Myocardial tissue characterisation using echocardiographic deformation imaging

STE deformation can be assessed on both global and segmental levels [17]. Global deformation measures have been compared to histological examination in a limited number of studies (Table 1) and are based on the assumption that the pathological process affects the myocardium of the assessed chamber in a global and even manner. This is not true for patients with hypertrophic cardiomyopathy (HCM), and it is therefore no surprise that global LV deformation parameters showed no significant change in studies (in patients with preserved LV ejection fraction (EF)) investigating the incidence of ventricular arrhythmias [18], myocardial fibrosis [13, 19] and histopathological changes pertinent to HCM [13]. However, in the same studies, septal deformation parameters showed significant correlations with the corresponding pathologies [13, 18, 19].

Conversely, pathologies globally affecting the myocardium have been shown to be significantly correlated with deformation parameters. Myocardial fibrosis in patients with advanced heart failure have been linked to the degree of change in LV global longitudinal strain (GLS) [20]. Similarly, correlations have been reported between myocardial fibrosis and GLS in patients with severe aortic stenosis (AS) [21‐23]. In patients with myocarditis, GLS and SR parameters of the LV have been shown to be correlated with the degree of cellular infiltrate during the acute phase [24, 25]. However, while global myocardial deformation has proven utility in diagnosis and prognosis [4, 26, 27], segmental deformation may provide tissue characterisation on a more regional level, allowing diagnosis of pathologies with a non-uniform distribution.

There has been only limited investigation into the relationship between deformation and other myocardial pathologies at the tissue level. In cardiac amyloidosis, GLS magnitude is inversely correlated with severity of amyloid deposition on endomyocardial biopsy [28]. Although we could not find direct comparisons with myocardial histology, in Friedreich’s ataxia cardiomyopathy, GLS (but not longitudinal strain rate) was associated with more advanced disease on imaging criteria [29]. To the best of our knowledge, a direct link between the underlying histology and echocardiographic deformation measures has yet to be established in Fabry disease, but impairment of GLS correlates with degree of late gadolinium enhancement on cardiac MRI [30].

The following sections discussing segmental STE deformation in hypertrophic cardiomyopathy, aortic stenosis, and dilated cardiomyopathy are summarised in Table 2.

HCM is characterised by histopathological changes including hypertrophy of cardiomyocytes, interstitial fibrosis, myocyte disarray, and small vessel disease (small intramural coronary arteriole dysplasia (SICAD)) [31], with examples shown in Fig. 3. Kobayashi et al. [13] investigated the correlation between STE deformation indices and underlying histopathology in 171 patients with HCM undergoing septal myectomy compared with patients with hypertensive heart disease. Comparing the histology of the resected basal septum with SR in the corresponding segment, they found decreasing magnitude of systolic and early diastolic longitudinal SR with increasing myocyte hypertrophy, interstitial fibrosis, SICAD, and fibre disarray. After adjustment for clinical and other echocardiographic variables, a linear relationship between longitudinal systolic SR was seen with myocyte hypertrophy and disarray, but not with SICAD or fibrosis, while early diastolic SR was independently associated with all four conditions.

The same group correlated STE deformation parameters with in vitro mechanics of isolated trabecular muscle in 122 HCM patients [12], and showed that resting tension (the tension measured when the muscle was mounted on the measurement instrument) and developed tension (the maximum tension developed after lengthening the muscle) were significantly correlated with longitudinal systolic strain and early diastolic SR. In multivariate analysis (which included degree of myocyte disarray and interstitial fibrosis), both systolic strain and early diastolic SR were independently associated with both resting and developed cardiomyocyte tension. Similarly, in a separate study of 46 HCM patients, longitudinal strain in the basal septum was modestly correlated with maximal tension developed by single cardiomyocytes isolated from resected septal myocardium [11].

STE strain was compared with histological analysis of basal septal myocardial biopsy obtained in 23 patients with AS undergoing aortic valve replacement surgery [22]. Histological assessment of fibrosis included quantitative measurement of fibrosis area and tissue microRNA miR-21 (which has a well-established association with interstitial fibrosis [32]). Myocardial fibrosis percentage area was significantly correlated with the deformation measures of longitudinal systolic strain and SR, and early diastolic SR, in the corresponding segment. Longitudinal systolic strain was also correlated with interstitial miR-21.

Samples from hearts explanted from patients with dilated cardiomyopathy (DCM) undergoing heart transplantation provide an invaluable opportunity to analyze STE associations, albeit in a select population. In a very elegant study by Cordero-Reyes et al., histological features, messenger RNA and protein expression, were compared with segmental STE deformation (longitudinal, circumferential and radial strain and SR measured at the LV apex, mid lateral wall, mid septum, and the right ventricular (RV) free wall) in 20 individuals with DCM [14]. A statistically significant association was found between myocyte diameter and longitudinal and circumferential strain and SR – segments with higher myocyte diameters had less deformation. However, in these patients with end-stage DCM, the range of values of fibrosis was minimal, and no association was seen between fibrosis percentage and segmental deformation. Expression of the calcium cycling proteins sarcoplasmic reticulum Ca²⁺ ATPase (SERCA2a) and phospholamban was correlated with segmental deformation measures, but again, likely due to the limited range of values, no association was seen with fibrosis-related mRNA and measured protein expression (collagen type I and III, transforming growth factor ß1, and phosphorylated Smad2/3). The titin isoforms N2B and N2BA tend to relate, respectively, to increased and decreased myocardial stiffness, and the study confirmed this by finding a negative correlation between deformation measures and N2B, and a positive correlation with N2BA. In a separate group of 8 patients, repeated measures were made in patients having LV assist device implantation, and this showed a strong association between change in SERCA2a and titin N2BA, and longitudinal and circumferential strain and SR (but not radial strain).

While segmental STE deformation imaging has an important role in the diagnosis of cardiac amyloidosis [33, 34], only one study assessed the degree of amyloid deposition in left ventricular tissue directly [34]. Segments in hearts from three cardiac transplantation patients was correlated with segmental STE LS; a total of 51 segments from three hearts were assessed, with the higher the degree of deposited amyloid in the cardiac tissue, the worse the segmental longitudinal strain. A single case report comparing an explanted heart in a patient with cardiac sarcoidosis noted a relationship between areas of impaired systolic strain in both ventricles and macro- and microscopic findings of granulomatous disease [35]. Although without histology correlation, in Fabry disease there is similarly a reduction in magnitude of segmental longitudinal strain in areas with disease involvement identified by cardiac MRI [30].

To summarize, the above-mentioned studies comparing STE deformation imaging with tissue analysis in patients with HCM, AS, DCM, and other cardiomyopathies suggest that a combination of LV echocardiographic imaging parameters can be used to better characterise underlying LV tissue architecture and function.

In the study by Cordero-Reyes et al. in patients with DCM [14], segmental STE deformation of the RV free wall had similar associations to those seen in the LV, namely increased myocyte diameter was associated with less deformation (longitudinal and circumferential strain as well as systolic and early diastolic SR), and there was no association between deformation measures and interstitial fibrosis or fibrosis related proteins. Increased SERCA2a was associated with more segmental deformation, and, again similar to the LV, there was a negative correlation between deformation measures and titin isoform N2B, and a positive correlation with titin isoform N2BA.

The left atrium (LA) has three distinct roles in LV filling – in addition to atrial contraction, it acts as a reservoir and a conduit for pulmonary venous return [36]. Atrial cardiomyocytes differ from those in the ventricle in many physiological and functional aspects [36]. These differences, when added to the unique geometry and structure of both atria, make it necessary to study the deformation and mechanics of atria independently from ventricles. Despite having some limitations regarding methodology and reference values [17], STE deformation imaging provides valuable information about atrial mechanics and its correlation with a range of cardiac conditions [36]. A recent consensus document regarding STE deformation analysis was published to standardize the methodology applied for the assessment of these chambers [37]. The difficulty in accurately co-localizing imaging and tissue samples in the atria, as well as the frequent involvement of the entire atrium in a disease process, means that studies tended to focus on global LA deformation.

Her et al. investigated LA deformation in 50 patients with mitral valve disease undergoing mitral valve replacement or repair, 28 of whom had mitral stenosis (MS) [15]. All patients with MS had rheumatic valve disease compared to only four patients with mitral regurgitation (MR), and 75% of patients with MS had atrial fibrillation (AF) compared to 50% of those with MR. LA global strain was significantly correlated with the degree of LA fibrosis, independent of age, rhythm, or type of mitral valve disease. LA strain rate was also univariately associated with fibrosis, but this association was lost on multivariate analysis.

The association between LA STE strain and LA fibrosis was investigated in another study of 46 patients with MR due to mitral valve prolapse undergoing surgery, all of whom were in sinus rhythm [38]. Cameli et al. reported a strong negative correlation between LA global longitudinal systolic strain and LA fibrosis, which remained as an independent association after accounting for LA volume index, LA ejection fraction and LA area. There was a significant correlation between LA global longitudinal systolic strain and endocardial thickness. This was one of the few studies to report standard measures of diagnostic accuracy, and LA longitudinal strain had a higher area under the Receiver Operating Characteristic (ROC) curve than LA volume index or E/e’ ratio as a predictor of LA fibrosis percentage > 50% (area under the ROC curve 0.89).

No studies were found directly correlating segmental or global right atrial deformation with underlying tissue characteristics.

Prior to the advent of CMR, integrated backscatter, which measures the ultrasonic reflectivity of the region of interest, was a major focus of tissue characterisation research [39]. Studies in small series of patients show that backscatter is correlated with fibrosis in undifferentiated cardiomyopathy [40], DCM [41], and AS [42], and correlated with fibrosis, disarray and myocyte diameter in HCM (albeit correlated with RV endomyocardial biopsy findings) [43]. There was also a weak correlation found in a port-mortem study between atrial backscatter measurements and atrial interstitial area [44]. However, backscatter has limited ability to reflect fibrosis in those with lower levels of myocardial fibrosis, such as coronary artery disease [45]. Although there are modest relationships between backscatter and STE deformation measures [46], backscatter has major limitations compared to STE: the need for an intrinsic reference frame, the actual position of the sample volume, the limited validated views, the effect of image setting and the presence of artifacts and other reflectors [17]. However, recent advances in computing power and machine learning may improve reproducibility of intensity-based data [47, 48]. It is likely we will see automated assessment that overcomes some of the challenges associated with current methods.

STE has a number of well-established limitations: it relies upon good image quality to allow accurate tracking and movement of a region of interest in and out of the image plane will affect strain measurement. This can be overcome using 3D imaging, which has its own limitations. In particular, there can be difficulty optimizing image quality over the whole 3D volume rather than a limited 2D slice, and frame rates are typically lower with 3D imaging. Less experienced sonographers also show lower levels of reproducibility of STE derived strain measures [49]. There also remains intervendor differences in STE derived measures, and this is particularly true of the segmental measures required for regional tissue characterisation [50, 51]. However, initiatives to standardize STE measurement across vendors are likely to provide improved reproducibility [17, 50, 51]. Finally, STE deformation only provides indirect assessment of tissue features as it relies on a change in tissue function. While this has the potential to be specific for certain conditions [52], in general, it will only allow detection of tissue abnormalities without specification of the cause of these– for example, impaired strain is unlikely to be able to distinguish amyloid infiltration from fibrosis on a segmental level.