Abstract:

Although there is substantial evidence for structural and functional abnormalities of the brain in schizophrenia, the role of these abnormalities in the clinical presentation of the disorder remains elusive. Patients can vary significantly in severity across various symptom dimensions, severity of cognitive deficit, and in their level or social and role functioning. Patients with schizophrenia also show inefficiency in working memory processing, such that when each patient maintains the maximum amount of information possible given their own capacity, high performing patients hyper-activate, and low performing patients hypo-activate, the dorsolateral prefrontal cortex (DLPFC) as compared to controls. This dissertation tests the hypothesis that inefficiency of the working memory network is related to the severity of clinical presentation. Partial least squares, a statistical technique that selects latent variables based on the covariance between sets of observed variables, is used to test this hypothesis in a group of recent-onset patients and in a group of patients at ultra-high-risk (UHR) for schizophrenia. A latent factor reflecting activation across the working memory network (including the thalamus, frontal and parietal cortices) was found to covary with a measure of role functioning across both patient groups, and in both working memory modalities (spatial and verbal). Disorganized and negative symptoms covaried with a latent factor reflecting distributed activation during the verbal task in the recent-onset patients (including the thalamus, frontal, and parietal cortices), and disorganized symptoms also covaried with a latent factor reflecting distributed activation during the spatial task in UHR subjects (including the thalamus, frontal and cingulate cortices). Interpreted within framework of dysconnectivity and disrupted cognition in schizophrenia, and taken together with prior work showing activation differences between high- and low-performing patients, these findings suggest that patients with a less severe and debilitating clinical profile are capable of mounting an increase in activation across the entire network (not only the DLPFC), which may compensate for reduced connectivity between nodes in the network in leading to normal performance on the task. At the same time, patients with more a more severe and debilitating form of illness do not show this compensatory increase in activation, perhaps reflecting an even greater degree of dysconnectivity, which in turn is associated with poor performance on the task. Implications of these findings with respect to intervention and assessment of treatment effects are also discussed.