Exercise stress CMR reveals reduced aortic distensibility and impaired right-ventricular adaptation to exercise in patients with repaired tetralogy of Fallot

Background The aim of our study was to evaluate the feasibility of exercise cardiac magnetic resonance (CMR) in patients with repaired tetralogy of Fallot (RTOF) and to assess right and left ventricular adaptation and aortic wall response to exercise in comparison with volunteers. Methods 11 RTOF and 11 volunteers underwent prospective CMR at rest and during exercise. A supine bicycle ergometer was employed to reach twice the resting heart rate during continuous exercise, blood pressure and heart rate were recorded. Bi-ventricular parameters and aortic stiffness were assessed using accelerated cine sequences and flow-encoding CMR. A t-test was used to compare values between groups. A Mann Whitney test was used to compare values within groups. Results In RTOF both ventricles showed an impaired contractile reserve (RVEF rest 36.2±8.3%, +1.3±3.9% increase after exercise; LVEF rest 53.8±6.1%, +5.7±6.4% increase after exercise) compared to volunteers (RVEF rest 50.5±5.0%, +10.4±7.1% increase after exercise, p = 0.039; LVEF rest 61.9±3.1%, +12.2±4.7% increase after exercise, p = 0.014). RTOF showed a reduced distensibility of the ascending aorta during exercise compared to volunteers (RTOF: 3.4±1.9 10-3.mmHg-1 vs volunteers: 5.1±1.4 10-3.mmHg-1; p = 0.027). Ascending aorta distensibility was correlated to cardiac work in the volunteers but not in RTOF. Conclusion RTOF showed an impaired contractile reserve for both ventricles. The exercise unmasked a reduced distensibility of the ascending aorta in RTOF, which may be an early sign of increased aortic rigidity.

RTOF showed a reduced distensibility of the ascending aorta during exercise compared to volunteers (RTOF: 3.4±1.9 10-3.mmHg -1 vs volunteers: 5.1±1.4 10-3.mmHg -1 ; p = 0.027). Ascending aorta distensibility was correlated to cardiac work in the volunteers but not in RTOF. PLOS  Introduction criteria were the need to perform a CMR exam in the follow-up of the pathology, absence of medical contraindications to physical exercise and the ability to hold their breath for approximately twenty seconds. Exclusion criteria were presence of an implanted cardiac device, history of heart rhythm disorder or heart failure or contraindications for CMR. Eleven volunteers were included as a control group with the following inclusion criteria: normal electrocardiographic findings at rest, no history of cardiovascular disease, the ability to undergo stress exercise and no cardiac symptoms. Exclusion criteria were presence of cardiac malformations and contraindications for CMR (Table 1).

CMR protocol
All subjects underwent CMR on a 1.5T magnet (Siemens, Avanto, Erlangen, Germany) using a 16 channel anterior surface coil array and a 9-element posterior spine-coil array. The supine bicycle exercise stress and CMR imaging protocol were described and validated in detail previously [25]. All images at rest and during stress were acquired during short breath-holds. For ascending aortic analysis, steady state free precession cine images were acquired at the level with the pulmonary trunk. The imaging plane was positioned perpendicular to the ascending aorta (AAo) and the proximal descending aorta (DAo) so that both could be measured on the same image. The difference in angulation between AAO and DAo never exceeded 15˚. The following parameters were used: TR = 21.7ms, TE = 1.36ms, α = 65˚, FOV = 280×340, slice thickness = 7 mm, pixel size = 1.26×1.26mm, GRAPPA 2, 30 to 40 phases were reconstructed. For the ventricular parameters, a short-axis balanced steady state free precession cine sequence with TR/TE of 2.3/1.0ms and a GRAPPA acceleration factor of 3 was used for imaging cardiac function. Additionally, 2 and 4 chamber cine sequence were acquired at rest. Velocity encoding gradient pulse sequences were acquired to assess the regurgitant fraction of the pulmonary valve and were positioned perpendicular to the pulmonary artery. The following parameter were used: through plane direction, maximum encoding velocity 200 cm/s, TR = 28.35ms, TE = 2ms, α = 30˚, slice thickness = 6 mm, FOV = 192x280, 60 reconstructed phases. Data were reconstructed in magnitude and phase. If flow artefacts were found on the SSFP images level with the ascending aorta, aortic compliance was measured by using magnitude images reconstruction acquired on the same plane as the cine images [23].

Supine bicycle exercise in the CMR room
For CMR exercise, a non-magnetic ergometer (Lode B.V., Groningen, the Netherlands) was mounted on the patient table. The position of the ergometer foot pedals was adjusted to the height of the subject. The heart rate was measured by electrocardiography. The training mode of the ergometer allowed us to adjust the workload in Watts and was independent of pedal rotations per minute defining. Initial workload was 25 Watts. It was increased by 25W steps every 2 min, with a minimum duration of 10 min. The revolutions per minute ranged from 50 to 70 in order to ensure that work load was independent of revolution per minute. Once the targeted heart rate was reached (double of the heart rate at rest), the subject was instructed to stop cycling and the table was moved to the center of the magnet for imaging. Considering that the heart rate and blood pressure diminished quickly, only two cine datasets were acquired immediately after the cycling stopped. The exercise heart rate was recorded at the time of imaging during each image acquisition. The exercise heart rate was calculated as the mean heart rate value measured during the first cine and the second cine data set acquired during the same exercise stop. The delay between the end of exercise and image acquisition was 20 ±7s. Before and after each exercise stop, the arterial diastolic and systolic blood pressures were measured using a brachial automatic sphygmomanometer (Maglife, Schiller Medical). The exercise blood pressure reported in the result section was the mean systolic and the mean diastolic pressure calculated during an exercise stop. The subject was then again moved out of the magnet to resume cycling, starting at the last exercise level for at least two minutes. The stress process was repeated until completion of stress CMR (Fig 1). Cardiac work was evaluated using the rate pressure product (RPP = systolic blood pressure x heart rate) [26, 27].

Data post processing
Argus software (Siemens Healthineers, Erlangen, Germany) was used to measure all ventricular parameters, i.e. ejection fraction, end diastolic/systolic volume, stroke volume (SV), cardiac output (CO) at rest and during stress. The segmentation was performed using a semi-automatic method initiated by visually selecting the end diastolic and end systolic frames, and manually tracing the endocardial borders. All data were analyzed by consensus between a junior radiologist and a senior radiologist specialized in CMR with over 10 years of experience. Aortic compliance (AC) data were analyzed using the QIR scientific image analysis software (Université de Bourgogne, Dijon, France) [23]. The contours of the aorta were automatically traced for all phases of the cardiac cycle. For this processing [28] we used the balanced steady state free precession cine images because these images provided a high degree of contrast between the aortic wall and lumen with manageable artifacts due to flow. For two RTOF, these data were not available due to flow artifact within the aortic lumen and were successfully replaced by flash cine images. User interaction was limited to the indication of only one point near the center of the  aorta on the first time frame. The AC is the absolute change in aortic area (Δarea) for a given pressure step (AC = Δarea/ΔP). Aortic distensibility (AD) was defined as the compliance divided by the minimum cross-sectional area. AC and AD rely on cross-sectional areas.

Statistical analysis
Continuous data with a normal distribution are expressed as mean±SD and non-normal distributed data as median (range). Categorical data are expressed as frequency or percentage. All data were analyzed with Prism 5 (GraphPad Software, Inc, San Diego, CA, USA). A nonparametric Kruskal-Wallis test followed by a post-hoc Bonferroni test was performed to compare ventricular parameters, AC and AD between exercise and rest, for RTOF versus volunteers. A Mann Whitney test was used to compare all values at rest and exercise within groups. Correlation between AC, AD and biventricular parameters were checked with a Spearman test. A p-value below 0.05 was considered to indicate statistical significance.

Study population and exercise test
The characteristics of the cohorts are shown in Table 1. There was no significant difference in age, weight, height, BMI and body surface between groups. One patient was excluded before CMR because of arrhythmia. All RTOF and volunteers completed the protocol. All CMR images were of good quality in both the volunteers and the patients. All subjects included in the study complete the CMR protocol in 45±5 min, all patients and volunteers were able to exercise and double their heart rate during exercise. Also, the exercise doubled the rest RPP in both RTOF (97.8 ± 39.4%) and volunteers (108.3 ± 55.8%) ( Table 2). There was no statistical difference between the two groups' rest and exercise parameters (Resting RPP and heart rate were respectively: 7 731±1 568 mmHg.

Ventricular parameters at rest and during exercise
Rest LVEF was reduced for the RTOF (53.8±6.1%), but the increase in response to exercise was significant (59.6±10.4% p = 0.015). There were no significant changes in SV index between rest and exercise in either population (36.6±8.9 ml/m 2 to 39.1±12.3 ml/m 2 , p = 0.14 for RTOF and 40.4±8.1 ml/m 2 to 42.4±7.6 ml/m 2 p = 0.24 in volunteers) (Fig 2). However, CO increased significantly during exercise (2.5±0.5 l/min/m 2 to 4.1±1.1 l/min/m 2 , p = 0.001 for RTOF and 2.7±0.7 l/min/m 2 to 4.6±1.1 l/min/m 2 , p = 0.001 in the volunteers) (Fig 3). The ability to adapt ventricular parameters to exercise represented by the change between the stress and resting values was not significantly different between the two groups for LVEF, SV and CO, reflecting  LV adaptation to exercise in both groups. Resting LVEF was normal in volunteers and increase significantly in response to exercise (61.9±3% to 74.1±4.1%; p = 0.0038). The volunteers showed a greater capacity to adapt their LVEF than the RTOF, and this difference was significant (+12.2±4.7% versus +5.7±6.4% respectively, p = 0.014) (Fig 4).
Concerning the RV, the RTOF showed no increase in their ejection fraction between resting 36.2 ± 8.3% and exercise 37.2 ± 12.2% p = 0.48. The difference in RVEF between rest and exercise was higher for the volunteers compared with the RTOF (+ 10.4±7.1% vs + 1.3±3.9%; p = 0.039; respectively) (Figs 2 and 4). These parameters reflect an absence of contractile reserve of the RVEF in RTOF. As expected, the volunteers adapted their RV function and volume to stress (Table 3).
Cardiac work correlated significantly with right and left CO indexed for RTOF and volunteers. The correlation coefficient was lower for RTOF (right CO: R = 0.45 p = 0.044, left CO: R = 0.66, p < 0.001 for RTOF and right CO: R = 0.75 p < 0.001, left CO: R = 0.92, p < 0.001 for volunteers) reflecting a greater adaptation to exercise in volunteers compared with the RTOF group (Table 3).

Discussion
This study showed that an exercise CMR using an ergometer is feasible within a reasonable delay in patients with RTOF operated in childhood. The exercise protocol was sufficient to double the resting RPP. The exercise stress CMR revealed a lack of contractile reserve of the RV in the RTOF. The study also showed a decrease in AC and AD during exercise in both groups but with no correlation between AD and RPP in RTOF. The exercise may therefore be a marker indicating a structural modification of the aortic wall. CMR is the reference method to quantify RV volumes and allow to assess cardiovascular adaptation to exercise in RTOF. More study will be needed to assess the impact of such parameters on patient selection for pulmonary valve surgery.

Exercise stress CMR
Exercise CMR can also be performed using a treadmill in the examination room next to the scanner. In this setting the patient exercised first and is moved to the examination table immediately after completion of the exercise phase [29]. However, with this protocol, a lot of time The resting RPP in patients and in volunteers was at 7 731 mmHg.min -1 and 8 321 mmHg. min -1 respectively within the standards. However, the RPP values during the exercise were at 15 412 mmHg.min -1 and 16 800 mmHg.min -1 in patients and controls respectively, this was in line with our objective to explore adaptation to exercise and not to test the patient's peak exercise capacity.
It accepted that CMR can be more reproducible in the analysis of the left ventricle on exertion than echocardiography. Also, an exercise stress CMR took more or less the same time to perform as exercise echocardiography [36]. Our study corroborated exercise CMR's potential capacity to assess both ventricles and the aortic arch within a similar time frame. CMR also provided good quality images in all volunteers as well as in patients.

Ventricular parameters
Baseline RVEDVi and RVESVi in the RTOF group were severely increased indicating that patients showed severe RV remodeling. In that group during exercise no increase in RVEDVi nor RVESVi was measured indicating a lack of contractile reserve in the right ventricle. The increase in cardiac output during exercise was only due to the heart rate increase. In the RTOF subjects LV parameters were preserved at rest but during exercise the LV showed a lower decrease of LVEDVi compared to the volunteers suggesting that LV contractile reserve is also impacted by the severe RV impairment. Adaptation of the RV to exercise in RTOF had not yet been extensively explored. However, a loss of adaptation of the RVEF is well documented [37] and has identified the pulmonary infundibulum, in particular in the anterior and lateral portion, to be the anatomical region which contracts the least. The study of Wald et al. Using CMR reported contrast uptake in these areas using CMR, suggesting a fibrous evolution of the myocardium [38]. AD = aortic distensibility. AAo = ascending aorta. DAO = descending aorta. All measurements are indexed. In RTOF at rest and on exertion a significant correlation was found between cardiac work and both biventricular cardiac output. This correlation is less stronger than these of volunteers. This clearly demonstrate a lack of adaptation of AD AAo with RPP. https://doi.org/10.1371/journal.pone.0208749.g006 In volunteers, exercise modification was in accordance with the literature reporting cardiac elongation force [39][40][41], namely a decrease in end diastolic volume and end systolic volume and an increase in the ejection fraction greater than 5%. The fact that the right and left SV did not increase significantly during exercise was consistent with the work of Belever et al. on adaptations of the heart during exercise in the supine position [42]. However, Bergovec et al. showed a significant increase in end diastolic volume and SV of the LV during exercise by radionuclide ventriculography [43]. More recent work by Stöhr et al. [44] confirmed that the SV increases early, at the beginning of the exercise, before reaching a plateau, after which the increase of CO is only related to the higher heart rate.

Why exercise stress CMR could be interesting for RTOF
Early childhood surgery is essential for patient survival [4] but pulmonary regurgitation and stenosis are the major complication during follow up. Pulmonary regurgitation is often associated with other myocardial damage factors such as death, sustained ventricular tachycardia and increase in NYHA class. This is the leading cause of adverse events [2,45,46]. Revalvulation is the method of choice for reducing the symptoms and morbidity associated with RV damage secondary to pulmonary regurgitation, and it has been shown to be effective on the ventricular parameters [47]. This procedure can be performed either by endovascular or conventional surgery [48]. Some asymptomatic RTOF profiles have been identified as good responders. These patients would indeed not require further treatment as they are considered as definitively cured: Patients who are over 35 years old never having needed pulmonary valvulation with a normal exercise capacity and a pulmonary valve diameter greater than 0.5 multiplied by z score [49]. However, currently, we do not know the optimal timing for revalvulation. There is a consensus on the indication for valve replacement in symptomatic patients, but at this point it may already be too late and ventricular dilatation could be fixed [50][51][52][53][54]. To date, pulmonary revalvulation is based on morphological parameters such as RVEDV but these criteria are still a matter of debate [55][56][57][58].

Aortic rigidity
In our study we showed that exercise unmasks a non-adaptation between AD and cardiac work in RTOF. That could be explained by the following hypothesis: surgery in childhood, aortic dextroposition or chronic ventricular damage may lead to secondary aortic wall damage. AD has been shown to be an early marker of damage to the smooth muscle fibers [59]. Measured lower adaptation to exercise might be an early sign of aortic rigidity increase and might corroborate with the higher incidence of aortic aneurysm and aortic wall histologic abnormalities in this population [60].
Therefore, the RTOF subjects probably fit into the framework of a more general cardiac condition as shown by the high number of cardiovascular events unrelated to RV damage [2,45,46]. Aortic rigidity is an important marker of risk for cardiovascular events, and increases faster with high mean blood pressure and a history of tobacco use [61,62]. Aortic rigidity can cause changes in the geometry of the aortic arch and LV remodeling [63,64] observed in the RTOF [65]. In comparison with the literature, Rutz et al. [66] found a lower AD at rest on AAo for RTOF compared to volunteers, suggesting an increased aortic rigidity in RTOF.
Our main limitation was a small number of subjects, a total of 22. However, Fallot's tetralogy remains a rare congenital heart disease, with an incidence of 1/3000 birth [14].
Another limitation was the fact that the image acquisitions were not carried out during exercise but less than a minute after pedaling stopped and with a short breath hold. We could not do better because the diameter of the magnet was too small to permit pedaling in the magnet. To compensate for this, exercise stress was monitored by the RPP calculation which showed a significant increase between rest and exercise and an average increase that was almost doubled in both the controls and the RTOF group, despite a broad dispersion. Selecting a control group using a paired matching procedure based on age and sex might have been helpful. VO2 consumption was not assessed and to the best of our knowledge there is no CE labeled device to do so in a CMR environment.
In conclusion, we have demonstrated the feasibility of performing an exercise test in an RTOF population under CMR condition within a reasonable time frame. With a dedicated protocol and a trained team this examination could be used in practice. In RTOF both ventricles were damaged by a lower contractile reserve. The ascending aorta showed early signs of increased rigidity during exercise, but not at rest.
Exercise CMR might be a promising tool for ventricular and aortic monitoring of RTOF.