Identification of Biological Subtypes of Friedreich Ataxia with Structural MRI-based Machine Learning

Area of Science:

• Neuroimaging
• Neurodegenerative Diseases
• Biostatistics

Background:

• Friedreich ataxia (FRDA) is a progressive inherited neurodegenerative disease with variable progression rates.
• Understanding FRDA heterogeneity is crucial for clinical trial design and effective treatments.
• Previous studies lacked subgroup identification for FRDA based on neuroimaging patterns.

Purpose of the Study:

• To stratify Friedreich ataxia (FRDA) participants using structural MRI and the Subtype and Stage Inference (SuStaIn) algorithm.
• To determine the biological and clinical relevance of identified FRDA subgroups.
• To investigate MRI-based patterns of neurodegeneration in FRDA.

Main Methods:

• Utilized structural MRI and clinical-demographic data from 565 participants (275 FRDA, 290 controls) from the ENIGMA-Ataxia working group.
• Applied the Subtype and Stage Inference (SuStaIn) algorithm to MRI biomarkers for subgroup identification.
• Assessed clinical and genetic relevance of subgroups using linear models.

Main Results:

• Identified three distinct FRDA subtypes: classical (66.5%), early cerebral (25.8%), and early cerebellar (7.64%).
• Classical subtype shows ascending damage (brainstem to cerebrum); early cerebral shows preceding cerebral atrophy; early cerebellar shows preceding cerebellar lobule atrophy.
• Advanced MRI-based disease stages correlated with symptom duration and severity, with subtype moderating this relationship.

Conclusions:

• The SuStaIn algorithm successfully identified three distinct, MRI-based subtypes of Friedreich ataxia (FRDA).
• These subtypes exhibit unique patterns of brain degeneration and varying associations with clinical severity.
• This stratification offers a biologically meaningful framework for understanding FRDA heterogeneity and guiding therapeutic development.