Non-reduction in hippocampal volume is associated with higher risk of psychosis

Previous research using MRI scans has shown reduced hippocampal volumes in chronic schizophrenia and first-episode psychosis compared to well subjects. There are few MRI volumetric studies of high-risk cohorts and no studies that have compared structural measures between high-risk subjects who later developed a psychotic illness and those who did not. Therefore, the question of whether structural changes to the hippocampi precede the onset of an acute psychotic episode has not been answered. Hippocampal and whole brain volumes of 60 people at ultra high-risk (UHR) of developing a psychotic episode (identified through state and trait criteria) were obtained through MRI scan and compared with subjects with first episode psychosis (FEP: n=32), and no mental illness (n=139). Thirty-three percent (n=20) of the UHR cohort developed a psychotic disorder during the 12-month period following the MRI scan. The UHR group as a whole, like the FEP group, had significantly smaller left and right hippocampal volumes than the normal comparison group. Contrary to our initial hypothesis, the left hippocampal volume of the UHR subjects who developed a psychotic disorder was larger than the UHR-non-psychotic subgroup and the FEP group, but no differences were found between the UHR-psychotic and normal groups. The right hippocampus of the UHR-non-psychotic group was significantly smaller than the Normal group but not different to the FEP group. Furthermore, a larger left hippocampal volume of the UHR cohort at intake was associated with the subsequent development of acute psychosis rather than smaller volumes. These results contradicted the expected outcomes, which had been influenced by the neurodevelopmental model of the development of psychosis and an earlier study comparing hippocampal volumes of first episode, chronic schizophrenia and normal populations. One implication of these results is that a process of dynamic central nervous system change may occur during the onset phase of schizophrenia and related disorders, rather than earlier in life as suggested by the neurodevelopmental model. Alternatively, selection factors associated with the UHR cohort may have created a bias in the results. Replication of these results is required in other high-risk cohorts. © 2002 Elsevier Science B.V. All rights reserved.