- Design
- Individual participant data mega-analysis plus meta-analysis
- Population
- 1,189 participants from 18 studies (476 non-responders, 427 responders, 286 controls)
- Primary outcome
- Neurometabolite differences by antipsychotic response
- Effect
- Medial frontal Glx Glass Δ 0.29; myo-inositol 0.35 (0.64 in treatment-resistant)
This mega-analysis pooled individual magnetic resonance spectroscopy data from 18 studies: 476 antipsychotic non-responders, 427 responders and 286 healthy controls, published in JAMA Psychiatry in September.
Compared with responders, non-responders had higher medial frontal glutamate (Glass Δ 0.21), glutamate plus glutamine (0.29), choline (0.22) and myo-inositol (0.35). The glutamate plus glutamine difference also appeared prospectively in first-episode psychosis (Δ 0.41), and myo-inositol was highest in those meeting treatment-resistance criteria (Δ 0.64). A wider meta-analysis of 23 studies (1,844 participants) confirmed raised choline and myo-inositol.
The effect sizes are small to moderate and overlap heavily between groups, so spectroscopy cannot yet tell you which patient will respond. The value is mechanistic: it supports the idea that some treatment-resistant psychosis is not a dopamine problem, and backs continued trials of glutamate-acting and anti-inflammatory approaches. Clinically, it strengthens the case for identifying non-response early and moving to clozapine on time.
- Do not order spectroscopy to predict antipsychotic response; this is a research signal, not a test.
- Assess response formally after an adequate dose and duration rather than switching repeatedly between dopamine antagonists.
- Offer clozapine once two adequate antipsychotic trials have failed; non-response may reflect a different biology.
- Check adherence and, where available, plasma levels before labelling a patient a non-responder.
Why it matters
It adds evidence that treatment-resistant psychosis may be biologically distinct, not just under-treated.
Don't overread it
The group differences are small and overlap widely; spectroscopy cannot classify individual patients.
The statistics, in plain English
Glass Δ expresses the difference between groups in units of the responder group's spread. A value of 0.2 to 0.35 is small: most non-responders and responders would have similar readings. The larger 0.64 in treatment-resistant patients is moderate, and the subgroup is smaller, so it is less certain.
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