Of mice and (wo)men: is this any way to test a new drug?

Mouse models are undeniably vital to our understanding of the molecular basis and pathogenesis of breast cancer. In addition, they provide valuable information for the development of novel antineoplastic agents with regard to basic pharmacokinetics, pharmacogenomics, and toxicity analysis. Their role in evaluating cancer drug efficacy, however, has been limited by inconsistencies in the translation of basic science into effective clinical treatments. Ideally, mimicking the intricacies of human tumorigenesis and metastasis in model systems would simplify the development of targeted therapeutics; however, modeling such interplay between the tumor and its surrounding stroma, matrix proteins, immune cells, endothelia, and lymphatics in mice is a seemingly insurmountable task. Xenograft models employ human tumor cells isolated from metastatic deposits that may have been passaged in vitro hundreds of times before injection into an orthotopic site of an immunodeficient mouse host. Although some of the original molecular and cellular pathways of the tumor cells are preserved in cell lines, the macroand microenvironment surrounding such tumors are artificial, and the true heterogeneity within the tumor is likely compromised. Their value in the historical development of cytotoxic chemotherapeutic agents, however, is supported by an extensive retrospective analysis from the National Cancer Institute. This study compared drug activity between xenograft model systems and phase II human participants. It demonstrated that clinical activity was absent in all six chemotherapy drugs in which xenograft activity was observed in less than one third of the xenografts tested, whereas 45% of the drugs that were active in more than one third of the xenografts were also active in clinical subjects. Although this is the best evidence to support the correlation between drug efficacy in preclinical xenograft mouse and early phase clinical studies in humans, it is tempered by the fact that only 17% of all cancer therapies that are evaluated in human studies will survive phase II evaluation. More recently, genetically engineered mouse models have become a promising alternative to traditional xenograft models in that they provide in situ tumor development in an immunocompetent setting. Their major limitation is their inability to replicate advanced cancer states and metastases—characteristics shared by participants in early-phase human clinical trials. Thus, their role in drug development yet remains to be defined. This can be particularly troublesome for monoclonal antibodies directed against human targets. Unless there is substantial cross-reactivity between mouse and human epitopes, host/antibody interactions cannot be modeled. Orthotopic transplantation of human tumor xenografts, on the other hand, promotes the establishment of high tumor volume and distant metastases. Although this seems preferable to other assessable systems, the technology is labor intensive, expensive, requires small-animal imaging to monitor response to therapy, and may not be easily reproducible. Although no one mouse model system can perfectly simulate human cancer biology, the ability of scientists to humanize these models will likely require a combination of such methodologies. As contemporary drug development takes aim at specific molecular targets, compared with their cytotoxic chemotherapeutic predecessors, it is fair to question whether mouse models are even appropriate for evaluating their efficacy. Despite such flaws, imperfect mouse models have provided important insights into breast cancer therapy. Cultured human estrogen receptor (ER)-positive breast cancer MCF-7 cells were found to be estrogen-dependent in vitro and in xenograft models. This model system was important in discovering the mechanisms of action and in vivo effects of antiestrogen therapy. Indeed, early studies of tamoxifen demonstrated its capacity to induce regression of ER-positive human breast cancer cell lines in nude mice. Such growth inhibition, however, was not observed in the ER-negative cell lines. These results were consistent with those of subsequent human clinical trials in which women with advanced, ER-positive, metastatic disease were found to have a significantly higher response rate to tamoxifen than did those with ER-negative tumors. Similar results were also observed in the adjuvant setting, and it is clear that tamoxifen significantly reduces the odds of recurrence and death resulting from breast cancer in women with ER-positive, but not ER-negative, disease. Studies in mice have also been crucial to the development of targeted cancer therapies against the epidermal growth factor receptor (EGFR) family, including HER-2 (human epidermal growth factor receptor 2). As the role of HER-2/neu in the pathogenesis and progression of breast cancer was defined, inhibitors against the surface membrane receptor were generated in mouse models. The 4D5 murine monoclonal antibody was shown to have specific dose-dependent antiproliferative effects in human tumor cell lines that overexpress HER-2. This led to the development of trastuzumab, a recombinant, humanized murine anti–HER-2 antibody that has demonstrated significant clinical benefit in the treatment of women with HER-2– overexpressing metastatic breast cancer and for adjuvant therapy in women with resected early-stage disease. Both ER and HER-2 have proven to be important prognostically and as biomarkers for JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 26 NUMBER 6 FEBRUARY 2