Solid tumors progress towards metastasis as their cells diversify in space and genotype. Unfortunately, by the time nearly all solid tumors are discoverable, the dynamics governing this diversification are obscured. To overcome this limitation, I will develop methodology that enables both reconstruction of key diversification dynamics throughout a tumor’s past and robustly predicts which current tumor cellular lineages, if unmitigated, will permit tumor progression towards poor clinical outcomes; i.e., a tumor ‘time machine’. In this proposal, I describe how I will construct such a time machine from phylodynamic methodologies, which synthesize phylogenetic and population dynamic models and were originally developed for the study of viral evolution and geographic spread. To make such methods suitable for the analysis of tumor sequencing data, I will extend these models to incorporate multiple attributes of tumor growth dynamics. I will apply these new tools at scale to analyze cancer evolutionary processes, and develop novel diagnostic tools to yield clinically actionable insights into tumor progression. Crucially, application of these methods relies on cancer single cell DNA sequencing data, which has recently gained prominence and reveals a much more detailed view of cancer evolution than previous tumor sequencing technologies. I hypothesize that these data will yield still greater insight when subjected to the rigorous, quantitative examinations enabled by the cancer-calibrated phylodynamic methodology I will develop. In applying these new tools, I will derive precisely-calibrated clinical timelines, including estimates of tumor initiation and the age of important clonal subpopulations, which will help us understand personalized tumor progression. I will disentangle heterogeneous cancer growth rates from within a single tumor, enabling both the prediction of the clinically most important variants and a new approach for discovering driver mutations. These phylodynamic methods will also illuminate how and when cancer cells disperse within a primary tumor as a function of genotype and/or environment, enabling us to chart the development of metastatic lineages. In addition to answering fundamental questions about the driving forces shaping tumorigenesis and cancer progression, this proposal aims to provide a powerful new set of methodologies that will drive a paradigm shift in the analysis of tumor single cell sequencing data in diverse basic science, clinical and translational cancer settings.

Project Narrative: Solid tumors progress towards metastasis as their cells diversify in space and genotype, but by the time nearly all solid tumors are discoverable, the dynamics governing this diversification are obscured. To overcome this limitation, I will develop methodology to construct a tumor ‘time machine,’ which enables both the reconstruction of evolutionary dynamics throughout a tumor’s past and robustly predicts which parts of a tumor will most rapidly progress in the future.