Introduction

Depression affects approximately 332 million people worldwide and is a leading cause of disability globally (Rong et al. 2025). According to Hasin et al. (2018), over 10.4% of U.S. adults met DSM-5 (Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition) criteria for major depressive disorder (MDD) within the previous 12 months, while the lifetime prevalence reached 20.6% (n = 36,309). Results from the STAR*D trial (Sequenced Treatment Alternatives to Relieve Depression) showed that remission rates decreased with each successive treatment step, reaching 36.8% after the first treatment, 30.6% after the second, 13.7% after the third, and 13.0% after the fourth. Overall, the cumulative remission rate was 67% (Rush et al. 2006).

Treatment-resistant depression (TRD) is a subtype of MDD characterized by a failure to respond to two or more antidepressant regimens despite adequate dose, duration, and treatment adherence. It is estimated that at least 30% of individuals with MDD develop TRD (McIntyre et al. 2023). The European Group for the Study of Resistant Depression (GSRD) TRD-III study (n = 1,410) showed that symptom severity, psychotic symptoms, suicidal risk, generalized anxiety disorder, inpatient status, a higher number of previously used antidepressants, more lifetime depressive episodes, and a longer duration of the current episode were associated with an increased risk of TRD. Patients whose depressive episode lasted three months or longer were 2.6 times more likely to develop TRD, and each additional depressive episode increased the risk by 15% (Kautzky et al. 2019). Poor treatment response in depression may reflect the heterogeneity of the disorder and the lack of treatments tailored to individual patient characteristics rather than true resistance. This highlights the need for more personalized treatment approaches and new therapies for patients who do not respond to currently available options. Several clinical factors have been associated with TRD, although its neurobiology and the mechanisms underlying different depression subtypes remain unclear (Ionescu et al. 2015).

The aim of this paper is to present current concepts regarding the pathogenesis of TRD and their therapeutic implications, as well as to discuss emerging therapeutic approaches that address the limitations of current first-line treatments. Traditional antidepressants, such as selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs), usually require at least two weeks to exert a therapeutic effect and are often associated with adverse effects, including sexual dysfunction, which can limit adherence (Lowe et al. 2021; Patacchini and Cosci 2021). These limitations underscore the need for faster-acting and more precise biomarker-guided treatment alternatives.

This article was prepared as a narrative literature review. A non-systematic search of the PubMed and Google Scholar databases was conducted to identify publications on TRD, its neurobiological mechanisms, and emerging treatment strategies. The search included English-language publications using combinations of the following keywords: treatment-resistant depression, major depressive disorder, ketamine, psilocybin, glutamate, neuroplasticity, brain-derived neurotrophic factor (BDNF), inflammation, kynurenine pathway, mitochondrial function, precision psychiatry, and biomarkers. Original research articles, systematic reviews, meta-analyses, narrative reviews, and relevant preclinical and clinical studies were included. Additional publications were identified through manual screening of the reference lists of selected articles.

Monoaminergic dysfunction

The monoaminergic hypothesis is one of the earliest frameworks explaining the pathophysiology of depression, proposing that symptoms arise from dysregulation of serotonin, noradrenaline, and dopamine in the central nervous system (CNS) (Delgado 2000; Hillhouse and Porter 2015). This concept is supported by observations that drugs increasing monoamine availability produce antidepressant effects (Delgado 2000). However, the model has important limitations. It does not account for the lack of response in a substantial proportion of patients or the delayed clinical improvement despite rapid neurochemical changes (Hillhouse and Porter 2015; Kajumba et al. 2024). Moreover, consistent evidence for a uniform monoamine deficiency in all patients is lacking, suggesting a more complex pathophysiology (Kajumba et al. 2024). Increasing attention has been given to dopamine, particularly in relation to reward processing, motivation, and anhedonia (Dunlop and Nemeroff 2007). The limited efficacy of predominantly serotonergic treatments further indicates that dopaminergic dysfunction may play a significant role, especially in TRD (Dunlop and Nemeroff 2007; Kajumba et al. 2024). Therefore, monoaminergic disturbances are now viewed as only one component of a broader, multifactorial model of depression (Hillhouse and Porter 2015).

Direct modulation of the serotonergic system through activation of postsynaptic 5-HT1A and 5-HT2A receptors may lead to rapid antidepressant effects, as observed with the psychedelic 5-methoxy-N,N-dimethyltryptamine (5-MeO-DMT) (Reckweg et al. 2022). This substance is a naturally occurring tryptamine found in several Amazonian plants and in the venom of the Sonoran Desert toad (Yost et al. 2022). 5-MeO-DMT mainly acts as an agonist at serotonin receptors, showing the highest affinity for the 5-HT1A receptor, which is involved in mood regulation, and for the 5-HT2A receptor, which is responsible for psychedelic effects (Kaufman et al. 2016; Kaumann and Levy 2006). Preclinical studies also suggest that activation of 5-HT1A may affect thalamocortical rhythms, neuroendocrine functions, and immune and anti-inflammatory processes that may play a role in the pathophysiology of TRD (Krebs-Thomson et al. 2006; Riga et al. 2016). The intensity of the psychedelic experience appears to correlate with therapeutic effects (Ott 2001). Unlike classical antidepressants, which indirectly increase monoamine levels, 5-MeO-DMT directly activates serotonin receptors, which may partly bypass resistance mechanisms seen in TRD. The pharmacokinetic properties of 5-MeO-DMT are also clinically important. Its metabolism depends on CYP2D6 activity, and genetic differences in this enzyme may influence the intensity of psychoactive effects and therapeutic response (Carbonaro et al. 2018; Holze et al. 2022). In addition, monoamine oxidase inhibitors (MAOIs) may increase systemic exposure and the risk of adverse effects (Shen et al. 2011; Jiang et al. 2016).

Glutamatergic dysfunction

Increasing evidence suggests that dysregulation of the glutamatergic system plays an important role in the pathophysiology of mood disorders, making it a potential target for new treatment strategies. It participates in most of the information-processing routes in the CNS (Sanacora et al. 2008; Szpręgiel and Bysiek 2024). In the CNS, glutamate is the main excitatory neurotransmitter. Activated presynaptic neurons release glutamate into the synaptic space, where it binds to metabotropic and ionotropic receptors, such as NMDA (N-methyl-D-aspartate), AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid), and kainate receptors, leading to further excitatory signaling. Glutamate is removed from the synapse mainly through uptake by astrocytes, where it is converted into glutamine and transported back to neurons as part of the glutamate/γ-aminobutyric acid (GABA)-glutamine cycle (Bak et al. 2006). When glutamate release exceeds the ability of astrocytes to clear it, excess glutamate may accumulate in the extrasynaptic space and contribute to excitotoxicity (Danbolt 2001). According to Candee et al. (2023), excessive glutamate release contributes to neuronal pathology through three main mechanisms:

1. Activation of the NMDA receptor leads to calcium influx into the cell, and excessive calcium accumulation may result in neuronal damage or death (Mark et al. 2001);

2. Glutamate in the extrasynaptic space binds to extrasynaptic NMDA, AMPA, and kainate receptors. Activation of extrasynaptic NMDA receptors inhibits the synthesis and release of BDNF, whereas AMPA receptor activation may increase BDNF levels in postsynaptic neurons (Aleksandrova et al. 2017);

3. Glutamate in the extrasynaptic space can activate microglial cells, promoting neurotoxicity by stimulating the release of pro-inflammatory cytokines, nitric oxide, and additional glutamate. This increases the risk of overwhelming astrocytes with excess glutamate (Haroon et al. 2017).

The glutamatergic pathway is involved in the mechanism of action of the rapid-acting antidepressant ketamine. Ketamine blocks NMDA receptors on GABAergic interneurons, decreasing inhibitory GABA signaling and thereby enhancing glutamate release from presynaptic neurons. Glutamate activates AMPA receptors on postsynaptic neurons, initiating a cascade of intracellular events involving the mTOR (mechanistic target of rapamycin) pathway, elevating BDNF levels (Antos et al. 2024). Non-ketamine NMDA receptor antagonists such as memantine and lanicemine showed limited antidepressant efficacy, suggesting that simple NMDA receptor blockade alone may be insufficient to reproduce ketamine’s antidepressant effects. Evidence indicates that ketamine’s therapeutic action may additionally involve downstream modulation of glutamatergic neurotransmission, enhanced AMPA receptor signaling, and promotion of synaptic plasticity. A double-blind study in TRD outpatients using forehead electroencephalography (EEG) demonstrated increased alpha power and reduced asymmetry in treatment responders, pointing to potential biomarkers of efficacy (Cao et al. 2019). Similarly, dynamic causal modeling (DCM) in TRD patients revealed that an enhanced excitatory-inhibitory balance correlated with a greater clinical response (Fagerholm et al. 2021). Pharmacokinetically, ketamine undergoes rapid hepatic metabolism via CYP2B6, CYP3A4, and CYP2C9, with norketamine and (2R,6R)-hydroxynorketamine (RR-HNK) as key metabolites. It has a short plasma half-life (2-3 h), while metabolites persist longer and likely contribute to sustained antidepressant effects. Pharmacodynamically, ketamine’s NMDA antagonism initiates downstream glutamate release, AMPA receptor activation, and BDNF signaling, facilitating synaptic plasticity and rapid mood improvement (Abuhelwa et al. 2022; Antos et al. 2024).

Another compound with potential in the treatment of TRD is psilocybin, a psychedelic that exerts antidepressant effects through the glutamatergic cascade (Haikazian et al. 2023; Lowe et al. 2021). After oral administration, psilocybin is rapidly converted into its active metabolite, psilocin, which – due to its structural similarity to serotonin – has a high affinity for serotonergic receptors, particularly 5-HT2A receptors (Dinis-Oliveira 2017; Patra 2016; Holze et al. 2022; Ling et al. 2022). Activation of postsynaptic 5-HT2A receptors on pyramidal neurons in the prefrontal cortex increases glutamate release and subsequent AMPA receptor activation, stimulating neuroplasticity-related pathways such as BDNF and mTOR (Aghajanian and Marek 1999; Scruggs et al. 2000; de Vos et al. 2021; Ling et al. 2022). Psilocin also modulates excitatory and inhibitory neurotransmission, altering brain connectivity observed in functional magnetic resonance imaging (fMRI) studies and affecting neural circuits involved in depression, including the default mode network (DMN) and amygdala (Lee and Roth 2012; Ling et al. 2022). These mechanisms are believed to underlie the rapid and sustained antidepressant effects of psilocybin (Haikazian et al. 2023; Ling et al. 2022; Raison et al. 2023). In summary, psilocybin has direct serotonergic and indirect glutamatergic effects, as well as other low-affinity molecular targets (Lowe et al. 2021; Ling et al. 2022).

GABA dysfunction in depression and GABA-glutamate interactions

GABA is present in most presynaptic and postsynaptic neuronal terminals. In addition to its inhibitory properties, GABA also acts as a trophic factor regulating proliferation, neuroblast migration, dendritic growth, and synapse formation during early embryonic development (Shao et al. 2021). Abnormalities in its release are associated with the development of neurological and psychiatric disorders (Shao et al. 2021). GABA synthesis requires the enzyme glutamic acid decarboxylase (GAD) and glutamate (GLU) as a substrate. Activation of the glutamatergic pathway by psilocybin acting on 5-HT2A receptors in the prefrontal cortex appears to be followed by excitatory transmission to subcortical structures, leading to changes in the activity of other neuronal populations, including GABAergic interneurons (Szpręgiel and Bysiek 2024). 5-HT1A receptors are also localized on GABAergic interneurons, and psilocybin shows lower affinity for these receptors as well. Therefore, its pharmacological effects may result from the interplay between inhibitory 5-HT1A and excitatory 5-HT2A receptor activation (Szpręgiel and Bysiek 2024). These observations suggest that the GLU/GABA ratio in both the medial prefrontal cortex (mPFC) and hippocampus may play an important role in modulating the pharmacological effects and adverse effects of psilocybin.

Neuroplasticity and neurogenesis

According to the review by Yang et al. (2020), impaired neuroplasticity is associated with abnormal changes in neurogenesis, axon branching, dendritic structure, and synaptic functioning. Abnormalities in neuroplasticity may be related to changes in the levels of neurotrophic factors, especially BDNF, whose synthesis and secretion depend on neuronal activity (Yang et al. 2020). It has also been suggested that neuronal plasticity may play an important role in the response to antidepressant treatment, as antidepressants have been shown to increase BDNF expression and signaling (Castrén and Kojima 2017; Yang et al. 2020). Stress may reduce BDNF synthesis at the transcriptional level, leading to disruptions in neuroplasticity, which may be one of the mechanisms contributing to the development of depression (Yang et al. 2020). In a randomized controlled trial conducted by Haile et al. (2014) involving 22 patients with TRD who received intravenous ketamine, it was found that 240 minutes after infusion, ketamine (unlike midazolam used as a control substance) significantly increased plasma BDNF levels in patients who responded to treatment compared to non-responders. In addition, BDNF levels at 240 minutes were negatively correlated with MADRS scores in patients receiving ketamine, but not midazolam. It was also shown that BDNF levels at 240 minutes significantly predicted MADRS scores up to 72 hours after ketamine infusion. Laboratory studies have shown that a single dose of either ketamine or psilocybin leads to the formation of new dendritic spines in the dorsomedial frontal cortex. The effect is long-lasting and persists for at least one month. Such findings may explain how, despite a short half-life in the body, psychedelics can cause sustained changes in behavior (Jiang et al. 2026).

In an exploratory placebo-controlled within-subjects study involving 28 healthy adults with no previous psychedelic experience, participants received oral psilocybin in two doses: first 1 mg (a subthreshold dose), and after 4 weeks 25 mg (an active dose). One month after the active dose, increases in cognitive flexibility, psychological insight, and psychological well-being were observed. Diffusion tensor imaging (DTI) analysis showed bilateral decreases in axial diffusivity in tracts connecting the prefrontal cortex with subcortical regions, which correlated with reduced brain network modularity in fMRI studies (Lyons et al. 2026).

HPA axis dysregulation

Beyond other mechanisms, dysregulation of the hypothalamic–pituitary–adrenal (HPA) axis is increasingly recognized as a key component of depression pathophysiology (Menke 2024; George et al. 2025). Patients with MDD frequently exhibit sustained activation of the stress response system, reflected in elevated cortisol levels and impaired negative feedback (Menke 2024; George et al. 2025). The hippocampus plays a central role in this process due to its high sensitivity to glucocorticoids and involvement in HPA axis regulation (Lei et al. 2025). Chronic stress and prolonged cortisol exposure lead to structural and functional alterations in the hippocampus, including reduced neurogenesis, dendritic remodeling, and synaptic dysfunction (Lei et al. 2025). These changes are associated with decreased hippocampal volume and impairments in memory and emotional regulation (Lei et al. 2025; Lin et al. 2024). Importantly, neuroimaging meta-analyses in TRD further confirm hippocampal structural and functional abnormalities, supporting the clinical relevance of these mechanisms (Miola et al. 2023). In turn, hippocampal dysfunction may weaken inhibitory control over the HPA axis, contributing to a self-perpetuating cycle of stress dysregulation (Menke 2024; Lei et al. 2025).

In a study conducted on 43 hospitalized patients at the Department of Adult Psychiatry, Poznań University of Medical Sciences, diagnosed with TRD during the course of bipolar disorder or major depressive episodes, total sleep deprivation (TSD) was used as an augmentation strategy for pharmacotherapy. Clinical response, defined as a sustained 50% reduction in Hamilton Depression Rating Scale (HDRS) scores, was achieved in 42.9% of patients, while 28.6% achieved remission (≤ 7 HDRS points by day 14). The study demonstrated a beneficial effect of chronotherapy, such as TSD with sleep phase advance, as an augmentation strategy for pharmacotherapy in TRD. This effect was associated with changes in cortisol, interferon γ, thyroid hormones, prolactin, and dopamine levels (Kurczewska et al. 2021).

Inflammation

Approximately one-third of patients with depression exhibit elevated inflammatory markers, supporting the concept of an inflammatory subtype of the disorder (Osimo et al. 2019; Pariante 2021; Křenek et al. 2023). Growing evidence suggests that neuroinflammatory mechanisms contribute to a worse prognosis, treatment resistance, and higher mortality rates in depression (Hassamal 2023). Patients with elevated C-reactive protein (CRP) levels above 3 mg/l may present a subtype of depression resembling “sickness behavior”, which includes anhedonia, apathy, reduced appetite, fatigue, excessive sleepiness, pain, suicidality, and cognitive dysfunction (Maes et al. 2012). Preclinical studies suggest that different types of chronic stress may contribute to the development of a melancholic subtype of depression, increase insulin resistance and enhance the production of pro-inflammatory cytokines. In animal models, elevated levels of peripheral inflammatory markers such as interleukin (IL)-1β, IL-6, and tumor necrosis factor α (TNF-α) are regularly observed (Hodes et al. 2015).

In the review by Yang et al. (2019), 10 articles involving a total of 540 participants (440 patients and 100 healthy controls) were analyzed. The study showed that inflammatory markers such as IL-6 and CRP/hs-CRP may have potential as predictive markers of treatment response in treatment-resistant depression. Raison et al. (2013) conducted a randomized, double-blind study involving 60 patients with moderate treatment-resistant depression (based on the Massachusetts General Hospital Staging Method). The study examined the effects of infliximab, a TNF antagonist given in three infusions (at baseline and at weeks 2 and 6, dose of 5 mg/kg), on depressive symptoms measured using the HDRS. The researchers also analyzed whether baseline inflammatory marker levels could predict treatment response. Infliximab was not effective in all patients with TRD. However, greater improvement was observed in patients with baseline hs-CRP levels above 5 mg/l. In this group, treatment response (defined as a ≥ 50% reduction in HDRS score) was achieved in 62% of patients treated with infliximab, compared with 33% of patients receiving placebo.

Strawbridge et al. (2019) conducted a longitudinal study involving 36 patients with TRD. The study examined depressive symptom severity and levels of 27 inflammatory proteins before hospitalization, after inpatient treatment, and during follow-up 3-12 months after discharge. The results showed that patients with TRD had higher levels of many inflammatory markers compared to the control group. In addition, elevated levels of IL-6, IL-8, TNF-α, CRP, and macrophage inflammatory protein-1 were associated with a poorer treatment response. Due to the small sample size, these findings need further confirmation in larger patient populations.

The kynurenine pathway plays an important role in linking inflammation with CNS dysfunction by reducing tryptophan availability and increasing the production of neurotoxic metabolites. During inflammation, activation of indoleamine 2,3-dioxygenase increases the conversion of tryptophan into kynurenine, reducing its availability for serotonin synthesis – a process known as the “kynurenine shunt”. Kynurenine can then be metabolized into neuroprotective kynurenic acid or neurotoxic quinolinic acid, which increases oxidative stress and glutamatergic excitotoxicity. Disturbances in this pathway have been observed in patients with depression and may play an important role in the pathophysiology of TRD (Réus et al. 2015).

In a study by Schwieler et al. (2016), cytokine levels and kynurenine pathway metabolites were assessed in patients with MDD treated with electroconvulsive therapy (MDD, n = 19; healthy controls [HC], n = 14). Patients showed lower levels of kynurenic acid and a higher quinolinic acid to kynurenic acid ratio (QUIN/KYNA), which may suggest increased NMDA receptor stimulation. Electroconvulsive therapy (ECT) treatment was associated with reduced quinolinic acid levels in 80% of patients and a lower QUIN/KYNA ratio, suggesting that the effectiveness of ECT may be partly related to modulation of the neurotoxic branch of the kynurenine pathway.

Recent studies have highlighted the role of the NOD-like receptor family pyrin domain-containing 3 (NLRP3) inflammasome as a potential link between chronic stress, neuroinflammation, and treatment-resistant depression (Ding et al. 2025; Han et al. 2024). Activation of the NLRP3 inflammasome in microglial cells promotes the release of pro-inflammatory cytokines, particularly IL-1β and IL-18, contributing to neuroinflammatory processes, impaired neuroplasticity, and neuronal dysfunction (Han et al. 2024). Experimental and clinical evidence suggests that NLRP3 inflammasome activation is associated with chronic stress exposure, microglial activation, and inflammatory dysregulation observed in depressive disorders (Han et al. 2024; Ding et al. 2025). Increasing evidence indicates that inhibition of NLRP3-related signaling pathways may represent a promising therapeutic strategy in mood disorders (Ding et al. 2025). In addition to selective inflammasome inhibitors, recent translational approaches have explored the repurposing of U.S. Food and Drug Administration (FDA)-approved drugs with potential NLRP3-modulating properties identified through machine learning and molecular simulation models, highlighting new possibilities for targeted anti-inflammatory interventions in psychiatric disorders (Agarwal et al. 2025).

According to current theories of depression pathophysiology, which include impaired neuroplasticity, peripheral and microglial inflammation, as well as disturbances in purine metabolism and cellular energy balance, increasing attention is being given to substances that may modulate allostatic processes. These therapeutic strategies include both anti-inflammatory drugs and compounds affecting energy metabolism and mitochondrial function. One of the investigated approaches is the modulation of inflammatory responses by inhibiting pro-inflammatory cytokines. TNF-α blockade with infliximab did not show clear effectiveness in the general population; however, potential benefits were observed in patients with elevated CRP levels. Promising results have also been reported for minocycline, a tetracycline antibiotic with strong anti-inflammatory properties and good penetration into the CNS. Its therapeutic potential was particularly observed in patients with elevated CRP levels (CRP ≥ 3 mg/l) (Nettis 2021; Nettis et al. 2021). Overall, these findings suggest that inflammatory dysregulation and other neuroimmunometabolic mechanisms may play an important role in the pathophysiology of a subgroup of patients with TRD and could represent promising therapeutic targets.

Cellular energy metabolism

Another important area of research includes substances affecting cellular energy metabolism. Creatine, a nitrogen-containing compound synthesized from arginine, glycine, and methionine, helps regenerate adenosine triphosphate (ATP) through the phosphocreatine system and supports brain energy metabolism under metabolic stress associated with depression. It has been suggested that creatine may enhance antidepressant treatment, especially when combined with psychotherapy or physical activity, by improving brain energy metabolism and reducing inflammation (Norouziasl et al. 2024). Preclinical studies suggest that creatine may exert antidepressant-like effects by improving mitochondrial bioenergetics, reducing oxidative stress, and modulating pathways related to neuroplasticity. However, clinical evidence in humans remains limited and heterogeneous (Juneja et al. 2024; Marshall et al. 2022).

Among mitochondrial modulators, N-acetylcysteine (NAC) currently has the strongest evidence of effectiveness (Liang et al. 2022). Beneficial effects have also been observed for acetyl-L-carnitine (ALCAR), with meta-analyses showing a reduction in depressive symptoms compared to placebo. Proposed mechanisms include epigenetic regulation and enhancement of mitochondrial plasticity (Veronese et al. 2018; Wang et al. 2014). Omega-3 fatty acids, especially eicosapentaenoic acid (EPA)-enriched formulations, have also demonstrated antidepressant potential in meta-analyses (Norouziasl et al. 2024).

Taken together, targeting cellular energy metabolism and mitochondrial function may offer a promising adjunctive treatment strategy for TRD. However, further large-scale clinical studies are needed to clarify the efficacy of these interventions.

Structural and functional brain abnormalities

Due to the rapidly growing number of studies investigating neuroimaging, genetics, epigenetics, and personalized treatment approaches in treatment-resistant depression, these areas are discussed only briefly in the present review, with emphasis placed on clinically relevant and emerging findings.

In an exploratory coordinate-based meta-analysis, Miola et al. (2023) included eight magnetic resonance imaging (MRI) studies involving 555 participants and reported three main findings. First, no significant voxel-based morphometry (VBM) differences were identified among patients with TRD, patients with TSD, and HC. Second, when resting-state fMRI studies were combined, a single cluster in the cerebellum/pons showed a marginal difference between TRD and HC. Finally, multimodal analyses revealed a significant cluster in the precentral/superior frontal gyrus when comparing patients with TRD and HC. These findings support the role of frontal regions in the pathophysiology of TRD; however, further studies using multimodal and task-based approaches are needed to better characterize the role of these regions in treatment resistance.

Other neuroimaging studies have also suggested abnormalities in fronto-limbic circuits involved in emotional regulation, reward processing, and cognitive control. Structural and functional alterations affecting the anterior cingulate cortex, hippocampus, and prefrontal regions may contribute to impaired stress regulation and reduced treatment responsiveness in TRD (Runia et al. 2022; Miola et al. 2023). However, currently available findings remain heterogeneous and have not yet resulted in clinically applicable neuroimaging biomarkers.

The default mode network (DMN) is a network of brain regions that shows higher activity during rest, for example when a person is awake with their eyes closed, and its activity decreases during goal-directed tasks (Raichle et al. 2001). The DMN includes regions of the dorsal and ventral medial prefrontal cortex, the medial and lateral parietal cortex, and parts of the medial and lateral temporal cortex (Sheline et al. 2009). The DMN plays an important role in internal mentation processes, which allow people to mentally simulate the past and the future, reflect on the mental states of others, and assign value to personally meaningful information (Andrews-Hanna 2012).

In a study by Sheline et al. (2009), fMRI was used to examine changes in DMN activity during an emotional reappraisal task in 20 patients with MDD and 21 HC. The results showed significant differences in DMN activity between patients with MDD and the control group, both during passive viewing of emotional stimuli and during emotion regulation tasks. In a systematic review by Runia et al. (2022), which included an analysis of 29 studies involving approximately 1,350 patients, changes within the DMN were identified as a potential neurobiological feature that may explain differences between TRD and treatment response in MDD. The review showed that TRD is associated with reduced connectivity (hypoconnectivity) within the DMN. Reduced functional connectivity (FC) was observed between different brain regions within the DMN and between DMN regions and other brain areas, including sensory and association areas as well as regions involved in executive and affective-limbic networks (Runia et al. 2022).

Genetic and epigenetic factors

Knowledge about the genetic basis of TRD remains limited. The most consistent findings so far come from candidate gene studies, although they focus only on selected polymorphisms. Genome-wide association studies (GWAS) in TRD have not identified any single genetic variants with significant effects so far. There is also growing interest in epigenetic mechanisms, such as DNA methylation, histone modifications, and non-coding RNA, which may influence the interaction between genetic and environmental factors and could become biomarkers of treatment response in TRD in the future (Płaza et al. 2025).

Increasing attention has also been given to stress-related genes involved in neuroplasticity and HPA axis regulation, including BDNF and FK506-binding protein 5 (FKBP5). Some studies suggest that epigenetic modifications associated with chronic stress and early-life adversity may influence antidepressant response and vulnerability to treatment resistance (Płaza et al. 2025; Halaris et al. 2021). However, the effect sizes of individual genetic and epigenetic markers remain relatively small, limiting their current clinical applicability.

The 5-HTTLPR polymorphism in the serotonin transporter gene (SLC6A4) is one of the most widely studied genetic variants related to depression and treatment response. Although large meta-analyses have not confirmed a strong association between this polymorphism and the risk of depression, some studies suggest its interaction with stress-related factors (Halaris et al. 2021). More consistent findings have been reported for antidepressant response: a meta-analysis of 49 studies showed that in Caucasian patients, SL and LL genotypes were more often associated with a better response to SSRI treatment and remission, while the S allele may be associated with a higher risk of adverse effects (Ren et al. 2020). The clinical usefulness of these findings remains limited because they explain only a small part of treatment response variability (Halaris et al. 2021).

Despite the limited predictive value of single polymorphisms, pharmacogenomic approaches combining multiple genetic variants are increasingly studied in the context of personalized treatment for TRD. Pharmacogenomic (PGx) testing analyzes the influence of genetic variants on a patient’s response to antidepressant treatment. The randomized GUIDED trial (Genomics Used to Improve DEpression Decisions) included 1,167 patients with major depression and an inadequate response to at least one antidepressant. Patients whose treatment was modified according to their pharmacogenomic profile showed higher response rates (28.5% vs. 16.7%) and remission rates (21.5% vs. 8.5%) compared to standard treatment. These findings suggest that PGx testing may help identify medications that are potentially unsuitable for a given patient, allowing for a more informed and personalized approach to treatment (Halaris et al. 2021).

Personalized therapies in the mechanistic treatment of comorbid diseases

Emerging evidence suggests that TRD should be viewed not only as a heterogeneous psychiatric disorder but also as a condition involving systemic inflammatory, neurotrophic, metabolic, and neuroendocrine dysregulation (Carvalho et al. 2020; Kas et al. 2025). Shared inflammatory and neurotrophic pathways observed across psychiatric and somatic disorders, including alterations in cytokine signaling, CRP, and BDNF, may contribute to overlapping mechanisms linking depression with cardiovascular and other chronic systemic diseases (Fioranelli et al. 2023; Medved et al. 2024). These findings support the growing concept of biomarker-informed treatment selection, in which inflammatory and neurobiological profiles could help identify biologically distinct subgroups of patients and guide more personalized therapeutic strategies in TRD (Béjar-Botello et al. 2026; Jones and Nemeroff 2021).

Such approaches may be particularly relevant in patients with coexisting inflammatory, cardiovascular, or metabolic disorders, where shared molecular mechanisms could influence both psychiatric symptoms and treatment outcomes (Carvalho et al. 2020; Fioranelli et al. 2023). Nevertheless, further validation studies are needed before biomarker-guided treatment strategies can be routinely implemented in clinical practice.

Although the clinical applicability of these approaches remains limited, ongoing advances in molecular psychiatry, pharmacogenomics, and translational medicine may contribute to the future development of mechanistically targeted and individualized treatment models.

Limitations

This narrative review has several limitations. It was not based on a strict search and study selection protocol, which may increase the risk of selection bias. The discussed disease mechanisms and treatment strategies come from studies with different methodologies, including preclinical, observational, neuroimaging, and clinical studies. Another limitation is the high biological and clinical heterogeneity of treatment-resistant depression, which makes it difficult to identify universal mechanisms of treatment resistance. Although many biomarkers and new treatment approaches show promising potential, their clinical use is still limited due to the small number of studies, small patient groups, and short follow-up periods. Many of the concepts presented require further confirmation in future research.

Conclusions

The findings reviewed in this paper suggest that TRD is not caused only by problems in the monoamine system. Increasing evidence shows that glutamate and GABA signaling, impaired neuroplasticity, chronic activation of the HPA axis, neuroinflammation, and disturbances in cellular energy metabolism also play important roles. Understanding these mechanisms not only improves our knowledge of TRD but also points to new treatment targets.

From a pharmacological perspective, treatments that affect the glutamatergic system and neuroplasticity are especially important. Ketamine is a well-known example of this approach. Interest in psychedelic-assisted therapies is also growing. However, these treatments are still being studied and are not currently considered standard treatments for TRD. The therapeutic potential of psilocybin and 5-MeO-DMT should be evaluated based on current clinical evidence.

Among psychedelic substances, psilocybin currently has the strongest scientific support. Several randomized clinical trials have shown that it can produce rapid and long-lasting antidepressant effects, especially when combined with structured psychotherapy. Research on 5-MeO-DMT is still at an early stage and is based mainly on observational studies and small early-phase clinical trials. Although the results are promising, there is not yet enough evidence to conclude that its effectiveness is comparable to that of psilocybin. Therefore, both psilocybin and 5-MeO-DMT should currently be considered experimental treatments that are being evaluated in controlled clinical settings rather than established treatments for depression.

Growing evidence also suggests that immune dysfunction and neuroinflammation play an important role in the development of TRD in at least some patients. Higher levels of inflammatory markers such as CRP, IL-6, and TNF-α are associated with poorer treatment outcomes and may identify a biologically distinct subgroup of patients. Disturbances in the kynurenine pathway and activation of the NLRP3 inflammasome may also play an important role by linking chronic inflammation with impaired neuroplasticity, altered glutamatergic signaling, and changes in neuronal function. Studies of anti-inflammatory drugs such as infliximab and minocycline suggest that their effectiveness may depend on the patient’s baseline level of inflammation. This highlights the potential value of inflammatory biomarkers in selecting patients for treatments targeting immune-related mechanisms.

Some patients may also benefit from interventions that improve mitochondrial function and cellular energy metabolism, such as creatine, N-acetylcysteine (NAC), and omega-3 fatty acids. In addition, the identification of biomarkers related to inflammation, brain network function, and genetic profile is becoming increasingly important, as it may help clinicians choose the most effective treatment for each patient.

Current research suggests that more effective treatment of TRD may require moving beyond an exclusive focus on the monoamine system and instead targeting the specific biological mechanisms that contribute to persistent symptoms in individual patients. Advances in biomarkers and precision psychiatry may eventually allow for more personalized treatment approaches and better outcomes for patients with TRD.

Disclosures

This research received no external funding.

Institutional review board statement: Not applicable.

The authors declare no conflict of interest.