Computer-Based Evaluation of α-Synuclein Pathology in Multiple System Atrophy as a Novel Tool to Recognize Disease Subtypes

Ain Kim,Koji Yoshida,Gabor G Kovacs,Shelley L Forrest

DOI: https://doi.org/10.1016/j.modpat.2024.100533

Abstract:Multiple system atrophy (MSA) is a neurodegenerative disorder with variable disease course and distinct constellations of clinical (cerebellar [MSA-C] or parkinsonism [MSA-P]) and pathological phenotypes, suggestive of distinct α-synuclein (αSyn) strains. Neuropathologically, MSA is characterized by the accumulation of αSyn in oligodendrocytic glial cytoplasmic inclusions (GCI). Using a novel computer-based method, this study quantified the size of GCIs, density of all αSyn pathology, density of only the GCIs, and number of GCIs in MSA cases (n = 20). The putamen and cerebellar white matter were immunostained with the disease-associated 5G4 anti-αSyn antibody. Following digital scanning and image processing, total 5G4-immunoreactive pathology (ie, neuronal, neuritic, and glial) and GCIs were optically dissected for inclusion size and density measurement and then evaluated applying a novel computer-based method using ImageJ. GCI size varied between cases and brain regions (P < .0001), and heterogeneity in the density of all αSyn pathology including the density and number of GCIs were observed between regions and across cases, where MSA-C cases had a significantly higher density of all αSyn pathology in the cerebellar white matter (P = .049). Some region-specific morphologic variables inversely correlated with the age of onset and death, suggestive of an underlying aging-related cellular mechanism. Unsupervised K-means cluster analysis classified MSA cases into 3 distinct groups based on region-specific morphologic variables. In conclusion, we developed a novel computer-based method that is easily accessible, providing a first step to developing artificial intelligence-based evaluation strategies for large scale comparative studies. Our observations on the variability of morphologic variables between brain regions and cases highlight (1) the importance of computer-based approaches to detect features not considered in the routine diagnostic practice, and (2) novel aspects for the identification of previously unrecognized MSA subtypes that do not necessarily reflect the current clinical classification of MSA-C or MSA-P.