Abnormalities in large-scale brain network dynamics in late-life depression with suicidal ideation: an EEG microstate analysis

Background

Patients with late-life depression (LLD) with suicidal ideation (SI) often have more explicit suicide plans, and suicide attempts among older adults are more highly lethal than in other age groups. Increasing evidence suggests that people with SI in depression exhibit abnormal brain network connectivity; however, the relationship between suicidal ideation in LLD and brain network dynamics is still unclear.

Methods

We recruited patients with LLD and SI (LLD-SI), patients with LLD without SI (LLD-NSI), and age-matched healthy older adults. We collected 64-channel resting state electroencephalography (EEG) recordings of all participants and used microstate analysis to explore large-scale brain network dynamics.

Results

We included 33 patients with LLD-SI, 29 patients with LLD-NSI, and 31 controls. We observed abnormal microstate parameters in the LLD-SI group, characterized by higher duration (p = 0.04), occurrence (p = 0.009), and contribution (p = 0.001) of microstate C (reflecting activity of the salience network), compared with the LLD-NSI group, as well as higher occurrence (p = 0.03) and contribution (p = 0.009) of microstate C compared with the control group. Furthermore, transition probabilities from microstate class A to D (r = −0.466, p = 0.04) and class D to A (r = −0.506, p = 0.02) (involving coupling and sequential activation of auditory and executive control network) were negatively correlated with completion time of Stroop Colour and Word Test Part C (a neuropsychological test of executive function) in the LLD-SI group.

Limitations

The sample size was relatively small, the cross-sectional nature of this study prohibited exploring the causal relationship between abnormal microstate dynamics and suicidal ideation, and we did not include medication-naive patients with first-episode LLD.

Conclusion

The study reveals altered microstate dynamics among patients with LLD-SI, compared with patients with LLD-NSI and controls. Our findings suggest that microstate dynamics could serve as potential neurobiomarkers for identifying SI in LLD.