Using magnetoencephalography to predict outcomes of glutamatergic psychosis treatment
Lead Research Organisation:
University College London
Abstract
Vision
This study will use magnetoencephalography (MEG) and computational modelling of MEG data to predict the outcomes – i.e., reduction in psychotic symptoms and/or improvement in cognition – of taking a glutamatergic treatment in individuals with early psychosis. My group's work in the first half of this project has shown that we can use modelling to infer reduced pyramidal (excitatory) neuron excitability from M/EEG data in people with psychosis, including in prodromal individuals prior to their first episode of psychosis. Ongoing work in my group and my collaborators in both humans and in genetic mouse models (with reduced pyramidal neuron function) is also revealing which M/EEG paradigms are most sensitive to revealing pyramidal neuron dysfunction, and also the reliability of these measures. These two findings – that pyramidal excitability is reduced even in very early psychosis, and which M/EEG paradigms are most sensitive to detect this reduced excitability – form the basis of the current project.
Objectives
We will recruit around 80 individuals with early psychosis for an experimental medicine study. They will be asked to take D-serine (n=50), a glutamatergic (NMDA) receptor modulator that is freely available over the counter as a health food supplement, or placebo (n=30) for 12 weeks. Participants and researchers will be double blind to medication status. We will obtain MEG scans and symptom and cognitive measures at baseline and after 12 weeks. In the MEG scan, lasting around one hour, participants will undergo auditory oddball, 40 Hz auditory steady state, and 'resting' paradigms. We will analyse these paradigms using dynamic causal modelling and estimate function of excitatory and inhibitory neurons in each participant.
Our primary objective is: to assess whether inferred excitatory neuron function at baseline predicts change in symptoms or cognition after 12 weeks of D-serine?
Our secondary objectives are: to assess whether
i) MEG data features related (in previous work) to excitatory neuron function at baseline predict change in symptoms or cognition after 12 weeks of D-serine? I.e., is modelling necessary for treatment outcome prediction?
ii) Can we demonstrate changes in inferred excitatory neuron function from baseline to the end of the study in those taking D-serine? Does this change correlate with symptom or cognition improvement?
Areas of focus
We are focusing on people with early psychosis because it is widely hypothesised that this group will be most responsive to glutmatergic treatments. We are focusing on both psychotic symptoms and cognitive symptoms because in previous work I have shown that both relate to the balance of excitatory and inhibitory function in cortex (Adams et al., 2022, Biol Psych). Furthermore, there is no established treatment for cognitive symptoms in psychosis. We are focusing on glutamatergic treatments because of the wealth of evidence implicating cortical NMDA receptors in psychosis risk.
Why it's important
Pharmaceutical companies have invested around $2.5B in glutamatergic treatments for psychosis but these have failed Phase 3 trials. Discovering a way to stratify patients by predicting their response to these treatments could unlock this enormous investment, for patient benefit.
Why it will succeed
My group have modelled these M/EEG paradigms in multiple data sets and are demonstrating validity and reliability of these measures in ongoing work. We are ideally placed to estimate excitatory neuron function as accurately as possible.
This study will use magnetoencephalography (MEG) and computational modelling of MEG data to predict the outcomes – i.e., reduction in psychotic symptoms and/or improvement in cognition – of taking a glutamatergic treatment in individuals with early psychosis. My group's work in the first half of this project has shown that we can use modelling to infer reduced pyramidal (excitatory) neuron excitability from M/EEG data in people with psychosis, including in prodromal individuals prior to their first episode of psychosis. Ongoing work in my group and my collaborators in both humans and in genetic mouse models (with reduced pyramidal neuron function) is also revealing which M/EEG paradigms are most sensitive to revealing pyramidal neuron dysfunction, and also the reliability of these measures. These two findings – that pyramidal excitability is reduced even in very early psychosis, and which M/EEG paradigms are most sensitive to detect this reduced excitability – form the basis of the current project.
Objectives
We will recruit around 80 individuals with early psychosis for an experimental medicine study. They will be asked to take D-serine (n=50), a glutamatergic (NMDA) receptor modulator that is freely available over the counter as a health food supplement, or placebo (n=30) for 12 weeks. Participants and researchers will be double blind to medication status. We will obtain MEG scans and symptom and cognitive measures at baseline and after 12 weeks. In the MEG scan, lasting around one hour, participants will undergo auditory oddball, 40 Hz auditory steady state, and 'resting' paradigms. We will analyse these paradigms using dynamic causal modelling and estimate function of excitatory and inhibitory neurons in each participant.
Our primary objective is: to assess whether inferred excitatory neuron function at baseline predicts change in symptoms or cognition after 12 weeks of D-serine?
Our secondary objectives are: to assess whether
i) MEG data features related (in previous work) to excitatory neuron function at baseline predict change in symptoms or cognition after 12 weeks of D-serine? I.e., is modelling necessary for treatment outcome prediction?
ii) Can we demonstrate changes in inferred excitatory neuron function from baseline to the end of the study in those taking D-serine? Does this change correlate with symptom or cognition improvement?
Areas of focus
We are focusing on people with early psychosis because it is widely hypothesised that this group will be most responsive to glutmatergic treatments. We are focusing on both psychotic symptoms and cognitive symptoms because in previous work I have shown that both relate to the balance of excitatory and inhibitory function in cortex (Adams et al., 2022, Biol Psych). Furthermore, there is no established treatment for cognitive symptoms in psychosis. We are focusing on glutamatergic treatments because of the wealth of evidence implicating cortical NMDA receptors in psychosis risk.
Why it's important
Pharmaceutical companies have invested around $2.5B in glutamatergic treatments for psychosis but these have failed Phase 3 trials. Discovering a way to stratify patients by predicting their response to these treatments could unlock this enormous investment, for patient benefit.
Why it will succeed
My group have modelled these M/EEG paradigms in multiple data sets and are demonstrating validity and reliability of these measures in ongoing work. We are ideally placed to estimate excitatory neuron function as accurately as possible.