Background
Non-ischemic dilated cardiomyopathy (NDCM) is a condition that manifests as the enlargement of the left ventricle (LV) or both ventricles with systolic dysfunction or abnormal loading conditions [
1]. As one of the leading causes of systolic heart failure (HF), NDCM principally affects young adults, leading to tragic outcomes. Early diagnosis, standard therapy, and timely follow-up have led to remarkable achievements in NDCM patients’ prognosis, with an increase of 5-year free event survival rate from 62 to 93% over the past three decades [
2]. Several studies have recently reported that numbers of NDCM patients, including children and adults, have experienced drastic improvement of LV systolic function with tailored medical treatments on serial echocardiography follow-up [
3‐
6].
With excellent spatial resolution and high reproducibility, cardiovascular magnetic resonance (CMR) is considered the non-invasive gold standard for evaluating cardiac function and myocardial contraction [
7]. Nowadays, CMR can provide a more specific diagnosis of cardiomyopathies than conventional techniques [
8]. Precise assessment and diagnosis could have important clinical implications, especially in preventing clinical adverse events, choosing optimal treatment regimens, and timing of transplantation in cardiomyopathy patients. Myocardial strain analysis or feature tracking (FT) imaging provides more sophisticated information on cardiac function over and beyond conventional CMR derived volumes and global contractile function (LV ejection fraction [LVEF]) [
9]. Most studies investigating regional LV function and motion abnormalities have focused on longitudinal dimension [
10,
11]. Yet, the largest degree of myocardial deformation occurs in the radial direction, so that radial deformation and radial synchrony are more likely to be sensitive markers for predicting the prognosis of NDCM patients. Few studies have reported deformation assessment as a prognosticator in reversible NDCM patients. The purpose of this study was to evaluate if strain parameters assessed by CMR FT analysis could predict the reversible NDCM.
Methods
Study population
A total of 120 newly diagnosed (symptoms occurred within 2 months) NDCM patients based on the 1995 World Health Organization/International Society and Federation of Cardiology criteria [
12] in our heart center from January 1st 2011 to June 30th 2016 were consecutively enrolled. Inclusion criteria were impaired systolic function (CMR LV ejection fraction (LVEF) ≤ 45%). Exclusion criteria were: (1) ≥ 50% stenosis of a major coronary artery or branch based on invasive coronary angiography or computed tomography angiogram; (2) estimated glomerular filtration rate (eGFR) < 30 ml/min/1.73 m
2; (3) HF secondary to chronic lung disease; (4) valvular disease of moderate or greater severity; (5) active myocarditis; (6) other cardiomyopathies (ischemic; stress-related; tachycardia; peripartum; metabolic or endocrine diseases) [
13]. In addition, patients with any contraindications failed to accomplish CMR test (n = 3), CMR images of inadequate quality (n = 3), implantation of cardiac resynchronization and/or defibrillator therapies or LV assist devices unable to accomplish the follow-up CMR test (n = 31) and heart transplantation (n = 4) were excluded. After exclusion, 79 patients were included in the final analysis. At enrollment, detailed medical data were obtained from all patients, including physical examination, blood laboratory tests, 12-lead electrocardiogram (ECG), and echocardiography. Optimal medical treatments were prescribed and maintained over follow-up.
A median interval of 36 months (interquartile range [IQR], 30 to 48 months) between the enrollment and follow-up CMR(n = 79) were used to classify patients into two groups. Recovery was defined as a current LVEF of ≥ 50% and a CMR LV end-diastolic volume indexed to body surface area (LVEDVI) within the normal range; and plasma N-terminal pro-B-type natriuretic peptide (NT-proBNP) concentration less than 250 ng/L [
13,
14]. The remaining NDCM patients were classified as unrecovered. Another 25 healthy healthy subjects without any other primary diseases and abnormalities of CMR images were selected from our CMR imaging database.
The study was approved by the Ethics Committee of Shanghai Jiaotong University Affiliated No.6 Hospital. Written informed consent was obtained from all participants.
CMR technique
ECG-gated CMR studies were performed on a 3T CMR system (Philips Healthcare, Best, The Netherlands). Balanced steady-state free precision (bSSFP) cines were acquired in three long-axis slices (each plane with 60 degrees interval along the central axis of the LV; four-, three- and two-chamber). Subsequent short-axis cines extending from the mitral valve ring to the apex were obtained to cover the entire LV (8 mm parallel slices with no gap; TR = 3.2 ms, TE = 1.5 ms, flip angle 45, an in-plane resolution was 1.9 × 1.9 mm2, acquisition matrix 232 × 219; 30 phases per cardiac cycle). Late gadolinium enhancement (LGE) imaging was performed 10 min after injection of 0.2 mmol/kg of contrast (gadobutrol/Gadovist; Bayer Healthcare, Berlin, Germany).
CMR analysis
The analysis of LV volumes (end-diastolic volume (LVEDV); end-systolic volume (LVESV)), LVEF, right ventricular (RV) ejection fraction (RVEF), and LV mass (LVM) were performed using standardized protocols and dedicated software (cvi42, Circle Cardiovascular Imaging, Calgary, Alberta, Canada). LVM was estimated at end-diastole and corroborated at end-systole, which excluded the papillary muscles. LV cardiac index (CI) was calculated according to the formula below: CI =
\(\frac{LVEDV-LVESV}{BSA\times1000}\times HR\). Height and weight were measured in all patients, and body surface area (BSA) was calculated using the Mosteller formula [
15].
CMR feature tracking (FT) analysis was performed on the standard acquired bSSFP cine images using cvi42 (Circle Cardiovascular Imaging). LV endocardial and epicardial borders were automatically tracked and manually corrected in three long-axis slices and in short-axis slices in the end-diastolic phase. RV endocardial and epicardial borders were manually traced in the end-diastolic phase in long-axis slices and in short-axis slices. Left atrial (LA) endocardial and epicardial borders were manually traced in the apical four-chamber and apical two-chamber views at end-systole [
16]. Global LV and RV global longitudinal strain (GLS) was derived from the long-axis cines, while short-axis cines were used to deduce global LV and RV global circumferential (GCS) and global radial strains (GRS) [
17]. Segmental LV peak circumferential strain (CS) and radial strain (RS) were obtained from three consecutive parts (basal, middle, apical) from the mitral annulus to apex. To reflect the base-to-apex CS and RS gradient, the ratios of apical, basal CS and RS (CS
api/bas, RS
api/bas) were calculated as apical strain divided by basal strain, respectively. Instantaneous LV peak torsion was defined as the maximum difference in rotation angle between the base and apex divided by the distance between apical and basal slices automatically (Additional file
1). Dyssynchrony was assessed by variability in time to peak strain. The SD16-TTPLS, SD16-FTPCS, and SD16-FTPRS were calculated based on the standard deviation of time to peak longitudinal strain (LS), CS, and RS in 6 basal, 6 mid-ventricle, and 4 apical segments of the LV, respectively (16 segments in all). LA strain values for each tissue point and the reservoir strain values were automatically derived.
Quantitative assessment of myocardial fibrosis was performed on LGE imaging data on short-axis images using cvi42 (Circle Cardiovascular Imaging). Normal myocardium was visually defined as a region of myocardium without any apparent LGE on visual inspection. The mean signal intensity and standard deviation (SD) were determined by drawing a region of interest (ROI) in a portion of the normal myocardium on each slice. The semi-automated greyscale threshold technique was performed by using 2, 4, 6 SD above the mean signal intensity for the normal nulled myocardium. Results were reported as the percentage of LGE mass to the total LVM [
18].
Intra- and inter-observer agreement
Data from 10 healthy subjects and 10 NDCM patients were applied to test inter- and intraobserver variability. 2 independent cardiologists (***1 and ***2) specialized in CMR were blinded to each other’s recordings and conducted separate CMR analyses. All CMR measurements were analyzed by both observers. Data from the separate acquisitions were used to test interobserver variability. In order to test intraobserver variability, the observers re-analyzed their own recordings (3 weeks apart).
Statistical analysis
Summary statistics of clinical and image data were expressed as mean ± SD, median (quartile25–quartile75) or percentage, as appropriate. Comparisons between groups were made using one-way ANOVA analysis for continuous, normally distributed data and Wilcoxon rank-sum test for continuous, non-normally distributed data. The bonferroni correction was made for multiple comparisons. The chi-square test or Fisher’s exact test was calculated for categorical variables. Univariate and multivariate logistic regression analyses were performed to identify prognostic predictors for the recovered condition. Considering the collinearity and clinical significance, 4 variables (heart rate (HR), BSA, RSapi/bas, SD16-FTPRS) were included in the model with the predictive accuracy of 81%. Receiver-operating characteristic curve analysis was used to identify parameters that were best fit in diagnosing the recovered model of NDCM. The best cutoff value was based on the maximum Youden index. The Kaplan Meier survival analysis was applied to calculate the recovery rate. The intraclass correlation coefficient (ICC) was used to determine inter- and intraobserver reproducibility. All calculations were performed using SPSS (version 22.0, Statistical Package for the Social Sciences, International Business Machines, Inc., Armonk, New York, USA and GraphPad Prism (version 8.0, Graph-Pad Software, San Diego, California, USA).
Discussion
The purpose of this study was to evaluate if strain parameters assessed by CMR FT analysis could predict the reversible NDCM. Several new findings helped us to comprehensively understand the complicated prognosis of NDCM patients: (1) 27% NDCM patients had normalized LV size and systolic function after a median 36 months; (2) conventional LV systolic function parameters were similar between the recovered and unrecovered NDCM groups at baseline; (3) myocardial contractility at apical and intraventricular mechanical coordination were much better in recovered patients at baseline using CMR FT analysis, of which RSapi/bas and SD16-TTPRS were independent predictors of recovered condition.
Several studies have shown that the recovered condition incidence ranges from 14 to 41% in NDCM patients [
3‐
5]. In our cohort, 27% of NDCM had a favorable outcome with normalization of LV size and systolic function after receiving standard treatments, which may shed light on the diversified prognosis of patients with NDCM.
LVEF, reflecting global LV systolic function, is a universal marker for routine risk stratification and therapeutic strategy decision in patients with NDCM. In this study, the CMR examination revealed no difference in LVEF at baseline in the two NDCM groups. LVEF only revealed the variation of global and accumulative LV volume and function, which later changes in the pathological process and lacks sensitivity and specificity to predict the disease’s subsequent prognosis. More delicate and accurate parameters are required to further discriminate NDCM patients.
LV deformation is expressed as strain, which represents the fractional or percentage change of a region of interest from its original dimension [
19]. In this study, strain was applied to reflect the heart’s systolic contractility through relative displacement in three different dimensions (longitudinal, circumferential, and radial). It was found that global strains (GLS, GCS, and GRS) were similar between two NDCM groups, which at least implied no remarkable difference in global systolic contractility. Furthermore, it was also found that all the global and segmental strains dramatically declined in NDCM patients, while segmental CS and RS disproportionately dropped. A similar decline of basal and middle segments but relative apical preservation in recovered patients resulted in higher CS
api/bas and RS
api/bas compared to unrecovered. In other words, the heterogeneity of contractile injury exists in patients with NDCM. Similar to our results, Bach et al. [
20] found the basilar-septum was a “sentinel” region injured earlier than other LV regions. In this study, the base-to-apex strain could help to differentiate the recovered NDCM from the unrecovered. With the spherical ventricular geometry variation, unrecovered patients had the most severely injured segments at the apical. The preserved apical strain is a compensatory mechanism that maintains LV systolic function. In the physiological state, the basal segment’s wall stress is higher than at the middle and apical segments due to the non-spherical ventricular geometry and the largest local radius of the LV curvature at the basal segment. For NDCM patients, the high wall stress at the basal segment will lead to cardiomyocyte necrosis and fibrosis. In addition, a greater diversity of myocardial fibers and matrix orientations at the apical segment compared with the basal segment could also contribute to the preserved deformation at the apical segment [
21,
22]. This finding implied that the difference of base-to-apex strain among NDCM patients might indicate a different prognosis.
Moreover, the decreased contractility of the localized part and the loss of consistency of regional myocardium could also affect systolic function, which may directly weaken the cardiac pump. Therefore, more research is focusing on mechanical synchronism. Our study demonstrated that significant intraventricular mechanical dyssynchrony in regional longitudinal, circumferential, and radial dimensions were detected in unrecovered patients, and the difference was more obvious in the radial dimension. On the contrary, better mechanical synchronization was found in recovered patients; however, it was not as homogeneous as in control subjects, whose septum contracts slightly earlier than the lateral and inferolateral walls [
22]. Intraventricular electrical conduction delay may exist in quite a few NDCM patients, while some HF patients with a narrow QRS duration (< 120 ms) may also exhibit significant mechanical dyssynchrony [
23,
24]. Also, we found some patients with a long QRS duration (> 120 ms) but recovered LV size and systolic function. Our findings indicated that intraventricular mechanical dyssynchrony rather than a complete left bundle block might have a crucial role in the prognosis in NDCM patients [
25]. The heart squeezes blood out of the LV cavity by any two of the opposing wall contracting almost at the same time. If any ventricular segment reaches peak strain earlier than others, there is an ineffective shifting of blood within the LV cavity resulting in smaller LV stroke volume. Moreover, the earlier contraction wall directly influences its opposing wall by overstretching the myocardium and causing increased wall stress and preload, which may further reduce its contractility [
26]. Also, this may be one of the potential reasons why recovered patients with relatively better mechanical accordance are more likely to have the contractility recovered.
To the best of our knowledge, this is the first study that reported on the use of CMR FT strain analysis for prediction of functional recovery in NDCM patients. We found 27 % of NDCM patients had recovered LV size and systolic function in this cohort. RSapi/bas and SD16-TTPRS assessed by CMR at presentation may prove to be non-invasive parameters for prediction of recovery in patients with NDCM. Global and segmental strains of the recovered NDCM patients will be analyzed and reported in our post-study, which may offer greater insight into the disease.
The present study has several limitations. As this was a single-center study, the sample size was modest. Though RSapi/bas and SD16-FTPRS predict recovery, the area under the curve for each variable was below 0.80. More markers such as native T1 or extracellular volume fraction could not be performed, as both were not available in the majority of the patients. It remains unknown how much longer the recovered NDCM patients could maintain their morphologic and functional recovery. Long-term follow-up and molecular studies are needed to make a preliminary stratification of those patients at presentation.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.