Abstract
Antipsychotic treatment resistance affects a third of people with schizophrenia and the underlying mechanism remains unclear. We used an fMRI emotion-yoked reward learning task, allied to prefrontal cortical glutamate levels, to explain the role of cognitive control in differentiating treatment-resistant from responsive patients. We investigated how reward learning is disrupted at the network level in 21 medicated treatment-responsive and 20 medicated treatment-resistant patients with schizophrenia compared with 24 healthy controls (HC). Dynamic Causal Modelling assessed how effective connectivity between regions in a cortico-striatal-limbic network is disrupted in each patient group compared to HC. Connectivity was also examined with respect to symptoms, salience and anterior cingulate (ACC) glutamate levels measured from the same region of the ACC. We found that ACC connectivity differentiated these patient groups, with responsive patients exhibiting increased top-down connectivity from ACC to sensory regions and reduced ACC drive to the striatum, while resistant patients showed altered connectivity within the ACC itself. In these resistant patients, the ACC drive to striatum was positively correlated with their symptom severity. ACC glutamate levels were found to correlate with ACC control over sensory regions in responsive patients but not in resistant patients. We suggest a central non-dopaminergic impairment that impacts cognitive control networks in treatment-resistant schizophrenia. This impairment was associated with disrupted reward learning and could be underpinned by aberrant glutamate function. These findings should form the focus of future treatment strategies (e.g. glutamatergic targets and giving clozapine earlier) in resistant patients.