Patient xenograft models: tool for preclinical phase II studies?

Jens Hoffmann
EPO – Experimental Oncology & Pharmacology Berlin-Buch GmbH
Robert-Rössle-Str. 10
13125 Berlin, Germany


Patient derived xenografts (PDX) are preclinical models for which clinical relevance has been described in many preclinical experiments. Several drug sensitivity screenings revealed an individual response to standard of care treatments comparable to historical data from clinical trials. Recently, preclinical phase II studies with up to 100 PDX models have demonstrated a strong correlation between tumor biology and treatment response.

Head and neck squamous cell cancer (HNSCC) or colon cancer are tumor types where a targeted therapy with cetuximab has been successfully introduced. Nevertheless, frequently observed treatment resistance caused by inherent genetic and cellular heterogeneity of advanced cancer calls for more intensified efforts to individualize treatment. 

As it is now generally accepted that tailored treatment approaches, including specific schedules or combinations, can improve the therapeutic outcome for target-specific compounds and chemotherapy, cancer care needs, next to “established treatment guidelines”, support from integrative translational research. The implementation of such approaches, starting already during diagnosis and surgery can help to establish a rationale for an individual treatment out from the numerous possible options.

Drug response in clinical trials with advanced HNSCC or colon cancer patients can frequently not be correlated with mutations, gene expression or other molecular markers as tissue samples are hardly repeatedly available. The development of personalized response predictors therefore still depend on the availability of surrogate models like PDX. Comprehensive studies that correlate pharmacodynamic data from PDX with systematic molecular tumor tissue characterization have improved our understanding of the disease and helped to design diagnostic procedures allowing optimal therapy for individual patients. These wet lab data together with molecular signatures are used for in silico modeling to identify new predictors for tailored therapies.

While PDX have demonstrated their value for identification and validation of predictive signatures, a benefit for individual treatment prediction is currently under evaluation. A larger implementation of PDX studies in personal therapy planning is still restricted by several factors – the time for development of such models, high costs, and the limited take rates which are especially for breast cancer still challenging. However humanization of the mouse models is currently under evaluation to improve take rates and help to overcome some of the limitations. As we found that engraftment rates of PDX tumors were strongly correlated with advanced stage of the patient tumor, these models could especially be of value for improving and personalizing treatment of advanced disease. Furthermore PDX provide an exceptionally broad basis for translational research and the development of targeted therapies in the future.