SNP-based HLA tagging, imputation, and association in adverse drug reaction of epilepsy patients from Hong Kong

Abstract

Human leukocyte antigen (HLA) genes control the regulation of the human immune system and are involved in infection and autoimmune-related complex diseases. Some HLA alleles (B*15:02, A*31:01 and B*57:01) have been associated with increased risk of cutaneous adverse drug reaction (cADR) in epilepsy patients taking antiepileptic drugs like carbamazepine and abacavir. However, traditional serotyping or Sanger sequencing for HLA genotyping may not be cost-effective for clinical application. A more efficient way may be to look for single nucleotide polymorphisms (SNP) that can tag HLA risk alleles with high sensitivity and specificity. Since haplotype structure is population-specific due to different linkage disequilibrium patterns, an investigation on pairwise SNP-HLA relationships was conducted in the Hong Kong population. This analysis arose from our program of research on epilepsy genetics and pharmacogenomics, for which we have built an in-house database including classical HLA serotyping, genome-wide SNP arrays, exome sequencing, and whole-genome sequencing. The discovery panel employed included 144 patients with both genome-wide SNP-array data and HLA*B or HLA*A serotypes, which enable a greedy search of highly correlated pairs of HLA risk alleles and SNPs through use of pairwise LD statistics. Common variants rs9265348, rs2532923, and rs114025781 were found to tag HLA risk alleles B*15:02, B*58:01, and B*13:01, respectively, with 100% sensitivity and >95% specificity. No HLA-A risk allele was found to be tagged by any single SNP. The results are being replicated in an independent sample, for which both HLA and SNP genotypes can be called directly from exome or whole genome sequencing. In addition, SNP2HLA was used to construct a Hong Kong population-specific reference dataset for HLA imputation. Using the GWAS sample, complete HLA-A and HLA-B allele information was inferred for a total of 408 epilepsy patients whose cADR status upon taking aromatic antiepileptic drugs was recorded. To determine how HLA genes affect cADR individually and jointly, this data will be combined with Caucasian populations from the International League Against Epilepsy Consortium on Complex Epilepsies for an overall trans-ethic meta-analysis.