Exercise-induced myocardial T1 increase and right ventricular dysfunction in recreational cyclists: a CMR study

This was a prospective observational cohort study approved by the medical ethical committees (Ethische toetsingscommissie Jessa Hospital, Hasselt, Belgium, Comité voor Medische Ethiek, UHasselt, Hasselt, Belgium and Comité de protection des personnes Sud-Ouest et Outre-mer II, Toulouse, France) and all participants gave their written informed consent. Recreational male cyclists performing the EDT 2021 (175 km, 3600 altimeters) ride were eligible for inclusion in this study. Before the race, each participant completed a questionnaire about their medical and training history (hours and kilometers in the last six months). No cyclist participated in any significant endurance event in the preceding week of the race or had regular training sessions within 72 h prior to rest CMR and the EDT 2021 ride. Exclusion criteria for admission were the intake of anti-inflammatory drugs and any systemic or cardiovascular disease.

According to the UCI guidelines, all participants were tested by a polymerase chain reaction (PCR) test within 72 h before the rest examinations to exclude the presence of SARS-CoV-2 RNA.

Following cardiac preparticipation evaluation by an experienced cardiologist, all participants performed an exercise test approximately one week before the EDT stage to evaluate their physical fitness. During the exercise test, an electronically braked cycle ergometer (Cylus2^®, General Electric GmbH, Bitz, Germany) with pulmonary gas exchange analysis (Metalyzer II^® 3B Cortex, Leipzig, Germany) was used. The participants started at 100 W during the first 3 min. Hereafter, the workload was increased by 40 W each 3 min and the test was performed until voluntary exhaustion. Participants were asked to cycle at least 70 rpm throughout the test. Oxygen uptake (VO₂), expiratory volume (VE), and respiratory exchange ratio (RER) were collected breath-by-breath and averaged every minute. At the end of the test, RER values (> 1.1) were evaluated to verify whether the test was performed at full capacity.

Maximal oxygen uptake (VO₂max) was the primary outcome of exercise capacity. In addition, maximal cycling resistance (W_max), maximal heart rate (HR_max), and test duration, defined as the corresponding load, heart rate, and minutes at the level of exhaustion, were reported. Recovery heart rate (HR_recovery) was noted following two minutes of rest after cessation of the exercise test. Capillary blood samples were obtained from the earlobe to analyze blood lactate concentrations (mmol/l). Blood lactate was measured every three minutes, at maximal exhaustion and after recovery (lactate peak) using a portable lactate analyzer (Accutrend Plus, Roche Diagnostics Limited, Sussex, UK).

Heart rate zones were determined using blood lactate concentrations of 2 and 4 mmol/l in combination with RERs of 0.92 and 1, respectively.

Each participant’s heart rate (HR) zones were individually assigned to the cycling computer device based on the maximal HR, and HR at the aerobic and anaerobic gas exchange thresholds during the exercise test. An HR monitor connected by Bluetooth to the cycling computer was used to ensure the correct registration of HR intensity over time and the ride duration. All participants were asked to cycle 2 h around their anaerobic gas exchange threshold (in HR zone 4) to induce a sufficiently intense exercise stimulus. The HR zones based on the individual exercise testing results gave a more correct and comparable measurement of the exercise intensity of each participant. The race activity files containing HR data and the ride duration were collected after the race.

HR data were analyzed for mean and maximal HR values and the intensity profile of each cyclist, including the percentage of total race time spent in HR zone 4. This approach was based on prior studies demonstrating a stepwise cTn increase between mean HR of 140 and 160 bpm and the duration of elevated HR as a predictor of exercise-induced cTn elevation (Stewart et al. 2016; Bjørkavoll-Bergseth et al. 2020).

CMR imaging (3 T Discovery MR 750w, GE Healthcare, Waukesha, WI) was performed one day before and 3–6 h after the race at the same time of the blood sample. Steady-state free precision (SSFP) cine, native T1 and T2 parametric mapping images were obtained. The time point post-exercise was chosen based on the highest troponin release post-exercise allowing enough time for myocardial edema to develop and be detectable, and representing the liberation of enzymes from damaged myocytes (Baker et al. 2019). In addition, a third CMR examination was repeated in a randomly selected subgroup of 8 cyclists approximately 18–20 h post-exercise to document the evolution over time and to confirm the presence of possible subclinical myocardial edema. The imaging parameters are reported in the ‘supplementary material 1’ file.

Native T1 and T2 maps were calculated offline pixel-by-pixel from a set of images with different T1- and T2-weighting by two observers (BP and BV) experienced in CMR using in-house Mevislab tool (Vs 2.5.1 MeVis Medical Solutions AG, Bremen, Germany) and commercially available software package (suiteHEART^®, Version 4.0.6, Neosoft, Pewaukee, Wisconsin, USA). Both readers were blinded to the scan (rest and post-exercise CMR), and clinical and biological data. Maps and error maps were calculated applying a validated retrospective motion correction method of the T1-weighted images (Tilborghs et al. 2019). A freehand region of interest (ROI) was drawn in the septum for global assessment using a fixed color code and range (lut royal; 650–1650 ms (ms) for T1 mapping, 0–80 ms for T2 mapping) and error maps to avoid evident susceptibility artifacts or adjacent blood pool and extra-myocardial tissue pixels (supplemental Fig. 1) (Schulz-Menger et al. 2020).

Cardiac ventricular volumes, function and LV mass were quantified at rest and post-exercise in consensus by two observers using analysis software (suiteHEART^®, Version 4.0.6, Neosoft, Pewaukee, Wisconsin, USA), according to standardized recommendations (Schulz-Menger et al. 2020). The software automatically defined the endo- and epicardial LV, and endocardial RV contours and was manually adjusted if the tracking was suboptimal. Trabeculae and papillary muscles were included in the LV cavity.

Two-dimensional transthoracic echocardiography (TTE) was performed using an ultrasound system TUS-AI 900/5L (Aplio i900, Software Version 6.5 Canon medical Systems, Otawara, Japan) with 2D phased array (PST-28BT, Canon medical Systems, Otawara, Japan) transducer. Atrial volumes, and left and right atrial and ventricular deformations were assessed off-line using commercially available software (UltraExtend NX, Software Version 1.0, Canon Medical Systems, Otawara, Japan).

All blood samples were collected one day before and at 3–6 h after the stage at the same time of the CMR examination, and analyzed by the Clinical Laboratory, University Hospital Pasteur 2, Nice, France. High-sensitivity cardiac Troponin I levels (cTnI) levels were determined using the cTnI-Ultra assay for the ADVIA Centaur CP (Siemens Healthineers, Erlangen, Germany) with an analytic limit of detection of 6 ng/L. The claimed 10% CV was 30 ng/L was with an upper reference limit of 40 ng/L. C-reactive Protein (CRP) and creatin kinase (CK) concentrations were analyzed using a Roche-Cobas system (Roche Diagnostics GmbH, Mannheim, Germany) with a normal range up to 5 mg/L and < 190 U/L respectively, according to the information provided by the manufacturer.

All statistical analyses were performed by IBM SPSS^® version 27.0 (IBM SPSS Statistics for Windows, Chicago, IL, USA). A Shapiro–Wilk test was used to test the normality of the data (p < 0.05). All continuous data characteristics were presented as mean ± standard deviation or absolute frequency. Comparisons between baseline and post-exercise values were tested using a paired two-sided Student t-test for normally distributed data and a Wilcoxon signed rank test for abnormally distributed data. The Friedman test was used to measure differences in T1 and T2 relaxation times between baseline, 3–6 h post-exercise, and 18–20 h post-exercise within a randomly selected sub-sample of 8 cyclists. Pairwise analyses (Dunn’s post-hoc comparison test) were performed when the Friedman test was statistically significant. A power calculation of our paired t-test was performed based on the standard deviation of a previous study (Tahir et al. 2020) using a significance level of 5% and a sample size of 16, which resulted in a power of 93% for T1 and 98% for T2 relaxation times.

Interobserver reproducibility for T1/T2 mapping measurements was assessed using intraclass correlation coefficient (ICC) estimates and Bland–Altman analysis.

Two participants recovered from COVID-19 infection more than one year before the EDT ride without any residual clinical symptoms. At the time of the EDT race, eleven participants were completely vaccinated, seven received one of two doses, and the last received an Ad26.COV2. S vaccine. None of the participants was vaccinated within the week before the EDT stage. All participants were in good health and active COVID-19 infection was excluded by a negative PCR test within 72 before the CMR at rest.

Eighteen cyclists successfully finished the race in 537 ± 58 min (range 453–635 min).

Battery problems (n = 2) and longer periods of missing data (n = 2) during the recording of the race activity file by the cycling computer device resulted in an evaluation of cycling intensity in 14 riders. The mean and maximal HR of the 14 cyclists were 131 ± 9 bpm and 167 ± 9 bpm, respectively. Cycling time in HR zone 4 (147 ± 59 min) was highly variable, with minimal and maximal times of 64 and 262 min, representing 29 ± 13% (range 10–47%) of total cycling time.

The cTnI values at rest were within normal ranges for all participants (1.9 ± 5.1 [0–18.8] ng/l). One cyclist had a CRP level of 9.5 ml/I, without clinical symptoms. cTnI level 3–6 h post-exercise was elevated above the cut-off level for acute myocardial infarction (> 30 ng/L) in 16 of the 18 cyclists (p < 0.001) (Fig. 1). Post-race CRP was significantly increased in five cyclists, including the two cyclists without significant cTnI elevation and three cyclists who had fallen during the race.

Post-race cTnI levels were elevated in 16 (89%) of our cyclists, similar to a meta-analysis evaluating biomarker changes after a strenuous endurance exercise (Sedaghat-Hamedani et al. 2015). Moreover, cTnI release was strongly correlated with post-race myocardial T1 relaxation time increase, as a potential marker of subclinical myocardial edema.

In contrast to previous CMR studies, we observed exercise-induced myocardial relaxation T1 increase, which can be explained by several reasons. First, participants performed a much longer (total race time 537 ± 58 versus 198 ± 162 min) and more intensive exercise session (cTnI 70 ± 42 versus 57 ± 86 ng/l) than those from Tahir et al. (2020). Second, the timing of cardiac imaging post-race may have affected the results. The time interval to post-race CMR was longer (306 ± 63 versus 144 ± 162 min) compared to a previous study using parametric mapping in triathletes (Tahir et al. 2020), and based on the time point of the highest cTnI release after the race as extensively documented (Shave et al. 2010; Aengevaeren et al. 2021a, b). Third, pixel-wise parametric mapping was used. This allowed the application of the recently revised 2018 Lake Louise criteria II (Ferreira et al. 2018) to detect myocardial edema with much high accuracy than the original Lake Louise criteria I with T2-weighted CMR imaging used in previous studies (Mousavi et al. 2009; O'Hanlon et al. 2010). Both T1 and T2 mapping can detect subclinical forms of myocarditis. Similar to previous studies, we observed no post-exercise T2 myocardial changes in our cyclists (Aengevaeren et al. 2020). There are several potential explications for this: (1) in contrast to T1 mapping, T2 mapping might be less sensitive to detect a subclinical level of myocardial edema post-exercise; (2) physiological confounders of myocardial T2 mapping, including regional and intersegmental variations and sensitivity to field inhomogeneities and magnetization transfer effects (Wiesmueller et al. 2020); (3) a dehydration status is associated with lower T1/T2 myocardial relaxation times (Luetkens et al. 2020). Although the post-race dehydration of the cyclists was not measured, a more hydrated status compared to rest CMR is almost impossible after such a strenuous and prolonged exercise. A higher HR was associated with lower T₂ values in a previous study (von Knobelsdorff-Brenkenhoff et al. 2013), and our study did not show a correlation between post-exercise HR and T1 increase. Although a pathological T1 and T2 mapping will increase specificity for diagnosing acute myocardial edema, having only a T1 mapping-based marker still support a diagnosis of acute myocardial edema with an area under the curve (AUC) of 89% (Ferreira et al. 2018). A recent meta-analysis to detect acute myocarditis confirmed a higher AUC of native T1 mapping compared to T2 mapping (0.95 versus 0.88) (Kotanidis et al. 2018). The mean T1 relaxation times at rest (1133 ± 48 ms) in our study were concord with the 95% tolerance interval of previous 3 T studies in this age category (von Knobelsdorff-Brenkenhoff et al. 2013; Roy et al. 2017).

The association between cardiac troponin levels and acute myocardial necrosis after endurance sports remains speculative. Post-race cardiac troponin elevations usually return to normal within 24 h suggesting no associated myocardial necrosis. A third CMR scan 18–20 h post-EDT race showed normalization of myocardial T1 relaxation time in a subgroup of 8 cyclists suggesting that just as reported for exercise-induced cardiac troponin changes, these alterations are transient (Aengevaeren et al. 2021a, b). An “increased cardiomyocyte membrane permeability” has been proposed as the major cause of post-exercise cTn release (Sedaghat-Hamedani et al. 2015). However, although there is no direct evidence of post-exercise cytotoxic myocardial oedema, a small degree of myocardial necrosis cannot be completely excluded because of post-exercise cTn elevations (Aengevaeren et al. 2021a, b).

The use of multiple imaging modalities in this study strengthened our findings. CMR showed reduced post-race RV function and increased volumes, while echocardiography demonstrated more pronounced post-race strain reductions of the right heart compared to the left heart. LV was less affected post-race with no significant post-race changes in LV function and cardiac volumes, as previously reported on TTE (La Gerche et al. 2012; Stewart et al. 2017). Exercise-induced RV functional limitations were more profound than LV, likely due to disproportionate RV wall stress during exercise (La Gerche et al. 2015). In contrast to other CMR studies (O'Hanlon et al. 2010; Aengevaeren et al. 2020; Tahir et al. 2020), the post-exercise RV impact on our cyclists can be explained by the higher intensity and longer duration of the exercise activity.

High intensity exercise in a prolonged duration cycling activity resulted in more post-race cTn release in our participants, as previously reported (Kleiven et al. 2019; Bjørkavoll-Bergseth et al. 2020; Aengevaeren et al. 2021a, b), while exercise-induced T1 myocardial modifications on CMR were transient, similar to cardiac troponin changes (Aengevaeren et al. 2021a, b). Myocardial inflammation and necrosis following strenuous exercise were worsened during viral infection in animal studies (Ilbäck et al. 1989). Therefore, our study adds to the body of evidence that it is important to avoid high-intensity exercise during (viral) infections. Exercise-induced subclinical edema may be an underlying mechanism of acute post-race RV dysfunction in recreational cyclists. Although the thin RV free wall prevents measurements of subclinical oedema with current parametric mapping techniques due to insufficient spatial resolution (Messroghli et al. 2017), it is a likely that oedema is also present in the RV wall given its expected diffuse nature and the greater hemodynamic impact of exercise on the RV. Biventricular interaction may be an additional mechanism to explain the observed reduction of RV function (La Gerche and Claessen 2019). A further prospective investigation is required to evaluate the recovery time for acute myocardial injury (ea. cTn-related T1 myocardial modifications and RV dysfunction) following (repeated) high-intensity cycling (De Bosscher et al. 2022; Eijsvogels Thijs and Aengevaeren Vincent 2022). An important shortcoming of previous studies is that only a single post-exercise assessment of associations between biochemical, functional, and structural myocardial changes has been evaluated. Detailed information about the recovery time for transient myocardial injury is lacking. Further imaging studies should address more precisely post-exercise myocardial T1/T2 kinetics and functional modifications, and their association with post-race cTn release and with exercise intensity and duration.

There are limitations to our study, including the relatively small sample size (n = 19), like many studies of this type. Also, due to above-mentioned technical problems a limited number of 14 files were available for HR zone measurements. The measurement of cycling time in HR zone 4 as a parameter of high intensity, was based on commercially available cycling computers and HR monitor devices, similar to previous studies (Bjørkavoll-Bergseth et al. 2020). The relationship between T1/T2 mapping and exercise parameters was not fully evaluated in this study due to undersampling of our study population and to the high variability of race intensity and duration. No late gadolinium enhancement and extracellular volume calculation were performed to exclude focal myocardial fibrosis. However, this was not the objective of our study and abnormal native T1/T2 mapping provide evidence for acute myocardial edema according to the updated 2018 Lake Louis Criteria (Ferreira et al. 2018). Hydration status was not measured in this study. Its effect on T1/T2 relaxation times could therefore not be assessed, although post-exercise dehydration might have contributed to reduced myocardial relaxations times (Luetkens et al. 2020). Finally, myocardial diffusion-weighted imaging to evaluate myocardial edema was not available on the MR system. Apparent diffusion coefficient maps showed improvement in the depiction of myocardial edema compared to T1/T2 mapping in a recent study (Moulin et al. 2020).

In conclusion, strenuous exercise causes cTnI increase, prolongation of myocardial T1 relaxation time and RV dysfunction in recreational cyclists, which showed interdependent correlation. Exercise-induced myocardial T1 increase may represent subclinical myocardial edema. More research is needed to determine the clinical significance (normal physiological changes or precursors to long-term remodeling) of these acute functional and structural myocardial changes following repeated high-intensity cycling.