Summary. In triple-negative breast cancer and high-grade serous ovarian cancer, the emergence of resistance to therapy is virtually inevitable and contributes to dismal long-term patient outcomes. The team will test the hypothesis that tumor ecosystems rapidly adapt to stress engendered by therapies, leading to the rapid emergence of resistance. As a corollary, blocking adaptive responses in tumor cells and the immune microenvironment will interdict the emergence of resistance. The objective is to monitor mechanisms underlying adaptive responses across temporal and spatial scales with single-cell precision, predict responses to untested combinatorial perturbations, and validate predicted drug combinations, fueling future clinical trials. An interactive team with diverse and complementary expertise and long collaboration history has been assembled: cancer and systems biology and therapeutics (Mills, contact PI, OHSU), computational biology/image analysis (Korkut, PI, MDACC; Goecks, OHSU), bioinformatics and systems biology (Liang, PI, MDACC), single-cell transcriptomics and proteomics (Mohammed, OHSU), molecular and anatomic pathology (Corless, OHSU; Sahin, MDACC), and ovarian and breast cancer translational research (Westin, MDACC; Mitri, OHSU). We will pursue two specific aims. Aim 1. Develop novel algorithms to create mechanistic maps of adaptive responses to therapeutic stress. The team will innovate algorithms to build data-driven and predictive models encompassing tumor cell signaling, microenvironment, and immune modulation. An extensive pre-existing longitudinal proteomics dataset of cell lines, xenografts, novel murine transplantable syngeneic models, PDXs, and patient samples will serve as the experimental data and constraints driving model construction. The modeling approaches will identify cellular vulnerabilities arising from adaptive responses to therapeutic stress and predict responses to untested combinatorial perturbations. The team will also determine whether therapeutic targeting “steers” proteomically heterogeneous tumors to a more therapeutically tractable homogenous state. For this purpose, we will use state- of-the-art multiplexed imaging-based proteomics technologies to formulate and implement data-driven models at spatial and single-cell precision. The single-cell, data-driven modeling will demonstrate how targeted therapies alter the tumor and immune microenvironment, leading to therapeutic vulnerabilities that new targeted therapy or immunotherapy combinations could exploit. Aim 2. Validate rational drug combinations targeting adaptive responses to therapy in relevant settings. The team will use cell lines, xenografts, PDXs, and novel murine transplantable syngeneic models to validate the therapeutic tractability of the rational drug combinations predicted by the data-driven models under Aim 1. Importantly, the experimental assessment will inform and improve the computational models through iterative data acquisition and subsequent remodeling. Novel therapy combinations will be assessed through clinical trials supported by other funds. The emerging principles and tools are highly applicable to other cancer lineages and could provide broad benefits.

Narrative Tumor cells and the microenvironment are rewired in response to targeted therapeutics, resulting in adaptive responses, including compensatory oncogenic pathway activation and changes in tumor ecosystem heterogeneity, leading to drug resistance and relapse. Blocking therapy-induced adaptive responses with rational drug combinations that are designed to interdict the emergence of resistance is predicted to yield more durable responses, and significantly improve patient benefits; however, the discovery of effective combinations is highly challenging due to the complexity and diversity of adaptive mechanisms within the tumor ecosystem. We will discover combination therapies that interdict adaptive responses in the tumor ecosystem through comprehensive drug response profiling from single cells and bulk tissues, data-driven mathematical modeling to predict responses to untested perturbations, subsequent in vitro/in vivo validation, and data-guided clinical trials.