The neural and molecular basis of working memory function in psychosis: a multimodal PET-fMRI study

Abstract

Working memory (WM) deficits predict clinical and functional outcomes in schizophrenia but are poorly understood and unaddressed by existing treatments. WM encoding and WM retrieval have not been investigated in schizophrenia without the confounds of illness chronicity or the use of antipsychotics and illicit substances. Moreover, it is unclear if WM deficits may be linked to cannabinoid 1 receptor dysfunction in schizophrenia. Sixty-six volunteers (35 controls, 31 drug-free patients with diagnoses of schizophrenia or schizoaffective disorder) completed the Sternberg Item-Recognition paradigm during an fMRI scan. Neural activation during WM encoding and WM retrieval was indexed using the blood-oxygen-level-dependent hemodynamic response. A subset of volunteers (20 controls, 20 drug-free patients) underwent a dynamic PET scan to measure [¹¹C] MePPEP distribution volume (ml/cm³) to index CB1R availability. In a whole-brain analysis, there was a significant main effect of group on task-related BOLD responses in the superior parietal lobule during WM encoding, and the bilateral hippocampus during WM retrieval. Region of interest analyses in volunteers who had PET/fMRI indicated that there was a significant main effect of group on task-related BOLD responses in the right hippocampus, left DLPFC, left ACC during encoding; and in the bilateral hippocampus, striatum, ACC and right DLPFC during retrieval. Striatal CB1R availability was positively associated with mean striatal activation during WM retrieval in male patients (R = 0.5, p = 0.02) but not male controls (R = -0.20, p = 0.53), and this was significantly different between groups, Z = -2.20, p = 0.02. Striatal CB1R may contribute to the pathophysiology of WM deficits in male patients and have implications for drug development in schizophrenia.

Fig. 1. Statistical parametric maps showing that a main effect of group in the striatum during working memory in first episode psychosis patients relative to healthy volunteers ((MNI coordinates: x = 30, y = 2, z = −8); F =  34.32, Z = 5.43, cluster size = 201, p < 0.001).

These findings survived family-wise error (FWE) correction on the basis of the peak-level extent (pFWE < 0.05). The colour bar shows the t statistic.

Fig. 2. Statistical parametric maps showing that cannabinoid 1 receptor availability, as determined by the distribution volume of [¹¹C] MePPEP, is significantly lower in the striatum in patients relative to controls.

Results are show using a height threshold p < 0.05 for visualization purposes. The colour bar shows the t statistic.

Fig. 3. The associtation between striatal CB1R availability and the striatal neural correlates of working memory retrieval.

a Association between the distribution volume of [¹¹C]MePPEP in the striatum (ml/cm³) and mean load-dependent striatal BOLD signal during WM retrieval (beta values) during working memory in the striatum in healthy volunteers and b patients with first episode psychosis. A Fisher r-to-z transformation indicated that these relationships were significantly different between groups, Z = −2.20, p = 0.02 (two-tailed).