Article Text
PDF
Abstract
Background Physiological cardiac adaptation in athletes is influenced by body surface area, gender, age, training intensity and sport type. This study assesses the influence of sport category and provides a physiological reference for sport category and gender.
Methods Three hundred and eighty-one subjects (mean age 25±5 years, range 18 to 39 years; 61% men) underwent cardiac MRI and ECG: 114 healthy non-athletes (≤3 training h/week) and 267 healthy elite athletes (mean 17±6.6 training h/week). Athletes performed low-dynamic high-static (LD-HS, n=42), high-dynamic low-static (HD-LS, n=144) or high-dynamic high-static sports (HD-HS, n=81).
Results Left ventricular (LV) end-diastolic volume (EDV) index (ml/m2) for non-athletes/LD-HS/HD-LS/HD-HS, respectively, was 101/107/122/129 in men and 90/103/106/111 in women. LV end-diastolic mass (EDM) index (g/m2) for non-athletes/LD-HS/HD-LS/HD-HS was, respectively, 47/49/57/69 for men and 34/38/42/51 for women. Left or right ventricular EDV ratios were alike in all groups. LV EDV/EDM ratios were similar in non-athletes/LD-HS/HD-LS athletes, and only lower in HD-HS athletes, disproving selective ventricular wall thickening in LD-HS athletes. Multivariate linear regression demonstrated HD-LS and HD-HS sport category coefficients (p<0.01) larger than those of training hours, gender and age (LV EDV/EDM coefficients for sport category LD-HS 6/0.75, HD-LS 16/7, HD-HS 21/17). ECG abnormalities were most frequent in HD-HS athletes and in male subjects.
Conclusions This study demonstrates a balanced cardiac adaptation with preserved ratios of LV/right ventricular volume (in all sport categories) and LV volume/wall mass (in LD-HS and HD-LS sports). Sport category has a strong impact on cardiac adaptation. HD-HS sports show the largest changes, whereas LD-HS sports show dimensions similar to non-athletes.
Statistics from Altmetric.com
Introduction
Increasing requests for cardiac MRI (CMR) investigations in athletes, either for screening purposes or to rule out pathology, require a reference framework that includes the main determinants of cardiac morphology.1 Numerous studies reporting on the physiological cardiac adaptation in athletes suggest that the most important factors are body surface area (BSA), gender, age, training intensity and the type of sport.2,–,4
Following the widely used Mitchell classification, sports can be characterised as being high or low in dynamic (endurance, isotonic) and static (strength/resistance, isometric) components.5 The low-dynamic high-static sports (LD-HS, strength sports) have particularly been subject to controversy. For decades, selective ventricular wall mass increase has been assumed in strength-trained athletes, and this hypothesis was supported by several reports,6,–,10 while other studies do not to observe this phenomenon.11,–,16 Moreover, a reference framework should take sport category into account when assessing the cardiac pathology in athletes as high-dynamic high-static (HD-HS) sports seem to represent the upper limits of physiological cardiac adaptation, greater than that caused by other sport categories.3 ,7 ,8 ,17 ,18
By investigating cardiac parameters on CMR in a large group of athletes of various sport categories, this study aims to assess the influence of sport category, alongside other major determinants, on cardiac adaptation, and to provide a reference framework for the physiological limits subdivided by the category of sport and gender.
Methods
Study population
Three hundred and eighty-one subjects (mean age 25±5 years, range 18 to 39 years; 61% men) were included: 114 healthy non-athletes and 267 healthy elite athletes, competing at national or international level. The non-athlete group serves as a baseline reference and exercised for a maximum of 3 h per week. Athletes exercised for a mean of 17±6.6 h per week. The four quadrants of the Mitchell classification comprise combinations of LD-HS, HD-LS (high-dynamic low-static) and HD-HS sport categories. The low-dynamic low-static (LD-LS) sports are assumed to show no cardiac adaptation – analogous with the control group – due to the low estimated percentage of the maximal oxygen uptake and muscle contraction.5 LD-LS sports were therefore not included in this study. Table 1 lists the individual sports and the corresponding number of subjects included. Inclusion threshold for elite athletes was based mainly on the level of competition and not on a fixed number of training hours, since the realistic numbers of training hours differ widely between different types of sports.
Numbers of subjects per sport and sport category
None of the subjects had evidence of hypertension or cardiovascular, pulmonary or metabolic disease as evidenced by medical and family history, blood pressure measurement, ECG and the CMR scan itself. All subjects gave a written informed consent and the study was approved by the institutional ethics committee.
Study protocol
CMR imaging was performed on a 1.5-T MRI scanner (Achieva, Philips, Best, The Netherlands) including the steady-state free precession (SSFP) cine images (two-chamber left ventricular (LV) and ventricular (RV), four-chamber, short-axis, LV and RV outflow tract) and quantitative flow (Q-flow) measurements of all four cardiac valves. The short axis was identified using the two-chamber left and four-chamber images and included the whole heart from ventricular apex up to and including the atria, spanning 14 to 20 slices (10 mm without interslice gap, 50 frames per cardiac cycle, 256×256 matrix, field of view 350 to 400 mm, repetition/echo time 3.2/1.6 ms, in-plane pixel size of 1.4 mm, flip angle 55', breath-held acquisition time of 18 heartbeats).
CMR analysis
Blinded observers used an established reliable and reproducible contour-tracing protocol19 to trace the endocardial and epicardial contours of both ventricles on a workstation (View Forum cardiac package version R5.1V1L2.SP3, Philips, Best, The Netherlands). In a previous study, this protocol demonstrated maximum interobserver and intraobserver disagreement of 8 and 5%, respectively. Furthermore, in a comparison of the Philips contour-tracing software to Medis software (QMass 7.1) disagreement between the observers was at most 5% in 30 test cases, showing good interplatform reproducibility (R2 0.99).19 The observers were trained with test cases to achieve the necessary expertise before performing the analysis for this study. The results were finalised after the approval of a blinded observer experienced in CMR. End-diastolic (ED) and end-systolic (ES) endocardial contours were used to calculate the ED volume (EDV) and ES volumes (ESV), ejection fractions (EF), stroke volumes and ED wall mass (EDM). LV and RV outflow tracts (LVOT/RVOT) were included in the endocardial borders and papillary muscles and trabeculae were excluded from the endocardial contours and therefore included in the blood volume.19 End-diastolic ventricular diameters and septal wall thickness were measured on the short-axis images.
ECG analysis
ECGs were analysed using an online rating system by two experienced sports cardiologists, following recent guidelines on the interpretation of ECGs in athletes and recommendations supported by leading experts in the field of sports cardiology.20 ,21 They were aware of gender, age and whether the subject was an athlete, but unaware of the sport category and the number of training hours. ECGs were classified as showing no abnormalities (normal), group 1 abnormalities (ie, common and training-related ECG changes) or group 2 abnormalities (ie, uncommon and training-unrelated ECG changes).20
Statistical analysis
Continuous data are presented as the mean ± SD and 95th percentile for quantitative cardiac parameters of both ventricles. Categorical data are presented as percentages. CMR volumes and wall mass were indexed to BSA, with BSA calculated by the Dubois and Dubois formula: BSA (m2)=0.20247×height (m)0.725×weight (kg)0.425.22 Differences between groups were assessed using the one-way analysis of variance (ANOVA) test with posthoc Bonferroni correction. A two-tailed p value of 0.05 was considered to be statistically significant. The simultaneous effect of sport category, the number of weekly training hours, age and gender on two different outcome variables (LV EDV and LV EDM) was assessed using multivariate regression analysis. For each outcome variable, a model was fitted and covariates were either retained or removed by a backward selection. Wald's p value threshold of 0.15 was considered statistically significant and was required for covariates to remain in the model as a predictor, as is the recommended practice.23 All statistical analyses were performed with R version 2.12.1.24
Results
Baseline characteristics of all four groups are presented in table 2. The major differences found at baseline are variations in training hours, gender differences and, to a lesser extent, age. These variables are included in the multivariate linear regression analysis. Note that LD-HS athletes were heavier and therefore had a larger BSA. Septal wall thickness (table 3) exceeded 12 mm in 10% of male athletes (1/27 LD-HS athletes, 17/57 HD-HS athletes), and 11 mm in 3% of female athletes (3/24 HD-HS athletes), representing the previously reported gender-specific cut-off values for the left ventricular hypertrophy (LVH).25 ,26 No one exceeded 16 mm septal wall thickness or 10 mm wall thickness beyond the septal wall.
Baseline characteristics
Quantitative ventricular parameters adjusted for body surface area
Table 3 lists the BSA-indexed ventricular volume, wall mass, dimensions and EF for both ventricles, as well as the ratio of LV EDV to LV EDM. This LV EDV/EDM ratio would differ from the non-athletes only if there was a selective increase of either the ventricular volume or the wall mass. There are three main findings that emerge as shown in table 3. First, a clear biventricular trend is revealed of increasing ventricular volume and wall mass from non-athletes to LD-HS athletes, HD-LS athletes and HD-HS athletes. Second, the increasing biventricular dilatation remains balanced with similar LV EDV/RV EDV ratios in all groups, ranging from 0.90 to 1.01. Third, the ratio of LV EDV to EDM was not different among non-athletic controls, LD-HS athletes and HD-LS athletes, and this would be smaller in LD-HS athletes if there was a selective ventricular wall thickening and larger in HD-LS athletes if there was a selective ventricular volume increase.Only HD-HS athletes showed a smaller LV EDV/EDM ratio (men 1.92/women 2.23) compared with the other groups (men 2.18 to 2.22/women 2.54 to 2.68), suggesting relatively more increase in the LV wall mass than in the LV volume.
To further illustrate these findings, the BSA-indexed LV EDV and LV EDM were plotted against each other as shown in figure 1. The HD-HS group shows the largest increases in LV EDV and LV EDM. The LD-HS and HD-LS groups are close to the line representing proportional increases of LV EDV and LV EDM, demonstrating a proportional, or balanced, increase of ventricular volume and wall mass. This is different from what would be expected under the Morganroth hypothesis, where the selective ventricular wall thickening in LD-HS athletes should have been observed with the LD-HS group deviating from the line of proportional increase towards the lower right corner of figure 1. This figure also illustrates that the LV EDV/EDM ratio in HD-HS athletes is lower than would be expected if LV EDV and LV EDM would increase proportionally.
BSA-indexed LV EDV and LV EDM by sport category. Bars represent 95% CI of the mean, which is represented by the centre of each cross; reference line represents the proportional increase of the left ventricular volume and wall mass (based on non-athletic group). HD-LS, high-dynamic low-static; HD-HS, high-dynamic high-static; LD-HS, low-dynamic high-static.
Multivariate linear regression model coefficients are listed in table 4. This analysis demonstrates that sport category is a highly significant contributor to changes in LV EDV and LV EDM, even when training hours, age and gender are also taken into account (p<0.01 for sport categories HD-LS and HD-HS). Age appears not to make a significant difference in this model (p=0.41 for LV EDV, p=0.86 for LV EDM). The training hours are retained as a relevant explanatory factor in the model, albeit with a coefficient of only 0.29 for LV EDV and 0.19 for LV EDM. Because this corresponds to the amount of change per training hour, the weekly training of 10 h would account for only 2.9 ml/m2 of LV EDV increase and 1.9 g/m2 of LV EDM increase. The coefficients for sport categories (ranging from 6 to 21 for LV EDV and from 0.75 to 17.1 for LV EDM) are considerably larger, representing a much greater difference corresponding to the sport category than to the number of training hours. As an illustration, in this model, an athlete's HD-HS status would predict an LV EDV increase of 21.3 ml/m2. The corresponding increase of LV EDV attributable to the number of training hours is much less, since the model coefficient indicates that every additional training hour will lead to an increase of only 0.29 ml/m2. In our study population, an HD-HS athlete would on average exercise for 19 h a week, which corresponds to an LV EDV increase of 5.5 ml/m2 (19×0.29). Even with much more training hours (40 h of weekly exercise would still only result in an increase of 12 ml/m2), the effect is still smaller than that of sport category (21.3 ml/m2).
Linear regression model coefficients
ECG parameters for all study groups are listed in table 5. Normal ECGs were only found in a minority of our study population (non-athletes/LD-HS/HD-LS/HD-HS combined men and women 33/29/18/10%, respectively). Group 1 ECG abnormalities, considered to be physiological and benign in athletes, are frequently present in athletic and non-athletic men and women (combined men and women per group, 66 to 89%). Group 2 ECG abnormalities are more frequent in athletes, and show a trend of increasing prevalence from non-athletes (2%) to LD-HS (5%), HD-LS (9%) and HD-HS (14%) in combined men and women. Early repolarisation (25 to 46%) and incomplete right bundle branch block (RBBB) (19 to 34%) seem to be fairly frequent in all groups. However, a pattern similar to the ventricular CMR parameters with higher prevalences in HD athletes can be noticed for sinus bradycardia (48-40-68-79%), first-degree atrioventricular (AV) block (2-2-6-9%), LVH (7-10-15-20%), T-wave abnormalities (0-5-9-14%) and ST-segment elevation (0-2-13-10%). Although the overall prevalence of group 1 and 2 ECG abnormalities was similar between male and female athletes, some ECG abnormalities are observed less often in women (male/female athletes: early repolarisation 42/14%, incomplete RBBB 36/19%, RBBB 5/0%, LVH 21/4%, first-degree AV block 9/0% and ST-segment elevation 15/1%).
ECG parameters
Discussion
This CMR study provides the physiological limits for the main sport categories for clinical reference purposes. All sport categories demonstrate a balanced increase of LV and RV volume. A balanced increase of the ventricular volume and wall mass is also observed in LD-HS and HD-LS sport categories, contradicting the hypothesis of separate adaptation patterns for LD-HS (‘strength-trained heart’) and HD-LS (more ‘endurance-trained heart’). Moreover, LD-HS sports appear to show remarkably little ventricular change. HD-HS athletes (endurance-strength athletes) have the greatest change in volume and mass with relatively more increase in LV wall mass than in volume.
The non-significant coefficient of LD-HS sport category in the regression models for LV EDM suggests that there is little difference between non-athletes and LD-HS athletes in the ventricular wall mass, confirming earlier studies that report little or no differences between sedentary controls and strength athletes.12 ,13 ,15 The small ventricular changes observed in high-static athletes are possibly due to the small amounts of dynamic exercise that many strength athletes perform as part of their training regime. It is unrealistic that an elite athlete's training regime would consist solely of high-static exercise. It should be noted that at inclusion, a majority of training hours (>75%) was required to be in the category to which an athlete was assigned.
There was a clear trend of larger ventricular volumes and wall mass from non-athletes to LD-HS, HD-LS and HD-HS athletes. Whereas predominantly HS sports seem to yield little cardiac adaptation, larger changes are observed in HD exercise. EDV and EDM of HD-HS sports exceed those of HD-LS sports and the LV EDV/EDM ratio is lower in HD-HS athletes. This suggests that there may be a synergistic effect of the volume load caused by high-dynamic exercise and the pressure load associated with high-static exercise, favouring an increase in the ventricular wall mass. The changes in LV EDV, LV EDM and LV/RV measures were proportional in LD-HS and HD-LS athletes and differed only in the extent of the changes rather than their pattern.
A similar trend of larger changes in HD athletes than in non-athletes and LD-HS athletes was observed in several ECG parameters: sinus bradycardia, first-degree AV block, LVH, first-degree AV block, T-wave abnormalities, ST-segment elevation. While bradycardia and first-degree AV block probably reflect the increased vagal tone, the higher prevalence of the latter three parameters in HD-HS athletes coincides with larger LV EDM and septal wall thickness, which exceeded the cut-off for LVH nearly exclusively in HD-HS athletes. Although the prevalence of group 1 and group 2 ECG abnormalities was similar between men and women, some parameters (early repolarisation, LVH, RBBB, first-degree AV block, and ST-segment elevation) were less common in women, as indicated in previous literature.27 It remains unclear if the presence of group 2 ECG abnormalities, which are considered to be uncommon and not related to training,20 represents pathology, since the athletes in our study did not report any clinical signs or symptoms and were considered to be healthy, all doing well at their respective sports.
Contribution of sport category, training hours, age and gender
The observed variation in the volume and wall mass is significantly determined by sport category, even when the number of training hours is taken into account using multivariate linear regression. This suggests that the increase of the ventricular wall mass and volume will be different for the same number of training hours of another sport category. At the same time, the number of training hours is also a statistically significant contributor to LV EDM and EDV, although the magnitude of the effect at realistic training levels is much smaller than that of sport category. There was no significant influence of age on the ventricular mass or on the ventricular volume in the multivariate model. As in other studies,18 ,28 ,29 male gender resulted in more pronounced increase of LV EDV and LV EDM, regardless of BSA.
Balanced adaptation in athletes
Recent studies, using echocardiography and CMR, are consistent with our findings of the proportional increase in LV volume and wall mass in athletes of LD-HS and HD-LS sport categories. This has been reported separately for football players30 and triathletes,31 as well as in a direct comparison of different sport categories, including LD-HS (power) athletes.4 However, the latter study also found a similar LV EDV/EDM ratio in HD-HS athletes. Dissimilar remodelling patterns were suggested in an echocardiographic study by Baggish et al, as correlations between LV volume and wall mass were present in endurance athletes while absent in strength-trained athletes. Nevertheless, the BSA-indexed LV diameter and wall mass were similar for strength- and endurance-trained athletes.10
The left-right balance in the cardiac adaptation pattern observed in our study has been reported in previous research32 and also by our own group.2 Although absolute RV volumes are consistently reported to be larger than those of the LV, La Gerche et al observed larger LV-RV differences in athletes on CMR, suggesting different LV and RV remodelling due to a disproportionate load on the RV.33 This difference turned out to be more pronounced during exercise, but was also present at rest. In a CMR study of 23 track runners by Perseghin et al, a disproportionately increased RV EDV was reported specifically in the sprint runners (n=14), whereas it was not observed in the marathon runners (n=9).34 While we were not able to demonstrate this difference at rest (see ratio measures as shown in table 3), we cannot rule out its existence during exercise.
What is already known on this topic
Physical exercise can result in changes in ventricular volume and wall mass. These changes seem to differ by sport category, but previous studies are conflicting on the pattern and the extent of changes for different sport categories.
What this paper adds
This paper demonstrates that the physiological cardiac adaptation is balanced for the left and right ventricles and the ventricular volume and wall mass in low-dynamic high-static and high-dynamic low-static sport categories and slightly skewed towards relatively more ventricular wall mass increase in combined high-dynamic high-static sports. Furthermore, sport category is an important determinant of cardiac adaptation independent from age, gender and the number of training hours.
The apparent balanced cardiac adaptation in athletes could be used in clinical practice to differentiate the physiological adaption from pathology, such as hypertrophic cardiomyopathy (HCM) by using the LV EDV/EDM ratio35 and arrhythmogenic right ventricular cardiomyopathy/dysplasia (ARVC/D) using the LV/RV volume ratio.36
Study limitations
Although the total sample size is relatively large, taking into account multiple covariates in regression analysis requires sufficient degrees of freedom. We therefore deliberately decided a priori to analyse the data using groups categorised by dynamic and static components, yielding sufficiently large group sizes. The numbers of subjects per individual sport would not allow for additional sport-specific tests in all sport categories. We considered this the best approach to ensure maximum validity and generalisability. Multivariate regression analysis makes the optimal use of the data available, and also facilitates the investigation of gender influence, despite smaller numbers of female athletes. This study cannot address the ethnic variation as the vast majority (96%) of the study population was Caucasian. As greater LV mass and wall thickness have been described in black athletes,25 ,26 ,37 ethnic variation should be further investigated. The limited ethnic variation and small female sample size in some sport categories may require additional testing to improve the clinical prediction of the extent of cardiac adaptation for any given gender, sport category and the number of training hours.
BSA of strength athletes was considerably different from that of the other groups. We used BSA-indexed outcomes and we do not know if the actual body composition would influence the observed results. Fat-free mass has been proposed as a more reliable body size scalar for cardiac dimensions.38 However, since LV EDV and LV EDM are indexed to the same value (either BSA or any other body composition variable), it does not affect the volume to mass ratio and therefore would not alter the findings of balanced adaptation.
Conclusion
Sport category has more effect on cardiac adaptation than on training hours, gender and age. HD-HS sports show the largest increase in the ventricular volume and wall mass, whereas LD-HS sports show no significant differences to non-athletes. Balanced cardiac adaptation is seen in LD-HS and HD-LS sports, with preserved ratios of LV volume/LV wall mass and LV/RV volume. LV/RV volume remains balanced in HD-HS sports, but LV wall mass increases relatively more than the volume. This study's results may serve as a reference in clinical practice when differentiating physiological adaption from pathology.
References
Footnotes
-
Competing interests None.
-
Ethics approval Medical-Ethical Comittee of the University Medical Center Utrecht.
-
Provenance and peer review Not commissioned; externally peer reviewed.