A Ventral Prefrontal-Amygdala Neural System in Bipolar Disorder: A View from Neuroimaging Research

Abstract

In the past decade, neuroimaging research has identified key components in the neural system that underlies bipolar disorder (BD). The ventral prefrontal cortex (VPFC) and amygdala are highly interconnected structures that jointly play a central role in emotional regulation. Numerous research groups have reported prominent structural and functional abnormalities within the VPFC and amygdala supporting their essential role in a neural system underlying the emotional dysregulation that is a core feature of BD. Findings in BD also include those in brain regions interconnected with the VPFC and amygdala, including the ventral striatum, hippocampus, and the cerebellum. Abnormalities in these regions may contribute to symptoms that reflect disruption in functions subserved by these structures, including motivational, mnemonic and psychomotor functions.

This article will first review leads from behavioral neurology that implicated these neural system abnormalities in BD. It will then review findings from structural imaging and functional imaging studies to support the presence of abnormalities within these neural system components in BD. It will also review new findings from studies using diffusion tensor imaging (DTI) that provide increasing evidence of abnormalities in the connections between these neural system components in BD. Emerging data supporting differences in this neural system during adolescence, as well as potential beneficial effects of treatment on structure and function will also be presented. Finally, the article will discuss the implications for future investigations, including those for early identification and treatment of BD.

Introduction

Bipolar disorder (BD) manifests as cycling between the acute emotional elevations of manic episodes and the acute dysphoria of depressive episodes. The sustained affective states of these episodes are accompanied by disruptions in a broad range of associated behaviors. These include disturbances in neurovegetative functions such as sleep, appetite and sexual drives, excesses or constriction in motivated behavior, thought abnormalities that are often colored by the affective state and can reach the level of psychosis, and disruptions in cognitive functions. A brain model for BD would ideally account for these characteristic emotional, neurovegetative, behavioral and cognitive abnormalities in the disorder, as well as its cyclic nature. Additionally, such a model would also account for abnormalities observed during periods between episodes. Though BD has long been thought of as an illness with return to excellent premorbid level of functioning between episodes, it is evident that many individuals with BD do not follow such a course and may experience protracted symptoms and disability even during euthymic periods^1–5.

Over the past decade, substantial effort has been put forth in neuroimaging research to understand the neural system abnormalities underlying BD. Significant progress has been made in identifying regional brain differences that likely contribute to the symptoms of acute episodes of BD. Below, evidence will be presented to support prominent abnormalities in a VPFC-amygdala neural system thought to play a key role in the emotional dysregulation of BD and that via its connections with structures, such as the ventral striatum, hippocampus, cerebellum and hypothalamus, could also contribute to the range of associated symptoms. Moreover, in recent years, subtle differences in this neural system, especially in the VPFC, have been identified during euthymic periods. This suggests trait disturbances that may confer susceptibility to features such as cycling that if characterized, could be targeted in preventive treatment efforts designed to bolster the resiliency of this neural system.

Leads from Behavioral Neurology

Behavioral neurology provided an initial guide to localizing key brain regions involved in BD. Since at least the 1800’s, affective and behavioral changes similar to those observed in BD have been reported in individuals with lesions in the ventral prefrontal cortex (VPFC)^6–11. [Please note that in this article the VPFC will be defined broadly to include orbitofrontal cortex (OFC), inferior and rostral frontal cortices, and the ventral and pregenual components of the anterior cingulate cortex (ACC).] In particular, these changes consist of difficulties in adaptively regulating motivational drives for engaging in pleasurable and risky behaviors despite awareness of the associated negative consequences, as can occur during manic episodes of BD^{6, 8, 11}. Furthermore, lesion studies have suggested that the valence of the mood symptomatology depends on the hemispheric lateralization of the lesion. Individuals with left hemisphere frontal lesions appear more likely to become depressed whereas those with right hemisphere lesions more often experience mania^12–14.

Studies of behaviors associated with seizures have also implicated brain regions involved in BD, particularly regions that likely exhibit excessive activity in the disorder. Flor Henry described BD-type symptoms in patients who had complex partial seizures with mesial temporal seizure foci encompassing the amygdala¹⁵. Bear and Fedio also observed BD-type symptoms associated with complex partial seizures of temporal lobe origin and additionally, they noted hemispheric lateralization with respect to the emotional valence of behavioral changes¹⁶. Paranoia and diminished humor, characteristic of depressive episodes, were associated with left-sided foci whereas elation, emotionality and hypersexuality, characteristic of manic episodes, were associated with right-sided foci.

Behavioral neurology highlighted the VPFC and amygdala as two key components in affect regulation as well as underscored the importance of hemispheric lateralization in the valence of emotional behavior. It suggested a model in which deficits in VPFC functioning and excesses in amygdala activity could produce BD-type symptoms. Given the significant reciprocal inhibitory connections between the VPFC and amygdala and the important role of the top-down VPFC regulation of amygdala responses in emotional behavior via these connections¹⁷, it suggested that disruptions in VPFC top-down regulation of the system could lead to the loss of affective homeostasis in BD. However, excesses in amygdala activity themselves could also contribute to the extremes in emotional behaviors seen in BD.

Literature from behavioral neurology has also suggested that disruptions in the white matter connections within this VPFC-amygdala neural system could contribute to the BD phenotype as well. Altered emotional and motivational states associated with lesions involving both intra-hemispheric fronto-subcortical connections and inter-hemispheric connections via the corpus callosum have been observed as early as the beginning of the 20^th century by Starr¹⁸. In support of these reports, genetic and postmortem studies have found glial cell, specifically oligodendrocyte, and myelination abnormalities in BD^19–26, leading to a recent surge in studies of white matter integrity and functional connectivity in BD.

In light of a postulated VPFC-amygdala neural system in BD, connections of the VPFC and amygdala point to other brain regions that may subserve additional functions disrupted in BD. These regions include the mesial temporal hippocampus, subcortical structures such as the ventral striatum and thalamus, the cerebellum, and the hypothalamus^27–33. Associated disruptions in the functioning of these structures could account for the range of symptoms observed in BD. For example, the hippocampus plays an important role in memory and therefore is implicated in the cognitive deficits observed in individuals with BD^34–38. Given the role of the ventral striatum in motivation and reward behaviors, ventral striatal abnormalities could contribute to the hedonic excesses of mania and anhedonia of depression³⁹. The cerebellum plays key roles in motor function and may influence the psychomotor disturbances of depression and mania. Additionally its known involvement in behavioral timing suggests that the cerebellum may influence emotional rhythmicity and timing and potentially mood cycling^40–42. The hypothalamus is important in the regulation of circadian rhythms including sleep, as well as appetite and sexual behaviors, implicating its influence in the neurovegetative symptoms of BD. Evidence of the involvement of these regions in BD from neuroimaging research are emerging as will be reviewed. Unfortunately, study of the hypothalamus in BD has been limited by the constraints of current imaging methods.

Behavioral neurology has provided important leads in the efforts to understand the neural system contributing to BD. Over the past decade, neuroimaging research has expanded upon these leads. It has identified the VPFC and amygdala as central components in the neural system underlying BD and the other connected brain regions as important contributors to the disruptions of this system. This article will review findings from structural imaging and functional imaging to support this neural system model of BD. First, the presence of abnormalities within the neural components involved in BD will be discussed, particularly in the VPFC and amygdala, as well as in the hippocampus, striatum, thalamus, and cerebellum. Then, diffusion tensor imaging (DTI) findings of abnormalities in the connections between the neural system components in BD will be reviewed. Finally, implications for future studies and for early identification and treatment of BD will be discussed.

Structural Neuroimaging Studies of BD

VPFC

In 1997, Drevets et al. provided a seminal report demonstrating reduced gray matter volume in left subgenual prefrontal cortex in adults with BD⁴³. Since that time, numerous groups have reported decreased volumes of frontal subregions in adults with BD^44–50. This has been striking as different research groups have utilized various methods for analyzing neuroimaging data^43–56. This view is supported by consistent findings in postmortem studies of decreased cellular density in the VPFC in BD^{21, 22}. A potential role for the VPFC in vulnerability to mood cycling is suggested by a reported association between greater magnitude of VPFC volume decreases in BD and rapid cycling ⁴⁴. However, it is also possible that VPFC volume decreases are a result of recurrent affective episodes. Further studies that clarify the role of specific subregions within the VPFC in BD, and their associations to cycling, will be important for understanding the role of the VPFC in the disorder. Additionally, future studies in the VPFC will also need to consider individual differences in its anatomy. For example, Fornito et al. point out that anatomy can vary across individuals, such as in the appearance of specific sulcal and gyral morphological patterns, which often has not been accounted for in morphological studies⁴⁶.

Among studies of the VPFC in adolescents with BD, there are some disparate findings, including reports of no differences in the VPFC^52–58. The variability in findings could be related to developmental influences on the VPFC. In both an earlier cross-sectional study and a more recent longitudinal within-subject study, Blumberg et al. and Kalmar et al. have noted age-related decline in VPFC volumes that appear accelerated in adolescents and young adults with BD, implicating differences in VPFC neurodevelopment in BD^{44, 53, 59}. This suggests an interaction between the maturation of VPFC during adolescence and the disorder, and possible abnormalities in the mechanisms that underlie structural maturation of VPFC during this epoch in BD⁵⁹. The potential progression of the VPFC differences over adolescence could affect the detection of VPFC differences in BD samples depending on the maturational stage of the study participants⁵⁶.

Effects of medication on VPFC volumes have been reported as well. Preclinical studies provide compelling evidence for neurotrophic and neuroprotective effects of mood-stabilizing medications that are used to treat BD, including lithium and valproate, on cortico-limbic structures⁶⁰. In 2000, Moore et al. reported cortical gray matter increases in individuals with BD following four weeks of lithium treatment⁶¹. Since that report, several cross-sectional studies found larger frontal volumes in individuals with BD treated with mood-stabilizing medications as compared to individuals with the disorder who were unmedicated at the time of study^{44, 62, 63}. A recent longitudinal within-subject study of healthy individuals found increases in prefrontal volume after four weeks of lithium treatment⁶⁴. In addition, increases in cortical gray matter have been reported in BD individuals treated with lithium⁶⁵.

Amygdala

While some studies have reported amygdala volume decreases in adults with BD, others have reported increases or no differences in volume among BD adults^66–69. The discrepant findings could relate to salient differences between clinical subtypes within BD, such as early-onset BD and BD with psychotic features, as well as differences in demographic characteristics, such as age and sex, developmental factors, duration of illness, and medication exposure⁶⁷. Evidence also indicate potential effects of genetic polymorphisms on amygdala volume, such as the 5-hydroxytryptamine transporter-linked polymorphic region (5-HTTLPR) genotype⁷⁰.

In contrast to the studies of BD adults, studies of BD adolescents have reported highly consistent findings of amygdala volume decreases^{57, 67, 71–73}. The differences in findings between adults and adolescents with BD may relate to fewer recurrences of affective episodes or shorter duration of illness in adolescents than in adults. They may also reflect less exposure to medications in adolescents compared to adults. Amygdala volumes have been reported to be larger in both adolescents and adults who were taking mood-stabilizing medication at the time of study as compared to those who were not, suggesting possible neurotrophic effects of these medications in the amygdala^{71, 74}. Alternatively, the differences may also indicate that early-onset BD represents a separable phenotype within bipolar disorders.

Associated Structures

The hippocampus has had variable findings of volume abnormalities with reports of significant or trend-level decreases or no differences in BD adolescents^{67, 71, 75, 76} and adults^{66–69, 77–82}. Multiple factors could contribute to the variable findings including differences related to sex⁷⁶ and increases in volume in association with mood stabilizers⁷⁵ as well as contribution from genetic variation^{83, 84}. The hippocampus is one of the first structures in which volume differences were associated with a specific genetic variation in BD. Chepenik et al. reported smaller hippocampus volumes with decreases most prominent in the left hippocampus in BD individuals carrying the brain-derived neurotrophic growth factor (BDNF) Val66Met allele, compared to healthy individuals and BD individuals not carrying the met allele⁸³. A longitudinal study of adults with BD also showed an association between the BDNF Val66Met polymorphism and temporal gray matter loss that included the left hippocampus⁸⁴. Chepenik et al. reported smaller hippocampus volumes with decreases most prominent in the left hippocampus in BD individuals carrying the BDNF met allele, compared to healthy individuals and BD individuals not carrying the met allele⁸³. These findings suggest genetic subtypes within BD that have particular vulnerability for volume deficits in the hippocampus.

Alterations in the other previously discussed brain regions implicated in BD have also been observed. Decreases in volume localized to the nucleus accumbens have been found⁵⁷. Studies have reported differences in basal ganglia volume or shape especially in individuals with BD who were unmedicated^{73, 85, 86}. Abnormalities in the cerebellar vermis associated with the number of prior affective episodes^87–89 and family history of BD⁹⁰ have been reported as well.

Functional Neuroimaging Studies of BD

VPFC

As functional neuroimaging studies capture regional brain activity that reflects the mental state at the time of scanning, such studies that have been performed during manic, depressed and euthymic states of BD have revealed both mood state- and trait-dependent features of the disorder. As predicted by the lesion studies described previously, mania has primarily been associated with deficits in VPFC activation that tend to be more lateralized to the right hemisphere across a variety of emotion and cognitive processing tasks ^91–95. Depressive states in BD have been more associated with relative increases in VPFC activation that tend to be left-lateralized^{92, 96, 97}. Increasingly studies across mood states in BD, and specifically of euthymic individuals, demonstrate VPFC abnormalities that suggest differences in VPFC functioning may be associated with the BD trait in adults^{92, 98–102}.

As noted with VPFC structural abnormalities, VPFC functional abnormalities may also progress over adolescence and young adulthood. In a preliminary cross-sectional study, Blumberg et al. noted that VPFC activation increased with age in healthy adolescents who were performing a color-naming Stroop task consistent with the view that the VPFC increasingly comes online with age in the service of providing adaptive inhibition of prepotent response tendencies¹⁰³. This age-related pattern was not seen in adolescents with BD suggesting an alteration in this functional maturation of VPFC in BD. In other studies of adolescents, VPFC deficits have been noted during commission errors on a stop signal task¹⁰⁴ as well as both increases and decreases in VPFC activation relative to controls during the performance of emotional processing tasks^{105, 106}. These studies suggest that some aspects of BD-related VPFC functional abnormalities may emerge during adolescence and may progress during this epoch in BD.

Mood-stabilizing pharmacotherapy has been associated with reduction of differences in VPFC activation. Deficits in activation were diminished in individuals taking mood-stabilizing medications as compared to those who were not taking medication at the time of scanning^{98, 107}. The mechanisms underlying the reversal of functional deficits are not known; their elucidation could prove important for developing new treatment strategies for BD. One possibility is raised by a recent study of the 5-HTTLPR polymorphism in BD demonstrating effects on regional brain function in the disorder that were most prominent in the ventral ACC, which has a high concentration of serotonin receptors¹⁰⁸. This suggests that mechanisms related to serotonergic transmission may have influences especially in the ventral ACC region in BD and individuals may respond to serotonergic interventions differently depending on their genetic composition.

Amygdala

Functional neuroimaging studies of individuals with BD at rest or performing emotional processing tasks during scanning have been relatively consistent in observing abnormal elevations in amygdala activation^{97, 98, 105–107, 109–115}. These findings have been reported in both adolescents and adults with BD, irrespective of mood state. This suggests that dysregulated amygdala activity may be an early trait feature of BD.

As observed in the VPFC, mood-stabilizing pharmacotherapy also has been associated with reduction of differences in activation in the amygdala. Excess activation in the amygdala appear to be blunted by mood-stabilizing medication^{98, 107, 111}. As in the VPFC, the mechanisms contributing to the reversal of functional differences in the amygdala are unclear. The reversal may reflect theorized mood-stabilizing effects achieved through restoration of homeostatic balance of excitatory glutamate as compared to inhibitory gamma amino butyric acid (GABA) activity¹¹⁶. This potential mechanism is supported by studies of anticonvulsant effects on synaptic neurotransmission in the amygdala. For example, lamotrigine has demonstrated effects on GABA transmission in the amygdala¹¹⁷ and valproic acid has been suggested to influence excitatory synaptic responses mediated by NMDA receptors in the amygdala¹¹⁸.

Associated Structures

Activation differences in other subcortical structures have also been reported in BD. The thalamus and striatum have shown functional abnormalities during both acute episodes and euthymic periods in BD adolescents^{103, 110} and adults^{97, 112, 119–121}. These abnormalities may also be reversed by medication^{120, 122}.

Functional neuroimaging studies have also noted abnormalities in the cerebellum among euthymic adults with BD^{112, 123, 124}. Decreased activation in the cerebellum has been observed in response to positive affect induction and during a counting Stroop interference task^{123, 124}. Additionally, abnormalities in cerebellar blood flow in BD have also been demonstrated¹¹². These abnormalities may be influenced by antipsychotic treatment¹²⁵.

White Matter Integrity and Connectivity

Decreases in glial cells, particularly oligodendrocytes, in both frontal regions and the amygdala, as well as down-regulation of oligodendrocyte- and myelin-related genes in frontal regions in BD, suggest the presence of white matter abnormalities in the disorder^19–25. Moreover, findings of alteration in gray matter brain regions that are highly interconnected, further support the presence of abnormalities within the white matter that connect these structures. Structural MRI studies have shown decreases in frontal white matter volume in adults with BD^{44, 50, 126–128}. In addition, abnormalities within the corpus callosum (CC) such as decreased mid-sagittal CC areas have been reported in adults with BD^{126, 127, 129–131}. This is of particular interest as the anterior component of the CC provides inter-hemispheric connections between the left and right prefrontal cortices thought to integrate hemispheric functioning and to play a key role in the integration of emotional and cognitive information¹³².

To further characterize the findings of white matter abnormalities observed by structural imaging, DTI imaging methods have recently been implemented in the study of BD. These methods may be more sensitive to detecting microstructural differences within white matter connections of regions involved in BD than structural MRI morphometry methods. Fractional anisotropy (FA) is one of the principal DTI measures that has been utilized and that provides an index of the coordinated directionality of fibers within white matter fiber bundles¹³³. DTI studies of BD have demonstrated differences in FA primarily in ventral frontal regions, in regions carrying fibers between the VPFC and the amygdala, hippocampus, striatum and thalamus as well as in the CC^134–142. As research efforts utilizing DTI progress, methods to isolate and intensively study specific white matter bundles of interest are emerging, especially those key in the neural circuitry of BD. For example, Wang et al. used methods to define the anterior region of the cingulum bundle¹⁴². This cingulum subregion is especially implicated in BD as it provides substantial connections between the regions discussed above. Wang et al. demonstrated decreases in FA in the anterior cingulum in adults with BD, as compared to a healthy group of adults¹⁴². Another bundle of particular interest in BD is the uncinate fasciculus, a ventral white matter structure that inter-connects the VPFC and amygdala. Utilizing tract-based spatial statistics, Versace et al. found abnormalities in FA in the uncinate fasciculus in adults with BD¹⁴⁰. The above findings support abnormalities in the ventral white matter that provides significant connectivity within the VPFC-neural system in BD.

Conclusions and Future Directions for the Field of BD Neuroimaging Research

Studies examining changes in behavior associated with lesions and seizures laid important groundwork for identifying the brain regions involved in BD. Since those studies, the field of BD research has advanced in understanding the structural and functional disruptions within the neural system contributing to BD and has progressed in the neuroimaging techniques used to study these disruptions. Recent developments in the field include studies that integrate multiple neuroimaging modalities to examine the relationships between gray and white matter and between structure and function. For example, a recent study combined structural and functional MRI (fMRI) methods to demonstrate a significant inverse association between volume decreases and amygdala activation increases in adolescents with BD suggesting a mechanistic relationship between the structural and functional abnormalities observed¹⁴³. Another study integrated fMRI functional connectivity and DTI methods to demonstrate associations between abnormalities in VPFC-amygdala coordinated activity and the structural integrity of the white matter that connects them suggesting that white matter abnormalities may contribute to functional disruptions in the brain circuitry of BD¹⁴⁴.

As neuroimaging research advances, a comprehensive neural system model for BD is evolving. Evidence suggests that the VPFC, amygdala, and their interactions are critical and central to the neuropathophysiology of the disorder. Data support influences by other brain structures interconnected with the VPFC and amygdala, such as the hippocampus, ventral striatum, thalamus, and cerebellum, on the neural system contributing to BD. The specific mechanisms of such influences are unclear. However, these structures may mutually affect one another’s development and function. Additionally, findings indicate effects of genetic, developmental, and treatment factors on this VPFC-amygdala neural system as well. In this VPFC-amygdala neural system model of BD, a complex interplay of different influences leads to disruptions within the neural circuitry that manifest in the range of BD symptoms, including affective episodes, mood cycling, cognitive disturbances, neurovegetative symptoms, and psychomotor disturbances. Though the proposed model highlights the central role of the VPFC and amygdala in BD, the potential contribution by other brain regions to the disorder are also important. Thus, efforts to understand such contributions should continue in conjunction with those to elucidate the roles of the VPFC and amygdala in BD. Such investigations may lead to mapping of specific BD symptoms, such as mood cycling or cognitive disturbances, onto certain components of the neural system underlying the disorder.

As a neural system model of BD emerges, there is great hope for the application of research findings to the timely and accurate diagnosis and effective treatment of BD. Work to translate findings from neuroimaging research to prevention and treatment strategies for BD has commenced. Findings of structural and functional brain abnormalities in adolescents with BD suggest that alterations in the development and function of some brain regions may emerge in adolescence and may progress during this epoch when the symptoms of BD often first present⁵⁹. In conjunction with reports that suggest the beneficial effects of medication on structural and functional abnormalities in BD, these findings emphasize the importance of early identification and treatment strategies. The differentiation of BD from other psychiatric disorders such as attention-deficit hyperactivity disorder and major depressive disorder among adolescents and young adults remains challenging. With the progress in the BD imaging field, future investigations can focus on direct comparisons between these disorders in adolescents and young adults. Such comparison studies could contribute significantly to improving diagnosis of BD in adolescents and young adults. Furthermore, studies of youths with BD, as well as emerging studies of unaffected family members at-risk for BD, could prove pivotal for the development of tools to identify and intervene early in those vulnerable to BD. Additionally, the emergence of studies that demonstrate influence of specific genetic variations on the brain circuitry of BD hold particular promise as they can point to specific molecular mechanisms that could be targeted for the development of new preventive and treatment strategies.