p < 0.01) (Figure 4). Based on the ROC curve analysis, the optimal cut-off value of MS for detection of HCM-HFpEF was 8 kPa (AUC = 0.993, sensitivity = 95%, specificity = 100%).

25Positive correlations were found between MS and parameters in echocardiography (E/e′, r = 0.783, p < 0.01; E/Vp, r = 0.616, p < 0.01; left atrial volume index [LAVI], r = 0.623, p < 0.01) and CMR (LGE, r = 0.804, p < 0.01; myocardial T1 pre-contrast, r = 0.711, p < 0.01; myocardial T1 post-contrast, r = 0.595, p = 0.01; ECV fraction, r = 0.447, p = 0.03). Correlation of MS with other parameters is summarized in Table 2. There was no correlation between MS and the global strain (r = 0.37, p = 0.27) or the septal strain (r = 0.40, p = 0.09). There was no correlation between MS and BNP (r = 0.41, p = 0.06).

26Among the 15 HV patients who were re-evaluated 3 months later, there was no statistical difference with the initial assessment (mean MS = 4.26 ± 1.36 kPa; p = 0.67). Moreover, Bland-Altman analysis showed good agreement between measurements: MS +0.08 kPa (upper limit of agreement: +0.89 kPa; lower limit of agreement: −0.73 kPA) (Supplemental Figure 2).

27The mean FA for the HV group (0.238 ± 0.068) was higher than that for the HCM-HFpEF group (0.133 ± 0.073; p < 0.01). Eighteen of 20 patients (90%) from the HCM-HFpEF group had an FA <0.155 whereas 56 of 60 patients (93%) from the HV group had an FA larger than this cut-off (AUC = 0.891, sensibility = 90%, specificity = 91.2%) (Figure 5).

28In this study, MS was assessed quantitatively using noninvasive SWI in HV and HCM patients with HFpEF. To our knowledge, this is the first study to assess MS quantitatively and noninvasively in both HV and pathological cases (HCM-HFpEF). This study showed that SWI allows us to establish values of MS in a HV population, MS increases strongly with age in the normal heart, and there is a large difference in MS between HV and HCM-HFpEF groups (cutoff = 8 kPa).

29In this study we were able to quantify MS aging. In the NORRE (Normal Reference Ranges for Echocardiography) study, Caballero et al. also found a gradual change with age of the main echocardiographic parameters of the diastolic function. In this study, which analyzed 449 HV echocardiographs, the E/e′ ratio increased from an average of 6.9 ± 1.6 in 20- to 39-year-old subjects to an average of 9.7 ± 2.8 in 60- to 79-year-old subjects, a change of approximately 50% with a fairly linear evolution. Myocardial aging was also evaluated by CMR on a human population or by invasive estimation on animal study. In 1991, Weger et al. have well demonstrated that the age-induced physiological myocardial fibrosis impacts on the cardiac function, including the ability of the ventricle to relax during the diastolic filling (auxotonic relaxation). Regarding HV patients who participated in our study, we also found a linear evolution of MS, allowing us to establish the change of MS with age.

30We also showed an MS difference between HV and HCM-HFpEF noninvasively. Zile et al. have shown on myocardial histologic explorations of HFpEF patients that an increase in passive MS is due to an architectural modification (increase of collagen and titin). Moreover, in a systematic review on HCM published in 2002, Barry J. Maron noted that the “LV myocardial architecture is disorganized […] with multiples intercellular connections often arranged on chaotic alignment and with expanded interstitial (matrix) collagen,” which is supported by previous work that tried to link myocardial histological explorations and MS. Our study shows that the abnormal MS of this characteristic pathological group (HCM-HFpEF) can be quantified noninvasively. In addition, SWI provided information on myocardial architecture through the analysis of the FA which revealed differences between the tissue architecture of the 2 groups, with a decrease of the physiological anisotropy in the HCM-HFpEF group. However, the clinical interest of this parameter must be further investigated and analyzed on more patients and other pathological groups.

31Beyond the analysis of viscoelastic properties and of tissue structure, the comparison of MS with the recommended ultrasound parameters used to assess the diastolic function seems to show that this quantitative parameter could help to distinguish patients with diastolic dysfunction from others. Obviously, the diastolic function analysis remains complex and it would be unrealistic to think that 1 quantitative parameter could define the LV diastolic function. For example, LAVI >34 ml/m2 is a key structural alteration allowing the definition the diagnostic of HFpEF, but Caballero et al. have also clarified that 15.1% of heathy people have a LAVI >34 ml/m2 whereas only 0.5% have an E/e′ ratio >15. Nevertheless, the MS assessment of patients with HFpEF has clearly helped to improve understanding of this disease. Performing the assessment noninvasively would refine our diagnostic capabilities and may help us to understand disease “at the bedside,” with a noninvasive approach.

32Beyond the diagnostic contribution that the evaluation of MS by SWI could represent, the prospects of therapeutic follow-up could be interesting. The current finding is that there is no specific medical treatment of diastolic dysfunction in HCM. This is probably due to the fact that there is as yet no medical treatment with a high expected efficacy in HFpEF, as recalled in the recent European guidelines: “No treatment has yet been shown, convincingly, to reduce morbidity or mortality in patients with HFpEF”. Noninvasive MS assessment could be a major tool for the development of novel treatments of HCM and/or HFpEF. Thanks to this noninvasive MS marker, the impact of certain treatments (angiotensin-converting enzyme inhibitors or mineralocorticoid/aldosterone receptor antagonists, for example) could be evaluated quantitatively.

33Finally, we have observed a good reproducibility of the MS assessment. Despite the small size of this analysis group (n = 15), which limits its interpretation, these results indicate that this technique could be used to evaluate a patient through longitudinal follow up. We did not re-evaluate the reproducibility of this technique on HCM-HFpEF groups because it is still difficult to estimate the impact of the disease evolution on the MS results, and the treatments of these patients were modified after the initial evaluation and could change their diastolic function (and maybe their myocardial structure).

34This was a monocentric study, and on a small sample. Therefore, it is difficult to extrapolate these results to a general population, which would require larger population groups. The gold standard of MS assessment remains an invasive measurement with a conductance catheter. From a regulatory point-of-view, this exploration could not be achieved on HV patients recruited exclusively for this study, notably because of the risks inherent to this invasive exploration. However, previous work on an animal model had shown a strong correlation between MS estimated by SWI and MS estimated through the end-diastolic strain-stress relationship,. In addition, the noninvasive echocardiographic evaluation of MS has been shown possible by extrapolating the LV volume-pressure curve (end-diastolic pressure volume relationship) knowing a single-beat point, based on Doppler and bidimensional data (volume),,. It will be interesting in future studies to compare the MS estimated by SWI with this noninvasive method. Concerning the HCM-HFpEF group, no catheterization was provided in the management of these patients during the time of the study. Another limitation of our study is the local evaluation of MS by the SWI. Indeed, we compared global functional parameter (E/e′, E/Vp, or LAVI) with a segmental parameter (MS of the ASB segment). Nonetheless, the same segment was analyzed for all people included in the study. This study was limited to the basal septum. The main reason is that we used a conventional phased array with an elevation focus at 70 mm which limited the generation of the shear waves at higher depth. To address other segments, the development of dedicated probes will be required.

35In this study, we quantitatively assessed the end-diastolic MS in adult HV patients using SWI in the HV and sarcomeric HCM with HFpEF patients. MS was found to increase with age, and a cut-off of 8 kPa allowed clear differentiation of these 2 groups. The FA obtained by SWI reflected the underlying tissue structure modifications. Thanks to this ultrasound technology, the noninvasive assessment of MS enables a new diagnostic option in cardiology. Future studies will aim to evaluate MS on other heart diseases, such as isolated HFpEF, HF with reduced EF, hypertension, diabetes cases with HFpEF, or other cardiomyopathies to determine the impact of this parameter in clinical practice.COMPETENCY IN MEDICAL KNOWLEDGE: This study is the first, to our knowledge, to quantify MS in patients using noninvasive SWI. It opens the path to the clinical evaluation of LV function using a new parameter relatively independent of loading. We expect this parameter to be robust, stable, and representative of the myocardial LV diastolic function.TRANSLATIONAL OUTLOOK: Future studies should address the evaluation of this new parameter in systolic and diastolic heart failure patients and whether this parameter could improve the diagnosis and prognosis of this population.

36COMPETENCY IN MEDICAL KNOWLEDGE: This study is the first, to our knowledge, to quantify MS in patients using noninvasive SWI. It opens the path to the clinical evaluation of LV function using a new parameter relatively independent of loading. We expect this parameter to be robust, stable, and representative of the myocardial LV diastolic function.

37TRANSLATIONAL OUTLOOK: Future studies should address the evaluation of this new parameter in systolic and diastolic heart failure patients and whether this parameter could improve the diagnosis and prognosis of this population.