Mouse models in oncogenesis and cancer therapy

Animal models have been critical in the study of the molecular mechanisms of cancer and in the development of new antitumor agents; nevertheless, there is still much room for improvement. The relevance of each particular model depends on how close it replicates the histology, physiological effects, biochemical pathways and metastatic pattern observed in the same human tumor type. Metastases are especially important because they are the main determinants of the clinical course of the disease and patient survival, and are the target of systemic therapy. The generation of clinically relevant models using the mouse requires their humanization, since differences exist in transformation and oncogenesis between human and mouse. Although genetically modified (GM) mice have been instrumental in understanding the molecular mechanisms involved in tumor initiation, they have been less successful in replicating advanced cancer. Moreover, a particular genetic alteration frequently leads to different tumor types in human and mouse and to lower metastastatic rates in GM mice than in humans. These findings question the capacity of current GM mouse carcinoma models to predict clinical response to therapy. On the other hand, orthotopic (ORT) xenografts of human tumors, or tumor cell lines, in nude mice reproduced the histology and metastatic pattern of most human tumors at advanced stage. Using ex vivo genetic manipulation of human tumor cells, ORT models can be used to molecularly dissect the metastatic process and to evaluate in vivo tumor response to therapy, using non-invasive procedures. Nevertheless, this approach is not useful in the study of the initial stages of tumorigenesis or the contribution of the immune system in this process. Despite ORT models are more promising than the most commonly used subcutaneous xenografts in preclinical drug development, their capacity to predict clinical response to antitumor agents remains to be studied. Humanizing mouse models of cancer will most likely require the combined use of currently available methodologies. © FESEO 2006.