Effects of a dissociative drug on fronto-limbic resting-state functional connectivity in individuals with posttraumatic stress disorder: a randomized controlled pilot study

Dissociation involves disruptions of consciousness, memory, identity and body awareness and can occur acutely after pharmacological challenge or chronically in disorders such as posttraumatic stress disorder (PTSD). Prior observational work has linked a dissociative subtype of PTSD to a pattern of "emotion overmodulation": increased prefrontal activation and reduced limbic engagement during symptom provocation, and elevated resting-state functional connectivity (RSFC) between medial prefrontal cortex (mPFC) regions and the amygdala. However, those findings derive from group comparisons and are therefore limited in establishing causal links between dissociation and fronto-limbic connectivity. Danböck and colleagues set out to test causality by experimentally inducing dissociation with a subanesthetic ketamine infusion and measuring its acute effects on RSFC between the amygdala and mPFC subregions. The trial compared ketamine (0.5 mg/kg over 40 min) to an active control, midazolam (0.045 mg/kg over 40 min), in individuals with PTSD and primarily without chronic dissociative symptoms, with RSFC assessed at baseline and during infusion. The authors hypothesised that ketamine would produce a stronger increase in amygdala–mPFC connectivity than midazolam, consistent with the emotion overmodulation model of dissociation.

This was a double-blind, randomised, controlled pilot study embedded in a larger trial. Twenty-eight participants meeting DSM-5 criteria for PTSD were randomised to receive either ketamine or midazolam; 26 participants completed the resting-state fMRI infusion session and constitute the analysed sample (ketamine n = 12; midazolam n = 14). Key exclusion criteria included lifetime bipolar disorder, borderline personality disorder, schizophrenia or schizoaffective disorder, current psychotic symptoms, moderate-to-high recent substance use disorder, significant traumatic brain injury, dementia, current suicide risk, or acute medical illness. The study procedures were IRB-approved and participants gave written informed consent. Participants underwent a 10-minute baseline resting-state scan (9:40 min acquisition) and a 40-minute resting-state scan during the intravenous infusion. To improve comparability between baseline and infusion and to capture temporal dynamics, infusion data were analysed in 10-minute segments (first, middle, last). Four participants had deviations in infusion timing (three ketamine, one midazolam) that shortened infusion to a minimum of 30 minutes, and for one participant infusion had begun ~7 minutes before the scan; the authors retained as much data as possible by extracting the first/middle/last 10-minute segments per individual and excluding segments with partially missing data. MRI was acquired on a Siemens 3T Prisma with a 32-channel head coil. Structural MPRAGE and multiband EPI functional scans were obtained (functional: TR = 1000 ms, TE = 30 ms, 2 mm isotropic voxels). Preprocessing used FMRIPrep v1.5.8, spatial smoothing (6 mm FWHM), and voxelwise denoising via nltools: regressors included average CSF and white matter signals, framewise displacement, six motion parameters plus their squares and derivatives, identified spike regressors, and linear and quadratic trends. The first five TRs of each sequence were excluded. Time series were extracted for bilateral non-overlapping functional parcels representing amygdala, ventromedial PFC (vmPFC), dorsomedial PFC (dmPFC), and anterior-medial PFC (amPFC), derived from meta-analytic coactivation maps. Static RSFC per segment was estimated as Spearman correlations between each mPFC parcel and the amygdala, Fisher z-transformed for analysis. For descriptive purposes, dynamic RSFC was visualised using the timecorr toolbox with a Laplace kernel (width 20). Statistical analysis used Bayesian multilevel regression models (brms in R) to test whether ketamine produced a stronger change in RSFC from baseline to infusion than midazolam. Three separate Gaussian models were fit for vmPFC–amygdala, dmPFC–amygdala, and amPFC–amygdala connectivity. Predictors were group (midazolam = 0, ketamine = 1), segment (categorical: baseline, first, middle, last 10 min), and their interaction; a random intercept accounted for repeated measures within subjects. Results are reported as regression coefficients (b), 89% credible intervals (CIs), and posterior probabilities of b being greater or less than zero (PP b>0, PP b<0). The authors considered effects significant when the 89% CI did not include zero. Model convergence diagnostics indicated acceptable Rhat and effective sample size.

Baseline comparisons indicated no group differences in RSFC for any of the examined mPFC–amygdala pairs. vmPFC–amygdala: Contrary to the pre-registered hypothesis, ketamine did not increase vmPFC–amygdala connectivity relative to midazolam. Instead, the ketamine group showed a greater transient decrease in vmPFC–amygdala RSFC from baseline to the middle 10-minute infusion segment compared with midazolam (interaction b = -0.14, 89% CI = [-0.26, -0.01]; PP b<0 = 96%). During that middle segment, vmPFC–amygdala RSFC was lower in the ketamine than the midazolam group (b = -0.13, 89% CI = [-0.24, -0.02]; PP b<0 = 97%). Within-group baseline-to-middle changes did not reach significance for either group (ketamine: b = -0.07, 89% CI = [-0.16, 0.02]; midazolam: b = 0.07, 89% CI = [-0.02, 0.15]). No above-threshold evidence was found for group differences in change from baseline to the first or last 10-minute infusion segments. dmPFC–amygdala: There were no notable group differences at baseline (b = 0.04, 89% CI = [-0.09, 0.18]) and no above-threshold evidence for group × segment interactions for the first, middle, or last infusion segments (first: b = -0.11, 89% CI = [-0.26, 0.05]; middle: b = -0.10, 89% CI = [-0.25, 0.05]; last: b = 0.02, 89% CI = [-0.13, 0.17]). amPFC–amygdala: Similarly, no baseline differences were observed (b = -0.03, 89% CI = [-0.19, 0.12]) and no above-threshold evidence for group differences in RSFC change from baseline to any infusion segment (first: b = 0.07, 89% CI = [-0.06, 0.19]; middle: b = -0.04, 89% CI = [-0.17, 0.08]; last: b = 0.09, 89% CI = [-0.04, 0.22]). In sum, the main measurable effect was a transient reduction in vmPFC–amygdala coupling during the middle portion of ketamine infusion relative to midazolam; other mPFC–amygdala connections showed no reliable drug-related changes.

The authors interpret these pilot results as evidence that acute pharmacological induction of dissociation with ketamine does not produce the predicted increase in fronto-limbic RSFC associated with the dissociative PTSD subtype. Rather, ketamine produced a transient decrease in vmPFC–amygdala connectivity during the middle 10 minutes of infusion. They note that this finding contrasts with prior correlational work that associated dissociation with increased top-down (vmPFC→amygdala) connectivity and suggest several reasons for the discrepancy. One major distinction is study design: observational comparisons between individuals with and without chronic dissociative symptoms may reflect shared aetiology or consequences of prolonged dissociation, whereas an acute pharmacological manipulation may reveal transient connectivity patterns directly linked to the dissociative state as it unfolds. The authors also highlight contextual differences: naturalistic dissociation is often triggered by aversive stimuli, cognitive overload, or tiredness, and those triggers themselves can modulate fronto-limbic circuitry. A pharmacological induction isolates the drug's effects from these contextual influences, so results need not match patterns seen during real-life dissociation. Neuropharmacology offers another possible account: naturally occurring dissociation might involve opioid mechanisms whereas ketamine acts primarily via glutamatergic (NMDAR) pathways; therefore, different neurotransmitter systems could produce qualitatively different neural signatures. The authors also point out that ketamine-induced dissociation may be less intense than real-life dissociation and that standard dissociation scales might not fully capture ketamine's subjective effects. Several limitations acknowledged by the authors temper interpretation. Dissociation was not rated during or immediately after the infusion scan, so it is unclear whether the observed vmPFC–amygdala decoupling temporally coincided with subjective dissociation. Using midazolam as an active control preserves blinding but constrains conclusions to relative drug effects; an inert placebo comparator would be needed to isolate ketamine-specific changes. Blinding was not formally assessed. Baseline scans were acquired with eyes open while infusion scans were eyes closed, which could interact with drug effects even though such eye-state differences mainly affect sensory networks not studied here. The authors did not apply additional correction for multiple comparisons beyond their pre-specified tests, and the sample size was modest, limiting power and generalisability. Nonetheless, they argue that small, rigorous pharmacological neuroimaging studies can provide useful discovery data that should be followed up in larger samples. Overall, the authors conclude that fronto-limbic alterations in dissociation may be context-dependent: dissociation can be associated with either increased or decreased amygdala–prefrontal coupling depending on how the state is induced and other boundary conditions, and that prefrontal involvement in dissociative phenomena remains an important area for further study. They recommend future research comparing multiple induction methods, including behavioural and pharmacological approaches, collecting time-resolved dissociation ratings, and testing inert control conditions to delineate mechanisms and boundary conditions of dissociation in PTSD.

This pilot randomised-controlled study is, to the authors' knowledge, the first to examine the acute effects of ketamine on fronto-limbic RSFC in individuals with PTSD. The findings do not support the hypothesis that induced dissociation necessarily corresponds to increased prefrontal–amygdala coupling (emotion overmodulation); instead, ketamine produced a transient decrease in vmPFC–amygdala connectivity in this sample. Divergent results across studies may reflect differences in experimental design, the circumstances eliciting dissociation, or distinct neurotransmitter systems involved. The authors propose further experimental comparisons of dissociation induction methods and the inclusion of dissociation ratings to clarify mechanisms and boundary conditions.