In the current study differences in amygdala connectivity were found between a cohort of subjects with bipolar I disorder in a depressed versus a manic/mixed state. These subjects had similar depression ratings during both scans. Consequently, differences in the two time points represent the presence versus absence of mania in the context of similar depressive symptoms rather than a contrast between manic and depressive syndromes per se; i.e., the activation changes seem to reflect changes in manic symptoms specifically. In fact, although the lack of differences in depression severity was not anticipated, it does control for depressive symptoms across the two time points.
The analysis of HC at baseline and four months showed increased negative correlation in the middle temporal gyrus and precuneus, both of which are part of posterior attentional networks. As the only difference in the HC group between time points was the familiarity of the task at the second scan; therefore, these activation changes may reflect the possibility that subjects were less distracted by the emotional images during the follow-up scan and therefore needed less modulation of posterior attentional regions by the amygdala. The right parahippocampal gyrus showed increased correlation with amygdala in the second scan and this could be related to anticipating emotional images. The differences in connectivity in the subjects with bipolar I disorder time did not include these regions and were therefore more likely related to changes in manic symptoms rather than familiarity with the task.
Without mania the subjects with bipolar I disorder showed greater functional connectivity between the right amygdala and right insula than during mania. The insula has been shown to be involved in the anticipation of emotional stimuli, the generation of empathy, and to activate in response to disgust (37
). Many of the IAPS photos involved mutilation or injury, and so would be expected to generate both disgust and empathy in subjects. Depression causes a heightened sensitivity to negative stimuli (40
)that may have caused these subjects to show a greater emotional response to negative emotions. This suggestion may partially explain the increased functional connection of these two regions during depression. One of the few prior neuroimaging studies of bipolar depression found increased activation in the insula compared with HC (44
). Given the overlapping role of insula and amygdala in the appraisal of emotions, a potential hypothesis is that they form a positive feedback loop that increases the negative valence of stimuli and contributes to the symptoms of depression. Increased connectivity with the insula was also observed in manic compared to HC subjects at baseline, suggesting that this region may play a role in bipolar I disorder across mood states. However the altered connectivity was found in the opposite hemisphere (left) and this activation difference was no longer found in the comparison between HC subjects and bipolar depressed subjects.
Subjects with bipolar I disorder also showed several changes in prefrontal regions across mood states. Bipolar manic subjects showed increased connectivity with the right middle frontal gyrus compared to HC, whereas bipolar depressed subjects showed increased connectivity with the right medial frontal gyrus and left middle frontal gyrus compared to HC subjects. Altered activation in ventral frontal regions has been found in prior neuroimaging studies of bipolar mania (1
). Using the larger sample that included the manic subjects used in the current study, Strakowski et al. (5
) found decreased activation in the left VLPFC compared to HC subjects. In addition, two prior functional connectivity studies of subjects with bipolar I disorder during mania found reduced connectivity between the amygdala and VLPFC (45
). These prior studies are consistent with the current findings of reduced connectivity between the right amygdala and VLPFC in manic patients. Previously, we hypothesized that dysfunction in connections between these anterior limbic network regions may disrupt emotional homeostasis (1
). The differences in connectivity of the right amygdala and VLPFC across mood states suggest that there may be different network dysfunction during mania and depression. Manic subjects showed increased connectivity with the left inferior frontal gyrus and decreased connectivity in the left and right middle frontal compared to depressed subjects. Further knowledge of network dysfunction during depression and mania could allow the development of neurofunctional markers of mood state.
There are several limitations to the current study. One significant limitation of functional connectivity is that it does not allow the determination of causal connections between regions. If activation between two brain regions, A and B, is correlated, then there are three possible explanations: (i) region A causes the simultaneous activation (or deactivation) in region B; (ii) region B causes the simultaneous activation in region A; (iii) or the correlated activation in region A and B is caused by a third region of the brain. Functional connectivity cannot distinguish among these possibilities. Although physiological artifacts were controlled using low-pass filtering and global correction, it would have been ideal to directly measure physiological parameters during the scans to more accurately control for these potential artifacts. While less than half of subjects had medication changes made between scans and there was no systematic pattern to these changes, it is still possible these had some effect on the results.
Another limitation to the current study is the relatively small number of subjects. Due to the many challenges in following patients with bipolar disorder, the study had may have been underpowered. Repetition effects are another potential limitation of this study. However, in order to assess the stability of this task over time, HC participants with a baseline and four-month follow-up scan were included in this analysis. None of the regions that showed changes across time in controls overlapped with those observed in the change from mania to depression.
Despite the limitations, the current results provide evidence for changes in cortical-amygdala functional connectivity across mood states during an emotional regulation process in bipolar I disorder. In that these connectivity differences occurred in the same individuals across manic and depressive episodes, they are likely to represent neurofunctional markers of mood state. Further longitudinal studies are needed to further clarify the role these regions play in the generation of mood symptoms in bipolar I disorder.