The value of non-motor features and genetic variants of Parkinson's disease for clustering Lewy body diseases

Introduction The use of non-motor Parkinson9s disease (PD) features and genetic PD variants for clustering analyses may refine the phenotypic classification of idiopathic Lewy body (LB) diseases. Methods One-hundred participants [n=7 E46K-SNCA (n=5 symptomatic and n=2 asymptomatic), n=4 PARK2, n=3 LRRK2, n=8 dementia with Lewy bodies (DLB), n=48 idiopathic PD (iPD), n=30 healthy controls (HC)] underwent a comprehensive evaluation of non-motor and motor PD features. Non-motor features were used to perform a hierarchical clustering analysis with patients and HC using a Scikit-learn toolkit. Results Clustering analysis suggested three clusters of subjects including Cluster 1 or “Normal-to-mild”: young iPD ( 60 years) classified together with the lowest symptomatic E46K-SNCA, PARK2 carriers and HCs, characterizing by having mild-to-moderate cognitive and motor disabilities with few axial symptoms; and Cluster 3 or “Severe”: old iPD (>60 years) classified together with all DLB and the most symptomatic E46K-SNCA carriers, characterized by having severe pattern-specific cognitive disabilities (visual attention, perception, processing speed, memory and executive functions) and severe motor PD manifestations with marked axial symptoms. Conclusions Our study supports the potential value of incorporating genetic PD variants in data-driven iPD classification algorithms and the usefulness of non-motor PD features, especially visual cognition abnormalities, to facilitate the identification of aggressive LB diseases.