Analysis of efficacy and side effects in CATIE demonstrates drug response subgroups and potential for personalized medicine.

OBJECTIVE: This study aims to improve understanding of antipsychotic non/response and assess the potential for personalized schizophrenia treatment.
METHODS: We used data from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE). Efficacy measures included the Positive and Negative Syndrome Scale (PANSS) and neurocognitive functioning. Side effect measures included weight, lipids, glucose, heart rate and QT prolongation. Latent class analysis was conducted for each of the five drugs on the individual treatment effects to study whether there were subgroups of drug responders. The posterior probabilities of belonging to a particular response group were correlated across drugs to examine if patients not responding to one drug are likely to respond to a different drug and whether response to one drug may help to predict response to another drug.
RESULTS: We identified four qualitatively distinct response groups: Optimal Responders, Average Responders, Global Responders and Non-Responders. Different patterns of correlations with demographics and clinical variables across classes provided further support for the validity of these groups. The low correlations between posterior probabilities of the same response groups across drugs implied that patients generally belonged to different response groups for different drugs.
CONCLUSIONS: Our results demonstrate the existence of subgroups of patients characterized by distinct patterns of drug response. Further, findings suggest that patients who experience a poor response to one drug may be an optimal responder to another antipsychotic. Taken together these findings demonstrate the potential to personalize schizophrenia treatment and highlight the importance of identifying better predictors of drug response.