A Preoperative Electroencephalography Signature for Predicting Treatment Response to Deep Brain Stimulation in Obsessive-Compulsive Disorder
Journal:
medRxiv
Published Date:
Jul 19, 2026
Abstract
Deep brain stimulation (DBS) is effective for treatment-refractory obsessive-compulsive disorder (OCD), but outcomes are heterogeneous and non-responders incur surgical and financial burden. We sought a scalable, non-invasive preoperative signature of DBS response. Using preoperative resting-state electroencephalography (EEG) from a randomized, double-blind, sham-controlled trial of nucleus accumbens/anterior limb of internal capsule DBS (N = 24; NCT04967560), we developed a machine learning approach tailored to small samples to identify lower relative delta power at a right fronto-temporal electrode as a robust predictor of greater six-month symptom reduction, explaining >40% of outcome variance and improving the response rate >20% over all-comers. This EEG signature was aberrant relative to healthy controls yet unrelated to baseline symptom severity, and predictive only under active, not sham, stimulation. Notably, it prospectively predicted outcomes in an independent cohort (7 of 8 patients). The signature generalized across eyes-open and eye-closed recordings, was corroborated by source-space magnetoencephalography, and showed excellent short- and long-term test-retest reliability. Mechanistically, this EEG signature was highly correlated with right fronto-temporal aperiodic exponent, and its predictive strength was concentrated in cortical regions enriched for inhibitory-neuron markers, linking the signature to excitation-inhibition balance. Moreover, longitudinal signature changes tracked clinical improvement, providing proof-of-concept support for its use as a target-engagement readout. These findings establish a scalable, biologically grounded EEG signature to guide patient selection for DBS in severe OCD.
