- Design
- cross-sectional plasma proteomic profiling with pathway enrichment analysis, false discovery rate corrected
- Population
- 1,881 Dutch adults: 469 current major depressive disorder, 918 remitted, 494 controls
- Primary outcome
- protein associations with diagnosis, symptom severity, and melancholic and atypical dimensions
- Effect
- 7 proteins for diagnosis; 18 for overall severity; 57 for atypical symptoms; none for melancholic symptoms
Plasma from 1,881 participants in the Netherlands Study of Depression and Anxiety — 469 with current major depressive disorder, 918 in remission and 494 controls — was profiled across 7,289 analytes on an aptamer platform, then tested against diagnostic status, symptom severity and the melancholic and atypical symptom dimensions.
The categorical comparison was thin: eight aptamers representing seven proteins separated current depression from controls after correction for multiple testing. Symptom dimensions did far better. Overall severity was associated with 18 proteins, and atypical symptoms — the appetite, weight and hypersomnia cluster — with 57. Melancholic symptoms produced no individually significant protein at all, though pathway-level analysis still flagged mitochondrial stress response, protein translation and complement signalling.
The pathways that recurred are not the ones a monoamine account of depression would predict: extracellular matrix organisation, collagen formation, complement cascade signalling, platelet activation and haemostasis for overall severity, and insulin-like growth factor transport for atypical symptoms. That pattern reads as inflammatory and metabolic rather than neurotransmitter-based, and it aligns with the long-standing clinical observation that atypical depression tracks with cardiometabolic disease.
Nothing here is orderable. There is no assay, no threshold and no evidence that any of these proteins predicts treatment response. The value for now is conceptual: it supports asking which symptoms a patient has rather than treating the diagnostic label as the biological unit.
- Record the symptom profile — appetite, weight and sleep direction — rather than only the diagnosis, since that is where the biology separated
- For a patient with atypical features, assess cardiometabolic risk actively; the implicated pathways are metabolic and inflammatory
- Do not order proteomic or inflammatory panels for depression — none of this is a clinical test
- Treat the absence of a melancholic protein signal as a limit of this dataset, not as evidence melancholia is not biological
- Expect symptom-dimension language in future trial stratification and read enrolment criteria accordingly
Why it matters
If the biology sits with symptom clusters rather than the diagnosis, the diagnostic category is the wrong unit for biomarker research.
Don't overread it
This is cross-sectional and associative — it cannot show that any of these proteins causes depression or predicts response to treatment.
The statistics, in plain English
With 7,289 analytes tested, most of the apparent hits in a raw analysis would be chance, which is why false discovery rate correction matters: only findings surviving it are reported. That seven proteins survived for the diagnosis while 57 survived for atypical symptoms, in the same sample, is a real difference in signal strength rather than a difference in sample size. The melancholic result is a null at the individual protein level, and pathway enrichment on non-significant proteins is exploratory — it generates hypotheses rather than testing them.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for psychiatry, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free