Right dorsolateral prefrontal cortex volumetric reduction is associated with antidepressant effect of low-dose ketamine infusion: A randomized, double-blind, midazolam-controlled PET-MRI clinical trial

Low-dose ketamine infusion has been shown to produce rapid antidepressant and antisuicidal effects in treatment-resistant depression (TRD), but response varies considerably across patients. Earlier research, including a meta-analysis of 2,050 TRD cases, reported a response rate near 45% to low-dose ketamine, and several neuroimaging studies have implicated the dorsolateral prefrontal cortex (DLPFC) in the neuropathology of TRD. Prior structural work has reported reduced DLPFC gray matter (GM) in TRD, and functional imaging has documented acute prefrontal activation after ketamine; nevertheless, whether ketamine induces short-term structural or metabolic changes in the DLPFC—particularly Brodmann area 46—and whether such changes relate to clinical improvement remains unclear. Li and colleagues set out to test whether a single low dose of ketamine (0.5 mg/kg) alters GM volume and glucose metabolism in the bilateral DLPFC and whether those changes are associated with reductions in depressive and suicidal symptoms. They randomised 48 adults with unipolar TRD and prominent suicidal ideation to ketamine or an active control (midazolam) and performed PET–MRI before infusion and on Day 3 postinfusion to examine voxel-based morphometry (VBM) and 18F‑FDG uptake in BA46. The authors hypothesised increases in DLPFC GM volume and glucose metabolism after ketamine and predicted that such neuroimaging changes would correlate with clinical benefit.

This was a randomised, double-blind, placebo-controlled trial in which 48 adults aged 20–64 with unipolar TRD and prominent suicidal ideation were allocated 1:1 to a single intravenous infusion of either 0.5 mg/kg ketamine or 0.045 mg/kg midazolam. TRD was defined as failure to respond adequately to at least two antidepressants; prominent suicidal ideation was defined as a score of ≥4 on item 10 of the Montgomery–Åsberg Depression Rating Scale (MADRS). Minimum overall severity cut-offs were MADRS ≥20 and HDRS‑17 ≥17. Permuted block randomisation was used, and midazolam was chosen as an active control because of similar pharmacokinetic properties and an equipotent dose rationale. Clinical assessments included the MADRS and HDRS at baseline, at 40, 80 and 240 minutes postinfusion, and on Days 2, 3, 5, 7 and 14. Suicidal symptoms were assessed with the Columbia‑Suicide Severity Rating Scale – Ideation Severity Subscale (C‑SSRS‑ISS) at baseline, 240 minutes and on Days 2, 3, 5, 7 and 14. Participants continued their existing psychotropic medications (add-on design). PET–MRI was performed before infusion and on Day 3 postinfusion using a 3 Tesla time-of-flight PET/MRI scanner. 18F‑FDG injections used a minimum dose of 5 mCi, adjusted by weight when >50 kg (0.1 mCi/kg), with scanning beginning about 1 hour after injection; high-resolution T1-weighted MRI was acquired simultaneously. VBM analysis employed the SPM12 longitudinal pipeline: within-subject registration to a midpoint image, Jacobian determinant computation, GM segmentation, DARTEL spatial normalisation to MNI space, modulation, and smoothing with an 8‑mm FWHM kernel. A GM probability threshold >0.2 was applied and the bilateral DLPFC (BA46) mask from the Brodmann atlas served as the region of interest. Glucose metabolism analysis coregistered 18F‑FDG PET to T1 images, normalised to MNI space, smoothed with a 12‑mm kernel, and calculated standardized uptake value ratios (SUVr) using the cerebellum as reference (cerebellar mask from the AAL atlas). Statistical analyses used ANOVA or Fisher’s exact test for baseline comparisons. Neuroimaging GLMs included age, sex and years of education as covariates, with total GM volume added for VBM models. The primary imaging approach was a two-way repeated-measures ANOVA (group × time); where appropriate planned contrasts (paired t-tests for within-group, two-sample t-tests for between-group) were used. Non-parametric threshold-free cluster enhancement (TFCE) with 10,000 permutations and small-volume Family-Wise Error (FWE) correction across the bilateral DLPFC mask set significance at peak-level p < 0.05. Mean cluster GM values and SUVr were extracted for partial correlation analyses (controlling for age, sex and education). Analyses were performed in SPSS 27 and MATLAB. (The extracted TFCE statistic reported in the results appears truncated in the source text.)

Forty-eight patients with TRD and prominent suicidal ideation were randomised evenly to ketamine or midazolam, with no baseline differences in age, sex, education, HDRS scores, suicide severity, BMI, history of suicide attempt, or psychiatric comorbidity. Clinically, on Day 3 postinfusion the ketamine group showed greater improvement than the midazolam group: HDRS (p = 0.029), MADRS (p = 0.004), and MADRS item 10 (suicidal item, p = 0.019). VBM revealed a small but statistically significant volumetric reduction in the right DLPFC (BA46) in the ketamine group relative to the midazolam group (small-volume FWE corrected p < 0.05). The reported TFCE cluster statistic is present but incompletely extracted in the text. In contrast, the 18F‑FDG SUVr analysis found no significant changes in glucose metabolism of the right DLPFC between baseline and Day 3 postinfusion across groups. Within the ketamine group only, there was a negative partial correlation (r = -0.51, p = 0.022) between change in total HDRS score and right DLPFC volume change, indicating that greater symptomatic improvement was associated with a smaller decrease in right DLPFC volume. No significant correlations were observed between right DLPFC volume changes and suicidal symptom measures: C‑SSRS‑ISS (r = -0.19, p = 0.21) and MADRS item 10 (r = -0.21, p = 0.18).

Li and colleagues noted that their main imaging finding ran counter to their original hypothesis: instead of an increase, ketamine was associated with a small reduction in GM volume of the right DLPFC on Day 3, and there were no detectable changes in glucose metabolism in that region at the same timepoint. They emphasised a negative association between HDRS improvement and right DLPFC volumetric change in the ketamine group, such that participants who improved most tended to show the least reduction in right DLPFC volume. The investigators discussed these results in the context of frontal asymmetry models of depression and earlier imaging work. Structural and functional asymmetry in the DLPFC has been implicated in major depression and TRD, and prior studies have reported compensatory or excessive right DLPFC activation in TRD. The authors related their findings to a mix of prior morphometric studies: some have reported increased GM in frontal regions after ketamine at earlier timepoints (for example, increases in the right inferior frontal cortex at 24 hours in one study), while others found no medication-specific effects or observed decreases over time in both drug and placebo groups. Such heterogeneous findings, they argued, complicate simple interpretations about directionality of volumetric change after ketamine. The authors proposed that ketamine’s antidepressant effect might relate to an ‘‘optimal’’ modulation of right DLPFC GM volume, where neither excessive increase nor excessive decrease is favourable—an idea invoked to reconcile the present negative correlation and the observation that a few responders showed small volume increases. They also considered prior functional PET and functional near-infrared spectroscopy findings from their group that support a role for right DLPFC dysfunction in TRD. Key limitations acknowledged by the investigators include the add-on design—participants continued baseline psychotropic medications, so imaging and clinical effects may reflect interactions with existing treatments; the choice of imaging timepoints (baseline and Day 3) which may miss earlier or later volumetric/metabolic changes reported by other studies (for example at 24 hours); and individual variability, exemplified by three ketamine responders who had small right DLPFC volume increases. The authors recommended longitudinal follow-up studies with additional timepoints to characterise the trajectory of GM and metabolic changes after ketamine. In their interpretation, low-dose ketamine produced rapid clinical improvements and a small but significant reduction in right DLPFC GM at Day 3, with a smaller volumetric decrease associated with greater symptomatic benefit, suggesting that modulation of right DLPFC structure may be involved in ketamine’s antidepressant mechanisms.