6.2.1 Limbic-Cortical-Striatal-Pallidal-Thalamic Circuit

Evidence from neuroimaging, neuropathology, and lesion analysis studies demonstrated that the limbic–cortical–striatal–pallidal–thalamic (LCSPT) circuit is involved in the pathophysiology of unipolar depression. The LCSPT circuit is related to emotional behavior based on its anatomical connectivity with visceral control structures that mediate emotional expression and regulation, such as the hypothalamus and periaqueductal gray. This circuit has two branches: one is the limbic–thalamic–cortical branch composed of the hippocampus, amygdala, mediodorsal nucleus of thalamus, and medial and ventrolateral prefrontal cortex, and another is the limbic–striatal–pallidal–thalamic branch(6). The caudate and putamen (striatum) and globus pallidus (pallidum) are organized in parallel to connect with limbic and cortical regions. The importance of LCSPT circuit alterations in the pathophysiology of MDD has recently been confirmed in a catecholamine depletion study(7). In a randomized, double-blind and placebo-controlled catecholamine depletion study, fifteen unmedicated MDD patients in full remission and thirteen healthy controls were included. The remitted MDD subjects showed increased metabolism of the LCSPT circuit in response to catecholamine depletion. But the healthy subjects showed decreased metabolism of this circuit or remained unchanged. This study demonstrated catecholaminergic dysfunction as a trait abnormality in MDD and the depressive and anhedonic symptoms resulting from decreased catecholaminergic neurotransmission may be related to increased activity within the LCSPT circuitry. Volumetric alterations have also been reported in this circuitry. A voxel-based morphometry study revealed significantly increased gray matter volume in the left paracentral lobule, left superior frontal gyrus, bilateral cuneus, and thalamus, which form LCSPT circuitry in first-episode, drug-naive MDD patients(8). These findings were out of confounding effects of the course of illness and treatment effects that may impact anatomic measurements and provided important insight into the early neurobiology of MDD. Lui and colleagues used resting-state seed-based functional connectivity MRI to evaluate functional connectivity alterations in patients with refractory and non-refractory MDD(9). These researchers found that refractory depression is associated with altered functional connectivity mainly in thalamo–cortical circuits, while non-refractory depression is associated with more distributed decreased connectivity in the limbic–striatal–pallidal–thalamic circuit. These results suggested that refractory and non-refractory depression were characterized by distinct functional alterations in distributed brain circuits. Though there is some imaging evidence suggesting the important role of the LCSPT circuit in the pathology of MDD, functional communications within this brain circuit and its regulatory effects on other regions of the brain are too complex and more work is required. Nevertheless, based on the currently available evidence, it has been hypothesized that the balance among the brain regions within the LCSPT circuit is disrupted in depression. This may be caused by decreased activity in the prefrontal cortex that impairs its regulatory (inhibitory) action on the limbic structures which, in turn, are overactive. This dysregulation may be responsible for clinical depressive symptoms, autonomic and neuroendocrine alterations, and other visceral functions. However, this hypothesis could explain some symptoms in MDD, but could not explain other cognitive functions, such as decreased attention and impairment in executive control(10). Thus, impairments within the LCSPT structures or in the interconnections among them could result in dysfunctions predisposing a man to depression, but it is difficult to explain all of the manifestations of depression.

6.2.2 Functional Connectivity Findings Related to Suicide in Depression

Suicide is a major global public health and social problem. MDD patients have a 2–12% risk of committing suicide in the lifetime(11). Suicide attempts, typically defined as self-destructive acts with some intent to end life, are strongly correlated with depression, and a history of attempts is one of the strongest predictors of completed suicide. Thus, studying the neurobiology of depressed patients with a history of suicide attempts or suicidal ideation (SI) is a promising strategy for learning about neurobiological factors that may confer risk for suicidal behavior and potentially for identifying an objective neurobiological marker of risk. Using the fractional amplitude of low-frequency fluctuation (fALFF) approach, Cao and colleagues (12) reported that MDD patients with a history of suicide attempt (SA) showed increased fALFF in the right superior temporal gyrus, left middle temporal gyrus, and left middle occipital gyrus compared with MDD patients without such history (nSA) and healthy controls (HC). Additionally, the SA group showed decreased fALFF in the left superior frontal gyrus and the left middle frontal gyrus compared with the nSA group. By conducting ROC (receiver operating characteristic) analysis, the authors claimed that fALFF in these two regions could serve as a potential predisposition to suicidal behavior in depression. In an ICA study, Zhang and colleagues explored resting-state functional connectivity changes in the default mode network (DMN) comparing thirty-five suicidal, eighteen non-suicidal depressed adolescents, and twenty-seven healthy controls(13). Compared with the healthy controls, all the depressed patients showed increased functional connectivity in DMN regions. Compared to the non-suicidal patients, the suicidal patients showed increased connectivity in the left cerebellum, left lingual gyrus and decreased connectivity in the right precuneus. These results highlighted the important role of the DMN in the pathophysiology of depression and suggested that abnormal functional connectivity in the DMN may be related to suicidal behavior in depressed adolescents. Since the amygdala is the key brain region involved in emotional and cognitive processing, Wei and colleagues compared whole-brain amygdala resting-state functional connectivity among first-episode MDD patients with SI, first-episode MDD patients without SI, and healthy controls(14). Compared with the non-SI and HC groups, the SI group showed altered resting-state FC between the amygdala and precuneus/cuneus. They suggested that the abnormal functional connectivity between amygdala and precuneus/cuneus might present a trait feature for suicide in first-episode MDD. Similarly, a resting-state FC analysis of the rostral anterior cingulate cortex (rACC) reported decreased FC between the rACC, the orbitomedial prefrontal cortex, and the right middle temporal pole (TP) in MDD patients with SI (15). Using network-based statistics (NBS) and graph-theoretical methods, Kim and colleagues (16) found decreased functional connectivity in a characterized subnetwork in MDD patients with SI. The subnetwork included the brain regions in the fronto-thalamic circuit, suggesting dysfunctions of decision-making and information integration in MDD patients with SI.

6.3.1 Default Mode Network

Within the last decade, many imaging studies on MDD-related alterations in brain network function have used different methodologies to clarify the dysfunction of these networks themselves and their interactions with other brain regions(17). One of the core networks involved in MDD is the DMN. The DMN (also known as the “task-negative network”) was initially identified as brain regions that showed consistently synchronized deactivation during tasks and prominent activation during rest(18). As research continues, researchers divided the DMN into an anterior subnetwork that centers on the medial prefrontal cortex (mPFC) and a posterior subnetwork that centers on the precuneus cortex (PCu) and posterior cingulate cortex (PCC)(19, 20). Both the anterior and posterior parts of DMN are implicated in spontaneous or self-generated cognition. However, the anterior DMN is correlated with self-referential processing and emotional regulation, partly through its connections with limbic areas, such as the amygdala. The posterior DMN is more related to consciousness and memory processing through the connections to the hippocampal formation(21–23). Except for the core regions, the DMN also includes the inferior parietal lobule (IPL), lateral temporal cortex (LTC), subgenual anterior cingulate cortex (sgACC), and the hippocampal formation (hippocampus and parahippocampal gyrus)(20, 24).

One of the most commonly used seed regions in seed-based analyses investigating the DMN is the mPFC in anterior DMN. One study found decreased functional connectivity of the dmPFC with the posterior subnetwork of DMN in MDD patients, which was consistent with the hypothesis of a dissociation between the anterior and posterior DMN in depression(25). Sheline and colleagues found that each of three networks (the central executive network, the DMN, and the affective network) showed increased connectivity to the bilateral dorsal medial prefrontal cortex in depressed patients (26). They suggested that this region, which they termed the “dorsal nexus,” is a functional hub with increased functional connectivity. These findings provided a potential mechanism to explain how symptoms of major depression arise concurrently from distinct networks, such as poor performance on cognitive tasks; rumination; excessive self-focus; increased vigilance; and emotional, autonomic, and visceral dysfunction.

Some brain regions such as the PCC or PCu in the posterior DMN were also commonly used as seed regions. Zhou and colleagues (27) reported increased connectivity of the PCC with other posterior DMN regions and the mPFC and OFC. Similarly, Alexopoulos and colleagues (28) found increased connectivity of the PCC with both anterior (sgACC, vmPFC) and posterior (PCu) regions of the DMN in medicated patients with late-onset depression. Further evidence was reported by another study in a larger group of unmedicated elderly MDD patients (29). The patients showed increased connectivity of the PCC with other nodes in the posterior DMN but decreased connectivity with the medial frontal gyrus before treatment. The functional connectivity with both the bilateral medial frontal gyrus and the dorsal ACC was increased after twelve weeks of antidepressant treatment, suggesting that antidepressant treatment could modulate the functional connectivity between anterior and posterior DMN regions.

Except for seed-based analysis, some studies used ICA analysis, which does not require a selection of regions of interest, to study brain functional connectivity in MDD. Many ICA studies focusing on the DMN have reported increased connectivity within several regions of the anterior DMN in MDD compared to healthy controls. One ICA study investigating the role of the DMN reported increased network functional connectivity in the subgenual cingulate, thalamus, orbitofrontal cortex, and precuneus in medicated depressed subjects(30). They also found that the subgenual cingulate was a prominent region in the network of depressed patients but not in the control group, suggesting that the presence of the subgenual cingulate in the DMN may be a finding unique to depression. Another ICA study included thirty-five first-episode, treatment-naive young adults with MDD and thirty-five matched healthy controls(31). They identified increased functional connectivity in the anterior DMN (dmPFC, vmPFC, pregenual ACC, and medial OFC) in MDD patients without the influence of disease course or medication. The results of the functional connectivity changes within posterior part of DMN in depression may be inconsistent. Two studies reported increased connectivity of the PCC/PCu in the posterior DMN in MDD (32, 33). However, Zhu et al. (31) found decreased functional connectivity in the PCC, PCu, and angular gyrus in depression. The inconsistent findings may be related to the sample size, characteristics of patients, such as age and symptom severity. The connections between anterior and posterior DMN have also been investigated using ICA. The researchers identified anterior and posterior subnetworks that were spatially independent and showed asynchronous activity patterns in depression. Furthermore, the antidepressant treatment normalized the increased connectivity in the posterior DMN but not in the anterior DMN(32).

6.3.2 Central Executive Network

The central executive network (CEN, also known as the “cognitive control network” or “cognitive executive network”) includes the dorsolateral prefrontal cortex (dlPFC), dorsal ACC, posterior parietal cortex, inferior temporal gyrus, and precentral gyrus(34, 35). The CEN is involved in attention-demanding cognitive tasks and shows increased activity in frontal and parietal regions associated with top-down modulation of attention and working memory tasks. The DMN and the CEN are often seen as opposite networks as the CEN is most active during cognitive tasks.

In the CEN, the dlPFC is important in the top-down regulation of emotional processing. Many studies used the dlPFC as a seed region to investigate functional connectivity changes in MDD. Ye and colleagues (36) reported increased functional connectivity with the right dlPFC in the left ACC, left parahippocampal gyrus, thalamus, and precentral gyrus in first-episode MDD patients. However, another study found decreased FC between right dlPFC and left cuneus, left lingual gyrus, and right ACC within the CEN in MDD patients, and the altered FC between dlPFC and right ACC was positively correlated with the executive function in MDD (37). Similarly, Alexopoulos and colleagues reported decreased connectivity of the dlPFC in the CEN in unmedicated late-life depression, and the resting functional connectivity in the CEN predicted poor remission rate after antidepressant treatment and persistence of depression, apathy, and dysexecutive behavior at the end of the treatment (28). Liston and colleagues (38) used resting-state fMRI to measure functional connectivity within and between the DMN and CEN in depressed patients before and after a five-week course of repetitive transcranial magnetic stimulation (TMS). Before treatment, depressed patients showed increased FC in the DMN and decreased FC in the CEN, and altered connectivity between these two networks. After treatment, TMS normalized depression-related subgenual cingulate hyperconnectivity in the DMN but did not modify connectivity in the CEN. TMS also induced anticorrelated connectivity between the dlPFC and medial prefrontal DMN nodes. They suggested that TMS selectively modulated functional connectivity both within and between the CEN and DMN, highlighting the potential role of the subgenual cingulate as a psychoradiological biomarker for predicting treatment response. Stange and colleagues (39) studied the functional connectivity changes within the CEN in remitted MDD using the dlPFC, inferior parietal lobule, and dorsal ACC as seeds. They reported that decreased connectivity in the entire CEN in remitted MDD patients was stable and reliable over time and was most pronounced from the right dlPFC and right inferior parietal lobule to the three bilateral CEN seeds. This study demonstrated that reduced connectivity within the CEN in MDD was stable over a short period of time, was present even in the remitted state. The use of a remitted MDD sample in this study allowed for increased confidence that CEN connectivity may act as a relatively trait-like factor that is not attributable to state-dependent depressed mood.

6.3.3 Salience Network

The salience network (SN) typically consists of the fronto-insular cortex, the dorsal ACC, the amygdala, and the temporal pole, and is involved in interceptive awareness, task-set maintenance, and detection of salient stimuli from the environment(40).

Neuroimaging, more specifically, the psychoradiological studies have reported abnormal functional connectivity in the SN in MDD patients. Manoliu and colleagues (33) performed ICA analysis of resting-state fMRI data to identify DMN, SN, and CEN in MDD patients. The MDD patients showed decreased connectivity of the right anterior insula in the SN. Moreover, decreased connectivity of the right anterior insula in the SN was associated with severity of symptoms and aberrant DMN–CEN interactions. These results suggested a link between altered salience mapping and abnormal coordination of DMN–CEN-based cognitive processes, which was in line with the insula’s involvement in switching between the DMN and the CEN reported by other studies(41, 42). The insular cortex and amygdala were the commonly used seed regions in the SN. Seed-based analyses investigating the insula reported that its connectivity was increased with the pregenual ACC(43) and the medial OFC(44). This increased connectivity to brain nodes of the anterior DMN is consistent with hyperconnectivity of the anterior DMN and supported the insula’s role in coordinating interactions between networks(41, 42, 45). One study also reported increased resting-state FC of the right anterior insula to right dlPFC and right PCC in depressed elderly patients with high apathy compared to non-apathetic depressed elderly, suggesting a biological signature of the apathy in late-life depression(46). The amygdala is another important node in the SN and is highly related to MDD. Many studies have reported decreased connectivity of the amygdala with various brain regions, including the hippocampus, parahippocampus, and precuneus in adolescent depression(47) and frontal areas, postcentral gyrus, and middle occipital gyrus in late-onset depression(48). Two studies also reported decreased resting-state FC between the amygdala and insula(49, 50), which was consistent with the uncoupling of the amygdala and insula from the SN as reported by ICA studies(33, 51).