Historically, preclinical tumor therapy models in mice have frequently been deficient in predicting subsequent clinical activity; over-prediction of clinical anti-tumor efficacy is common. Several approaches are being made in an attempt to improve the clinical relevance of preclinical models, and include the use of genetically engineered mouse models (GEMMs) of cancer or patient derived xenografts (PDXs). Here we summarize, in the context of breast cancer, another approach, namely, the development of postsurgical models of either macroscopic or microscopic metastatic disease to mimic metastatic or adjuvant therapy. To do so we used in vivo selected metastatic variants of established human breast cancer cell lines such as MDA-MB-231. Testing antiangiogenic drugs such as the oral tyrosine kinase inhibitor (TKI) sunitinib alone or combined with chemotherapy in models involving treatment of established primary tumors invariably resulted in demonstrable anti-tumor activity VSports手机版. In contrast, identical treatments of postsurgical mice with advanced metastatic disease did not: survival times were not prolonged. This reflects multiple failed phase III trials of sunitinib based therapies in metastatic breast cancer patients. However, using a VEGF pathway targeting antibody drug instead of a TKI, with (paclitaxel) chemotherapy, resulted in anti-tumor activity in the metastatic setting, partially reflecting prior clinical results of the E2100 phase III trial of weekly paclitaxel plus bevacizumab. Other experiments involving postsurgical adjuvant treatment of early stage disease foreshadowed the phase III clinical trial failures of adjuvant bevacizumab in colorectal or breast cancer. In contrast, some investigational metronomic oral chemotherapy protocols alone or in combination with an antiangiogenic drug demonstrated potent activity in the advanced metastatic setting; these encouraging results have yet to be validated in randomized phase III clinical trials, which are underway based on some encouraging phase II clinical trial results. We have also observed circumstances where mice with advanced systemic disease, when successfully treated so as to prolong survival, sometimes relapse with brain metastases, reflecting a similar clinical phenomenon. Given our overall findings, we suggest that using preclinical mouse tumor models which mimic postsurgical adjuvant or metastatic therapy may be a promising strategy to help improve the ability to predict subsequent clinical outcomes. .