Objective: We aimed to categorize Parkinson’s disease (PD) using k-means clustering based on thalamic subnuclei volumes measured by structural magnetic resonance imaging (sMRI).

Background: The thalamus, a relay nucleus in the input to the cerebral cortex, is divided into about 30 subnuclei according to the characteristics of the constituent neurons and input/output fibers. It is known that motor pathways, limbic pathways, and sensory pathways other than smell all pass through the thalamus and are also involved in the pathogenesis of PD. FreeSurfer (https://surfer.nmr.mgh.harvard.edu/) enabled us to calculate thalamic subnuclei volumes, based on which we may categorize patients with PD.

Method: We recruited 105 patients with sporadic PD (Male 56, median age 69.0, IQR 65.0-74.2) aged under 80 years and excluded patients with onset before 50 years old. All patients received 3T MRI. Cognition was assessed by Mini-Mental State Examination (MMSE), the Frontal Assessment Battery (FAB), and the Japanese version of the Montreal Cognitive Assessment (MoCA-J). Thalamic subnuclei volumes were calculated by FreeSurfer and the average of both side of thalamic subnuclei was used for analysis. After normalizing each thalamic subnuclei, k-means clustering was performed based on thalamic subnuclei volumes, and the optimal number of clusters was determined by Hubert index and elbow plot. Comparisons between clusters were conducted using the Kruskal-Wallis. Each thalamic subnuclei was further analyzed using multiple regression analysis (MRA), adjusting covariates including age and gender, to determine their association with cognitive function. Specific thalamic subnuclei including lateral dorsal nucleus (LD), lateral pulvinar nucleus (PuL), medial mediodorsal nucleus (MDm) were selected as covariates according to the previous reports.

Results: Significant differences between clusters were found for each cognitive scale (MMSE and MoCA-J, p < 0.01; FAB, p < 0.001). MRA showed that LD (p = 0.0003-0.045), PuL (p = 0.009-0.02), and age (p = 0.001-0.03) were significant factors in MMSE, MoCA-J, and FAB, while no significant association was found between MDm and those cognitive scales.

Conclusion: This study highlights that thalamic subnuclei may reflect cognitive function in PD. The identified thalamic subnuclei may serve as potential biomarkers for assessing cognitive variation in PD. Validation in external datasets is necessary to confirm these findings.

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/sub-grouping-of-parkinsons-disease-based-on-thalamic-subnuclei-volume/