Using blood methylomes to predict response to amisulpride in the first-episode psychosis in the OPTiMiSE cohort

Treatment of schizophrenia relies heavily on the use of antipsychotic drugs. Their efficacy is at present determined by lengthy trial-and-error approach, calling for more efficient strategies based on personalized medicine. Here, we present a prospective study of 116 first-episode-psychosis (FEP) patients from the OPTiMiSE cohort, aiming to identify blood epigenomic biomarkers predicting response to amisulpride and to shed light on involved mechanisms by linking the observed methylation patterns to genetic variation and gene expression. The analysis of 210 paired (baseline and follow-up) blood methylomes revealed 67 regions stably differentially methylated between good and bad responders and 197 regions with response-specific dynamics. The former were primarily enriched in functions related to neurotransmission and synapse assembly, the latter in immunity and inflammation. Baseline methylation values of three of these candidate regions, situated within HOXA, HTR2A and PRR5 genes, were selected as good predictors (10x cross-validated Matthews correlation coefficient = 0.81) of amisulpride response in our cohort. Screening for associations between the methylation of the selected regions and the genetic variants (SNPs) in a 1MBp surroundings revealed a high degree of genetic control for HTR2A, but not for HOXA or PRR5 regions. Whereas we detected multiple correlations between methylation and gene expression, few were temporally stable, such as the correlation between HOXA5 and SKAP2 expression, a gene affecting susceptibility to schizophrenia. Our findings demonstrate the strengths of prospective design in response-biomarker research and suggest that epigenetic variation associated with antipsychotic response is shaped by both the environmental and genetic factors.