Distributed brain network hypoconnectivity linked to suicide attempt history and childhood sexual abuse in depression

Introduction: Suicide is a leading cause of death among individuals with major depressive disorder (MDD), and a prior suicide attempt remains the strongest predictor of future suicide mortality. Although resting-state functional magnetic resonance imaging (fMRI) studies have identified regional functional connectivity (FC) abnormalities associated with suicidality, the whole-brain connectome architecture specifically differentiating depressed patients with and without a lifetime history of suicide attempts has not been fully characterised. How such network-level alterations relate to suicidal ideation severity and specific childhood maltreatment subtypes also remains unclear.

Aims: This study aimed to (1) identify distributed FC subnetworks distinguishing MDD patients with a suicide attempt history from those without, (2) evaluate whether identified FC patterns correlate with suicidal ideation and childhood trauma dimensions, and (3) characterise topological properties of attempt-related subnetworks using graph-theoretical approaches.

Methods: Resting-state fMRI data were acquired from 204 adults: 61 MDD patients with at least one lifetime suicide attempt (suicidal depression, SD), 62 without such history (non-suicidal depression, NSD), and 81 healthy controls (HC). Whole-brain FC matrices were constructed using the Schaefer 400-region cortical parcellation mapped onto seven canonical networks. Between-group differences were tested using network-based statistics (NBS; primary threshold t = 3.3, 5000 permutations, family-wise error [FWE]-corrected p < 0.05), controlling for age, sex, and education. Graph-theoretical metrics were computed at whole-brain and subnetwork levels. Partial correlations examined associations of subnetwork FC with suicidal ideation (Beck Scale for Suicide Ideation, SSI) and childhood trauma subtypes (Childhood Trauma Questionnaire, CTQ), with Bonferroni correction within each group.

Results: NBS revealed a significant subnetwork of reduced FC in SD relative to NSD (pFWE = 0.038; 82 nodes, 118 edges), spanning visual, somatomotor, dorsal attention, frontoparietal, and default mode networks. SD additionally showed widespread hypo- and hyperconnected subnetworks compared with HC (pFWE = 0.001 and 0.023, respectively), whereas NSD exhibited no significant FC differences from HC. Within this subnetwork, graph-theoretical analysis demonstrated reduced strength (pFWE = 5.05 × 10⁻⁵), global efficiency (pFWE = 2.93 × 10⁻⁴), and betweenness centrality (pFWE = 5.69 × 10⁻⁸) in SD versus NSD, whereas whole-brain graph metrics did not differ between groups. In the MDD group, mean subnetwork FC was inversely associated with SSI scores (r = −0.327, pFWE = 0.004). Somatomotor–visual and default mode within-network FC also correlated negatively with suicidal ideation (r = −0.278 and r = −0.275; pFWE < 0.05). In SD, somatomotor within-network FC was inversely associated with childhood sexual abuse severity (r = −0.402, pFWE = 0.027). These findings remained robust after adjusting for depressive severity, medication status, and illness duration.

Conclusions: Suicide attempt vulnerability in MDD is characterised by selective hypoconnectivity across sensorimotor, attentional, and self-referential systems rather than global network disorganisation. The association between reduced somatomotor connectivity and childhood sexual abuse suggests that trauma-related disruption of bodily representation systems may contribute to long-term suicide risk. These connectome-level patterns may inform biomarker development for suicide risk stratification in depression.

References

No references