Neural Correlates of Self-Directed Violence: A Large-Sample Resting-State fMRI Study from UK Biobank

ImportanceSuicide remains a leading cause of death worldwide, underscoring the importance of identifying neurobiological markers associated with suicidal behaviors. While non-suicidal self-directed violence (NSSDV) is an established risk factor for future suicide attempts (SA), the underlying neural distinctions between these behaviors remain insufficiently understood. ObjectiveTo examine functional connectivity (FC) changes that may differentiate SA from NSSDV based on resting-state functional magnetic resonance imaging (rs-fMRI). Design, setting, and participantsIn this case-control study, cross-sectional data of participants with a history of self-directed violence (SDV) were stratified into SA (n = 579) and NSSDV (n = 491) groups based on self-reported questionnaires. Both cross-sectional data and rs-fMRI data were collected from the UK Biobank between October 2016 to January 2024. Main outcomes and measuresFC matrices across different brain regions were analyzed. A general linear regression model was employed to identify significant FC correlations with SA. An exploratory analysis examined the correlation between FC and SDV frequency and interval. ResultsA total of 1,070 participants with a history of self-directed violence (SDV) were stratified into SA and NSSDV groups. Compared to NSSDV, the SA group exhibited decreased FC in bilateral amygdala and right putamen, alongside increased FC in the right caudate. Higher frequency of SDV was correlated to increased FC between the orbitofrontal cortex (OFC) and occipital regions, whereas individuals reporting SDV within the past 12 months showed decreased FC between the OFC and both the rectus gyrus and temporal lobe. Conclusions and relevanceThis large-sample rs-fMRI study highlights distinct FC alterations associated with different dimensions of self-harm. The limbic and reward circuits appear particularly relevant in differentiating SA from NSSDV and in capturing variations tied to SDV frequency and recency. These insights advance our understanding of suicidal behaviors neurobiological underpinnings and may inform targeted risk stratification.