Polygenic Scores Derived From Neuroimaging Endophenotypes Predict Outcomes To Psychotherapy And Medication Treatments For Major Depressive Disorder

Background Biomarkers to guide optimal treatment selection are lacking for major depressive disorder (MDD). Despite some promising findings, candidate gene and genome-wide association studies (GWAS) of antidepressant response have met with little success and none has focused on differential outcome to mechanistically different treatments. Development of such biomarkers would allow to better match patients with their most favourable treatment and has direct implications for the development of precision biology-based clinical practice. Methods The overall aim of this work is to provide a framework for developing easily accessible predictors of individualized treatment response. Using a neuroimaging-based genomics approach, we investigated whether polygenic scores (PGS) derived from single nucleotide polymorphisms (SNPs) influencing hippocampus (HC) volume from the Enhancing NeuroImaging Genetics through Meta-Analysis (ENIGMA) study (Hibar et al., 2015) could predict treatment-specific outcome in an independent cohort; the Prediction of Remission in Depression to Individual and Combined Treatments (PReDICT) sample (B. W. Dunlop et al., 2012). PReDICT enrolled treatment-naive patients and randomized them to three antidepressant treatments; CBT, ESC and duloxetine (DUL). As such, it is the largest single-site MDD randomized trial comparing CBT to antidepressant medication (ADM) ever performed. Results HC-PGS predicted differential outcomes (P = 0.00053 and R2 = 4.6) to CBT vs. ADM in PReDICT. In addition after predicting randomly permuted response status 1000 times with the best-fit HC-PGS, only one P-value was lower than those initially achieved (pperm Discussion The approach used in this work contributes to the identification of molecular pathways possibly critical for antidepressant outcomes and offers novel insights into MDD pathophysiological subtypes. We demonstrate that combining neuroimaging and genetic markers is essential to identify predictors of antidepressant response and may allow selection of a specific treatment for specific patients.