Moderate intensity supine exercise causes decreased cardiac volumes and increased outer volume variations: a cardiovascular magnetic resonance study

Twenty-six healthy volunteers (six women) aged 30±8 years (mean±SD) (range 19–59) underwent CMR at rest and during exercise with one-legged knee extensions. A custom built MR-compatible ergometer provided concentric resistance during knee extension by a rope and pulley system which was integrated with a mechanically braked flywheel. A strap connected to a variable weight system provided resistance and weight was added to achieve an exercise level at approximately 40 beats per minute (bpm) higher than the subjects’ resting heart rate. The subjects were connected at the ankle to the axle of the flywheel by a rope and the extension phase of the exercise turned the flywheel. Gravity returned the leg to the starting position and a gearing system on the axle returned the rope to the starting position at the end of each duty cycle.

Six subjects underwent a total of five CMR scans to investigate the reproducibility of volumetric measurements during exercise and the potential effects of different respiratory phases as well as differences in exercising muscle mass. The scans were divided into two sessions with a 1.5 hour rest outside the scanner between them. Session 1 included a) CMR at rest; b) CMR with 1-legged exercise at end-expiratory breath hold; and c) CMR with 2-legged exercise at end-expiratory breath hold. Session 2 included a) CMR with 1-legged exercise at end-expiratory breath hold; and b) CMR with 1-legged exercise at end-inspiratory breath hold with the instructions to keep an open glottis and avoid Valsalva-like increases in intra-thoracic pressures.

A 1.5T scanner (Philips Achieva, Philips, Best, The Netherlands) with a 5 channel cardiac coil was used to scan all subjects in the supine position. A balanced steady-state free-precession (bSSFP) sequence with retrospective ECG gating was used to acquire images of the heart (repetition time typically 3.0 ms, turbo factor 16, echo time 1.5 ms, flip angle 60°, reconstructed to a spatial resolution of 1.4 × 1.4 mm, acquired temporal resolution typically 50 ms reconstructed to 30 ms, and slice thickness 8 mm with no slice gap). After defining the long-axis orientation of the heart, short-axis images covering the heart from the base of the atria to the apex of the ventricles were obtained. Breath-hold during imaging during exercise was typically 6 s for long-axis images and 8–10 s for each short-axis slice. An ECG-triggered phase-contrast sequence was used to measure blood flow in the aorta (repetition time 8.6 ms, echo time 6.4 ms, 150 cm/s velocity encoding, slice thickness 8 mm). The measurement plane was positioned perpendicular to the vessel. Heart rate was obtained from the ECG during image acquisition.

All measurements were done using the software Segment 1.8 (http://​segment.​heiberg.​se) [24]. Left and right atrial volumes were measured in short-axis images at the time of ventricular end-diastole and ventricular end-systole. Left ventricular mass (LVM), end-diastolic volume (LVEDV), end-systolic volume (LVESV) and stroke volume (LVSV) were measured in short-axis images using planimetry, by manual delineation of endocardial and epicardial borders of the left ventricle. Papillary muscles were not included in LVM measurements. Right ventricular end-diastolic volume (RVEDV), end-systolic volume (RVESV) and stroke volume (RVSV) were measured in short-axis images by manual delineation of the right ventricular endocardial and epicardial border.

Total heart volume (THV) was measured in short-axis images by planimetry [22] and was defined as the volume of all structures within the pericardium, including myocardium, blood pool, atria, pericardial fluid and the proximal parts of the great vessels.

Atrio-ventricular plane displacement (AVPD) was determined from CMR long-axis images as previously described [15]. Longitudinal pumping of the left and right ventricle was calculated as the distance travelled by the AV-plane multiplied by epicardial short-axis area of the ventricle [25]. Radial pumping was determined from the total heart volume variation (THVV) [15]. Longitudinal and radial contribution to SV (%) was calculated as longitudinal pumping divided by SV and radial pumping divided by SV.

Statistical analysis was performed using SPSS statistics 20 (IBM, Chicago, IL, USA) and a p-value <0.05 was considered statistically significant. Paired t-tests were used to test for changes between rest and exercise. Wilcoxon non-parametric test was used to test for differences between rest and exercise in the subgroup of six subjects who underwent a total of five scans to investigate the reproducibility of measurements. Results are presented as mean ±SEM unless stated otherwise. Inter-observer variability was determined for the left ventricular measurements in ten subjects during rest and exercise.

Heart rate increased significantly during exercise from 60±2 to 94±2 bpm (p<0.001). Left atrial volumes at end-diastole decreased from 39±3 to 35±3 mL (p<0.05) as did right atrial end-diastolic volumes from 65±5 to 56±4 mL (p<0.05). At ventricular end-systole, where the atria reach their largest volumes, left atrial volumes were unchanged (84±4 and 85±5 mL, p=0.72) and right atrial volume decreased significantly from 124±7 to 103±7 mL (p<0.001) with exercise.

Left ventricular EDV remained unchanged during exercise (186±8 to 185±8 mL, p=0.81) and LVESV decreased from 82±4 to 74±4 mL (p<0.05) (Figure 2A-B). Left ventricular SV increased from 104±4 to 111±5 mL (p<0.05, Figure 2C). For the right ventricle, both RVEDV and RVESV decreased from 203±9 to 185±10 mL for RVEDV and from 100±6 to 77±6 mL for RVESV(p<0.001 for both) (Figure 2D-E) but the increase in right ventricular SV from 104±4 to 108±5 mL was not significant (p=0.18) (Figure 2F). Both left and right ventricular ejection fraction (LVEF and RVEF) increased during exercise from 56±1 to 60±1% and from 52±1 to 59±1% respectively (p<0.01 for both). Cardiac output increased from 6.2±0.3 to 10.4±0.5 L/min (p<0.001) mainly due to the increase in heart rate (Figure 3A-B). Interestingly, total heart volume decreased significantly during exercise by −30±8 mL, (p<0.01) corresponding to a 4±1% decrease of volume (example shown in Figure 4 and in Additional file 4). As expected, LVM was unchanged from rest to exercise (115±6 to 114±6 g, p=0.62).

Left ventricular AVPD and RVAVPD remained unchanged during exercise. Left ventricular AVPD was 14.6±0.3 mm at rest and 15.3±0.5 mm during exercise (p=0.06) and the RV AVPD was 20.9±0.6 mm for both rest and exercise (p=0.90).

Left ventricular longitudinal contribution to SV (%) remained unchanged at approximately 60% (59±1% at rest and 60±2% at exercise, p=0.74). Right ventricular longitudinal contribution (%) decreased from 81±2 to 75±2% (p<0.05) (Figure 5A-B) due to the decrease in RV end-diastolic volume. Total heart volume variation increased during exercise from 5.9±0.5 to 9.7±0.6% (p<0.001) (Figure 5C).

For the six subjects participating in repeated scans there were no differences in THV, RVEDV or left and right SV between the first and second exercise session with one leg. Left ventricular EDV increased more during the second exercise session when compared to rest; a 5±5% increase during the first session and a 11±4% increase during the second session (p<0.05). When comparing end-expiratory breath hold with end-inspiratory breath hold, only LVEDV differed between sessions. When compared to rest there was a 5±5% increase at end-expiratory breath hold and a 15±7% increase at end-inspiratory breath hold (p<0.05). During exercise using two legs, left and right ventricular EDV did not differ when compared to one-legged exercise. Left ventricular EDV increased by 5±5% and 7±7% respectively (p=0.35), and RVEDV decreased by −4±6% and −2±7% respectively (p=0.17). Right ventricular SV, however, increased more during exercise using two legs. When compared to rest the increase was 3±10% with one leg and 12±9% with two legs.

Results are presented as mean ±SD. At rest, inter-observer variability for LVM was 9±5 g, EDV −1±4 mL and SV −2±5 mL. During exercise imaging was more difficult and the image quality was lower, which is reflected by a slightly larger variability; LVM 7±10 g, EDV −3±16 mL and SV 0±9 mL.

The inconsistent results of previous studies [4‐7, 11, 12, 14, 26] may be explained by differences in imaging modalities and, perhaps most important, body position. The results of this study are in line with other studies of supine exercise showing unchanged LVEDV [5, 9, 10, 27, 28]. The significant decrease in RVEDV differs from previous studies of RV volumes during supine exercise using radionuclide ventriculography [13] and CMR [14, 28] where RVEDV remained unchanged during moderate intensity exercise (mean HR in these studies were 112,120 and 100 bpm respectively). However, in line with our results the study by Mols et al. [13] used radionuclide ventriculography and showed a decreased RVEDV at workloads at a HR of 127 bpm and above. The differences between the present study and previous CMR studies may be explained by differences in exercise protocol where we acquired breath-hold images during leg exercise whilst Holverda et al. [28] used non-breath hold imaging and Roest et al. [14] allowed the subjects to rest for the short period of image acquisition. Free-breathing may decrease image quality and rest during image acquisition will allow HR to decrease making interpretation of results more difficult.

In line with studies by Bevegård et al. [29] the present study showed that CO increased significantly due to increased HR whilst SV only increased by 8%. During the early stages of exercise in healthy subjects, the increase in HR is primarily caused by a decreased parasympathetic tone whereas the sympathetic effects are not seen until later stages [30]. As the exercise bouts of the study were short, the increased HR with only a small increase in SV is likely caused by parasympathetic withdrawal. Furthermore, the increased venous return caused by the supine position lead to maximal filling of the ventricles already at rest, which would explain the discrepancy between our study and exercise studies performed in the upright position. Our study would then be more representative of exercise in the supine position such as swimming or perhaps in micro gravitational environments such as space flight.

In contrast to a previous study of upright exercise on an ergometer cycle where the left ventricular valve displacement was significantly increased during exercise [20], our results showed unchanged LV AVPD and longitudinal contribution to LVSV. Right ventricular valve displacement (RV AVPD) remained unchanged but together with the decreased volume of the right ventricle, the right ventricular longitudinal contribution to SV was significantly decreased. Furthermore, total cardiac pumping became significantly more radial during exercise as shown by the increased THVV when exercising both with one and two legs, as well as during end-expiratory and end-inspiratory breath hold. This is in contrast to a hypothesis previously suggested by our group [22] where we expected cardiac longitudinal pumping to increase and radial pumping to decrease. Increased radial pumping as seen in the present study may theoretically increase the amount of energy spent on moving surrounding tissues and thus decrease the energy efficiency of the heart. However, for the left ventricle, Riordan and Kovács [31] showed that radial pumping may actually be important for diastolic suction during the rapid filling phase. Exercise requires rapid mass transfer from the atria to the ventricle, and it is possible that the increased radial pumping seen in the right ventricle may actually improve cardiac pumping efficiency due to enhanced diastolic suction.

It is possible that our findings of increased THVV only relates to exercise in the supine position such as swimming, and it would be of interest to perform similar studies during upright exercise.

Ventricular volumes and THV were reproducible between the first and second exercise session, and also when imaging was performed at end-inspiratory breath hold as well as during exercise with two legs. The differences seen in LVEDV between the first and second exercise session with one leg as well as between end-expiratory and end-inspiratory breath hold is probably best explained by individual variations that are more distinguishable in the small population. As shown in Figure 2 there is some variability between individuals for all variables and when only assessing six subjects results may fall out as statistically significant although not physiologically relevant.

Heart failure is a complex syndrome and diagnosis can be especially challenging at early stages. Cardiac MR during physical exercise may become useful for assessing patients with normal ejection fraction and suspected heart failure to investigate if cardiac function and filling are affected during low and medium intensity exercise. Furthermore, exercise CMR may also be used to asses patients with congenital heart disease such as Tetralogy of Fallot before and after surgery.

Exercise heart rate in our healthy volunteers only increased by approximately 40 bpm over resting HR and it is possible that a higher exercise HR may yield different results. The study population included to test for reproducibility of exercise measurements was small (n=6) and the results of the statistical tests of this subpopulation on reproducibility should be interpreted with caution. Furthermore, the study was performed in the supine position limiting the interpretation of our results to supine exercise such as swimming, but it may also be applicable for conditions of microgravity, such as space flight.