Neutralizing Anti-Tgf-Beta Antibodies Elicit Heterogeneous Therapeutic Responses In A Panel Of Murine Metastatic Breast Cancer Models

Overexpression of transforming growth factor-βs (TGF-βs) correlates with metastasis and poor prognosis in many advanced cancers, and TGF-βs have pro-oncogenic effects on nearly every cell type in the ecosystem of advanced tumors. Based on these observations and encouraging results in preclinical models, strategies to block TGF-β signaling are in early phase clinical oncology trials. To date however, preclinical studies supporting the development of anti-TGF-β therapeutics in cancer have focused around a few well-characterized mouse models which do not capture the heterogeneity of the human disease. To assess the generalizability of these findings within a given tumor type, we have assembled a panel of transplantable mouse models of metastatic breast cancer. Using this panel of models, we tested the efficacy of a pan-TGF-β neutralizing antibody (1D11), against the metastatic endpoint using fully immunocompetent mouse hosts to capture the contribution of anti-tumor immune responses. We observed therapeutic efficacy or trend to efficacy in 5 models (InhibMet), no effect in 2 models (NoEff) and an undesirable stimulation or trend to stimulation of metastasis in 4 models (StimMet). This heterogeneity in therapeutic responses suggests that it will be critical to develop good predictive biomarkers for patient selection in clinical trials using TGF-β antagonists. Plausible candidate biomarkers suggested by the existing literature, such as p53 mutation status, claudin-low status, or TGF-β protein expression, did not correlate with therapeutic response, so we applied integrated genomic discovery approaches to the panel. We find significant differences in patterns of gene expression, gene polymorphism/mutation and copy number variation between the StimMet and InhibMet models. Notably, transcriptomic analyses of the untreated primary tumors show that these segregate by response to therapy in principal component analysis, after removal of mouse strain and tumor origin (spontaneous vs genetically-engineered) as factors. Analysis of differentially-expressed genes suggests that InhibMet models are characterized by higher TGF-β pathway activation, higher angiogenesis, poor immune cell infiltration/activation, and other markers of tumor aggressiveness such as higher tumor cell proliferation and survival. The data point to fundamental differences in tumor biology between the two classes of model and suggest that it will be possible to generate biomarkers that predict therapeutic response to TGF-β pathway antagonists. Citation Format: Yu-an Yang, Kathleen C. Flanders, Binwu Tang, Miriam R. Anver, Anand Merchant, Howard Yang, Maxwell Lee, Scott Lonning, John M. McPherson, Lalage M. Wakefield. Neutralizing anti-TGF-β antibodies elicit heterogeneous therapeutic responses in a panel of murine metastatic breast cancer models. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4094. doi:10.1158/1538-7445.AM2015-4094