Effective deep brain stimulation for obsessive-compulsive disorder and Tourette Syndrome increases network-wide neural variability

Timon Merk, Kalman A. Katlowitz, Tommy Liu, Sandesh Reddy, Anthony K. Allam, Jane H. Bentley, Kasra Mansourian, Sarah Soubra, Thomas Hamre, Rick R. Hanish, Garrett P. Banks, Arjun Tarakad, Steven Bellows, Nora Vanegas‐Arroyave, Eric A. Storch, Sarah R. Heilbronner, Sameer A. Sheth, Nicole R. ProvenzamedRxiv 2025 · DOI

This research paper, recently posted to medRxiv, explores the neurophysiological mechanisms behind Deep Brain Stimulation (DBS) for psychiatric and movement disorders. By analyzing long-term brain recordings from patients with comorbid OCD and Tourette Syndrome, the authors suggest that "healing" may look less like a new rhythm and more like a healthy dose of chaos.

TL;DR

Successful Deep Brain Stimulation (DBS) for OCD and Tourette Syndrome is characterized by a significant increase in "neural variability"—specifically, the disruption of rigid, highly predictable 24-hour rhythms in the brain. This suggests that therapeutic response is linked to restoring physiological flexibility across subcortical networks.

What Problem Does This Paper Address?

While DBS is an established therapy for treatment-resistant obsessive-compulsive disorder (OCD) and Tourette Syndrome (TS), we still don't fully understand how it works at a circuit level. Most studies focus on immediate, local effects of electricity on neurons.

However, recent evidence suggests that psychiatric symptoms fluctuate on a "chronobiological" scale (circadian rhythms). The authors previously found that in OCD patients, successful DBS in the Ventral Striatum (VS) disrupted overly rigid neural rhythms. This paper tackles a critical follow-up question: Is this "rhythm disruption" a local phenomenon unique to the VS, or does it generalize to other brain regions like the Globus Pallidus internus (GPi) used to treat Tourette's?

Methods

The researchers utilized a rare clinical setup: four patients implanted with bilateral DBS leads in two distinct targets (VS and GPi) to treat comorbid OCD and TS.

  • Chronic Sensing: Using the Medtronic Percept neurostimulator, they recorded Local Field Potentials (LFPs) continuously for over 900 days.
  • The "9 Hz Band": They focused on power fluctuations in the 9 Hz range (the theta/alpha border), a known biomarker in these circuits.
  • Predictability Modeling: To quantify "variability," they used Linear Autoregressive (AR) models. Essentially, they tested how well they could predict today’s neural power based on yesterday’s data. A high $R^2$ value meant a rigid, predictable rhythm; a low $R^2$ indicated high variability (disruption).
  • Network Analysis: They used Granger causality and cross-correlation to determine if the VS and GPi were "talking" to each other or simply reacting to a third, unseen driver.

Datasets

The study relies on a unique, longitudinal dataset:

  • Human Subjects: 4 adults with severe, treatment-resistant OCD and comorbid TS.
  • Duration: Mean follow-up of 662 days, with some recordings exceeding 900 days.
  • Clinical Metrics: Yale-Brown Obsessive-Compulsive Scale (Y-BOCS II) and Yale Global Tic Severity Scale (YGTSS).

Key Insights and Ideas

  • The "Rigidity" of Disease: In the pre-DBS state, patients exhibited highly predictable, 24-hour cycles of neural activity. The authors hypothesize that these rigid rhythms may be a hallmark of the "pathological state."
  • Variability as a Biomarker: In patients who responded to treatment, DBS "shattered" these predictable rhythms. The brain activity became more variable and less predictable. This aligns with the broader biological theory that healthy systems function optimally with a degree of "physiological noise" or flexibility.
  • Network-Wide, Not Just Local: The disruption wasn't confined to the site of stimulation. If DBS was effective for TS, the GPi showed increased variability; if effective for OCD, the VS showed it. In responders, both regions often showed disruption.
  • The Timescale of Discovery: Interestingly, the 9 Hz coupling between brain regions was invisible in short, high-resolution recordings (seconds). It only became apparent when looking at "longitudinal snapshots" over hours and days. This highlights the importance of chronic, long-term sensing in neuromodulation research.

Key Results

  • Predictability Drop: In clinical responders, the AR model's ability to predict neural power ($R^2$) plummeted. For example, in one responder, GPi predictability dropped from ~40% to ~6% post-stimulation.
  • Responder Classification: Neural variability accurately classified responder vs. non-responder status across all four patients. Non-responders maintained stable, rigid rhythms throughout the study.
  • Patient-Specific Directionality: While the fact of disruption was consistent, the "flow" of information was not. Granger causality showed that in one patient, the VS drove the GPi, while in others, the direction was reversed or ambiguous.

Limitations and Future Work

  • Sample Size: With only four patients (n=4), the findings are compelling but require validation in larger cohorts.
  • Causality vs. Correlation: While the disruption predicts clinical response, it is still unclear if the increased variability is the cause of improvement or a byproduct of a brain returning to a healthy state.
  • Closed-Loop Potential: The authors suggest these circadian biomarkers could eventually be used for "closed-loop" DBS, where the device automatically adjusts stimulation levels based on the "rigidity" or "flexibility" of the patient's neural rhythms in real-time.

Summary updated

Abstract

Our findings demonstrate that increased low-frequency neural variability within the VS and GPi predicts symptom improvement in OCD and TS, respectively. These findings extend our previous results in OCD alone and recapitulate the notion that biological systems function optimally with a healthy degree of physiological variability.
BibTeX
@misc{merk2025effective,
  author = {Timon Merk and Kalman A. Katlowitz and Tommy Liu and Sandesh Reddy and Anthony K. Allam and Jane H. Bentley and Kasra Mansourian and Sarah Soubra and Thomas Hamre and Rick R. Hanish and Garrett P. Banks and Arjun Tarakad and Steven Bellows and Nora Vanegas‐Arroyave and Eric A. Storch and Sarah R. Heilbronner and Sameer A. Sheth and Sameer A. Sheth and Nicole R. Provenza},
  title = {{Effective deep brain stimulation for obsessive-compulsive disorder and Tourette Syndrome increases network-wide neural variability}},
  year = {2025},
  doi = {10.1101/2025.10.09.25337425},
  howpublished = {medRxiv},
}
Figure