Obi L. Griffith1, Anneleen Daemen1, Steffen Durinck1, Laura M. Heiser1, Francois Pepin1, Lakshmi R. Jakkula1, Nicholas J. Wang1, James Korkola1, Heidi Feiler1, Gordon Mills2, Paul T. Spellman1, Joe W. Gray1.
1. Life Sciences Division, LBNL, Berkeley, California, USA.
2. Dept. of Systems Biology, MD Anderson Cancer Center, Houston Texas, USA
The clinical and genomic heterogeneity of breast cancer necessitates development of personalized therapeutic strategies. Hundreds of new candidate agents are under investigation for use in well-defined patient subpopulations. Cell lines mirror many of the molecular characteristics of the tumors from which they were derived, and are thus a good preclinical model for the study of drug response in cancer. We hypothesize that correlating the responses of a panel of breast cancer cell lines to FDA approved and investigational therapies with their molecular characteristics will reveal biomarkers that can be used to guide clinical trials in terms of patient stratification. A collection of 72 breast cancer cell lines was assembled representing all known molecular subtypes. Cell lines were tested for response to 111 therapeutic compounds by growth inhibition assays. In addition, nine molecular profiling datasets were collected for the cell line panel assessing copy number, expression, transcriptome sequence (RNA-seq), exome sequence, methylation, protein abundance, and mutation status. Classification signatures for drug response were developed using the Random Forests approach. In order to determine the importance of the different molecular data sets, classifiers were built on the data sets separately as well as on the combined data. All data types resulted in optimal response prediction for at least some specific drug compounds. Around one third of drug compounds showed a transcriptional subtype-specific response. For a significant fraction of drugs, the ROC curve (AUC) obtained with molecular data increased by at least 0.1 compared to the predictive power of breast cancer subtype alone.
This work was supported by the Director, Office of Science, Office of Biological & Environmental Research, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231; by NCI grants P50 CA058207; U54 CA112970; U24 CA126477; NHGRI U24 CA126551, and by SU2C-AACR-DT0409.
Back to Mini-Workshop: Cancer Genomics