Fine mapping the association between the Major Histocompatibility Complex and Barrett’s oesophagus/Oesophageal adenocarcinoma


Year:

Session type:

Theme:

Laura Chegwidden1,Johannes Schumacher2,Rebecca Fitzgerald3,Thomas Vaughan4,Stuart MacGregor5,Puya Gharahkhani5,Janusz Jankowski6,Ian Tomlinson1,Claire Palles1
1Institute of Cancer and Genomic Sciences, University of Birmingham,2University of Bonn,3Medical Research Council (MRC) Cancer Unit, Hutchison-MRC Research Centre and University of Cambridge,4Division of Public Health Sciences, Fred Hutchinson Cancer Research Center,5QIMR Berghofer Medical Research Institute,6The Royal College of Surgeons Ireland and The University Hospitals of Morecambe Bay Hospitals NHS Trust

Abstract

Background

rs9257809 is associated with Barrett’s Oesophagus (BO) and Oesophageal Adenocarcinoma (OAC) risk (OR=1.23;95%CI=1.14-1.31;P=5x10-9)[Gharahkhani et al.,2016].  This common variant has the largest effect size of all BO/OAC SNPs and maps to the Major Histocompatibility Complex (Human Leukocyte Antigen (HLA) in humans). It is unclear which gene is affected. HLA imputation suggested class I HLA alleles might be associated with risk but did not explain the association as well as the lead tagging SNP. Chr6 tagging SNP data available from four BO/OAC GWAS studies (10,279 cases and 17,159 controls), provided a large dataset for fine-mapping. Expression quantitative trait loci (eQTL) data from multiple gastrointestinal tract tissue types is available from GTEx Portal (https://www.gtexportal.org/home/), providing an opportunity to determine which disease-associated alleles in the locus alter gene expression.

Method

SNP2HLA [Jia et al.,2013] was used to impute HLA classical alleles and amino acid positions in HLA genes. Association testing was performed in PLINK, adjusting for significant principal components.  Meta-analysis was performed using GWAMA. GTEx data was interrogated for gene expression effects of all SNPs in LD (r2>0.2) with rs9257809.

Results

None of the imputed classical HLA alleles reached genome wide significance (P=5x10-8).  HLA-DPB1*1001 was the most significantly associated classical HLA allele.  GTEx data shows rs9257809 is associated with expression of Zinc Finger Protein 57 (ZFP57), known to be involved in regulating gene imprinting via methylation maintenance (P=2x10-14).

Conclusion

Evidence would support rs9257809 as the functional variant in the locus.  We plan to validate eQTL findings using data from BO/OAC samples from Oesophageal Cancer Clinical and Molecular Stratification Study (OCCAMS) and perform in vitro assays to determine how rs9257809 affects ZFP57 and what the role of this gene is in BO/OAC predisposition.  We also aim to increase the sample size for SNP2HLA analysis by including BO/OAC cases and controls identified in UKBiobank for further fine mapping.