Pathway-based polygenic risk scores for schizophrenia and associations with clinical and neuroimaging phenotypes in UK Biobank

Abstract

AbstractBackgroundSchizophrenia is a heritable psychiatric disorder with a polygenic architecture. Genome-wide association studies (GWAS) have reported an increasing number of risk-associated variants and polygenic risk scores (PRS) now explain 17% of the variance in the disorder. There exists substantial heterogeneity in the effect of these variants and aggregating them based on biologically relevant functions may provide mechanistic insight into the disorder.MethodsUsing the largest schizophrenia GWAS to date, we calculated PRS based on 5 gene-sets previously found to contribute to the pathophysiology of schizophrenia: the postsynaptic density of excitatory synapses, postsynaptic membrane, dendritic spine, axon, and histone H3-K4 methylation gene-sets. We associated each PRS, along with respective whole-genome PRS (excluding single nucleotide polymorphisms in each gene-set), with neuroimaging (N>29,000; cortical, subcortical, and white matter microstructure) and clinical (N>119,000; psychotic-like experiences including conspiracies, communications, voices, visions, and distress) variables in healthy subjects in UK Biobank.ResultsA number of clinical and neuroimaging variables were significantly associated with the axon gene-set (psychotic-like communications: β=0.0916, pFDR=0.04, parahippocampal gyrus volume: β=0.0156, pFDR=0.03, FA thalamic radiations: β=-0.014, pFDR=0.036, FA posterior thalamic radiations: β=-0.016, pFDR=0.048), postsynaptic density gene-set (distress due to psychotic-like experiences: β=0.0588, pFDR=0.02, global surface area: β=-0.012, pFDR=0.034, and cingulate lobe surface area: β=-0.014, pFDR=0.04), and histone gene-set (entorhinal surface area: β=-0.016, pFDR=0.035). In the associations above, whole-genome PRS were significantly associated with psychotic-like communications (β=0.2218, pFDR =1.34×10−7), distress (β=0.1943, pFDR =7.28×10−16), and FA thalamic radiations (β=-0.0143, pFDR=0.036). Permutation analysis carried out for these associations revealed that they were not due to chance.ConclusionsOur results indicate that genetic variation in 3 gene-sets relevant to schizophrenia (axon, postsynaptic density, histone) may confer risk for the disorder through effects on a number of neuroimaging variables that have previously been implicated in schizophrenia. As neuroimaging associations were stronger for gene-set PRS than whole-genome PRS, findings here highlight that selection of biologically relevant variants may address the heterogeneity of the disorder by providing further mechanistic insight into schizophrenia.