Glucose homeostasis requires hormonally and neurally mediated regulatory actions that contribute to the correct functioning of the brain and the peripheral tissues. Glucose is not only the main energy source for the maintenance of neural and non-neural cellular activity; it also acts as a signaling molecule. Therefore, glucoregulatory mechanisms are key to ensuring an appropriate glucose supply to meet the metabolic needs of the central nervous system as well as the peripheral tissues.

The antagonistic effects of the pancreatic hormones insulin and glucagon, the activity of the hypothalamic–pituitary–adrenal axis, and the components of the autonomic nervous system help to maintain blood glucose levels within a physiological range depending on the energy status of the organism. Alterations in physiological glycemic levels have deleterious consequences, increasing both morbidity and mortality rates. To prevent marked blood glucose oscillations, glucose sensors in several locations accurately sense the glucose concentrations in the extracellular space and set in motion regulatory mechanisms needed to maintain glucose homeostasis. Glucokinase (GK) has been shown to act as a glucose sensor in the hypothalamic neurons of both humans and rats (9, 25, 26), being involved in the regulation of energy homeostasis, feeding behavior, glucose metabolism, and the autonomic nervous system (27–29).

In humans under physiological situations, glucagon-like peptide-1 (GLP-1) was shown to significantly reduce glucose metabolism in a selective and temporal manner in the hypothalamus and the brainstem, areas involved in the control of food intake and glucose sensing (30). These observations suggest that reversible and brief glucose hypometabolism in these brain areas may induce important biological effects.

In the brain, virtually all glucose is oxidized to CO₂ and H₂O through glycolysis and the mitochondrial respiratory machinery. The relationship between O₂ consumption and CO₂ production in the brain is close to one, from which it follows that carbohydrates and glucose, in particular, are the exclusive substrates of oxidative metabolism in the brain (31). Changes in brain glucose metabolism detected using several technical approaches have been reported in AD patients. Currently, functional neuroimaging techniques such as positron emission tomography (PET) provide in vivo measurements of glucose metabolism with very high sensitivity. This imaging methodology allows us to identify functional changes in brain metabolic activity even before the appearance of clinical symptoms. At present, the use of ¹⁸F-FDG for PET imaging is approved in Europe and the USA for the study of glucose metabolism in certain cardiovascular, neurological, and oncological diseases (32).

In PET functional imaging, 2-deoxy-2-(¹⁸F) fluoro-D-glucose (¹⁸F-FDG) competes with glucose to bind to transporters localized in the cell membranes of neurons and astrocytes. The ¹⁸F-FDG intracellular concentration relates to the hexokinase activity and glycolysis of brain cells, and changes in glycolytic activity are associated with neurodegenerative diseases. Therefore, ¹⁸F-FDG-PET provides quantitative tomographic images of the distribution of neuronal metabolism, allowing the determination of the changes in regional hypometabolism observed in several neurological diseases. Thus, PET neuroimaging is a valuable tool for the early diagnosis of neurodegenerative diseases, which are characterized by marked alterations in brain glucose metabolism ( Table 1 ). Regional brain glucose hypometabolism measured by ¹⁸F-FDG-PET imaging by itself indicates a reduction in cellular glucose utilization but this cannot be directly attributed to a particular type of cell (neuronal or non-neuronal) or whether is related to functional deafferentation or to cellular loss. Despite this caveat, considering that the affected areas have been well characterized, by means of other complementary techniques, by glial reactivity and neuronal dysfunction or death, it is generally accepted that brain glucose hypometabolism is mostly an overall reflection of neuronal impairment or death. Regional brain hypometabolism in neurodegenerative diseases is well-described in animal models of AD, in which hypometabolism is found in the hippocampus and temporo-parietal cortex. Thus, ¹⁸F-FDG-PET imaging in patients revealed that glucose metabolic reductions in the parieto-temporal, frontal and posterior cingulate cortices were a hallmark of AD (33). This focal alteration in metabolism can also be observed in brain areas affected by stroke and in epileptic foci. Likewise, brain glucose hypometabolism is typical in the striatum in PD models. In psychiatric pathologies such as major depression, robust hypometabolism is generalized. Furthermore, ¹⁸F-FDG-PET imaging can be performed in experimental models of disease. Thus, in a transgenic mouse model of tauopathy (34) as well as in GSK-3β-overexpressing mice (35), brain glucose hypometabolism is reflected by a reduction in ¹⁸F-FDG ( Figure 2 ).

Insulin action inducing glucose uptake in the CNS differs from that inducing that in the peripheral tissues due to the brain’s lower content of GLUT-4, which is the only insulin-sensitive glucose transporter. However, GLUT-1 and GLUT-3’s presence and action are more relevant in the brain than in peripheral tissues. This selective tissue distribution of glucose transporters is necessary for the metabolic activities occurring in physiological situations and might explain the alterations observed in pathophysiological conditions (34, 35)

Brain glucose uptake was traditionally considered insulin-independent (41, 42). It is currently known that insulin acts in concert with IGF-1 on astrocytes controlling brain glucose metabolism (43, 44). Thus, insulin and IGF-1 synergistically stimulate a mitogen-activated protein kinase/protein kinase D (MAPK/PKD) pathway, resulting in the translocation of GLUT-1 to the cell membrane through multiple protein–protein interactions.

The ablation of insulin receptors in astrocytes reduces the glucose-induced activation of hypothalamic proopiomelanocortin (POMC) neurons. Accordingly, astrocytic insulin signaling co-regulates hypothalamic glucose sensing and systemic glucose metabolism. Thus, insulin signaling in hypothalamic astrocytes co-controls CNS glucose sensing and systemic glucose metabolism via the regulation of glucose uptake across the blood–brain barrier (BBB) (45). In this context, the role of hypothalamic GK as a glucose sensor involved in the control of satiety and body weight is noteworthy. At this point, it is important to know the functional interrelations between GK and GLUT-1 in POMC neurons.

Despite the different physiological consequences, the molecular events through which insulin acts on the brain are similar to those in the periphery ( Table 2 ).

Insulin’s actions on the brain contribute to regulating energy expenditure, glucose homeostasis, feeding behaviors, reproduction, cell proliferation and differentiation, neuroprotection, neuromodulation, and learning and memory (10). When referring to insulin-induced proliferative effects, the interaction of insulin and IGF-1 at the receptor molecular level must be considered. Insulin can act on IGF-1 receptors at high concentrations. Likewise, IGF-1 can also interact with the insulin receptor. In addition, there are hybrid receptors in which one of the α-subunits binds to insulin, while the other binds to IGF-1. These findings, increasing the plasticity of both molecules in growth and metabolic activities, may be relevant under physiological and pathological situations.

Insulin is also a potent neuroprotective agent, mainly inhibiting apoptosis, beta amyloid toxicity, oxidative stress and ischemia (46–50). Moreover, insulin facilitates learning and memory by modulating hippocampal synaptic plasticity, which is importantly regulated by glutamatergic neurotransmission, which mediates long-term depression (LTD) by reducing AMPA receptors in the postsynaptic membrane while stimulating long-term potentiation (LTP). Glutamate also upregulates GABA receptors in the postsynaptic and dendritic membranes of the neurons (60–62). Impaired memory ability and reduced hippocampal synaptic plasticity were shown to be restored by insulin treatment in an experimental model of diabetes mellitus. Additionally, it was reported that IGF-1 augments hippocampal synaptic transmission by a mechanism involving AMPA receptors and PI3K activity (63).

Insulin also affects APP metabolism, activating α-secretases and regulating Aβ levels by promoting Aβ transport to the neuronal gap. Insulin-degrading enzyme (IDE) is responsible for degrading not only insulin but also Aβ, having a higher affinity for the hormone than for Aβ. Then, insulin can prevent the formation of Aβ fibrils, stimulating the internalization of Aβ oligomers and, therefore, inhibiting their binding to neurons, protecting the synapses from Aβ oligomers (64). In states of insulin resistance, the protective role of insulin regarding Aβ accumulation is diminished, and, in turn, Aβ deposits downregulate the action of insulin. Consequently, Aβ peptides inhibit the binding of insulin to its receptors, altering insulin-induced signaling pathways (65, 66). It was shown that tau protein phosphorylation increased significantly in an animal model of insulin resistance induced by fructose; in AD transgenic mice, insulin resistance induced by diet facilitated brain Aβ formation.

AD is a neurological disorder that causes significant memory loss and progressive dementia, accompanied by the presence of amyloid plaques, neurofibrillary tangles and amyloid angiopathy, as well as the widespread loss of neurons and synapses (74). More than 40 million people worldwide suffer from AD. The prevalence is increasing rapidly, being expected to exceed 130 million patients by 2050 (51). Impaired insulin secretion and/or insulin resistance were estimated to affect 425 million patients worldwide in 2017, with 90% of the patients having a diagnosis of T2DM (75).

Preclinical studies reported that the induction of acute hyperglycemia in a mouse model of AD, using glucose clamps and in vivo microdialysis techniques, increased Aβ and lactate production in the hippocampal interstitial fluid (ISF), considered biomarkers of altered neural activity. These effects were exacerbated in aged AD mice with advanced Aβ pathology. Furthermore, and based on the altered ISF Aβ levels and neuronal activity observed after the pharmacological manipulation of ATP-sensitive potassium (K⁺-ATP) channels in the hippocampus, they suggested the involvement of these channels in mediating the response to hyperglycemia. Therefore, K⁺-ATP activation might mediate the response of hippocampal neurons to hyperglycemia by coupling metabolism with neuronal activity and ISF Aβ levels (76). Likewise, in vitro studies showed that Aβ1-42 incubated with hippocampal slices produced a significant decrease in glucose uptake (77), which was related to impaired glycolysis but not to altered mitochondrial function (70, 78).

Insulin resistance is a risk factor for the development of AD, being a common finding in AD patients independent of T2DM. Furthermore, peripheral insulin resistance might also be accompanied by central insulin resistance, IGF-1 resistance, and IRS-1 and IRS-2 dysfunction, presumably as a consequence of Aβ oligomers further contributing to cognitive decline (79). It was proposed that peripheral and central insulin resistance crosstalk through a “liver–brain axis”, promoting cognitive dysfunction in T2DM (80).

Clear-cut experimental evidence shows that inflammatory responses are closely associated with the development of insulin resistance in obesity and T2DM (81), increasing the risk of AD. Accordingly, in the cerebrospinal fluid (CSF) of patients with AD, high concentrations of interleukin 6 (IL-6) have been reported (82). Studies on experimental animals suggest that inflammation interacts with the processing and deposition of β-amyloid peptide (83). Increased levels of inflammatory cytokines alter hippocampal synaptic plasticity and the components of spatial learning (84). Chronic inflammation may contribute to the development of insulin resistance and T2DM, as well as the association of both AD and T2DM (85). Additionally, augmented brain TNFα and Aβ contents in obese hyperinsulinemic patients facilitate the formation of amyloid plaques (86).

Alterations in some insulin signaling pathways such as PI3K/Akt and GSK-3 are present in central inflammation and insulin resistance (51). The PI3K pathway inhibits the formation of IL-12 by dendritic cells, whereas GSK-3 is a kinase involved in tau hyperphosphorylation and the modulation of Aβ metabolism (87). Some interactions occur between this protein and the insulin signaling pathway. IR activation phosphorylates and inhibits GSK-3β (52). In AD and T2DM, GSK-3β activity is increased, phosphorylating the IR and IRS-1 (53). In addition, the inhibition of GSK-3 stimulates the production of anti-inflammatory cytokines such as IL-10, and decreases proinflammatory cytokines such as IL-1β, IL-6 and IFN-ϒ in response to Toll-like receptors (88). These findings have been observed in AD patients and in animal models. Accordingly, the PI3K/Akt/GSK-3 pathway may play a relevant role by mediating the inhibitory effect of insulin on inflammation.

¹⁸F-FDG-PET imaging has revealed reduced glucose uptake in several brain areas in AD patients (89, 90). This brain glucose hypometabolism is present in both nondiabetic and diabetic patients with AD. However, it is important to remember that glucose metabolism dysfunction in diabetic patients increases the risk of cognitive impairment (91). In transgenic mice overexpressing the tau protein, marked hippocampal glucose hypometabolism and a significant beneficial effect of insulin improving cognitive function were reported (34).

Parkinson’s disease (PD) is a neurodegenerative disorder caused by a progressive deterioration of the dopaminergic neurons of the substantia nigra in the midbrain (92). PD patients exhibit widespread cortical hypoperfusion and reduced brain glucose metabolism (71). As in AD patients, these metabolic abnormalities are expressed in the initial stages of the disease, suggesting that they might be important for an early diagnosis of the disease or for initiating potential treatment.

Several pieces of epidemiological and molecular evidence show that insulin alterations such as DM are related to PD (93). The high density of insulin receptors in the dopaminergic neurons of the ventral tegmentum and substantia nigra suggests that those neurons might be targets for the action of insulin (54, 94–97). In addition, insulin receptors’ mRNA expression and immunoreactivity in the neurons of the substantia nigra are reduced in PD patients (98), and dysfunctional insulin-mediated signaling occurs before the death of dopaminergic neurons. Therefore, it seems that a relationship between insulin dysregulation and PD exists.

Approximately half of the patients with a diagnosis of PD are either glucose-intolerant and/or diabetics (99). Data obtained from postmortem studies, experimental animal models and cell cultures support the idea that the neuroinflammatory processes related to microglial activation, astrogliosis, and lymphocyte infiltration are involved in the loss of neurons in PD (100–102). Considering the role of neuroinflammation in insulin resistance and the role of the components involved in insulin signaling transduction mediating inflammatory processes, it could be deduced that neuroinflammation may facilitate the crosstalk between insulin dysfunction and PD. In this context, the antidiabetic drug pioglitazone, acting through peroxisome proliferator-activated receptor gamma (PPAR-Υ) receptors, was interestingly shown to reduce neuroinflammation, suggesting that it might protect dopaminergic neurons in PD (103).

Insulin dysfunction has also been related to the progression of PD. It was shown that dopamine synthesis in the synaptosomes of the striatum in diabetic rats was significantly lower than that in the controls, and insulin resistance reduced the release and clearance of dopamine by dopaminergic neurons (104).

Insulin has regulatory effects on the expression and activity of tyrosine hydroxylase, the limiting enzyme responsible for dopamine synthesis (105). Thus, tyrosine hydroxylase activity and dopamine concentrations are reduced in the PD brain; reduced tyrosine hydroxylase mRNA expression was reported in the dopaminergic neurons of diabetic rats (106). In addition, insulin has been shown to increase the expression of the dopamine transporter and to promote presynaptic dopamine reuptake in the tegmentum ventral and substantia nigra (107, 108).

¹⁸F-FDG-PET imaging studies showed that glucose metabolism is elevated in the putamen, an increase that is usually accompanied by low ¹⁸F-DOPA uptake. Other brain regions of the associative cortex show a decline in glucose metabolism that increases as the disease progresses. Importantly, progressive cortical hypometabolism in temporo-parietal and occipital regions marks the transition from normal cognition to dementia (109).

Huntington’s disease (HD) is a neurodegenerative genetic disorder characterized by the abnormal formation of polyglutamine aggregates, known as huntingtin, which progressively results in neuron dysfunction and tissue loss in the striatum and cerebral cortex. The clinical manifestations of HD include disordered movements, cognitive decline, and psychiatric symptoms (55). Additionally, HD is not only associated with altered insulin metabolism and diabetes mellitus, but also has a neuroinflammatory component that plays a relevant role in the development and progression of the disease (110). Insulin-sensitizer drugs, such as thiazolidinediones, and anti-inflammatory drugs, such as inhibitors of cyclooxygenase (COX-2), have been shown to have protective effects in experimental models of HD (111, 112).

The incidence of DM is higher in HD patients (113), who have abnormal glucose tolerance (114), likely as a consequence of reduced β-cell mass, insulin content, and β-cell replication, and altered exocytosis (115). The glibenclamide treatment of diabetes in a transgenic mice model of HD showed beneficial effects, suggesting a stimulatory action on insulin exocytosis (116).

Insulin plays a key role in regulating genes involved in the pathogenesis of HD, and insulin dysregulation affects the neuropathology of HD (117). The activity of the protein kinase Akt, one of the main components of the insulin signaling pathway, is altered in HD. The activation of Akt in the insulin signaling cascade inhibits cell death in primary cell cultures of the striatum. Results from postmortem studies with HD samples demonstrate that caspase-3 degrades Akt, abolishing its effect on cell survival (55).

Another mechanism underlying the effect of insulin on the pathogenesis of HD is the ability of this hormone to promote the clearance of huntingtin aggregates. Although neurons have a certain ability to clear these protein aggregates, the continuous production of the mutant protein, as occurs in HD, compromises this ability. We suggest that a vicious circle between the pathogenesis of HD and insulin-related alterations might develop, contributing to the progression of the disease. In addition, metformin is considered by some authors as therapeutically useful for neurological diseases such as HD (118).

¹⁸F-FDG-PET imaging studies in HD revealed an early and marked reduction in glucose metabolism in the caudate and putamen nucleus that, in further advanced stages, extends to other brain areas such as the thalamus and brain cortex (119).

Epilepsy is a heterogeneous neurological chronic disease characterized by recurrent spontaneous seizures. It is one of the most important neurologic syndromes, with a high prevalence, affecting around 0.5–2% of the world’s population. Among the different forms of this disease, temporal lobe epilepsy (TLE) is the most prevalent in adults. The development of epilepsy is associated with a wide spectrum of neuronal alterations, including neuroinflammation, neurochemical imbalances, synaptic modifications in specific brain areas, and metabolic activity disturbances. ¹⁸F-FDG-PET imaging techniques, commonly used in TLE patients, have been demonstrated to be highly sensitive, allowing for the localization of the epileptic focus. During the seizure or ictal phase of the disease, the brain metabolism and cerebral blood flow increase in the epileptic focus. By contrast, after the seizure, during the interictal phase, the epileptogenic zone is characterized by pronounced glucose hypometabolism (120).

Under physiological conditions, brain metabolism is mainly fueled by glucose through both aerobic and anaerobic pathways (121). It was suggested that altered glucose metabolism is probably a key initiating factor promoting epileptogenesis. Energy deficiency could underlie the brain hypometabolism featured by ¹⁸F-FDG-PET imaging in patients and in animal models of epilepsy (122, 123).

In epilepsy, during seizure, brain glucose hypermetabolism occurs (124–126), and the metabolic rate of glucose and oxygen consumption increases. It was suggested that the aerobic pathway is unable to supply the amount of energy required to deal with the demands imposed by the seizure. The energy necessary to sustain a neuronal-hyperactivity-induced seizure might be obtained via increased blood glucose uptake and/or increased glycolytic astrocyte activity coupled to the lactate shuttle (121). The inhibition of lactate dehydrogenase (LDH) reduces hippocampal lactate, eliciting neuronal inhibition and reducing seizures and epileptiform activity (127). It has long been known that a ketogenic diet is a nonpharmacological alternative treatment for children with refractory epilepsy. Less is known about its effectiveness in adult patients. Nevertheless, the mechanisms underlying the antiepileptogenic effect of ketogenic diets are yet to be fully unveiled. Furthermore, ketogenic diets seem to act through many distinct and unrelated mechanisms that ultimately converge in an overall antiepileptogenic effect. Nevertheless, it seems that ketogenic diets, by leading to the generation of ketone bodies as alternative fuel, reduce hippocampal lactate by weakening the astrocyte–neuron lactate shuttle. Thus, both LDH inhibition and ketogenic diets seem to converge in at least one common mechanism related to their antiepileptogenic properties.

Among the various animal models of epilepsy that have been developed, the most frequently used are those that induce partial seizures through kindling or status epilepticus (SE). Kindling is the procedure through which the progressive development of seizures is achieved in response to an initial subconvulsive electric or chemical stimulus applied repeatedly and intermittently (128).

In addition to its clinical application, ¹⁸F-FDG-PET imaging has been used to characterize several animal models of epilepsy, such as the lithium–pilocarpine status epilepticus and amygdala electrical kindling models (12, 129). For studies in those models as well as in other animal models of epilepsy, ¹⁸F-FDG-PET imaging reveals a significant reduction in cerebral glucose metabolism. In the silent stage of the pentylenetetrazol (PTZ) kindling model (130), this glucose hypometabolism was considered to be a reliable and predictive index for the epilepsy outcome, clearly correlating hypometabolism imaging in the entorhinal cortex and epileptogenesis with spontaneous seizures ( Figure 2 ). This finding suggests that a reduction in brain glucose metabolism is effectively associated with convulsive seizures. Although the mechanisms underpinning brain glucose hypometabolism in epilepsy are not yet fully understood, it was suggested that epilepsy-induced brain hypometabolism could be due to neural loss (131), the alteration of cerebral blood flow (132), or the deafferentation of epileptic neurons (133).

Altered insulin signaling in the brains of genetic-absence epilepsy rats from Strasbourg was found (134), suggesting a role for this hormone in this disorder and a possible relationship of it with brain glucose alterations.

Even though the mechanisms underlying brain glucose hypometabolism in epilepsy and their potential contributions to the neuropathology of epilepsy are still unknown, it is reasonable to suggest the involvement of multiple and likely concurrent processes such as neuronal excitability dysfunction, neuroinflammation and microglial activation, disturbed brain metabolism, neuronal death, and/or disruption of cerebral blood flow.

Schizophrenia is a psychiatric disorder characterized by abnormal behavior and psychotic symptoms such as hallucinations, false beliefs, confused thinking, reduced social engagement, and emotional expression. At the neurochemical level, these patients have alterations in dopaminergic and glutamatergic neurotransmission. Although the mechanisms by which they occur are unknown, progressive inflammation and neurodegeneration have been suggested as the potential main mechanisms contributing to this disease (135, 136). Similar to the other neuropathologies discussed in this review, alterations in the secretion of and sensitivity to insulin also occur in schizophrenic patients (56, 137). Schizophrenic individuals have elevated circulating insulin levels, which corresponds to a state of insulin resistance, and are more susceptible to suffering from T2DM (138), and they have altered tolerance to glucose overload. Additionally, schizophrenic patients with normal body weights and corporal mass indices have anomalous tolerance to glucose, and they have abnormalities in lipid and energy metabolism (139).

It is known that the signal transduction mediated by the insulin receptor located in the prefrontal dorsolateral cortex is reduced, as is the autophosphorylation of the insulin receptor (138). Clozapine administration in animal models of insulin resistance reduced the signaling deficiency for the insulin receptor (138).

Patients with schizophrenia were found to have frontal hypometabolism in ¹⁸F-FDG-PET studies, and this was found to be strongly associated with cognitive deficits (140).

Robust evidence supports the idea that DM is associated with changes in mood and anxiety (141). Many studies have established a marked relationship between depressive alterations and changes in insulin metabolism. The prevalence of depression in T2DM patients is three times greater than that in the general population (142). This is also observed in the pediatric population, with depression being two to three times more prevalent in diabetic children and adolescents than in their nondiabetic counterparts. Mood alterations during the postpartum period have also been related to the abrupt decrease in circulating insulin levels after delivery (143).

Insulin is known to have effects on serotonin neurotransmission, which plays an important role in behavior and mood disorders. Most of the antidepressant drugs act by increasing the synaptic concentrations of serotonin and other monoamines such as dopamine and norepinephrine (144). Insulin metabolism alters brain monoamine activity both in diabetic humans and in animal models of DM (11). The serotonin content in the hypothalamus and brain stem of streptozotocin-diabetic rats is reduced (145), and changes in the expression and function of brain dopamine and serotonin receptors were also observed in the brain cortices of such animals (139). Thus, although only the density of the 5HT2a receptor was increased, both 5HT2a and 5HT1a receptors showed a reduced sensitivity in response to their agonists, which was restored to the values found in the control animals after insulin treatment (139).

Depression has negative effects on glycemic control, being a high-risk factor for the development of DM (146). It was suggested that the persistent activation of the hypothalamus–pituitary–adrenal gland axis in depression may facilitate the occurrence of T2DM (147).

Depression is also associated with neuroinflammatory and neurodegenerative processes (148). Thus, the levels of proinflammatory cytokines are increased in the blood plasma, CSF, and samples of postmortem brains of patients with severe depression; interestingly, treatment with antidepressant drugs normalizes the elevated levels of cytokines (149). The administration of cytokines, such as IL-1β, induces anxiety behavior and depression, whereas treatment with anti-inflammatory agents, such as long-chain polyunsaturated fatty acids, produces antidepressant effects (150, 151).

One of the mechanisms described as responsible for the depressive state induced by inflammation is altered brain monoaminergic neurotransmission. Accordingly, the central administration of IL-1β increases the production of dopamine, norepinephrine and serotonin, which also occurs in diabetic brains (150).

As previously mentioned, insulin dysfunction plays a crucial role in neuroinflammatory processes. Thus, neuroinflammation might be a common phenomenon that connects the neurodegenerative and neuropsychiatric diseases with insulin dysfunction (152, 153).

Molecular, functional and structural changes have been detected in the brains of major-depressive-disorder patients, mainly by means of imaging techniques such as magnetic resonance imaging (MRI) and positron emission tomography (PET) (154). An anomalous default mode network, as revealed by resting-state functional MRI, is probably associated with aberrant metabolic and serotonergic functions. These findings indicate that further investigations are important to shed light on the serotonergic network system associated with the behavior and genetic variations in major depressive disorder.