Pathogenesis and neurobiology of depression

Article written and reviewed by Dr Luca Moretti

Depression does not have a single identifiable biochemical cause. Contemporary knowledge instead describes a multifactorial and multiscale pathogenesis in which genetic predisposition, development, environmental experiences, stress response, neurotransmission, synaptic plasticity, immune and metabolic systems, circadian rhythms and the functional organization of brain networks can interact in different combinations.

Most of the strongest biological evidence concerns major depressive disorder, but studies use populations, diagnostic definitions and phenotypes that are not always perfectly comparable. Findings identified in research should therefore not be automatically generalized to every form of depression or interpreted as alterations that must necessarily be present in an individual patient.

The same clinical diagnosis can in fact be reached through very different combinations of symptoms. Two patients who meet the criteria for a major depressive episode may differ in anhedonia, anxiety, insomnia or hypersomnia, psychomotor slowing or agitation, appetite, cognition, psychotic features, stress response and comorbidity. This phenotypic heterogeneity is consistent with the absence of a single neurobiological alteration shared by all cases.

Theories developed during the twentieth century instead tended to seek a dominant dysfunction. The most influential was the monoamine hypothesis, according to which depression would result from insufficient activity of serotonin, norepinephrine or dopamine. Models centered on the hypothalamic pituitary adrenal axis, BDNF, inflammation, glutamate and neuroplasticity were subsequently proposed. None of these models, considered in isolation, explains the entire disorder.

The current model does not deny the involvement of these systems, but places them within a complex causal network. An alteration observed during an episode may represent a preexisting vulnerability, a consequence of stress or illness, a mechanism that contributes to symptom persistence, the effect of medications or comorbidities, or a combination of these factors. Demonstrating a biological association therefore does not amount to demonstrating causality.

Response to treatment must also be interpreted cautiously. The fact that an antidepressant modifies a particular neurotransmitter system and reduces symptoms does not demonstrate that the disorder was originally caused by a deficit in that system, just as the effectiveness of acetylsalicylic acid for pain does not imply that pain is caused by a lack of the drug. Pharmacology can, however, identify circuits and biological processes that can be therapeutically modulated.

It is also essential to distinguish findings obtained at the group level from their applicability to an individual. Mean differences in cortisol, cytokines, hippocampal volume, striatal activity or BDNF may be statistically significant between large groups of patients and controls while their distributions substantially overlap. A mean difference of this type does not automatically constitute a diagnostic biomarker.

In light of current evidence, depression can therefore be interpreted as a disorder of the adaptive regulation of systems that integrate emotion, reward, motivation, cognition, stress response, sleep and behavior. The mechanisms described in the following sections represent complementary levels of this architecture rather than mutually exclusive theories.

Genetic predisposition, environment and epigenetics

The genetic component of major depressive disorder is documented by family and twin studies, but it does not follow a Mendelian model. Estimates from twin studies suggest moderate heritability, generally in the range of 30 to 40%, indicating that an important proportion of variability in susceptibility depends on nongenetic factors and their interactions with inherited factors.

Modern genome wide studies have radically changed the concept of biological predisposition. There is no single “depression gene”: risk is highly polygenic and derives from the contribution of a very large number of common variants, each individually associated with an extremely small increase in the probability of disease.

The large trans ancestry study published in 2025 by the Psychiatric Genomics Consortium, which included hundreds of thousands of individuals with depression and millions of controls from numerous populations, identified hundreds of independent genomic associations distributed across hundreds of loci. The result strengthens the view that depression emerges from a network of biological processes rather than from a single molecular abnormality.

Genes implicated by genomic studies converge mainly on processes expressed in the central nervous system, including neuronal functions, development, intracellular signaling and synaptic organization. This does not mean, however, that each identified locus already has a defined causal function: a genome wide association locates a region statistically correlated with the phenotype and requires subsequent functional studies to identify the variants, genes and mechanisms actually responsible.

Polygenic risk scores aggregate the effects of numerous variants and can quantify part of genetic predisposition in research settings. Their individual discriminative ability, however, remains insufficient to diagnose the disorder or accurately predict its onset. Performance is also influenced by the ancestral population in which the scores were developed and by differences among the phenotypic definitions used in different studies.

Genetic predisposition interacts with the environment throughout life. Childhood maltreatment, early adversity, stressful events, violence, social isolation, financial hardship and chronic stress are associated with an increased risk of depression. The association is probabilistic: most people exposed to a single adverse event do not necessarily develop a depressive disorder, and major depression can also occur without a clearly recognizable precipitating event.

The relationship between genes and environment is more complex than a model in which a single genetic variant specifically increases sensitivity to stress. The best known example concerns the serotonin transporter polymorphism 5 HTTLPR, initially proposed as a moderator of the effect of stressful events. Subsequent studies and meta analyses have produced inconsistent results, illustrating the limitations of simple gene environment interactions based on individual candidate polymorphisms.

A further level is represented by epigenetics. DNA methylation, histone modifications, chromatin regulation and noncoding RNAs can modulate gene expression without changing the DNA sequence. Stress and other environmental exposures can alter these processes in experimental models, providing a biologically plausible mechanism through which environmental experiences may leave persistent molecular effects.

In humans, however, interpretation of epigenetic data is complex. Many studies use blood or other peripheral tissues, whereas epigenetic modifications are strongly dependent on cell type. A difference detected in leukocytes cannot automatically be considered representative of the prefrontal cortex, hippocampus or other brain structures.

In addition, medications, smoking, diet, physical activity, obesity, sleep, age and numerous medical conditions can modify the epigenome. It is therefore not always possible to establish whether an observed difference precedes depression, is a consequence of the episode or is an effect of exposures associated with the illness.

The contemporary genetic model is therefore one of distributed susceptibility. The genome helps determine the probability that particular neurobiological systems will respond to development and the environment in a certain way, but it does not directly encode an inevitable clinical outcome.

This framework also helps explain why the same environmental exposure can produce different outcomes in different people and why different combinations of predisposition and experience can converge on clinically similar depressive presentations.

Serotonin, norepinephrine and the monoamine hypothesis

The monoamine hypothesis is the best known historical biochemical model of depression. It emerged in the 1960s from observations of the effects of drugs that altered monoamine neurotransmission and from the subsequent development of tricyclic antidepressants and monoamine oxidase inhibitors.

The original formulation proposed that depression was determined by insufficient functional activity of norepinephrine, serotonin or both at cerebral synapses. The subsequent development of selective serotonin reuptake inhibitors contributed to the spread of the idea of a serotonergic deficit, often simplified in nonspecialist communication as a “chemical imbalance”.

This representation is not supported by contemporary evidence. A uniform reduction in the amount of brain serotonin has not been demonstrated in patients with depression, and there is no value for serotonin, its metabolite 5 HIAA, or other parameters of the serotonergic system that reliably distinguishes people with major depressive disorder from those without depression.

Studies of monoamine metabolites in cerebrospinal fluid, blood or urine have produced heterogeneous results. Moreover, peripheral concentrations do not necessarily represent neurotransmitter availability at the specific brain synapses involved in mood regulation.

Particular attention has been devoted to 5 HIAA, the principal metabolite of serotonin. Lower cerebrospinal fluid concentrations have been associated in some studies primarily with dimensions such as impulsivity and suicidal behavior, but the finding is neither sufficiently specific nor reproducible to diagnose depression or quantify suicide risk in an individual.

Another argument against the simple synaptic deficit model comes from the different time scales of pharmacological and clinical effects. SSRIs inhibit the serotonin transporter rapidly, producing neurochemical changes during the early stages of treatment, whereas the full clinical response generally develops over a much longer time scale.

This interval has led to the hypothesis that the antidepressant effect depends primarily on progressive adaptations downstream of the initial change in neurotransmission: receptor modifications, intracellular signaling, gene expression, synaptic plasticity and functional reorganization of circuits may be more relevant than the simple amount of monoamine present in the synaptic space.

Serotonergic neurotransmission nevertheless remains biologically important. Serotonergic neurons of the raphe project widely to the cortex, limbic structures, hypothalamus and other regions, modulating emotional processing, responses to stimuli, appetite, sleep, social behavior and numerous other functions. Saying that depression is not caused by a simple serotonin deficiency therefore does not mean that serotonin is irrelevant.

Similarly, norepinephrine, produced mainly by neurons of the locus coeruleus, modulates arousal, attention, responses to salient stimuli and adaptation to stress. Alterations of the noradrenergic system have been described in subgroups of patients and represent a therapeutic target of noradrenergic and serotonin norepinephrine antidepressants, without identifying a universal noradrenergic deficit.

Chronic response to antidepressants includes adaptations of autoreceptors and postsynaptic receptors, changes in monoaminergic circuit activity and interactions with glutamate, GABA, BDNF and stress systems. The direction of these adaptations varies according to pharmacological class and cannot be condensed into a single receptor sequence common to all antidepressants.

The monoamine theory therefore retains historical and pharmacological value, but does not provide a complete etiological explanation. The serotonergic and noradrenergic systems are better interpreted as modulators of brain networks that participate in depression together with numerous other molecular and circuit systems.

This change in perspective also avoids a common error: the efficacy of monoaminergic antidepressants does not logically imply that the disorder is caused by a deficiency of the substances those drugs modulate. Therapeutic efficacy instead demonstrates that acting on these systems can modify a complex pathological state.

Dopamine, reward, motivation and anhedonia

The dopaminergic system is particularly relevant for understanding some symptom domains of depression, especially anhedonia, reduced motivation, loss of initiative and alterations in reward based learning. Here too, however, a simple and uniform “dopamine deficiency” has not been demonstrated.

Dopaminergic neurons of the ventral tegmental area project to the nucleus accumbens, striatum, prefrontal cortex and other structures, forming a central component of the mesocorticolimbic reward circuits. These systems contribute to reward anticipation, the motivation required to obtain reward, associative learning and modulation of stimulus salience.

Anhedonia is not a unitary phenomenon. It is possible to distinguish at least the capacity to anticipate a reward, motivation to expend effort to obtain it, hedonic experience during consumption and learning from the outcome. These components involve partly different circuits and may be altered to different degrees in individual patients.

Functional neuroimaging studies show, at the group level, a reduced response in components of the striatum and cingulate regions during some phases of reward processing. Recent meta analyses confirm alterations in both reward anticipation and reward receipt, while showing differences related to the nature of the stimulus and the experimental paradigm used.

These findings provide a plausible neurobiological basis for symptoms such as loss of interest and reduced motivation, but they do not allow the conclusion that every patient with anhedonia has the same dopaminergic alteration or that a single neuroimaging test can directly measure the severity of clinical anhedonia.

Dopamine metabolites have also been studied historically, especially homovanillic acid, or HVA. Some studies have found reduced concentrations in cerebrospinal fluid in subgroups of patients, whereas others have not confirmed the finding. Methodological and biological variability prevents HVA from being used to diagnose depression or identify clinical subtypes.

HVA should therefore not be interpreted as a “measurement of brain dopamine”. Its concentration in cerebrospinal fluid reflects overall dopamine metabolism through complex processes and does not provide a selective measure of the activity of mesolimbic synapses relevant to motivation.

Clinical observations from Parkinson disease and the use of drugs that modify dopamine have also contributed to interest in this system. Depression is common in Parkinson disease, but the relationship involves neurodegeneration, involvement of nondopaminergic systems, disability, inflammation, sleep disorders and other factors. It therefore cannot be used as proof of a single dopaminergic mechanism.

Some antidepressants, such as bupropion, modify dopaminergic and noradrenergic neurotransmission, while certain pharmacological augmentation strategies indirectly involve dopamine receptors. This evidence supports the therapeutic relevance of the system without defining a specific common causal alteration.

Dopamine also interacts with glutamate, GABA, serotonin, endogenous opioids and numerous neuropeptides within reward circuits. System function depends not only on the amount of dopamine released, but also on signal timing, different receptors, circuit context and previous experience.

For this reason, the contemporary model of anhedonia is primarily circuit based and computational: depression may change the way the brain assigns value to rewards, learns from outcomes and decides how much effort to invest to obtain a possible benefit.

Dopaminergic pathophysiology is therefore an important but not autonomous component of depression. Its greatest conceptual usefulness lies in explaining specific clinical dimensions rather than defining a single mechanism responsible for the entire depressive disorder.

Glutamate, GABA, synaptic plasticity and BDNF

The discovery of the rapid antidepressant effects of ketamine helped shift attention beyond monoaminergic systems toward glutamatergic neurotransmission and mechanisms of synaptic plasticity. Glutamate is the principal excitatory neurotransmitter in the central nervous system and acts through ionotropic and metabotropic receptors distributed throughout most brain networks.

Among ionotropic receptors, NMDA receptors and AMPA receptors play a fundamental role in excitatory neurotransmission and plasticity. The balance between NMDA and AMPA activity, the synaptic and extrasynaptic localization of receptors, interneuron function and glutamate glutamine cycling contribute to regulation of network excitability and stability.

No simple glutamate concentration that is “too high” or “too low” has been identified as characteristic of depression. Studies using magnetic resonance spectroscopy and other methods have found variable differences among brain regions and clinical populations. Their meaning depends on location, phase of illness, treatment, age and methodology.

Experimental models of chronic stress show changes in dendritic structure and synapses in regions such as the prefrontal cortex and hippocampus. These data contributed to the hypothesis that persistent stress burden may impair the ability of circuits to adapt and remodel in response to experience.

Ketamine, an NMDA receptor antagonist, can produce antidepressant effects much more rapidly than traditional monoaminergic antidepressants in selected patients. Numerous experimental models attribute this effect to rapid reorganization of glutamatergic neurotransmission, relative enhancement of AMPA signaling and activation of intracellular pathways involved in the formation and stabilization of synapses.

Among these pathways, the BDNF TrkB cascade, mTORC1, ERK, Akt and other signaling systems have been studied. In animal models, activation of these pathways can promote protein synthesis, dendritic spine formation and recovery of synaptic connectivity impaired by stress.

Translation to humans, however, requires caution. The rapid clinical effect of ketamine demonstrates that modulation of glutamatergic neurotransmission can rapidly change depressive symptoms, but it does not demonstrate that the disorder is caused in all patients by a deficit of mTOR, BDNF or synaptogenesis.

The principal inhibitory neurotransmitter, GABA, has also been studied in depression. Some investigations using spectroscopy and neurochemical analyses have described alterations in GABAergic neurotransmission, but results are heterogeneous and depend on the region studied and clinical subtype.

The functional balance between glutamatergic excitation and GABAergic inhibition is essential for cortical network function. Alterations of GABAergic interneurons can change circuit synchronization and indirectly affect glutamate release and plasticity, but there is no clinical parameter capable of measuring this balance in an individual patient.

A central role in models of plasticity is attributed to brain derived neurotrophic factor, BDNF, a neurotrophin involved in neuronal survival, synaptic maturation and activity dependent plasticity. Stress and antidepressants modify BDNF expression in numerous experimental models.

Meta analyses have found lower mean serum BDNF concentrations in patients with depression than in controls and changes associated with treatment. However, circulating BDNF derives from multiple sources, is influenced by platelets, physical activity, metabolism, age and analytical methodology, and does not represent a direct measure of BDNF concentration in brain structures.

Similarly, the Val66Met variant of the BDNF gene can affect trafficking and activity dependent secretion of the protein, but associations with depression risk and treatment response are insufficient to allow reliable clinical predictions in an individual.

Adult hippocampal neurogenesis has been extensively studied in animal models. Some antidepressants increase neurogenesis, and certain experiments suggest that this phenomenon contributes to specific behavioral effects. In humans, however, the extent of adult hippocampal neurogenesis and its causal role in depression remain debated and cannot be used as an established clinical explanation.

The more robust concept today is therefore synaptic and network plasticity rather than simple neurogenesis. Stress, neurotransmitters, trophic factors, inflammation and metabolism can converge on the ability of synapses to effectively modify their strength and organization.

This convergence explains why plasticity is one of the main bridges among different pathogenic models. Apparently distinct systems, such as cortisol, serotonin, glutamate and cytokines, can all modify neuronal excitability, gene expression and the organization of synaptic connections in different ways.

Hypothalamic pituitary adrenal axis and stress response

The relationship between depression and stress has led to intensive study of the hypothalamic pituitary adrenal axis, or HPA axis, the principal neuroendocrine system involved in responses to stressors. Activation of the axis begins with hypothalamic release of CRH and vasopressin, which stimulate pituitary ACTH secretion and subsequently cortisol production by the adrenal cortex.

Cortisol exerts feedback through glucocorticoid and mineralocorticoid receptors present in many brain regions, including structures involved in emotional and cognitive regulation. Under physiological conditions, this system makes it possible to adapt metabolism, behavior and cardiovascular and immune functions to environmental demands.

In some patients with depression, alterations in cortisol secretion, circadian rhythm, the awakening response and glucocorticoid feedback have been described. Abnormalities are more evident in certain clinical phenotypes, but they are not present in all patients and their direction is not uniform.

Melancholic and psychotic forms have frequently been associated with greater HPA axis activity, whereas other phenotypes may show different patterns. This helps explain why studies that treat major depressive disorder as a single biologically homogeneous entity can produce apparently discordant results.

The dexamethasone suppression test has had major historical importance. Dexamethasone should suppress ACTH and cortisol secretion through glucocorticoid feedback; some patients with severe depression show incomplete suppression. Sensitivity and specificity are insufficient, however, and the test is not used to diagnose major depressive disorder.

Isolated cortisol measurements likewise have no diagnostic value. Secretion varies with time of day, awakening, sleep, physical activity, medications, contraceptives, obesity, smoking, physical illness and numerous other factors. There is also marked physiological interindividual variability.

Prolonged exposure to glucocorticoids can modify synaptic plasticity, metabolism, immune function and memory in experimental models. The hippocampus, prefrontal cortex and amygdala are particularly sensitive to stress mediators and in turn participate in regulation of the endocrine response.

It is not correct, however, to describe human depression through an invariant chain in which stress causes hypercortisolemia, hypercortisolemia reduces BDNF and this inevitably causes neuronal atrophy. Each step in this sequence shows heterogeneity, bidirectionality and numerous modulating factors.

The concept of allostatic load provides a more appropriate model. A stress response that is useful in the short term can become biologically burdensome when repeatedly activated or when it fails to deactivate properly. Neuroendocrine, autonomic, immune and metabolic systems are then exposed to persistent regulation that may increase vulnerability to both psychiatric and physical disorders.

The effects of stress also depend on the stage of life at which exposure occurs. Early adversity may interact with sensitive phases of brain development and maturation of stress systems, whereas stressors in adulthood act on circuits that are already developed. This contributes to differences in individual vulnerability.

Stress and depression also have a bidirectional relationship. An adverse event may precede the episode, but depression itself can increase conflict, isolation, occupational difficulties and health problems, generating additional stressors and maintaining a cycle of mutual reinforcement.

The HPA axis should therefore be interpreted as one of the major systems integrating environment and brain, not as a diagnostic test or a universal mechanism. Its importance lies primarily in linking exposure to stress with neuroendocrine adaptation, immunity, metabolism and brain plasticity.

Inflammation, immunometabolism and the kynurenine pathway

Numerous studies have documented an association between depression and immune activation. Meta analyses of case control studies show mean differences in several peripheral mediators, including C reactive protein, interleukin 6 and other cytokines. The magnitude of these effects is, however, modest and highly heterogeneous.

It is therefore not correct to state that depression is universally an “inflammatory disease”. Some patients have a phenotype associated with greater inflammatory activity, whereas many others show no significant alterations. Inflammation may also be secondary to or amplified by obesity, smoking, physical inactivity, sleep disorders, infections, chronic diseases and medications.

Peripheral immune mediators can communicate with the central nervous system through different routes, including vagal, endothelial and humoral signaling. Cytokines can modify microglial activity, metabolism, neurotransmission and stress responses, creating potential interactions between immunity and brain circuits.

The biological importance of the immune system is also demonstrated by conditions in which strong inflammatory stimulation can cause depressive symptoms. Immunomodulatory therapies and some inflammatory diseases can be associated with affective manifestations, but these conditions represent specific models and do not demonstrate that every depressive episode has the same immune origin.

One of the main interfaces between immunity and neurotransmission is tryptophan metabolism. This amino acid can be used for serotonin synthesis or metabolized through the kynurenine pathway. Enzymes that are also regulated by inflammatory signals can increase flux toward kynurenine formation.

Kynurenine is subsequently metabolized into several biologically active compounds. Among them, kynurenic acid and quinolinic acid exert different effects on glutamatergic neurotransmission; the latter can act at NMDA receptors and has neuroactive properties that have fueled interest in this pathway in depression.

Metabolic partitioning does not occur in the same way in all cells. Microglia, astrocytes and peripheral cells express different enzymes of the kynurenine pathway, so interpretation of plasma concentrations is not equivalent to direct measurement of metabolism within specific brain circuits.

It is also an oversimplification to state that inflammation causes depression simply by “using up tryptophan” and reducing serotonin. Kynurenine metabolism involves multiple metabolites with different effects and interacts with glutamate, oxidative stress, mitochondrial function and immune activity.

Microglial activation has been studied using postmortem tissue, animal models and molecular imaging techniques. Findings suggest brain immune changes in some conditions, but do not demonstrate a uniform microglial pattern and currently provide no marker suitable for clinical practice.

Astrocytes also potentially contribute to pathophysiology through control of ionic homeostasis, energy metabolism, synaptic support and the glutamate glutamine cycle. Glial alterations have been described in several studies, but their causal significance and distribution across different depressive phenotypes remain incompletely defined.

Inflammation and oxidative stress can interact. Reactive oxygen and nitrogen species are normally produced during cellular metabolism and neutralized by antioxidant systems; a persistent imbalance can interfere with membranes, proteins, DNA and mitochondrial function. Biomarkers of oxidative stress are altered in some studies of depression, but have no diagnostic value.

Inflammation has also been proposed to contribute particularly to symptoms such as anhedonia, fatigue and slowing through effects on reward circuits and energy metabolism. These associations are biologically plausible and supported by some studies, but they do not yet allow an inflammatory subtype to be defined by a single laboratory value.

From a therapeutic perspective, interest in anti inflammatory drugs does not imply that they are standard treatments for depression. Study results are variable, and the balance between benefit and risk depends on the drug, population and degree of inflammation. An elevated CRP alone is not an indication to initiate anti inflammatory antidepressant therapy.

The immunology of depression is therefore a particularly clear example of biological heterogeneity: a mechanism may be important in a subgroup of patients without being necessary to define the disorder as a whole.

Brain circuits, connectivity and neuroimaging

Depression is not associated with a characteristic focal brain lesion. Contemporary studies instead describe distributed changes in functional circuits and networks involved in emotional regulation, reward, self referential thought, stimulus salience and cognitive control.

Among the most studied regions are the medial and dorsolateral prefrontal cortex, anterior cingulate cortex, orbitofrontal cortex, insula, amygdala, hippocampus, thalamus and striatum. These structures do not function in isolation, but constitute nodes of interconnected networks.

Analyses by the ENIGMA consortium have made it possible to study thousands of magnetic resonance scans using harmonized protocols. In major depressive disorder, group level structural differences have been observed in some cortical and subcortical regions, including a lower mean hippocampal volume, particularly evident in certain populations with recurrent depression.

The magnitude of structural effects is generally small, however, and the distributions of patients and controls overlap widely. A patient with depression can therefore have a hippocampal volume well within the range observed in the healthy population and vice versa.

ENIGMA cortical studies have identified mean differences in thickness or surface area in regions such as the orbitofrontal, cingulate and insular cortices, with partly different patterns in adults and adolescents. These findings also do not constitute a diagnostic anatomical signature.

Functional neuroimaging has shifted attention from individual structures to connectivity. Among the networks most extensively investigated are the default mode network, salience network and frontoparietal cognitive control networks. Mean alterations in connectivity within and between these networks have been reported in numerous studies.

The default mode network includes regions that are particularly active during self referential thoughts and mental activity not directly oriented toward an external stimulus. Altered regulation of this network has been linked to rumination and excessive negative self referential processing, although it is not a specific mechanism exclusive to depression.

The salience network contributes to identifying biologically and psychologically relevant stimuli and to transitions between internal and external processing. Frontoparietal networks instead contribute to cognitive control, working memory and top down regulation of emotional responses.

Prefrontal and cingulate regions interact with the amygdala and other limbic structures during emotional stimulus processing. Different responses to negative stimuli and difficulties in regulating those responses have been described in depressed groups, but the pattern depends on the experimental paradigm and clinical characteristics.

The ventral striatum and other nodes of the reward network show reduced activation during reward anticipation or receipt in several studies. Recent meta analyses support the presence of mean alterations in mesocorticolimbic circuits, consistent with anhedonia and reduced motivation.

One of the main problems is the considerable heterogeneity among neuroimaging studies. Scanners, sequences, preprocessing, statistical analyses, medications, illness duration, age, symptoms and comorbidities can influence results. Small single center studies also have a high risk of overestimating effects.

Large multisite datasets and machine learning algorithms can identify distributed patterns with accuracy above chance in research samples. This does not mean, however, that a diagnostic magnetic resonance imaging test already exists: generalizability across populations, scanners and clinical settings remains a fundamental obstacle.

The main value of neuroimaging is therefore to understand how specific clinical dimensions may emerge from coordinated network dysfunction. Anhedonia, rumination, cognitive difficulties and emotional dysregulation do not necessarily have to be traced back to the same circuit.

This perspective also underlies neuromodulation. Techniques such as transcranial magnetic stimulation act on specific cortical targets and the networks connected to them, demonstrating that circuit modulation can have clinical effects. The efficacy of these interventions strengthens the network model but does not identify a single anatomical abnormality responsible for the disorder.

Circadian rhythms, sleep, metabolism and emerging mechanisms

The circadian system coordinates variations in biological functions over 24 hours through a central pacemaker located mainly in the suprachiasmatic nucleus of the hypothalamus and peripheral oscillators present in numerous tissues. Light, sleep schedules, activity, food intake and social interactions contribute to synchronization of the biological clock.

Sleep alterations are among the most common manifestations of depressive episodes. They may include initial insomnia, nocturnal awakenings, early morning awakening or hypersomnia. Sleep disorders are not merely a symptom, however: insomnia and irregular rhythms are also associated with a greater subsequent risk of depression.

Studies using actigraphy and accelerometry show associations between major depressive disorder and integrated changes in sleep, physical activity and circadian rhythm. These domains are strongly interdependent, and it is not always possible to determine which alteration represents the cause and which the consequence of an episode.

The circadian system also regulates cortisol and melatonin secretion, body temperature, energy metabolism, immune function and neurotransmission. Persistent desynchronization may therefore simultaneously influence several systems involved in depressive pathophysiology.

Depression with a seasonal pattern is one of the clearest clinical examples of the relationship among photoperiod, biological rhythms and mood. The efficacy of light therapy in this context provides further evidence of the functional importance of circadian systems, without implying that every depression is caused by a disorder of the biological clock.

Variants in circadian clock genes and alterations in the expression of rhythmic genes have been studied, but there is no circadian molecular profile that can be used to diagnose depression. Biological rhythms can also be profoundly modified by behavior, shift work, light exposure, medications and social habits.

Another area of research concerns energy metabolism and mitochondria. Mitochondria produce ATP, regulate redox signaling, calcium metabolism and apoptotic processes, and interact with stress and immunity. Differences in mitochondrial parameters have been described in mood disorders, but results are heterogeneous.

Studies of mitochondrial DNA copy number in blood, for example, have not identified a sufficiently stable pattern to become a clinical biomarker. Peripheral parameters can also reflect cellular composition, inflammation, age, medications and many other conditions.

Systemic metabolic alterations are common in patients with depression, and depression, obesity, diabetes and cardiovascular diseases have bidirectional associations. Insulin resistance, adipokines, inflammation and lifestyle can create biological connections, but there is no single depressive metabolic phenotype.

The microbiota gut brain axis is also the subject of intensive research. The intestinal microbiota can influence nutrient metabolism, production of metabolites, the intestinal barrier, the immune system, the vagus nerve and the metabolism of neuroactive precursors. Observational studies have described differences in microbial composition between depressed groups and controls.

These differences must be interpreted with particular caution because diet, medications, geography, age, body mass index, intestinal transit and numerous environmental factors profoundly modify the microbiome. Associations observed in humans do not demonstrate that a particular microbial composition causes major depressive disorder.

Microbiota transfer experiments in animals and studies with probiotics are interesting research tools, but there is currently no microbiota test capable of diagnosing depression or selecting a validated individual treatment on the basis of a specific bacterial profile.

Nutrition and physical activity can also interact with many of the pathways described, including inflammation, metabolism, BDNF, vascular function and the microbiota. These mechanisms probably contribute to the bidirectional relationships between physical and mental health, but are difficult to isolate in a single causal chain.

Additional systems under investigation include endocannabinoid, opioid, cholinergic and orexinergic signaling and several neuropeptides. Some may contribute to specific symptom dimensions or become therapeutic targets, but their role is not yet sufficiently defined to be integrated into a clinical diagnostic model.

Emerging mechanisms should therefore be evaluated according to the robustness of the evidence rather than biological plausibility alone. The history of depression research shows that many initially promising associations have not retained the same strength when replicated in larger and methodologically more rigorous populations.

Biomarkers and the integrated pathogenic model

Despite decades of research, there is currently no approved diagnostic biomarker capable of confirming or excluding major depressive disorder in an individual patient. Diagnosis remains clinical and is based on symptoms, duration, functional impairment, longitudinal course and exclusion of appropriate alternative diagnoses.

This principle applies to both historical and more modern parameters. 5 HIAA, HVA, norepinephrine, cortisol, the dexamethasone test, BDNF, CRP, interleukins and kynurenine metabolites can provide research information, but none has the performance required to diagnose depression in clinical practice.

The same consideration applies to genetics and epigenetics. Polygenic risk may be statistically associated with depression in large populations, but there is very extensive overlap between affected and unaffected individuals and it does not deterministically identify who will develop the disorder.

Structural and functional neuroimaging show reproducible differences in some large studies, but individual effect sizes are generally insufficient. Magnetic resonance imaging is used in the evaluation of a patient with depression when there is a clinical indication to investigate neurological or structural disease, not to demonstrate the presence of depression itself.

Even combining multiple parameters with machine learning algorithms has not yet produced a sufficiently validated and generalizable test to replace clinical diagnosis. A model may perform well in the dataset in which it was developed and lose substantial accuracy when applied to different centers, populations or prevalences.

One fundamental reason is the biological heterogeneity of the diagnosis. If different phenotypes share the same nosological label, it is unlikely that a single biomarker will be altered in the same way in every patient. Future approaches may therefore focus more on symptom dimensions, biological subtypes or multimodal combinations.

The absence of a diagnostic biomarker does not imply that depression lacks a biological basis. It means that its biology is distributed, dynamic and substantially overlaps with normal human variability and other psychiatric disorders. Many systems implicated in depression are normal physiological systems whose regulation may be quantitatively or contextually altered.

The pathogenic model most consistent with contemporary evidence therefore begins with vulnerability. Polygenic background and developmental experiences help determine how brain circuits, the stress axis, the immune system and metabolism respond to subsequent events.

Psychological or biological stressors can modify the HPA axis, sleep, autonomic activity, inflammation and behavior. These changes can in turn alter neurotransmission, energy metabolism and synaptic plasticity. Persistence of these changes may modify the functioning of networks involved in reward, cognitive control and emotional processing.

The clinical result may manifest as anhedonia, depressed mood, rumination, cognitive difficulties, psychomotor changes and neurovegetative symptoms. The precise combination varies from person to person because the combination of biological and psychological mechanisms contributing to the presentation also varies.

Once an episode has begun, symptoms can in turn modify the systems that contributed to vulnerability. Insomnia, isolation, inactivity, altered eating, substance use and social stress can further promote circadian, metabolic and immune dysfunction, creating maintenance loops.

Therapeutic response can interrupt these loops at different levels. Psychotherapy modifies cognitive, behavioral and emotional processes and is associated with functional changes in networks; antidepressants act initially on specific molecular targets but produce progressive adaptations; ketamine rapidly influences glutamatergic neurotransmission and plasticity; neuromodulation directly affects brain circuit activity.

The fact that biologically different treatments can produce remission is consistent with the existence of multiple routes into the same pathological system. It is therefore not necessary to assume that all treatments correct a single fundamental biochemical abnormality.

This framework also explains why precision medicine in psychiatry remains a goal rather than a fully achieved reality. Identifying reliable combinations of clinical, genetic, molecular and circuit characteristics capable of predicting the optimal treatment for an individual patient remains one of the major areas of contemporary research.

In conclusion, depression should be regarded as a multifactorial, polygenic, systemic and circuit based disorder. Serotonin, norepinephrine, dopamine, glutamate, GABA, BDNF, cortisol, immunity, metabolism and circadian rhythms participate in a network of biological interactions, but none of these elements alone represents “the cause” of depression.

The most appropriate pathophysiological sequence is therefore not a linear chain from biochemical cause to symptom, but a dynamic network in which genetic and environmental vulnerability modifies systems of adaptation and plasticity; these influence the functioning of circuits that regulate emotion, cognition, motivation and homeostasis; the resulting symptoms in turn modify environment and biology, contributing to possible persistence of the disorder.

Bibliography
1. Malhi GS et al. Depression. The Lancet. 407(10540), 2026: 1738-1756.
2. Marx W et al. Major depressive disorder. Nature Reviews Disease Primers. 9(1), 2023: 44.
3. Adams MJ et al. Trans-ancestry genome-wide study of depression identifies 697 associations implicating cell types and pharmacotherapies. Cell. 188(3), 2025: 640-652.e9.
4. Fries GR et al. Molecular pathways of major depressive disorder converge on the synapse. Molecular Psychiatry. 28(1), 2023: 284-297.
5. Moncrieff J et al. The serotonin theory of depression: a systematic umbrella review of the evidence. Molecular Psychiatry. 28(8), 2023: 3243-3256.
6. Duman RS et al. Synaptic plasticity and depression: new insights from stress and rapid-acting antidepressants. Nature Medicine. 22(3), 2016: 238-249.
7. Vreeburg SA et al. Major depressive disorder and hypothalamic-pituitary-adrenal axis activity: results from a large cohort study. Archives of General Psychiatry. 66(6), 2009: 617-626.
8. Osimo EF et al. Inflammatory markers in depression: a meta-analysis of mean differences and variability in 5,166 patients and 5,083 controls. Brain, Behavior, and Immunity. 87, 2020: 901-909.
9. Molendijk ML et al. Serum BDNF concentrations as peripheral manifestations of depression: evidence from a systematic review and meta-analyses on 179 associations (N=9484). Molecular Psychiatry. 19(7), 2014: 791-800.
10. Schmaal L et al. Subcortical brain alterations in major depressive disorder: findings from the ENIGMA Major Depressive Disorder working group. Molecular Psychiatry. 21(6), 2016: 806-812.
11. Schmaal L et al. Cortical abnormalities in adults and adolescents with major depression based on brain scans from 20 cohorts worldwide in the ENIGMA Major Depressive Disorder Working Group. Molecular Psychiatry. 22(6), 2017: 900-909.
12. Bore MC et al. Distinct neurofunctional alterations during motivational and hedonic processing of natural and monetary rewards in depression: a neuroimaging meta-analysis. Psychological Medicine. 54(4), 2024: 639-651.
13. Kang SJ et al. Integrative Modeling of Accelerometry-Derived Sleep, Physical Activity, and Circadian Rhythm Domains With Current or Remitted Major Depression. JAMA Psychiatry. 81(9), 2024: 911-918.
14. Calarco CA et al. Whole blood mitochondrial copy number in clinical populations with mood disorders: A meta-analysis. Psychiatry Research. 331, 2024: 115662.
15. Kennis M et al. Prospective biomarkers of major depressive disorder: a systematic review and meta-analysis. Molecular Psychiatry. 25(2), 2020: 321-338.
16. Marx W et al. Diet and depression: exploring the biological mechanisms of action. Molecular Psychiatry. 26(1), 2021: 134-150.
17. Dollish HK et al. Circadian rhythms and mood disorders: Time to see the light. Neuron. 112(1), 2024: 25-40.
18. Otte C et al. Major depressive disorder. Nature Reviews Disease Primers. 2, 2016: 16065.