In this study, using an activation paradigm involving the matching of emotional facial expression, we found (i) significant attenuation of right lateral and left orbitofrontal (BA47) and right dorsolateral (BA9) cortical activation, and (ii) a significant increase in activation in the left lateral orbitofrontal cortex (BA10) in the bipolar depressed subjects compared to controls. Additionally, as opposed to the control group, the bipolar depressed subjects did not show any significant (within-group) amygdala activation, but the between-group difference was not significant.
The activation of what appears to be orbitofrontal region BA47 seen in our control subjects using the current paradigm has been previously reported in other normal control samples (31
). No study, to our knowledge, has used this paradigm with bipolar depressed subjects. In bipolar depressed subjects in the current study, there was a lack of activation of this brain region. Five functional neuroimaging studies have previously probed orbitofrontal function in patients with bipolar disorder during mania, and they report a striking convergence of findings of reduced activation in this brain region (25
). Further, in one [15
O positron emission tomography study involving six euthymic bipolar subjects, decreased bilateral OFC activity at rest was also noted (47
). These studies, in conjunction with the current study, suggest that BA47 hypoactivation may occur in all bipolar mood states and thus represent a possible physiologic trait marker of the disorder. One study using voxel-based morphometry found reductions in gray matter density in subjects with bipolar disorder, including reductions in right inferior frontal gyrus (BA47) (48
). Whether an enduring structural deficit (decreased BA47 gray matter density) underlies functional abnormalities seen in BA47 across mood states deserves further study.
The orbitofrontal brain region has been implicated in processing emotional salience (35
) and ‘drive’ (49
). It has been speculated that the OFC is involved in regulating the highest level of control of behavior, especially in relation to emotion, through pathways between the OFC and autonomic systems that govern visceral responses associated with affective stimuli (50
). There is extensive anatomical connectivity between OFC, amygdale, and cingulate (36
). Orbital prefrontal areas are also connected with temporal polar and entorhinal temporal cortex and thus have connections with the limbic cortex (52
). Neuroimaging studies have demonstrated a role for medial and lateral regions of the OFC in mood regulation (55
) and in associative emotional memory functions (35
). Orbitofrontal hypo-activity has been reported in the literature to be associated with both euphoric and depressive mood states (23
). The OFC may thus participate in aspects of emotional processing/expression that are not exclusively related to a specific valence of emotion. Lack of normal functioning in this region might result in dysregulation of mood, to either pole.
The bipolar depressed group showed an area of activation in another orbitofrontal area, the left anterior prefrontal cortex (BA10), that was not observed in the control subjects while performing the face-matching task. This region of activation is important in that it demonstrates that neither mood state per se
nor medications per se
globally dampen brain activity in the bipolar depressed subjects. Our finding of an increased left prefrontal cortical activation in bipolar depression is consistent with another fMRI study that assessed 10 bipolar depressed subjects using a different neurocognitive task (25
): a relative increase in activation in left ventral prefrontal cortex was found in this study in the bipolar depressed compared to the bipolar euthymic group. This left prefrontal activation is not typically seen in the face-matching task. This regional activation appears to be limited to the bipolar depressed state, as activation was not seen in the controls in this study, nor in manic subjects using the same paradigm in another study by our group (34
Activation of BA10 has been reported to occur in association with conflict resolution and decision making (63
). Further, this region is recruited as cognitive tasks become more difficult (64
). It is possible that in the depressed state, increased cognitive effort was expended by patients to perform the task and the increased activation is a reflection of this increased effort. This region was also shown to activate in another study of unmedicated bipolar patients (performing an attentional task), suggesting again that medications are not solely responsible for this effect (66
). In that study, however, bipolar subjects were euthymic. Thus, the state versus trait activation of BA10 requires further study. The exact role of BA10 in mood regulation, if any, is not known and will require further evaluation in both normal and bipolar depressed subjects with paradigms that can specifically probe this brain region.
The significant amygdala activation reported previously in a small number of manic patients contrasts sharply with the lack of detectable activation seen in the current bipolar depressed sample and suggests that activation/responsivity of this brain region may vary as a function of mood state. Given the small size of the structure, it is possible there was no activation in the bipolar depressed group due to a threshold effect. However, we tested this more directly by reducing the threshold of activation to p < 0.1, and we were still unable to detect any amygdala activation in the bipolar depressed subjects. While the current study suggests a reduction of activation during bipolar depression, the between-group differences were not significant with random effects analysis. As the structure is small, our sample was small, and there were only two blocks of the match faces condition, there may have been inadequate power to detect between-group differences. One other published fMRI study used a paradigm that, like ours, specifically probed amygdala function in depressed bipolar subjects, but amygdala activation was not reported in patients or controls in that study (24
). Prior fMRI studies in bipolar subjects in either the manic or euthymic state have suggested significantly increased amygdala activation in equally small subject samples with either mania or euthymia (22
) compared to control subjects. Blunted amygdala activation during bipolar depression may represent a state marker of a core physiologic change of amygdala reactivity. A larger sample may be necessary for more definitive results.
Both control and bipolar depressed groups showed substantial bilateral fusiform activation as well as activation of right DLPFC (BA9), suggesting that both groups used working memory circuits to perform the task. However, activation of right DLPFC was significantly less in the bipolar depressed compared to control subjects, suggesting a failure to fully activate secondary brain regions involved in the task. Visuospatial tasks involving a working memory component have been reported to activate DLPFC in normal subjects (68
). The task we used was not designed to explicitly test cognitive function of working memory and we had no a priori
hypothesis regarding DLPFC function when designing the current study. Additionally, we had a limited extent of coverage of the dorsal prefrontal cortex in this protocol, with the more dorsal regions not covered. However, it is known that the DLPFC is a brain region involved in willed actions (70
) and is integrally involved with the ‘holding in mind’ of information on which a future action will be based (71
). Functional MRI studies involving DLPFC function have reported reduced DLPFC activation (67
) or no differences in DLPFC activation in euthymic bipolar versus control subjects (73
). In the current study, there were no behavioral differences between groups in performing the faces task, despite the fact of significant between-group activation. Studies using fMRI with neuropsychological paradigms assessing working memory in bipolar depressed subjects may further delineate functional deficits in this brain region that are state versus trait related.
There are several limitations of the current study. First, our sample size is small, involving 11 bipolar depressed and 17 control subjects, and thus it may not be possible to extrapolate our findings to the larger population of bipolar depressed subjects. It is, however, one of the larger in the very few published fMRI studies involving subjects with bipolar depression. While only two bipolar (and no control subjects) had a history of substance abuse, most bipolar patients studied were on antimanic and antidepressant medications at the time of scanning. The impact of these medications on blood flow in the frontal lobes and amygdala has not been comprehensively studied. Antimanic agents such as divalproex sodium (in patients with epilepsy) and lithium have both been shown to decrease overall cerebral blood flow (74
) or to have no effect (78
). The majority of patients in the current study were taking these mood stabilizers. However, despite taking these medications, significant activation was seen in some brain regions in the bipolar group, including fusiform and orbitofrontal (BA10) and DLPFC (BA9) areas. Additionally, in the three published studies using face-matching paradigms in bipolar subjects on these same medications but in either a manic or euthymic state (22
), amygdala or frontal lobe regional brain activation was significantly greater in the bipolar than control subjects despite medication exposure. Thus, medication status alone does not appear to fully account for the current findings. Rather, mood state may play a role in mediating some state-related brain imaging findings, or bipolar disorder per se
may be associated with some persistent trait-related changes as well, and these could have complex interactions with medications.