Prevalence and pattern of cardiovascular magnetic resonance late gadolinium enhancement in highly trained endurance athletes

We recruited 93 triathletes from local triathlon clubs by invitation to participate in our study, between March 2015 and December 2017. Inclusion criteria were: (a) highly-trained endurance athletes (20–45 years) who had trained for a minimum of 12 h per week in the previous 5 years, (b) no cardiac symptoms or cardiovascular disease risk factors and (c) no family history of sudden cardiac death. The control group comprised 72 age- and gender-matched individuals. They were eligible if they only performed recreational sports of mild-moderate intensity for less than 3 h per week, and if they had no known history of cardiovascular disease.

The subjects were evaluated through the following tests: detailed anamnesis, physical examination, blood test and 12-lead surface electrocardiograms (ECG), and also a maximal cardiopulmonary exercise test in an upright cycloergometer (Ergoselected 100, Ergoline, Bitz, Germany) following an incremental ramp protocol. Study exclusion criteria were the presence of heart disease, renal failure (creatinine > 1,5 mg/dL), claustrophobia or any other contraindication for CMR. Lifetime training history and current training load were registered by direct reporting of the athletes. The study protocol complied with the declaration of Helsinki, institutional review board and ethics committee approval were obtained for this study. All subjects provided written informed consent.

All athletes underwent CMR imaging with either a 3 T (n = 65) or a 1.5 T (n = 28) CMR scanner (Magnetom Trio or MagnetomAera, respectively; Siemens Healthineers, Erlangen, Germany), using a dedicated cardiac coil and ECG gating. Control subjects were scanned with the same 3 T scanner. After standardized CMR planning [26], conventional balanced steady-state free-precession cine imaging in a short-axis (8 mm slice thickness, 2 mm gap), covering the left ventricle (LV) and right ventricle (RV) from above the atrioventricular groove to the apex, in 2- and 4- chamber view were acquired [27]. In addition, after 10–15 min of 0.15 mmol/Kg gadobutrol (Gadovist®, Bayer Healthcare, Berlin, Germany) intravenous bolus injection, LGE images were obtained with standard phase-sensitive inversion recovery sequence matching cine images. Finally, in a subgroup of 28 athletes, whose studies were performed in the 1.5 T scanner, T1-mapping was also added using a modified look-locker inversion recovery sequence (MOLLI 5–3-3) in short axis (basal, mid ventricular and apical) and 4-chamber view. They were acquired at baseline and 15 min after a bolus of gadobutrol to quantify the ECV. Blood samples were drawn at the same moment of the CMR study and the hematocrit was determined. The acquired images were digitally stored in DICOM format.

CMR images analysis were performed with specialized software (Argus, Siemens Medical Solutions, Germany) and were blindly evaluated by two experienced investigators, who had more than twelve years of CMR clinical expertise. Cardiac function quantification was performed using the summation of discs method; LV endo- and epicardial borders and RV endocardial borders were manually traced to quantify volumes at end-diastole (EDV) and end-systole (ESV), as well as the LV mass. Data were normalized to the subject’s body surface area (BSA). LGE was identified in contrast sequences and was deemed significant if the LGE pattern and extent was visually identified by both clinicians at least in two consecutive images. Native T1 and post-contrast T1 values were measured in regions of interest drawn in the LV septum (midventricular) and in the LV cavity (blood pool) to quantify ECV [ECV = (1 - Hematocrit) x (ΔR1_myocardium/ΔR1_blood)] [28]. Areas of focal LGE were excluded in order to evaluate these parameters unbiased from the presence of LGE.

Characteristics of the male and female triathletes and control subjects are shown in Table 1. All four groups presented a similar distribution of age. Female athletes had lower BSA than male athletes, and BSA was also lower in athletes as compared to controls (p < 0.05). As expected, athletes showed a lower heart rate and higher maximal oxygen consumption (VO_2max) as compared with the control group. All controls had a normal ECG and athletes presented with exercise-induced adaptive ECG changes such as incomplete right bundle branch block (male vs female: 20.4% vs 18.2%; p = 0.78) and mild or moderate sinus bradycardia (male vs female: 61% vs 64%; p = 0.81). No ECG pathological abnormalities were found in controls or in the athlete’s population.

Table 2 depicts cardiac dimensions and function as well as the results of contrast CMR in the different groups. Athletes showed a slight reduction of biventricular ejection fraction and larger biventricular volumes, particularly in the RV, resulting in a higher RV EDV-to-LV EDV ratio in athletes as compared to controls subjects (0.97 ± 0.1 vs 0.89 ± 0.1; p < 0.001). Male athletes presented larger biventricular and right atrial cavities as compared to female athletes (p < 0.05). LV mass and bi-atrial size were also larger in athletes as compared to controls.

The presence of LGE was significantly more prevalent in athletes than in controls: 35 of 93 athletes (37.6%) showed LGE [17 of 49 males (34.7%) and 18 of 44 females (40.9%)], while only two of 72 (2.8%) controls had evidence of LGE. Thus, the presence of LGE was markedly increased in highly trained endurance athletes (in both genders) as compared to age- and gender-matched control subjects. All LGE⁺ triathletes had a similar pattern consisting of a small volume of focal LGE confined to the inferior interventricular septum, commonly where the RV attaches to the septum – insertion point or hinge point (Figs. 1 and 2). Other specific LGE patterns, such as subendocardial, transmural or subepicardial, were not identified in our population of healthy asymptomatic athletes.

Table 3 depicts population characteristics and CMR parameters for athletes with and without LGE. Both groups were of similar age, height, weight or BSA and had a similar training load.

The exercise-induced increase in systolic blood pressure was equivalent in LGE+ and LGE- athletes. Among men, their competition history revealed that those athletes with LGE⁺ had a trend toward more hours of training per week (14.9 ± 2.6 vs 13.3 ± 2.3; p = 0.06). Nevertheless, LGE⁺ athletes (both genders) showed no differences in LV and RV volumes and ejection fraction as compared to LGE⁻ athletes.

In a subgroup of 28 athletes (71% female), that were studied in the 1.5 T scanner, T1 mapping sequences were also included in the contrast CMR protocol. Globally, ECV mean values were 26.0 ± 2.3%, and within the range of previously reported normal intervals (ECV 25.3 ± 3.5%) [29, 30].

Additionally, 13 of those 28 athletes (46.4%) had LGE confined to the interventricular septum. Those LGE+ athletes, had higher ECV at remote LV myocardium (areas where focal LGE was not identified) as compared to LGE- athletes (27.1% ± 2.2 vs 25.2% ± 2.1; p < 0.05), despite still within the reference limits of normality. These data are shown in Fig. 3. ECV above the reference limits of normality were only found in 2 of 28 athletes and they were not associated with gender (p = 0.82), training load (p = 0.59), or with structural (considering LV EDV; p = 0.32) or functional (considering left ventricular ejection fraction (LVEF); p = 0.11) ventricular changes. Concordance was high between the two observers for ECV values (intraclass correlation coefficient -absolute agreement- > 0.85).

As expected, our results are in line with previous data proving that long-term endurance training promotes cardiac remodelling that consists of significant biventricular and biatrial dilatation in both genders and a slight reduction of biventricular ejection fraction (at rest). These changes are part of a common adaptation in trained athletes, known as the athlete’s heart [2, 3, 31, 32]. A mild decrease in resting LVEF and right ventricular ejection fraction (RVEF) was generally associated with more significant LV and RV enlargement; this is typically considered as adaptive since dilated ventricles have increased volume- and contractile reserve [31, 33].

Our study provides additional evidence that highly trained endurance athletes show a significantly higher prevalence of focal LGE than age- and gender-matched control subjects [4, 11, 23]. The prevalence of LGE, as a potential marker of focal fibrosis, was up to 37.6% in our group of athletes, ten-fold higher than that observed in the control group. This prevalence was significantly higher than reported in other series of athletes [11, 24, 34]. Even though Tahir et al. [34] did not find focal LGE in female athletes, we found a similar prevalence between male and female athletes (p > 0.05), as also previously described in other reports [24]; this could be due to the homogeneity of our cohort of athletes. Additionally, we did not find differences regarding training load between female and male athletes, unlike Tahir et al. [34], who noted that their female athletes reported less training load.

In our study, all athletes with LGE showed a similar pattern, consisting of a small volume of focal LGE confined to the interventricular septum, commonly where the RV attaches to the septum (the inferior insertion point or hinge point). This pattern corresponds to a typical non-ischemic substrate. Other specific LGE patterns, such as subendocardial, transmural or subepicardial, were not identified in our population.

This observation may be because our population was constituted by completely asymptomatic, otherwise healthy, young athletes, while in previous studies the athlete’s cohort was older [11, 12, 23, 34, 35] or had abnormal findings on their regular screening check-ups [36]. Breuckman and colleagues [11] studied a cohort of 102 veteran athletes and found that 12% had LGE, of which almost half showed a coronary artery disease pattern. Bohm et al. [12] in 33 endurance athletes found 3 of them having subepicardial LGE, which is typical for old myocarditis/pericarditis. Wilson and colleagues [23] in a small cohort of 12 veteran athletes found that 42% exhibited a non-coronary LGE pattern. They also included 17 young athletes in which no LGE was found. Likewise, Merghani et al. [35] included 152 master athletes, of which 14% revealed LGE. Nevertheless, they did not find relationship between myocardial fibrosis and exercise intensity, years of training, or number of competitions. Only in the study of Schnell et al. [36], athlete’s population was younger than ours. This study included 7 asymptomatic athletes who all showed LGE; including four with pathological T-wave inversions and two with ventricular arrhythmias on a screening exercise test. To our understanding, ours is the largest cohort of young, healthy and asymptomatic athletes published, and we assume that, unlike other studies, we probably did not find clearly pathological LGE patterns since our athletes were strictly recruited by invitation to participate in a research project, and not because they presented any type of clinical or electrical abnormality.

Regarding the presence of focal fibrosis in the insertion point, La Gerche et al. [32] has suggested that the RV may remodel slightly more than the LV in endurance athletes, since RV wall stress increases more than LV wall stress during exercise and this places an additional pressure load on the RV. Focal LGE in the RV inferior interventricular septum has been related to RV pressure overload and systolic pulmonary hypertension [37, 38]. Specifically, the presence of focal LGE in the RV inferior interventricular septum, as well as the extent of hyperenhancement, has been inversely related to measures of RV systolic function in patients with severe symptomatic pulmonary artery hypertension [37]. However, the clinical significance and the real impact on outcome of the presence of small regions of LGE in the RV insertion point, in asymptomatic subjects, are still uncertain and these aforementioned prognostic implications should not be extrapolated.

LGE⁺ athletes showed no significant differences in LV and RV volumes or function as compared to LGE⁻ athletes. However, other authors [11, 23, 24, 34] have reported that those athletes with LGE had been competing in endurance sports for longer and had greater RV EDV and lower RVEF. Additionally, our athletes did not show a hypertensive response to exercise and, unlike Tahir et al. [34], we did not find significant differences regarding the peak exercise systolic blood pressure between LGE+ and LGE- athletes.

Our data demonstrates balanced biventricular and biatrial dilatation with a slight reduction of biventricular ejection fraction in athletes with and without LGE; this may arise, given the inclusion criteria, from the fact that our athletes practise the same endurance sport discipline and have similar training load. Thus, all this data potentially suggests that this LGE pattern might be another feature of the athlete’s heart, related to local mechanical stress due to the exercise-induced cardiac overload. While LGE has been correlated to focal fibrosis in chronic infarction, LGE dynamics are still controversial and not fully understood in other settings. The presence of LGE indicates that the local matrix and fibre structure has changed, similarly to what happens indeed in histologic fibrosis and scar; however, LGE could also be present due to localized structural remodelling regardless of the presence of real fibrosis. Experimental models of endurance training might shed some light in the real clinical significance of the observed LGE in athletes.

Considering that the increased LV mass, observed in athletes, is due to an expansion of the cellular compartment, a decreased ECV is to be expected [39]. In fact, McDiarmid et al. presented data showing that increasing degrees of training load linearly increase myocyte hypertrophy and inversely ECV in the athletic group in a cohort of 30 athletes (athletes vs controls; 22.5% ± 2.6 versus 24.5% ± 2.2; p = 0.02) [39].

Nevertheless, in our study we did not find a decrease in ECV in athletes. And when we separately analysed those athletes with focal LGE confined to the inferior insertion point, we observed that they had slightly higher ECV at remote LV myocardium than those without LGE, despite still being within the limits of normality [29, 30]. These data are in line with previous published literature [34] suggesting that the potential myocardial fibrosis might involve the entire myocardium of athletes, with focal LGE, and would not be confined only to the LGE areas. There might be other explanations to the increase of ECV; Coelho-Filho and colleagues [40] worked with hypertensive patients with LV mass within the gender-specific normal range, and found that their ECV was significantly higher than in controls. They suggested that expansion of the ECV could precede significant increase of LV mass. Nevertheless, our patients did not have hypertension, which indeed was an exclusion criterion for the study, and there were no significant differences in LV mass between LGE+ and LGE- athletes.

Further studies with long-term follow-up are clearly warranted to understand if these slightly higher ECV values are another feature of the athlete’s heart or if this may be able to early identify athletes who are starting to potentially develop diffuse interstitial fibrosis and have indeed an abnormal adaptation to training.

The quantification of the training load by assessing self-reported training load represents a potential limitation, because these parameters depend on the individual perception and accuracy of athletes. Currently, no long-term follow-up data on outcomes is available to assess the prognostic and clinical implications of the observed LGE. The T1 mapping sequence was available in only 30% of athletes, among whom 71% were female; thus, findings need to be confirmed in larger populations.