Objective:

(1) To identify neuroanatomical subtypes of patients with Parkinson’s disease (PD) and to define how they differ in motor symptoms. (2) To investigate how the neuroanatomy and motor symptoms of these subtypes change over time. 

Background:

Novel approaches to parse the heterogeneity in PD based on neurobiological mechanisms are needed to improve the efforts in precision medicine. Here, we used multimodal imaging data to identify PD subtypes and their progression.  

Design/Methods:

T1-weighted, diffusion tensor imaging data and clinical motor scores were obtained from the Parkinson’s Progression Markers Initiative database (www.ppmi-info.org/data) for 61 PD patients at both baseline and 48-month follow-up, and 70 healthy controls (HC) at baseline only. We extracted mean cortical thickness (CTh, reflecting cortical morphometry) from 68 cortical regions, and mean quantitative anisotropy (QA, reflecting the axonal integrity) from 45 white matter tracts. Using Mahalanobis distance, multivariate summary scores of CTh (M_CTh) and QA (M_QA) were derived and selected as features for heterogeneity through discriminative analysis. Group differences in neuroanatomical variables (M_CTh, M_QA), and motor symptoms between HC and PD subtypes were tested using ANOVA. To assess the longitudinal changes, a linear mixed-effect model was used.

Results:

Three PD subtypes were identified. Subtype 1 showed no detectable neuroanatomical abnormalities and mild motor symptoms at baseline. However, over time, these patients demonstrated substantial axonal degeneration with worsened motor symptoms. Subtype 2 showed cortical atrophy and severe motor symptoms at baseline, but with significant axonal degeneration and absence of changes in motor symptoms over time. Subtype 3 characterized by axonal damage and severe motor symptoms at baseline, displays no neuroanatomical or motor symptom deterioration over time. 

Conclusions: