Rare genetic variant contributions to multiple cancers

Abstract

Clinical and family-based studies have long identified strong associations between rare genetic variants, typically defined as genetic variants with minor allele frequencies < 1% in a population, with pathogenic effects on specific genes and hereditary cancer predisposition syndromes. For example, the lifetime risk of diagnosis across a variety of cancers, including breast and ovarian, among carriers of rare pathogenic BRCA1 or BRCA2 variants estimated to be many times higher than average population risk for those cancers. However, known rare pathogenic variants in established cancer predisposition genes still only account for a small proportion of the total cancer heritability estimated by twin studies. Some of these contributions can be attributed to common genetic variants, typically defined as genetic variants with minor allele frequencies ≥ 1% in a population. Cancer genome wide-association studies (GWAS) have identified many hundreds of genomic loci associated with multiple cancer types including a subset of at least 130 where common variants are associated with multiple cancer types. While these common variants do explain a large proportion of cancer heritability cancer, there are still substantial gaps. Genetic previously unknown rare variants in known cancer predisposition genes or other genes are a potential source of genetic contributions to risk in the population. However, the relationships between these rare variants, and cancer risk at the population level is currently not well understood due to a historical lack of the large sequencing datasets needed to conduct population-based analyses of rare variants. Only a handful of previous studies have been conducted in this area, and most have either 1) focused on a small number of cancers or 2) focused on a limited number of cancer predisposition genes. With the generation of these datasets in studies with linked cancer diagnosis data, more comprehensive examinations of the contributions of rare genetic variants to risk across multiple cancer types is becoming feasible. In this dissertation, we attempted to extend our knowledge of rare variant contributions to cancer risk across multiple cancers and how we may leverage these variants to improve risk prediction models. We conducted exome-wide gene-based mixed-model rare coding variant analyses in two large cohorts to identify genes where rare variants have pleiotropic associations with over 70 cancer types, including cancers that are both common and rare in the population (Chapter 1). We then further characterized the magnitudes and directions of associations between the pleiotropic genes we identified and 27 individual cancer types. We also estimated the contributions of rare variants to breast, colorectal, and prostate cancer heritability (Chapter 2) and their utility in genetic risk prediction models for those cancers (Chapter 3). In summary, this work can expand our understanding of shared rare variant genetic architecture across multiple cancers and how this knowledge might be used to improve risk stratification for some of the most common cancer types. The results of this dissertation suggest that rare coding variation in specific genes may be related to risk across a wide range of cancers and that rare coding variants in a small number of genes play small but important roles in population-level breast, colorectal, and prostate cancer risk.