Reference values of left heart echocardiographic dimensions and mass in male peri-pubertal athletes

Abstract

Background

Several articles have proposed reference values in healthy paediatric subjects, but none of them has evaluated a large population of healthy trained adolescents.

Design

The study purpose was to establish normal echocardiographic measurements of left heart (aortic root, left atrium and left ventricular dimensions and mass) in relation to age, weight, height, body mass index, body surface area and training hours in this specific population.

Methods

We retrospectively evaluated 2151 consecutive, healthy, peri-pubertal athletes (100% male, mean age 12.4 ± 1.4 years, range 8–18) referred to a single centre for pre-participation screening. All participants were young soccer athletes who trained for a mean of 7.2 ± 1.1 h per week.

Results

Left ventricular internal diameters, wall thickness, left ventricular mass, aortic root and left atrium diameters were significantly correlated to age, body surface area, height and weight (p < 0.01). Age, height, weight and body surface area were found associated with chamber size, while body mass index and training hours were not. Inclusion of both age and body size parameters in the statistical models resulted in improved overall explained variance for diameters and left ventricular mass.

Conclusion

Equations, mean values and percentile charts for the different age groups may be useful as reference data in efficiently assessing left ventricular parameters in young athletes.

Introduction

The physiological, morphologic and functional changes induced in the heart as a result of regular physical training and sport activities are less evident in teens and young junior athletes, as they are less physically mature and exposed to fewer years of intense training¹ compared with adults. The goal of the pre-participation screening (PPS) is to differentiate between physiological adaptive changes and an initial pathology. Transthoracic echocardiography has become the standard non-invasive imaging modality in screening only for presence of clinically suspicious anomalies, uncommon and training-unrelated electrocardiogram (ECG) abnormalities and on discovery of a positive family history.^2,3 The correct interpretation of data strongly relies on the available echocardiographic reference values, but paradoxically the latter are derived from adult series or refer to small paediatric populations, with high variability in the methodology. The lack of robust standardized nomograms based on a large set of healthy young athletes may hamper the accuracy of estimating the presence of cardiac abnormalities, or introduce a bias in the PPS decision-making process.⁴ Moreover, standardized methodologies in performing normalization for Z-score calculation, especially in the presence of heteroscedasticity and non-linear relationships, are lacking.⁵ Recently, Z-scores have been published exclusively for normalized pulse wave Doppler and tissue Doppler imaging in paediatric echocardiography.⁶ We aimed to provide echocardiographic left heart reference values and Z-score equations to better define the physiologic limits of left heart remodelling in a large and homogenous population of trained peri-pubertal athletes and compare our data with the existing literature that evaluated untrained subjects.

Methods

Study design

This retrospective study includes all the consecutive junior soccer players (8–18 years old, trained for at least nine months, 100% male, all Caucasian) evaluated for PPS at the Sports Medicine Institute of Rome between January 2008 and March 2009. The soccer players were engaged in training 7.3 ± 1.2 h per week, including warm-up, technical/tactical skills, aerobic training reaching at least 75% of the maximal heart rate, strength training and cool down. All athletes underwent a conventional Italian PPS, which by law must be performed every 12 months in the case of competitive athletes, and includes ECG, medical history and physical examination. In addition, all athletes underwent a complete echocardiographic study, as previously described in two papers evaluating the overall study population.^7,8 All athletes with a normal echocardiographic evaluation were included. Exclusion criteria were: the presence of any abnormal echocardiographic findings, including any form of cardiomyopathy, bicuspid aortic valve, mitral valve prolapse, inter-atrial septal defect, and even minor defects such as patent foramen ovale. We also excluded children with a body mass index (BMI) for age >2 (obese subjects, based on World Health Organization charts) because the presence of a high level of fat mass in adolescents can introduce a bias.⁹ This investigation complied with the international guidelines related to patient confidentiality and research ethics.¹⁰ Parents signed a written informed consent at the moment of PPS and the Institutional Review Board approved the study.

Echocardiography

All consecutive subjects underwent a conventional two-dimensional (2D) Doppler transthoracic echocardiography, performed with a commercially available machine (Siemens Acuson X300) equipped with a 5.1 and an 8.4 MHz probe. M-mode, 2D echocardiographic data, as well as pulsed, continuous and colour Doppler flow mapping were digitally recorded for off-line analysis and reviewed by an experienced cardiologist with a strong expertise in sport cardiology and echocardiography. The parameters of left ventricular (LV) geometry were measured off-line in 2D, including interventricular septum (IVS), posterior wall thickness (LVPW) and LV end-diastolic (LVEDD), end-systolic (LVESD) diameters and linear fractional shortening. A combination of long-axis and short-axis was used to ascertain that the short-axis or minor-axis measurement was perpendicular to and crossed the mid-points of the ventricular septum and posterior wall and that the LV short-axis geometry was circular throughout the cardiac cycle.¹¹ LV mass (LVM) was calculated using the corrected American Society of Echocardiography convention¹¹ proposed by Devereux: LV mass (LVM, g) = 0.8 × 1.04 × ((LVEDD + LVPW + IVS)³− LVEDD³) + 0.6. Fractional shortening was calculated with the assumption of a cylindrical shaped LV, using the conventional formula [(LVEDD – LVESD)/LVEDD]*100. Fractional shortening is related to the effect of age. For a given level of end-systolic stress, fractional shortening decreased by 0.26%/year in children and adolescents,¹² so we decided to test this variable instead of LV ejection fraction, which does not show the same relationship with age.

Height and weight were measured to the nearest 0.1 cm and 0.1 kg using a wall-mounted stadiometer and a digital scale, respectively. BMI was calculated as weight (kg) divided by height² (m²). Body surface area (BSA) was calculated applying the Haycock formula¹³ as it is recognized as the most appropriate for describing the growth of cardiovascular structures.¹⁴

Statistical analysis

Clinical, demographic, and echocardiographic data are presented as mean ± SD for continuous variables and as absolute numbers and percentages for categorical variables. The degree of association between left heart measurements and age, echocardiographic and anthropometric data was quantified by means of Pearson’s correlation coefficient (r²). Allometric normative equations for diameters were obtained using linear regression, univariate or multiple using stepwise forward approach, after log-transforming dependent (diameters) and independent (age, BSA, BMI, height, weight and training) variables. A set of nomograms displaying the upper limits of normal for each of the left heart measurements was obtained by adding two SDs to the mean predicted diameter. Standardized Z-score equations were defined based on the regression coefficients and the mean square error.

For the sake of clarity, results and graphs referring to age have been presented coupling consecutive years, although age has been considered as a continuous variable in all the performed statistical analyses.

Variables measured in different age groups were compared using analysis of variance test with post-hoc Bonferroni correction for multiple comparisons.

To assess the reproducibility of the echocardiographic measurements, image analysis was repeated in 20 randomly selected study subjects by an experienced cardiologist (author EC) and then repeated two weeks after. Inter- and intra-observer variability was expressed as the coefficient of variation, defined as the absolute difference of the corresponding pair of repeated measurements in percentage of their mean in each subject and then averaged over the entire group.

p-values < 0.05 were considered significant. Statistical analysis was performed using SPSS version 17.0 (SPSS, Chicago, IL, USA).

Results

Anthropometric and echocardiographic data

Among the 2261 subjects initially screened in a previous study,⁸ 2151 (95%) subjects were included in the study. The remaining 110 subjects were excluded from the study because of any abnormal echocardiographic findings, BMI for age >2 or incomplete study. Table 1 presents the demographic and anthropometric data for participants included in the final analysis based on age distribution. In Table 2, the left heart echocardiographic parameters included are presented, based on age range. The distribution of the same data is depicted in Figure 1, and suggests a positive relation between age and LVEDD, LVESD, LVM, aortic root (AoR) and left atrial (LA) diameter. Conversely, only a weak association could be qualitatively assessed between age and IVS, LVPW or fractional shortening.

Relationship between anthropometric and echocardiographic data

Table 3 provides the results of the univariate analysis between echocardiographic measurements and the anthropometric/demographic variables, separately for raw data (top) and after log-transforming both the dependent and independent variables. The allometric approach slightly improved the variance explained by the models. BSA displayed a better association with cardiac measurements, more so than height or weight and particularly BMI, while training was not significantly associated with any of the dependent variables. Fractional shortening was not associated with any of the anthropometric measurements, and only a weak inverse relation with age was found, comparable to the existing literature.¹²

Allometric models, obtained using BSA and age, are described in Table 3. Model fitting was good particularly for LVEDD, LVESD, AoR and LVM (explained variance ranging from 37% to 67%), with a positive association between the echocardiographic measurements and both BSA and age. Weaker associations were found for ventricular wall thickness (IVS and LVPW) and LA diameter (LAD), but only with BSA and not age. The allometric model was not capable of describing the variance observed within the fractional shortening measures.

Normative equations and Z-scores

Normative equations on the basis of model coefficients for allometric scaling are described in the footnote to Table 4. The upper limits of echocardiographic measurements as a function of BSA and age are shown in Figure 2. For IVS, LVPW, LAD and fractional shortening the upper limits curves are independent of age, consistently with the results in Table 2.

Standardized Z-scores can be calculated using the equations listed in Table 5. For instance, in a 14-year-old subject, with a BSA = 1.60 and LVEDD = 56 mm, the relevant Z-score should be calculated as the difference between the measured and the expected values, normalized by the mean square error: (ln(56) – (3.52 + 0.07·ln(14) + 0.35 ln(1.60)))/0.11, resulting in Z = 1.42, which is not indicative for LV dilation (Z-score < 2). The same case is graphically shown in Figure 3. For a given BSA the upper limit of LVEDD (Z-score = 2) can be read from this nomogram and, in this case of a non-dilated LVEDD, the diameter falls under the nomogram line.

Comparison of our allometric models with existing literature

Figure 4 depicts a qualitative comparison between the allometric models described in Table 4 and those described by Kampmann et al.¹⁵ for LV diameters, and Kaldarova et al.¹⁶ for AoR, in a similar cohort of subjects. In all cases, only BSA, and not age, was entered as an independent variable. A good agreement between the allometric models was found for both LVEDD and LVESD, with an over-estimation by the model from the literature for the youngest patients with the largest BSA, and under-estimation in older patients with reduced BSA. As far as the AoR is concerned, the association here reported with BSA was much weaker than that reported in the literature, leading to non-negligible under- and over-estimation in subjects with higher or lower BSA, respectively.

Inter-observer variability was 2.2% for LVEDD, 3.9% for LVESD, 3.9% for AoR, 5.1% for LAD, 6.5% for IVS and 3.1% for LVPW. Intra-observer variability was 1.69% for LVEDD, 2.1% for LVESD, 1.46% for AoR, 1.42% for LAD, 3.3% for IVS and 3.2% for LVPW. Importantly, both the inter- and intra-observer variability were well below 10%.

Discussion

The importance of reference values and Z-score

We have presented the reference values for a set of left heart parameters in a large population of healthy peri-pubertal athletes, and the association between this set of cardiac measurements and anthropometric variables in this unique study population. The nomograms allow determination of the Z-score ≤ 2 (≤95th percentile) as a function of age and BSA. The Z-score of a measurement is the number of standard deviations of that value from the mean value. If a measurement is equal to the population mean, the Z-score is 0; a Z-score of +2 or −2 corresponds to the 95th and the 5th percentile, respectively (i.e. two standard deviations above or below the mean).⁴ The main benefit of using Z-scores has been the absence of any dependence on a predetermined association between the size of a structure and BSA¹⁷ and Z-scores are particularly useful in tracking allometric growth over time.¹⁸ This new set of nomograms is helpful to sport medicine clinicians in the recognition of mild LV abnormalities and the quantification of their progression over time in athletes aged ≤18 years as Z-score = 2 is easily visually assessed. Nomograms for left heart dimensions and wall thickness are crucial for diagnostic purpose at PPS to better differentiate the normal growth and physiological adaptation to physical training from pathologic abnormalities.

Limitation of the previously proposed reference values

Based on the criticisms of the previous reference values as highlighted by Cantinotti et al.⁴ and Mawad et al.,⁵ the published normal ranges are affected by scarce consideration of confounders, such as gender, and discrepant normalization of data to account for the effects of body size and age. These limits were overcome as the reference population included a large sample size of healthy males with a 10-year age range and similar training characteristics; furthermore all echocardiographic measurements were standardized according to the ASE recommendations and Z-scores were defined. Our normative data may be useful for diagnosis at PPS and in the follow-up of athletes who fall into a grey zone between athlete’s heart and an initial cardiac abnormality.

The peri-pubertal athlete versus non-athlete population

The comparison between our population and a non-athlete population, estimated by the equation generated by previous published articles,^15,16 advocated that in this age range physical training does not largely affect the size of the left ventricle, in terms of LVEDD and LVESD, as measured by 2D echocardiography. The suboptimal overlap in the normative surfaces shown in Figure 4 can be explained by the lack of anthropometric adjustments in the normative equations retrieved from the literature, leading to a large mismatch for the most extreme body size. Conversely, in the case of AoR, the discrepancy between the two models was apparent. This could have been due to the different population used to build the models, but also could be an effect of training. Specifically designed studies are needed to corroborate this hypothesis. Molecular, morphologic and functional cardiac adaptations to athletic training are known as athlete’s heart. While moderate exercise can be mostly beneficial, intense exercise induces molecular modifications that could even be detrimental and translated in a functional decline in systolic and diastolic cardiac function.¹⁹ An exaggerated reactive oxygen species production has been linked to this effect of intense training, but an inter-individual variability exists in the redox imbalance that can be counterweighted by antioxidants intake. This can be morphologically and functionally translated in an inter-individual variability of the acute right ventricle (RV) impairment evaluated by speckle-tracking echocardiography after a trail-running race, independent of prior training.²⁰ While adult athlete’s heart has been extensively studied with different imaging modalities,^21,22 the paediatric physiological adaptation to intensive training is less evident and some studies failed to identify clinical features of athlete’s heart in prepubertal subjects,²³ while another reported a significant increase only in LV and LA dimensions and AoR diameter.²⁴ Among different sports, endurance training influences the growing heart of male preadolescent athletes with an addictive increase in RV dimensions and in biatrial volumes, with a preserved RV and biatrial function, thus enforcing the hypothesis of a physiological adaptation to physical activity.^25,26 Gender difference exists in athlete’s heart, with the female sex showing lower left ventricular mass and thicknesses compared with males, without geometrical differences in terms of LV remodelling.²⁷ Despite these different reports, a recent meta-analysis²⁸ focused on electrical and structural adaptations of the paediatric athlete’s heart, concluding that paediatric athletes had significantly greater LV diameters and thickness, LVM and LAD. Even if the age was a positive predictor of those parameters, the magnitude, prevalence and distribution of such changes are dependent on the chronological age of the paediatric athlete.²⁸ No specific values for age class or BSA were available, nor nomograms to accurately derive Z-score values. Therefore, our LV nomograms could be used to evaluate peri-pubertal children undergoing PPS. Due to the noteworthy amount of training per week (see Table 1) and to the specific training of soccer players, caution should be applied when using these nomograms to evaluate athletes practising sports other than soccer, or female athletes or non-Caucasian athletes. Independently from the Z-score value obtained for a specific parameter, the sports physician must take care of other parameters, such as physical examination, family history and ECG findings, and evaluate whether further evaluation are required to investigate for pathologic cardiovascular disorders associated with sudden cardiac death in athletes. Furthermore, echocardiography as a diagnostic tool has limitations in sensitivity and specificity, in particular inter-observer variability among physicians remains a major concern and can hamper results. The actual innovations in imaging technology, including three-dimensional echocardiography, speckle tracking echocardiography and cardiac magnetic resonance, have enhanced the diagnostic capabilities of the current imaging modalities and made possible the correct identification of the athlete’s heart grey zone.^29,30

Potential limitations

The generalizability of our results may be limited by the homogeneity of the study population with respect to sex, ethnicity and practised sports. The lack of peri-pubertal female athletes, non-Caucasian athletes and athletes practising other sports except soccer represent a relevant limitation of this study. The cardiovascular impact of soccer is a mix of high dynamic and moderate static demand. Although these findings cannot be applied in sports different from soccer, the current data can give a general idea of the degree of remodelling also in other sports.

Also, we decided to focus only on conventional parameters of the left heart that are always evaluated in a standard echocardiography; still tissue Doppler and speckle tracking imaging are promising tools in athlete’s heart evaluation, but are not universally adopted in PPS evaluation.^30,31 Other authors have already published normal values of AoR diameter,³² pulse wave Doppler and tissue Doppler imaging⁶ in paediatric populations; however, to our knowledge no such nomograms exist in trained peri-pubertal athletes. The evaluation of the right heart will be the object of other ongoing research in a sub-group of this study population, where different RV parameters were assessed. The effect of training could only be evaluated on the basis of normative equations, obtained in large sets of peri-pubertal athletes and non-athletes, given the lack of a group of matched controls. The stage of athletes’ puberty has not been assessed by a proper scale, such as the Tanner scale, therefore the term ‘peri-pubertal’ covers subjects in a heterogeneous age range of from 8 to 18 years.

Conclusion

In this study we present reference values for left heart measurements in peri-pubertal athletes with Z-score for normalized data. We are confident that we have successfully reduced the influence of age and body surface area and this set of nomograms will ease the interpretation of 2D echocardiography for the sport medicine physician at PPS. These data should be further integrated with other important factors such as ECG data, family history, ventricular morphology and symptoms in order to clear the athlete for PPS.

Author contribution

EC, FM, LC and FP conceived the study, participated in data collection, analysis and manuscript drafting, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. FM performed the statistical analysis, participated in the manuscript drafting and revised it critically. FS, EG, AN, FQ, MM, MR and CF participated in acquisition of data, analysis, interpretation and manuscript drafting, revising it critically for key intellectual content. All authors read and approved the final version of the study. LC and FP are joint last authors.

Acknowledgements

The authors are extremely grateful to Dr Elaine Tyndall for her critical reading. This study has been approved by the local institutional review board. The authors declare that the data of the research article are original.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by unrestricted grant from Villa Stuart Sport Clinic, FIFA Medical Centre of Excellence and partially funded by Sapienza University of Rome (grant number 2016/43780) to Elena Cavarretta.

References