Ketamine Effects on EEG during Therapy of Treatment-Resistant Generalized Anxiety and Social Anxiety

Anxiety disorders such as generalized anxiety disorder (GAD) and social anxiety disorder (SAD) are common, often persistent, and frequently resistant to conventional treatments; one-third of SAD patients, for example, remain treatment resistant, producing substantial impairment and economic burden. Ketamine, an NMDA receptor antagonist, has shown rapid efficacy across several treatment-resistant neurotic conditions including depression, obsessive–compulsive disorder, and post‑traumatic stress disorder, but the neural mechanisms underlying its anxiolytic effects remain unclear. Prior human and animal work has reported mixed effects of ketamine on EEG rhythms (reductions in delta, theta and alpha, with increases in gamma in some studies), and neuroimaging and pharmacological evidence implicates glutamatergic dysfunction in SAD; these observations motivated the current study. Shadli and colleagues set out to examine concurrent changes in anxiety symptoms and frontal EEG in treatment-refractory GAD and SAD patients after double‑blind administration of three ascending ketamine doses and an active control (midazolam). The primary aim was to test whether ketamine produces dose-related symptom improvements together with specific EEG changes, in particular alterations in frontal alpha asymmetry and Higuchi's fractal dimension, and band‑specific power shifts (decreases in delta/alpha/beta and increases in theta/gamma) that might relate to clinical response.

This was a double‑blind, within‑subject study in which 12 patients with treatment‑refractory DSM‑IV GAD and/or SAD received three ascending subcutaneous ketamine doses (0.25, 0.5, and 1.0 mg/kg) and an active control, midazolam (0.01 mg/kg). Midazolam was randomly inserted into the dosing sequence and the ketamine doses were given in ascending order, with one week between sessions. The study had ethics approval and was prospectively registered; participants remained on their existing psychotropic medications but did not change doses or start new treatments during the trial. Inclusion criteria included age ≥18 years, HAM‑A ≥20 and/or LSAS ≥60, failure to respond to at least two antidepressant courses, and absence of current depression (MADRS <20). Exclusions covered severe medical conditions, pregnancy/lactation, certain medications (MAOIs, thyroxine, stimulants), and active suicidal ideation. Twelve participants (4 male, 8 female; mean age 31 years) completed the protocol; all had SAD, 10 had GAD, and two had panic disorder. Clinical assessments comprised the Fear Questionnaire (FQ) and the Hamilton Anxiety Rating Scale (HAM‑A) administered predose and at 1, 2, 24, 72, and 168 hours postdose; EEG was recorded only predose and at 2 hours postdose. Tolerability and dissociative symptoms were monitored, and vital signs were measured up to 2 hours postdose. For analysis, midazolam was treated analytically as equivalent to 0 mg/kg ketamine. EEG was recorded from frontal and central leads (Fp1, Fp2, F7, F3, Fz, F4, F8, Cz) referenced to the left mastoid, using a cap and a 256 Hz sampling rate, then downsampled to 128 Hz for offline processing. Participants alternated 1‑minute eyes‑open and eyes‑closed epochs for a total of 10 minutes. Processing included blink removal, artefact rejection, FFT with overlapping Hanning windows, log transformation of power, and averaging to yield band powers for delta (1–3 Hz), theta (4–6 Hz), alpha1 (7–9 Hz), alpha2 (10–12 Hz), beta (25–34 Hz), and gamma (41–53 Hz). Frontal alpha asymmetry (FAA) was computed as ln(R) − ln(L) for F8:F7 and F4:F3, and Higuchi fractal dimension (HFD) was calculated with kmax = 8. Statistical analysis used repeated‑measures ANOVA (channel, frequency, dose as within‑subjects factors) and polynomial contrasts; stepwise and multiple regression analyses assessed relationships between EEG changes and symptom change.

Twelve participants completed the protocol. Clinically, 8 of 12 patients (67%) showed >50% reduction in HAM‑A and/or FQ scores at 2 hours after the 0.5 or 1.0 mg/kg ketamine doses. The FQ showed a dose‑related improvement: ANOVA reported a significant dose effect (F(2.67,29.41) = 3.80, P = .024) and a significant linear trend (F(1,11) = 7.12, P = .022), with a quadratic trend approaching significance suggesting a possible ceiling or reduction at 1.0 mg (F(1,11) = 4.68, P = .053). By contrast, HAM‑A scores did not change significantly (all F ≤ 1.2, all P ≥ 0.3). The extracted text notes that only predose and 2‑hour postdose anxiety data are considered alongside EEG. EEG analyses indicated dose‑related decreases in low‑frequency power and increases in higher frequencies. Specifically, ketamine significantly but nonlinearly reduced delta power, and sometimes theta, at lateral frontal sites (F7, F4 and particularly F8), while higher frequencies (beta and especially gamma) tended to increase; the greatest low‑frequency reductions were observed at fronto‑polar sites and diminished more posteriorly. Changes in Higuchi fractal dimension were minimal, non‑significant, and opposite to the hypothesis, and there were no systematic changes in frontal alpha asymmetry for either alpha1 or alpha2 bands. Analyses linking EEG to clinical response identified theta changes at right frontal locations as the most relevant predictor. Stepwise regressions across sites and bands extracted change in theta power at F4 as the single significant predictor of FQ change. Forcing F3, Fz, F4, F8 and Cz into a multiple regression produced a model explaining 17% of variance in FQ change, about 5% more than F4 alone; within that model, a small unique contribution (3%) was attributed to contralateral F3, about 9% variance was shared among Fz, F4, F8 and Cz, and 5% was unique to F4, with F8 and Cz contributing no unique variance. The authors report that delta and gamma bands, though substantially altered by ketamine, showed little or no correlation with FQ improvement. A post‑hoc power note indicated >80% power at alpha = 0.05 for mean differences between ketamine 1 mg/kg and midazolam for theta and gamma bands with sample sizes ranging from 3 to 11 (per electrode/band comparisons reported).

Shadli and colleagues interpret their principal finding as a dose‑related reduction in right frontal theta (maximal at F4 and around 0.5 mg/kg) that appears to mediate ketamine’s rapid anxiolytic effect on phobic anxiety as measured by the FQ. They emphasise that although ketamine produced larger changes in other bands (notably decreases in delta and increases in gamma), these did not predict clinical improvement, suggesting a specific role for right frontal theta reductions in the therapeutic effect. The lack of change in frontal alpha asymmetry and in Higuchi fractal dimension ran counter to the investigators’ predictions. They note that FAA may reflect a trait vulnerability (related to aversion/withdrawal) rather than a state marker of anxiety, and that HFD has been more consistently linked to depression than to anxiety. The absence of an HFD or FAA effect with ketamine, and the absence of an HFD increase with midazolam, are discussed in this context. The authors consider consistency and discrepancies with prior literature: reductions in waking delta and increases in gamma are broadly similar to some earlier reports, but the reported waking delta decrease differs from findings of increased slow‑wave activity during sleep after ketamine; the investigators suggest this may reflect functional differences between waking and sleep delta or a rebound increase in sleep following an acute daytime reduction. They also attribute variability across studies to differences in dose, recording site, patient population and small samples. Methodological limitations are acknowledged explicitly: small sample size, lack of a true inert placebo (the study used an active control), and that measured blood drug levels were not analysed. They recommend further work with larger samples and carefully matched healthy controls. Finally, the authors relate their results to an anxiolytic biomarker they have developed (goal‑conflict rhythmicity), which is manifest in the theta band at right frontal sites; they suggest ketamine’s right frontal theta effect could reflect action on a similar brain system that is homologous to rodent anxiolytic‑related theta. They caution, however, that midazolam produced little effect in their patients and that it is therefore unclear whether ketamine is acting on the same theta‑dependent system as conventional anxiolytics or on a related but distinct right‑frontal mechanism. Overall, the investigators conclude that right frontal theta changes are a plausible biomarker of anxiolytic action and that their double‑blind patient data support preclinical and human evidence linking right frontal EEG changes to anxiolytic treatments.