Neuroimaging studies of major depressive disorder (MDD) and bipolar disorder (BD) have shown abnormalities of brain structure, function, and receptor pharmacology that appear specific to subtypes broadly defined according to family history and age-at illness-onset. In some cases these abnormalities appear episode- or mood state-dependent, while in others they persist beyond symptom remission. The former set of findings putatively reflect areas where physiological activity (measured as regional cerebral
blood flow or glucose metabolism) increases or decreases to mediate or respond to emotional and cognitive manifestations of the depressive syndrome. In contrast, the abnormalities which persist independently of mood-state may guide research aimed at identifying markers of vulnerability to MDD.

In early onset, familial MDD and BD, PET imaging abnormalities that persist into remission include abnormal glucose metabolism in the amygdala, the prefrontal cortex (PFC) ventral to the genu of the corpus callosum (i.e., "subgenual" PFC) and the
dorsomedial/ dorsal anterolateral PFC, and reductions in serotonin 1A receptor binding potential in the midbrain raphe and mesiotemporal cortex (hippocampus/amygdala). Each of these findings has been associated with histopathological changes in post mortem microscopic studies of primary MDD and BD cases, and are in some cases associated with reductions in grey matter volume (measured by in vivo MRI and post mortem neuromorphometric studies). These brain regions have been shown by
electrophysiological, lesion analysis, or PET/fMRI brain mapping studies to play major roles in emotional behavior.

During major depressive episodes additional metabolic changes suggest that brain structures involved in mediating emotional and stress responses [e.g., amygdala, hypothalamus, and periaqueductal grey (PAG)] are pathologically activated, brain areas that appear to modulate or inhibit emotional expression are also activated [e.g., posterior orbital cortex], and areas implicated in attention and sensory processing are deactivated (e.g., areas of the dorsal PFC and parietal-occipital cortex). During antidepressant drug treatment, these episode-dependent changes reverse in treatment-responders. Persistence of abnormal metabolism in the amygdala and orbital cortex during treatment has been associated with having a high risk for depressive relapse.
 
In contrast, depressives with age at illness-onset over 55 show a different set of neuroimaging abnormalities, seen in MRI images as more numerous and extensive MR signal hyperintensities in the deep and periventricular white matter (as compared with age-matched, healthy and early-onset depressive controls). Post mortem studies of such lesions have shown arteriosclerosis, gliosis, white matter necrosis, and axon loss within the affected areas, but not in surrounding tissue where the MRI signal appears normal. Functional imaging studies have confirmed that cerebral blood flow is abnormally decreased in the areas where patches or large caps of white matter hyperintensities (WMH) are evident in MR images. The personal and family risk factors for developing such WMH patches and late-onset depression are also risk factors for cerebrovascular disease, and the areas where infarction has been associated with an increased risk of major depression (left frontal lobe, striatum) are the areas most commonly affected by WMH in late-onset MDD. These observations suggest that cerebrovascular disease plays a major role in the pathogenesis of late-onset
depression.

Studies examining early onset depressives who have no family history of MDD or BD, familial loading for alcoholism, or depression arising secondary to alcohol dependence or personality disorders have not found significant differences with respect to healthy controls. In the case of subjects with familial loading for alcoholism the neuroimaging data are particularly variable, suggesting the clinical heterogeneity observed within this group may be reflected by pathophysiological heterogeneity.

Neuroimaging technology may ultimately prove useful for characterizing phenotypic subtypes of familial MDD and BD. It remains less clear whether image data will have the specificity and sensitivity needed to identify affected and at-risk subjects (i.e., before illness onset) in individual cases to guide genetic studies. The major obstacle in this regard has been that the abnormalities reviewed above have relatively small effect sizes (generally between 0.5 and 1.5). Neuroimaging approaches for classifying individual cases may nevertheless be able to rely on the covariance between multiple imaging measures for guiding discriminant analyses.

References:
1) Drevets WC, Gadde K, Krishnan R.  Neuroimaging Studies of Depression.In:  The Neurobiological Foundation of Mental Illness.  Charney DS, Nestler EJ, Bunney BJ (eds.)  Oxford University Press, 1999
2) Drevets, W.C., Price, J.L., Simpson, J.R., Todd, R.D., Reich, T., Vannier, M., Raichle, M.E. Nature 1997; 386: 824-827
3) Öngür D, Drevets WC, Price JL.1998. Proc Nat Acad Sci USA, 95: 13290-95
4) Drevets WC. Integration of structural and functional neuroimaging in depression research. In:  Psychiatric Neuroimaging Strategies: Research and Clinical Applications. Dougherty D, Rauch S, Rosenbaum J (Eds) American Psychiatric Press, Inc. Washington DC: In Press, 1999