Abstract BACKGROUND: Triple-negative breast cancer (TNBC) is a breast cancer subtype that is not defined by targetable molecular markers. Since all the aberrations that exist and ultimately contribute to the tumor phenotype converge, from a functional point of view, in the final status of the phosphorylation of the proteome in a given moment, we sought to interrogate the phosphoproteome with two aims: 1) establish a taxonomy of TNBC based on measurable markers that predict clinical course; 2) reduce the phosphoproteome that characterizes the bad- from good-prognosis cases to its targetable, driving kinases, in order to define rational therapeutic approaches in TNBC. We chose mass spectrometry as a discovery approach, because of its coverage, sensitivity, dynamic range and specificity. Here we report the part of phosphoproteomic analysis used for establishing a new taxonomy of TNBC. METHODS: We performed quantitative phosphoproteomics from a discovery set of 34 frozen tumor samples divided in two sets paired by classic prognosis factors: 1, with 13 patients relapsed in less than 3 years; 2, 21 patients, no relapse in 10 years follow up. Raw data were processed with Maxquant and cases clustered using a supervised hierarchical approach. Kinases (predicted with linear domain consensus analysis tools using a PHOSIDA) and phospho-sites discriminating the set 1 from 2 were validated by immunohistochemistry in 113 consecutive TNBC cases, spotted in TMAs, with 14 years follow up. RESULTS: Overall, 15000 unique phosphopeptides were identified. Supervised clustering of subset 1 vs. 2 showed that 161 and 541 peptides were significantly more phosphorylated in the subset 1 and 2, respectively (FDR<0.15). Consensus domain analysis predicted kinases CDK1,2,4,6, CLK1, DAPK3/ZIPK, PAK2, RSK1, p70S6K, AKT, PKCEpsilon and PIM1 were driving the profile of subset 1. Those kinases, plus 9 phosphosites against which antibodies were available, were interrogated in the validation set. Kinase validation was performed directing an antibody against the active kinase form, if available; if not, against the total levels. The staining of kinases and the candidate phosphosites was analyzed with an Ariol, scored with a continue value from 0 to 3, and divided in quartiles for each staining probe. So far, 12 probes have been quantified. The prognosis of patients with upper quartile staining vs. the remainder was analyzed with the log-rank test and cox proportionate hazards model for each probe. Five of them showed statistically significant association with relapse (pSTAT3Tyr705, 6.1 vs. 10.1 years, p = 0.024; CYCB1, 7 vs. 10.3 years, p = 0.027; CDK6, 7 vs. 10.3 years, p = 0.013; pp70S6KThr389, 7.2 vs. 10 years, p = 0.042 and PRKCEpsilon, 7.1 vs. 10.2 years, p = 0.021). A combined variable integrating upper-quartile staining from either of those 5 probes occurring in 67.4% patients, identified almost all patients that experienced relapse: 48.3% of them relapsed (median time 7.6 years) vs. the reminder 32.6% of which only 13.8% relapsed (median time 11.9 years), log rank p = 0.001 INTERPRETATION: High-throughput mass-spectrometry is a powerful tool for generating a disease taxonomy, and can be translated to routine techniques like IHC. The signature constituted by pSTAT3Tyr705, CYCB1, CDK6, pp70S6K and PRKCEpsilon constitutes the most parsimonious signature to date able to detect all the early TNBC patients that relapse in the long term. Citation Format: Ivana Zagorac, Tamara Mondejar Tevar, Jesús Sánchez Ruíz, Albert Heck, Maarten Altelaar, Renske Penning, Harm Post, Gonzalo Gómez López, David Pisano, Javier Muñoz Peralta, Javier Muñoz Peralta, Manuel Morente, Luis Manso, Miguel Quintela-Fandiño. A phosphoproteomic portrait of triple - negative breast cancer: functional taxonomy. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr B9.