TRANSETHNIC ANALYSIS OF HIGH-RESOLUTION HLA ALLELES AND COMPLEMENT 4 STRUCTURAL POLYMORPHISMS IN SCHIZOPHRENIA

Background Genome-Wide Association Studies (GWAS) have produced strong evidence for association of Schizophrenia (SCZ) in the extended Major Histocompatibility Complex (chromosome 6q), including the Human Leukocyte Antigen (HLA) region. It has been reported that structural polymorphisms in the C4 gene can explain the signal in this region, but it has not been determined by direct high-resolution genotyping whether HLA alleles have independent effects. We have used a novel next-generation sequencing method to achieve HLA typing with 8-digit (4-field) resolution over classical HLA genes (A, B, C, DRB1, DRB3-5, DQA1, DQB1, DPA1 and DPB1) in cohorts of three ethnicities; and assayed C4 structural polymorphisms in a subset of the sample to permit imputation in the entire cohorts. These data permit analysis of the association of SCZ with HLA alleles while accounting for C4 variants. Methods Subjects included 21,592 individuals: SCZ cases/controls from the MGS and GPC cohorts (European-ancestry: 5,762 cases, 5,558 controls; African-American: 2,671 cases, 2,373 controls), and 1,670 Taiwan families (3,340 parents, 1,728 affected offspring). HLA genotyping was carried out using the Mia Fora NGS Flex HLA Typing Kit (Immucor), which uses long-range PCR to capture 11 HLA loci (the entire gene in most cases) followed by high-depth, multiplexed Illumina sequencing. C4 polymorphisms were assayed using ddPCR using published primers and calling algorithms, in 940 individuals representing all ancestries and genotyping platforms. Analyses are almost completed, including imputation of C4 structural alleles into all samples (HIBAG), followed by frequentist association tests of HLA alleles with and without conditioning on C4 alleles, taking ancestry covariates into account (PLINK). It is expected that imputation of HLA genotypes can also be improved for each ancestry based on the high-resolution genotyping of large cohorts. Results All analyses are in progress and will be completed prior to the Congress. Missing data rates across loci are ~1%, and Mendelian error rate in trios is low. C4 alleles can be imputed with high accuracy in all combinations of ancestry and genotyping platform. Imputation accuracy was over 90% for all HLA genes. Discussion The results of this study will provide important new information about whether HLA alleles have effects on SCZ risk, independent of their linkage disequilibrium with C4 structural polymorphisms. The study will also provide improved resolution of imputation of both C4 polymorphisms and of HLA alleles.