45-OR

Abstract

GWAS reported associations between HLA region SNPs and multiple cancers whose mechanisms remain unexplored. In most cases, whether the associated SNP represents a proxy for an already known HLA association, and if so, whether it is a non-functional proxy or possibly the causal SNP underlying the HLA association, is unknown. To gain mechanistic insight into the reported associations, we compiled cancer associations listed in NHGRI-GWAS catalog and selected ones found in candidate gene studies. We also included hits from the HLA region in breast cancer GWAS. We mapped 32 of them on HLA haplotypes using 103 HLA-typed IHWG reference cell lines, examined their independence, and assessed their functionality by bioinformatics tools and existing empirical evidence. Most SNPs were not exclusive to HLA alleles/haplotypes/lineages, but the SNPs associated with lymphoid malignancies, lung cancer, and nasopharyngeal cancer showed some correlations. One common finding was that most of the cancer-associated SNPs were correlated with MICA/B or HLA class I/II gene expression (especially HLA-DQA1), and some had functional effects on other genes. The non-HLA genes involved in BCL signaling pathway (BAK1 ‘BCL2-antagonist/killer 1’; BAG6 ‘BCL2-associated athanogene 6’), stress response (MICA; HSPA1A/B; BAG6/BAT3), and DNA damage repair (MDC1) appeared to be involved in more than one cancer as causal pathways. Two SNPs showed LD (rs130067/prostate cancer and rs1265100/breast cancer hit in the HLA region), but all others were independent. Overall results showed that there are multiple cancer risk markers within the HLA region, they are independent markers for different cancers, and not simply proxy markers for certain HLA types. This analysis suggests that besides cancer immune surveillance mediated by specific HLA types, involvement of the diverse and rich set of non-HLA genes within the HLA region should be considered in mechanistic studies of HLA and cancer associations.