Introduction
Stress cardiac magnetic resonance imaging (CMR) offers effective cardiac prognostication in patients with stable chest pain syndromes and is frequently performed using pharmacological agents [1, 2]. Exercise stress testing, however, provides a physiological challenge to the myocardium, with the ability to link physical activity to symptoms and ischemia [3–5]. CMR, while being the clinical gold standard for assessing cardiac morphology and systolic function, remains underutilized as an exercise imaging modality, mostly because of the limited availability of MR-compatible exercise equipment [6]. Several investigators demonstrated exercise stress CMR using an in-room treadmill system followed by a rapid transfer of the subject into the scanner with subsequent imaging [7–12]. On the other hand, MR-compatible equipment enabling imaging at 1.5 tesla (T) while the patient is at peak exercise stress was shown to be feasible for the evaluation of blood flow dynamics and ventricular function in healthy volunteers [13–20]. Another proof-of-concept study reported on a 3T compatible ergometer, which allowed the assessment of cardiac morphology and function in healthy subjects [6]. Apart from safety concerns, the improvement of the signal-to-noise ratio, and subsequently the spatial and temporal resolution, makes high-field imaging particularly attractive. Due to the limited availability of MR-compatible exercise equipment for high magnetic field strengths; however, publications showing the feasibility of cardiac exercise stress testing at 3T in patients with suspected coronary artery disease (CAD) are lacking. Therefore, the purpose of this study was to evaluate a newly designed 3T conditional pedal ergometer for cardiac stress imaging using rapid CMR acquisition techniques prior to and immediately after maximal exercise stress in healthy volunteers and in patients with known or suspected CAD.
Methods
Study population
Ten healthy volunteers and 11 patients with known or suspected CAD were recruited for this single-center, prospective, feasibility study. Healthy subjects met the inclusion criteria of age between 20 and 80 years and no medical history of cardiovascular, metabolic, or neurological disease. Inclusion criteria for patients included the same age limits as for healthy volunteers and an additional referral to elective invasive coronary angiography (CAG) at the local department of cardiology based on known or suspected CAD with moderate to high pretest probability for CAD. Exclusion criteria for both healthy subjects and patients included physical exercise restrictions and general contraindications for CMR. The study complies with the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the local ethics committee. Written informed consent was obtained from all participants before inclusion into the study.
Equipment and exercise protocol
Exercise testing was conducted using a newly developed, commercially available ergometer, which is compatible with MR scanners up to 7T (Ergospect GmbH, Cardio Step Module). The pedal ergometer consists of a 2-cylinder hydraulic system connected to an air pressure system for power control. The resistance that must be overcome for movement of the exercise pedal is generated by hydraulic throttling of a piston-cylinder system. For this purpose, a hydraulic throttle valve pneumatically piloted by means of a differential pressure valve is used. A control unit regulates the air pressure, by which the hydraulic system resistance can be regulated continuously with manual force adjustment. All structural components, and power and control systems were constructed using different kinds of plastic and aluminum in order to prevent artifacts during the CMR measurements. The pedal ergometer is mounted on the scanner table, permitting supine positioning of the patient. Velcro straps were used to fix patients’ feet to the pedals, and patients were in addition secured to the scanner table using a special harness in order to prevent movement artifacts. This enables CMR measurements during exercise within the isocenter of the MR scanner without any repositioning of the patient. With this setting, post-exercise imaging could be started immediately after the end of exercise. The incremental maximal exercise protocol started with pedaling against a pressure of 2 atmospheres (atm) at a cadence of 30 revolutions per minute (rpm) and progressing by 0.5 atm every 2 minutes up to volitional exhaustion, which was characterized as a significant and persistent loss of cadence (≥ 10 rpm despite verbal encouragement to continue). Heart rate and work rate were recorded continuously.
Cardiac magnetic resonance imaging
Cardiac magnetic resonance imaging scans were performed on a 3T MR scanner (Magnetom Skyra, Siemens Healthineers AG). The CMR protocol included:
- Plethysmography-triggered first-pass perfusion-balanced steady-state free precession sequence (bSSFP/true fast imaging with steady-state precession [trueFISP]) at rest conditions during intravenous administration of 0.15 mmol/kg of Gd-DO3A-butriol (Gadovist®, Bayer Vital), which was administered using an automatic injector (Spectris Injection System, Medrad) at 2 mL/s, followed by 20 mL of saline flush. Three short-axis slices and one long-axis slice were acquired;
- Real-time cine imaging at rest and immediately after the end of exercise, using a non-gated, free-breathing bSSFP acquisition for volumetric measurements. A short-axis stack covering the left ventricle (LV) and 3 long-axis images (a 3-chamber, a 2-chamber, and a 4-chamber slice) were acquired;
- High-resolution cine imaging post exercise using retrospectively plethysmography-triggered bSSFB bright-blood sequences with breath hold for the detection of any regional LV wall motion abnormalities (WMA). Three short-axis and 3 long-axis slices were acquired;
- Late gadolinium enhancement imaging (LGE) by using a plethysmography-triggered single-shot phase-sensitive inversion recovery (PSIR) sequence with bSSFP readout. A short-axis stack covering the LV and 3 long-axis images were acquired.
The imaging parameters of the respective sequences are shown in Table 1.
|
First-pass perfusion |
Real-time cine |
Segmented cine |
Single-shot PSIR |
|
|
Type |
2D |
2D |
2D |
2D |
|
Repetition time [ms] |
2.30 |
2.70 |
3.00 |
3.40 |
|
Echo time [ms] |
1.15 |
1.22 |
1.26 |
1.69 |
|
Flip angle [°] |
10.00 |
36.00 |
39.00 |
40.00 |
|
Acquisition matrix |
192 × 150 |
128 × 66 |
156 × 256 |
192 × 256 |
|
Field of view [mm] |
360 × 322 |
320 × 380 |
308 × 380 |
285 × 380 |
|
Slice thickness [mm] |
6 |
8 |
8 |
7 |
|
Voxel size [mm3] |
1.875 × 1.875 × 6 |
2.97 × 2.97 × 8 |
1.48 × 1.48 × 8 |
1.48 × 1.48 × 7 |
|
Number of slices and orientation |
3 SAX, 1 LAX |
10 SAX covering LV, 3 LAX |
3 SAX, 3 LAX |
10 SAX covering LV, 3 LAX |
|
Parallel imaging mode |
GRAPPA |
GRAPPA |
GRAPPA |
GRAPPA |
|
Parallel imaging factor |
2 |
3 |
2 |
2 |
|
phase encoding step per cardiac cycle |
75 |
22 |
15 |
84 |
|
Temporal resolution [ms] |
199 |
60 |
44.25 |
509 |
|
Fat saturation |
No |
No |
No |
No |
|
Averages |
1 |
1 |
1 |
1 |
Image analysis
The overall subjective image quality of real-time cine images at rest and post-exercise condition was graded by 2 radiologists (A.M. and M.S.) with several years of experience in cardiovascular imaging (12 and 22 years, respectively), blinded to the results of each other as well as to clinical information. Subjective image quality grading was based on a 4-point Likert scale (1 = excellent image quality, 2 = good image quality, 3 = acceptable diagnostic despite impairments by artefacts, 4 = non-diagnostic). Volumetric evaluation of rest and post-exercise real-time cine images was performed using standard software (ARGUS, Siemens Erlangen, Germany). Contouring of LV endo- and epicardial borders was performed semi-automatically by both observers, and mean values were used for statistical analysis.
High-resolution cine images at rest and post-exercise conditions were visually analyzed by both observers. Regional WMAs were defined as deterioration in regional wall motion within a myocardial segment after exercise relative to the wall motion observed during rest imaging. Thereby wall motion was labeled as hyperkinetic, normal, hypokinetic, akinetic, or dyskinetic. Myocardial ischemia was defined if ≥ 3 out of 16 American Heart Association (AHA) segments showed new WMA [21]. Each observer was blinded to the ratings of the other.
Statistical analysis
Statistical analysis was conducted using Statistical Package for Social Sciences (SPSS) version 24.0 (IBM AG). Categorical variables are presented as frequencies and corresponding percentages. The Shapiro-Wilk test was applied to test for normal distribution (ND). All results are expressed as mean ± standard deviation (SD) (if ND) or as median with interquartile range (IQR) (if not ND). Correlations of continuous variables were calculated with Pearson’s or Spearman’s rank test as appropriate. Differences of the recorded parameters were compared with Student’s t test for paired or unpaired variables (ND) or Mann-Whitney U test (if not ND). The interclass correlation coefficient (ICC) from a 2-way mixed model for single measures was used to compare image quality scores between observers. For all tests, a 2-tailed p value < 0.05 was considered statistically significant.
Results
Ten healthy volunteers (mean age 44.9 ± 16 years, 9 [90%] males) and 11 patients (mean age 60.1 ± 8.9 years, 8 [73%] male) completed the feasibility study. The mean age of the overall study cohort was 52.9 ± 14.65 years (17 [81%] males). Detailed demographic characteristics of the study cohort are shown in Table 2.
|
Characteristic |
All subjects (n = 21) |
Healthy (n = 10) |
Patients (n = 11) |
P |
|
Age [years] |
52.9 ± 14.6 |
44.9 ± 16 |
60.1 ± 8.9 |
0.013 |
|
Female |
4 (19%) |
1 (10%) |
3 (27%) |
0.512 |
|
Male |
17 (81%) |
9 (90%) |
8 (73%) |
|
|
Body mass index [kg/m2] |
26.5 ± 4.1 |
25.9 ± 4.6 |
27 ± 3.7 |
0.852 |
|
Body surface area [m2] |
1.89 ± 0.17 |
1.89 ± 0.2 |
1.9 ± 0.13 |
0.106 |
|
Heart rate at rest [bpm] [IQR] |
62 [62–73] |
65 [60–73] |
67 [62–77] |
0.973 |
|
Heart rate postexercise [bpm] [IQR] |
111 [104–143] |
125 [105–166] |
111 [100–125] |
0.349 |
Exercise testing by pedal ergometer focused on the maximum possible stress level of healthy volunteers and patients. The primary reason for stopping exercise was the insurmountable resistance level (all of the healthy volunteers, 8/11 of the patients). Two (22%) patients reported breathlessness, and 1 patient stopped exercise because of mild angina pectoris symptoms. Twenty-four percent of the participants (5/21) met 85% of the age-predicted maximal heart rate (as calculated by 85% × [220 – age]), whereof 2/10 (20%) were healthy volunteers and 3/11 (27%) were patients. Seventy percent of the age-predicted maximal heart rate (as calculated by 70% × [220 – age]) was achieved for 71% (15/21) of the participants (6/10 healthy volunteers, 9/11 patients).
Cardiac imaging at rest and immediately post exercise was successfully completed in all of the subjects. The mean examination time for completing the whole study protocol including localizers, first-pass-perfusion imaging, and real-time and high-resolution cine imaging at rest and post-exercise including a test run with the pedal ergometer as well as LGE imaging was 22.3 ± 3.5 min. Absolute exercise time using the multistage strain protocol was on average 7.8 ± 3.2 min. Post-exercise real-time cine imaging was completed within 3 ± 1 s after the end of exercise, and post-exercise high-resolution cine imaging under breath hold was finished after an additional 53 ± 8 s.
Image quality
Median subjective image quality of real-time cine studies at rest was 1 (IQR 1–2) and 2 (IQR 2–2.5) for post-exercise real-time cine (p = 0.001). Images of real-time cine studies under rest and post-exercise conditions were evaluated with good to excellent inter-observer reliability (ICC = 0.999 and 0.862, respectively, all p < 0.0001; Fig. 1). Image quality did not differ significantly between patients and healthy volunteers (p > 0.05).
Figure 1. Representative cardiac magnetic resonanse images of a 48-year-old healthy volunteer. Left column: Real-time cine of basal short-axis slice at rest end-diastolic (A) and end-systolic (a). Second column: Real-time cine post-exercise end-diastolic (B) and end-systolic (b). Third column: high-resolution cine at rest end-diastolic (C) and end-systolic (c). Right column: late gadolinium enhancement short-axis slice (D) and 4-chamber view (d). Heart rate at rest 62 bpm, heart rate post-exercise 111 bpm.
Cardiac reserve and regional wall motion abnormalities post-exercise
Heart rate, LV ejection fraction, stroke volume, as well as cardiac output and cardiac index increased significantly from rest to post-exercise conditions in healthy volunteers as well as patients (all p < 0.0001). The increase in stroke volume was mediated by a reduction in end-systolic volume (ESV) in healthy volunteers and patients (p < 0.002 and p < 0.0001, respectively). End-diastolic volume (EDV) did not change with exercise (all p > 0.066; Tables 3 and 4, Fig. 2).
|
Rest (n = 21) |
Post exercise (n = 21) |
P |
|
|
Heart rate [bpm] [IQR] |
62 [62–73] |
111 [104–143] |
0.0001 |
|
EF [%] |
62 ± 9 |
71 ± 8 |
0.0001 |
|
EDV [mL] |
140 ± 32 |
142 ± 30 |
0.094 |
|
EDV [mL/m2] |
74 ± 16 |
75 ± 16 |
0.067 |
|
ESV [mL] |
55 ± 20 |
42 ± 21 |
0.0001 |
|
ESV [mL/m2] |
29 ± 11 |
22 ± 12 |
0.0001 |
|
SV [mL] |
85 ± 21 |
101 ± 19 |
0.0001 |
|
SV [mL/m2] |
45 ± 10 |
53 ± 9 |
0.0001 |
|
CO [L/min] |
5.6 ± 1.5 |
12.5 ± 4 |
0.0001 |
|
CI [L/min/m2] |
2.9 ± 0.7 |
6.6 ± 1.9 |
0.0001 |
|
Rest conditions |
Post-exercise conditions |
P |
|
|
Healthy volunteers (n = 10) |
|||
|
Heart rate [bpm] [IQR] |
65 [60–73] |
125 [105–167] |
0.0001 |
|
EF [%] |
64 ± 8 |
73 ± 6 |
0.005 |
|
EDV [mL] |
136 ± 34 |
138 ± 40 |
0.242 |
|
EDV [mL/m2] |
72 ± 15 |
73 ± 15 |
0.186 |
|
ESV [mL] |
48 ± 14 |
36 ± 13 |
0.002 |
|
ESV [mL/m2] |
25 ± 7 |
19 ± 7 |
0.002 |
|
SV [mL] |
88 ± 28 |
102 ± 24 |
0.0001 |
|
SV [mL/m2] |
46 ± 13 |
54 ± 11 |
0.0001 |
|
CO [L/min] |
5.6 ± 1.9 |
13 ± 5 |
0.0001 |
|
CI [L/min/m2] |
2.9 ± 0.9 |
7 ± 2.5 |
0.0001 |
|
Patients with known or suspected CAD |
|||
|
Heart rate [bpm] [IQR] |
62 [62–77] |
111 [100–125] |
0.0001 |
|
EF [%] |
59 ± 9 |
70 ± 10 |
0.0001 |
|
EDV [mL] |
145 ± 30 |
146 ± 29 |
0.263 |
|
EDV [mL/m2] |
77 ± 16 |
76 ± 17 |
0.244 |
|
ESV [mL] |
61 ± 24 |
47 ± 26 |
0.0001 |
|
ESV [mL/m2] |
32 ± 13 |
25 ± 15 |
0.0001 |
|
SV [mL] |
84 ± 14 |
100 ± 14 |
0.0001 |
|
SV [mL/m2] |
44 ± 7 |
53 ± 7 |
0.0001 |
|
CO [L/min] |
5.6 ± 1 |
11.7 ± 1.8 |
0.0001 |
|
CI [L/min/m2] |
2.9 ± 0.6 |
6.2 ± 0.9 |
0.0001 |
Figure 2. Graphs show variation of heart rate at rest and immediately post-exercise (A), ejection fraction (B), end--diastolic volume index (C), end-systolic volume index (D), cardiac output (E), cardiac index (F), and as assessed by free-breathing real-time cine imaging. Error bars = mean ± standard deviation; ***p < 0.0001; NS — non-significant.
Two of 11 patients showed inducible regional WMA on post-exercise high-resolution cine images, without evidence of any ischemic scar in the LGE sequences. One patient showed hypokinetic WMA in the anteroseptal segments at midventricular and apical LV level (4/16 AHA segments) on post-exercise cine images. Subsequent CAG of this patient revealed a 90% proximal left anterior descending artery — first diagonal artery bifurcation stenosis, resulting in percutaneous coronary intervention with stent implantation (Fig. 3). Another patient with hypokinetic WMA of the anterior segments at basal and midventricular level (3/16 AHA segments) under stress imaging conditions subsequently needed stenting of a 90% stenosis of the Ramus intermedius. All the other patients did not showed any newly regional WMA on post-exercise high-resolution cine images or any coronary stenosis requiring intervention on subsequent CAG.
Figure 3. Representative cardiac magnetic resonance imaging images of a 65-year-old female patient with suspected coronary artery disease. Upper row: High-resolution cine post-exercise end-diastolic on a basal short axis slice (A), midventricular short-axis slice (B), and apical short-axis slice (C). Lower row (a–c): High-resolution cine post-exercise end-systolic frames show hypokinetic wall motion abnormality in the anteroseptal segments of midventricular (b) and apical slice of the left ventricle (c), as illustrated by reduced end-diastolic to end-systolic wall thickening in bullseye diagram (D). Percutaneous coronary angiography on the following day depicted a 90% proximal left anterior descending artery — first diagonal artery bifurcation stenosis (E, arrowhead), which was intervened by stent implantation. Heart rate at rest imaging 59 bpm, heart rate at post-exercise imaging 125 bpm.
Discussion
To the best of our knowledge, this study is the first to demonstrate the feasibility of physical stress imaging by a 3T compatible pedal ergometer in healthy volunteers and patients with known or suspected CAD. The main study findings are the following:
- Our newly developed 3T compatible pedal ergometer enables exercise testing in order to image cardiac responses to dynamic stress;
- The current results suggest a reliable ability to image myocardial ischemia in patients with CAD triggered by pedal ergometer exercise inside the bore of the magnet;
- Free-breathing real-time cine images acquired immediately post exercise while the patient is at their peak heart rate provides good image quality.
While CMR is rarely used for imaging cardiac exercise stress in general, the use of exercise equipment, particularly at higher magnetic field strengths, still seems to be very limited. 3T CMR, however, has become widespread and offers advantages over 1.5T for a broad range of applications, for example perfusion imaging and LGE [22, 23]. Furthermore, the optimized signal-to-noise ratio and image resolution at 3T compared to 1.5T permit the detection of small and slight volumetric/ /structural alterations [24]. On the other hand, for CMR at higher magnetic field strength, greater efforts are needed to prevent artefacts particularly in the setting of ECG triggering [25, 26]. In addition, stress imaging may worsen heart rate variability and arrhythmias, leading to skipped cardiac cycles with the potential of losing diagnostically important information. Consequently, several study groups are working on alternative gating methods like Doppler ultrasound, acoustic triggering, or self-gating [27–31]. Another widespread technique, the feasibility of which was shown earlier [32–34], is peripheral pulse gating (photoplethysmography), as was used in the current study.
It was shown previously that exercising outside of the MR scanner with subsequent fast transition into the scanner for image acquisition can be performed within 60 s of peak exercise [6–12]. However, hemodynamic recovery occurs rapidly, and caution is required when interpreting data acquired more than 4 s post exercise. This is especially important in the younger, healthy population, but also clinical patients show a comparable decline in heart rate [6, 9]. Our results demonstrate that post-exercise real-time cine imaging can be completed within 3 ± 1 s after the end of exercise to comprehensively image cardiac responses to dynamic stress. Regarding imaging of myocardial ischemia, exercise-induced regional WMA is known to begin to diminish immediately following cessation of exercise [35]. For this reason, we used rapid post-exercise, high-resolution cine imaging in this study, in order to perform this imaging step within the first minute after exercise cessation. Thus, newly emerged regional WMA could be detected in 2 patients who subsequently underwent coronary intervention. It is necessary to confirm these results in a future study with a significantly larger number of patients.
In line with previous reports, our data suggest that supine positioning within the bore of the magnet in combination with leg elevation can produce near maximal EDVs because both healthy subjects and patients showed no EDV reserve during supine exercise. Along with an increase in heart rate, cardiac reserve was mediated by a significant reduction in ESV from rest to post-exercise conditions. This resulted in an increase of the cardiac index for healthy people and patients by a factor of more than 2. In fact, conventional upright cycling showed major ESV changes compared to supine exercise, reflecting recruitment of a greater muscle mass [12]. However, prior investigations using an MR-compatible ergometer showed reduced metabolic demand by supine exercising [6, 16]. Moreover, the stepwise resistance approach used in our study enables a steady-state workload. Because the heart rate has the strongest association with myocardial work of all non-invasive measures, its quantification is crucial during cardiac function studies [35]. Admittedly, only 24% of the participants in the current study met 85% of age-predicted maximal heart rate whereas 71% achieved 70% of age-predicted maximal heart rate. However, our exercise testing by pedal ergometer focused on the maximum possible stress level of participants, with an insurmountable resistance level being the primary reason for aborting the exercise (18/21 participants). Maximum heart rates achieved in this way are at a comparable level to previous MR studies using supine exercise testing while still allowing reliable synchronization of MR data acquisition by peripheral pulse gating [6, 13, 36–38] as well as good subjective image quality for post-exercise real-time Cine images.
Conclusions
Our findings suggest that exercise stress CMR using the proposed 3T MR conditional pedal ergometer is clinically feasible. It enables reliable imaging of functional response in healthy subjects as well as myocardial ischemia triggered by dynamic stress in patients with known or suspected CAD. For the latter, in particular, confirmation of this finding in larger trials is essential.
Funding
This study was supported by K-Regio, project number WIF-275-01-00002/01-0042.
