Objective:

In this study, we investigated whether dynamic patterns of large-scale limbic-prefrontal synchrony using hidden Markov models (HMM) are associated with depression severity and assessed how epileptiform activity modulates these dynamic connectivity patterns. 

Background:

Depression has been linked to altered communication within the limbic and prefrontal networks. Dynamic approaches, such as HMM can capture transient network states and provide insight into how brain connectivity unfolds over time. 

Design/Methods:

We analyzed two-hour continuous iEEG recordings from 12 subjects with depth electrodes implanted in the amygdala, hippocampal head, hippocampal tail, orbitofrontal cortex, and anterior cingulate cortex. Interregional dynamic functional connectivity (dFC) was calculated using the imaginary part of the phase locking value (iPLV) for each 5-second epoch of EEG data for each canonical frequency band. HMM was applied to the dFC to identify recurring network states, with the optimal number of states determined via the Akaike Information Criteria (AIC). Fractional occupancy and mean dwelling time were correlated with depression severity scores. False discovery rate (FDR) correction was applied for multiple comparisons. 

Results:

Among the four inferred states, one state that was characterized by significant involvement of the beta-band iPLV showed a significant association with depression severity. Fractional occupancy (r = -0.70, p(FDR) = 0.04) and mean dwell time (r = -0.69, p(FDR) = 0.05) of this state decreased with higher depression scores. Epileptiform discharges were associated with reduced dwell time and decreased self-transition probability in this state, suggesting that interictal activity disrupts the stability of beta-dominated network configurations. 

Conclusions:

These findings indicate that reduced stability of a beta-dominated network state within the broader limbic-prefrontal cortical networks is linked to greater depression severity. Epileptiform discharges appear to destabilize this state, potentially exacerbating mood dysregulation. HMM-based analysis of iEEG connectivity provides a promising framework for identifying dynamic neural signatures of mood disorders.