DYNAMIC CAUSAL MODELING OF TRIPLE-NETWORK CONNECTIVITY VARIES BY SUICIDE ATTEMPT HISTORY

Background

The triple-network model of psychopathology implicates dysregulated interactions among salience, default mode, and executive control networks in mood disorders and suicidal behavior. We examined whether suicide attempt (SA) history in major depressive disorder (MDD) or bipolar disorder (BD) is associated with altered triple-network effective connectivity using resting-state magnetoencephalography (MEG).

Methods

Adults with unipolar or bipolar depression (n=37; 29 MDD, 8 BD) completed restingstate MEG (275-channel CTF). Participants were grouped by lifetime SA history (SA+ n=13; SA− n=24). Regions of interest were selected a priori as canonical triple-network nodes: posterior cingulate cortex (PCC; default mode), bilateral anterior insula (AI; salience), and bilateral lateral prefrontal cortex (LPFC; executive control). Source time series were analyzed using Dynamic Causal Modeling (SPM12). Four biologically plausible effective connectivity architectures were compared using Bayesian model selection (BMS). Parametric Empirical Bayes tested for parameters differentiating SA+ and SA− [posterior probability (Pp) > 0.95].

Results

BMS favored the model architecture comprising reciprocal connections between all three nodes. Relative to SA−, SA+ participants showed reduced backward AI-to-PCC connectivity (Ep = −0.46; Pp = 1). SA+ participants also exhibited reduced PCC excitatory synaptic gain (Ep = −0.35; Pp = 1) and reduced PCC intrinsic self-coupling (Ep = −0.49; Pp = 1).

Discussion

Suicide attempt history was associated with altered salience-to-default mode feedback and altered PCC intrinsic dynamics, consistent with the triple-network model of psychopathology. These findings support a circuit-level account of suicidal behavior in mood disorders and motivate evaluation of effective connectivity markers in larger samples.