RESTING-STATE EEG SIGNATURES OF SUICIDAL IDEATION SEVERITY AND INTENSITY IN TREATMENT-RESISTANT DEPRESSION: PRELIMINARY BASELINE DATA FROM TWO ONGOING CLINICAL TRIALS

Background

Suicidal ideation (SI) is prevalent among persons with treatmentresistant depression (TRD) and is associated with increased all-cause mortality risk. Identifying scalable neurophysiologic correlates of SI is important for improving risk characterization and for identifying treatments that normalize suicide-relevant circuitry. Resting-state electroencephalography (EEG) spectral power is a scalable index of circuit dynamics that may differentiate graded SI burden.

Methods

Baseline EEG recordings from two ongoing phase II TRD trials were analyzed, integrated intravenous racemic ketamine plus internet-based cognitive behavioral therapy (KET-CBT, NCT06480500; n = 30) and psilocybin-assisted psychotherapy dose comparison (PSI-1V2, NCT06341426; n = 29). Analyses were conducted within each trial and in a pooled baseline dataset. Resting eyes-open (EO) EEG was prespecified as the primary condition to standardize wakeful vigilance, and eyes-closed (EC) EEG was analyzed as a sensitivity condition to evaluate state dependence. Morlet time-frequency decomposition yielded log10 band power in delta, theta, alpha, beta, and gamma bands, summarized across whole-scalp and a priori regions of interest. SI burden was indexed using the ColumbiaSuicide Severity Rating Scale (C-SSRS) SI severity score modeled as an ordinal predictor. EO-only and EC-only models tested monotonic severity-power trends using robust standard errors, adjusting for age, sex, depressive symptom burden, number of comorbid diagnoses, and total exposure to stimulants, benzodiazepines, serotonergic psychedelics, and glutamatergic agents. Paired EO+EC models tested severity-by-eye-state interactions with participant-clustered standard errors. Within each severity level, C-SSRS intensity (1-25) was modeled with severity-by-intensity effects and eye-state dependence. False discovery rate (FDR) q-values are reported.

Results

In the pooled dataset, eleven participants reported no SI and among those with nonzero severity, C-SSRS severity spanned 1-5 (mean ± standard deviation; 1.83 ± 1.02) and intensity spanned 7-18 (13.10 ± 2.77). SI severity is positively associated with EO alpha power across global and regional summaries (whole-brain: β = 0.169, p < 0.001, q = 0.002; frontal: β = 0.134, p = 0.02, q = 0.008; central: β = 0.163, p < 0.001, q = 0.004; posterior: β = 0.196, p < 0.001, q = 0.003), as well as EO posterior beta power (β = 0.108, p < 0.001, q < 0.001). Severity effects differed by eye-state, with significant severity-by-eye-state interactions most prominent in whole-brain and frontal theta and posterior beta. Trialstratified analyses were directionally compatible with pooled estimates, including an EO posterior beta association in KET-CBT (p = 0.002, q < 0.05) and an EO posterior delta association in PSI-1V2 (p = 0.001, q < 0.05). Within-severity analyses identified a robust intensity effect in KET-CBT at severity level 2, where higher intensity predicted lower EC posterior beta power (p < 0.001, q > 0.55).

Conclusion

In TRD, baseline SI burden is associated with graded differences in resting EEG spectral power, most consistently in EO alpha and beta features, with robust eye-statedependent modulation in theta and beta bands. Severity and within-severity intensity exhibited partially distinct EEG associations, supporting longitudinal evaluation of whether treatment-related changes in EO spectral signatures track SI improvement and enhance SI stratification beyond symptom ratings.