For clinicians thinking about immunopsychiatry clinical applications, the practical question is no longer whether the immune system matters. It is how treatment-resistant depression inflammation, C-reactive protein (CRP) in psychiatric disorders, and kynurenine pathway depression biology should change what we notice in the room. I want to start this one with a confession. For most of my career, when a client’s depression would not move, I reached for the same toolbox everyone else does. Switch the SSRI. Augment with an atypical. Try an SNRI. Add bupropion. Consider esketamine. Somewhere along the way I learned to ask about thyroid, about sleep apnea, about substance use, about the things we are trained to rule out. What I did not routinely ask was whether the depression in front of me was, at least in part, an inflammatory event. That omission is starting to look like a mistake.
Across the past several editions, this brief has kept circling the same idea: psychiatric illness is less a deficiency of one chemical and more a disorder of disrupted circuits, plasticity, and regulation. Immunopsychiatry is the next floor down in that same building. It asks a question that should have been obvious decades ago. If the brain is an organ, and inflammation affects every other organ, why would the brain be exempt? It is not. And the data have been accumulating quietly for years while most of us kept treating monoamines.
The association between depression and inflammation is one of the more replicated findings in biological psychiatry. A meta-analysis spanning clinical and community samples found that depression was positively associated with C-reactive protein, interleukin-1, and interleukin-6, with the relationships strongest in clinically depressed patients but still present in the general population (Howren et al., 2009). A larger pooled analysis of more than five thousand patients and five thousand controls confirmed elevated inflammatory markers in depression and, importantly, noted greater variability in those markers, which is a hint that we are looking at a subgroup rather than a uniform feature of the illness (Osimo et al., 2020).
That subgroup matters. Using the American Heart Association cutoff for elevated cardiovascular risk, a CRP above 3 mg/L, roughly a quarter to a half of depressed patients qualify as “inflamed,” depending on the population studied (Osimo et al., 2020). In treatment-resistant samples the proportion climbs. In the infliximab trial I will return to shortly, 45 percent of treatment-resistant patients had a CRP above 3 mg/L (Raison et al., 2013). So this is not a fringe phenomenon affecting a handful of unusual clients. It is a meaningful slice of the people sitting in our waiting rooms, and it appears to concentrate in exactly the population that frustrates us most: the ones who do not get better.
There is also a recognizable clinical face to inflammatory depression. Elevated CRP tracks with the somatic, atypical symptom cluster, fatigue, hypersomnia, leaden paralysis, increased appetite, weight gain, and anhedonia. That is worth sitting with. Anhedonia, the symptom I devoted an entire edition to as the reason our “successful” treatments still fail, shows up again here, riding alongside the inflammatory signature. The reward circuit and the immune system are not strangers.
It is one thing to observe a correlation. It is another to explain it. And the mechanism is what convinced me this is more than epidemiologic noise, because it connects directly to the glutamate and plasticity story this brief keeps returning to.
Start with tryptophan. It is the precursor for serotonin, and it is also the entry point for the kynurenine pathway. Under normal conditions, most tryptophan stays available for serotonin synthesis. But pro-inflammatory cytokines, interferon, interleukin-6, tumor necrosis factor, activate the enzyme indoleamine 2,3-dioxygenase, which diverts tryptophan away from serotonin and down the kynurenine pathway instead (Capuron & Miller, 2011; Dantzer et al., 2008). So step one is almost poetic in how it undercuts the old model: inflammation actively manufactures a monoamine deficit by stealing the substrate.
It gets more interesting downstream. In the inflamed brain, microglia, the resident immune cells, metabolize kynurenine into quinolinic acid, which is an agonist at the NMDA glutamate receptor and is frankly neurotoxic at sustained concentrations (Haroon et al., 2017). Meanwhile the protective branch of the pathway, kynurenic acid produced by astrocytes, gets relatively crowded out. The net result is excess glutamatergic drive at NMDA receptors and a shift toward neurotoxic metabolites. Anyone who has read the prior editions on ketamine, osavampator, and the AMPA-BDNF-mTOR axis should feel a jolt of recognition here. Inflammation is not operating in some separate compartment. It is pushing on the exact glutamatergic machinery we have been trying to modulate from the other direction. It is, in a sense, the same conversation arriving from the immune side of the house.
And there is a clue hiding in the interferon literature. Patients treated with interferon-alpha for hepatitis or melanoma develop depression at strikingly high rates, with elevated kynurenine-to-tryptophan ratios and increased cerebrospinal fluid kynurenine metabolites (Capuron & Miller, 2011). That is about as close to a controlled human experiment as we are likely to get: give someone an inflammatory cytokine, and a substantial fraction become clinically depressed. The arrow points from inflammation to depression, not merely the reverse.
This is where I have to do the thing I always do, which is to slow down before the enthusiasm outruns the evidence. Because the treatment story is genuinely mixed, and the way it is mixed is instructive.
The landmark study is Raison and colleagues’ randomized controlled trial of infliximab, a TNF-alpha antagonist, in treatment-resistant depression (Raison et al., 2013). On the headline outcome, it failed. Infliximab was no better than placebo across the whole sample. If you stopped reading at the abstract’s primary endpoint, you would conclude that anti-inflammatory treatment for depression does not work. But the prespecified analysis of baseline inflammation told a different story. Among patients with a baseline CRP above 5 mg/L, infliximab outperformed placebo, with response rates around 62 percent versus 33 percent. Among patients with low CRP, infliximab actually did slightly worse than placebo. The drug did not treat depression. It treated inflamed depression, and it may have done a disservice to the rest.
That single finding reframes the entire field. The lesson is not “anti-inflammatories are antidepressants.” The lesson is that giving an anti-inflammatory to a non-inflamed brain is the wrong tool, and pooling those patients together guarantees a null result. We have spent years running trials that mix inflamed and non-inflamed clients, then expressing surprise when the signal washes out.
The augmentation literature shows the same fingerprint. Meta-analyses of celecoxib added to standard antidepressants have reported impressively large effects, with pooled standardized mean differences in the range of negative 0.8, which would be remarkable if you took it at face value (Köhler-Forsberg et al., 2019). But the trials are small, heterogeneous, and several come from a narrow set of research groups. When a rigorous, adequately designed trial augmented vortioxetine with celecoxib and attempted to stratify by CRP, it found no benefit, and baseline CRP did not predict response (Baune et al., 2021). A 2026 systematic review applying GRADE methodology concluded that the certainty of evidence is low and that celecoxib cannot be recommended as routine practice for depression (Simon et al., 2026). One thoughtful critique even walked through how the celecoxib meta-analysis pooled endpoint scores with change scores, an apples-and-oranges error that can inflate an effect size into something that looks better than it is (Andrade, 2025).
I am not telling you this to be deflating. I am telling you because it is the honest shape of the data, and because the pattern of failure is itself the most useful finding in the field. The trials that ignore inflammation status fail or produce unstable effects. The analyses that account for it find a subgroup that responds. That is not a reason to abandon the hypothesis. It is a reason to test it correctly.
This is why the field has pivoted toward stratification, and why the more interesting current trials no longer ask whether anti-inflammatory treatment works for depression. They ask which depressed patients have an immune signature, and whether treating that signature changes outcomes. The MINDEP study of minocycline augmentation found benefit specifically in patients with CRP at or above 3 mg/L in post hoc analysis (Nettis et al., 2021). Larger trials are now prospectively splitting patients into high-CRP and low-CRP strata before randomizing, which is exactly the design the infliximab data demanded a decade ago.
There is also a quieter, more immediately practical thread running through this literature, and it does not require a single experimental drug. Inflammatory biomarkers may help us choose among the antidepressants we already have. In analyses of existing data, patients with higher CRP responded better to escitalopram than to nortriptyline, and adding bupropion to an SSRI improved outcomes specifically in patients with CRP at or above 1 mg/L (Raison, 2026, as discussed in AJMC). That is a striking idea for everyday practice. A cheap, ubiquitous blood test that most of our clients have had drawn a dozen times might help us pick the next medication rather than guessing. We are not there as standard of care, and I want to be careful not to oversell a CRP level as a prescription. But the direction is real, and it costs almost nothing to start paying attention to.
The immune story does not stop at mood. Anti-inflammatory strategies have been studied across the psychiatric map, and the pattern of differential response by condition is itself revealing. In schizophrenia, the agents with the most supportive signal are minocycline and aspirin rather than celecoxib, while in mood disorders celecoxib, pioglitazone, and statins carry more of the weight (Jeppesen et al., 2020). That divergence suggests we are not looking at one inflammatory mechanism but several, expressed differently across disorders. It maps, loosely, onto the cognitive and circuit themes from the schizophrenia editions: neuroinflammation and microglial activation are increasingly implicated in the disorganization and cognitive erosion that the older dopaminergic model never addressed.
I would put immunopsychiatry in the same category as the muscarinic and plasticity stories I have written about before. It is not a revolution that replaces everything we do. It is a layer we have been ignoring, and integrating it makes the rest of the picture more coherent rather than less.
Let me bring this down to the chair, because that is where this brief lives. I am not about to start infusing TNF antagonists in my office, and neither should you. The evidence does not support handing anti-inflammatories to every depressed client, and there is a real cost to treating non-inflamed brains with drugs that carry cardiovascular and gastrointestinal risk. Precision cuts both ways.
But a few things have changed in how I practice:
The reason this matters comes back to the throughline of everything I write here. We keep mistaking symptom suppression for restoration of function. Inflammatory depression may be one of the clearest examples of why so many of our treatments stall short of recovery: we are aiming at the monoamine when the problem is partly immune, partly metabolic, and partly glutamatergic, and a serotonin reuptake inhibitor was never built to touch any of that. The immune system has been sitting in the diagnostic chair the whole time. We just were not looking at it.
The molecules are not the hard part. Identifying the right client is. That is the work.
Andrade, C. (2025). Anti-inflammatory treatments for depression: Perspectives on how to read a meta-analysis critically. Journal of Clinical Psychiatry, 86(4), 25f15901.
Baune, B. T., Sampson, E., Louise, J., Hori, H., Schubert, K. O., Clark, S. R., Mills, N., & Fourrier, C. (2021). No evidence for clinical efficacy of adjunctive celecoxib with vortioxetine in the treatment of depression: A 6-week double-blind placebo controlled randomized trial. European Neuropsychopharmacology, 53, 34–46.
Capuron, L., & Miller, A. H. (2011). Immune system to brain signaling: Neuropsychopharmacological implications. Pharmacology & Therapeutics, 130(2), 226–238.
Dantzer, R., O’Connor, J. C., Freund, G. G., Johnson, R. W., & Kelley, K. W. (2008). From inflammation to sickness and depression: When the immune system subjugates the brain. Nature Reviews Neuroscience, 9(1), 46–56.
Haroon, E., Miller, A. H., & Sanacora, G. (2017). Inflammation, glutamate, and glia: A trio of trouble in mood disorders. Neuropsychopharmacology, 42(1), 193–215.
Howren, M. B., Lamkin, D. M., & Suls, J. (2009). Associations of depression with C-reactive protein, IL-1, and IL-6: A meta-analysis. Psychosomatic Medicine, 71(2), 171–186.
Jeppesen, R., Christensen, R. H. B., Pedersen, E. M. J., Nordentoft, M., Hjorthøj, C., Köhler-Forsberg, O., & Benros, M. E. (2020). Efficacy and safety of anti-inflammatory agents in treatment of psychotic disorders – A systematic review and meta-analysis. Brain, Behavior, and Immunity, 90, 364–380.
Köhler-Forsberg, O., Lydholm, C. N., Hjorthøj, C., Nordentoft, M., Mors, O., & Benros, M. E. (2019). Efficacy of anti-inflammatory treatment on major depressive disorder or depressive symptoms: Meta-analysis of clinical trials. Acta Psychiatrica Scandinavica, 139(5), 404–419.
Nettis, M. A., Lombardo, G., Hastings, C., Zajkowska, Z., Mariani, N., Nikkheslat, N., Worrell, C., Enache, D., McLaughlin, A., Kose, M., Sforzini, L., Bogdanova, A., Cleare, A., Young, A. H., Pariante, C. M., & Mondelli, V. (2021). Augmentation therapy with minocycline in treatment-resistant depression patients with low-grade peripheral inflammation: Results from a double-blind randomised clinical trial. Neuropsychopharmacology, 46(5), 939–948.
Osimo, E. F., Pillinger, T., Rodriguez, I. M., Khandaker, G. M., Pariante, C. M., & Howes, O. D. (2020). 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, 901–909.
Raison, C. L., Rutherford, R. E., Woolwine, B. J., Shuo, C., Schettler, P., Drake, D. F., Haroon, E., & Miller, A. H. (2013). A randomized controlled trial of the tumor necrosis factor antagonist infliximab for treatment-resistant depression: The role of baseline inflammatory biomarkers. JAMA Psychiatry, 70(1), 31–41.
Simon, M. S., Schön, C., & Baune, B. T. (2026). Celecoxib adjunct to antidepressants for major depressive disorder: Systematic review and meta-analysis. Journal of Affective Disorders Reports.
Disclosure: I am a paid speaker for Bristol Myers Squibb (Cobenfy), AbbVie (Vraylar), and Axsome (Auvelity). The opinions and views expressed here are my own and do not necessarily reflect those of these organizations.
Schedule an appointment today to get diagnosed, receive a prescription, and continue your journey towards mental peace.
