PO-435 Proteomic classification of triple negative breast cancers

Abstract

IntroductionTriple negative (TN) breast cancer is an aggressive form of breast cancer for which no targeted treatment currently exists. The identification of potential therapeutic targets is hampered by the heterogeneity observed among TN breast cancer tumours. The aim of the present study is therefore to classify TN cancer. Earlier studies have classified TN breast cancers mainly based on transcriptomic and genomic data. Proteins play a major role in cellular functions and could give important clues for therapeutic development. The originality of this work was therefore1 the use of proteomic data for the classification and2 taking into account protein variability in the classification. This enabled us to study not only the protein expression, but also protein activation, by analysing protein phosphorylation.Material and methodsWe analysed two datasets of TN breast cancer samples that were both part of the Rational Therapy for Breast Cancer (RATHER) consortium (www.ratherproject.com). First, TN breast cancer tumour samples collected from patients in the Netherlands Cancer Institute (NKI), Amsterdam (n=64). Second, TN breast cancer tumour samples collected from patients at the Addenbrooke’s Hospital, Cambridge, UK (n=31). Protein expression was measured for 116 proteins using Reverse Phase Protein Arrays. The NKI dataset served as the training set and the Cambridge dataset served as the validation set. Unsupervised classification methods were applied to the training set and differential analyses for the identified groups were applied at protein, transcriptome and clinical levels.Results and discussionsTwo TN groups were identified in the training set. Each group was characterised by a specific set of proteins. The groups were robust to protein subsampling, i.e. they did not change if a part of the proteins were removed from the analysis. Importantly, the same two groups were found in the validation set, confirming the biological relevance of the two groups. Eighteen proteins were found to be crucial for identifying the two groups. Among these 18 proteins, eight were phosphorylated (i.e. 45%, compared to 32% in the original dataset).ConclusionOur results stress out the importance of including proteomic data in TN breast cancer classifications. In order to better understand the underlying molecular mechanism in TN breast cancer, we now plan to study the relations between genomic, transcriptomic and proteomic data for the two identified TN groups.