Dynamic brain state instability as a candidate indicator of conversion risk from major depressive disorder to bipolar...

Early identification of bipolar disorder (BD) is challenging, particularly when the first episode presents as depression and is misdiagnosed as major depressive disorder (MDD). Misclassification delays appropriate treatment. While prior studies focused on static imaging features, emerging evidence suggests that transient, frequency-specific neural dynamics may offer more sensitive candidate indicators.

We included 87 patients with MDD and 55 with BD. During follow-up, 33 MDD patients converted to BD (tBD), and 54 remained unipolar (UD). Baseline resting-state MEG and MRI data were analyzed using the DyNeMo model to characterize millisecond-level oscillatory states. Group differences in dynamic metrics were assessed with permutation ANCOVA.

Six dynamic brain states were identified. Significant group effects emerged for the switching rate (SR) of State 2 and for both fractional occupancy (FO) and SR of State 4. tBD and UD groups exhibited elevated SR in States 2 and 4 compared to the BD and HC groups, with tBD showing the highest SR in State 4.

The sample size of this study was limited, diagnostic transitions in UD could not be fully verified, and residual confounding from education level and medication use cannot be completely excluded.

Heightened instability of the low-power background network characterizes tBD and may reflect disrupted baseline neural homeostasis. This dynamic feature may serve as a candidate neuroimaging indicator for identifying individuals at elevated risk of conversion from UD to BD.