Functional Coupling and Longitudinal Outcome Prediction in First-Episode Psychosis

Background Clinical outcomes following a first episode of psychosis (FEP) are highly heterogeneous between patients. The identification of prognostic biomarkers would greatly facilitate personalized treatments. Psychosis patients often display brain-wide disruptions of inter-regional functional coupling (FC), with some being linked to symptom severity and remission. FC may thus hold prognostic potential for people experiencing psychosis. Methods Ninety antipsychotic-naive FEP patients (51% female, 15-25 years) were randomized to receive either antipsychotic or placebo tablets for 6 months alongside psychosocial interventions. A subset of these patients also completed functional magnetic resonance imaging (55 with usable data at baseline and 37 at 3 months), which was used to evaluate whether baseline FC, or 3-month change in FC, could predict 6- and 12-month changes in symptoms and functioning, quantified using the Brief Psychiatric Rating Scale and the Social and Occupational Functioning Assessment Scale, respectively. We considered three different cross-validated prediction algorithms: (i) connectome-based predictive modelling; (ii) kernel ridge regression; and (iii) multilayer meta-matching. Results All algorithms showed poor performance in predicting patients' 6- and 12-month changes in symptoms and functioning (all r mean < 0.3), and no model achieved significance via permutation testing (all p > 0.05). Conclusions Our findings suggest that brain-wide measures of FC may not be suitable for predicting extended clinical outcomes over a 6- to 12-month period in FEP patients.