This study revisits data from the Vanderbilt “DBS in early-stage PD” pilot clinical trial (NCT00282152; IDEG050016; IRB#040797), which was a prospective, randomized, single-blind clinical trial evaluating bilateral STN-DBS in patients with early-stage PD (aged 50–75 years; PD medication duration 1–4 years; no history or evidence of dyskinesia or motor fluctuations) [9]. All 14 subjects with complete data for the prior electrode localization study [1] are included in this analysis, alongside clinical outcomes for subjects randomized to optimal drug therapy (ODT) in the pilot trial [9]. The present study also includes an independent cohort of 29 advanced-stage PD patients who received STN-DBS as standard care and participated in a long-term outcomes study at Vanderbilt University Medical Center (IRB#181198). The standard of care DBS subjects were recruited at least two years post-surgery. All subjects provided written informed consent for participation.

In the early-stage PD cohort, motor progression was defined as the change in the Unified Parkinson’s Disease Rating Scale part III score (UPDRS-III) from pre-operative baseline to 24 months measured after a one-week therapeutic washout (baseline: OFF medications; 24 months: OFF medications and OFF stimulation), which was blindly rated from video-recorded motor examinations at the conclusion of the trial [9]. In the advanced-stage PD cohort, pre-operative (ON medications) and longitudinal post-operative (ON medications, ON stimulation) UPDRS-III motor examinations were video-recorded and blindly rated. Long-term motor symptom changes were defined as the percent change from the pre-operative UPDRS-III to the longitudinal post-operative UPDRS-III. UPDRS-III scores for both cohorts do not include rigidity which cannot be evaluated via video recording. Levodopa equivalent daily doses (LEDDs) were calculated for both cohorts as previously described [10].

Electrodes for both cohorts were localized with Lead-DBS [11, 12] using pre-operative T1-weighted and T2-weighted magnetic resonance (MRI) scans and post-operative computed tomography (CT) scans. The same processing pipeline for localizations previously reported for the early DBS cohort [1] was applied to the standard of care DBS cohort. Group visualizations were performed using Lead-Group [13].

The motor progression sweet spot and connectivity models were defined using the early-stage PD cohort as previously published [1]. Critically, both models were used exactly as previously published with all model parameters unchanged [1].

Sweet spot model: model definition (as carried out in [1]): the sweet spot identified in the DBS in early-stage PD pilot clinical trial was defined by voxels covering at least 3 electric fields (i.e., E-fields) with a vector magnitude > 0.2 V/mm [1]). Model validation (present study): we tested whether this published sweet spot would be able to account for variance in long-term clinical outcomes in a common advanced-stage PD cohort. To do so, E-field magnitudes for each patient in the standard of care, advanced-stage PD cohort were spatially rank-correlated with the sweet spot following published methods [14, 15]. This led to a Spearman correlation coefficient for each E-field, denoting how strongly it overlapped with the sweet spot in an interval between -1 and 1. These coefficients were then correlated with clinical improvements across the cohort [12].

Optimal connectivity model: the same concept was applied to the a priori published connectivity model. Model definition (as carried out in [1]): the optimal connectivity model was identified using the DBS in early-stage PD pilot clinical trial cohort with the outcome of two-year motor progression (change from baseline to 24 months on examiner-blinded UPDRS-III score measured after a one-week therapeutic washout). Structural tract definitions of this model were based on an extended version of the DBS Tractography Atlas [16, 17], which was built using manual tractography results that were based on the HCP 1065 template provided with the DSI Studio software [18]. The HCP 1065 template was constructed from a total of 1065 subjects’ diffusion MRI data from the Human Connectome Project (2017 Q4, 1200-subject release). The cohort consisted of 575 females and 490 males, with ages ranging from 22 to 37 years (mean = 29; Q1 = 26, median = 29, Q3 = 32). Each of these streamlines were assigned a ‘Fiber-R-Score’ by correlating peak E-field magnitudes along the streamline with motor progression scores across the early DBS cohort. Model validation (present study): E-field magnitude values of the advanced-stage PD cohort were superimposed with the a priori published tract model [1] and average magnitudes were multiplied with the correlation weight of each streamline they intersected following the approach of [14, 15]. Resulting coefficients (‘Weighted Means of Fiber R-scores’) were then correlated with clinical improvements across the advanced-stage PD cohort.

The de-identified data from the standard of care DBS cohort will be made available upon reasonable request. The de-identified data and related study documents from the ‘DBS in early-stage PD’ pilot clinical trial are not being publicly shared at this time as they are currently being used for the development of a proprietary, multicenter, phase III, pivotal clinical trial (IDE G050016).