Multimodal Magnetic Resonance Imaging and Machine Learning Uncovers Distinct Progression Patterns in Friedreich Ataxia.
Authors
Affiliations (11)
Affiliations (11)
- School of Psychological Sciences, The Turner Institute for Brain and Mental Health, Monash University, Clayton, Victoria, Australia.
- School of Translational Medicine, Monash University, Melbourne, Victoria, Australia.
- Bruce Lefroy Centre for Genetic Health Research, Murdoch Children's Research Institute, Parkville, Victoria, Australia.
- Department of Paediatrics, University of Melbourne, Parkville, Victoria, Australia.
- Balance Disorders and Ataxia Service, Royal Victorian Eye and Ear Hospital, East Melbourne, Victoria, Australia.
- Bionics Department, University of Melbourne, Melbourne, Victoria, Australia.
- NeuroMovement Laboratory, Bionics Institute, East Melbourne, Victoria, Australia.
- Neurology Department, Alfred Hospital, Monash University, Prahran, Victoria, Australia.
- QIMR Berghofer Medical Research Institute, Brisbane, Queensland, Australia.
- Department of Neurology, University of Campinas, Campinas, Brazil.
- Department of Neurosciences, School of Medicine, University of São Paulo at Ribeirão Preto (USP-RP), Ribeirão Preto, Brazil.
Abstract
Friedreich ataxia (FRDA) is a rare neurodegenerative disorder with heterogenous clinical progression, complicating prognosis and trial design. Neuroimaging offers objective biomarkers of disease progression, yet variability in progression patterns remains poorly understood. The objective of this study is to identify distinct neurodegenerative progression patterns in FRDA using longitudinal multimodal magnetic resonance imaging (MRI) and to evaluate associations with clinical, demographic, and genetic factors. Longitudinal structural and diffusion MRI data from 54 patients with FRDA and 57 controls were analyzed. Annualized progression rates of macrostructural (volumetric) and microstructural (diffusion) features across cerebellum, brainstem, and spinal cord regions were clustered using Gaussian Mixture Models. Following model selection, clusters were evaluated for biological plausibility of longitudinal imaging trajectories, bootstrap and subsampling reproducibility, and consistency of case-control composition. Associations with demographic, genetic, and clinical variables were examined, and Random Forest modeling assessed predictors of cluster membership. Four statistical clusters were identified, three of which represented robust, biologically interpretable progression patterns characterized by predominant microstructural degeneration, predominant macrostructural atrophy, and minimal measurable progression. The microstructure- and macrostructure-dominant patterns were enriched for FRDA participants, whereas the minimal-progression pattern contained more controls. GAA1 repeat length was the only variable consistently associated with cluster membership, with larger expansions observed in the microstructure-dominant pattern. Disease duration and clinical progression rates did not significantly differentiate progression patterns. Longitudinal multimodal MRI shows distinct neurodegenerative progression patterns in FRDA that are not fully captured by conventional clinical measures. This data-driven framework provides a basis for investigating imaging-derived disease heterogeneity and its potential relevance to participant stratification in clinical trials. © 2026 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.