Neuroinflammation in Psychiatric Disorders: Cytokines, Microglia, and Clinical Treatment Implications

For most of the twentieth century, the brain was considered an 'immune-privileged' organ — largely walled off from peripheral immune processes by the blood-brain barrier (BBB). Psychiatric disorders were understood almost exclusively through the lens of monoamine neurotransmitter dysfunction or psychodynamic formulations. Over the past three decades, however, a convergence of epidemiological, molecular, neuroimaging, and clinical-trial evidence has fundamentally challenged this separation. Neuroinflammation — loosely defined as the activation of the brain's innate immune system, including microglial cells, astrocytes, and locally produced cytokines — is now recognized as a transdiagnostic mechanism implicated in major depressive disorder (MDD), schizophrenia spectrum disorders, bipolar disorder, post-traumatic stress disorder (PTSD), and obsessive-compulsive disorder (OCD), among others.

This is not a fringe hypothesis. Meta-analyses encompassing thousands of patients consistently demonstrate elevated peripheral inflammatory markers in psychiatric populations, and PET neuroimaging studies have confirmed central microglial activation in vivo. The clinical significance is substantial: approximately one-third of patients with MDD show elevated inflammatory biomarkers, and this subgroup tends to respond poorly to first-line monoaminergic antidepressants but may respond to anti-inflammatory adjuncts. Understanding neuroinflammation is therefore not merely an academic exercise — it has direct implications for treatment selection, prognostication, and the development of novel therapeutics.

This article provides a detailed, research-informed review of the neuroinflammatory mechanisms implicated in psychiatric disorders, translating neuroscience into clinical relevance. It covers cytokine signaling, microglial biology, neuroimaging evidence, genetic and epigenetic factors, biomarker research, and the current state of anti-inflammatory treatment strategies. Throughout, animal model findings are clearly distinguished from human evidence, and the limitations of this rapidly evolving field are honestly addressed.

Cytokines: The Molecular Messengers

Cytokines are small signaling proteins produced by immune cells (and, within the CNS, by microglia and astrocytes) that regulate inflammation, cell survival, and intercellular communication. The cytokines most consistently implicated in psychiatric disorders are the pro-inflammatory cytokines: interleukin-6 (IL-6), interleukin-1β (IL-1β), tumor necrosis factor-alpha (TNF-α), and interferon-gamma (IFN-γ). Anti-inflammatory cytokines such as IL-10 and IL-4 serve regulatory roles and are often found to be relatively reduced in psychiatric populations.

Peripheral cytokines access the brain through several well-characterized pathways:

• Circumventricular organs: Regions lacking a complete BBB (e.g., area postrema, median eminence) allow direct cytokine entry.
• Active transport: Saturable transport systems carry IL-1β, IL-6, and TNF-α across the BBB.
• Vagal afferents: The vagus nerve detects peripheral inflammatory signals (particularly from the gut) and relays them to the nucleus tractus solitarius, which projects to limbic and cortical regions.
• BBB disruption: Chronic systemic inflammation can increase BBB permeability, allowing larger molecules and even immune cells to enter the CNS parenchyma.
• Endothelial activation: Brain endothelial cells express cytokine receptors and, when activated, produce secondary inflammatory mediators (prostaglandins, nitric oxide) within the CNS.

Microglia: The Brain's Resident Immune Cells

Microglia constitute approximately 5-12% of all cells in the brain and are its primary innate immune cells. Derived from yolk sac progenitors during embryogenesis (not from bone marrow like peripheral macrophages), microglia perform continuous surveillance of the neural environment. In their homeostatic state, they engage in synaptic pruning, clearance of cellular debris, secretion of neurotrophic factors such as BDNF, and modulation of synaptic plasticity — functions critical for normal brain development and maintenance.

When activated by pathogen-associated or damage-associated molecular patterns (PAMPs/DAMPs), peripheral cytokine signaling, or psychosocial stress, microglia undergo morphological and functional transformation. The traditional M1/M2 polarization framework — where M1 represented a pro-inflammatory, neurotoxic phenotype and M2 a neuroprotective, reparative phenotype — has been largely abandoned in favor of a spectrum model. Single-cell RNA sequencing studies have revealed that activated microglia exist along a continuum of phenotypic states, expressing overlapping pro- and anti-inflammatory markers simultaneously. Clinicians and researchers should be cautious of oversimplified M1/M2 framing that still appears in some clinical literature.

Chronically activated microglia release excess glutamate, reactive oxygen species (ROS), quinolinic acid (a neurotoxic tryptophan metabolite), and pro-inflammatory cytokines. These mediators can damage neurons and oligodendrocytes, impair synaptic plasticity, and disrupt white matter integrity. In animal models, sustained microglial activation via lipopolysaccharide (LPS) injection produces a behavioral phenotype remarkably similar to major depression — anhedonia, psychomotor retardation, social withdrawal, and cognitive impairment — collectively termed sickness behavior.

The Kynurenine Pathway: A Critical Interface

One of the most clinically significant downstream effects of neuroinflammation involves the kynurenine pathway of tryptophan metabolism. Pro-inflammatory cytokines, particularly IFN-γ, upregulate the enzyme indoleamine 2,3-dioxygenase (IDO), diverting tryptophan away from serotonin synthesis and toward kynurenine. Within the CNS, kynurenine is metabolized along two branches:

• Neurotoxic branch (in microglia): Kynurenine → 3-hydroxykynurenine → quinolinic acid (QUIN), an NMDA receptor agonist and source of oxidative stress.
• Neuroprotective branch (in astrocytes): Kynurenine → kynurenic acid (KYNA), an NMDA receptor antagonist with potential neuroprotective properties.

In psychiatric populations with elevated inflammation, the balance shifts toward the neurotoxic branch. Elevated QUIN/KYNA ratios have been documented in the cerebrospinal fluid of patients with MDD and suicidal ideation, and in postmortem brain tissue from suicide completers. This pathway provides a mechanistic link between inflammation, glutamate excitotoxicity, serotonin depletion, and psychiatric symptomatology.

Major Depressive Disorder (MDD)

The evidence linking neuroinflammation to MDD is the most extensive in psychiatry. A landmark meta-analysis by Dowlati et al. (2010) analyzing 24 studies found significantly elevated blood levels of IL-6 (effect size d = 0.62) and TNF-α (d = 0.40) in individuals with MDD compared to healthy controls. Subsequent larger meta-analyses, including Köhler-Forsberg et al. (2017) encompassing over 80 studies, confirmed elevations in IL-6, TNF-α, IL-1β, and C-reactive protein (CRP), with CRP elevations above 3 mg/L present in approximately 25-35% of patients with MDD.

Critically, not all depressed patients are inflamed. The neuroinflammatory subtype of depression is characterized by a specific symptom profile: pronounced fatigue, psychomotor retardation, appetite increase, hypersomnia, anhedonia, and cognitive sluggishness — overlapping substantially with the DSM-5-TR specifier for atypical features. Patients with melancholic features show a different and sometimes opposing inflammatory profile. This heterogeneity is clinically important for treatment selection.

The interferon-alpha (IFN-α) treatment model provides some of the strongest causal evidence in humans. Approximately 25-40% of hepatitis C patients treated with IFN-α develop clinically significant depressive episodes, with symptom onset correlating with cytokine elevation. Pretreatment with SSRIs partially prevents this, supporting a causal role for cytokine-induced neurotransmitter changes.

Schizophrenia Spectrum Disorders

Meta-analytic evidence from Miller et al. (2011), analyzing 40 studies, found elevations in IL-6, TNF-α, and IL-1β in both first-episode and chronic schizophrenia patients compared to controls. Importantly, some cytokine elevations (e.g., IL-6) appear to be state markers that normalize with antipsychotic treatment, while others (e.g., IL-12, IFN-γ) remain elevated regardless of treatment status, suggesting they may be trait markers.

Postmortem studies have found increased microglial density in the prefrontal cortex, hippocampus, and temporal cortex of individuals with schizophrenia, though these findings are confounded by medication history, cause of death, and postmortem interval. The complement system — particularly complement component 4 (C4) — has received attention following the landmark genome-wide association study by Sekar et al. (2016), which found that structurally distinct C4 alleles associated with schizophrenia risk lead to increased complement deposition on synapses and excessive synaptic pruning via microglia-mediated phagocytosis. This provides a plausible neurodevelopmental mechanism linking immune dysfunction to the cortical thinning and reduced synaptic density observed in schizophrenia.

Bipolar Disorder

Bipolar disorder shows a complex inflammatory trajectory. Meta-analytic data (Modabbernia et al., 2013) indicate that inflammatory markers are most elevated during manic and depressive episodes (IL-6 and TNF-α in particular) and partially normalize during euthymia, though they remain above healthy-control levels even in remission. This pattern is consistent with a neuroprogressive model in which repeated mood episodes drive cumulative neuroinflammatory damage, potentially explaining why episode frequency predicts cognitive decline and treatment resistance in longitudinal studies.

PTSD and Anxiety Disorders

Elevated CRP, IL-6, and TNF-α have been documented in PTSD in multiple meta-analyses. Early-life adversity — a robust risk factor for PTSD — is associated with persistent inflammatory priming, likely mediated through epigenetic modifications to glucocorticoid receptor genes (discussed below). Neuroimaging studies suggest that inflammation-related changes in the anterior cingulate cortex (ACC) and amygdala may contribute to the threat hypervigilance and emotional dysregulation characteristic of PTSD.

One of the most important advances in this field has been the development of neuroimaging tools that can detect neuroinflammation in living human brains. The primary modality is positron emission tomography (PET) using radioligands that bind to the translocator protein (TSPO), a mitochondrial protein upregulated in activated microglia.

PET Studies of Microglial Activation

First-generation TSPO ligands such as [¹¹C]PK11195 showed inconsistent results due to low signal-to-noise ratios. Second-generation ligands — including [¹¹C]PBR28, [¹⁸F]FEPPA, and [¹¹C]DPA-713 — have provided clearer evidence:

• MDD: Setiawan et al. (2015), using [¹⁸F]FEPPA PET, found significantly elevated TSPO binding (indicative of microglial activation) in the prefrontal cortex, anterior cingulate cortex, and insula of medication-free MDD patients compared to controls. Importantly, TSPO elevation was greatest in patients with longest untreated illness duration, supporting the neuroprogression hypothesis. A follow-up study by the same group found that TSPO binding was approximately 26% higher in the ACC of depressed patients than in controls.
• Schizophrenia: Results have been mixed. A meta-analysis by Plavén-Sigray et al. (2018) of 16 PET studies found that when accounting for the TSPO rs6971 polymorphism (which affects ligand binding affinity), evidence for elevated microglial activation in schizophrenia was not statistically significant. This contrasts with peripheral cytokine data and highlights the distinction between peripheral and central inflammation.
• Bipolar disorder: Limited PET data exist. Haarman et al. (2014) reported elevated [¹¹C]PK11195 binding in the right hippocampus of bipolar patients, but replication with second-generation ligands is needed.

fMRI and Inflammation

Functional MRI studies using experimental inflammatory challenges have been highly informative. Harrison et al. (2009) administered typhoid vaccination to healthy volunteers and demonstrated that the resulting mild inflammatory response produced altered neural activity in the subgenual anterior cingulate cortex (sgACC) — a region consistently implicated in depression — and that this activity change correlated with mood deterioration. Eisenberger et al. (2010) showed that endotoxin-induced inflammation increased amygdala reactivity to social threat cues and disrupted ventral striatal responses to reward — paralleling the threat sensitivity and anhedonia observed in inflammatory depression.

Structural MRI findings complement this picture. In a study of over 14,000 participants from the UK Biobank (Kitzbichler et al., 2021), higher CRP levels were associated with reduced gray matter volume in the medial prefrontal cortex, insula, and temporal regions, and with disrupted connectivity in the default mode network and salience network — circuits implicated in rumination and interoceptive processing, respectively.

Limitations of Neuroimaging Evidence

Important caveats must be acknowledged. TSPO PET has significant limitations: TSPO is expressed not only on microglia but also on astrocytes and endothelial cells, making it a nonspecific marker. The rs6971 genotype creates high-, mixed-, and low-affinity binders, complicating group comparisons. Sample sizes remain small (typically n = 15-30 per group). Novel radioligands targeting more specific microglial markers — such as colony-stimulating factor 1 receptor (CSF1R) — are in early clinical testing and may offer improved specificity.

Genetic Architecture

Genome-wide association studies (GWAS) have identified considerable overlap between immune-related genetic loci and psychiatric risk. The Psychiatric Genomics Consortium's cross-disorder analyses have shown that variants in the major histocompatibility complex (MHC) region on chromosome 6 are associated with schizophrenia, bipolar disorder, and MDD. As noted above, the C4 complement gene within the MHC region has emerged as a particularly compelling risk locus for schizophrenia, with larger C4A gene copy number and higher C4A expression associated with increased schizophrenia risk (Sekar et al., 2016).

Other immune-relevant genetic variants implicated in psychiatric risk include:

• IL-6 gene polymorphisms (rs1800795): The G allele is associated with higher IL-6 production and has been linked to depression risk in some candidate gene studies, though effect sizes are small and replication inconsistent.
• FKBP5 (rs1360780): This gene encodes a co-chaperone of the glucocorticoid receptor. The risk allele is associated with enhanced cortisol-driven inflammatory responses and increased PTSD vulnerability following childhood adversity.
• Toll-like receptor (TLR) pathway genes: Variants in TLR4 and downstream signaling molecules have been associated with depression risk in some studies, consistent with the role of innate immune activation in neuropsychiatric pathology.

Epigenetic Mechanisms

Epigenetic modifications — changes in gene expression without alterations in DNA sequence — represent a critical interface between environmental exposures and neuroimmune function. The most studied mechanism is DNA methylation, particularly at the NR3C1 gene (encoding the glucocorticoid receptor). Reduced NR3C1 methylation at specific CpG sites has been associated with childhood maltreatment in both human studies (McGowan et al., 2009, examining postmortem hippocampal tissue of suicide victims with childhood abuse histories) and animal models (Weaver et al., 2004, demonstrating that maternal care quality in rats altered hippocampal NR3C1 methylation).

The clinical consequence is glucocorticoid resistance — a state in which cortisol fails to adequately suppress immune activation through its normal anti-inflammatory negative feedback mechanism. This creates a feed-forward cycle: early adversity → epigenetic reprogramming of the HPA axis → chronic low-grade inflammation → increased psychiatric vulnerability across the lifespan. Longitudinal cohort studies, including the Dunedin Multidisciplinary Health and Development Study, have confirmed that childhood adversity predicts elevated CRP levels in adulthood, even after controlling for adult health behaviors.

Histone modifications and microRNA-mediated regulation of inflammatory gene expression are emerging areas of research, with preliminary evidence that specific microRNAs (e.g., miR-155, miR-146a) are dysregulated in the peripheral blood and, in some postmortem studies, the prefrontal cortex of individuals with MDD and schizophrenia.

The identification of reliable inflammatory biomarkers in psychiatry holds enormous promise for stratified treatment, early detection, and outcome prediction. However, the current state of the evidence requires honest appraisal.

Most Studied Biomarkers

• C-reactive protein (CRP): The most accessible and best-studied inflammatory biomarker in psychiatry. CRP is an acute-phase reactant produced by the liver in response to IL-6 stimulation. It is inexpensive, widely available, and has established reference ranges. In MDD, meta-analytic data show that approximately 25-35% of patients have CRP >3 mg/L (a level associated with elevated cardiovascular risk and, in psychiatric contexts, with specific symptom profiles and differential treatment response). In clinical trials, CRP has demonstrated moderate utility as a treatment-selection biomarker (discussed below).
• IL-6: The most consistently elevated cytokine across psychiatric diagnoses. It crosses the BBB, activates the HPA axis, and drives CRP production. However, serum IL-6 is influenced by BMI, smoking, sleep, time of day, and acute infection, limiting its specificity.
• TNF-α: Elevated in MDD and schizophrenia, but less consistently than IL-6. TNF-α has direct neurotoxic effects and can induce apoptosis in oligodendrocytes, potentially contributing to white matter abnormalities seen in mood disorders.
• Kynurenine metabolites: The QUIN/KYNA ratio in CSF or plasma shows promise as a marker of neuroinflammatory burden and glutamatergic excitotoxicity, but requires specialized assays not available in routine clinical settings.

Clinical Utility — Current Limitations

Despite promising research findings, no inflammatory biomarker is currently recommended for routine psychiatric diagnostic use in any major clinical guideline (APA, NICE, CANMAT). The reasons are important to understand:

• Low diagnostic specificity: Elevated CRP or IL-6 can reflect obesity, cardiovascular disease, autoimmune conditions, infection, smoking, or normal aging — not only psychiatric pathology.
• Within-disorder heterogeneity: Only a subset of patients with any given psychiatric diagnosis show elevated inflammatory markers. CRP does not distinguish MDD from bipolar depression, PTSD, or medical comorbidity-driven inflammation.
• Lack of validated thresholds: While CRP >3 mg/L is used in cardiovascular medicine and has been applied in psychiatric research, no psychiatric-specific cutoffs have been validated for treatment selection in large-scale prospective trials.
• Temporal instability: Within-person CRP and cytokine levels fluctuate with illness state, medication changes, and extraneous factors. Single-point measurements are less informative than longitudinal trajectories.

That said, some researchers and clinicians are beginning to use CRP as a pragmatic clinical tool to inform treatment decisions, particularly for patients with treatment-resistant depression. The rationale is based on secondary analyses of randomized trials (described below) and is most defensible when CRP is considered alongside clinical presentation rather than in isolation. This represents a reasonable, if not yet guideline-endorsed, application of available evidence.

CRP-Stratified Antidepressant Selection

Perhaps the most clinically actionable finding in this field comes from secondary analyses of the CO-MED trial (Combining Medications to Enhance Depression Outcomes). Jha et al. (2017) found that among patients with baseline CRP >1 mg/L, the combination of escitalopram + bupropion outperformed escitalopram monotherapy, while among patients with CRP <1 mg/L, the reverse pattern was observed. This suggests that dopaminergic/noradrenergic augmentation may be particularly beneficial in the context of inflammation-associated anhedonia and fatigue. The INSIGHT trial by Chamberlain et al. similarly showed that CRP moderated response to different antidepressant classes.

In the PANDA trial (Khandaker et al., 2022), CRP was prospectively tested as a stratification biomarker. Results were mixed: while CRP predicted overall prognosis (higher CRP = worse outcomes), it did not clearly differentiate between antidepressant classes in the primary analysis. This underscores that CRP-guided prescribing, while conceptually compelling, requires further prospective validation.

Anti-Cytokine Agents

Monoclonal antibodies targeting specific cytokines have been tested in psychiatric populations, primarily as adjuncts:

• Infliximab (anti-TNF-α): Raison et al. (2013) conducted a randomized controlled trial of infliximab versus placebo in treatment-resistant MDD. Overall, infliximab did not outperform placebo. However, in a pre-specified subgroup with CRP >5 mg/L, infliximab showed significant antidepressant effects (response rate ~62% vs. ~33% for placebo). This trial was pivotal in establishing the concept of an 'inflammatory biotype' of depression that may respond to targeted anti-inflammatory treatment.
• Tocilizumab (anti-IL-6 receptor): Case reports and small open-label studies in patients with comorbid inflammatory conditions and depression have shown mood improvement, but no large RCT in a primary psychiatric population has been completed.
• Sirukumab (anti-IL-6): Post-hoc analyses of trials in rheumatoid arthritis patients found that sirukumab improved depressive symptoms independently of improvements in pain or physical functioning, supporting a direct CNS effect of IL-6 blockade.

NSAIDs and COX-2 Inhibitors

A meta-analysis by Köhler-Forsberg et al. (2019) of 36 RCTs evaluating anti-inflammatory agents as antidepressant adjuncts found a significant overall effect (SMD = -0.55, 95% CI: -0.75 to -0.35). Celecoxib, a COX-2 selective inhibitor, has the strongest evidence base as an adjunct to SSRIs, with an NNT estimated at approximately 4-6 based on pooled trial data. However, cardiovascular risks associated with chronic COX-2 inhibition limit long-term use, and most trials were ≤8 weeks in duration.

Minocycline

Minocycline, a tetracycline antibiotic with documented anti-inflammatory and neuroprotective properties (including direct microglial inhibition), has been tested in both depression and schizophrenia. A meta-analysis of adjunctive minocycline in MDD (Rosenblat & McIntyre, 2018) found a significant antidepressant effect (SMD ≈ -0.78), though based on a small number of trials with limited sample sizes. In schizophrenia, results have been mixed — a Cochrane review found insufficient evidence to recommend routine adjunctive use, though some trials showed improvement in negative symptoms.

Omega-3 Fatty Acids

EPA (eicosapentaenoic acid) has anti-inflammatory properties through competition with arachidonic acid in prostaglandin synthesis pathways. A meta-analysis by Liao et al. (2019) found that EPA-predominant formulations (≥60% EPA) at doses of ≥1g/day had a significant antidepressant effect as adjuncts (effect size d ≈ 0.50), particularly in patients with elevated baseline inflammatory markers. DHA-predominant formulations did not show antidepressant effects, which is consistent with the differing anti-inflammatory profiles of these two fatty acids.

Psychotherapy and Lifestyle Interventions

Anti-inflammatory approaches are not limited to pharmacology. Regular aerobic exercise is one of the most potent anti-inflammatory interventions available, reducing CRP and IL-6 by approximately 20-30% in meta-analytic data. The antidepressant effect of exercise (SMD ≈ -0.70 to -1.0 in recent meta-analyses) may be partially mediated through these anti-inflammatory effects, in addition to BDNF upregulation and HPA axis normalization.

Cognitive behavioral therapy (CBT) and mindfulness-based interventions have also been shown to reduce inflammatory markers. A meta-analysis by Shields et al. (2020) found that psychotherapy was associated with modest but significant reductions in IL-6 and CRP, though effect sizes were smaller than those seen with exercise. Whether these anti-inflammatory effects of psychotherapy mediate clinical improvement (versus being an epiphenomenon of mood improvement) remains an open question.

The neuroinflammation field is rapidly growing but is also prone to oversimplification in popular and even clinical discourse. Correcting these misconceptions is important for accurate understanding and appropriate clinical application.

• "Depression is caused by inflammation." This is an overgeneralization. Inflammation is one pathophysiological pathway contributing to depression in a subset of patients — likely 25-35% based on CRP data. It is not the sole or even primary cause in most cases. The monoamine, glutamatergic, neurotrophic, and neural-circuit hypotheses remain relevant and complementary. Inflammation is best understood as a moderator and amplifier of other risk factors, not a singular causal agent.
• "Microglia are either M1 (bad) or M2 (good)." As discussed, this binary classification is outdated. Single-cell transcriptomic studies have identified dozens of distinct microglial activation states. Microglia perform essential homeostatic functions — including synaptic pruning and neurotrophic support — that are critical for brain health. The goal of treatment should be to normalize aberrant microglial activation, not to suppress microglial function broadly.
• "High CRP means you need anti-inflammatory treatment." CRP is influenced by numerous non-psychiatric factors (obesity, infection, autoimmune disease, smoking, sleep disruption). Elevated CRP in a psychiatric patient requires clinical interpretation. It may inform treatment decisions, but it does not constitute a definitive indication for anti-inflammatory therapy outside of research settings.
• "Neuroinflammation means the brain is 'infected.'" Neuroinflammation in psychiatric disorders is typically sterile — not caused by pathogens. It represents dysregulated activation of endogenous immune processes, often driven by chronic stress, peripheral metabolic dysfunction, or genetic vulnerability. The term 'inflammation' can mislead patients (and clinicians) into assuming an infectious etiology.
• "Anti-inflammatory medications should replace antidepressants." Current evidence supports anti-inflammatory agents as adjuncts to standard treatment, not replacements. No anti-inflammatory monotherapy has demonstrated sufficient efficacy to be recommended as first-line psychiatric treatment. Replacing evidence-based antidepressants with anti-inflammatory agents based on current evidence would be premature and potentially harmful.

Despite significant progress, the neuroinflammation field faces substantial challenges that must be honestly acknowledged.

Key Limitations

• Correlation vs. causation: Most human evidence is cross-sectional. While experimental models (IFN-α treatment, endotoxin challenge) provide causal evidence, these represent acute inflammatory insults. The causal role of chronic low-grade inflammation in naturally occurring psychiatric disorders is harder to establish definitively.
• Peripheral-central disconnect: Blood cytokine levels may not accurately reflect CNS inflammation. The PET evidence for central microglial activation in schizophrenia, for example, is weaker than peripheral cytokine data might suggest. Brain-specific biomarkers accessible through non-invasive means remain a critical unmet need.
• Confounding variables: Psychiatric patients often have higher rates of obesity, smoking, sedentary behavior, poor sleep, and medical comorbidities — all of which independently elevate inflammatory markers. Disentangling the contribution of 'psychiatric inflammation' from 'lifestyle-related inflammation' is methodologically challenging.
• Publication bias and small samples: Many early cytokine studies in psychiatry had small sample sizes (n < 50) and may be subject to publication bias favoring significant findings. Larger, pre-registered studies are needed.
• Medication effects: Antipsychotics (particularly clozapine and olanzapine), lithium, and some antidepressants have immunomodulatory properties. Distinguishing illness-related from treatment-related inflammatory changes is often impossible in cross-sectional studies of medicated patients.

Research Frontiers

• Immune-metabolic subtypes: Efforts to define biologically distinct subtypes of depression (e.g., 'immuno-metabolic depression' characterized by elevated CRP, increased BMI, fatigue, and hypersomnia) are underway using large-scale cohort data and machine-learning approaches. The Netherlands Study of Depression and Anxiety (NESDA) has been particularly productive in characterizing these subtypes.
• Gut-brain-immune axis: The gut microbiome profoundly influences systemic immune tone. Dysbiosis (altered gut microbial composition) has been associated with depression, anxiety, and schizophrenia, and may drive neuroinflammation through increased intestinal permeability ('leaky gut') and systemic endotoxemia. Probiotic and dietary interventions targeting the gut-immune interface are an active area of clinical research.
• Next-generation neuroimaging: Novel PET radioligands targeting CSF1R, P2X7 receptors, and other microglia-specific markers may overcome the limitations of TSPO imaging. Simultaneous PET-MRI protocols may allow correlation of microglial activation with circuit-level functional changes in the same scanning session.
• Precision immunopsychiatry: The ultimate goal is to move toward an approach in which inflammatory biomarkers guide treatment selection — matching patients with the 'inflammatory biotype' to anti-inflammatory augmentation strategies and sparing non-inflamed patients from unnecessary immunomodulation. The BIODEP study and the planned INCA trial represent steps in this direction.

Translating this evidence base into practical clinical recommendations requires balancing the promise of the neuroinflammation hypothesis against the limitations of current evidence. The following points represent reasonable clinical takeaways for practitioners:

• Consider inflammatory contributors in treatment-resistant presentations: When a patient with MDD fails two or more adequate antidepressant trials, assessing for elevated CRP (a simple, inexpensive blood test) is reasonable. CRP >3 mg/L in a patient with fatigue-predominant, anhedonic depression may suggest an inflammatory contribution worth addressing.
• Screen for medical and lifestyle inflammatory drivers: Obesity, metabolic syndrome, obstructive sleep apnea, sedentary behavior, smoking, and poor diet are all potent drivers of systemic inflammation. Addressing these factors is likely to benefit both psychiatric and physical health outcomes and should be considered part of comprehensive psychiatric treatment.
• Exercise as anti-inflammatory medicine: Recommending regular aerobic exercise (150+ minutes/week of moderate-intensity activity) is one of the most evidence-based anti-inflammatory interventions available and has well-established antidepressant effects. Framing exercise as a biological intervention affecting inflammatory pathways (not merely a 'lifestyle suggestion') may improve patient engagement.
• Be cautious with off-label anti-inflammatory prescribing: While the evidence for celecoxib adjunction is growing, the cardiovascular risks of chronic COX-2 inhibition, the gastrointestinal risks of NSAIDs, and the lack of large-scale prospective trials in psychiatric populations mean that routine anti-inflammatory prescribing for psychiatric indications is premature. Such decisions should be individualized and carefully weighed.
• Incorporate inflammatory thinking into case formulation: Understanding that childhood adversity, chronic psychosocial stress, and social isolation have inflammatory consequences provides a biologically grounded framework for understanding how psychosocial factors 'get under the skin' to produce psychiatric symptoms. This can enrich both biological and psychotherapeutic formulations.
• Stay informed but skeptical: The neuroinflammation field is advancing rapidly. New targets, new imaging tools, and new clinical trial data emerge regularly. Practitioners should follow this literature with interest but apply it critically, recognizing that many promising findings from early studies may not replicate in larger, more rigorous trials.

The recognition that neuroinflammation plays a significant role in psychiatric disorders represents a genuine paradigm expansion — not a replacement of existing models, but a critical addition to them. Cytokine signaling, microglial activation, kynurenine pathway dysregulation, and their interactions with monoamine, glutamatergic, and neurotrophic systems provide a mechanistic framework for understanding why chronic stress, medical illness, and social adversity confer psychiatric risk, and why certain patients fail to respond to conventional treatments.

The evidence base is strongest for major depressive disorder, where approximately one-quarter to one-third of patients show an inflammatory profile associated with specific symptom patterns, differential treatment response, and potentially modifiable with anti-inflammatory strategies. In schizophrenia, the complement-mediated synaptic pruning hypothesis offers a compelling neurodevelopmental mechanism, though the clinical translation is less advanced. Across diagnoses, the concept of an 'inflammatory biotype' that cuts across traditional diagnostic categories is gaining traction and may prove more useful than disorder-specific approaches.

At the same time, the field must guard against premature clinical application based on incomplete evidence. No inflammatory biomarker is currently validated for psychiatric diagnosis or treatment selection in major guidelines. Anti-inflammatory treatments are adjunctive, not primary. And the heterogeneity within and across psychiatric diagnoses means that neuroinflammation is relevant for some patients, some of the time — not a universal explanation for mental illness. The trajectory toward precision immunopsychiatry is promising, but the destination has not yet been reached.