Prediction of Drug Efficacy in Colon Cancer Preclinical Models Using a Novel Ranking Method of Gene Expression

The presence of stromal cells in tumors is altering the significance of molecular profiling when using standard methods of gene expression quantification. We developed a novel normalization method to rank target gene expression in tumor samples by comparisons with reference samples representing the different cell types found in a tumor. The score for each target gene obtained after normalization, is aimed to be predictive of targeted therapies efficiency. We performed this qPCR analysis on human colorectal cancers to demonstrate the importance of reference samples to obtain accurate data and on a collection of patient-derived xenografted (PDX) colon tumors treated with Cetuximab (anti-EGFR) to demonstrate that the calculated EGFR score is predictive of Cetuximab efficacy. Interestingly, the score allowed to select an efficient treatment in a PDX model refractory to standard of care. This method is opening a novel way to predict targeted therapy efficiency which could be extended to several tumor types, and to unlimited target genes.