Data-driven trajectories of atrophy explain clinical heterogeneity across Lewy body diseases
Journal:
medRxiv
Published Date:
Jul 8, 2026
Abstract
Background: Lewy body diseases (LBD) collectively share alpha-synuclein Lewy pathology, yet present wide clinical heterogeneity, with overlapping motor and non-motor features and progression patterns that challenge traditional diagnostic boundaries. Methods: To resolve this spatiotemporal heterogeneity at the biological level, we applied a data-driven atrophy progression framework to MRI data from 833 individuals across Parkinsons disease (PD), dementia with Lewy bodies (DLB), and prodromal idiopathic REM sleep behaviour disorder (iRBD) using the Subtype and Stage Inference (SuStaIn) algorithm. Findings: Four transdiagnostic subtypes (A: Early cortico-limbic/late basal ganglia, B: Early basal ganglia/late limbic, C: Early temporo-limbic/late basal ganglia, and D: Early basal ganglia-cingulate/late cortex) emerged, each defined by a distinct spatiotemporal progression of atrophy that explained cognitive, motor, and psychiatric variability. An early cortico-limbic/late basal ganglia subtype represented a dementia-prone subtype across clinical diagnoses, with limbic involvement associating with the emergence of visual hallucinations. Interpretation: These biologically relevant spatiotemporal atrophy subtypes provide an interpretable stratification of patients with LBD, with potential to refine prognosis, improve clinical trial stratification, and guide precision therapeutic approaches. Funding: This work was made possible by an Ignition grant from the University of Sydney and University College London (Global Engagement Fund). Keywords: Dementia, Heterogeneity, Lewy body diseases, Subtypes, Neurodegeneration, Machine learning.
