Shared and symptom-specific local field potential spectral signatures of motor symptoms in Parkinson's disease

Adaptive deep brain stimulation (aDBS) for Parkinson's disease (PD) aims to improve treatment efficacy by adjusting stimulation amplitude based on a neurophysiological feedback signal reflecting the patient's clinical state. Although beta band (± 13–35 Hz) spectral power from local field potential (LFP) recordings is the best-characterized physiomarker to-date, correlations with motor symptom severity remain modest. To explore whether combining spectral information from multiple frequencies improves explained variance in motor symptom severity, we applied canonical correlation analysis (CCA) to the full power spectral density (1–100 Hz) of LFP recordings from 67 patients with PD undergoing subthalamic nucleus DBS, alongside clustered and individual items of the Movement Disorders Society Unified Parkinson's Disease Rating Scale part III (UPDRS-III). CCA redundancy indices quantified that over 20% of variance in UPDRS-III scores was explained by a linear combination of spectral power across frequencies, compared to ~10% by beta power alone. Beta frequencies contributed most strongly and positively to total UPDRS-III scores, with slightly stronger contributions of low-beta (~13–25 Hz) than high-beta (~26–35 Hz) ranges. Low-frequency (<10 Hz) and finely-tuned gamma activity at half the stimulation frequency (62–63 Hz) contributed negatively. Symptom-specific analyses revealed a clear double beta peak associated with bradykinesia, while rigidity was primarily associated with low-beta power. Tremor showed a less distinctive spectral pattern. CCA results were highly similar for contralateral and ipsilateral symptoms. Overall, CCA findings revealed both shared and symptom-specific spectral patterns underlying PD motor symptoms and demonstrate the added value of a broader spectral physiomarker.