This study provides evidence of an association between depressed mood and alterations in gray matter density in bipolar disorder. The regions that demonstrated decreased density included bilateral dorsomedial and right dorsolateral prefrontal cortex. Moreover, we found evidence of increased gray matter density in the parahippocampal gyrus. Importantly, all patients had been medication free for at least 2 weeks before the scan, which removed any confounding effects of current or recent psychotropic medication.
Our findings contradict the assumption that mood state is not associated with acute structural cerebral changes. Thus, attempts to elucidate cerebral correlates of depression should include not only metabolic but also structural measures. Historically, structural changes in the human brain have been attributed to developmental changes or neurodegenerative disease. Thus, psychiatric illnesses, such as schizophrenia (Hulshoff Pol et al., 2001
) and bipolar disorder (Soares et al., 2005
; Nugent et al., 2006
) have been associated with focal gray matter density changes that are often attributed to disease course. Only recently has there been increasing appreciation of the potential of other factors such as degree of exposure to the potential neurotrophic properties of psychotropic medications to affect cerebral structure (Moore et al., 2000
; Sheline, 2003
We significantly extend previous demonstrations of neuroplasticity, to show that depressed mood state may be associated with structural changes. The changes in gray matter density that we observed overlap with areas of metabolic change in acute depression. For example, decreased prefrontal metabolism has also been demonstrated in depression (Buchsbaum et al., 1997
; Ketter et al., 2001
; Brooks et al., 2009
). However, we did not detect differences in the subgenual prefrontal cortex, which has been implicated as a structure that is vital to mood-state changes in depression (Mayberg, 1997
; Seminowicz et al., 2004
). This suggests that some regions may demonstrate metabolic change in the absence of structural changes.
At the least, our findings suggest that mood state should be accounted for in morphometric studies of bipolar disorder. Several prior studies that have reported gray matter density differences in patients with bipolar disorder relative to healthy controls have included depressed patients in their samples. For example, Lyoo et al. (2004)
found evidence of decreased gray matter density in the anterior cingulate, left medial frontal gyrus, and right precentral gyrus. Lyoo et al.'s sample of 39 bipolar patients included 17 who were depressed at the time of scanning. More recently, Bearden et al. (2007)
explored the effects of lithium on gray matter density in bipolar disorder and found that patients taking lithium exhibited increased gray matter density in the left cingulate and paralimbic association cortices. In Bearden et al.'s study, 30% of the bipolar patients taking lithium met criteria for depression and 50% of those not taking lithium were depressed. The results of the present study raise the question of the extent to which heterogeneous mood states may have influenced the results in these studies.
One possible explanation of our findings can be derived by considering related cerebral networks. The DLPFC has played a significant role in the conceptualization of depression. For example, the hypothesis of reciprocal limbic-cortical function as an explanation of cerebral metabolic changes hinges on reciprocal responses among higher-level cortical regions (e.g., DLPFC) and lower-level limbic regions involved in emotional control (Mayberg et al., 1999
). In studies of depression, cortical hypometabolism (in the DLPFC and cingulate cortex) and limbic hypermetabolism (in the amygdalae) support the reciprocal limbic-cortical hypothesis (Liotti and Mayberg, 2001
; Ketter et al., 2001
). Some researchers have suggested that cognitive therapeutic interventions affect higher level, or DLPFC, changes whereas pharmacological interventions have an effect on lower level, or limbic, activity (Liotti and Mayberg, 2001
The corticolimbic hypothesis provides an explanation of how our observed gray matter changes could contribute to depressive episodes. The orbitofrontal cortex, where we observed bilateral decreases in density, has extensive reciprocal connections with the amygdala, thalamus, and the subgenual prefrontal cortex (SGPFC) and is often associated with emotional homeostasis (Mega et al.,1997
). If structural changes in gray matter density in the orbitofrontal cortex disrupted emotional homeostasis, such changes might generate a cascade of functional abnormalities in other regions, such as the limbic system and, in particular, the anterior cingulate and subgenual prefrontal cortex.
Interestingly, the medial orbitofrontal cortex (BA 11) enjoys a reciprocal relation with the SGPFC and efferent connections from the parahippocampal gyrus (Adler et al., 2006
). This network is critical for affective domains partially because of integral connections with the amygdala. One hypothesis posits that decreased inhibitory feedback between medial, dorsal, and inferior frontal cortex and elements of the anterior limbic network contribute to the emotional dysregulation observed in bipolar disorder (Haznedar et al., 2005
). Our study revealed decreased gray matter density in both the superior frontal gyrus and the parahippocampal gyrus, but no evidence of density changes in the amygdala. As noted previously, there is conflicting evidence regarding amygdala volumes in adults with bipolar disorder. Differences in age, chronicity, and treatment could account for variation in amygdala volume in adult patients with bipolar disorder, and thus may have contributed to the absenceof amygdala density changes associated with depression.
Key data that highlight the importance of networks involving the anterior limbic system illustrate relations between cerebral metabolism and treatment response. For example, successful pharmacological treatment of depression can be associated with a normalization of cerebral metabolism in the DLPFC and the anterior cingulate (Kennedy et al., 2001
). Response to paroxetine in unipolar depression increases cerebral metabolism in the dorsolateral, ventrolateral, and medial aspects of prefrontal cortex as well as anterior cingulate and parahippocampal regions (Kennedy et al., 2001
). Pharmacological response data are consistent with findings from deep brain stimulation studies, in which subcortical tract stimulation results in normalized cerebral blood flow in the dorsolateral and subgenual prefrontal cortices (Mayberg et al., 2005
). In addition, patients with bipolar disorder who were treated with lithium exhibited increased gray matter density in the bilateral cingulate cortex (Bearden et al., 2007
4.1. Study limitations
There are several characteristics of the current study that limit its generalizablity. First, our design was between subjects, so there remains a possibility that even though groups were closely matched, they differed on some other unmeasured aspect. One such factor that could have influenced our results is lifetime exposure to various psychotropic medications with varying neurotrophic properties. Given the substantial interindividual differences in treatment of bipolar disorder, controlling for this across subjects is a substantial challenge that could be avoided by utilizing a within-subject design. Thus a more compelling demonstration of reversible neuroplasticity would demand a within-subjects design controlling for lifetime exposure to psychotropic medications, preferably with gray matter changes associated with a change from euthymia to depression and back to euthymia. Of course, such a study would be fraught with additional challenges based on time effects and the likely contaminating effects of medication or therapy interventions to address the depression.
Although our depressed and euthymic groups were matched on several variables, the gender balance differed in that the depressed group was predominantly female and the euthymic group was predominantly male. Research regarding gender differences in gray matter density in bipolar disorder is lacking, but some research suggests such differences may exist. For example, adolescent females with bipolar disorder were found to have larger gray matter volumes in lateral and medial orbitofrontal cortices than their male counterparts (Najt et al., 2007
). To the extent that such results hold in adult bipolar disorder patients, they would have worked against our pattern of results, as the depressed group (which was mostly female) exhibited lower prefrontal density than the euthymic group (which was mostly male). Even so, we are unable to define the precise role that gender may have played in our findings.
Finally, our study may underestimate the extent of gray matter changes associated with depression. Insufficient sensitivity may have arisen from limited sample size and the small size of regions of specific interest. Thus, if there are mood-state-dependent changes in gray matter density, we may not have detected them because of insufficient power.
The association between gray matter density and mood state extends our knowledge of changes associated with depression beyond altered blood flow and metabolism. Moreover, the presence of mood-state-dependent gray matter variations suggests a possible mechanism through which longer-term changes in structural volumes could occur. Should our findings of mood-state-related changes in gray matter density extend to other mood states, such as mania, structural changes could provide a partial explanation of the dramatic clinical mood and functional alterations observed in mood disorders and may need to be accounted for in studies of functional changes during depression. For example, when regional abnormalities of N
-acetylaspartate, choline, and glutamate/glutamine are considered (Winsberg et al., 2000
), alterations in gray matter density may be one part of a constellation of profound, albeit unmeasured, cortical changes in depression that need to be considered. Further studies are needed to better characterize structural cerebral changes with depression and their relationships to the functional cerebral and dramatic clinical changes seen during depressive episodes.