Myocardial strain analysis by cardiac magnetic resonance 3D feature-tracking identifies subclinical abnormalities in patients with neuromuscular disease and no overt cardiac involvement

Abstract

Aims

Cardiovascular magnetic resonance (CMR) is valuable for the detection of cardiac involvement in neuromuscular diseases (NMDs). We explored the value of 2D- and 3D-left ventricular (LV) myocardial strain analysis using feature-tracking (FT)-CMR to detect subclinical cardiac involvement in NMD.

Methods and results

The study included retrospective analysis of 111 patients with NMD; mitochondrial cytopathies (n = 14), Friedreich’s ataxia (FA, n = 27), myotonic dystrophy (n = 27), Becker/Duchenne’s muscular dystrophy (BMD/DMD, n = 15), Duchenne’s carriers (n = 6), or other (n = 22) and 57 age- and sex-matched healthy volunteers. Biventricular volumes, myocardial late gadolinium enhancement (LGE), and LV myocardial deformation were assessed by FT-CMR, including 2D and 3D global circumferential strain (GCS), global radial strain (GRS), global longitudinal strain (GLS), and torsion. Compared with the healthy volunteers, patients with NMD had impaired 2D-GCS (P < 0.001) and 2D-GRS (in the short-axis, P < 0.001), but no significant differences in 2D-GRS long-axis (P = 0.101), 2D-GLS (P = 0.069), or torsion (P = 0.122). 3D-GRS, 3D-GCS, and 3D-GLS values were all significantly different to the control group (P < 0.0001 for all). Especially, even NMD patients without overt cardiac involvement (i.e. LV dilation/hypertrophy, reduced LVEF, or LGE presence) had significantly impaired 3D-GRS, GCS, and GLS vs. the control group (P < 0.0001). 3D-GRS and GCS values were significantly associated with the LGE presence and pattern, being most impaired in patients with transmural LGE.

Conclusions

3D-FT CMR detects subclinical cardiac muscle disease in patients with NMD even before the development of replacement fibrosis or ventricular remodelling which may be a useful imaging biomarker for early detection of cardiac involvement.

Graphical Abstract

Synopsis of the study design, key findings and conclusions.

Open in new tabDownload slide

Introduction

Neuromuscular diseases (NMDs) include a wide spectrum of conditions with variable degrees and phenotypes of cardiac involvement.^1,2 Cardiovascular magnetic resonance (CMR) is the modality of choice to assess the presence of cardiac involvement in NMD.^1,3 CMR allows the accurate measurement of ventricular volumes and function to detect even subtle systolic dysfunction or ventricular dilation, while multiparametric imaging with tissue characterization allows detection of fibrosis by late gadolinium enhancement (LGE).^4,5 We have recently described the different cardiac phenotypes in NMD by CMR, as well as the patterns and prognostic value of myocardial LGE in patients with known NMD.⁴

Feature-tracking CMR (FT-CMR) is a novel technique that allows quantification of motion and strain using a standard steady-state in free-precession (SSFP) sequence, which forms part of a routine left ventricular (LV) study protocol.⁶ This means that a comprehensive assessment of myocardial mechanics can be reached via the standard CMR examination.⁶ 2D or 3D FT-CMR can be used to independently interrogate the longitudinal, radial, or circumferential shortening, providing thus distinct information on the deformation of endocardial, mid-wall, and epicardial myocardial fibres, respectively.^7,8 2D or 3D FT-CMR can detect subclinical myocardial disease in hypertrophic cardiomyopathy (HCM)⁹ or storage diseases, i.e. Anderson–Fabry disease,¹⁰ but its role in NMD remains largely unexplored.

In this study, we explored for the first time the value of 2D and 3D FT-CMR to detect subclinical myocardial involvement in patients with various forms of NMD. We also explored whether the assessment of myocardial strain by FT-CMR allows the detection of subclinical disease in distinct myocardial layers (i.e. epicardial, mid-wall, and endocardial) in the various forms of NMDs.

Methods

We have formerly described the value of LGE by CMR in NMDs.⁴ The study population included 111 consecutive patients with a diagnosis of an NMD who were referred to our Unit (CMR Unit, Royal Brompton Hospital, NHS, London, UK), in the period 2002–20, to screen for cardiac involvement. The study included every eligible patient with NMD and available CMR scan data. Patient’s medical records were searched for demographics, information on disease background, and adverse cardiac events after the CMR scan. Medical imaging datasets of CMR studies were retrieved for independent reviewing and analysis. In addition, 57 CMR studies from age- and sex-matched healthy controls were also analysed for comparison.

The study was registered as a Clinical Audit by the Quality and Safety Department of the Royal Brompton Hospital. The main aim of the study was to explore (i) the value of 3D-FT by CMR to detect subclinical cardiac involvement in NMD and (ii) to explore whether the various NMD subtypes differentially affect the various strain subtypes (i.e. longitudinal, radial, and circumferential). Secondly, we also explored (i) the relationship between 3D-FT strain subtypes and LGE presence and (ii) the prognostic value of 3D-FT CMR for adverse clinical events.

Cardiac magnetic resonance studies

Digitally archived CMR studies were retrospectively reviewed. All the examinations were performed on a 1.5 T scanner (Magnetom Sonata, Avanto, or Aera Siemens, Erlangen, Germany). Imaging protocols included SSFP breath-hold cines for the assessment of biventricular volumes and function, and LGE sequences for the detection of myocardial fibrosis, as per the recommendations of the Society for CMR.¹¹ More details on the exact examination protocol followed have been previously published.⁴

All studies were analysed using semi-automated software (CMR tools; Cardiovascular Imaging Solutions, London, UK). Earlier established age- and sex-adjusted reference ranges for biventricular volumes and ejection fractions were used to classify the presence of dilation or systolic dysfunction. In addition, LGE presence, LGE type (subendocardial, subepicardial, and mid-wall), and LGE extent using the 17 myocardial segments model were assessed.

LV 2D and 3D strain analyses were performed in all patients and healthy controls using a dedicated FT-CMR tool provided by the CVI42 software (Version 5.13.3, Circle Cardiovascular Imaging Inc., Calgary, Canada). All analyses were performed by an experienced CMR operator blinded to clinical information. Electrocardiography-gated balanced SSFP short-axis (SAX) stack and three long-axis (LAX) views with at least 25 phases were selected from each CMR study. End-diastole and end-systole were identified as the phases with maximum and minimum volumes, respectively. The LV endocardial and epicardial contours along with the anterior and inferior right ventricular insertion points were defined in all slices. Papillary muscles and trabeculae were reliably excluded. The basal SAX contours were deleted in the phases where the LVOT was visible. 2D and 3D strain analyses were automatically performed by the FT software in all slices containing both endocardial and epicardial contours. LV contours were manually adjusted, and the algorithm was reapplied in case of inaccurate tracking. Strain curves and a 16-segment polar map were generated for each strain analysis. The following strain parameters were collected for the purposes of our study: 2D global circumferential strain (GCS) SAX, 2D global radial strain (GRS) SAX, 2D global longitudinal strain (GLS) LAX, 2D-GRS LAX, 3D-GCS, 3D-GRS SAX, and 3D-GLS (Figure 1). 3D global strain values were calculated as average from the respective 16-segment analyses. Per-segment values included: 2D peak radial strain SAX, 2D peak circumferential strain SAX, 2D peak radial strain LAX, 2D peak longitudinal strain LAX, 3D peak radial strain, 3D peak circumferential strain, and 3D peak longitudinal strain. Circumferential and longitudinal strains were expressed as negative values since they represent a shortening of the myocardium in systole (along the LV LAX and the LV circumferential axis, respectively). On the contrary, the radial strain was expressed as a positive value, as it represents the thickening of the myocardium in systole (toward the centre of the LV cavity and perpendicular to the LV LAX). Torsion was also automatically calculated as the difference of rotation between the most basal and apical slices, normalized for the distance between the two planes, and expressed in degrees/cm. The coefficient of variation for 3D longitudinal strain measurements was 7.8%, for 3D radial strain 2.8%, and for 3D circumferential strain 8.4%.

Statistical analysis

Continuous variables that were normally distributed are presented as mean ± SD, while non-normally distributed as median (interquartile range). Comparisons of study characteristics between different groups of patients were performed using an unpaired t-test or Wilcoxon signed-rank test as appropriate for two groups. Comparisons of continuous variable between three or more groups were done by ANOVA or Kruskal–Wallis as appropriate. Categorical variables were compared by using the χ² test as appropriate. A two-tailed P-value <0.05 was considered statistically significant for all comparisons. SPPS statistical package version 22.0 was used for all statistical analyses.

Results

Patient demographics

Overall, 111 patients with various forms of NMD were included in the study. In particular, five main subgroups were identified: 15 patients were diagnosed with a Becker/Duchenne’s muscular dystrophy (i.e. BMD/DMD, Duchenne’s carriers, Emery–Dreifuss dystrophy, and unclassified forms of muscular dystrophy), 27 with a myotonic dystrophy (Type 1 and Type 2 myotonic dystrophy), 14 with mitochondrial cytopathy, 27 had Friedreich’s ataxia (FA), and the remaining 28 were grouped as other NMD (Charcot–Marie Tooth, facioscapulohumeral myopathy, motor neurone disease, nemaline myopathy, and scapuloperoneal myopathy). Clinical characteristics of the study population and conventional CMR data analyses have been previously published,⁴ and also summarized in Table 1. In addition, a total of 57 CMR age- and sex-matched healthy control studies were also analysed for comparison (mean age 38 years, range 5–71, 65% male).

Comparison of global strain parameters between NMD and healthy controls

We compared 2D and 3D global strain values in the NMD cohort and age/sex-matched healthy controls (Figure 2). Among the 2D global strain analyses, only GCS SAX and GRS SAX were significantly impaired in the NMD cohort (P = <0.05). There was no significant difference in 2D-GRS LAX and 2D-GLS LAX between the two populations (P = NS). Likewise, torsion values were not significantly different in the NMD patients. Interestingly, 3D-GRS, 3D-GCS, and 3D-GLS values were all significantly affected in the NMD cohort (P < 0.0001) and had a better diagnostic performance in differentiating healthy controls from NMD patients as assessed by the c-index (Figure 2I).

Comparison of 3D global strain parameters between NMD subgroups and healthy controls

3D global strain values (i.e. GRS, GCS, and GLS) in the five NMD subgroups defined earlier and in healthy controls were compared (Figure 3A–C). The most abnormal 3D global strain values were found in the muscular dystrophy subgroup, followed by FA and mitochondrial disease. Among the 111 NMD patients, we identified 61 patients (55%) with overt cardiac involvement, in the form of a dilated cardiomyopathy (DCM, n = 23), HCM (n = 24), hypokinetic non-dilated cardiomyopathy (n = 2), or another cardiomyopathy (n = 2). The 3D global strain parameters were compared between patients with overt cardiomyopathy, patients with no obvious cardiac involvement, and healthy controls (Figure 3D–F). Interestingly, all 3D global strain values were significantly affected not only in the cardiomyopathy subgroup but also among patients with NMD and no overt cardiac involvement (P = 0.0001 vs. healthy controls). Illustrative examples of strain values in a healthy subject, an NMD patient without cardiac involvement (NMDc−) and an NMD patient with cardiac involvement (NMDc+) are shown in Figure 4.

Patterns of impaired LV mechanics as assessed by 3D strain analyses in NMD

We then assessed 3D strain values for each of the 16-segments in all NMD patients and healthy controls. We calculated the average of the 16-segments for each 3D strain parameter in all healthy controls and identified: (i) 10th and 25th percentiles of the 3D radial strain 16-segment averages (positive values) and (ii) 75th and 90th percentiles of the 3D GLS and GCS 16-segments averages (negative values). Bull’s eye plots with 16 myocardial segments were created for each 3D strain parameter in the four main NMD subgroups to represent the distribution of strain abnormalities (Figure 5). Notably, 3D radial strain values were below 10th percentile (red) in most segments, with the exclusion of the basal lateral segments. On the contrary, 3D longitudinal and circumferential strain values were less affected in all NMD groups; 3D longitudinal strain values were entirely below 75th percentile (green) in patients with FA.

Correlation between per-segment 3D strain values and LGE

We finally explored the relationship of LV strain values with LGE presence and pattern (Figure 6). We assessed 3D strain per-segment in myocardial segments with and without LGE (a total of n = 1776 myocardial segments) in the overall population of patients with NMD (Figure 6A–C). In the LGE-positive segments, there were significantly lower (less positive) 3D radial strain values and higher (less negative) 3D circumferential strain values, compared with the LGE-negative segments (P = 0.014 and P = 0.0007, respectively). No significant difference was observed in the 3D longitudinal strain values between the two groups (P = NS). 3D LV strain indices and LGE pattern correlation showed that all 3D global strain values were significantly affected in patients with transmural LGE. 3D GLS was mainly impaired in patients with subendocardial LGE, 3D GRS in the presence of mid-wall/subepicardial pattern, and 3D GCS in the case of subepicardial LGE.

Prognostic value of 3D strain in NMD

Follow-up data for clinical events were available in 57 patients who were followed-up in our Institution (mean follow-up 49 months, range 1–151 months). Overall, 21 patients reached the composite clinical endpoint of death, heart failure development, or permanent pacemaker/intracardial defibrillator implantation. In survival analysis, patients with 3D radial strain <25% had a significantly worse prognosis compared with those with 3D radial strain ≥25% (P = 0.039, χ² = 4.245). Similarly, the subgroup of patients with 3D circumferential strain ≥−15% had a significantly higher risk for the composite clinical endpoint vs. the patients with 3D circumferential strain <15% (P = 0.012, χ² = 6.274). The respective Kaplan–Meier curves are shown in Figure 7.

Discussion

In the present study, we explored the value of strain analysis by FT-CMR in a large cohort of patients with NMD and made several important observations: (i) 3D strain by FT-CMR is impaired in patients with NMD, and has superior diagnostic accuracy vs. 2D strain in classifying NMD patients from healthy individuals; (ii) not all types of NMD patients have the same degree of LV strain impairment, with muscular dystrophy patients being more severely affected; (iii) 3D strain values are impaired in patients with NMD compared with healthy age- and sex-matched controls, even in the absence of overt NMD-related cardiac involvement, as defined by LV dilation, replacement fibrosis, or systolic dysfunction; and (iv) strain analysis of distinct myocardial fibres (as assessed by 3D-GLS, 3D-GRS, and 3D-GCS) correlates with the presence of replacement fibrosis in the respective endocardial, mid-wall, and epicardial layers of LV myocardium.

Cardiac involvement in NMD is typically manifested as the development of an overt cardiomyopathy or a conduction disease.² Given that, there are underlying molecular and cellular changes in dystrophinopathies, laminopathies, myofibrillar, or mitochondrial myopathies, and other forms of NMD lead to progressive myopathy, it is likely that subclinical myocardial disease occurs too. Nonetheless, a diagnosis of cardiomyopathy in NMD is made only in the presence of an overt cardiomyopathy phenotype. In everyday clinical practice, cardiomyopathy is typically defined by the presence of morphological changes in LV and the development of hypertrophic or DCM phenotypes.² In addition to these extreme phenotypes, we⁴ and others² have demonstrated that NMD-related myocardial involvement could manifest in other cardiomyopathy forms, e.g. with mild systolic dysfunction with or without LV dilation, LV dilation in the absence of systolic dysfunction, or isolated replacement fibrosis.

Since cardiac death remains one of the major causes of mortality in patients with NMD, it is important to identify patients with subclinical myocardial dysfunction to intervene early in the course of the disease to prevent end-stage heart failure development.^12,13 Earlier studies employing strain analysis by echocardiography, have shown that GLS is reduced in small cohorts of patients with FA¹⁴ or Duchenne muscular dystrophy patients.¹⁵ A major limitation of echocardiography is that strain analysis is typically used to assess long-axis function by GLS. However, the function of epicardial or mid-wall myocardial fibres (which are typically affected in NMD) is better assessed using radial or circumferential strain analysis, whose assessment by echocardiography carries technical limitations and is less reliable.

Strain analysis by FT-CMR is a promising tool to detect early impairment in myocardial mechanics, even in the absence of overt systolic dysfunction.¹⁶ CMR overcomes the limitations of poor image quality, while its excellent spatial resolution facilitates the tracking of mid-wall or epicardial myocardial fibres and the assessment of radial or circumferential strain. Our findings demonstrate that the use of 3D FT-CMR is superior to 2D strain measurements in diagnosing NMD-related cardiac involvement. In addition, analysis of all individual myocardial segments shows that the GCS and GRS are correlated with regional LGE myocardial involvement. This is not unexpected given that LGE distribution in NMD typically involves mid-wall or epicardial myocardial layers. Interestingly, the LV basal anterolateral segment had the best 3D circumferential strain values compared with other segments in almost all NMD subgroups and the control group. The presence of higher strain values in the basal lateral wall is a well-known observation reported also in a earlier study that provided reference ranges for 3D FT-CMR.¹⁷ This may reflect the complex mechanics of the subepicardial fibres during systole with a more prominent circumferential shortening in this ventricular region, possibly even in the presence of myocardial fibrosis (as in dystrophinopathies). The observed differences in the LGE pattern and the 3D global strain parameters could be explained by the different orientations of the myocardial fibres that are known to control the LV mechanics.¹⁸

The value of FT-CMR in NMD has been shown earlier in a longitudinal study with serial CMR scans of Duchenne muscular dystrophy patients¹⁹; Siddiqui et al.¹⁹ showed that 3D-GCS was the best predictor of developing cardiac involvement before the onset of LV dysfunction. Importantly, our findings also suggest that compared with age- and sex-matched healthy individuals, strain analysis by 3D FT-CMR is impaired in patients with NMD even in the absence of overt cardiac involvement (defined by LV dilation, replacement fibrosis, or systolic dysfunction). This is a unique observation, particularly pertinent for the management and/or the administration of therapeutic interventions in patients with NMD to prevent cardiac disease progression.

Limitations

This is a retrospective observational study performed in a single tertiary centre and, as such, selection and referral bias are likely. Our cohort includes a mix-up of patients with different NMD and disease status, therefore, may not be fully representative of all populations with NMD or a particular subgroup. Besides, NMD-related cardiomyopathy exhibits age-related penetrance and variable expressivity. In addition, the diagnostic and prognostic yields of parametric imaging in this population were not assessed as T1 and T2 mapping were not routinely performed in all patients, according to departmental protocols. Finally, follow-up data were available only for those patients who had their continuous care in our centre.

Conclusions

In conclusion, in one of the largest available cohorts with patients with NMD we provide evidence of reduced 3D FT-CMR derived myocardial deformation indices in patients with NMD before the development of myocardial replacement fibrosis or ventricular remodelling. Among the available imaging biomarkers, 3D-GCS and 3D-GRS may be particularly relevant for the early detection and monitoring of disease progression in patients with NMD and patient selection for therapeutic interventions.

Funding

None.

Data availability

Data is available upon request from the corresponding author.

References

Author notes