Association Between Polygenic Risk Score for Schizophrenia and Neurocognitive Measures in the Western Australian Family Study of Schizophrenia (Wafss)

Abstract

Background Schizophrenia (SZ) is a clinically heterogeneous disorder with multifactorial causes including a significant genetic contribution. Recent, large scale GWAS have described a polygenic risk score (PRS) for SZ which accounts for ~ 7% of variation on the liability scale (Bulik-Sullivan et al, 2015). The aim of this study was to ascertain whether the PRS also contributed to a range of psychological measures which are associated with SZ. Methods Our cohort comprised a set of 127 multiplex families recruited through a proband with SZ (n=536 including 161 SZ cases) and a control group of 30 unaffected families (n=121). We examined neurocognitive measures across 6 domains: General cognitive ability (National Adult Reading Test & Shipley Institute of Living Scale IQ); verbal learning and memory (Rey Auditory Verbal Learning Test); sustained attention (Continuous Performance Task, degraded stimulus and identical pairs); speed of information processing (inspection time); executive function (Controlled Oral Word Association Task) and personality factors (Schizotypal Personality Questionnaire & Temperament and Character Inventory) (Hallmayer et al, 2005). Whole genome sequence (WGS) data (Illumina 10X platform, 15X coverage) were available for a subset of the multiplex families (n=320). We mapped all 102,636 PRS SNPs to the called WGS data by rsID; 97,507 passed QC and were used to calculate a PRS for each individual (Purcell et al, 2009). Correlation between PRS and each neurocognitive measure was conducted in SOLAR adjusting for the relationship matrix, age, sex and educational attainment. Results In the multiplex families, all psychological measures were associated with SZ in the expected direction (FDR corrected at α 0.05). Comparison with the control families showed that the Shipley measure of current IQ, and the identical pairs measure of sustained attention were significantly impaired in unaffected relatives of those with SZ compared the healthy control cohort (FDR-corrected at α 0.05) indicating a shared genetic component with SZ. PRS was significantly correlated with the identical pairs task, both inspection time tasks, and personality measures after FDR correction with the expected direction of effect. Discussion Recent studies have shown associations between PRS and neurocognitive measures in healthy populations, including IQ measures in a population based sample of children (Hubbard et al 2016) and working memory in young adults (Hatzimanolis et al 2015). Our findings add to these data and support the presence of shared common genetic factors between SZ and measures of sustained attention, speed of information processing and personality factors in a familial cohort with SZ. This information may help to inform the selection of endophenotypes for SZ.