The application of exercise stress cardiovascular magnetic resonance in patients with suspected dilated cardiomyopathy

In this single-center prospective outcome study, patients were recruited between April 2016 and August 2018 for exercise stress CMR if they had imaging features of DCM on transthoracic echocardiogram or resting CMR. To be included in this study, patients did not have heart failure (New York Heart Association (NYHA) I) and had to satisfy at least one of the following two criteria using age- and sex-specific Asian CMR references values [2, 9]:

Patients with other cardiomyopathies such as hypertrophic cardiomyopathy, hypertensive heart disease, arrhythmogenic right ventricular dysplasia, LV non-compaction cardiomyopathy and ischemic cardiomyopathy were excluded. Ischemic cardiomyopathy was defined as the presence of myocardial infarction on CMR or hemodynamically significant coronary artery disease assessed with either invasive or computed tomography coronary angiography, as guided by conventional recommendations [2, 3, 6].

The physical activity levels in all participants were evaluated using the established General Practice Physical Activity Questionnaire. The self-administered questionnaire described the type and amount of physical activity in the workplace, the number of hours spent on different activities and their usual walking pace. The activity levels were classified into four categories: inactive, moderately inactive, moderately active and active [10].

Genetic sequencing was performed in all the patients with suspected DCM. Genomic DNA was extracted from patient’s whole blood using Chemagic DNA blood kit (PerkinElmer, Inc. Waltham, Massachusetts, USA) following the manufacturer’s protocol. Quality and quantity of the extracted DNA were assessed by Labchip Ds (PerkinElmer, Inc), an automated ultraviolet-visible spectrophotometer. Purified DNA was targeted enriched for genes related to DCM using TruSight Cardio kit (Illumina, San Diego, California, USA) according to the manufacturer’s protocol. Pooled libraries were sequenced using Illumina MiSeq (v2 kit) or NextSeq 500 (Mid Output v2 kit) benchtop sequencers using paired-end, 150 bp reads. Raw sequencing data were demultiplexed, trimmed and mapped to UCSC GRCh37/hg19 reference genome as previously described [11]. Variants were called using GATKv3.3 HaplotypeCaller and UnifiedGenotyper annotated using CardioClassifier, a semi-automatic classification on inherited cardiac conditions defined according to the American College of Medical Genetics and Genomics and the Association for Molecular Pathology [12]. The pathogenicity of the variants for each patient was further confirmed by a cardiologist who is an expert in genetics and DCM (SAC). He was blinded to the imaging and other clinical data.

Baseline breath-hold CMR at 1.5 T was performed in all healthy subjects and patients before initiating exercise stress (Siemens MAGNETOM Aera, Siemens Healthineers, Erlangen, Germany). Balanced steady-state free precision (bSSFP) cines were acquired in the standard long-axis (2-, 3- and 4-chamber) and short-axis extending from the base to the apex (8 mm thick and 2 mm gap; TE 1.2 ms, TR 3 ms; field of view 280-320 mm; acquired voxel size 1.6 × 1.3 × 8.0 mm³; 30 phases per cardiac cycle).

Exercise stress was performed with a programmable supine ergometer (Lode BV, Groningen, The Netherlands) fitted onto the CMR scanner table. The protocol has been described in detail previously [8]. In brief, the participants cycled with an initial workload of 25 W, at a cadence of at least 70 rpm for 1 min. Workload was gradually increased by 25 W every minute until exhaustion. Real-time prospective ECG-gated bSSFP cines in the short axis were acquired at every stage of the exercise (8 mm thick and 2 mm gap; TE 0.99 ms; TR 2.3 ms; field of view 225 x 300 mm; acquired matrix size 68 × 128 pixels; acceleration factor 4; acquired voxel size 3.3 × 2.3 × 8.0 mm³; temporal resolution 39.1 ms). As part of our protocol, patients taking beta-blockers were advised to stop taking them 2 days before the stress test. All healthy subjects and patients with suspected DCM successfully completed the exercise stress CMR protocol and were included in the analysis of this study.

In patients with suspected DCM, replacement and diffuse myocardial fibrosis imaging was assessed using late gadolinium enhanced (LGE) imaging and myocardial T1 mapping, respectively. LGE imaging was performed at 8 min after 0.1 mmol/kg of gadobutrol (Gadovist; Bayer Pharma AG, Germany) was administered. An inversion-recovery fast gradient echo sequence was used, and the inversion time was optimised to achieve appropriate nulling of the myocardium. Myocardial T1 mapping was performed using the Modified Look-Locker Inversion-recovery (MOLLI) sequence (flip angle 35; minimum TI 100 ms; TI increment of 80 ms). Native and 15-min post-contrast T1 maps were acquired using a heartbeat acquisition scheme of 5(3)3 and 4(1)3(1)2, respectively.

All image analysis was performed using standardized protocols at our NHRIS CMR Core Laboratory (CVi42, Circle Cardiovascular Imaging, Calgary, Canada). LV mass and cardiac volumes at baseline (breath-hold short axis cine images) and each exercise stage (real-time short axis cine images) were assessed in all patients. The exercise stress protocol demonstrated excellent scan-rescan (0.2 ± 05 L/min/m²) and inter-observer (LVEDV: 2.8 ± 5.2 mL; LVESV: − 0.5 ± 5.2 mL) reproducibility [8]. Exercise-related parameters examined were relative change in exercise LVEF, relative change in exercise CI and peak exercise CI expressed as age- and sex-specific percentiles.

Myocardial strain was analysed in the breath-hold cine images using the Tissue Tracking Plugin. Circumferential and radial strain were measured in the short axis cine images, whilst longitudinal strain was measured in the three long-axis views [13]. The amount of replacement fibrosis was quantified using a signal intensity threshold greater than two (2-SD) and five times (5-SD) the standard deviation above the mean value in a normal region of myocardium sampled in the same short-axis slice. Areas of artefacts or contamination (blood pool and/or epicardial fat) were manually excluded. The higher threshold was accurate compared with histological fibrosis whilst the lower threshold predicted adverse prognosis in patients with DCM [14, 15]. Diffuse myocardial fibrosis was assessed using extracellular volume fraction (ECV), estimated from the native and 15-min post contrast T1 map [16]. Interstitial volumes were defined as a product of ECV and myocardial volume (myocardial volume = LV mass/1.05 g/mL), a measure that correlated very well with histological fibrosis (r = 0.87, P < 0.001) and had prognostic value in patients with aortic stenosis [17].

Available clinical outcome data in all patients with suspected DCM were reviewed in June 2019. The primary outcome was a composite of all-cause mortality, cardiac decompensation and ventricular arrhythmias. Cardiac decompensation was defined as objective evidence of decreased LVEF and elevated brain natriuretic peptides, corroborated with heart failure symptoms; and ventricular arrhythmias were defined as sudden cardiac deaths, ventricular tachyarrhythmias (sustained and non-sustained) and ventricular fibrillation. These are clinically relevant end-points in DCM that were carefully adjudicated by an experienced cardiologist (CWLC), who was blinded to the imaging, genetics and clinical data.

Distribution of continuous variables was assessed using the Shapiro-Wilk test. Data were presented in either mean ± standard deviation or median [inter-quartile range], as appropriate. Depending on the normality of the distribution, parametric Student’s t test and 1-way ANOVA or the nonparametric Mann-Whitney U test and Kruskal-Wallis test were used to compare groups of continuous data. Categorical data were compared using the χ² test. The prognostic association of the peak exercise CI threshold was assessed using the Kaplan-Meier analysis. Statistical analyses were performed using GraphPad Prism 8.0.1 (GraphPad Software, Inc., San Diego, California, USA) and SPSS Version 24 (Statistical Package for the Social Sciences (SSPS), International Business Machines, Inc., Armonk, New York, USA). A two-sided P-value <0.05 was considered as statistically significant.

Most of the healthy subjects (76%) were at least moderately active. Cardiac volumes, LV mass and systolic function (including myocardial strain measures) were within normal limits (Table 1). Peak exercise CI was higher in males than females (9.0±1.3 versus 8.2±1.2 L/min/m²; P < 0.001) and correlated negatively with age (r = − 0.30; P < 0.001). There were no associations between peak exercise CI and intensity of physical activity on self-administered questionnaire, body mass index, heart rate or blood pressures (P > 0.05). Reference ranges in peak exercise CI were established in the healthy subjects, stratified by sex (Fig. 1). The peak exercise heart rate achieved was 149 [137 to 160] beats per min, corresponding to 82 [76 to 86] % of the age-predicted maximal heart rate. All achieved at least 50% of the age-predicted maximal heart rate.

A total of 60 asymptomatic patients with suspected DCM were recruited (Table 1). Most of them were at least moderately active based on the self-administered questionnaire. We defined confirmed DCM on the basis of positive genotypic and phenotypic (G+P+) expressions. There were 13 (21%) patients with confirmed DCM: 92% of pathogenic gene mutations were in the TTN gene (premature stop, frameshift and essential splice: I-band, n = 3; A-band, n = 9) and one missense in MYH7. The clinical and CMR characteristics were presented in Table 2. In the patients with confirmed DCM, 15% had a positive family history of cardiomyopathy and 38% had non-ischemic replacement fibrosis on CMR.

No patients with confirmed DCM had a peak exercise CI more than 35th percentile specific for age and sex (Fig. 2). Applying this threshold in G-P+ patients (n = 47), those patients with peak exercise CI less than 35th percentile (n = 16; 34%) shared similar characteristics as confirmed DCM in terms of demographic profiles, activity levels and myocardial strain values (Table 2). Of the 29 patients with low peak exercise CI, most did not have pathogenic gene mutation (n = 16; 55%) or fibrosis on CMR (n = 19; 66%). Approximately half had only one abnormal phenotype (13 had impaired LVEF and one had a dilated LV). Conversely, G-P+ patients with peak exercise CI greater than 35th percentile (n = 31; 66%) were younger, more active, had enlarged right ventricles and increased myocardial strain values compared to confirmed DCM (Table 2). Of the 31 patients, 87% and 80% of them were able to achieve peak CI exceeding 50th percentile and 80th percentile, respectively. Of note, the change in exercise LVEF and the extent of non-ischemic fibrosis were not able to distinguish between confirmed DCM and those with high exercise capacity (Table 2).

Over 21 ± 9 months of follow-up, there were 7 events (event rate of 6.0 events/100 patient-years). All seven patients developed evidence of cardiac decompensation during follow-up: LVEF decreased by 10–29% and NT-proBNP concentrations were elevated between 973 and 6701 pg/mL (all patients had normal renal function). One patient underwent implantation for LV assist device, one was advised for cardiac resynchronization therapy with defibrillation, and the third patient had an arrhythmic event. This patient developed episodes of non-sustained ventricular tachycardia, the longest was 7 s that occurred during sleep. She eventually received an implantable cardioverter-defibrillator. There were no deaths. In the 7 patients with adverse events, one in two had a pathogenic gene mutation and/or non-ischemic type replacement fibrosis on CMR (Additional file 1).

Despite small numbers and limited follow-up, we found prognostic significance of exercise CMR: all 7 events occurred in patients with low exercise capacities whilst those with peak exercise CI more than 35th percentile had no events (P = 0.004; Fig. 3). Of importance, there was no difference in event distribution stratified by either genotype status (P = 0.780) or fibrosis status on CMR (P = 0.118; Fig. 3). The other exercise-related parameters (heart rate and systolic blood pressure) were similar between those with and without adverse events (Additional file 1).

CMR has a well-established clinical role in ascertaining the underlying etiology in DCM [23]. Unlike cardiopulmonary exercise test or echocardiography, exercise stress CMR can assess cardiac function, myocardial mass, tissue characterization and exercise capacities in a single modality. Unlike exercise LVEF and other imaging parameters, peak exercise CI is a measure that incorporates a physiologic parameter (heart rate) in the assessment of exercise capacity. This could partially explain the ability of peak exercise CI to distinguish between pathological DCM and physiological exercise-induced cardiac remodeling.

DCM remains a challenging diagnosis in some patients despite advances in cardiac imaging. Particularly in patients with imaging features of DCM and do not have a pathogenic gene mutation, exercise stress CMR may increase the diagnostic confidence of distinguishing between pathological DCM and physiological exercise-induced cardiac remodeling. Further investigations are needed to discern the role of exercise stress imaging in differentiating between physiology and pathology in highly trained athletes with increased cardiac volumes and mildly impaired LVEF [24].

The event rates were low because of the short follow-up duration and the patients were asymptomatic (NYHA I) at the time of recruitment, which precluded further analysis to examine the incremental and independent prognostic value of the exercise stress imaging compared to other established prognostic markers. Despite these limitations, we observed that only patients with low peak exercise CI developed adverse prognosis. In the absence of any well-validated diagnostic criteria of DCM, we had used very stringent definition of confirmed DCM (positive genotype and phenotype). As a consequence, the sample size is relatively small, but an irrefutable diagnosis of DCM is essential to test the potential of low peak exercise CI. Most of the patients with suspected DCM were males. However, the peak exercise CI was expressed as age- and sex-specific percentiles to minimize any impact (if any) on the overall validity of the study. As echocardiography was not performed in the study, the diagnostic and prognostic value of echocardiographic parameters such as global longitudinal strain and diastolic function cannot be compared with peak exercise CI.