Structural imaging studies of amygdala volume in MDD are inconsistent, with some reporting increased amygdala size [121], others noting decreased size [21], and others showing no difference from normal controls [122, 123]. In BPD, as in MDD, volumetric studies have yielded discrepant results, with reports of volume reduction [124–128], perhaps reflecting excitotoxicity with volume loss, alongside studies citing no volume differences [129–131]. Taken together, MDD and BPD structural imaging studies do not converge on a consistent finding regarding amygdala volume.

The amygdala has been viewed as the subcortical structure from which fear and perhaps anger may emerge. Amygdala activity is typically studied after exposure to a fear-inducing stimulus. In MDD, however, amygdala hyperactivity has been consistently reported [132] even at rest, perhaps due to internally generated thoughts of anxiety or sadness [133]. Similarly, increased amygdala activity is found in MDD during REM sleep, when conscious processing of stressors is not occurring. Subjects with MDD show exaggerated response to increasingly sad faces in the left amygdala and other areas that process facial emotion compared to healthy controls [134]. There is less clarity regarding amygdala response to positive stimuli [134, 135].

While increased amygdala activity in response to negative stimuli has been consistently reported in MDD, there are contradictory findings regarding amygdala activity in BPD. fMRI studies in BPD do not show increased amygdala activity when at rest as in MDD. However, increased amygdala activity is shown in BPD in response to specific types of stimulus [136] (e.g., “unresolved” life events), emotional faces [137], scenes of threat and suffering [138], positive and negative emotional pictures [139], and scripts [140]. Hazlett and colleagues [139] also reported an increase in amygdala activity to emotional stimuli. Contrary to earlier studies, however, Goodman et al. (in press) found baseline amygdala activity (at rest) to be higher in BPD subjects when compared to healthy controls. Similar amygdala hyperactivity is seen in impulsive aggressive personality-disordered subjects in response to emotional faces [141]. Furthermore, prolonged amygdala activity is seen when subjects are exposed to negative stimuli such as electrodermal stimulation [142]. These findings suggest overlap in amygdala hyperactivity in both disorders, but with differences at rest and potential variation according to type of emotional stimuli. In addition, BPD patients seem to show particularly robust responses to other emotions, including anger [143]. However, contrary to these previous findings, a meta-analysis of the neural correlates of negative emotionality in BPD found that BPD subjects showed less amygdala activity than control subjects in response to negative emotional stimuli [144] (Table 2.2).

Numerous studies in MDD have noted volume reductions in subgenual and pregenual ACC. The subgenual ACC is involved in the subjective experiencing of [123] and is viewed as a critical structure in the pathogenesis of MDD. The left subgenual ACC is reported to have 20–40 % gray matter volume reductions [145], but despite these volumetric decreases, some studies suggest that there is hyperactivity of the remaining subgenual tissue, which decreases to normal with effective antidepressant treatment [133] and is the target of deep brain stimulation [146]. However, a meta-analysis of neuroimaging data on altered emotion and cognition in MDD reported hypoactivity in subgenual tissue when not at rest [147]. Additionally, subgenual measures, such as decreased pretreatment responsivity to negative words, have been associated with treatment outcome in cognitive behavior therapy [148]. In the pregenual ACC, findings regarding the effect of treatment have been less consistent [149].

The ACC has also been a region of interest in BPD. Evidence suggests decreased gray matter volume and increased white matter volume in rostral [150] and subgenual [151] cingulate in individuals with BPD but no current MDD compared to healthy controls. Functional imaging studies in BPD have tended to show decreased activation of the ACC in response to provocation. Schmahl and colleagues [136] noted in 12 BPD subjects (one with current MDE and 11 with history of MDD) diminished activation of the perigenual ACC with induction of pain. Several other functional imaging studies in BPD also show decreased activation of the ACC in response to provocation [143, 152, 153]. Silbersweig and colleagues [154], using a behavioral inhibition task during the induction of negative emotion with fMRI, demonstrated decreased activation in the subgenual ACC and orbitofrontal cortex (OFC), with increases in amygdala activity, prompting their group and another [148] to propose that BPD sits at the “intersection of cognition and emotion” and ponder whether this constellation of impaired regions is specific to BPD.

Pharmacologic probes have also shown decreased metabolic activity in the ACC and OFC in response to the serotonergic challenge in BPD patients with impulsive aggression [155, 156] and with affective instability [157] compared to healthy controls. Decreased coupling of resting metabolism between the OFC and ventral ACC has been reported by our group [130]. A recent case study of a patient with schizencephaly [158] resulting in a primary ACC and secondary OFC lesion, who prominently manifested symptoms of BPD, supports the notion of important interconnections between these two brain regions in the development of BPD, but not MDD.

Taken together, these studies suggest similar decreases in ACC volume in MDD and BPD. By contrast, while there appears to be an overlap in the anatomic region of aberrant processing (ACC and adjacent OFC) between MDD and BPD, differences exist in the functional responses of these brain regions: In MDD they are generally hyperreactive, only when corrected for volume loss, while in BPD they appear to be under-responsive.

Decreased hippocampal volume has been reported in MDD in most, but not all, studies, with 8–19 % difference from normal controls [21, 122, 159, 160]; for a meta-analysis of these findings, see Videbech and Ravnkilde [161]. A recent meta-analysis by Kempton et al. [162] also reports decreased hippocampal volume in MDD. Volume loss appears to be inversely related to time spent depressed [163]. However, hippocampal volume loss is also found in other disorders, such as PTSD and schizophrenia.

In BPD, hippocampal volume loss has been reported in some [164–166] studies, but appears to be associated with extent of trauma [167] and abuse history [129], reflecting comorbidities with PTSD rather than specificity to BPD itself [168]. An exception to this, however, is a recent study [166] that found hippocampal volume reductions in BPD to be inversely correlated with aggressive but not impulsive symptomatology. Recent meta-analyses also support previous findings of hippocampal volume decrease in BPD [128, 168] (Table 2.3).

Other neuroimaging findings not cited by the Hasler review but implicated in MDD include increased metabolism in the posterior cingulate [133], a region believed to function as a sensory association cortex where processing of affective salience occurs, and decreased cerebral blood flow and metabolism in the dorsal medial PFC, whose impairment affects the ability to modulate emotional responses. The ventrolateral PFC, lateral orbital regions, and insula are reported to show increased metabolism in MDD; however, these findings appear to be state dependent and to improve with treatment [149].

In BPD, similar findings of posterior cingulate activation were noted by New and colleagues [156] in their 5-HT challenge study. However, other findings include volume loss [152] in the region and diminished uptake with PET scanning in BPD females with dissociation and history of childhood sexual trauma, phenomena which complicate the clinical picture and obscure the direct contribution of BPD symptomatology to the posterior cingulate findings [169].