Theta Deep Brain Stimulation for Cognitive Enhancement in Parkinson's Disease

Objective: The objective of the proposed research is to test the ability of theta stimulation of the STN to modulate cognitive processing in PD patients by stimulating the ventral, putatively associative territory of the STN.

Hypothesis: The hypothesis is that theta stimulation of the ventral (putatively associative) STN will improve performance on cognitive tasks compared to a control condition in which stimulation is delivered to the dorsal (putatively sensorimotor) STN.

Methods: Power analysis - with previously reported moderate effect sizes (Cohen's d=0.57), the investigators anticipate the need to recruit 27 subjects for a within-subject design with a power of 0.8 and an alpha level of 0.05.

Subjects: Patients with deep brain stimulation systems will be recruited from the Vancouver General Hospital DBS clinic. Subjects will be included if they have STN DBS, pre- and post-operative imaging (to allow electrode reconstruction), and are at least three months post-operative. Subjects will be excluded if they are unable to complete the cognitive task (due to language barriers or dementia) or if they have significant DBS related complications.

Study design: This is a double blind, randomized, cross-over within-subject repeated measure design assessing the interaction between stimulation location and behavioural modification. Subjects will be blinded to stimulation condition, and the individual administering the working memory task will also be blinded, as a neutral third party will program the DBS settings for each condition.

Electrode reconstruction and stimulation location: For each subject, the electrode is reconstructed using the Lead-DBS pipeline. A pre-operative MRI and a post-operative CT are co-registered and non-linearly normalized into standard space, and electrodes are identified and reconstructed using the PaCER algorithm, with manual refinement in native space where required. Directional lead orientation is determined from the CT artifact.

The protocol specified an individualized, connectivity-guided selection of stimulation contacts, with a pre-specified fallback to simplified ventral and dorsal configurations in the event that differential activation of the associative and sensorimotor networks proved unachievable given the anatomical location of the electrode or the electrode model. Review of the electrode reconstructions of enrolled and upcoming subjects indicated that configurations satisfying the connectivity criteria did not exist for a sufficient proportion of subjects. The fallback was therefore invoked, and after the first two subjects all stimulation was delivered using standardized ventral and dorsal montages.

These montages are defined by contact position along the lead rather than by individualized connectivity modelling. The ventral (associative) condition uses the most distal ring contact of each lead as cathode against the implantable pulse generator case as anode (left lead 0-/C+; right lead 8-/C+). The dorsal (sensorimotor) condition uses the most proximal ring contact as cathode in the same monopolar configuration (left lead 3-/C+; right lead 11-/C+). Stimulation is bilateral in both conditions. Because contact numbering follows a consistent dorsoventral ordering along the implanted leads, this fixed scheme produces a maximally separated ventral-versus-dorsal contrast within every subject's lead, independent of individual anatomy, and avoids confounding stimulation location with subject-specific programming decisions.

The ventral and dorsal designations are therefore a priori geometric labels, not verified per-subject network assignments. What was actually stimulated in each subject is established post hoc. For each subject and each theta condition, the electric field is simulated using the exact delivered montage, amplitude and pulse width, and overlap of the stimulation volume with the associative and sensorimotor territories of the STN is quantified in standard space using the DISTAL atlas, expressed as absolute volume, as a proportion of total stimulation volume, and as electric-field-weighted overlap. Structural connectivity between the stimulation volume and cortical regions of interest is estimated from normative structural connectomes as the number of streamlines traversing the stimulation volume and terminating in the region of interest; functional connectivity is estimated from normative resting-state connectomes as the Pearson correlation between the stimulation volume and the region of interest.

Baseline cognitive assessment: Each subject will undergo a baseline cognitive screening examination (Montreal Cognitive Assessment, MOCA), and that will be used as a covariate in future analysis.

Theta DBS: Subjects will come into the clinic on their regular medication schedule. The time since their last medication will be obtained, along with the participants levodopa equivalent daily dose (LEDD) to be used as a covariate to help control for the known non-linear contribution of dopamine to working memory. They will undergo a computerized working memory task while being stimulated in each of four following conditions:

• OFF
• Target condition: Theta (6 Hz) stimulation of the ventral (associative) STN.
• Anatomical control condition: Theta (6 Hz) stimulation of the dorsal (sensorimotor) STN.
• Frequency control condition: The subject's own clinically optimized high-frequency program.

These conditions will be randomized within each subject by creating a random integer (1-4) with no repeats. In the two theta conditions, pulse width is held at 60 µs and amplitude is set to the subject's clinical amplitude, matched across the two conditions so that delivered charge is equivalent and only the stimulation site differs. The clinical condition uses the subject's own programmed contacts, amplitude, pulse width and frequency. After changing the stimulation setting, there is a wash-out/wash-in period of at least 10 minutes prior to commencement of the working memory task. The subject will be asked if there is any unwanted stimulation side-effect (for example paresthesia). If the subject has paraesthesia or otherwise complains about the stimulation, the amplitude will be reduced by 0.5mA. If stimulation is not possible, the patient will be excluded from the study. Immediately before the working memory task commences, subjects will undergo a brief motor examination, consisting of items from the Unified Parkinson's Disease Rating Scale III. At the end of the session, the subjects will be placed back on their clinically optimized program and will continue to take their medication according to their normal schedule.

Working Memory Task: A modified Sternberg task will be presented using a laptop computer.

Statistical analysis: The primary outcome will be working memory performance during each stimulation condition as assessed by a linear mixed effect model. The model will be specified as working memory performance ~ age + MoCA + stimulation condition + baseline cognition*stimulation condition + stimulation order + (1|subject). The hypothesis is that there will be a main effect of stimulation condition on working memory performance, with post-hoc one sample t-tests demonstrating an improvement in the theta associative STN condition relative to the other conditions. An interaction effect of baseline cognition with stimulation condition is also expected to be observed. Based on previous literature, it is hypothesized that subjects with lower baseline cognition will have more improvement during associative STN region stimulation.

A secondary analysis will relate working memory performance to what was actually stimulated in each subject. For each subject, the degree to which the ventral and dorsal montages achieved the intended anatomical dissociation is quantified as overlap of the modelled electric field with the associative and sensorimotor STN territories, and as connectivity between the stimulation volume and dorsolateral prefrontal, primary motor and supplementary motor cortex. These measures will be tested as moderators of the effect of stimulation condition on working memory performance. The hypothesis is that subjects in whom the ventral montage produced greater associative-territory engagement will show greater working memory improvement during ventral theta stimulation.

By including the stimulation order variable, the investigators will adjust for the possibility that subjects performance will decline over time (due to fatigue) irrespective of the stimulation condition. Supplementary analysis will assess for motor performance during each stimulation condition, with the hypothesis that high frequency stimulation will have a reduction in motor scores compared to low frequency stimulation. Post-hoc analysis will also be performed to assess the relationship between task performance and functional and structural connectivity between the STN and associative networks, assuming there is some variability in individual connectivity strengths.

Inclusion criteria

• Those with STN DBS devices
• Those with pre- and post-operative imaging
• Those that are at least 3 months post-operative

Exclusion criteria

• Those unable to complete the cognitive task (due to language barriers or dementia)
• Those with significant DBS complications