Identification and validation of novel breast cancer predisposition genes

Abstract

The genetic causes of the majority of hereditary breast cancer families remain unresolved (BRCAx families) and lack of this information compromises primary and secondary cancer prevention for the affected women and their family members. Despite intensive efforts, no major new breast cancer predisposition genes with equivalent impact have been discovered since the identification of BRCA1 and BRCA2 in the 1990s. Research from our laboratory provided direct demonstration that the remaining hereditary causes of breast cancer are not due to a few BRCA1-like genes (high penetrance and high carrier frequency) but instead by numerous moderate penetrance genes, each accounting for only a small fraction of families. This finding highlighted the need for very large case-control studies as the only way of resolving the missing breast cancer heritability. In this thesis I aimed to carry out such a study.

Using existing germline whole exome and targeted sequencing data, 162 candidate genes were selected for validation in ~8,000 unrelated BRCAx cases and cancer-free controls. Combined with the existing targeted sequencing data of ~4,000 cases and controls, this analysis represents one of the largest targeted sequencing studies of its kind. A significant association with breast cancer predisposition was confirmed for known breast cancer genes PALB2, CHEK2 and ATM. Among the 162 candidate genes, nearly twice as many had more LoF variants in the cases than the controls, compared to those with more variants in the controls than the cases, demonstrating a high enrichment for genuine breast cancer predisposition genes. Most novel candidate genes appeared to convey only low to moderate risks, with a total of 35 genes identified that had an OR>2. Despite the very large sample size, the number of carriers of LoF variants for any candidate gene was still small, further demonstrating the extreme genetic heterogeneity of BRCAx families, and that case-control data is still insufficient on its own to claim the discovery of a specific new breast cancer gene. Nevertheless, this large-scale case-control data will be a valuable guide for future validation. Many of the candidate genes such as NTHL1, WRN, BAP1, PARP2 and CDK9 play essential roles in DNA damage repair, consistent with the function of all the established breast cancer genes. An intriguing exception is CTH which is involved in the trans-sulfuration pathway, and if confirmed, would be the first moderate penetrance breast cancer predisposition not involved in DNA repair. This study also re-evaluated some contentious genes and provided strong evidence to reject RINT1 and RECQL as breast cancer predisposition genes. On the other hand, by combining both case-control data with tumour sequencing data, RAD51C was identified as a triple-negative breast cancer predisposition gene.

This study has shown that because of the high genetic heterogeneity of BRCAx families, future validation studies will require either an extremely large sample size (>10,000 subjects), and/or the inclusion of approaches providing independent evidence, such as tumour sequencing to identify bi-allelic inactivation and specific mutational signatures.