Electroencephalography-Based Clustering Reveals Robust Neurophysiological Subtypes in Parkinson's Disease

Area of Science:

• Neuroscience
• Biomedical Engineering
• Genetics

Background:

• Parkinson's disease (PD) exhibits significant clinical heterogeneity, making conventional scales insufficient for understanding neural mechanisms.
• Scalp electroencephalography (EEG) offers a scalable method to measure brain activity for PD stratification, but EEG-based subtypes remain underexplored.

Purpose of the Study:

• To identify neurophysiological subtypes of Parkinson's disease using EEG-derived features.
• To investigate the clinical, cognitive, gait, and genetic correlates of these identified subtypes.

Main Methods:

• EEG recordings were analyzed from 116 PD patients and 30 controls during rest and walking.
• Spectral power, aperiodic components, and temporal complexity were extracted and analyzed using dimensionality reduction and clustering.
• Internal validity and stability were assessed via leave-one-out resampling, with clinical and genetic data compared across clusters.

Main Results:

• A robust three-cluster solution revealed distinct neurophysiological profiles based on spectral slowing, signal complexity, and aperiodic activity.
• Clusters showed no differences in motor severity or disease duration but differed in cognitive performance and genetic makeup.
• One cluster had preserved cognition and more LRRK2-mutation carriers; two cognitively impaired clusters had distinct electrophysiological signatures.

Conclusions:

• EEG-based clustering identifies reliable PD subtypes reflecting heterogeneity not captured by standard clinical assessments.
• These findings support EEG as a scalable tool for mechanism-based PD stratification.
• This approach has implications for improving prognosis and designing precision clinical trials for Parkinson's disease.