Abstract The role of EMT in cancer has been well reported and has been shown to prime cells for invasion and metastasis. EMT can be adopted or reversed (i.e. mesenchymal to epithelial transition, MET) by cells, revealing plasticity that can also lead to stemness and drug resistance. Although it is appreciated that EMT is not a binary process of two extremes but instead a continuum of intermediate states of partial EMT phenotypes, these are poorly defined. Given that intermediate EMT cancer states are viewed as critical for understanding and clinically targeting EMT processes, our aim was to dynamically capture and characterize intermediate EMT states in TGF beta treated lung cancer cells and clinical specimens. With single cell analysis we identified 4 distinct transition states. These states patterned an EMT axis featuring: (1) an epithelial, (2) a partial EMT and (3) a mesenchymal state that branches off to (4) a subset of phenotypically stem-like cells. Transition was reflected by gradual changes in E Cadherin, Vimentin, CD44 and CD24 levels. “True” mesenchymal state (E Cad- Vim+) was isolated to the stem-like cells (CD44hiCD24lo), which were negative for Twist protein expression. To interrogate the dynamism of EMT and MET processes, we performed TGF beta withdrawal experiments, which showed that most cells were able to reverse their behavior. Simultaneous analysis of 30 parameters with CyTOF confirmed the 4-state transition, offering deep dynamic views of a variety of cell surface, signaling, cell cycle, and transcriptional markers. We then proceeded to computationally identify and define additional states of the EMT/MET spectrum that are visited by transitioning cells in a step-wise manner, in order to construct a lung cancer EMT/MET proteomic landscape. To tackle this, we applied CCAST, an algorithm that employs decision trees to identify and discretize homogeneous cell subpopulations among heterogeneous single cell data. When we assembled the high dimensional data in a timely order, we were able to visualize the emerging states that cells visited during their transitions. This revealed the existence of more than one possible EMT/MET trajectories as well as the existence of a transient subpopulation of cells with a distinct phenotype (Twist+, CD34+, E Cad-, Cytokeratin 7+, pEGFR-). Finally, CyTOF analysis and projection of 2 lung adenocarcinoma specimens on the constructed EMT/MET landscape confirmed the existence of states observed in our cell line studies, including the aforementioned Twist+ transient subpopulation. In summary, we provide a lung cancer cell proteomic map that dissects EMT phenotypic plasticity. Clinically, this type of EMT/MET proteomic mapping could help identify, predict and target EMT mechanisms known to have a role in cancer progression, drug resistance and disease recurrence. Citation Format: Loukia G. Karacosta, Benedict Anchang, Samuel Kimmey, Matt van de Rijn, Joseph B. Shrager, Sean C. Bendall, Sylvia K. Plevritis. Identifying dynamic EMT states and constructing a proteomic EMT landscape of lung cancer using single cell multidimensional analysis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4997.