In adults with major depressive disorder and chronic suicidality, a 6‑week low‑dose oral ketamine regimen produced a transient reduction in centro‑parietal alpha power immediately post‑treatment that returned toward baseline by four‑week follow‑up, while theta and low‑beta bands showed delayed changes emerging after treatment cessation. Decreases in occipital theta correlated with improvements in depression and stress scores, suggesting EEG spectral shifts may track clinical response.
Papers cited by this study that are also in Blossom
Background
Ketamine has considerable therapeutic potential in alleviating major depressive disorder and chronic suicidality. However, the clinical diagnosis of neuropsychiatric disorders requires more robust diagnostic criteria. Electroencephalography (EEG) has shown promise in classifying depressive and suicidal patients from healthy individuals. The present study aimed to identify changes in the spectral properties of EEG in patients with major depressive disorder and chronic suicidality after completing the 6-week Oral Ketamine Trial on Suicidality with follow-up occurring 4 weeks after final ketamine treatment and determine associations between EEG spectral output and clinical symptoms.
Methods
Participants (n = 25) had 4-minute eyes closed resting state EEG recorded at frontal, temporal, centro-parietal, and occipital regions. Spectral analysis was performed with Welch’s power spectrum density method, and the power of 4 distinct frequency bands was analyzed: theta, alpha, low-beta, and high-beta. Correlation analyses between changes in clinical symptoms and spectral power were conducted using Spearman’s ranked correlation.
Results
Between pre- and posttreatment, only centro-parietal alpha power decreased. Between posttreatment and follow-up, centro-parietal alpha increased again in addition to increases in temporal alpha, centro-parietal and temporal theta, and occipital low-beta and decreases in occipital theta and temporal low-beta. Additionally, the decrease of occipital theta positively correlated with clinical subscales for depression and stress.
Conclusions
EEG spectral analysis revealed significant changes in theta, alpha, and low-beta frequency bands. Alpha band showed initial changes after treatment; however, this trended back toward baseline levels after the treatment cessation. In contrast, theta and low-beta showed significant power changes only after the treatment had ended.
Major depressive disorder (MDD) is highly prevalent and a leading contributor to suicide risk, yet clinical diagnosis and prognosis still rely largely on symptom reports and clinical screening. To improve objective characterisation of MDD and suicidality, prior research has explored neurophysiological biomarkers. Electroencephalography (EEG) — especially spectral (frequency-band) analysis of resting-state signals — has shown promise for distinguishing depressed or suicidal individuals from healthy controls and for tracking effects of antidepressant interventions, including ketamine. This study aimed to characterise changes in EEG spectral power following a 6-week course of low-dose oral ketamine in adults with MDD and chronic suicidality, and to test whether EEG changes associate with clinical outcomes. Based on earlier ketamine-EEG work, the investigators hypothesised that at the end of treatment (Post) versus baseline (Pre) there would be: i) decreased theta power across regions, ii) decreased alpha power in the centro-parietal region, and iii) decreased low- and/or high-beta power in the centro-parietal region; they further predicted that at 4-week follow-up (FUP) EEG power would trend back towards pre-treatment values alongside clinical measures.
This report used data from the open-label Oral Ketamine Trial on Suicidality (OKTOS). Participants received six weekly oral, sub-anaesthetic ketamine doses over six weeks with flexible titration from 0.5 up to 3.0 mg/kg. Assessments occurred at three principal timepoints: Pre (up to 14 days before the first dose), Post (4–7 days after the final dose) and FUP (28–32 days after the final dose). Of 32 trial completers, 25 participants were included in the EEG analyses because of withdrawals or missing EEG sessions. The analysed cohort comprised 11 males and 14 females aged 22–71 years (mean 46.41, SD 14.12). All participants had DSM‑5 MDD and chronic suicidality; most (92%) were taking psychotropic medications and 76% had additional psychiatric comorbidities. Resting-state EEG comprised 4 minutes of eyes-closed recording using a 32-channel BioSemi ActiveTwo system sampled at 1024 Hz. Channels were grouped into four regions for analysis: frontal, temporal, centro-parietal and occipital (channel lists provided in the paper). Pre-processing used an in‑house pipeline (EEG‑pyline) and standard steps: 0.5–30 Hz band-pass filtering, removal of ocular artefacts via signal-space projection using recorded EOG channels, segmentation into consecutive 5-second epochs, automated epoch/channel rejection with Autoreject, and visual screening of global field power. Delta band (1–3.9 Hz) was excluded because it failed a reliability check across epochs. Spectral analysis estimated power spectral density (PSD) with Welch’s method (2-second Hamming window, 50% overlap). PSD values (µV2/Hz) were averaged for theta (4–7.9 Hz), alpha (8–12 Hz), low-beta (12.1–18 Hz) and high-beta (18.1–30 Hz) bands for each region. The primary clinical outcome was change in suicidality measured by the Beck Scale for Suicide Ideation (BSS); secondary measures were the DASS‑21 subscales (depression, anxiety, stress). Paired comparisons across timepoints used the Wilcoxon signed‑rank test due to non-normal distributions. Associations between change scores on EEG bands and clinical measures were tested with Spearman rank correlations (two‑tailed; significance p < 0.05).
A c c e p t e d M a n u s c r i p t
Our study is the first to examine electrophysiological changes in adults with major depression and chronic suicidality following a low-dose ketamine treatment with oral administration. Whilst previous studies have confirmed the feasibility and tolerability of oral ketamine as a treatment for chronic suicidality, the present study reports significant changes in the spectral properties of EEG. Alpha band power displayed changes throughout the trial, during the active treatment phase and after the treatment had ended. Theta and low-beta band powers showed changes only after the treatment period had ended, between post-treatment and later follow-up. Theta band was also correlating the most with clinical symptoms. A c c e p t e d M a n u s c r i p t
Major depressive disorder (MDD) is one of the most prevalent and disabling psychiatric disorders globally, affecting one in six adults in their lifetime. Besides the significant burden conferred by MDD alone, it is a common co-morbidity to physical and mental disorders, and importantly, it is associated with an increased risk of suicide. Clinically, suicidality is characterised by recurrent and intrusive contemplations, wishes, and preoccupations with death and suicide. Annually, more than 700,000 deaths in the world are attributed to suicide, of which the high prevalence of MDD substantially contributes to the 10.6 and 7.2 years of life lost in men and women, respectively. Currently, the clinical diagnosis of MDD and suicidality rely on self-report measures in combination with clinical screening. To establish more robust diagnostic criteria that can address the significant variability in disorder pathogenesis and presentation, considerable research effort has been focused on finding biomarkers for neuropsychiatric disorders such as MDD and suicidality (for reviews, see. The spatio-temporal profile of neural circuit dynamics, and information processing within the brain, is predominantly defined by the dynamic balance of excitatory and inhibitory transmission of neurons). An emergent property of excitatory and inhibitory neural transmission that lends itself to measurement is the synchronised activity that arises from populations of neurons. Electroencephalography (EEG) is a non-invasive tool frequently used to measure these brain activity signals, which owing to technological advancements, has garnered renewed interest as a valuable method for determining biomarkers for depression and drug response. In recent clinical studies, EEG has shown promise in classifying depressive and suicidal patients from healthy individuals, and in assessing ketamine as a treatment for depression. Ketamine, an NMDA receptor (NMDAr) non-competitive antagonist, has shown considerable therapeutic potential in alleviating MDD and suicidality at sub-anaesthetic, low-doses (0.5 -3.0mg/kg;. Mechanistically, these effects are associated with NMDAr-mediated changes in excitatory and inhibitory neurotransmission (i.e., glutamatergic and GABAergic activity, respectively) in the medial prefrontal cortex, anterior cingulate cortex, and hippocampus, driving a period of plasticity-induced structural and functional remodelling. One of the most common techniques to acquire information from EEG data is to transform the signals from time domain to frequency domain and describe the signals using well-known frequency bands: delta, theta, alpha, beta, and gamma (de Aguiar Neto and Rosa, 2019). However, delta and gamma bands can be artifactual. This method, termed band power or spectral analysis, of both global and region-specific EEG signals formed from cortical and sub-cortical circuits has yielded some valuable insights into the effect of ketamine on neural circuit dynamics. A study conducted bycompared EEG power spectra between 111 depressed and 526 healthy subjects, reporting increased power in theta, alpha, A c c e p t e d M a n u s c r i p t and beta bands at parietal and occipital regions for the depressed group. Moreover, alpha band power has been found to be greater in frontal and parietal regions for unmedicated MDD patients compared with healthy controls. In addition, increased relative theta band power has been found in healthy individuals with higher suicidality in frontal and central regions. Abnormally higher theta, alpha, and beta band powers could display an imbalance of excitatory and inhibitory transmission in neural circuits which thereby leads to a psychiatric state or disorder. A randomised double-blind trial including participants with treatment-resistant MDD found that intravenous sub-anaesthetic dosage (0.5mg/kg) of ketamine decreased power in theta, alpha (parietal region), and beta (centro-parietal region) bands immediately post-infusion while decreasing depressive symptoms and suicidality; in contrast, 240 minutes after the infusion the power increased in all the bands towards the baseline values (de la. Another study byfound alpha band power to be increased at frontal regions 240-250 min after sub-anaesthetic ketamine infusion for participants with treatment-resistant depression. In a recent randomised, double-blind, active placebo-controlled crossover trial, resting state EEG was recorded during infusion of ketamine (subanaesthetic dose) or active placebo (remifentanil) in 30 participants with MDD and the observed spectral changes included an increase in theta and decreases in delta, alpha, and beta band powers, but none of these changes predicted treatment response. This present study aimed to identify changes in the spectral properties of EEG in patients with chronic suicidality (and diagnosed with MDD) after a 6-week low-dose oral ketamine treatment and determine if there is an association between EEG spectral output and clinical symptoms. Considering previous findings, it was hypothesised that compared to pre-treatment, EEG patterns at the post-treatment (6-week) timepoint would show a significant i) decrease of theta band power in all regions (frontal, temporal, centro-parietal, occipital), ii) decrease of alpha band power in centro-parietal region, and iii) decrease of low-and/or high-beta band power in centroparietal region. Four weeks after the final ketamine treatment (i.e., follow-up at 10-week), it was predicted that the direction of changes in EEG power spectra would trend back towards the pretreatment values (i.e., increase of power in the aforementioned bands at the same regions) and the pattern of clinical symptoms would follow suit.
Create a free account to open full-text PDFs.
Can, A. T., Hermens, D. F., Dutton, M. et al. · Translational Psychiatry (2021)
Dutton, M., Can, A. T., Beaudequin, D. et al. · Journal of Affective Disorders (2022)
Papers in Blossom that reference this study
Brown, R. E., Lissemore, J. I., Shinozuka, K. F. et al. · Journal of Affective Disorders (2026)
Alpha band: The principal, pre-specified finding was a decrease in centro‑parietal alpha power from Pre to Post. Between Post and FUP, centro‑parietal alpha power increased again toward baseline. Temporal alpha also increased between Post and FUP; however, the Pre-to-Post decrease at the temporal region was not statistically significant. Overall, no significant difference in alpha was observed when comparing Pre to FUP, consistent with a transient modulation during the treatment period. Theta band: No significant theta power changes emerged immediately after treatment (Pre vs Post). Significant changes appeared between Post and FUP: occipital theta power decreased, while temporal and centro‑parietal theta power increased. The paper notes these delayed theta changes differ from some prior ketamine studies that measured EEG shortly after infusion, reflecting the longer post-treatment interval used here. Beta bands: High‑beta showed no statistically significant differences across any of the timepoint comparisons. Low‑beta exhibited region‑specific Post-to‑FUP changes: a decrease at the temporal region and an increase at the occipital region. The abstract and discussion also report an increase in occipital low‑beta between Post and FUP. Clinical correlations: Changes in EEG power did not correlate with the primary clinical outcome (BSS) for any timepoint pair. Several significant correlations were found between EEG changes and the DASS‑21 subscales. Notable examples (timepoint pair, region, band, Spearman r, p): DASS‑Depression (Post–FUP) with occipital theta r = 0.431, p = 0.032; frontal low‑beta r = -0.464, p = 0.019; frontal high‑beta r = -0.571, p = 0.003. DASS‑Anxiety showed correlations such as Pre–Post occipital theta r = -0.522, p = 0.007 and Pre–FUP temporal alpha r = -0.455, p = 0.022. DASS‑Stress yielded several correlations including Pre–Post occipital theta r = -0.659, p = 0.0003, Pre–Post temporal theta r = -0.585, p = 0.002, and Post–FUP occipital theta r = 0.526, p = 0.007. The extracted text lists multiple other region‑ and band‑specific correlations with DASS subscales (many in the moderate range, p < 0.05). Additional results and sample characterisation: From the larger trial, responders (defined by BSS change) numbered 17 (68%) and non‑responders 8 (32%); prolonged responders were 11 (44%) and prolonged non‑responders 14 (56%), but subgroup EEG analyses were not reported. The study did not identify delta‑band results due to reliability concerns. High‑beta produced no significant timepoint effects.
Anijärv and colleagues interpret the principal finding as support for a transient modulation of alpha power by low‑dose oral ketamine: centro‑parietal alpha decreased during the active treatment phase and reverted toward baseline by the 4‑week follow‑up. The authors note this pattern is consistent with prior ketamine EEG work that observed short‑lived alpha effects, and they therefore characterise the alpha changes here as transient. Theta-band results were more complex. Changes in theta power did not appear immediately after the treatment but emerged between Post and FUP, with occipital theta decreasing and temporal/centro‑parietal theta increasing. The investigators propose two tentative interpretations: the occipital theta decrease could reflect delayed positive neuroplastic effects of ketamine, whereas increases in other regions might indicate remission of treatment effects and return of depressive neurophysiology. Because these patterns are mixed and differ from studies that measured EEG soon after infusion, the authors describe theta findings as inconclusive and in need of replication. Regarding beta activity, high‑beta showed no reliable changes, while low‑beta exhibited regional Post‑to‑FUP shifts. The discussion links beta‑band activity to cognitive constructs such as sense of agency and to anxiety, suggesting possible pathways by which ketamine's clinical effects might be mediated, but the authors emphasise that these mechanistic inferences are preliminary. On clinical associations, none of the EEG changes correlated with the primary suicidality measure (BSS), but multiple correlations were found with DASS‑21 subscales, notably several moderate correlations involving occipital theta and various beta measures. The authors caution that the absence of BSS correlations and the pattern of DASS associations leave the utility of resting EEG as a biomarker for ketamine’s long‑term antidepressant or anti‑suicidal effects unresolved. The investigators acknowledge several limitations that temper inference: the relatively small sample (N=25 for EEG analyses), the open‑label design without healthy controls or placebo/active comparator, heterogeneity introduced by concurrent psychotropic medications, variable daytime timing of EEG recordings, and the focus on chronic suicidality which may not generalise to other populations. They conclude that the results show significant EEG spectral changes after long‑term low‑dose oral ketamine in this cohort but argue that randomised controlled trials with larger samples and comparator groups are required to establish whether EEG can serve as a robust biomarker of sustained clinical response.
This study formed part of a larger open-label clinical trial, Oral Ketamine Trial on Suicidality (OKTOS), conducted at the Thompson Institute between August 2018 and November 2019. The intervention consisted of 6 weeks of flexible-dose (titration from 0.5 up to 3.0 mg/kg) treatment with oral ketamine and a 4-week follow-up phase without the treatment (nil ketamine). All participants received a total of six oral, sub-anaesthetic doses of ketamine for 6 weeks: one dose per week. The trial had three major timepoints (Fig(1) 'pre-treatment' or 'Pre' (i.e., up to 14 days prior to first treatment); (2) 'post-treatment' or 'Post' (i.e., 4-7 days after final ketamine treatment); (3) 'follow-up' or 'FUP' (i.e., 28-32 days after final ketamine treatment).
The trial included a broad range of clinical, neurophysiological, neuroimaging, and biochemical measures (i.e., MRI, EEG, blood tests, urinalysis, clinical scales). This study utilised data only from EEG recordings and two clinical scales. The primary clinical outcome measure was reduction in suicidality with ketamine treatment and this was determined by Beck Scale for Suicide Ideation (BSS). Secondary clinical measures used in this study included the Depression Anxiety Stress Scale 21 (DASS-21). In total, assessments at each major timepoint took approximately 8 hours. The clinical scales and biophysical data (i.e., blood tests, urinalysis) were the first items to be collected and remaining measures were acquired based on the availability of shared resources, and thus, not collected in any specific order. For this reason, EEG data acquisition occurred at variable times during the day between 8:00 am -6:00 pm. Participants were offered fuel vouchers to assist in the travel to their assessments and weekly treatments at the Thompson Institute if requested.
Ethics approval was obtained through the Bellberry Limited Ethics Committee (2017-12-982) and ratified by the University of the Sunshine Coast Human Research Ethics Committee (A181101). All study activities were conducted in accordance with the Australian National Statement on Ethical Conduct in Human Research () and conform with the Declaration of Helsinki ethical principles for medical research involving human subjects. This study was registered on the Australian Clinical Trials Registry:. The anonymised data that support the findings of this study are available from the corresponding author upon reasonable request.
Of the 32 participants who completed the trial, 25 were included in the current study due to participant withdrawal (n=2 after week 6) or missing electrophysiological recording session(s) (n=1 for Pre, n=1 for Post, n=3 for FUP). These 25 participants included 11 males and 14 females who were aged from 22 to 71 (mean = 46.41, standard deviation = 14.12). The participants were suffering from chronic suicidality and were referred to the trial by their local general practitioners. All participants had a diagnosis of MDD (DSM-V) and 76% had co-morbid mental disorders, 92% reported ongoing use of psychotropic medications (seefor details).
Participants were seated in a quiet, dimly lit room. Four minutes of eyes closed resting state EEG data was acquired according to standard pharmaco-EEG procedureswith BioSemi ActiveTwo 32-channel system (Biosemi B.V, Amsterdam, Netherlands) at a sampling rate of 1024 Hz. Scalp electrodes (Ag/AgCl active electrodes impedances < 40 kΩ) were localised according to the international 10/20 layout. Six additional electrodes were placed including two (left and right) mastoids and four electrooculographic (EOG) channels for obtaining horizontal and vertical eye movements. A c c e p t e d M a n u s c r i p t
An in-house EEG data processing pipeline, EEG-pyline, was used for signal pre-processing and spectral analysis efforts. The specific code (i.e., Jupyter notebook) used for this study can be found in the 'studies' folder within the GitHub repository.
All signals were filtered with 0.5-30 Hz band-pass filter (FIR with Hamming window, one-pass, zerophase, non-causal). EOG artefacts, including eye movements, were removed by computing signalspace projection (SSP) vectors using specific EOG channels acquired during EEG recording and applying these projections to the EEG signal to remove the artefacts. The previously mentioned steps were done using the MNE package for Python. The time series was divided into equal-sized consecutive 5-second epochs without any overlap between epochs. Epochs were cleaned from artefacts in all channels using the Python package Autoreject. The resulting EEG signals for each participant were screened visually using global field power plots to check whether the magnitude of a signal is in similar scale throughout the whole signal, thus ensuring the removal of large artefacts.
The processed EEG signals were transformed into the frequency domain by estimating power spectrum density (PSD) using Welch's method, with a 2-second Hamming window (50% overlap). Average PSD values (µV 2 /Hz) were calculated for four frequency bands: theta (4-7.9 Hz), alpha (8-12 Hz), low-beta (12.1-18 Hz), and high-beta (18.1-30 Hz). All the 32 channels were averaged into four brain regions: frontal (Fp1/2, AF3/4, F3/4, F7/8, Fz), temporal (FC5/6, T7/8, CP5/6, P7/8), centro-parietal (FC1/2, C3/4, Cz, CP1/2, P3/4, Pz), and occipital (PO3/4, O1/2, Oz). Additionally, signal reliability was checked for each band by calculating z-scores using median and absolute median deviation across all the epochs to determine if the PSD values fluctuated across time. As the participants were recorded in resting state, the power spectra were expected to not change more than two absolute median deviations across epochs. Delta band (1-3.9 Hz) did not pass this reliability test within most of the subjects, thereby was not included in further analysis.
Correlation coefficients for changes between timepoints (i.e., Pre, Post, and FUP) on two selfreported measures; the BSS and DASS-21, and EEG power at all four frequency bands were calculated. DASS-21 includes three different subscales: DASS-D for depression, DASS-A for anxiety, and DASS-S for stress. For calculating the coefficients, Spearman correlation was used due to the non-parametric distribution of the data. Thresholds for the coefficients were set as the following magnitudes: ± 0.3-0.5 for low correlation; ± 0.5-0.7 for moderate correlation; and ± 0.7-1.0 for high correlation. A c c e p t e d M a n u s c r i p t
To evaluate EEG power differences across different timepoints (i.e., Pre, Post, and FUP), a Wilcoxon signed-rank test was used. The non-parametric test was required since the EEG data did not meet the parametric assumptions (i.e., not normally distributed), and participants were compared to themselves across timepoints (i.e., paired samples). For statistical testing of the Spearman correlation coefficients, two-tailed t-test distributions were used, with significance set at p < 0.05. The statistical analysis and data visualisation were performed with the support of Pandas, NumPy, SciPy, Matplotlib, Seabornpackages for Python. The EEG data is summarised and presented by measures of central tendency and dispersion -both by median (M) with interquartile range (IQR) and mean with standard deviation (SD).
Theta band displayed significant changes only between post-treatment and follow-up timepoints..3)
When comparing high-beta band power between pre-treatment-vs-post-treatment, post-treatmentvs-follow-up, or pre-treatment-vs-follow-up, no statistically significant differences were found.)
None of the power spectra changes within the four bands correlated with change in BSS among the three timepoint pairs (i.e., Pre-Post, Post-FUP, Pre-FUP). However, the secondary clinical outcome measure, DASS-21, showed several correlations with the EEG spectral data. Change in DASS-D correlated with Post-FUP change in occipital theta (r = 0.431, p = 0.032; Fig 6 ); frontal low-beta (r = -0.464, p = 0.019); frontal high-beta (r = -0.571, p = 0.003), and with Pre-FUP change in occipital lowbeta (r = -0.421, p = 0.036); temporal high-beta (r = -0.504, p = 0.010); and centro-parietal high-beta (r = -0.420, p = 0.036). Change in DASS-A correlated with Pre-Post changes in occipital theta (r = -0.522, p = 0.007) and with Pre-FUP changes in centro-parietal theta (r = -0.404, p = 0.045); temporal alpha (r = -0.455, p = 0.022); and temporal high-beta (r = -0.443, p = 0.027). Change in DASS-S correlated with Pre-Post changes in temporal theta (r = -0.585, p = 0.002); occipital low-beta (r = -0.506, p = 0.010); occipital theta (r = -0.659, p = 0.0003); temporal low-beta (r = -0.528, p = 0.007), and with Post-FUP changes in occipital theta (r = 0.526, p = 0.007, Fig), and with Pre-FUP changes in temporal alpha (r = -0.402, p = 0.046); temporal high-beta (r = -0.595, p = 0.002); and occipital high-beta (r = -0.399, p = 0.048). (Supplementary Tables
The main objective of this study was to explore spectral changes of EEG in patients with MDD and chronic suicidality following 6 weeks of low-dose oral ketamine treatment. Secondly, it was to investigate correlations between changes in EEG spectral results and changes in clinical measures of suicidality, depression, anxiety, and stress. The most conclusive finding was the hypothesised alpha band power change at centro-parietal region. Between pre-treatment and post-treatment timepoints, the alpha band power decreased at centro-parietal region. Following the final ketamine treatment between post-treatment and followup, alpha band power increased again at the same region. Similar changes were also reported in recent study investigating low-dose ketamine as treatment for depression. Additionally, our results showed alpha band power significantly increased between the post-treatment and follow-up timepoints at the temporal region. However, at that same region a decrease in power between pre-treatment and post-treatment was not statistically significant. As alpha band power has been found to be higher in depressive subjects compared to healthy controls, our findings suggest that low-dose oral ketamine decreases alpha band power at centro-parietal region, but the changes revert close to pre-treatment levels after ketamine treatment has ceased. We did not find any statistically significant change between alpha band power between pre-treatment and follow-up, which suggests the effects of ketamine are transient in terms of changes in alpha activity. For theta band, we did not find any power changes immediately following the treatment period (i.e., post-treatment), but there were significant changes between post-treatment and follow-up timepoints -power was decreased at the occipital region and increased at temporal and centroparietal regions. Interestingly, de la Salle et al. () reported an increase in theta band power following final ketamine treatment. However, these results are not directly comparable, as changes were measured 2 hours post-treatment, in comparison to 28-32 days following final ketamine treatment in OKTOS. Other clinical studies (which did not include ketamine treatment) have found theta power to be higher in participants with depression compared to healthy controls. For example,found theta power to be increased at parietal and occipital regions, whilereported this at frontal and central regions. Based on these studies, the occipital theta decrease observed here could represent a delayed positive treatment effect due to ketamine-induced neuroplastic changes. In contrast, increase in temporal and centro-parietal theta power would suggest the opposite -that treatment effects have ceased, and the depressive pathophysiology returns. Thereby, theta band power changes in our study are inconclusive, and more studies are needed to confirm the function of theta band power in depression and suicidality, and whether this is ameliorated by ketamine treatment. High-beta band showed no significant power changes across timepoints (Fig), but low-beta power between post-treatment and follow-up decreased at temporal region and increased at occipital region. Our findings for beta bands were not consistent with previous findings by de la; potentially due to the timing of measurements at post-treatment and follow-up. In terms of cognitive functions, a recent study found that low-beta (and alpha) band powers at parieto-occipital region reflect one's sense of agency, whereas distorted A c c e p t e d M a n u s c r i p t sense of agency has been suggested to be linked to several neuropsychiatric disorders including depression. Beta band activity has also been related to anxiety previously, which itself has been used to predict suicidal ideation; indicating an indirect relationship. Therefore, ketamine therapeutic effect on depression and suicidality reported here could be mediated by changes in beta band activity that influence sense of agency and/or symptoms of anxiety, but further studies are needed for more precise conclusions.
Whilst none of the power spectra changes correlated with the main clinical outcome (BSS), several significant relationships with the secondary measure (DASS-21) were found. One of the most relevant correlation findings was the theta band power at occipital region showed a moderate positive correlation with DASS-S and weak positive correlation with DASS-D between post-treatment and follow-up timepoints. This is relevant because the changes at occipital region for theta band power between post-treatment and follow-up were also statistically significant (Fig). There exists a paucity of research examining long-lasting electrophysiological effects of ketamine in suicidal population. For this reason, the relationship between ketamine's effect on both clinical and EEG measures remain unclear, but future studies with a larger sample size and randomised control study design should examine the utility of EEG as a biomarker in predicting long-term antidepressant response to ketamine. Additionally, the lack of relationship between EEG spectral changes and clinical symptoms could be due to low-frequency EEG oscillations having stronger correlation with psychotomimetic effects rather than antidepressant effects of ketamine. In this study, there were several limitations which may have impacted the results. Firstly, the sample size was relatively small, which may have impacted the results. Secondly, this study focused only on the whole group-level ketamine treatment effects without categorically examining the differences between (i) responders (n=17; 68%) and non-responders (n=8; 32%), according to BSS changes from Pre to Post timepoints; and (ii) prolonged responders (n=11; 44%) and prolonged non-responders (n=14; 56%), according to BSS changes from Pre to FUP timepoints. According toprolonged response' was defined by larger than 50% improvement in BSS score from Pre to Post/FUP timepoints or BSS score less than 6 in Post/FUP timepoint. Thirdly, the neurological effects of concurrent medications could not be excluded as most of the participants were taking various psychotropic medications throughout the trial. Fourthly, this study focused on chronic suicidality (i.e., patients with suicidality for at least six months), and there may be different outcomes if this period of duration changes. Finally, the current study design does not include healthy controls, placebo or active comparator groups for comparison which would be necessary to draw more robust conclusions regarding the spectral changes of EEG. This study found that long-term low-dose oral ketamine treatment caused significant changes in EEG power spectra in adults with MDD and chronic suicidality. However, further studies of low-dose oral ketamine in suicidality are needed using a randomised control study design and larger sample size to address the limitations of current open-label trial. A c c e p t e d M a n u s c r i p t