Heterogeneous Genetic Architecture in Lung Cancer Risk

Title: Heterogeneous genetic architecture in lung cancer risk Project Summary/Abstract Lung cancer is a multifactorial disease driven by environmental exposures, inherited germline genetic variants, and an accumulation of somatic genetic events. Most lung cancer cases are attributed to cigarette smoking. Approximately 10% of lung cancer cases have a family history of at least one first-degree relative.

Genome-wide association studies (GWAS) have identified about 45 susceptibility loci directly influencing lung cancer risk. The genetic architecture in individuals with a family history of lung cancer differs from those with sporadic lung cancer, although there are common genetic loci influencing both cases. The magnitude of the

association between smoking and lung cancer risk varies by family history status. There is also a familial component to smoking behaviors, which may confound the association of lung cancer risk with smoking in hereditary cases. Each locus, however, exerts only a modest effect, and most of the heritability remains unexplained. New

approaches are therefore needed to address the problem of missing heritability. Since relatively weak signals of a genetic risk factor are difficult to detect, an improvement would be to jointly model multiple genetic factors to determine the impact of the risk. Using machine-learning, high-order interaction models among numerous

genetic features in subsets of cases defined according to the family history status, also considering smoking behaviors can help illustrate how these genetic variants in each stratified subset jointly influence lung cancer risk. The overarching aims of this proposal are to characterize the shared and distinct genetic contributions to

lung cancer development and provide better insight into homogeneous and heterogeneous genetic architectures of lung cancer etiology across family history status in different populations. The specific aims of the proposed research are to identify the hierarchical combination of multiple genetic variants for lung cancer susceptibility

using machine-learning and to explore in depth the impact of genetic factors in lung cancer GWAS of European- and African-ancestry populations. This proposal capitalizes on existing well genotype-phenotype repositories of sporadic and familial lung cancers using subset-based and machine-learning association methods in different

populations. This proposal addresses a comprehensive approach to improve our understanding of how homogeneous and heterogeneous genetic architectures and biological mechanisms contribute to lung cancer development and how specific genetic effects, additive, dominant, or recessive, explain lung cancer risk. Finally,

we can identify subgroups of individuals at high-risk based on the individual’s genetic characteristics, with potential implications for the prevention, screening, and management of lung cancer.