Novel transcriptome-based polygenic risk score for depression is associated with neural processing of social stimuli

Background: Convergent data from imaging and postmortem brain transcriptome studies implicate corticolimbic circuit (CLC) dysregulation in the pathophysiology of depression. To more directly bridge these lines of work, we generated a novel transcriptome-based polygenic risk score (T-PRS), capturing subtle shifts towards depression-like gene expression patterns. We then mapped this T-PRS onto CLC function and related depressive symptoms in a non-clinical sample of young adults. Methods: Genetic, self-report, and neuroimaging data were available in 482 Duke Neurogenetics Study participants (226 men; age 19.78+/-1.23). T-PRS was generated based on common functional SNPs shifting gene expression in the brain towards a depression-like state. We used multivariate partial least squares regression to map T-PRS onto whole-brain activity patterns during perceptual processing of social stimuli (i.e., human faces). Posthoc univariate analyses followed up on the link between T-PRS and amygdala reactivity to neutral and threatening faces. For comparison, we generated a PRS summarizing depression risk variants identified by the Psychiatric Genomics Consortium (PGC-PRS). Sex was modeled as moderating factor. Results: T-PRS was associated with male-specific reductions in neural response to neutral faces in a widespread network of cortical and subcortical regions (multivariate p=0.03) including the amygdala (beta=-0.14, p=0.04). These results mirrored patterns associated with PGC-PRS independently of sex (ps