Objective: We aimed to conduct image-based, data-driven classification of Parkinson’s disease (PD) progression using event-based modeling (EBM) and derive neuropathological insights from the results.

Background: Recent clinicopathological evidence suggest the existence of distinct subtypes regarding manifestation of prodromal biomarkers, speed of motor and nonmotor progression, and early cognitive impairment.

Method: This retrospective study included 565 patients with PD and propensity-matched 150 healthy participants, recruited to generate the cortical thickness at baseline. 16 regions of interest (9 areas of cerebral cortex and 7 areas of deep gray matter) were extracted from the T1-weighted magnetic resonance images (MRI). Clinical information, including REM sleep behavior disorder (RBD), levodopa equivalent daily dose (LEDD), Korean version of the Montreal Cognitive Assessment (K-MOCA), and clinical milestones (e.g. motor fluctuation, dyskinesia, wearing-off, freezing) were collected at the time of MRI. The Subtyping and Staging Inference (SuStaIn) algorithm was used to generate two distinct image-guided PD progression subtypes and clinical implications were sought thereafter through comparative analysis.

Results: The average disease duration was 6.7 years, and no sex difference was found between the two subtypes. We found two distinct neurodegenerative disease trajectories: deep grey (DG) to cortex (CO) versus CO to DG. Notably, the CO-first subtype demonstrated a higher prevalence of RBD (p = 0.009) and peak-dose dyskinesia (p = 0.024) compared to the DG-first subtype. Additionally, the progression was faster for CO-first subtype, leading to a significantly higher LEDD at an earlier stage (CO-first: 5.1 stage and 90.4mg/d LEDD annual increase, DG-first: 3.4 stage 75.2mg/d LEDD annual increase). Interestingly, the sensorimotor and auditory network-related areas and the amygdala, were selectively involved firsthand.

Conclusion: Our findings dictate the existence of two divergent subtypes in PD progression radiologically and has clinical implications beyond imaging finding. Longitudinal studies on the evolution of these subtypes are required to confirm.

« Back to 2024 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/data-driven-image-based-subtyping-and-staging-of-neurodegeneration-in-parkinsons-disease/