Arrhythmic Mitral Valve Prolapse Phenotype: An Unsupervised Machine Learning Analysis Using a Multicenter Cardiac MRI Registry
Ralph Kwame Akyea,Stefano Figliozzi,Pedro M Lopes,Klemens B Bauer,Sara Moura-Ferreira,Lara Tondi,Saima Mushtaq,Stefano Censi,Anna Giulia Pavon,Ilaria Bassi,Laura Galian-Gay,Arco J Teske,Federico Biondi,Domenico Filomena,Vasileios Stylianidis,Camilla Torlasco,Denisa Muraru,Pierre Monney,Giuseppina Quattrocchi,Viviana Maestrini,Luciano Agati,Lorenzo Monti,Patrizia Pedrotti,Bert Vandenberk,Angelo Squeri,Massimo Lombardi,António M Ferreira,Juerg Schwitter,Giovanni Donato Aquaro,Gianluca Pontone,Amedeo Chiribiri,José F Rodríguez Palomares,Ali Yilmaz,Daniele Andreini,Anca-Rezeda Florian,Marco Francone,Tim Leiner,João Abecasis,Luigi Paolo Badano,Jan Bogaert,Georgios Georgiopoulos,Pier-Giorgio Masci
DOI: https://doi.org/10.1148/ryct.230247
Abstract:Purpose To use unsupervised machine learning to identify phenotypic clusters with increased risk of arrhythmic mitral valve prolapse (MVP). Materials and Methods This retrospective study included patients with MVP without hemodynamically significant mitral regurgitation or left ventricular (LV) dysfunction undergoing late gadolinium enhancement (LGE) cardiac MRI between October 2007 and June 2020 in 15 European tertiary centers. The study end point was a composite of sustained ventricular tachycardia, (aborted) sudden cardiac death, or unexplained syncope. Unsupervised data-driven hierarchical k-mean algorithm was utilized to identify phenotypic clusters. The association between clusters and the study end point was assessed by Cox proportional hazards model. Results A total of 474 patients (mean age, 47 years ± 16 [SD]; 244 female, 230 male) with two phenotypic clusters were identified. Patients in cluster 2 (199 of 474, 42%) had more severe mitral valve degeneration (ie, bileaflet MVP and leaflet displacement), left and right heart chamber remodeling, and myocardial fibrosis as assessed with LGE cardiac MRI than those in cluster 1. Demographic and clinical features (ie, symptoms, arrhythmias at Holter monitoring) had negligible contribution in differentiating the two clusters. Compared with cluster 1, the risk of developing the study end point over a median follow-up of 39 months was significantly higher in cluster 2 patients (hazard ratio: 3.79 [95% CI: 1.19, 12.12], P = .02) after adjustment for LGE extent. Conclusion Among patients with MVP without significant mitral regurgitation or LV dysfunction, unsupervised machine learning enabled the identification of two phenotypic clusters with distinct arrhythmic outcomes based primarily on cardiac MRI features. These results encourage the use of in-depth imaging-based phenotyping for implementing arrhythmic risk prediction in MVP. Keywords: MR Imaging, Cardiac, Cardiac MRI, Mitral Valve Prolapse, Cluster Analysis, Ventricular Arrhythmia, Sudden Cardiac Death, Unsupervised Machine Learning Supplemental material is available for this article. © RSNA, 2024.