In vivo effects of ketamine on glutamate-glutamine and gamma-aminobutyric acid in obsessive-compulsive disorder: proof of concept

Obsessive-compulsive disorder (OCD) involves frontostriatal circuit dysfunction and is commonly treated with serotonin reuptake inhibitors (SRIs), which typically require 2–3 months to produce partial benefit. Earlier research has increasingly implicated glutamatergic abnormalities in OCD pathophysiology, and there is growing evidence for GABAergic disturbances as well. Ketamine, an NMDA receptor antagonist that modulates glutamatergic signalling, has been shown in prior work to reduce OCD symptoms rapidly and robustly; however, the in vivo neurochemical effects of ketamine in people with OCD had not been characterised. Rodriguez and colleagues set out to measure dynamic changes in medial prefrontal cortex (MPFC) glutamate+glutamine (Glx) and gamma-aminobutyric acid (GABA) during intravenous ketamine versus saline infusions in medication-free adults with OCD. The primary hypothesis was that ketamine would increase MPFC Glx relative to saline; exploratory analyses examined whether ketamine would also affect MPFC GABA levels.

This proof-of-concept, randomized crossover study delivered two 40‑minute intravenous infusions (ketamine 0.5 mg/kg and saline) to each participant at least 1 week apart, with infusion order randomised. Seventeen adults with a primary DSM-IV diagnosis of OCD and Yale-Brown Obsessive-Compulsive Scale (YBOCS) ≥16 were enrolled; after excluding one extreme MRS outlier, 16 participants comprised the final MRS sample. Key exclusion criteria included comorbid major depression (HDRS-17 ≥25), bipolar, psychotic or eating disorders, recent substance abuse or dependence, current psychotropic medication use, pregnancy, and MRI contraindications. Diagnoses were confirmed with the SCID-IV and clinical interviews. Magnetic resonance data were acquired on a General Electric 3.0‑T scanner while participants lay supine. A 2.5 × 3.0 × 2.5 cm3 MPFC voxel (approximately 18.8 cm3) was positioned anterior to the genu of the corpus callosum, encompassing portions of Brodmann areas 24, 32 and 10 including the pregenual anterior cingulate cortex. Each scanning session lasted about 90 minutes and included a structural T1 SPGR plus six 13‑minute proton MRS frames (one baseline [0–13 min], three during the infusion, and two post‑infusion), with a 2‑minute gap between frames for automatic shimming. The J‑editing PRESS difference method optimised for GABA detection was used (TE/TR 68/1500 ms; 256 interleaved excitations per edited condition), and metabolite signals were expressed semi‑quantitatively as ratios to the unsuppressed voxel tissue water (Glx/W and GABA/W). Spectral data from the eight‑channel phased‑array coil were quality‑assessed, combined into time‑domain signals, and the editing on/off signals subtracted to yield difference spectra. Metabolite peaks were fitted in the frequency domain using a Levenberg–Marquardt nonlinear least‑squares routine modelling resonances as pseudo‑Voigt functions; goodness of fit metrics and visual inspection were used for quality control. Voxel tissue segmentation and metabolite correction were not performed because the principal focus was within‑subject changes over time. Clinical symptom measures included the OCD‑visual analogue scale (VAS) at multiple intra‑ and post‑infusion time points, baseline YBOCS, and HDRS‑17. Glx/W (primary) and GABA/W (exploratory) were analysed using mixed effects linear models with fixed effects for infusion type, infusion order, MRS data frame, and responder status, and random effects for subject and scan nested within subject. Where overall F‑tests reached p<0.05, post hoc tests examined individual 13‑minute acquisitions; these post hoc tests were treated as exploratory and were not corrected for multiple comparisons. Carryover was assessed with paired t‑tests on baseline Glx/W and GABA/W between first and second infusions, and no evidence of neurochemical carryover was found, permitting collapse across phases.

After excluding one extreme Glx/W outlier, 16 participants contributed viable MRS data. The final sample had substantial OCD severity (mean YBOCS 26, SD 4.2) and an average illness duration of 17.3 years (SD 9.7). Most participants (11/16; 69%) had no other psychiatric comorbidity; two had comorbid depression with baseline HDRS‑17 scores of 13 and 16. No participants were receiving medication or psychotherapy at the time of scanning. In the MPFC voxel, mixed model analyses showed no significant difference between ketamine and saline over time in Glx/W (F=0.65; df=6,141; p=0.689). By contrast, GABA/W showed a modest overall difference across the six successive 13‑minute acquisitions (F=2.16; df=6,146; p=0.048). Analyses restricted to the first treatment phase yielded similar directionality but did not meet the pre‑specified significance threshold for GABA/W (F=1.90; df=6,66; p=0.094). Post hoc tests identified a specific time window driving the GABA/W effect: GABA/W increased significantly at approximately 60–73 minutes post‑infusion in the ketamine condition compared with saline (t(146)=2.38, p=0.02) and compared with baseline (t(72)=2.22, p=0.03). The change in GABA/W from baseline to 60–73 minutes post‑infusion correlated positively with reductions in OCD symptoms as measured by the OCD‑VAS at multiple assessments during and up to 7 days after the infusion, with Pearson r values ranging from 0.46 to 0.63 and p‑values from 0.01 to 0.06. No significant correlations were found between GABA/W changes and dissociative or psychotic symptoms as measured by the Clinician Administered Dissociative Symptoms Scale and the Brief Psychiatric Rating Scale.

Rodriguez and colleagues report that a single subanaesthetic ketamine infusion did not increase MPFC Glx/W over time in unmedicated adults with OCD, contrary to their primary hypothesis and to some prior MRS studies in other populations. Instead, the principal neurochemical effect observed was an increase in MPFC GABA/W, driven predominantly by a single time point about 1 hour after infusion. The authors note that this pattern differs from some studies in healthy volunteers and depressed subjects that documented Glx or glutamate changes, and suggest that Glx changes in OCD may be smaller or less readily detected with the methods used. The increase in GABA/W is discussed in light of emerging evidence for GABAergic abnormalities in OCD, including transcranial magnetic stimulation findings and genetic associations implicating GABBR1. The authors note prior reports of baseline MPFC GABA deficits in unmedicated OCD, and propose that the observed ketamine‑related GABA change could reflect engagement of cortical GABAergic interneurons; this is consistent with animal and human work indicating ketamine can activate subpopulations of GABAergic interneurons and alter cortical oscillatory activity. The positive correlation between GABA/W change and acute and short‑term improvement in OCD symptoms is described as counterintuitive and needing replication; the investigators speculate that larger GABA/W changes during ketamine may index greater underlying GABAergic abnormalities that are engaged by the drug. Several limitations acknowledged by the authors temper the conclusions. The sample size was small, limiting statistical power and the ability to examine potential confounds such as gender. The J‑editing MRS method cannot separate glutamate from glutamine, so changes in one component could be obscured; whole‑tissue MRS may miss transient increases in synaptic glutamate release that do not alter tissue pools. The GABA peak includes contributions from mobile macromolecules, so ketamine‑induced macromolecular changes might influence the observed signal. Large voxel sizes required for reliable GABA quantification introduce variability in tissue composition, which could mask Glx changes. Although no neurochemical carryover was detected, clinical carryover cannot be entirely ruled out, and the psychoactive effects of ketamine make blinding difficult in crossover designs. To address these issues, the authors recommend larger samples, parallel‑group designs with active controls that mimic dissociative effects, and MRS methods capable of distinguishing glutamate and glutamine (for example higher‑field or multi‑dimensional techniques) as well as additional post‑infusion timepoints within 1–3 days. In summary, the study provides preliminary evidence that ketamine increases MPFC GABA/W but does not alter Glx/W in this OCD sample, and that GABA/W changes were associated with short‑term symptom change; the authors call for further research to clarify how glutamatergic and GABAergic abnormalities in fronto‑striatal circuits relate to OCD pathophysiology and treatment response.