Polygenic heterogeneity in antidepressant treatment and placebo response

Abstract

The genetic architecture of antidepressant response is poorly understood. Polygenic risk scores (PRS), exploration of placebo response and the use of sub-scales might provide insights. Here, we investigate the association between PRSs for relevant complex traits and response to vortioxetine treatment and placebo using clinical scales, including sub-scales and self-reported assessments. We collected a clinical test sample of Major Depressive Disorder (MDD) patients treated with vortioxetine (N = 907) or placebo (N = 455) from seven randomized, double-blind, clinical trials. In parallel, we obtained data from an observational web-based study of vortioxetine-treated patients (N = 642) with self-reported response. PRSs for antidepressant response, psychiatric disorders, and symptom traits were derived using summary statistics from well-powered genome-wide association studies (GWAS). Association tests were performed between the PRSs and treatment response in each of the two test samples and empirical p-values were evaluated. In the clinical test sample, no PRSs were significantly associated with response to vortioxetine treatment or placebo following Bonferroni correction. However, clinically assessed treatment response PRS was nominally associated with vortioxetine treatment and placebo response given by several secondary outcome scales (improvement on HAM-A, HAM-A Psychic Anxiety sub-scale, CPFQ & PDQ), (P ≤ 0.026). Further, higher subjective well-being PRS (P ≤ 0.033) and lower depression PRS (P = 0.01) were nominally associated with higher placebo response. In the self-reported test sample, higher schizophrenia PRS was significantly associated with poorer self-reported response (P = 0.0001). The identified PRSs explain a low proportion of the variance (1.2-5.3%) in placebo and treatment response. Although the results were limited, we believe that PRS associations bear unredeemed potential as a predictor for treatment response, as more well-powered and phenotypically similar GWAS bases become available.

Fig. 1. Overview of the samples and study design.

The clinical test sample included 2191 genotyped MDD patients treated with vortioxetine or placebo from 7 randomized, double-blind, placebo-controlled clinical trials. 907 vortioxetine-treated patients and 455 placebo-treated patients of European genetic ancestry passed genotyping quality control (QC). Response was quantified using different clinical scales and sub-scales, resulting in 6 response measures for both the vortioxetine and the placebo group. The self-reported test sample of European ancestry participants suffering from depression or bipolar disorder identified 642 vortioxetine-treated participants based on the AESES questionnaire. In the statistical analysis, PRSs for treatment response, diseases, and symptom traits were calculated for everyone in both samples. Vortioxetine and placebo response association with these PRSs were estimated.

Fig. 2. Treatment response PRSs association with vortioxetine response.

A Bar plot showing the coefficients of self-reported responders PRS and clinically assessed response PRS association with vortioxetine response in the clinical and self-reported test samples. B Odds ratios for vortioxetine response given by HAM-A PA improvement in the clinical test sample for different quantiles of clinically assessed response PRS.

Fig. 3. Schizophrenia PRS association with vortioxetine response.

A Bar plot showing the coefficients of schizophrenia PRS association with placebo and vortioxetine response in the clinical and self-reported test sample. B Odds ratios for vortioxetine response in the self-reported sample for quantiles of schizophrenia PRS.

Fig. 4. Clinically assessed response and subjective well-being PRSs association with placebo response.

Bar plot showing the coefficients for all placebo and vortioxetine response outcome measures in the clinical test sample association with A clinically assessed response PRS and B subjective well-being PRS. C Odds ratios for placebo response in the clinical test sample for clinically assessed response PRS. D Odds ratios for placebo response and vortioxetine response in the clinical test sample for subjective well-being PRS.