Abstract
Endometrial cancer is the most common gynaecological cancer in New Zealand and the incidence is increasing as the population ages. Genetic predictors of endometrial cancer risk that allow early detection of the disease are important for prevention and improved management strategies. Mutations in the mismatch repair (MMR) genes MLH1, MSH2, MSH6, PMS1 and PMS2 are known to confer increased risk in a proportion of endometrial cancer cases, and the mutation spectrum includes copy number variants (CNVs). There are several other genes encoding proteins that act in the MMR pathways, but to date the evidence for their involvement in endometrial cancer predisposition is limited. This study aims to 1) To identify genes in the MMR pathway that are overlapped by CNVs in endometrial cancer cases, 2) To compare the CNV frequency of common and rare CNVs between endometrial cancer cases and controls, 3) To identify regions in the genome that are associated with endometrial cancer risk, and 4) To identify biological pathways that are enriched for genes overlapped by CNVs in cases and controls. Genome-wide scanning of CNVs was performed using Illumina610k single nucleotide polymorphism data from a large cohort of ~1300 endometrioid endometrial cancer cases and ~600 population-based female controls. Up to four CNV calling algorithms (CNVPartition, QuantiSNP, PennCNV and GNOSIS) were used to identify CNVs, and where possible, novel CNVs were validated by quantitative PCR.
This study has identified and confirmed deletions that disrupted known endometrial cancer susceptibility genes, MSH2 and MSH6. We also identified novel variants in several other mismatch repair pathway genes, including duplications overlapping TGFBR3 and MUTYH, and a deletion in RPA3, that are predicted to disrupt the coding sequence of TGFBR3 and RPA3. These results suggest that other genes within MMR pathway may be disrupted by mutations in a proportion of endometrial cancer cases and warrant further investigation. Three approaches were carried out to assess CNV load in cases and controls that defined CNV regions based on overlap with CNVs in the Database of Genomic Variants, overlap with discrete CNV clusters, and overlap with known coding genes. Results suggested that although there was no difference in the number of total CNVs between cases and controls, rare CNVs were shown to be more frequent in cases. For example, cases were found to have twice as many rare deletions as controls (p-value=2.2x10-16). Pathways analysis of genes overlapping these rare CNVs did not identify major gene networks in the endometrial cancer cases that were distinct from those found in the controls. However, some genes affected in the cases were found to form a network with genes, such as VEGF and FSH, which are known to play an important role in endometrial cancer development. Results from a genome-wide association study, identified several genes, including NF-κB, FSH and TK1 that were differentially affected by CNVs in cases and controls, but the relationship between these genes and endometrial cancer is unclear. Future studies will focus on confirming CNV load, genome wide association and pathway results in a larger cohort with age matched controls and characterising the contribution of individual variants with lymphablastoid cell line transcriptome analysis.
In conclusion, this study has identified areas for further investigation that have the potential to provide new insights into roles of CNVs in endometrial cancer susceptibility. These include three genes, RPA3, TGFBR3, and MUTYH, that have not previously been associated with endometrial cancer, and genomic features such as CpG islands, G4 quadruplexes and miRNAs that may have contributed to the development of endometrial cancer in this cohort. This thesis also highlights a number of limitations associated with studies of this kind, such as cohort size and cohort design, which must be overcome in order to successfully elucidate genetic contributors to this disease.