Abstract 350: Virtualization of drug testing by predictive systems biology modeling for optimal drug treatment of cancer cells and drug repositioning

Proceedings: AACR Annual Meeting 2014; April 5-9, 2014; San Diego, CA Traditionally, tumors have been classified based on their tissue of origin and their histopathological characteristics and consequently patients are divided into groups where they are treated identically despite substantial differences in genetic profiles. Identification of genes causally implicated in cancer development has led to the development of ‘stratified’ medicine, which adjusts a patient’s therapy based on biomarkers that identify some of the patients expected not respond to a specific therapy. However, many tumor genome sequencing projects have shown that many more gene mutations drive the development of cancer than previously thought, making every tumor unique. These results in low success rates and thereby high cost in drug approval as the identified drugs are only effective for particular patient groups while large patient cohorts have no clear clinical benefit. In the early process of drug development, drugs are therefore screened on large cancer cell line collections to define drug applicability and to determine potential tumor targets. Here, we report the results of a computational modeling platform, ModCell, allowing the prediction of individual drug effects using large-scale genomic and transcriptomic data to virtualize such cell line screening. We validated this approach two-fold: on publicly available data from the cancer cell line encyclopedia, that comprises pharmacological profiles for anti-cancer drugs across a library of cancer cell lines as well as on cancer cell line culture experiments. We provide evidence that ModCell is able to reproducibly predict the effects of individual drugs with a confounding 80% accuracy, to predict combinatory drug action and to identify new applications for existing drugs. Thus, computational modeling using ModCell can improve todays drug development by accelerating and partly replacing work which would have otherwise be conducted in the laboratory and in the clinic. Citation Format: Alexander Kuehn, Felix Dreher, Svetlana Peycheva, Reha Yildiriman, Verena Lehmann, Thomas Kessler, Christoph Wierling, Hans Lehrach, Bodo MH Lange. Virtualization of drug testing by predictive systems biology modeling for optimal drug treatment of cancer cells and drug repositioning. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 350. doi:10.1158/1538-7445.AM2014-350