Using Multimodal MRI Techniques to Understand the Role of Hippocampus in Schizophrenia

According to our hippocampal metaplasticity model for schizophrenia (SZ), reduced glutamate signaling in dentate gyrus could lower the LTP threshold in its target region CA3, thus generates increased associative function in CA3, resulting in memories with psychotic content. The loss of mnemonic functions in dentate gyrus could decrease its pattern separation function. Multimodal MRI techniques including proton magnetic resonance spectroscopy (1H-MRS), cerebral blood volume (CBV), cerebral blood flow (CBF) and functional MRI at 3T were used to examine the metaplasticity model and probe the role of hippocampus in schizophrenia both with (SZ-on) and without (SZ-off) antipsychotic drug treatment. Single-voxel localized scalar (J) difference editing sequence was used in 1H MRS to measure glutamate (Glu), GABA and N-acetylaspartate (NAA) concentrations in the left hippocampus in normal control (NC), SZ-on and SZ-off groups. Significant decreases in Glu and NAA concentrations relative to creatine (Cr) were found in SZ group, which confirms our hypothesis of decreased dentate gyrus glutamatergic output in SZ. High resolution vascular-space occupancy (VASO) technique using Gd-DTPA as contrast agent acquired CBV maps with resolution of 0.78mm x 0.78 mm x 4 mm, and found increased CA3+CA1+Sub relative CBV value normalized by thalamus CBV in SZ group compared to NC group, which suggests a basal hypermetabolic state in these hippocampal subregions in SZ. Pseudo-continuous arterial spin labeling (pCASL) was used to measure whole brain CBF at standard resolution, and we found hippocampal CBF had positive correlation with PANSS total scores. In the acquired equivalence (AE) fMRI study, reduced midbrain and hippocampal activity was found in both trained and transfer tasks in SZ-on group compared to NC group. The behavior data also support this finding although lacking statistical significance, which may indicate a hyperactivity-induced inefficiency in memory processing in hippocampus. Some of our results support while some modify our original hypotheses. The reduced glutamate concentration in left hippocampus can be interpreted as deriving from the putative NMDA receptor lesion in dentate gyrus in SZ. And that the hypoglutamatergic lesion in dentate gyrus would be sufficient to generate reduced hippocampal glutamate detected by MRS is demonstrated in our NR1-KO mouse study. The increased basal neuronal activity in CA3 and CA1 is supposed to decrease the efficiency of the pattern completion function within CA3, resulting in hyperassociative memory. The positive correlation between hippocampal CBF and PANSS total reveals that the increased CBF in hippocampus is related to psychosis. The AE fMRI result also gives a support to our hypothesis. We could not distinguish the SZ-on and SZ-off groups in our 1H MRS and VASO studies. This might be due to the limited sample sizes of these studies. In the AE fMRI study, the SZ-off group had better performance than the SZ-on group and the brain activity of the SZ-off group was more similar to the NC group compared to the SZ-on group, which contradicts our prediction. The age range and the match between groups shall be considered more strictly in our future studies.