Hepatocyte growth factor inhibition: a novel therapeutic approach in pancreatic cancer

Pancreatic stellate cells (PSCs, which produce the stroma of pancreatic cancer (PC)) interact with cancer cells to facilitate PC growth. A candidate growth factor pathway that may mediate this interaction is the HGF-c-MET pathway.

Our studies suggest that targeted therapy to inhibit stromal-tumour interactions mediated by the HGF-c-MET pathway may represent a novel therapeutic approach in PC that will require careful modelling for optimal integration with existing treatment modalities.

Effects of HGF inhibition (using a neutralising antibody AMG102) alone or in combination with gemcitabine were assessed (i) in vivo using an orthotopic model of PC, and (ii) in vitro using cultured PC cells (AsPC-1) and human PSCs.

We have shown that human PSCs (hPSCs) secrete HGF but do not express the receptor c-MET, which is present predominantly on cancer cells. HGF inhibition was as effective as standard chemotherapy in inhibiting local tumour growth but was significantly more effective than gemcitabine in reducing tumour angiogenesis and metastasis. HGF inhibition has resulted in reduced metastasis; however, interestingly this antimetastatic effect was lost when combined with gemcitabine. This suggests that gemcitabine treatment selects out a subpopulation of cancer cells with increased epithelial-mesenchymal transition (EMT) and stem-cell characteristics, as supported by our findings of increased expression of EMT and stem-cell markers in tumour sections from our animal model. In vitro studies showed that hPSC secretions induced proliferation and migration, but inhibited apoptosis, of cancer cells. These effects were countered by pretreatment of hPSC secretions with a HGF-neutralising antibody but not by gemcitabine, indicating a key role for HGF in PSC-PC interactions. Conclusions: Our studies suggest that targeted therapy to inhibit stromal-tumour interactions mediated by the HGF-c-MET pathway may represent a novel therapeutic approach in PC that will require careful modelling for optimal integration with existing treatment modalities.