Non-specific myocardial fibrosis in young competitive athletes: clinical significance and risk prediction by a powerful machine learning-based model

E. Androulakis,S. Marwaha,N. Dikaros,R. Bhatia,H. MacLachlan,S. Fyazz,N. Chatrath,A. Merghani,G. Finocchiaro,S. Sharma,M. Papadakis

DOI: https://doi.org/10.1007/s00392-024-02550-y

IF: 6.138

2024-10-15

Clinical Research in Cardiology

Abstract:Background Non-specific myocardial fibrosis (NSMF) is a heterogeneous entity. We aimed to evaluate young athletes with and without NSMF to establish potentially clinically significance. Methods We analysed data from 328 young athletes. We identified 61 with NSMF and compared them with 75 matched controls. Athletes with NSMF were divided into Group 1 ( n = 28) with 'minor' fibrosis and Group 2 ( n = 33) with non-insertion point fibrosis, defined as 'major'. Athletes were followed-up for adverse events. Finally, we tested various machine learning (ML) algorithms to create a prediction model for 'major' fibrosis. We created 4 different classifiers. Results Athletes of black ethnicity were more likely to have a subepicardial pattern (OR: 5.0, p = 0.004). Athletes with 'major' fibrosis demonstrated a higher prevalence of lateral T-wave inversion (TWI) ( 500/24 h, p = 0.046; non-sustained VT, p = 0.043). Athletes with 'minor' fibrosis demonstrated higher right ventricular volumes ( p = 0.013), maximum Watts ( p = 0.022) and maximum VO 2 ( p = 0.005). Lateral TWI ( p = 0.026) and VO 2 < 44 mL/min/Kg ( p = 0.040) remained the only significant predictors for 'major' fibrosis. During follow up, athletes with 'major' fibrosis were 9.1 times more likely to exhibit adverse events (OR 13.4, p = 0.041). All ML models outperformed the benchmark method in predicting significant MF, best accuracy achieved by the random forest classifier (90%). Conclusions Lateral TWI and reduced exercise performance are associated with higher burden of fibrosis. Fibrosis was associated with increased ventricular arrhythmia and adverse events. A comprehensive assessment can help develop a ML-based model for significant fibrosis, which could also guide clinical practice and appropriate CMR referrals.

cardiac & cardiovascular systems