Abstract
Abstract Many anti-cancer drugs act by directly binding or modifying DNA, or through indirect binding during processes that rely on the chromatin architecture. Mis-regulation of the expression of epigenetic factors may perturb chromatin organization and contribute to drug resistance mechanisms. Here, we aim to investigate the power of chromatin regulators and epigenetic factors as predictive markers for the efficacy of single chemotherapy drugs. We exploit transcriptome data from breast cancer cell lines and a validation set comprised of breast cancer patient derived xenograft (PDX) models that are either sensitive or resistant to individual chemotherapy agents, to correlate mRNA expression of chromatin regulators to drug efficacy. We then use Random Forests, a non-parametric machine learning method, to model the cell lines’ response to different compounds. The model is then used to predict the patients’ response to each drug, identifying which genes contribute the most to the prediction. We then compare our predictive gene signature to commercially available prognostic kits. Random Forest analysis revealed a 28-gene signature (termed EPOCH28) containing several histone chaperones, histone variants, histone modifying enzymes, among other chromatin regulators, which is highly predictive of drug efficacy. Indeed, our model accurately predicts 11 of 14 resistant PDX models and 15 out of 17 sensitive PDX models (∼ 84%), with an average confidence level of over 73%. Importantly, we find that the EPOCH28 gene signature is specific, and is thus not a general prognostic signature for patients who will benefit from any chemotherapy. In line with this, the EPOCH28 outperforms the breast cancer prognostic gene signatures of Mammaprint and Oncotype DX, suggesting that while these commercial tests may help guide clinical decisions, it will be critical to consider additional factors that can help identify which drug will be most beneficial to an individual patient. The integration of chromatin regulators as clinical biomarkers, in particular in the context of predictive markers for the response to a single drug, will help guide clinical decisions and treatment options for breast cancer. Importantly, our approach is general and thus can be applied to other chemotherapy drugs, including those in clinical trials to develop companion diagnostics, and for other cancer types. Citation Format: Zachary A. Gurard-Levin, Vera Pancaldi, Laurence OW Wilson, Elisabetta Marangoni, Sergio Roman-Roman, Alfonso Valencia, Paul Cottu, Genevieve Almouzni. Epigenetic profiling of chemotherapy sensitivity. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr LB-155. doi:10.1158/1538-7445.AM2015-LB-155
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have