Exercise is broadly beneficial for human health and is one of the most potent modifiable lifestyle risk factors for many diseases, including cancer. Cancer is a highly prevalent disease and is one of the leading causes of death worldwide. While cancer incidence is a stochastic process, many factors influence the probability that an individual will develop cancer, including genetics, environmental exposures and lifestyle factors. An estimated two-thirds of cancer deaths in the U.S. can be attributed to modifiable risk factors such as smoking, diet and exercise. Exercise is the strongest positive modifiable risk factor and has been linked to almost all cancer types and stages of disease progression. Individuals who exercise more have reduced risk of developing cancer and improved clinical outcomes. However, even within a specific cancer type, tumors appear to respond differently to exercise. Indeed, the molecular mechanisms by which exercise exerts its effect on cancer outcomes are almost entirely unclear. My dissertation aims to fill this fundamental gap in our understanding of cancer etiology, uncovering how exercise affects diverse host and tumor molecular landscapes, as well as clinical outcomes. To enhance our molecular understanding of exercise oncology, I have separated my research into three chapters. These chapters cover a variety of study designs, including large cross-sectional patient cohorts, prospective longitudinal clinical trials, and experimental mouse studies. Each study design provides its own unique advantages, complementing each other and coming together to reveal novel insights into exercise oncology.In Chapter 1, I present the study of a large cross-sectional cohort of 5,150 patients with linked tumor genomic sequencing from 38 different cancer types and clinical annotation of post-diagnosis exercise dose and other important covariate lifestyle behaviors such as smoking, alcohol, and diet. Leveraging the large sample size and diversity of cancer types, we investigated both pan-cancer and cancer type-specific exercise-associated modulation of the cancer genome. Tumors differed in mutation burden, mutational signatures, and specific driver mutations in an exercise dose-dependent manner. The direction and magnitude of these associations varied across cancer types, yet exercisers had reduced hazard of all-cause mortality for all cancers combined and for multiple individual cancer types. Our data show exercise promotes genome maintenance, which may have broad implications for understanding how exercise suppresses tumor pathogenesis and potentially other common age and lifestyle-related diseases. To our knowledge, this study is the first to characterize the link between exercise and human tumor genomic profiling.
In Chapter 2, I present the study of a decentralized, digital prospective longitudinal clinical trial of 13 patients with prescribed exercise therapy. The prescribed exercise regimen between cancer diagnosis and surgical resection allows for the control of exercise dose in humans linked to high-quality clinical data. We performed longitudinal multiparametric profiling of host physiology, plasma, gut microbial composition, and tumor tissue before, during, and after exercise therapy intervention. Time-series analyses revealed hundreds of host molecular changes in the plasma proteome and metabolome and gut microbiome involved in a diverse-array of biological processes. System-wide changes were paralleled by modulation of core tumor gene expression pathways notably tumor cell cycle regulation, stress response, and metabolism. Integrative network analyses revealed the complexity of the host–tumor interaction under exercise therapy regulation, elucidating novel mechanistic insights. Variability at baseline and in response to treatment emphasized highly personalized responses to uniform exercise therapy. Our study provides an example of the application of a digital approach to generate a longitudinal high-definition dataset providing a framework of the integrative effects of exercise therapy with considerable translational and discovery potential. To our knowledge, this study is the first deep longitudinal host-tissue molecular characterization of patient response to exercise therapy.
In Chapter 3, I present a mouse study of tumor xenografts from seven human breast cancer cell lines and syngeneic grafts from one mouse breast cancer cell line, with the cell lines representing a range of breast cancer subtypes. Tumors derived from different cell lines displayed differential growth phenotypes, with some tumors growing faster and others growing slower in response to exercise treatment. The tumors also had distinct genomic, transcriptomic, and proteomic changes with exercise. These molecular changes pointed to perturbations in common biological pathways, including DNA repair. The integration of exercise-associated molecular alterations and growth phenotypes across multiple breast cancer subtypes provides further evidence that, indeed, the effect of exercise on cancer is context dependent, not only varying by cancer type, but also by subtype, adding another layer of complexity to the field of exercise oncology.
