Behavioural and clinical biomarkers of the human bed nucleus of stria terminalis from direct neural recordings.

Identifying novel neuromodulatory targets for deep brain stimulation (DBS) in psychiatric disorders is an urgent clinical need. Equally critical is the discovery of simple oscillatory biomarkers that bridge behaviour and clinical symptoms, enabling personalized treatment strategies. The bed nucleus of the stria terminalis (BNST), a pivotal output structure of the amygdala, is a potential candidate for DBS due to its key role in regulating fear, emotional valence, and prosocial behaviour. However, owing to the small size, its neural dynamics and functional contributions are poorly understood, precluding behavioural-clinical relevance. In a cross-sectional design, we acquired BNST neural recordings from 23 patients with depression undergoing DBS during two tasks: pain perception with painful/non-painful scenarios and an affect task with emotionally valenced images. We first localized the electrode contacts in the BNST and using their neural recordings for further analysis. We subjected the preprocessed data to time frequency decompositions to find condition differences. The significant clusters were then used to link to the behavioural ratings and clinical symptom severity. Furthermore, cross-frequency interactions were also undertaken. Pain perception elicited late theta and alpha activity (∼1s), with theta activity linked to subjective pain ratings and alpha correlating with anxiety/depression scores and anxiety symptoms post-DBS. Negative imagery induced early theta (∼250 ms), resembling previously reported amygdalar responses, and which link to valence ratings, depression and anxiety symptom severity. These results reveal distinct BNST dynamics in depression: early theta for rapid threat processing and late theta/alpha for complex socio-cognitive responses. Task-dependent theta and alpha activity linked behavioural profiles and symptom severity, highlighting BNST's role in behaviourally and clinically relevant oscillatory patterns, contributing novel insights for advancing precision neuromodulation strategies.