Untargeted-metabolomics identifies key metabolites for atrial fibrosis and recurrence of atrial fibrillation indication across populations

Introduction Atrial fibrosis serves as a key substrate for both atrial cardiomyopathy and atrial fibrillation (AF).Cardiomyocyte injury triggers the proliferation and differentiation of interstitial cells, such as fibroblasts, which subsequently secrete extracellular matrix components like collagen. Objective This study developed a predictive model combining plasma metabolomics with clinical data, followed by experimental validation of the fibro-genic functions of identified metabolites. Methods Untargeted metabolomics and bioinformatics analyses were conducted on plasma samples from 254 AF patients at a Province Hospital.Predictive models were constructed using machine learning algorithms to analyze: 1) metabolites alone, 2) clinical parameters alone, and 3) integrated metabolite-clinical combinations for fibrosis detection and recurrence prediction.Prospective validation was performed on 103 patients from three hospitals in Eastern China.Functional validation employed optical mapping and Western blot to investigate the effects of suberic acid on cardiomyocytes and fibroblasts. Results Metabolomics analysis identified 12 differential fatty acid metabolites, achieving ROC-AUC values of 0.83/0.68 in the training/validation sets for atrial fibrosis prediction.Clinical parameters (age and AF type) yielded ROC-AUC values of 0.79/0.67.The LASSO-integrated model (combining 5 metabolites and 2 clinical factors) demonstrated superior performance with ROC-AUC values of 0.85/0.85, supported by decision curve analysis confirming its clinical utility.For recurrence prediction, suberic acid (OR = 3.56), Succinimide (OR = 0.67), and age (OR = 1.06) achieved an ROC of 0.73.External validation provided ROC values of 0.79 for atrial fibrosis detection and 0.83 for recurrence prediction.Experimental validation confirmed that suberic acid disrupts cardiomyocyte calcium handling and enhances fibroblast collagen synthesis, as revealed by optical mapping. Conclusion In conclusion, integrating untargeted plasma metabolomics from AF patient cohorts (Discovery, n = 251; Validation, n = 103) with clinical variables resulted in a robust LASSO-logistic regression model capable of predicting both underlying atrial fibrosis and the risk of AF recurrence.Functional studies further identified SA as a key metabolite contributing directly to arrhythmogenesis, promoting pro-fibrotic fibroblast differentiation and disrupting cardiomyocyte calcium handling and electrical conduction.These findings position SA not only as a novel biomarker but also as a potential mechanistic driver of AF.Plasma metabolic biomarkers offer a non-invasive method for assessing atrial fibrosis and can aid in selecting patient-specific ablation strategies and upstream treatments.