Novel electroencephalographic biomarkers for the prediction of responders to an experimental glutamatergic agent in patients with schizophrenia

Comprehensive Summary

This study by Siekmeier and Coyle investigates whether pre-treatment EEG biomarkers can identify a subgroup of patients with schizophrenia who are likely to respond to a novel glutamatergic treatment. The authors analyzed pre-treatment resting-state and photically stimulated EEG data from two failed Phase III clinical trials of the mGluR2/3 agonist pomaglumetad methionil, applying gradient-boosted machine-learning models to a large set of EEG features. They identified distinct constellations of EEG markers, encompassing measures such as beta/gamma ratios, gamma activity, and power law exponents, that predicted treatment response with over 90% accuracy across cognitive, negative, and positive symptom domains. Importantly, these EEG patterns predicted response specifically to pomaglumetad and not to standard antipsychotic treatment, indicating drug-specific rather than generic treatment responsiveness. In the discussion, the authors emphasize that these biomarkers likely reflect a biologically meaningful glutamatergic endophenotype within schizophrenia and argue that such system-level EEG signatures may represent a final common pathway linking heterogeneous genetic risk to clinical response.

Outcomes and Implications

This research is important because it addresses a central limitation in schizophrenia therapeutics: the failure of mechanistically novel drugs in heterogeneous clinical trial populations despite clear benefit in a subset of patients. By providing a noninvasive, scalable method to identify responders, the work has direct relevance for improving clinical trial design and for advancing psychiatry. Clinically, EEG-based stratification could enable targeted use of glutamatergic agents for patients most likely to benefit, particularly for negative symptoms and cognitive deficits that are poorly treated by current dopamine-based antipsychotics. While the approach is not yet ready for clinical implementation, the authors suggest it could be incorporated into Phase II and Phase III trials in the short term, with broader clinical adoption resting on validation by independent cohorts.