Alterations in functional activation in euthymic bipolar disorder and schizophrenia during a working memory task

Abstract

Dysfunctions in prefrontal cortical networks are thought to underlie working memory (WM) impairments consistently observed in both subjects with bipolar disorder and schizophrenia. It remains unclear, however, whether patterns of WM-related hemodynamic responses are similar in bipolar and schizophrenia subjects compared to controls. We used fMRI to investigate differences in blood oxygen level dependent activation during a WM task in 21 patients with euthymic bipolar I, 20 patients with schizophrenia, and 38 healthy controls. Subjects were presented with four stimuli (abstract designs) followed by a fifth stimulus and required to recall whether the last stimulus was among the four presented previously. Task-related brain activity was compared within and across groups. All groups activated prefrontal cortex (PFC), primary and supplementary motor cortex, and visual cortex during the WM task. There were no significant differences in PFC activation between controls and euthymic bipolar subjects, but controls exhibited significantly increased activation (cluster-corrected P < 0.05) compared to patients with schizophrenia in prefrontal regions including dorsolateral prefrontal cortex (DLPFC). Although the bipolar group exhibited intermediate percent signal change in a functionally defined DLPFC region of interest with respect to the schizophrenia and control groups, effects remained significant only between patients with schizophrenia and controls. Schizophrenia and bipolar disorder may share some behavioral, diagnostic, and genetic features. Differences in the patterns of WM-related brain activity across groups, however, suggest some diagnostic specificity. Both patient groups showed some regional task-related hypoactivation compared to controls across the brain. Within DLPFC specifically, patients with schizophrenia exhibited more severe WM-related dysfunction than bipolar subjects.

Figure 1

“On” block schematic. In each 5.6‐s trial, subjects were presented with four random stimuli, followed by a fifth (after the red disc) that was either in previous set of four 50% of the time. During the last 500 ms interval, subjects were asked to make a button press response to indicate whether the fifth design was contained in the set of four prior designs. “On” blocks were made up of five of these 5.6‐s WM trials, so each block lasted 28 s.

Figure 2

Overlays of within‐group activation for the WM task. The overlays of average activation within groups show the similarities and differences in activation between control subjects (blue) and patients with schizophrenia (green) in the top panel, between control subjects (blue) and euthymic bipolar subjects (red) in the middle panel, and patients with schizophrenia (green) and euthymic bipolar subjects (red) in the bottom panel. Average maps were thresholded at Z = 2.3 with a cluster corrected significance threshold of P = 0.05. L, left; R, right; A, anterior; P, posterior. The same within‐group activation maps are provided for each diagnostic group separately as Supporting Information Figure 1.

Figure 3

Contrasts showing significantly greater (FWE cluster‐corrected P < 0.05, Z > 2.3) activation in (1) controls compared to patients with schizophrenia, (2) controls compared to patients with bipolar, and (3) patients with schizophrenia compared to patients with bipolar during the WM task. Color indicates Z statistic values. L, left; R, right; S, superior; I, inferior; P, posterior; A, anterior.

Figure 4

Mean percent change of the BOLD signal in the DLPFC of each subject group. Raw mean values are displayed next to each data point. Error bars represent 95% confidence intervals for the mean.