An investigation of intrinsic, task-interim resting state BOLD signal correlations with Broca's and Wernicke's areas in schizophrenics and in subjects at high genetic risk

RESULTS: Schizophrenic patients’ t-tests (p < 0.01) consistently exhibited stronger positive correlations with Wernicke’s area (BA 22) than the NC or HR subjects in the right hemisphere (RH) superior parietal lobule (Brodmann area/BA 7), frontopolar BA 10, insular, supplementary motor, and inferotemporal cortices, and left hemisphere (LH) lingual (BA 18) and calcarine (BA 17) gyri and stronger positive correlations with Broca’s area (BA 44/45) in LH BA 40, angular area 39, lingual (BA 18) and calcarine (BA 17) gyri, and the insula, as well as the RH superior parietal lobule. In contrast, HR subjects displayed greater widespread correlations with the two language areas studied compared to the SZ patients or NC subjects. HR subjects exhibited stronger positive correlations with Wernicke’s area than did SZ patients in RH motor (BA 4) and premotor (BA 6), precuneate (BA 19), and inferofrontal cortices and LH medial temporal, mid-temporal, and insular cortices and the hippocampus, and stronger positive correlations with Broca’s area in cingulate areas (BA 24/31/33), and the LH superior parietal lobule. Both HR and SZ subjects tended to exhibit greater correlations with Broca’s and Wernicke’s areas versus NC subjects in infero- and medial frontal cortices, RH insula, and cingulate areas. Interestingly, however, the strongest differences in Broca/Wernicke correlations were consistently observed in the cerebellar pyramids, culmen, tonsils, and inferior semilunar lobule, in both SZ versus NC (3.405 < t < 6.348) and HR versus NC subjects (3.167 < t < 5.086). These findings suggest substantial hyperconnectivity of the cerebellum and isocortical regions with respect to Broca’s and Wernicke’s areas in SZ patients and HR subjects. DISCUSSION: This study demonstrates the utility of examining resting state information extracted from properly designed task-based BOLD fMRI data sets. In such data sets, resting and task-based noise is uncorrelated, providing a window into not simply the cognitive response to the task, but information relevant to the underlying brain network responsible for information processing. Our results provided evidence for two important findings. First, high-risk subjects appear to display more diffuse language-area connectivity than schizophrenic or normative subjects, particularly in medial temporal, occipital, pericentral, and prefrontal regions. Conversely, the present study’s patient population exhibited a degree of diffuse connectivity between that of the high-risk and normative subjects. These findings may indicate that the schizophrenic and HR (prodromal?) resting-state networks differ, within a single individual, only in that the latter is a more constrained version of the former. This finding would be consistent with the breakdown of the “multiple constraint organization” hypothesis of presenting schizophrenics [10] which suggests that SZ cortical networks exhibit excessive connectivity and are thus “effectively random”. Our second finding – that patients appear to effect greater cerebellocortical connectivity in the resting state, compared to high-risk and normative subjects – also supports prior studies that suggest rhombencephalic involvement in the motor coordination, sensory integration, and affective symptoms of schizophrenia [11, 12]. However, the correlation between cerebellar recruitment and symptomatology was not investigated in the present study, and further investigation is required to substantiate or refute such a link. REFERENCES: