Polygenic scores from the MiR137 pathway explain variability in cognitive performance in patients with schizophrenia and controls.

Abstract

Introduction Genome-wide association studies have identified rs1622579, a common variant located within an intronic region of MiR137, as a risk variant for schizophrenia. MiR137 has a hypothesized role in neural cell proliferation, migration and maturation. Cognitively, this variant is associated with deficits in verbal episodic memory, and altered fronto-amygdala connectivity. As MiR137 interacts with several other genome wide associated variants (e.g. ZNF804A, TCF4, CACNA1C, CSMD1 and WBP1L), we tested whether these cognitive effects were associated with polygenic risk in a set of identified downstream targets of MiR137. Methodology Polygene scores were calculated by (1) selecting all identified downstream genetic targets of MiR137 (Hill et al., 2014); (2) selecting risk SNPs within each of these genes based on risk estimates from the Psychiatric Genomics Consortium (Ripke et al., 2011), and (3) ascribing a polygenic risk score to each participant based on the total number of risk alleles they carried. To define risk, we used 3 arbitrary p-value thresholds (p < 10-5, p < 0.05, and p < 0.5) from the PGC1 case-control analysis. To estimate these polygene effects on cognitive deficits associated with SZ (IQ, episodic and working memory, attention control, and theory of mind) linear regression analyses was performed, with age and gender entered as variables of no interest. We carried out three separate analyses: all cases and controls (n=586), patients with a broad diagnosis of psychosis including non-schizophrenia psychosis (n=489) and patients with schizophrenia and schizoaffective disorder (n=379). Results MiR137 polygenic risk scores were significantly higher in the broad and narrow diagnosis groups compared to the controls (F(2,1081) =5.52, p<0.01). Poorer performance on measures of IQ, memory, and attention were each found to be associated with higher MiR137 polygene scores thresholded at p=0.05, including paired associated learning (p<0.001); faces, letter-number sequencing (LNS) sustained attention to reaction time (SART) (p<0.01); full-scale IQ, logical memory (LM) and spatial working memory (SWM) (p<0.05). The amount of variance explained on the measures varied between 1% and 2.6% Discussion These data support a modest but significant influence of this MiR137 'pathway' on cognition, across multiple measures of cognition. These widespread cognitive effects are consistent with MiR137’s previously described role in brain development. Greater variance was explained when all cases and controls were included, rather than considering narrow, broad, and controls separately. Extending on previous findings, MiR137's effects on cognition were significantly more apparent when considered in combination with the other risk variants with which it interacts rather than when the rs1622579 risk variant was considered in isolation.