Blood biomarker changes and relationships after low dose oral ketamine treatment for post-traumatic stress disorder (PTSD)

Trauma exposure is common and roughly 10% of exposed people develop post-traumatic stress disorder (PTSD), characterised by re-experiencing, avoidance, negative affect and hyperarousal. Standard treatments include exposure-based psychotherapies and pharmacotherapy with selective serotonin reuptake inhibitors, but remission and recovery rates are limited. Ketamine, an N-methyl-D-aspartate (NMDA) receptor antagonist, has shown promise in treatment-resistant depression and PTSD, predominantly in intravenous (IV) studies. The authors note practical and cost advantages of low dose oral ketamine and indicate that their recent open-label oral ketamine trial for PTSD produced response rates comparable to IV studies, but emphasise that very little is known about the blood biomarker changes that accompany ketamine treatment in PTSD patients. L. and colleagues set out to begin filling this gap by analysing circulating levels of several candidate biomarkers before and after a six-week, low dose oral ketamine regimen in people with PTSD. The targets examined were brain-derived neurotrophic factor (BDNF), vascular endothelial growth factor A (VEGF-A), serotonin, FK506 binding protein 51 (FKBP51) and a panel of cytokines (IL-1β, IL-2, IL-4, IL-6, IL-12p70, IL-17A and TNFα). The stated aim was to identify biomarker changes across treatment and follow-up and to explore relationships between biomarker trajectories and clinical measures of PTSD and wellbeing.

This analysis used samples and data from the open-label Oral Ketamine Trial on PTSD (OKTOP). The clinical trial delivered low dose oral ketamine once weekly for six weeks in a titrating-up manner; participants were allowed to stay on previously prescribed psychotropic medications (ketamine acted as an augmentative treatment). Participants were eligible for the biomarker analysis if they had a DSM-5 PTSD diagnosis (confirmed with the Clinician-Administered PTSD Scale for DSM-5, CAPS-5) and available blood samples at baseline (pre-treatment) and at least one post-treatment timepoint. The cohort for this biomarker study comprised 25 participants, with sample counts of n = 25 pre-treatment, n = 19 post-treatment (~2 hours after the sixth/final dose), n = 24 at 1-week follow-up (1–7 days after final dose), n = 23 at 1-month follow-up (28–32 days after final dose) and n = 14 mid-treatment (~2 hours after the third dose) included for some analyses. Treatment response in the trial was defined as ≥ 50% reduction in PCL-5 score from pre-treatment. Clinical outcome measures collected at pre-treatment, post-treatment and follow-up timepoints included the PTSD Checklist for DSM-5 (PCL-5), the Depression, Anxiety and Stress Scale (DASS-21) and the WHO-5 Well-Being Index, in addition to CAPS-5 at baseline for eligibility. Non-fasting whole blood was drawn at midday/early afternoon into serum separator and EDTA tubes, processed to serum and plasma, aliquoted and stored at −80 °C; samples were processed promptly or kept at 4 °C for up to 3.5 hours before centrifugation. Biomarker quantification used multiplex ProcartaPlex/Luminex assays for BDNF and VEGF-A (assessed in both serum and plasma), high-sensitivity ProcartaPlex assays for the panel of cytokines (measured in plasma), and ELISAs for serotonin (serum) and FKBP51 (plasma). Limits of quantification and assay detectability were reported: cytokine detectability ranged roughly from 72% to 95% depending on the analyte, and FKBP51 was within detectable range in 66% of samples. Values below assay lower limits of quantification (LLOQ) were assigned half the LLOQ for statistical analyses. Statistical analyses were performed in R (v4.3.3). The Skillings–Mack test, a non-parametric repeated-measures test that accommodates missing data, was used to screen for biomarker changes across five timepoints. Linear mixed models (lme4) were used to assess change-from-baseline patterns, modelling participant as a random effect and timepoint as a fixed effect; change values were derived from log-transformed biomarker concentrations. Immune biomarker data were normalised and clustered to identify inflammatory subgroups; Spearman rank correlations were computed for biomarker–clinical relationships and p-values were adjusted for multiple comparisons with the Benjamini–Hochberg false discovery rate method.

Sample and assay performance: The analysed cohort included 25 participants with varying completeness of longitudinal samples (see Methods). Assay detectability varied by analyte: BDNF, VEGF-A and serotonin were within detectable ranges for all tested samples, whereas cytokine detection rates ranged from ~72% to 95% across targets and FKBP51 was detectable in 66% of samples. Values below LLOQ were imputed as half-LLOQ. Overall repeated-measures screening (Skillings–Mack) identified that plasma BDNF, serum VEGF-A, IL-2, IL-4 and IL-6 showed significant changes at some point during the trial. However, substantial inter-individual variability was evident across most biomarkers. Change-from-baseline mixed-model analyses focused on serum measures (BDNF and VEGF-A) because plasma results for these targets were not significant in more detailed analyses. Serum VEGF-A showed statistically significant decreases from baseline at multiple timepoints: mid-treatment (decrease 0.062 log10 pg/ml, p = 0.030), post-treatment (decrease 0.052 log10 pg/ml, p = 0.043), follow-up 1-week (decrease 0.065 log10 pg/ml, p = 0.007) and follow-up 1-month (decrease 0.050 log10 pg/ml, p = 0.038). Serum BDNF exhibited small but significant decreases from baseline at mid-treatment (decrease 0.216 log10 pg/ml, p = 0.010), follow-up 1-week (decrease 0.157 log10 pg/ml, p = 0.026) and follow-up 1-month (decrease 0.142 log10 pg/ml, p = 0.046). BDNF, VEGF-A and serotonin levels from plasma did not show significant change over treatment in the more detailed analyses. Cytokines and inflammatory clustering: Normalisation and clustering of immune biomarkers identified two stable inflammatory subgroups across timepoints: a “high” inflammatory group (n = 12) and a “low” inflammatory group (n = 13). The two groups differed significantly across most cytokines (IL-1β, IL-2, IL-4, IL-12p70, IL-17A and TNFα at all four major timepoints; IL-6 at pre-treatment and follow-up 1-week). Ketamine treatment did not appear to produce consistent changes in circulating cytokine levels overall, and inflammatory group membership did not correlate significantly with participant demographics, PTSD severity, comorbid depression or perceived stress, nor with treatment response in this cohort. Biomarker–clinical correlations: Several correlations between biomarkers and clinical scales were reported. At pre-treatment, serum BDNF correlated negatively with CAPS-5 scores (Spearman r = −0.487, p = 0.014; adjusted p = 0.080) and FKBP51 correlated positively with PCL-5 (r = 0.497, p = 0.016; adjusted p = 0.072); these adjusted p-values did not meet conventional significance thresholds after multiple-testing correction. Serum VEGF-A at post-treatment correlated negatively with PCL-5 (r = −0.509, p = 0.011; adjusted p = 0.039), a result that survived adjustment. Serotonin levels were positively correlated with WHO-5 wellbeing scores after ketamine treatment (post-treatment r = 0.632, p = 0.005; adjusted p = 0.022), with weaker and non-significant adjusted correlations at later follow-ups. BDNF and VEGF-A serum levels were positively correlated with each other across timepoints (examples include pre-treatment r = 0.415, p = 0.039, adjusted p = 0.149; post-treatment r = 0.691, p = 0.001, adjusted p = 0.006). The extracted text reports that serotonin and FKBP51 were negatively correlated at pre- and post-treatment, but the numerical statistics presented in the extraction appear internally inconsistent, so the exact strength and direction of those correlations are not clearly reported in the provided text. FKBP51 levels were consistently higher in the high-inflammatory group, with differences reaching statistical significance at pre-treatment and follow-up 1-month. Clinical response counts: The study reports responder/non-responder counts for the 25 participants by timepoint using the ≥ 50% PCL-5 reduction definition: post-treatment 15 responders / 3 non-responders / 7 missing, follow-up 1-week 15/8/2, and follow-up 1-month 12/11/2.

L. and colleagues present what they describe as the first broad blood biomarker analysis in people with PTSD undergoing ketamine treatment. The main reported biological findings were small but statistically detectable decreases in serum BDNF and serum VEGF-A after low dose oral ketamine, and a robust correlation between BDNF and VEGF-A across timepoints. The authors interpret these changes in the context of prior literature that links BDNF and VEGF-A to neurotrophic and neuroplastic processes implicated in antidepressant response and note preclinical evidence that ketamine’s sustained antidepressant actions require interactions between BDNF and VEGF-A. They suggest that the observed decreases in serum BDNF and VEGF-A after symptom improvement may reflect a normalising process in PTSD, where baseline BDNF is often elevated relative to other psychiatric disorders. The investigators also report distinct inflammatory subgroups within their PTSD cohort (roughly half with low-level systemic inflammation and half without), and they found that circulating cytokine concentrations did not change consistently across ketamine treatment and follow-up. FKBP51 emerged as associated with inflammatory group status and showed positive correlations with PTSD symptom measures at baseline. Serotonin levels were positively associated with self-reported wellbeing after treatment, and the authors note a novel reported relationship between circulating serotonin and FKBP51 that was present during active PTSD and immediately after treatment but not at later follow-up. The authors position these biomarker–clinical relationships as biologically plausible and worthy of further investigation. Key limitations are acknowledged: small sample size, incomplete sample sets for some participants, multiple comparisons that reduced statistical power after correction, the open-label (non-randomised, unblinded) trial design that may bias self-reported outcomes, and the allowance for concurrent psychotropic medications that could confound biomarker changes. The authors stress that several correlations lost significance after p-value correction and that many findings require replication in larger, independent cohorts with more comprehensive and sensitive biomarker panels. They also highlight the value of longitudinal, within-subject analyses given the substantial interpersonal variability observed in biomarker levels. Overall, the authors conclude that their results add preliminary evidence of serum BDNF and VEGF-A decreases after ketamine treatment in PTSD and identify several biomarker–clinical relationships (including serotonin, FKBP51 and inflammatory subgroup differences) that merit further study, while emphasising that definitive conclusions await larger, controlled investigations.