Personalizing deep brain stimulation for Parkinson's disease

Conventional deep brain stimulation (DBS) for Parkinson's disease (PD) delivers constant stimulation, regardless of symptom fluctuations driven by levodopa intake. Adaptive DBS (aDBS) adjusts stimulation using a neurophysiological signal (‘physiomarker’), but how its behavior is shaped by surgery, stimulation and medication, remains unclear.
This thesis investigates the most commonly used physiomarker, the power of beta oscillations, and their response to deep brain stimulation (DBS) activation, stimulation amplitude, and levodopa, to personalize aDBS.
Using data from the AI-DBS study, a prospective cohort study (n = 100), we determined the duration of the electrophysiological ‘stun effect’. Similar to the clinical stun effect, the electrophysiological stun effect wanes over several weeks following DBS surgery. Only once the electrophysiological stun effect has waned does beta suppression with increasing stimulation amplitude become observable, so beta dynamics recorded in the first weeks after surgery are unreliable.
Next, we show that beta power captures levodopa cycles and that this is related to the expected pharmacodynamics of levodopa: it decreases after levodopa intake and returns to baseline just before the next dose. However, the frequency band showing the largest levodopa-related fluctuation during active DBS varies considerably between patients. A personalized physiomarker may therefore reflect levodopa-related symptom fluctuations better than a fixed frequency band.
A case series of 12 patients shows that adaptive DBS based upon a personalized physiomarker improves symptoms in a selected group of patients. The thesis provides a practical summary of personalizing subthalamic nucleus aDBS for PD using data-driven methods in the discussion.