High-frequency stimulation (HFS) of the subthalamic nucleus (STN) is known to be effective in Parkinson's Disease (PD) patients, whereas low-frequency stimulation (LFS) is not. We examined single unit spike train activity in the contralateral STN during HFS and LFS of the STN in 8 PD patients during deep brain stimulation (DBS) surgery. We were able to record in the contralateral STN during STN stimulation, where stimulation pulses produce less artifactual interference with single unit data from the recording zone. Combined with the ability to cluster and remove stimulation-related artifact, we were able to better observe and define the features seen in HFS-induced changes to pathological beta oscillation, burstiness, and firing rate.

We hypothesized, and indeed found, that HFS drives a repetitive network-wide modulation that counteracts systemic pathologic neuronal activity that is not seen with LFS. We also found that this physiological modulation is greatest where DBS (with HFS) happens to make the greatest clinical impact, within the dorsal-most portion of the motor subunit, 1-2mm below the AC-PC plane, in posterolateral STN. We targeted this location based upon our prior observation (Cheng et. al. 2006, 2010) that patients with the most dramatic clinical improvements had their electrodes implanted within this location. We call this location the "sweet spot" for STN PD DBS. Within this "sweet spot", we find modulation of pathological PD dynamics as follows: 1) STN HFS appears to suppresses STN firing counteracting PD STN hyperactivity, whereas STN LFS does not reliably modulate firing rates; 2) firing rate suppression is accompanied by a stereotyped pulse-by-pulse decrement in the propensity to spike in the first 1-2 ms directly after each high frequency stimulation pulse; and 3) there is a stimulation frequency-locked increased probability to spike 3-4ms after each HFS pulse. These findings are not seen with LFS.

With our collaborators, we are now beginning to model the downstream effect of these findings upon beta power, burstiness, and firing rate using biophysical modeling methods and control algorithms. We believe that this work may lead to greater mechanistic understanding of why HFS, but not LFS, modulates the pathological dynamics of Parkinson's Disease. In addition, with the means to detect specific physiological signals related to HFS from the "sweet spot" that appear to correlate with clinical PD improvement, we hope to improve surgical localization and postoperative programming for DBS systems.