Prognostic relevance of cancer-associated fibroblasts in human cancer.

Prognostication is an integral part of cancer diagnostic and helps oncologists to guide treatment decisions and therapy intensity. Accumulating evidence suggest that the stroma compartment also contains independent prognostic information, best exemplified by the impact of immune cells and cells of the vasculature on cancer progression. Similarly, strong experimental evidence exist that stromal fibroblasts, often designated cancer associated fibroblasts (CAFs), are actively involved in tumorigenesis. Thus, it can be anticipated that the molecular repertoire of CAFs is likewise important for the clinical behavior of the tumor. In this review we present recent studies addressing the prognostic impact of CAFs, with the focus on human lung and breast cancer. Several single markers have been suggested, either CAF specific or CAF derived, that in immunohistochemical studies have demonstrated independent association with survival. This includes members of the platelet derived growth factor receptor (PDGFR) family, CAF-markers like podoplanin and fibroblast activation protein (FAP) as well as transcription factors (FoxF1) and secreted factors (matrix metalloproteinases (MMPs), SPARC). However, most studies are based on explorative evaluations on single patient cohorts and require further validation. Using a more comprehensive approach, microarray studies have been employed to create gene expression signatures that detect an activated fibroblast state. These "stroma signatures" have been applied to identify specific CAF features associated with prognosis in several independent data sets of breast and lung cancer patients. Early studies in breast cancer have also demonstrated that fibroblast features influence therapy response. Thus, many strategies have been used to present encouraging proof-of-concept findings that CAFs could be exploited for prognostication. However, these studies also highlight the difficulties to conclusively define an "activated stroma" and to identify the individual factors involved in clinically relevant tumor-stroma interactions.