Abstract/Project Summary While prognosis for early stage breast cancer (BC) has improved dramatically, 20-30% of patients recur at distant sites and ultimately succumb to their disease. To date, the spatial and temporal patterns of BC relapse have been difficult to predict. Moreover, when and how metastatic potential is determined is largely unknown. To address these questions, we have developed spatial computational and mathematical models of tumor progression to infer the ‘time’ of metastatic seeding. Application of these techniques to paired primary breast cancers with matched metastases, yielded quantitative evidence for early metastatic seeding, often 2- 4 years before the primary lesion is detectable (Nat Gen 2020), consistent with experimental and clinical data indicating that BC cells can disseminate early and persist. In parallel, we defined the rates and routes of relapse across 11 Integrative Clusters (ICs) in an analysis of 2,000 early-stage BCs with long-term follow-up (METABRIC cohort: Nature 2019; 2012). These include two triple negative BC subgroups with distinct relapse trajectories and four ER+/HER2- ICs (1, 2, 6 and 9) with high and persistent risk of relapse up to 20 years after diagnosis. Collectively, these high-risk subgroups account for 26% of all ER+/HER2- tumors and the majority of BC relapses. The pattern of copy number amplification (CNA) and overexpression in these high-risk ICs echoes that seen for HER2+ BC, each harboring druggable clonal genomic drivers. This breakthrough discovery led to a biomarker-driven clinical trial evaluating new targeted therapies in early-stage high-risk BC patients. However, the definitive drivers and mechanisms of progression in these subgroups have yet to be characterized. Moreover, how the local tissue microenvironment (TME) varies across the ICs and contributes to immune suppression, dissemination, dormancy and relapse is unknown We hypothesize that the oncogenic drivers of the high-risk ICs drive tumor progression and relapse by dictating immune contexture and remodeling the cell surface glycoproteome, potentiating tolerogenic cell states– as further functionally evaluated in Projects 2 and 3 in collaboration with Michael Angelo and Michael Bassik. Additionally, we hypothesize that these amplicons confer intrinsic endocrine resistance, necessitating new therapeutic strategies. We test these hypotheses in the following Specific Aims: Aim 1- Characterizes the TME in longitudinal BC cohorts and evaluate the association between IC, glycosyltransferase expression, response to therapy and relapse. Aim 2 - Quantifies the dynamics of BC relapse and the response to clinical therapies across the ICs. Aim 3 - Measures clonal dynamics and identify mechanisms of resistance to targeted and endocrine therapies in high-risk ER+ BC patient-derived organoids from primary and metastatic lesions.