Associations between echocardiographic parameters and aerobic performance in young athletes

Cardiac morphology can be shaped according to chronic and acute exercise and provides important findings regarding cardiac response to different types of performance. However, cardiac remodelling is a multidimensional process involving intertwined structural, functional, and haemodynamic components, making it difficult to explain with a single criterion and requiring a comprehensive assessment. The aim of this study is to investigate the effects of echocardiographic (ECHO) parameters on 30-15 interval fitness test (IFT) performance.

Twenty-three active athletes participated in the study. All ECHO measurements were performed by a specialist cardiologist using standard clinical methods, and the IFT was administered to participants after the measurements. In the analysis, ECHO variables representing the structural characteristics of the left ventricle, systolic-diastolic function, and aortic/atrial measurements were evaluated, with priority given to parameters indexed according to body surface area (BSA). Relationships between IFT performance and ECHO parameters were examined by establishing parsimonious linear regression models appropriate for small sample sizes; models were evaluated using Akaike Information Criterion (AICc)-based model comparison and model averaging, results were checked for robustness using leave-one-out cross-validation (LOOCV), and the SHapley Additive exPlanations (SHAP) approach, consistent with the model, was used for interpretability.

In the AICc-based model comparison, the parsimonious model with the highest support consisted of the body composition (BMI) + posterior wall (PW)_I + aortic strain (AORT_STR) components and explained approximately 46% of the variance in IFT performance (mean adj. R ² = 0.461; LOOCV root mean squared error (RMSE) = 1.811; mean ΔAICc = 0; mean Akaike weight = 0.0476). Variable importance analyses in the best model set revealed that the variables most consistently associated with IFT were BMI (Σw = 0.46; weighted mean-SHAP- = 0.154; β =  - 0.068) and posterior wall thickness indexed by BSA (PW_I) (Σw = 0.35; weighted mean-SHAP- = 0.211; β = 0.348). This was followed by IVS_I (Σw = 0.191; weighted mean-SHAP- = 0.072; β = 0.124), A wave (Σw = 0.177; weighted mean-SHAP- = 0.080; β = 0.849) and AORT_STR (Σw = 0.166; weighted mean-SHAP- = 0.077; β = 0.013) followed. Overall, the distribution of model weights across multiple models indicated significant model uncertainty, while BMI and indexed structural measures provided the strongest and most consistent signal in relation to IFT.

In this context, it has been concluded that body composition is also a determining factor in the relationship between cardiac morphology and IFT performance. Large prospective studies are important for determining this complex relationship.