Exercise and Cardiac Remodeling in Normal and Athletic States Geometrical remodeling of the mitral and tricuspid annuli in response to exercise training: a 3-D echocardiographic study in elite athletes

Intense exercise exposes the heart to signi ﬁ cant hemodynamic demands, resulting in adaptive changes in cardiac morphology and function. Nevertheless, the athletic adaptation of the atrioventricular valves remains to be elucidated. Our study aimed to characterize the geometry of mitral (MA) and tricuspid (TA) annuli in elite athletes using 3-D echocardiography. Thirty-four athletes presented with functional mitral regurgitation (FMR) were retrospectively identi ﬁ ed and compared with 34 athletes without mitral regurgitation (MR) and 34 healthy, sedentary volunteers. 3-D echocardiographic datasets were used to quantify MA and TA geometry and lea ﬂ et tenting by dedicated softwares. MA and TA areas, as well as tenting volumes, were higher in athletes compared with controls. MA area was signi ﬁ cantly higher in athletes with MR compared with those without (8.2±1.0 vs. 7.2±1.0 cm 2 /m 2 , P < 0.05). Interestingly, athletes with MR also presented with a signi ﬁ cantly higher TA area (7.2±1.1 vs. 6.5±1.1 cm 2 /m 2 , P < 0.05). Nonplanar angle describing the MA ’ s saddle shape was less obtuse in athletes without MR, whereas the values of athletes with MR were comparable with controls. The exercise-induced relative increases in left ventricular (35±25%) and left atrial (40±29%) volumes were similar; however, the increment in the MA area was disproportionately higher (63±23%, overall P < 0.001). The relative increase in TA area (40±23%) was also higher compared with the increment in right ventricular volume (34±25%, P < 0.05). Atrioventricular annuli undergo a disproportionate remodeling in response to regular exercise. Athletic adaptation is characterized by both annular enlargement and increased lea ﬂ et tenting of both valves. There are differences in MA geometry in athletes presented with versus without FMR. NEW & NOTEWORTHY We have characterized the annular geometry of mitral and tricuspid valves in elite athletes using 3-D echocardiography. We have found that exercise-induced remodeling of the atrioventricular annuli comprises a disproportionate dilation of annular dimensions and increased lea ﬂ et tenting of both valves. Moreover, we have demonstrated a more pronounced saddle shape of the mitral annulus in athletes without mitral regurgitation, which was not present in those who had mild regurgitation.


INTRODUCTION
Regular, intense physical exercise results in complex adaptive changes in the morphology and function of the heart (1). Nevertheless, the vast majority of data are related to the ventricles and the atria (2). A balanced dilation of the cardiac chambers is a characteristic and unequivocal feature of an athlete's heart. Although ventricular and atrial dilation can significantly affect atrioventricular annular geometry and related valvular competency in various pathological states, less is known about the exercise-induced alterations in the shape and function of the mitral and tricuspid annuli in elite athletes (3).
Primary valvular heart diseases show the same prevalence in athletes than in the general population (4). However, mitral and tricuspid insufficiency is reported more frequently among athletes, suggesting predominantly secondary (functional) origin (5). The classical ventricular type of functional regurgitation is mainly characterized by increased leaflet tenting, whereas a notably enlarged annulus rather refers to atrial functional regurgitation (3). The athlete's heart shares several features of both types of functional regurgitation: pronounced ventricular and atrial dilation may even exaggerate their impact on annular geometry and valvular function. Thus, a mixed type of functional regurgitation can be assumed in athletes presented with mitral or tricuspid insufficiency.
Our current study aimed to characterize the geometry of mitral and tricuspid annuli in elite athletes compared with healthy, sedentary volunteers using three-dimensional (3-D) echocardiography. We hypothesized that atrioventricular annuli undergo a disproportionate remodeling in contrast to cardiac chambers, and there is a difference in the mitral annular geometry between athletes presented with versus without mitral regurgitation.

METHODS
As part of our Center's complex sports cardiology screening program, we retrospectively identified healthy athletes with 3-D transthoracic echocardiographic datasets available for detailed analysis of the left and right hearts. Using this cohort, we have selected those athletes who presented with at least mild mitral regurgitation (MR). Exclusion criteria were 1) presence of any primary cause for MR revealed at the first report or during the review process of the previously acquired datasets (including prolapse, billowing, cleft, etc.), 2) suboptimal 3-D dataset image quality for the analysis of the annuli, and 3) noncompatible image source, as tricuspid annular (TA) quantification is available only by a single vendor. An additional age-matched cohort of athletes was selected from our database with the same sex distribution but no mitral regurgitation. An age-matched healthy, sedentary population (<3 h of exercise/wk) with the same sex distribution served as the control group having neither MR nor tricuspid regurgitation (TR). Detailed medical history and training regime were obtained along with a standard physical examination and 12-lead ECG. Two-dimensional (2-D) and 3-D echocardiography and then cardiopulmonary exercise testing (CPET) were performed on all athletes on the same day. All participants provided written, informed consent to study procedures. Our study is in accordance with the Declaration of Helsinki and approved by the Medical Research Council (ETT-TUKEB No. 13687-0/2011-EKU).

Cardiopulmonary Exercise Testing
CPET for peak oxygen uptake (V _ O 2 and V _ O 2 /kg) quantification was performed on a treadmill until the exhaustion of the athletes. The volume and composition of the expired gases were analyzed breath by breath using an automated cardiopulmonary exercise system (Respiratory Ergostik, Geratherm, Bad Kissingen, Germany). Subjects were encouraged to achieve maximal effort, which was confirmed by respiratory exchange ratio, by lactate curves by regular fingertip lactate measurements every 2 min during the test, and by reaching the predicted maximal heart rate and a plateau in V _ O 2 .

2-D and 3-D Echocardiography
Transthoracic echocardiographic examinations were performed on a commercially available ultrasound system (E95, 4Vc-D probe, GE Vingmed Ultrasound, Horten, Norway). A standard acquisition protocol consisting of 2-D loops from parasternal, apical, and subxiphoid views was applied. Beyond conventional echocardiographic examination, ECGgated full-volume 3-D datasets reconstructed from four cardiac cycles optimized for the left or the right heart were obtained for further analysis on a separate workstation (Fig.  1). Left ventricular (LV) internal diameters and wall thicknesses; left atrial (LA) 2-D end-systolic volume; mitral inflow velocities and E wave deceleration time; systolic (s 0 ), early diastolic (e 0 ), and atrial (a 0 ) velocities of the mitral lateral and septal annulus; average E/e 0 ; right ventricular (RV) basal short-axis diameter, tricuspid annular plane systolic excursion (TAPSE), fractional area change (FAC), and TA velocities; and right atrial (RA) 2-D end-systolic volume were measured according to current guidelines (6).
MR was graded based on the measurement of vena contracta width (VCW), preferably on zoomed color parasternal long-axis view loops at the narrowest portion of the regurgitation jet as it emerges beyond the orifice. Mild MR was defined as having a measurable vena contracta but with a width of <0.3 cm as per current guidelines (7). "Non-MR" was defined as an absolutely not or just barely detectable jet, no flow convergence, and not having measurable vena contracta with a faint, incomplete, or no continuous-wave Doppler signal. Similarly, TR grading was based on measuring the VCW. TR jet was established either on apical fourchamber or in parasternal right ventricular (RV) inflow views. Mild TR was defined as a small jet but with measurable VCW (<0.3 cm). Lack of TR was defined as an absolutely not or just barely detectable jet, no flow convergence, and not having measurable VCW with a faint, incomplete, or no continuous-wave Doppler signal. 3-D datasets focused on the left heart were processed using semiautomated, commercially available softwares (4D LV-Analysis 3, TomTec Imaging, Unterschleissheim, Germany, and EchoPAC v204, 4D Auto LAQ, GE). We determined LV end-diastolic volume index (EDVi), end-systolic volume index (ESVi), stroke volume index (SVi), and LV mass index (LV Mi). To assess global LV function, ejection fraction (EF), global longitudinal strain (GLS), and global circumferential strain (GCS) were also calculated. Concerning the 3-D quantification of the LA, we measured maximal (LAV max ) and minimal (LAV min ) volumes and the volume at the onset of atrial contraction (LAV preA ). Using these volumetric data, we calculated LA total emptying fraction as 100 Â (LAV max -LAV min )/LAV max , LA passive emptying fraction as 100 Â (LAV max -LAV preA )/LAV max , and LA active emptying fraction as 100 Â (LAV preA -LAV min )/LAV preA , as parameters of LA reservoir, conduit, and contractile function, respectively. The analysis of the 3-D datasets focused on the right heart was performed using 4D RV-Function 2 software (TomTec Imaging). We quantified 3-D RV EDVi, ESVi, and SVi, EF, and septal and free wall two-dimensional (2-D) longitudinal strain as well. Concerning the 3-D quantification of the RA, we used the same software as for the LA (EchoPAC v204, 4D Auto LAQ, GE) due to the lack of dedicated RA software package. Similarly, we measured maximal (RAV max ) and minimal (RAV min ) volumes and the volume at the onset of atrial contraction (RAV preA ). Using these volumetric data, we calculated RA total emptying fraction as 100 Â (RAV max -RAV min )/RAV max , RA passive emptying fraction as 100 Â (RAV max -RAV preA )/RAV max , and RA active emptying fraction as 100 Â (RAV preA -RAV min )/RAV preA , as parameters of RA reservoir, conduit, and contractile function, respectively.

3-D Quantification of Mitral and Tricuspid Annuli
Mitral annulus (MA) was quantified by a commercially available software (EchoPAC v204, 4D Auto MVQ, GE) previously validated against other mitral valve quantification softwares and also against transesophageal image acquisition and subsequent analysis (8). First, the automatic selection of enddiastolic and end-systolic time points was confirmed by visual assessment of cardiac cycle events. Then, the orientation of the 3-D dataset was aligned to conform with prespecified standard views. A midsystolic reference frame was automatically selected, which was subsequently used to report static parameters. Six anatomical landmark points were placed using the two standard orthogonal planes (commissural and long-axis views), and then, the software created the 3-D model of the mitral valve annulus and leaflets. If needed, model contours could be modified manually throughout the cardiac cycle. Reported MA and leaflet parameters are as follows (also presented graphically on Fig. 2): 3-D MA area, 2-D MA area (projected 2-D area at the level of the best-fit plane), MA perimeter, MA anteroposterior diameter, MA posteromedial-anterolateral diameter, commissural diameter, intertrigonal distance, MA sphericity index (the ratio between anteroposterior and posteromedial-anterolateral diameters), MA height, MA nonplanar angle (describing the MA saddle shape), annulus height to commissural width ratio (ACHWR, describing the MA saddle shape), MA systolic excursion, MA maximal systolic velocity, MA area fraction, mitral-aortic angle, anterior and posterior leaflet lengths, areas, angles, tenting height, area, and volume.
TA was quantified by commercially available software (EchoPAC v204, 4-D Auto TVQ, GE). First, the automatic selection of end-diastolic and end-systolic time points was confirmed by visual assessment of cardiac cycle events. Then, the orientation of the 3-D dataset was aligned by the placement of tricuspid valve center point and adjusting RV long-axis to conform with prespecified standard views. A midsystolic reference frame was automatically selected, which was subsequently used to report static parameters. Five anatomical landmark points were placed using the two standard orthogonal planes (4-chamber and RV 2-chamber views), and then, the software created the 3-D model of the tricuspid valve annulus and leaflets. If needed, model contours could be modified manually throughout the cardiac cycle. Reported TA and leaflet parameters are as follows (also presented graphically on Fig. 3): 3-D TA area, 2-D TA area (projected 2-D area at the level of the best-fit plane), TA area fraction, TA perimeter, TA diameter on fourchamber view, TA diameter on two-chamber view, TA major and minor axes, sphericity index (the ratio between four-chamber view and two-chamber view diameters), TA systolic excursion, maximal tenting height, and tenting volume.
Parameters were normalized to body surface area (BSA) calculated by the Mosteller formula (9).

Statistical Analysis
Statistical analysis was performed using dedicated software (StatSoft Statistica, v. 12, Tulsa, OK). Continuous variables are presented as means ± SD, whereas categorical variables are reported as frequencies and percentages. Concerning LV EDVi, LAVi max , 3D MA area, RV EDVi, RAVi max , and 3-D TA area indices, individual values of athletes were normalized to the mean value of the control group to calculate their relative increase. After verifying the normal distribution of each variable using the Shapiro-Wilk test, athlete groups were compared with unpaired Student's t test or Mann-Whitney U test for continuous variables and chisquare or Fisher's exact test for categorical variables, as appropriate. Multiple-group comparisons were performed using ANOVA (with Fisher's post hoc test) or the Kruskal-Wallis test (with Dunn's post hoc test). The relative increase in the aforementioned parameters was compared using oneway repeated-measures ANOVA (with Fisher's post hoc test). The Pearson or Spearman test was computed to assess the correlation between continuous variables. Multivariate linear regression analysis was performed in athletes to find independent determinants of MA and TA 3-D area index. To avoid multicollinearity, tolerance was set at >0.4. A twosided P value of <0.05 was considered as statistically significant. The intra-and interobserver variabilities were evaluated by intraclass correlation coefficient (ICC) values. To assess the intraobserver reproducibility of the presented key parameters, the experienced first operator (A.F.) of the offline measurements repeated the 3-D analysis in a randomly selected subset of 5-5-5 subjects from each group blinded to previous results. The interobserver variability was deter-mined by 3-D analysis of the same subjects by a second experienced operator (A.K.) in a blinded fashion.

RESULTS
Four hundred and twenty-five healthy athletes were identified with a previous echocardiographic examination with both left heart-and right heart-focused 3-D datasets. From this cohort, 42 (9.9%) athletes presented with at least mild MR. Eight athletes were excluded due to either suboptimal image quality for MA quantification (n = 6) or noncompatible image source vendor (n = 2). Thus, 34 athletes (male/female: 26/8) formed the final study population (MR athlete group). Furthermore, 34 age-matched athletes (non-MR athlete group) and 34 healthy, sedentary individuals (control group) with the same sex distribution (male/female: 26/8) were selected from our database ( Fig. 4; Table 1).
Basic anthropometric, hemodynamic, and training specific data of the study groups are summarized in Table 1. Our athlete populations consisted of mixed or endurance-trained individuals, predominantly soccer players (28%), water polo players (28%), and swimmers (25%). Athletes had a higher height, weight, and lower resting heart rate than the sedentary control group. Athletes without MR demonstrated significantly higher systolic blood pressure compared with the other two groups. Athletes have been participating in competitive sports for an average of 15 years; at the time of the echocardiographic investigation, they trained an average of 17 h/wk. BSA, heart rate, competitive years, weekly training hours, and CPET-derived peak exercise capacity did not differ between the two athlete groups (Table 1).
Conventional 2-D and 3-D echocardiographic parameters are shown in Table 2. As expected, there were significant differences between the athlete groups and the control group concerning LV and RV morphological and functional parameters. LV and RV EDVi and ESVi were significantly higher in the athletes compared with controls; on the other hand, these did not differ between the MR and non-MR athlete groups. LV Mi, LV SVi, and RV SVi values were the highest in the athlete group with MR compared with non-MR and control groups. LV EF, GLS, GCS, RV FAC, and RV EF showed significantly decreased resting values in the athletic groups than controls. Notably, LV GLS was higher in the MR athlete group compared with the non-MR group. TAPSE and RV septal and free wall longitudinal strain did not differ between groups. 3-D LA and RA maximal volume indices were higher among athletes with MR, even compared with non-MR athletes. In terms of LA reservoir and contractile function, there was no difference between the groups; however, LA passive EF referring to conduit function was significantly lower in the MR athlete group compared with controls. Although RA contractile function was similar, athlete groups presented with lower RA reservoir and conduit function. Moreover, RA total emptying fraction was significantly lower in the MR athlete group compared with non-MR athletes. Concerning diastolic function, there was no difference in either mitral inflow velocities or early diastolic mitral annular velocities between our study groups. In contrast, late diastolic annular velocities (a 0 ) were lower in athletes than in controls ( Table 2).
In the MR athlete group, all individuals presented with mild MR, and no moderate regurgitation was established (VCW: 0.22 ± 0.07 cm). Interestingly, 74% of athletes with MR also had mild TR (VCW: 0.19 ± 0.07 cm). In the non-MR group, 9% of athletes presented with mild TR (VCW: 0.12 ± 0.04 cm).
We have compared the exercise-induced relative dilation of the LV, the LA, and the MA in the overall athlete cohort. Although the relative increase in LV EDVi and LAVi max was comparable, the MA 3-D area index's increment was disproportionately higher, with an average enlargement of over 60%. Concerning the right heart, the relative increase in the 3-D TA area index was significantly higher than in RVEDVi but showed no difference compared with RAVi max . The increment in RAVi max was higher compared with RVEDVi ( Fig.  5). We investigated the relative geometrical changes in female athletes (n = 16) separately and found that the increment in LAVi max (46 ± 33%) was significantly higher compared with LVEDVi (32 ± 21%) but was still lower than MA 3-D area index (61 ± 30%, overall P < 0.001). In terms of the right heart, the increment in TA 3-D area index (35 ± 25%) was similar to RVEDVi (35 ± 25%), but both were lower compared with RAVi max (49 ± 45%, overall P < 0.001).
The three groups differed significantly from each other with regard to all of the parameters (3-D and 2-D area indices, perimeter, anteroposterior, posteromedial-anterolateral, commissural diameters, intertrigonal distance) describing the size of the MA, as athletes having MR had significantly higher values even compared with athletes without MR (Table 3; Fig. 6). The athlete groups had higher annulus height and less obtuse mitral-aortic angle. Athletes without MR had a more pronounced MA saddle shape, as suggested by the significantly less obtuse nonplanar angle and higher annulus height to commissural width ratio (AHCWR, Fig. 6). Interestingly, athletes with MR were rather similar in this regard to sedentary controls. Sphericity, annular excursion, and MA area fraction did not differ between the athlete groups. Anterior leaflet length and area were significantly higher in the MR athletes even compared with non-MR athletes; however, posterior leaflet length and area did not show a difference between the two athlete groups. The anterior leaflet angle was significantly lower in the MR athlete group compared with non-MR athletes. Tenting height, area, and volume indices were higher in both athlete groups compared with the control group (Table 3; Fig. 6).
with both non-MR athletes and controls (Table 4 and Fig. 7). TA perimeter, two-chamber view diameters, and major axis diameters were comparable between athlete groups but still larger than controls. TA sphericity, area fraction, and systolic excursion were similar between groups. Maximal tenting height and tenting volume index were significantly higher in the athlete groups compared with the sedentary controls (Table 4; Fig. 7).
RAVi max , and V _ O 2 /kg), RAVi max was the only independent predictor of TA 3-D area index, with a cumulative R value of 0.648 (P < 0.001; Table 5).
Despite using transthoracic 3-D echocardiographic images for annular quantification, good intra-and interobserver variabilities could be assessed concerning all of the key measurements (intraobserver and interobserver ICC values with 95% confidence intervals, respectively;

DISCUSSION
To the best of our knowledge, our study is the first to characterize the annular geometry of mitral and tricuspid valves in elite athletes using 3-D echocardiography. We have found that exercise-induced remodeling comprises a significant dilation of annular dimensions and increased leaflet tenting of both atrioventricular valves. MA showed a disproportionate enlargement compared with both the LV and the LA, whereas TA was more extensively increased than the RV. Moreover, we have demonstrated a more pronounced saddle shape of the MA in athletes without MR, which was not present in those athletes who had mild MR.
The mitral and tricuspid annuli are the fibrous continuity between ventricular and atrial myocardium and leaflet tissues. These saddle-shaped structures' function is to minimize leaflet stress and optimize coaptation throughout the corresponding phase of the cardiac cycle (12). 3-D echocardiography revolutionized our understanding of the MA geometry and dynamics, highlighting important pathophysiological differences between primary and secondary mitral valve diseases. Although MA dilation can be present both in patients with mitral valve prolapse (Barlow disease or fibroelastic  deficiency) and in patients with ventricular functional MR, the latter group is rather characterized by a stiff, hypodynamic MA. The flattening of the saddle-shaped MA is also considered as an adverse response to different pathological stimuli (11,12). The added value of 3-D echocardiography is evident not just in terms of describing the nonplanar MA geometry and eliminating the underestimation of MA size (when assessed by linear measurements) but also in the more accurate quantification of MR and supporting a more personalized diagnostic approach and subsequent therapeutic decisions (11). Recently, 3-D modeling and systolic tracking of the TA became available using 3-D echocardiography, and it promises similar advantages compared with conventional assessment (13).
Regular physical exercise exposes the cardiovascular system to often extreme hemodynamic demands, resulting in significant changes in cardiac morphology and function (14). Although the ventricular myocardium's dynamic adaptation is definitely the most widely known aspect when considering an athlete's heart, alterations of the interconnected valvular apparatus should not be overlooked. During intense exercise, the atrioventricular valves have to support an adequate diastolic filling while also keeping their systolic competency to maintain the "one-way" circulation in the face of a wide range of intracardiac pressures and ventricular outputs (15).
Still, data are scarce even concerning MA and TA linear dimensions in athletes. In accordance with our current results, it was previously shown that athletes have significantly higher MA linear diameters than sedentary controls. Moreover, a more pronounced MA enlargement exists in athletes presenting with MR (5). Although functional mitral and tricuspid regurgitation rarely exceed the mild degree in athletes, a higher overall prevalence of MR and TR is reported in the literature (5). In our cohort, the vast majority of athletes with functional MR had concomitant TR, and interestingly, the MR group is presented with higher TA dimensions as well. This phenomenon implies that athletes prone to more excessive dilation of the MA are susceptible to a more pronounced TA remodeling. Of note, it has been suggested that alterations in TA geometry can be present even in the case of degenerative processes affecting the MA primarily (16). The mitral valve apparatus also shows a distinct change in shape if we compare the two athlete groups: although in athletes with MR, the parameters referring to the MA saddle shape, namely, annulus height to commissural width ratio (AHCWR) and nonplanar angle, were comparable with the values of sedentary controls, increased AHCWR and a less obtuse nonplanar angle were found in athletes without MR. These latter indicate a more pronounced MA saddle shape in that group, which is, according to our knowledge, a unique finding across the spectrum of physiological or pathophysiological  alterations of the MA (11). However, the anatomical advantages of this phenomenon are not completely clear. We may hypothesize that it may serve as an adaptive change to maintain proper coaptation during the different exercise levels when chamber geometry and intracardiac pressures continuously change. The MA nonplanarity of athletes presented with MR did not significantly differ from sedentary controls, suggesting that in this population, the mild regurgitation may originate at least partially from the "insufficient" geometrical adaptation of the MA to regular, intense exercise. Interestingly, the anterior leaflet area, but not posterior leaflet area, was significantly higher in athletes with MR compared with those without. According to previous data, mechanical stretch by papillary muscle displacement can induce leaflet hyperplasia (17). However, as the enlargement of the leaflets is usually proportional in pathological states, the asymmetrical growth of the mitral valve leaflets may be a specific manifestation of this "adverse" athletic adaptation accompanied by MR.
Athletes with MR showed tendentially higher LV volumes; significantly higher LV GLS, LAVi max , and RAVi max ; and significantly lower RA total emptying fraction than athletes with no MR. Although such mild MR (assessed during resting conditions) certainly cannot be considered a hemodynamically relevant stimulus, these observations refer to what can be seen in the case of pathological LV volume overload (18,19). These findings can originate in two ways: 1) an athlete with a predisposition to a more pronounced exercise-induced chamber dilation may also be more prone to higher MA dimensions and a consequential MR, and 2) there is an inherent anatomical cause of the MR that may significantly increase during exercise, resulting in more excessive cardiac remodeling. Nevertheless, further studies involving stress echocardiography are warranted to clarify this classical "chicken or the egg" dilemma. In accordance with our previous publication (2), we found a relatively more pronounced biatrial dilation in female athletes. Future studies with higher case numbers should also target sex differences in this regard.
MA and TA areas showed at least moderate correlations with the corresponding ventricular and atrial volumes, confirming these exercise-induced dilatative processes' interconnected nature. A modest but significant increase in the tenting volumes of both atrioventricular valves could also be demonstrated, related to the ventricular volumes. Accordingly, a mixed type of functional MR and TR (showing the characteristics of both atrial and ventricular types) was established in our examined cohort of athletes. MA and TA 3-D areas also correlated with CPET-derived exercise performance; in the case of the MA, V _ O 2 /kg was even found to be an independent determinant. These findings rather support that the atrioventricular annular dilation is still one of the adaptive aspects of the complex physiological cardiac response to regular, intense exercise training. Notably, however, atrial volumes were independent determinants of annular areas, whereas ventricular volumes were not. These results resonate with recent publications showing that the RA dilation is the major determinant of TA size in patients with atrial fibrillation and/or functional TR (20,21). Longitudinal studies with long-term follow-up are required to establish the (patho)physiological link between exercise-induced cardiac remodeling, functional valvular regurgitation, and, eventually, occurrence of atrial fibrillation.

Limitations
Several limitations have to be acknowledged. First, the study cohort's size is relatively limited, mainly because we aimed to investigate elite athletes with available good-quality 3-D acquisitions and presenting with MR and compare them with those without. Second, clinical relevance and therapeutic consequences of such quantifications would be related to athletes with primary valvular heart diseases and higher grades of regurgitation; however, athletes with such disorders were absent in our database. Nevertheless, we plan a longterm follow-up of our athletes to investigate the progression/ regression dynamics of the established alterations and the potential occurrence of clinical adverse events. 3-D echocardiographic quantification of the mitral valve is primarily designed for transesophageal images, where spatial and temporal resolutions are higher. However, athletes are generally presenting with exceptionally good acoustic windows, and data have shown that results derived from transthoracic and transesophageal approaches are rather interchangeable (8). Intraobserver and interobserver variability also confirmed the reliability of our measurements. The generalizability of our results to other sport disciplines, to amateur athletes, and to different age categories remains unknown.

Conclusions
In our retrospective cohort study of young elite athletes, we have shown that beyond the dilation of the cardiac chambers, atrioventricular annuli may undergo a disproportionate remodeling in response to regular, intense exercise training. Athletic valvular adaptation is characterized by both annular enlargement and increased leaflet tenting of both the mitral and tricuspid valves. There are also specific differences in MA geometry between athletes presented with or without functional MR. Further research is warranted in athletes competing in other sport disciplines, in master athletes, and in athletes with primary valvular heart diseases.