SCHIZOPHRENIA POLYGENIC RISK SCORE PREDICTS ANTIPSYCHOTIC TREATMENT RESPONSE IN PATIENTS WITH FIRST EPISODE PSYCHOSIS

The genetic basis of antipsychotic drug efficacy is likely polygenic in nature. Genetic risks of schizophrenia may also be related to antipsychotic drug response. The Psychiatric Genomics Consortium (PGC) Genome-Wide Association Study (GWAS) provided evidence of association with schizophrenia risk for many Single Nucleotide Polymorphism (SNP) across the genome. We examined whether Polygenic Risk Scores (PRS) based on the PGC GWAS are predictive of antipsychotic efficacy in four cohorts of patients with first episode psychosis (total n=510): 1) Zucker Hillside Hospital First Episode schizophrenia trial (ZHH-FE), 2) European First Episode Schizophrenia Trial (EUFEST), 3) Spanish First Episode Psychosis study (SFEP), and 4) the clinical trial as part of the Center for Intervention Development and Applied Research at ZHH (CIDAR). The discovery cohort was the ZHH-FE with 77 patients (mixed ethnicity) randomized to risperidone or olanzapine. Three replication cohorts were: 1) EUFEST with 141 patients (all Caucasian) randomized to five antipsychotics; 2) SFEP with 192 patients (all Caucasian) on various antipsychotics; and 3) CIDAR with 100 patients (mixed ethnicity) randomized to risperidone or aripiprazole. Genotyping was performed using the Illumina Omni-1Quad (EUFEST and ZHH-FE) or Illumina Infinium HumanOmniExpressExome platform (CIDAR and SFEP). SNP imputation was conducted with IMPUTE2 against the full 1000 Genomes v3 reference panel. PRS was computed based on the results of the PGC GWAS using PRSice software for the discovery cohort with thresholds at PT<5E-8, 0.001, 0.01, 0.05, 0.10, 0.20, 0.50. Based on the findings from the discovery cohort, PRS was computed for the three replication cohorts using a threshold of PT<0.01. Symptom measure was the total score of Brief Psychiatric Rating Scale (BPRS) for ZHH-FE, SFEP, and CIDAR, or Positive and Negative Symptoms Scale (PANSS) for EUFEST. Hierarchical linear regression was performed on the 3-month symptom score with the PRS as the predictor while controlling for age, sex, and baseline symptom score. Genomic principal component scores were also covaried to control for population stratification for ZHH-FE and CIDAR. In the ZHH-FE cohort, higher PRS at the thresholds of PT<0.01, 0.05, 0.10, 0.20, and 0.50 significantly predicted higher symptom scores at 3-month follow-up, explaining 6–8% of the variance (all p's<0.05). PT<0.01 gave the strongest result in the discovery sample, and was used to replicate the findings in the other three cohorts. Higher PRS significantly predicted worse symptoms in EUFEST and SFEP cohorts, explaining 3.5% and 3.7% of variance, (p's<0.01), but not in the CIDAR sample. Combining the four cohorts in a meta-analysis, PRS was significantly predictive of 3-month symptom scores (pooled partial r=0.18, p=0.002). Higher PRS was associated with higher symptom scores at 3-month follow-up, suggestive of less improvement in treatment. The overall results remained significant when only European ancestry individuals were included in the analysis. These findings suggest that polygenic risk scores for schizophrenia may also be related to antipsychotic drug response. Patients with higher polygenic risk scores tended to have less improvement with antipsychotic drug treatment. Further analysis is needed to elucidate a more refined genomic profile for antipsychotic drug response.