Early cardiac involvement detected by cardiac magnetic resonance feature tracking in idiopathic inflammatory myopathy with preserved ejection fraction

Cardiac involvement is common in idiopathic inflammatory myopathy (IIM) but often subclinical. Cardiac magnetic resonance (CMR) is a promising tool in detecting cardiac involvement in patients with IIM. The aim of this study was to assess cardiac involvement in IIM patients by CMR feature tracking (CMR-FT). Thirty-seven IIM patients and 25 controls were enrolled in this retrospective study. The left ventricular (LV) functional parameters such as volume and ejection fraction were measured. Global and regional LV peak strain (PS) in radial, circumferential and longitudinal directions were derived from cine images. Left atrial (LA) volume, longitudinal strain and strain rate (SR) parameters and LA reservoir function, conduit function and booster pump function were assessed, respectively. IIM patients with preserved LVEF showed significantly reduced global and regional LV PS in longitudinal direction (all p < 0.05). Compared with controls, LA reservoir and conduit function were significantly impaired in IIM patients (all p < 0.05). The global LV longitudinal PS, LAVpre-ac and SRe were independent predictors of IIM. By Pearson’s correlation analysis, the LV global radial, circumferential and longitudinal PS were all correlated to LVEF in IIM patients (r = 0.526, p < 0.001 vs. r = − 0.514, p < 0.001 vs. r = − 0.288, p = 0.023). CMR-FT based LV and LA deformation performance could early detect cardiac involvement in IIM patients with preserved LVEF.


Introduction
Idiopathic inflammatory myopathies (IIM) are a heterogeneous group of immune-mediated chronic diseases characterized by muscle weakness and inflammatory cell infiltrates in skeletal muscle [1,2]. The frequency of cardiac involvement varies between 9 and 72%, which depends on patient selection and investigations employed [3][4][5][6]. Noteworthy, the occurrence of heart failure or malignant arrhythmia could be fatal [7]. Accumulating evidence have shown subclinical cardiac involvement in IIM patients could be more frequent than its sympotomatic manifestation. Thus, a reliable technique is needed to detect myocardial injury in the early stage of IIM to prevent disease progression.
Traditional screening for cardiac involvement in IIM includes electrocardiography (ECG), biochemical markers of myocardial damage, radionuclide ventriculography and echocardiography [7][8][9][10]. ECG, biochemical markers and echocardiography are easy to operate, but the abnormalities were unspecific. Radionuclide ventriculography requires supplementary CT to image structures, which has radiation Wangyan Liu and Yinsu Zhu have contributed equally to this work.

3
and is time-consuming. The gold standard for detecting myocardial injury is endomyocardial biopsy. But it is invasive and is low sensitivity [11]. In a word, cardiac involvement in IIM patients is not rare, and a valuable technique is needed.
Cardiac magnetic resonance (CMR) is excellent in myocardial tissue characterization, which is useful for preclinical diagnosis of cardiac involvement in patients with IIM [12], but the disadvantage is that both late gadolinium enhancement and T1mapping technology require patients to inject contrast agent. CMR cine imaging is the gold standard technique for assessing ventricular volumes and ejection fraction. CMR feature tracking (CMR-FT) is a developing contrast-free quantitative method that uses CMR cine images and it is able to quantify systolic and diastolic myocardial deformation in different orientations. Left ventricular (LV) strain parameters could early detect cardiac involvement in patients with ischemic or non-ischemic cardiomyopathy prior to a reduction in the left ventricle ejection fraction (LVEF) [13][14][15][16]. Furthermore, left atrium (LA) acts as a crucial role in the regulation of LV filling, involving reservoir, conduit and booster pump functions [17]. LA size, function and strain parameters could reflect LV diastolic dysfunction. Some studies have confirmed CMR-FT based atrial performance analysis could help to identify LV diastolic dysfunction in hypertensive cardiomyopathy, hypertrophic cardiomyopathy, myocardial infarction and heart failure and could indicate the presence of atrial myopathy and related complications [17][18][19]. However, studies focused on LA deformation in IIM patients have rarely been reported.
Therefore, in the current study, we aim to investigate whether CMR-FT could early detect cardiac involvement in IIM patients with preserved LVEF.

Study population
The study cohort retrospectively enrolled 48 patients with IIM underwent CMR from October 2017 to September 2020. All patients were included based on the 2017 EULAR/ACR Classification and Diagnosis for IIM [20], meanwhile, suspected myocardial injury on presenting with at least one of the following clinical situations: (1) elevated cardiac troponin T (cTnT) > 15 ng/L, (2) clinical symptoms as concerns to heart conditions (palpitations, chest pain, congestive heart failure et al.), (3) pericardial effusion, and (4) ECG abnormalities except for sinus tachycardia. The exclusion criteria included coronary artery disease (n = 2), cardiomyopathy (n = 4), LVEF < 50% (n = 1), congenital heart disease, heart valve disease and unqualified CMR images (n = 4). Finally, 37 patients with IIM were eligible for the study. A total of 25 sex-matched controls were recruited. The main cardiovascular risk factors such as hypertension, hyperlipidemia were also matched between IIM patients and the controls. Informed consent was obtained from all subjects according to the Declaration of Helsinki. The study protocol was approved by the Institutional Ethics Committee for Medical Research (Clinical characteristics of IIM patients with cardiac involvement: 2020-SR-228).

CMR imaging protocol
All participants underwent cine CMR imaging on a 3 T whole-body scanner (MAGNETOM Skyra, Siemens Healthcare, Erlangen, Germany) with an 18-channel phase-array coil using ECG gating. The balanced steady-state free precession (b-SSFP) sequences were performed to obtain cine CMR images, including a stack of parallel short-axis slices covering the entire LV and three LV long-axis slice (2-, 3-, and 4-chamber views) images (field of view 340-380 mm, repetition time 3.4 ms, echo time 1.4 ms, matrix size 208 × 188 mm, voxel size 1.6 × 1.6 × 8.0 mm 3 , bandwidth, 962 Hz/Px, flip angle 47°, slice thickness 8 mm, inter slice gap 2 mm, acquisition in end-inspirational breath-hold). Twenty-five frames are acquired for both short-axis and long-axis cine images.

CMR image analysis
All CMR images were analyzed using CVI42 (version 5.0, Circle Cardiovascular Imaging Inc. Calgary, Canada). The endo-and epicardial contours of LV myocardium were manually traced at end-diastole and end-systole on shortaxis b-SSFP cine images. Papillary muscles and moderator bands were excluded from volumes. Cardiac volumetric and functional parameters, including LV end-diastolic volume, LV end-systolic volume, and LVEF were automatically generated. All the volumetric parameters were indexed to body surface area (BSA), which was calculated by the Mosteller formula.
LA volumetric parameters, including LA maximal (LAVmax, at the LV end-systole), pre-atrial contractile (LAVpreac, at the LV diastole before LA contraction) and minimal volume (LAVmin, at the late LV end-diastole after LA contraction) were calculated using the following equation: L min corresponds to the shorter long-axis length of the LA either from the 2-chamber or the 4-chamber view [21,22]. Image measurements of LA area and length on 2-and 4-chamber views of a participant are shown (Fig. 1). LAVmax, LAVpre-ac and LAVmin were all normalized to BSA.
LA total, passive and booster EF were calculated from LA volumes according to the following equations [23,24]:

Myocardial strain analysis by CMR-FT
The CMR-FT analysis was performed on the acquired b-SSFP cine images. A set of cine images in short-axis and three long-axis views (2-, 3-, and 4-chamber views) were loaded into the feature tracking module. All endoand epicardial contours of LV at end-diastole were manually depicted ( Fig. 2) with subsequent automatic tracking throughout the cardiac cycle. The contours were manually adjusted if needed. After defining the RV insertion points within the LV in short-axis images, the LV global and regional (basal, medial and apical) peak strain (PS) in radial, LAV pre−ac circumferential and longitudinal mode were automatically derived by the software. LA endocardial contour was manually traced in the 4-and 2-chamber views at the phase of maximal LA volume before mitral valve opening. Pulmonary veins and LA appendage were excluded from the LA endocardial borders (Fig. 3). LA endocardial longitudinal strain parameters included LA reservoir function [total strain (ε s ), peak positive SR (SRs)], LA conduit function [passive strain (ε e ), peak early negative SR (SRe)] and LA booster pump function [active strain (ε a ), late peak negative SR (SRa)].

Reproducibility analysis
For the assessment of the inter-observer reproducibility, 25 IIM patients and controls randomly selected were assessed by two experienced and double-blinded observers (WYL with 8 years of CMR experience, CJF with 2 years of CMR experience). To determine the intra-observer variability, 25 IIM patients and controls randomly selected were repeated by one of the investigators (WYL) 1 month later.

Statistical analysis
Continuous data were presented as mean ± standard deviation, categorical data was expressed in frequencies and percentage. To test the normality of distribution of continuous variables, the Kolmogorov Smirnov test was used.   Strain curves (B, E) and average strain rate curves (C, F) were demonstrated. LA strain and strain rate parameters were obtained as shown in (B and C), respectively. LA left atrium, LV left ventricle Differences in continuous variables between two groups were analyzed using unpaired Student's t test and analysis of covariance adjusted by age. Categorical variables were compared by Chi square test or Fisher exact-test. Univariate and multivariate binary logistic regression analyses were performed to identify independent risk factors of IIM. IIM diagnostic models were generated by forward and conditional method using multivariate binary logistic regression. Receiver operating characteristic curve analysis was performed to examine the diagnostic accuracy of these models. Intra-observer and inter-observer variability were assessed by the intraclass correlation coefficient. All statistical analyses were performed with SPSS 21.0 (SPSS, Chicago, USA). All tests were 2 sided, and p < 0.05 was considered significant.

Patient characteristics
All baseline characteristics are summarized in Table 1. The LV function parameters, including LV end-diastolic volume index (LVEDVI), LV end-systolic volume index (LVESVI), LV stroke volume index (LVSVI), showed no differences between the normal controls and patients with IIM (all p > 0.05). Although, LVEF showed significant difference between IIM patients and controls, all the subjects' LVEF were more than 50%.

Comparison of the LV strain parameters between patients with IIM and controls
The LV strain parameters of all subjects are shown in Table 2. Compared with the controls, the magnitude of global LV longitudinal PS was decreased in patients with IIM (− 12.87 ± 2.62% vs − 14.88 ± 1.79%, p = 0.002).   Fig. 4.

LA volumetric and deformation parameters assessed by CMR-FT
As shown in Table 3

Univariable and multivariable binary logistic regression of volumetric and deformation parameters of LA and LV
Univariable and multivariable binary logistic regression results were shown in Table 4. The global LV longitudinal PS, LAVpre-ac and SRe were independent predictors of IIM. Three models were builded to evaluate diagnostic accuracy.

Receiver operating characteristic analysis of the three models
As shown (Fig. 5

Intra-observer and inter-observer variability
The intraclass correlation coefficient (ICC) of the LV and LA strain parameters were shown in Table 5. There were excellent inter-observer (0.939-0.971) and intra-observer (0.978-0.995) agreements in global LV myocardial PS. For regional LV myocardial PS, the inter-and intra-observer ICC values were 0.784-0.967 and 0.893-0.995, respectively. LA strain parameters also showed excellent inter-observer (0.888-0.924) and intra-observer (0.846-0.979) agreements.

Disscussion
In the present study, we applied CMR-FT technique to evaluate LV and LA deformation performance in IIM patients. The main findings were as follow: (1) The damaged LV strain in IIM patients mainly involved global and regional LV PS in longitudinal direction; (2) LA reservoir function and conduit function were impaired in IIM patients; (3) LA volumetric parameters and systolic function parameters showed significant difference between IIM patients and the controls. This was the first report for detecting cardiac involvement in patients with IIM using LA strain and functional parameters. LA strain and functional parameters derived from CMR-FT could early detect cardiac involvement in IIM patients with preserved LVEF besides LV strain parameters.
In the present study, we found the magnitude of global and regional LV longitudinal PS were decreased in IIM patients with preserved LVEF compared with the controls [25,26]. LV longitudinal PS represents the contraction of the longitudinal myocardial fibers in the LV subendocardium, which is the most susceptible layer to various stressors, longitudinal strain parameters therefor have the sensitivity to detect subtle systolic dysfunction before conventional changes in LVEF [8]. Previous studies have demonstrated that IIM patients showed impaired LV myocardial microvascular dysfunction, and that the abnormal LV myocardial deformation was associated with microvascular dysfunction in other diseases [25,27]. Thus, we presume that the magnitude reduction in longitudinal myocardial strain of IIM patients in this study may be related to LV myocardial microvascular ischemia. Interestingly, our study showed that there were regional alterations in LV radial and circumferential strain and these alterations were not associated with a significant modification of global radial and circumferential strain. This phenomenon might portend an evolving disease towards the more external myocardial layers.
The frequency of LV diastolic dysfunction in IIM patients was 42% [7]. The principal role of the LA is to modulate left ventricular filling due to three basic functional elements: 1. Reservoir function (collection of pulmonary venous reflux during LV systole) 2. Conduit function (passage of blood flow to LV during early diastole) 3. Contractile booster pump function (active filling of LV during late diastole). LA reservoir and conduit function reflect left ventricular diastolic dysfunction, while impaired booster pump function reflect dysfunction of left atrial contraction [28]. The reservoir and conduit functions make the main contribution to LV filling during early diastole, while the booster pump function is the basis for active LV filling during late diastole [17]. In this study, LA reservoir function and LA conduit function showed significant difference between controls and IIM patients. That is to say, our results verified that IIM patients may have LV diastolic dysfunction. This may be caused by elevated LV filling pressure, increased LV wall stiffness and abnormal LA-LV coupling. The suggested main mechanisms of cardiovascular involvement in IIM include the effects of traditional cardiovascular risk factors, systemic and local inflammation of the disease. Hypertension, diabetes mellitus and ischemia heart disease are more prevalent in IIM patients than in general population [29]. Increased cardiac load forces adaptive myocardial remodelling to uphold the perfusion of the peripheral tissues in IIM patients. Cardiac remodelling leads to hypertrophy and enlargement of heart chambers [30]. Hence, in IIM patients, cardiac involvement could lead to elevated filling pressure, LV diastolic dysfunction, and LA remodelling. Therefore, LA strain parameters can early detect cardiac involvement in IIM patients.
All LA volumetric analysis, total, passive and active LA emptying fractions showed significant difference between IIM patients and the controls. LA enlargement and LAEF impaired happened in the early stage of IIM were confirmed in this study, which providing an optional method to evaluate LV diastolic dysfunction by CMR in the situation without commercially available software to perform CMR-FT. In our study, the global LV PS in radial, circumferential and longitudinal directions all show significant correlation with LVEF. Myocardial deformation assessed by tracking analysis allow the assessment of subtle alteration in cardiac (atrial or ventricular) performance, that are not immediately disclosed by the simple assessment of LVEF [14,15]. Early detection of cardiac injury can promote early intervention of heart failure, conduction defects, arrhythmias, and inflammatory infiltrate and improve survival rate of IIM patients.
All three models were proved to be useful for diagnosing cardiac involvement of IIM. The global LV longitudinal PS were included in all models. Previous studies demonstrated that it is a sensitive and reliable index to diagnose disease [25,26]. LAVpre-ac indexed were included in model 2 and model 3, which indicated it was also a sensitive parameter in detecting cardiac involvement. Both LAVpre-ac index and SRe correspond to LA conduit function. Previous study confirmed that there is a relationship between LA volumetric parameters and the severity of LV diastolic dysfunction [18]. Because the left atrium is directly exposed to LV pressures during diastole, elevated LV pressures result in an increase in LA pressure to maintain adequate LV filling. This compensatory LA pressure elevation leads to stretching of the LA wall, resulting in LA dilatation. The LAVpre-ac index was significantly lower in patients with mild LV diastolic dysfunction.

Limitation
The limitations of this study include the following: First, this was a retrospective and single center study with small sample size, and potential center-specific bias cannot be excluded. Second, we did not compare tissue characteristic parameters with strain parameters, because some previous studies have confirmed it. Third, IIM patients were older than the controls, with regard to this, data analysis were adjusted for age. Finally, CMR parameters were not compared with clinical and follow-up data (biochemical markers of myocardial damage, ECG abnormalities). The prognostic value of cardiac dysfunction in IIM detected by CMR-FT needs further exploration.

Conclusion
Impaired LV and LA deformation could be detected by CMR-FT in IIM patients. The global LV longitudinal PS, LAVpre-ac index and SRe can early detect cardiac involvement in IIM patients with preserved LVEF. Furthermore, LV strain parameters could reflect the severity of LV systolic dysfunction.