Integrating multi-polygenic scores for enhanced prediction of antidepressant treatment outcomes in an East Asian population

Abstract

Major Depressive Disorder (MDD) significantly impacts global public health, yet the effectiveness of antidepressant treatments varies widely across individuals. This study addresses an important gap in the literature by examining how multi-polygenic scores (PGSs) can improve predictions of selective serotonin reuptake inhibitor (SSRI) treatment outcomes in an East Asian population-a region where pharmacogenomic studies have been limited. We analyzed two Taiwanese cohorts: the Taipei Veterans General Hospital cohort (VGHTP, N = 177) and the National Health Research Institutes cohort (NHRI, N = 245), all receiving SSRIs. PGSs for 108 traits potentially relevant to SSRI treatment outcomes were derived from large-scale genome-wide association studies using PRS-CS and PRS-CSx, incorporating data from multiple ancestries. We combined these PGSs with demographic and clinical variables (e.g., baseline severity of depression, medication dosage) and employed generalized linear mixed models with L1-penalization (glmmLasso), as well as machine and deep learning algorithms, to identify and evaluate predictors. Our results revealed several important PGS predictors, notably related to insomnia, multisite chronic pain, and higher levels of inflammatory biomarkers, which consistently correlated with lower treatment efficacy. While the ensemble model achieved a modest area under the curve (AUC) of 0.631 for predicting responders/non-responders, integrating early improvement in depressive symptoms considerably boosted predictive accuracy (AUC = 0.859) for identifying remitters/non-remitters by week 8. These findings underscore the value of multi-PGS approaches and highlight the necessity of expanding pharmacogenomic research to non-European populations. Future studies with larger, diverse cohorts and additional biomarkers may further advance individualized therapeutic strategies and alleviate the burden of depression worldwide.