Online Program

Return to main conference page
Friday, September 14
Fri, Sep 14, 9:15 AM - 9:55 AM
Atrium
Poster Session

Patient-Driven Tumor Xenograft-Based Multigene Prediction Model of Response to Anti-Cancer Therapeutics in Cancer Patients (300725)

*Youngchul Kim, Moffitt Cancer Center 

Keywords: patient-driven tumor xenograft model, anti-cancer drug response, gene expression model

Cytotoxic chemotherapy and targeted therapy play a major role with surgery, radiotherapy, and immunotherapy in the treatment of cancer. Responses of cancer patients to drugs of those anticancer therapies vary because of the substantial heterogeneity in molecular characteristics of their tumors even with histologically same cancer type1. A successful personalized anticancer therapy will then greatly depend on predictive cancer biomarkers that can accurately select patients who will benefit from the anticancer drugs. Recently, patient-derived tumor xenograft (PDX) on which surgically derived patient’s tumor is implanted has been recognized to better inform therapeutic development strategies than the cancer cell lines. In this study, we developed a pipeline, so-called PDXGEM to construct a multi-gene expression model (GEM) for predicting response to anti-cancer therapeutic agents in cancer patients on the basis of data on gene expression and drug sensitivity from a pan-cancer PDX cohort. As a proof-of-cancer study, we applied the PDXGEM to build GEMs for an anti-cancer agent, paclitaxel, for breast cancer and for a targeted therapy, cetuximab, for colorectal cancer. For paclitaxel, PDXGEM built 66-genes based GEM with 13 breast cancer PDX models. The GEM was validated to predict pathological responses of a breast cancer cohort (n=127). Prediction scores of the cohort was significantly contrasted between patients who are responsive to paclitaxel-based chemotherapy and those not responsive validate the GEM (p=0.022). Next, PDXGEM for cetuximab resulted in 882-genes based GEM and its retrospective validation on 70 colorectal cancer patients yielded a significant difference in prediction scores between the responsive and the non-responsive (p=0.031). In conclusion, PDX-based drug response prediction models can be developed and validated successfully for certain cancer types, it will significantly improve overall response rates and therapeutic quality in patients with the