Project Summary/Abstract Patients with seemingly identical pancreatic neuroendocrine tumors (PanNETs) often differ in their clinical outcomes (emergence of metastases, treatment response, survival, etc.), suggesting that these malignancies may in fact be of different subtypes that are currently unknown and indistinguishable by standard diagnostics (e.g., histology or genomics). PanNETs have been difficult to further classify or risk-stratify by traditional genomic or transcriptomic methods alone. Currently, follow-up monitoring after surgical resection is mostly based on radiographic imaging alone. There are no curative therapies once metastases occur, no protein markers of early detection and metastatic risk, and no established molecular means of surveillance. We hypothesize that PanNETs can be better classified according to proteome signatures. We propose to uncover new proteome-based subtypes by deep proteomic analysis that better define these tumors. In preliminary studies, we have shown that proteomic profiling can distinguish and sub-classify various tumors and identify protein signatures characteristic of primary or metastatic lesions. In this proposal, we will first elucidate the deep proteomes of well differentiated PanNETs of histologic grades ranging from G1 to G3. We will study tissues from two clinical outcome cohorts, a “low risk” group (tumors did not show metastases for at least 5 years after surgery) and a “high risk” group (tumors developed subsequent metastases, but are otherwise indistinguishable by current diagnostic means). By examining both primary and metastatic lesions, we will identify protein markers that differentiate these two outcomes and signatures that distinguish primary from metastatic lesions. Spatial heterogeneity within and between tumors, neopeptide/neoprotein markers, and phosphoproteomic signaling pathways will also be examined. Extensive proteomic and integrated proteogenomic analyses will be performed to define proteome-based subtypes within PanNETs and between primary and metastatic lesions. We will validate marker panels in independent cohorts by immunohistochemistry and correlate with clinical outcomes. Based on the protein-panel signatures, we will be able to develop risk stratification models that predict metastatic propensity of PanNETs. Our study will have significant impact on understanding pancreatic neuroendocrine tumors. The likelihood of success of this proposal is high, as we have already discovered new cancer subtypes in our preliminary studies. The project will benefit from the strong scientific and clinical expertise of the project team and the high volume of rare pancreatic neoplasms treated at MSKCC. Our envisioned proteome-based risk stratification may explain the clinical conundrum of why patients with currently seemingly similar PanNETs exhibit strikingly different metastatic propensity, treatment response, and length of survival. New proteomic subtyping may inform future therapy deve...