We use functional neuroimaging to determine local neuronal activity in patients with schizophrenia and other neuropsychiatric disorders during performance of working memory and abstract reasoning tasks as well as during performance of matched sensorimotor control tests and other contrast tasks. Using a 2-back version of an N-back continuous working memory task, we have re-confirmed previous findings (derived with a more complex task, the Wisconsin Card Sorting Test) of hypofunction of prefrontal cortex. As in those previous studies, the aberrant activation pattern appears to occur even in some patients who perform relatively well on the task. We are also examining changes in cognitively-evoked activation across several time points during a medication-free period in individual patients and determining the relationship between the physiological changes and concomitant changes in symptomatology and cognitive performance. Control subjects show considerable consistency of areas activated by the N-back task, both across subjects and across time points within subjects. In contrast, patients show a general picture of hypoactivation, but also show more heterogeneity across subjects and across time points. Further analyses of the timecourse of these changes are ongoing. We have also explored the functional connectivity among various nodes of the working memory system using eigenimage analysis and have shown that more than half the intergroup variance (across controls and patients) was explained by a single pattern showing inferotemporal-hippocampal-cereballar loading for patients versus dorsolateral-prefrontal-cingulate activity for controls. Expression of this pattern perfectly separated all patient scans from the control group, a finding prospectively validated in two new data sets, suggesting that it may be a trait marker. Another pattern accounting for 20% of the variance demonstrated that expression of the working-memory-related network was significantly more variable than that in control subjects, suggesting a systems-level explanation for the findings of hypofunction. We have also shown that even when patients with schizophrenia are compared with young normal subjects with equally poor performance, they still have diminished prefrontal cortical response while performing tasks with a strong working memory component. Moreover, patients with schizophrenia still show reduced activation of the prefrontal cortex when compared with healthy, elderly subjects matched with them on a one-to-one basis for WCST performance. In this schizophrenic group, reduced activation in the dorsolateral prefrontal cortex (DLPFC) correlated with more perseveration during the WCST. In contrast, in the elderly subjects DLPFC activation is preserved and does not correlate with performance, suggesting that other mechanisms may account for their cognitive impairment on this task. We have initiated a number of cross-modal neuroimaging studies in schizophrenia. We have found that prefrontal N-acetyl-aspartate magnetic resonance spectroscopy signal predicts impaired WCS rCBF activation not only in the prefrontal cortices of our patients, but also in other nodes in the working memory system. This relationship is found in patients, but not in control subjects, and not with NAA in other brain regions. Finally, we have found that the degree to which patients' prefrontal cortices are dysfunctional predicts the degree to which their subcortical dopamine stores are aberrantly elevated. We have also initiated a new project in Williams syndrome to elucidate genetic effects on the brain that produce cognitive dysfunction. Preliminary results suggest abnormal lateralization during verbal fluency.
