Network Action of Subcallosal Cingulate White Matter Deep Brain Stimulation

Abstract

Deep brain stimulation (DBS) within the subcallosal cingulate cortex (SCC) alleviates symptoms of depression through an unclear therapeutic mechanism. Precise stimulation of SCC white matter (SCCwm) is thought to be necessary to achieve therapeutic response, and clinical recordings can now be used to test this hypothesis. In this paper we characterized the where, what, and how of SCCwm-DBS immediate effects, its network action, at therapeutic stimulation frequencies. First, using simultaneous LFP and EEG, we determined whether the effects of SCCwm-DBS are local at the SCC and/or remote at downstream cortical regions. We then charactized the spatial pattern effected by DBS across high-density EEG, finding multi-oscillatory response modes. Finally, we demonstrated that these modes are spatially consistent with white matter tracts targeted during surgical implantation. These results clarify the immediate actions of SCCwm-DBS as broad low-frequency power increases in brain regions downstream to stimulated white matter. This quantitative characterization of SCCwm-DBS network action has implications for future clinical trials, and may accelerate adaptive therapy optimization.

Decoding depression during subcallosal cingulate deep brain stimulation

“…Right-SCC PSDs exhibited significantly more variability than Left-SCC PSDs (Figure 3b). The peak found at 25 Hz was found during in vitro recording as a voltage-sensitive artifact [24]. This peak is included in traditional β oscillation definitions (15 Hz to 30 Hz) but is ignored as a part of mismatch compression corrections β * (15 Hz to 20 Hz) [24].…”

“…Local field potentials (LFPs) were recorded with the Activa PC+S™ (Medtronic PLC., MN, USA), a dual-channel recording device that measures differential LFP [27, 24]. Two channels were recorded at 422 Hz, one each in left and right SCC (Figure 1c).…”

“…To remove distortions related to mismatch compression, an extra preprocessing pipeline was applied in the frequencydomain representation of all recordings, unless otherwise specified. This pipeline removes broad-spectrum features, narrows frequency bands for oscillations, and measures the level of gain compression driven by stimulation-related distortions in the presence of heterogeneous tissue between the two ∂ LFP recording electrodes [24].…”

“…Recordings are transformed into power spectral densities (PSDs) using the Welch algorithm with Blackman-Harris window of 2 s, 0% overlap and 2 10 = 1024 FFT bins. The PSD is then preprocessed to remove mismatch compression distortions [22, 24] arising from chronic impedance variability. Average (median) oscillatory power is calculated within five modified band windows: δ (1-4Hz), θ (4-8Hz), α (8-14Hz), β * (14-20Hz), γ 1 (35-50Hz).…”

“…Physiologic signals from the brain networks underlying depression are needed [20], and recent DBS advances enable low-profile recording from around the stimulation target alongside active DBS therapy [21, 22, 23]. Differential LFP recording devices capable of recording alongside active therapy provide a powerful window alongside therapy, but require great care to remove stimulation-related distortions[22, 24]. Additionally, small and heterogeneous patient cohorts typical of antidepressant DBS studies require supervised machine learning to identify links of interest in the presence of unavoidable, confounding correlations.…”