First-Episode Psychosis Reduced CSF C4A-C1Q Immune Coupling

TL;DR: The complement story in schizophrenia has been narrated as “too much pruning.” A new CSF-and-plasma study sharpens it. In healthy brains, C4A and C1Q move together. In first-episode psychosis, they decouple — and the near-twin protein C4B does not show the same disease pattern. The interesting signal is not inflammation rising as a whole; it is C4A behaving differently from its biological look-alike.

Source: Translational Psychiatry (2026) | Arjmand et al.

For years, the complement model in schizophrenia has been summarized in three words: too much pruning. The genetics pointed at the major histocompatibility complex region — and specifically at C4A copy number — and the working interpretation was that excess synaptic tagging during late adolescence helped trigger psychosis. Compelling story. Hard to translate. Genetic association by itself cannot tell you what C4A is actually doing inside a living human brain.

This paper takes a step toward that translation by measuring complement and inflammation proteins in CSF and plasma from first-episode psychosis patients. The result is more specific, not less.

Why C4A Keeps Showing Up When C4B Usually Does Not

The complement hypothesis got its initial traction from a genetic mapping result and the observation that some complement proteins help tag synapses for pruning during development. That made C4A a plausible mechanistic candidate — but the field has been stuck with the harder question ever since: how is C4A actually behaving in living patients, and why does the closely related protein C4B not follow the same pattern?

The two proteins are near-twins biochemically. If “complement activation” were the relevant disease driver, you would expect C4A and C4B to do similar things in patient samples. They do not, and that asymmetry has been underexplored compared to the much louder C4A genetics story.

This study moves the question from DNA to fluid biology. By measuring complement proteins alongside a broad inflammatory panel in both CSF and plasma — and by including a substantial antipsychotic-naive subgroup — the authors ask whether C4A occupies a distinct immune network in first-episode psychosis. Their answer is yes.

The Decoupling That Mattered

113 first-episode psychosis patients and 90 healthy controls contributed CSF, plasma, and inflammatory profiling. The first striking result was relational, not absolute.

In controls, C4A and C1Q moved together — exactly what you would expect if the classical complement pathway were operating in coordinated fashion. C4A and C1Q correlated positively at z = 0.41, p < 0.001. In first-episode psychosis, that coordination disappeared. The same comparison dropped to z = 0.09, p = 0.40.

This is a more interesting kind of abnormality than a one-marker elevation. It implies that the network architecture around C4A may be altered in early psychosis, not just its level. Disease was not adding inflammation on top of normal coordination — it was breaking the coordination itself.

C4A Drifted Across the Inflammatory Network. C4B Did Not.

The paper’s central claim becomes clearer when the analysis expands beyond C1Q to a 48-protein inflammatory panel. In healthy controls, C4A tended to show predominantly negative associations across the network in CSF, while C4B and C1Q skewed positive. C4A already occupied a different niche under healthy conditions.

In psychosis, the whole C4A pattern shifted in a directionally robust way (z = 3.81, p < 0.0001). C4B, tested with the same approach, showed a non-significant negative shift. One protein moved as the disease moved. The biologically nearest other protein did not.

That split is the point. If you take the schizophrenia complement story seriously, this is exactly the kind of evidence you want — not “inflammation happened” but one candidate molecule behaving in a way that fits prior genetics and diverges from its closest look-alike.

Plasma Echoes the Brain, But Imperfectly

The same OLINK inflammatory panel was measured in plasma, which matters because blood is far easier to collect than CSF and any biomarker translation has to cross that bridge. The broad directional shift for C4A survived in plasma. The underlying protein-by-protein architecture did not — it changed substantially between compartments.

The warning embedded in that result is important. A blood marker can echo a central signal without reproducing the biology that makes the central signal interesting. The compartments here were related enough to be encouraging, but different enough to show that plasma cannot be treated as a plug-and-play substitute for CSF in this domain. Future translational work will probably need paired blood-CSF studies, not blood alone.

The authors also tested whether a combined complement-plus-inflammatory profile tracked clinical severity. One CSF principal component hinted at a relationship with negative symptoms, but it did not cleanly survive multiple-comparison correction. The paper’s strongest contribution remains mechanistic, not yet clinical.

What This Reframes for Synaptic-Pruning Models

The most provocative reading is that C4A-driven synaptic tagging may become uncoupled from its usual complement context in early psychosis. The paper does not prove that directly. It fits a broader model in which C4A contributes to schizophrenia risk through a brain-specific immune specialization rather than through a general inflammatory storm.

That has consequences for how the field should talk about its own data. “Complement activation” is the kind of phrase that smooths over exactly the asymmetry this paper exposes. Disease-relevant biology may live in which complement relationships stay intact, which ones break, and in which compartment those breaks appear — not in the broad direction of an inflammatory marker panel.

What This Result Cannot Yet Settle

The study is cross-sectional. The cohorts are clinically heterogeneous. First-episode psychosis is not identical to schizophrenia, and not every first-episode patient ultimately receives a schizophrenia-spectrum diagnosis. Medication exposure complicates inflammatory measurements, even though the 36 antipsychotic-naive patients help separate disease biology from treatment biology.

Early psychosis is also biologically messy in ways that go well beyond complement. Stress, sleep disruption, substance exposure, immune state, and treatment timing all shape inflammatory measurements at exactly this clinical stage. That is why the selective C4A pattern is interesting but not yet diagnostic.

The right next step is not “the same measurement, larger sample.” It is longitudinal — follow first-episode patients through diagnostic clarification, symptom course, treatment exposure, and cognitive outcome, and ask whether the C4A–C1Q decoupling stays broken, normalizes, or marks a subgroup with a distinct trajectory. That kind of follow-up would also clarify whether the CSF signal and a more accessible plasma signal can ever be aligned closely enough to support clinical translation.

For now, the headline finding is unusually disciplined for the schizophrenia immune literature: the disease-relevant signal is not the whole complement system rising. It is one specific protein behaving differently from its near-twin, in the brain compartment, at the disease’s earliest visible stage. That kind of specificity is usually what the field looks like when it is getting closer to a real mechanism, not further from one.