POLYGENIC RISK SCORE ANALYSIS OF ANTIDE-PRESSANT TREATMENT RESPONSE: RESULTS FROM THE CAN-BIND-1 COHORT

The genetic architecture of antidepressant response remains poorly understood. Here, we used polygenic risk scores to investigate the genetic overlap between psychiatric and non-psychiatric traits and response to antidepressant treatment. The Canadian Biomarker Integration Network for Depression (CAN-BIND-1) sample of 211 major depressive disorder (MDD) patients was used as our target dataset. The CAN-BIND-1 included patients that were treated with escitalopram for 16 weeks. Response and remission status were assessed on Weeks 8 and 16. Those who were non-responders in Week 8 were also given augmentation therapy with aripiprazole (antipsychotic). We constructed polygenic risk scores using GWAS summary statistics from psychiatric and non-psychiatric disorders (e.g., schizophrenia, bipolar, post-traumatic stress syndrome or PTSD, neuroticism) obtained through the Psychiatric Genomics Consortium (PGC) using the clumping and thresholding method (with PRSice v2). Next, we assessed the associations between PRSs with antidepressant remission and symptom improvement at Weeks 8 and 16 in our CAN-BIND-1 sample co-varying for age, sex, drug-arm, and the first three genomic principal components to adjust for population stratification. Bonferroni correction for multiple testing and 10000 permutation tests were also applied to mitigate overfitting. PRS for PTSD negatively correlated with antidepressant symptom improvement (Beta= -29.7 (9.71), p-value=0.015, p(Threshold)=0.05, Nsnps=17950) and remission at Week 8 (OR=0.08 [0.013-0.42], p-value =0.017, p(Threshold)=0.05, Nsnp=17950). PRS for MDD was negatively correlated with antidepressant remission at Week 8 (OR=0.38 [0.18-0.78], p-value=0.043, p(Threshold)=0.05, Nsnp=20793). However, PRS for MDD showed a positive correlation with antidepressant symptom improvement (Beta=5.56 (2.22), p-value=0.04, p(Threshold)=0.0001, Nsnp=206) and remission at Week 16 (OR=1.86 [1.18-2.90], p-value=0.025, p(Threshold)=0.0001, Nsnp=206). None of the PRS associations survived the Bonferroni threshold (0.001). Our results show a trend that higher polygenic loading for PTSD and MDD is associated with worse antidepressant symptom improvement and remission confirming previous findings. Interestingly, when antipsychotics were added as augmentation therapy with antidepressants, the PRS for MDD was associated with better symptom improvement and remission. While this finding should be validated in larger samples, augmentation therapy with antipsychotics may be needed for individuals with high genetic liability for MDD.