Documentation of HLA-DRB1*1302 As a Maternal Prenatal Risk Factor for Autism Using Imputation in a Maternal-Fetal Incompatibility Model

Background:  The Major Histocompatibility Class (MHC) HLA-DR, and in particular its highly polymorphic beta chain gene, HLA-DRB1, have been studied in autism. The association of autism with HLA genes and haplotypes suggests that an underlying dysregulation of the immune system in autism may be mediated at least in part by HLA genes (Torres AR et al, 2012). Early studies by Reed Warren and others raised the question of whether an HLA-DRB1 gene acting in the mother during pregnancy might contribute to autism in her fetus. Later, Lee et al, 2006, analyzed two datasets and reported that in one of them, individuals with autism and their mothers but not their fathers had a significantly higher frequency of HLA-DR4 alleles compared with controls. We followed up these studies using TDT in maternal trios (maternal grandparents & mothers) as well as TDT in case trios, and found an odds ratio of 4.7 (95% CI: 1.3 to 16.2) for transmissions of HLA-DR4 to mothers of individuals with autism from maternal grandparents, but we did not see increased transmission to the individuals with autism. In the present study, we used the very large Simons Simplex dataset to extend these findings in silico by HLA imputation of GWAS data.

Objectives: To attempt a more in-depth study of the HLA region in autism by a newer method.

Methods:  We used HLA*IMP2, Dilthey et al. 2013, and HIBAG, Zheng et al. 2013, to impute HLA alleles for HLA-A, -B, -C, -DPB1, -DQA1, -DQB1, -DRB1, -DRB3, -DRB4, and -DRB5 in individuals with autism and their parents from GWAS data available in the Simons Simplex Collection Version 14 on the Illumina Human 1M Duo dataset. Families were limited to those of European descent. All individuals’ data were imputed without familial information.  We tested for association with autism by the Maternal-Fetal Genotype Incompatibility Test as implemented in the Mendel software. We received IRB permission for this study.

Results:  HLA-DRB1*1302 showed a significant effect (p<3.6 x 10-4) under the non-inherited maternal antigens model, in which offspring without the allele whose mothers who were heterozygous for the allele had increased risk.  No alleles showed significant RHD-type incompatibility effects (i.e., heightened risk for heterozygous offspring of mother without the allele). This result remained significant after correction for multiple comparisons.

Conclusions: The present study supported action of an HLA-DRB1 allele in autism as a maternally acting gene allele in a very large dataset.  Maternally acting gene alleles are key determinants of the intrauterine environment during pregnancy.  They may be of particular importance very early during pregnancy when the ratio of maternal to fetal mass is largest. This is important for the pathogenesis of autism as well as for potential treatment, since the prenatal period is the earliest opportunity for treatment of autism.