PDXGEM: Patient-Derived Tumor Xenograft based Gene Expression Model for Predicting Clinical Response to Anticancer Therapy in Cancer Patients

Background Cancer is a highly heterogeneous disease and shows varying responses to anti-cancer drugs. Although several approaches have been attempted to predict the drug responses by analyzing molecular profiling data of tumors from preclinical cancer models or cancer patients, there is still a great need of developing highly accurate prediction models of response to the anti-cancer drugs for clinical applications toward personalized medicine. Here, we present PDXGEM pipeline to build a predictive gene expression model (GEM) for cancer patients’ drug responses on the basis of data on gene expression and drug activity in patient-derived xenograft (PDX) models. Results Drug sensitivity biomarkers were identified by an association analysis between gene expression levels and post-treatment tumor volume changes in PDX models. Only biomarkers with concordant co-expression patterns between the PDX and cancer patient tumors were used in random-forest algorithm to build a drug response prediction model, so called PDXGEM. We applied PDXGEM to several cytotoxic chemotherapy as well as targeted therapy agents that are used to treat breast cancer, pancreatic cancer, colorectal cancer, or non-small cell lung cancer. Significantly accurate predictions of PDXGEM for pathological and survival outcomes were observed through extensive validation analyses of multiple independent cancer patient datasets obtained from retrospective observational study and prospective clinical trials. Conclusion Our results demonstrated a strong potential of utilizing molecular profiles and drug activity data of PDX tumors for developing a clinically translatable predictive cancer biomarkers for cancer patients. PDXGEM web application is publicly available at http://pdxgem.moffitt.org .