Effects of diazepam on hippocampal blood flow in people at clinical high risk for psychosis

Abstract

Elevated hippocampal perfusion has been observed in people at clinical high risk for psychosis (CHR-P). Preclinical evidence suggests that hippocampal hyperactivity is central to the pathophysiology of psychosis, and that peripubertal treatment with diazepam can prevent the development of psychosis-relevant phenotypes. The present experimental medicine study examined whether diazepam can normalize hippocampal perfusion in CHR-P individuals. Using a randomized, double-blind, placebo-controlled, crossover design, 24 CHR-P individuals were assessed with magnetic resonance imaging (MRI) on two occasions, once following a single oral dose of diazepam (5 mg) and once following placebo. Regional cerebral blood flow (rCBF) was measured using 3D pseudo-continuous arterial spin labeling and sampled in native space using participant-specific hippocampus and subfield masks (CA1, subiculum, CA4/dentate gyrus). Twenty-two healthy controls (HC) were scanned using the same MRI acquisition sequence, but without administration of diazepam or placebo. Mixed-design ANCOVAs and linear mixed-effects models were used to examine the effects of group (CHR-P placebo/diazepam vs. HC) and condition (CHR-P diazepam vs. placebo) on rCBF in the hippocampus as a whole and by subfield. Under the placebo condition, CHR-P individuals (mean [±SD] age: 24.1 [±4.8] years, 15 F) showed significantly elevated rCBF compared to HC (mean [±SD] age: 26.5 [±5.1] years, 11 F) in the hippocampus (F(1,41) = 24.7, p_FDR < 0.001) and across its subfields (all p_FDR < 0.001). Following diazepam, rCBF in the hippocampus (and subfields, all p_FDR < 0.001) was significantly reduced (t(69) = -5.1, p_FDR < 0.001) and normalized to HC levels (F(1,41) = 0.4, p_FDR = 0.204). In conclusion, diazepam normalized hippocampal hyperperfusion in CHR-P individuals, consistent with evidence implicating medial temporal GABAergic dysfunction in increased vulnerability for psychosis.

Fig. 1. ASL preprocessing and analysis pipeline.

A Diagram showing pipeline for region-of-interest (ROI) analysis. T1 images were preprocessed (1) and run through the MAGeT Brain toolbox (2), generating masks for grey matter, whole hippocampus, and hippocampal subfields. Using T1- > T2- > PD transformations calculated from registration (3 & 4), these masks were resampled (5) to CBF space (single resampling step) to allow for sampling of rCBF in native space (6). B Demonstration of optimum registration between CBF map, masks, and T1 images. C Schematic showing steps for voxel-wise analysis for CHR-P diazepam vs. placebo. A study-specific template was generated (1) from participant-averages (averaged structural scans from both drug conditions), and the CBF maps were resampled (2) to common space (single resampling step calculated from PD- > T2- > T1-> template transformations/deformations). hipp hippocampus, PD proton density, rCBF regional cerebral blood flow, ROI region-of-interest.

Fig. 2. Region-of-interest rCBF findings.

Absolute rCBF for HC and CHR-P participants (under placebo and diazepam) for A total grey matter, B hippocampus, and C hippocampus subfields, including D CA1, E subiculum, and F CA4/DG. CHR-P clinical high risk for psychosis, DG dentate gyrus, rCBF regional cerebral blood flow, ns non-significant; * < 0.05; *** < 0.001.

Fig. 3. Voxel-wise grey matter rCBF findings.

T-statistic map of drug condition (diazepam vs. placebo) effect on grey matter rCBF at the whole-brain level in CHR-P individuals from voxel-wise linear mixed effects models, thresholded and displayed at 5% FDR. Peak regions with t-statistic > 5 have been labeled. Color bars denote t-statistics which reflect 5% FDR threshold (i.e., ±2.498) and less (i.e., up to ±6.68) for both contrasts (diazepam < placebo in blue/green and placebo < diazepam in yellow/red). N.B. there were no significant voxels at 5% FDR threshold for placebo < diazepam. DLPFC dorsolateral prefrontal cortex, MCC middle cingulate cortex, NAc nucleus accumbens, OFC orbitofrontal cortex, rCBF regional cerebral blood flow.

Fig. 4. Association between baseline clinical characteristics and diazepam-induced hippocampal rCBF changes.

Pearson correlations between change in absolute hippocampal rCBF by diazepam vs. placebo and baseline clinical characteristics (at assessment visit): A positive symptoms (n = 23), B negative symptoms (n = 21), C cognitive functioning (n = 19), D social functioning (n = 23), E role functioning (n = 24), F anxiety symptoms (n = 22), and G depression symptoms (n = 21). N.B. for panels D & E a higher score denotes less impairment, whilst for all other panels a higher score/time denotes higher symptom severity. Shaded light green areas reflect 95% confidence intervals. The number of participants differs between panels due to the removal of outliers or missing data. rCBF regional cerebral blood flow.