Six studies focused on LV structure [12, 14, 15, 16, 17, 18]. As structural parameters are largely influenced by BSA [19, 20] only indexed parameters were compared: LV end-diastolic diameter (LVEDD), LV end-systolic diameter (LVESD), interventricular septal thickness (IVS), LV wall thickness (LVWT), LV posterior wall thickness (LVPWT), mean wall thickness (MWT), relative wall thickness (RWT), LVM, LV length, LV end-diastolic volume (LVEDV), and LV end-systolic volume (LVESV). LVEDD was significantly higher in athletes in two studies [14, 15]. The latter also reported a significantly higher LVEDS. Binnetoglu et al. [16] reported similar mean values for LVEDD and LVESD in athletes and controls, except for basketball players. Athletes’ IVS was significantly higher in three studies [14, 15, 16], LVPWT was significantly increased in five studies [14, 15, 16, 17, 18]. Sulovic et al. [18] reported a higher RWT in athletes. In addition, static exercising athletes had a significantly higher RWT compared to dynamic exercising athletes. The same for LVM held true in this study and in three other studies [14, 15, 17, 18]. Binnetoglu et al. [16] reported the same findings in soccer players vs. tennis players and controls and a significantly increased LVM in wrestlers vs. tennis players. In summary, there is evidence of structural cardiac adaptations in junior athletes. These adaptations cannot be exclusively attributed to either dynamic or static types of sport (see Supplementary Table 6).

LV systolic function is represented by stroke volume (SV), LV ejection fraction (EF), fractional shortening (FS), cardiac output (Q), cardiac index, peak systolic myocardial velocity (S’), LV myocardial performance index (Tei index), concentricity, and sphericity index.

Endurance athlete’s EF was significantly increased in the study by Rundqvist et al. [14]. Divided into dynamic and static types of sport, Sulovic et al. [18] reported a significantly higher EF in dynamic exercising athletes and a significantly reduced EF in static exercising athletes. Contrary to these findings, three studies did not report significant differences [15, 16, 17]. The same is reported for FS [15, 16].

Diastolic function is represented by eleven parameters: mitral annulus plane systolic excursion (MAPSE), peak early LV diastolic filling velocity (E), peak late LV diastolic filling velocity (A), E/A ratio (E/A), deceleration time of E (DT), early diastolic myocardial velocity (E’), late diastolic myocardial velocity (A’), E/E’ ratio (E/E’), and E’/A’ ratio (E’/A’).

Significant differences for E and A are reported by Binnetoglu et al. [16] and Sulovic et al. [18]. Swimmers’ E was significantly higher than the other athletes and the CG [16]. Sulovic et al. [18] reported a significantly higher E in dynamic compared to static exercising athletes, and a significantly reduced E in the latter, compared to controls. Wrestlers [16] had a reduced A compared to controls as well as static exercising athletes compared to dynamic exercising athletes and controls [18]. E/A was significantly higher in athletes in the study by Rundvqist et al. [14]. Summarized, the studies reported contradictory findings regarding LV function (Supplementary Table 7).

Two studies assessed RV structure [3, 14] with the following parameters: right ventricular outflow tract (RVOT) assessed in the parasternal long-axis view (PLAX) and the parasternal short-axis view (PSAX), RVOT distal diameter, RV basal diameter, RV mid-cavity diameter, RV end-diastolic area, and RV end-systolic area. All RV parameters in D’Ascenzi et al.’s [3] study failed significance (indexed to BSA) except for RV end-systolic area. RV parameters by Rundqvist et al. [14] were significantly higher in athletes compared to controls (p 0.01). In summary, there are conflicting results on the effect of exercise on RV structure (Supplementary Table 8).

Diastolic parameters as tricuspid annular plane systolic excursion (TAPSE), E/A, E’, A’, E/E’, and E’/A’ were assessed by two authors [3, 14]. RV systolic function was assessed with two parameters: S’ and RV fractional area change (FAC). Only TAPSE, indexed to BSA, was significantly higher in athletes whereas RV FAC was significantly reduced [14]. Summarized, there are conflicting results on the effect of exercise on RV function (Supplementary Table 9).

D’Ascenzi et al. [13] reported results of biatrial remodelling. Four other authors [14, 15, 17, 18] assessed LA diameter and LA volume. There were no significant differences in the study by D’Ascenzi et al. [13] and Sulovic et al. [18]. Rundqvist et al.’s [14] study showed an increased LA diameter and volume. Soccer players in the study by Zdravkovic et al. [15] had an increased diameter compared to controls. Summarized, three out of five studies reported increased left atria dimensions in athletes (Supplementary Table 10).

Two authors reported results of RA structure [13, 14]. Parameters assessed were: RA area, RA diameter, and RA volume. Soccer players in the study by D’Ascenzi et al. [13] had a significantly larger RA volume compared to controls. Rundqvist et al. [14] observed a significantly increased RA area and diameter in endurance athletes. Summarized, there is evidence of the influence of exercise on right atrial structure (Supplementary Table 11).

Three studies [12, 16, 17] reported results on four-chamber longitudinal strain. Whereas Beaumont et al. [12] did not observe significant differences between soccer players and controls, basketball players in the study by Binnetoglu et al. [16] had a significantly lower strain (p 0.0001) compared to soccer players, swimmers, wrestlers, and controls but not tennis players. Controls in the study by Simsek et al. [17] had a significantly lower strain compared to runners and wrestlers. The authors observed the same for the two- and three-chamber view as well, with a significantly lower strain in controls. The global longitudinal strain (GLS) as an overall marker of LV function was assessed in three studies [14, 16, 17]. Binnetoglu et al. and Rundqvist et al. [14, 16] reported GLS as an average strain of 18 segments (four-, two-, and three-chamber view, two walls each, subdivided into basal, mid, and apical segments). Simsek et al. [17] however, reported GLS as an average strain of 15 segments.

Again, basketball players in the study by Binnetoglu et al. [16] presented the lowest GLS compared to other study groups (p 0.001). Simsek et al. [17] did not observe a significant difference in GLS between runners and wrestlers but a significant difference between the two athlete groups and controls was reported. Rundqvist et al. [14] on the contrary did not observe a significant difference in GLS between endurance athletes and controls.

Two studies [12, 16] reported results on circumferential and radial strain but for different LV segments. Beaumont et al. [12] measured circumferential and radial strain at the mitral valve or basal level, respectively, and mid-ventricular at the mid-papillary muscle level [12]. Circumferential strain differed significantly at both levels between soccer players and controls with higher values in soccer players. They did not observe significant differences regarding radial strain. In contrast, Binnetoglu et al. [16] reported global circumferential and radial strain, measured at the anteroseptal, anterior, lateral, posterior, inferior, and septal wall but did not state at which segmental level [16]. The combined group of athletes showed a significantly lower circumferential strain (p 0.04), however, post-hoc analysis did not reveal significant differences between groups. Only Beaumont et al. [12] assessed rotational and twist mechanics and found a significant difference between soccer players and controls in counter-clockwise apical rotation. In summary, three out of four studies reported higher strain values in athletes with results being influenced by the type of sports (Supplementary Table 12).

D’Ascenzi et al. [3] and Rundqvist et al. [14] reported RV longitudinal strain values assessed at the four-chamber view. Both authors took measurements of the RV free wall only, subdivided into basal, mid, and apical segments, and did not observe significant differences between athletes and controls (p $\mathrm{>}$ 0.05), see Supplementary Table 13.

Two of six studies reported results on 2D STE of the LA, however, D’Ascenzi et al. [13] and Rundqvist et al. [14] did not investigate the same LA parameters. D’Ascenzi et al. [13] reported results on peak atrial longitudinal strain (PALS), which is a measure of LA deformation during the reservoir phase, and peak atrial contraction strain (PACS), which is the myocardial strain during atrial systole [21, 22]. They did not report significant differences between athletes and controls. Rundqvist et al. [14] assessed LA total strain measured at the four- and two-chamber view with subdividing the LA into six segments each [23] and also did not observe significant differences between athletes and controls either. Summarized, there is no evidence of an influence of exercise on LA strain, (Supplementary Table 13).

Only D’Ascenzi et al. [13] reported results on 2D STE of the right atrium, assessed at the four-chamber view with subdividing RA into six segments. Analogous to LA function, PALS and PACS of the right atrium were assessed. The authors did not find significant differences between athletes and controls. Summarized, there is no evidence for an influence of exercise on RA strain (Supplementary Table 13).

Overall, five of six studies observed increased LV dimensions. These results are in line with other authors [4, 5, 7, 8, 24, 25, 26]. Mc Clean et al. [4], reported increased LV morphometry in a meta-analysis involving $>$14 000 junior athletes. LVEDD, LVDS, IVSD, RWT, LVM (p $\le$ 0.001), and LVPWT (p $\le$ 0.01) differed by 5.6–27.6% from non-athletic controls. Krysztofiak et al. [25] observed a significantly increased LVEDD, LVPWT (p 0.001) and IVS (p 0.01) in 791 boys and girls of different types of sports (5–18 years) and Sharma et al. [7] significantly higher IVSD, LVPWT, LVWT, LVEDD. Non-significant LV structure in soccer players compared to controls by Beaumont et al. [12] were explained by participants maturity status, referring to Nottin et al. [27] who concluded that a sufficient maturity status has to be reached to elicit an exercise-induced increase in LV structure. This observation is not in line with results by Mc Clean et al. [4] who observed a significant influence of exercise on cardiac dimensions during as well as before puberty. Ayabakan et al. [5], reported a significantly increased IVSD, LVPWT, LVMI (p 0.001), and RWT (p 0.007) in 22 pre-pubertal male swimmers (10–12 years). These results, as well as results by D’Ascenzi et al. [3], interfere with the hypothesis by Nottin et al. [27], leading to the assumption that adaptations do take place before and during puberty – however adaptations might be accelerated by hormonal influences during puberty. Additionally, other factors contribute to cardiac adaptations, such as genetics, training history and intensity.

LV hypertrophy, defined by an LVM $>$95th percentile can be differentiated into eccentric (RWT $\le$0.42) and concentric (RWT $>$0.42) hypertrophy [28]. Traditionally, it is believed that dynamic stimuli result in eccentric and static stimuli in concentric hypertrophy, respectively [29, 30]. Binnetoglu et al. [16] observed LV hypertrophy in 45.9% of all athletes with 29% being eccentric and 16.1% concentric. Interestingly, the type of LV hypertrophy was not a function of the underlying training stimulus. More than one third (35.6%) of swimmers (static-dynamic) as well as 39.1% of wrestlers (static) presented eccentric hypertrophy whereas concentric hypertrophy was observed in 28.9% of swimmers and only 4.3% of wrestlers. Sulovic et al. [18] reported eccentric hypertrophy in 79.4% of athletes in the dynamic group and 54.05% of the static group. Concentric hypertrophy was prevalent in 20.6% of athletes of the dynamic and 49.95% of the static groups. On the contrary, Simsek et al. [17] reported results that support the traditional view. Participants in this study [17] were slightly older compared to Binnetolgu et al. [16] and Sulovic et al. [18]. Aformentioned influences like hormonal status, genetics and training history could play a role in this adaptive process.

Regarding LV wall dimensions, a LVWT or IVSD $>$12 mm represents the upper limit in males, and $>$11 mm the upper limit in females [7]. Binnetoglu et al. [16] reported 98.9% of male athletes having IVSD $>$12 mm. Other authors [14, 15, 17, 18] reported none of the athletes exceeding this cut-off which is in line with Pelliccia et al. [6] who reported LVWT $>$12 mm in only 16 of 947 male elite Olympic athletes (13–49 years) as well as results by Sharma et al. [7] who reported LVWT $>$12 mm in 38 out of 720 male elite athletes (15.7 $\pm$ 1.4 years). None of the studies included in this review reported IVSD or LVWT $>$11 mm for females. During puberty, testosterone levels in males exceed female levels by up to 15 times [31]. That is why structural changes in males may be more pronounced than in females.

In conclusion, exercise does have an impact on LV structure in young athletes. This impact is influenced by athletes’ age, hence pubertal and hormonal status and also by training volume and intensity. Most studies observed significantly increased LV diameter, LVWT, and LVM in athletes.

Results regarding LV function did not clearly state significantly different results between athletes and controls or within the athletic groups. None of the studies reported any adverse results regarding a significantly impaired LV function. Three authors [12, 14, 18] noticed an improved systolic function by a significantly increased EF in soccer players and dynamic sports, respectively. Rundqvist et al. [14] observed a significantly improved diastolic function (E/A) as well as Sulovic et al. [18] in endurance-trained athletes (E). This result is in line with other studies [5, 24, 32, 33]. Unnithan et al. [33] compared 22 highly trained soccer players (12 $\pm$ 0.3 years) and 15 controls (11.7 $\pm$ 0.2 years) and noticed a significantly higher E in soccer players at rest as well as during submaximal exercise on the cycle ergometer. The authors concluded that exercise leads to an improved diastolic function in highly trained athletes even at a young age. Ayabakan et al. [5] and Rundqvist et al. [34] noticed significantly improved diastolic function in pre-pubertal male swimmers (10–12 years) and endurance athletes (13–19 years), respectively. Gajda et al. [35] examined 12 swimmers at a ultramarathon-relay with TTE before, during the competition and during recovery, e.g., 48 hours after the competition. During recovery, LV EF and SF were significantly increased compared to baseline measurements and during the competition. On the contrary, Pavlik et al. [32] compared male children, adolescents, and adults with a significantly improved diastolic function in adolescents (15–18 years) and adults (19–60 years), only. McClean et al. [4] and Sharma et al. [7] did not report a significantly improved diastolic function at all, and Sulovic et al. [18] noticed a reduced diastolic function compared to controls (E, A) in static training athletes.

In conclusion, results are controversial and do not allow a clear statement. Regarding LV systolic function, there are studies reporting improved results in young athletes but also no significantly different results compared with controls. The same is for LV diastolic function. If significantly increased results were reported, they were reported in endurance athletes but not in strength-trained athletes.

Two studies assessed RV structure [3, 14] with conflicting results. D’Ascenzi et al. [3] only noted a significant increase in RV end-systolic area index in swimmers compared to controls. All other parameters failed statistical significance when indexed to BSA. Rundqvist et al. [14] observed a significantly increased RVOT, RV basal diameter index, and RV end-systolic area index. RV adaptation is expected in athletes as the RV works hand-in-hand with the LV [36, 37]. Strength training, on the contrary, does not affect the RV to the same extent that endurance exercise does, and pulmonary vasculature is protected by high pressures [30, 38]. Current literature does not provide better insight into RV structure in children and adolescents. Only one study could be found that assessed RV structure in this age group [39]. Allen et al. [39] reported RVWT and RV cavity in 77 swimmers (32 females), aged 10.8 (5–17) years. All participants exceeded the 95th percentile of reference values for RVWT, and most of the participants for RV cavity. La Gerche et al. [38] noticed a significant increase in RV volume right after a competition in 40 adult athletes (37 $\pm$ 8 years) and hypothesize a strong impact of endurance exercise on the right heart’s structure. Comparisons of RV end-diastolic and end-systolic areas were significantly higher in adult athletes vs. controls [40] and endurance athletes vs. strength-trained athletes [41]. In conclusion, results are controversial. Results in adults point towards an influence of predominantly dynamic but not strength exercise. Further studies are needed to confirm these results in the younger age group.

Two studies investigated RV function in junior athletes [13, 14]. Rundqvist et al. [14] found a functional remodelling in endurance-trained athletes whereas D’Ascenzi et al. [3] observed no differences between swimmers and controls for most parameters and a significantly reduced RV FAC in swimmers. La Gerche et al. [38, 42] confirmed these results in adults at rest and immediately after a competition. Thus, the slightly reduced resting function that was preserved during exercise rather bears a contractile reserve but does not represent impaired RV function [42]. Reduced RV function in highly trained athletes immediately after a competition mostly recovered after one week but long-term structural remodelling is likely [38]. The adverse consequence of this is ventricular arrhythmia, which is observed in trained adults, associated with a longer duration of exercise [29, 38, 43]. To prevent this adverse adaptation in children and adolescents, a closer observation of junior athletes is needed—especially in a longitudinal setting.

Five studies examined LA structure [13, 14, 15, 17, 18], and two studies assessed RA structure [13, 14]. All studies except for two [13, 18] reported significantly increased LA and RA dimension and volumes which is in line with studies in adult athletes [44, 45, 46, 47, 48, 49]. During exercise, the LA adapts to pressure and volume overload, which leads to LA dilatation [45]. Pelliccia et al. [48] observed marked atrial dilatation ($\ge$40 mm) in 20% of n = 1777 adult athletes but defined this as a physiological adaptation to exercise as only 0.8% of athletes presented with supraventricular arrhythmias. LA adaptation was largely associated with the LV as 1 mm in LV dilatation induced a 0.4 mm increase in LA diameter. Furthermore, LVWT, BSA, and age contributed to LA adaptation. For the RA, increased diameter and volume were also regarded as a physiological adaptation [13, 49]. Gjerdalen [47], however, noticed significantly more tricuspid regurgitations in adult athletes vs. controls (n = 343/58% vs. 17/36%).

In conclusion, LA diameter and volume, RA volume, area, and diameter were increased in athletes, indicating a significant influence of exercise. As a consequence of LA dilatation, atrial flutter or fibrillation could arise as complication. Therefore, more focus should be placed on atrial examination to detect adverse adaptations as early as possible.

Two of four studies reported improved LV function in junior athletes. In the study by Simsek et al. [17], this difference did not depend on the types of sports, as there were no significant differences between endurance and strength athletes. On the contrary, Binnetoglu et al. [16] observed a significantly reduced strain in basketball players compared to other groups and controls whereas Rundqvist et al. [14] did not observe significant differences in strain between endurance athletes and controls. The latter is in line with no significant results reported by other authors [33, 52, 53] who compared 22 soccer players (12.0 $\pm$ 0.3 years) with 15 controls (11.7 $\pm$ 0.2 years) at rest and during submaximal exercise [33], and 76 cross country skiers (12.1 $\pm$ 0.2 years) with 25 controls (12.1 $\pm$ 0.3 years) [52]. No significant differences for GCS and GRS were reported by Charfeddine et al. [54] in a cohort of 33 soccer players (13.19 $\pm$ 1.2 years) and 20 controls (12.9 $\pm$ 2.1 years). Furthermore, GLS turned out to be significantly reduced in athletes (20.68 $\pm$ 2.05 vs. 22.99 $\pm$ 2.32, p 0.001). De Luca et al. [55] support these results. They recognized reduction in LV strain as an early sign of LV dysfunction in n = 50 athletes (soccer, cycling, basketball, 14–19 years). In conclusion, results are conflicting regarding LV function. Contrary to the majority of studies [14, 33, 52, 53, 54], two studies presented in this review [12, 17] reported an improved LV function in junior athletes.

Two studies assessed RV function in junior athletes [3, 14] and did not observe significantly different values in athletes and controls. Furthermore, there was a tendency toward lower strain values in athletes vs. controls. Bjerring et al. [52] observed a significantly reduced RV GLS (28 $\pm$ 4 vs. 31 $\pm$ 3, p 0.001) in 76 cross country skiers (12.1 $\pm$ 0.2 years) compared to 25 controls (12.1 $\pm$ 0.3 years) and a negative correlation of RV GLS with VO${}_{2max}$ (r = –0.22, p 0.06) and the amount of exercise (r = –0.24, p 0.05). They hypothesize that the RV might be affected by cardiac fatigue through exercise which is explained by a reduced resting cardiac function in athletes after strenuous exercise. This cardiac fatigue might affect the RV earlier than the LV [38, 52]. In conclusion, results of this review do not point towards a significant influence of exercise on RV longitudinal strain. However, only two studies out of six reported RV longitudinal strain. There is evidence that the RV is affected by strenuous exercise. As the RV bears the potential to elicit arrhythmogenic cardiomyopathies, the focus of further studies should be placed of the assessment of RV function.

Two studies investigated atrial function by 2D STE [13, 14]. D’Ascenzi et al. [13] examined biatrial function in swimmers vs. controls and did not report significant differences. Rundqvist et al. [14] observed non-significant differences in LA strain between endurance athletes and controls. One study was identified that examined LA function in n = 595 highly trained soccer players (25.1 $\pm$ 4.6 years) and n = 47 controls (26.2 $\pm$ 6.5 years) with no significant differences between groups [47]. Furthermore, in athletes with enlarged atria, LA function was still preserved. In conclusion, results do not point towards a significant influence of exercise on LA strain parameters. However, only two studies of six investigated LA strain. Only one study examined RA function and did not observe a significant influence of exercise. As both atria, are affected by higher blood volume in athletes, and the LA also by an increase in pressure during exercise, assessing atrial function is of importance.