SA91 - INCREASED PREDICTED C4A EXPRESSION IS ASSOCIATED WITH COGNITIVE DEFICIT IN BOTH SCHIZOPHRENIA AND ALZHEIMER'S DISEASE

Abstract

Background Recently, Sekar et al. (Nature, 2016) reported that the top GWAS hit for schizophrenia in the Major Histocompatibility Complex (MHC) locus implicated structural variants in the C4 gene which are associated with C4A expression and may affect synaptic pruning. Synapse loss is associated with cognitive decline in a number of disorders, including Alzheimer's disease and schizophrenia. We hypothesised that C4A expression predicted from C4 structural alleles is associated with cognition. We tested this in the Western Australian Family Study of Schizophrenia (WAFSS) (n=770, including 402 schizophrenia cases and 368 controls), and the Australian Imaging, Biomarkers and Lifestyle Study (AIBL) (n=1,112, including 211 Alzheimer's disease cases, 133 with mild cognitive impairment and 768 controls). Methods Sekar et al. showed that C4 structural alleles could be imputed from GWAS data with an accuracy of ~70%. We used 222 reference haplotypes with C4 structural alleles measured by Sekar et al. to impute C4 alleles in all individuals using Beagle. C4A expression was predicted for each individual using the regression coefficients calculated by Sekar et al. Composite measures of cognition were available for both WAFSS and AIBL (Hallmeyer et al., Am J Hum Genet. 2005; Burnham et al., J Alzheimers Dis. 2015). Linear regression was used to assess their relationship with predicted C4A expression. P values for these relationships were calculated using linear mixed models to account for the relatedness between some WAFSS participants, and repeated measurements in the AIBL dataset. All analyses were adjusted for case status, age, sex, and education. Results Although mean C4A expression was not significantly different between schizophrenia cases and controls in the WAFSS (P=0.7), cases were overrepresented in those with the highest C4A expression levels. There was a small but consistent association between increased cognitive deficit and increased levels of C4A in both the WAFSS and the AIBL samples (P Discussion These data suggest that C4 structural variants previously shown to be associated with schizophrenia may specifically implicate the cognitive deficit which is a core feature of schizophrenia. Further, our finding that a similar relationship between C4A levels and cognition exists in patients with schizophrenia and Alzheimer's disease suggests that the locus may be pleiotropic and have implications for a range of neuropsychiatric disorders.