Proteomic predictors of arrhythmic risk in hypertrophic cardiomyopathy
Y Zhao, H Cui, J SongAbstract
Background
Hypertrophic cardiomyopathy (HCM) is genetically heterogeneous with variable clinical outcomes. Postoperative arrhythmia (atrial fibrillation/ventricular tachycardia) remains challenging to predict using clinical variables alone, and the molecular determinants underlying arrhythmic risk are incompletely understood.
Purpose
To define genotype-associated myocardial proteomic remodeling in HCM and to determine whether integrating proteomic markers with clinical parameters improves prediction of postoperative arrhythmia.
Methods
Interventricular septal tissue from 175 HCM patients undergoing septal myectomy (MYH7-mutant, MYBPC3-mutant, genotype-negative) and 20 healthy controls was profiled by Astral-DIA mass spectrometry. Differential expression, gene set enrichment, and WGCNA were used to identify clinically relevant proteomic programs. For arrhythmia prediction, 7,417 proteins and 60 clinical variables were analyzed with a train/test split (n=122/53). Candidate features were selected by LASSO and logistic regression (intersection retained) and entered into multivariable logistic regression models (clinical-only vs integrated); performance was assessed by AUC and visualized by a nomogram.
Results
All HCM subtypes exhibited extracellular matrix remodeling with genotype-linked differences: MYBPC3 hearts (greater IVS thickness) were enriched for elastic fiber pathways, whereas genotype-negative cases (greater LVPW thickness) showed activation of coagulation-related proteins. WGCNA identified 14 modules associated with clinical traits, including immune-related proteins linked to LVPW thickening and proteostasis/cytoskeletal remodeling correlated with IVS hypertrophy and disarray. For postoperative arrhythmia prediction, a model including pre-operative arrhythmia, age, PAD, eGFR plus six proteins (BCAS2, PI4KB, PLEKHA2, RBM22, SRPK1, TOE1) achieved AUC 0.867/0.794 (train/test). An integrated clinical–proteomic model improved performance to 0.888/0.856, indicating added predictive value from proteomic features. The selected proteins implicate RNA processing, kinase signaling, and stress-response pathways in arrhythmic susceptibility.
Conclusion
Myocardial proteomics reveals genotype- and phenotype-associated remodeling programs in HCM and, critically, improves prediction of postoperative arrhythmia when combined with clinical variables. This integrated framework supports more precise risk stratification and nominates mechanistically relevant protein biomarkers for arrhythmic complications.