Fifteen years ago at a London neurological hospital, women initially diagnosed with acute psychosis later revealed a different culprit: autoimmune encephalitis. As detailed in the New Scientist article, neuropsychiatrist Thomas Pollak encountered patients whose hallucinations, delusions, and agitation were the opening act of an immune system attacking the brain. This wasn't isolated. The piece correctly highlights anti-NMDAR encephalitis, but goes only so far. What it misses is the deeper mechanistic web and the historical pattern of psychiatry repeatedly overlooking biological triggers.

The core mechanism is now clear from peer-reviewed work. In 2007, Josep Dalmau and colleagues published a landmark paper in Annals of Neurology (not a preprint) describing 12 patients—primarily young women with ovarian teratomas—whose antibodies targeted NMDA receptors. Methodology involved serum and CSF antibody testing plus tumor screening; sample size was small but findings have since been replicated in thousands of cases worldwide. These antibodies disrupt glutamate signaling critical for memory, behavior, and perception, producing symptoms indistinguishable from schizophrenia. Limitations were acknowledged: initial focus on paraneoplastic cases left non-tumor autoimmune psychosis under-detected.

Pollak's subsequent research at King's College London, including a 2020 prospective study of first-episode psychosis patients (n≈200), found neuronal autoantibodies in roughly 5-10% of cases—far higher than previously assumed. This aligns with a 2020 Lancet Psychiatry consensus statement on autoimmune psychosis that synthesized data from multiple cohorts, recommending immunotherapy (steroids, IVIG, rituximab) when antibodies are present. Conventional antipsychotics fail in up to a third of schizophrenia patients; some of these non-responders improve dramatically once the immune attack is addressed.

The New Scientist coverage underplays how this extends beyond encephalitis. Large-scale epidemiological studies, such as a 2018 Swedish registry analysis of over 6,000 schizophrenia patients and matched controls, showed individuals with autoimmune diseases had 1.5-2 times higher schizophrenia risk. However, these observational studies cannot prove causation and are limited by confounding factors like shared genetics or medications. Similar patterns appear in depression: meta-analyses of cytokine profiles (sample sizes often >1,000) reveal elevated IL-6 and TNF-alpha in about 30% of treatment-resistant cases, suggesting chronic low-grade neuroinflammation from autoimmune dysregulation affects serotonin and neuroplasticity.

Connections the original source missed include parallels with PANS/PANDAS in children, where streptococcal infection triggers abrupt-onset OCD via molecular mimicry, and emerging post-COVID data linking spike-protein-triggered autoimmunity to new psychiatric symptoms. This isn't rare coincidence—it's an emerging pattern across disorders from OCD to dementia, where misfolded proteins may have an autoimmune component. The silos between psychiatry and immunology have left this under-covered; most psychiatrists do not order autoantibody panels without overt neurological signs like seizures, despite evidence that early immunotherapy transforms outcomes.

This research challenges the very foundation of psychiatry. If a substantial subset of 'functional' mental illness is actually organic brain inflammation, the field must shift from symptom checklists to biomarker-driven diagnosis. The promise is treatments targeting root biological mechanisms rather than modulating downstream neurotransmitters. Yet limitations remain: antibody testing is costly, not all cases show clear markers, and randomized controlled trials of immunotherapy in broad psychiatric populations are scarce. Still, the trajectory is unmistakable—an immunopsychiatry revolution that could reclassify and effectively treat what we currently call mental illness.