Nature and Nurture: It Takes a [Dysfunctional] Village to Raise a Tumor

There is a growing appreciation in the field of cancer research of the importance of the microenvironment to influence tumor growth. In the early 1970s, Judah Folkman put forth the concept that because tumors are dependent on angiogenesis for growth, targeting the vessels in the surrounding tumor microenvironment may prevent tumor progression (Folkman, 1972). As Folkman described, “the interplay between tumor cells and endothelial cells constitutes a highly integrated ecosystem,” and this concept of the tumor “ecosystem” has now been extended beyond the vasculature to include many diverse facets of the microenvironment, such as the immune system and the extracellular matrix, including most recently the concept of the cancer stem cell niche (Sneddon and Werb, 2007). Underlying this interest in tumor microenvironment is the idea that tumor cells under certain conditions cannot grow or survive in the absence of microenvironmental support. Because the microenvironment can be co-opted to the tumor without the accumulation of genetic changes, it may be easier to target with chemotherapy without developing resistance. In the paper “Nf1-dependent tumors require a microenvironment containing Nf1+/− and c-kit dependent bone marrow” in this issue of Cell, groups from Indiana University and the University of Texas Southwestern present an elegant use of genetically engineered mouse models to carefully dissect the role of inflammation in the tumor microenvironment. Based on their findings, they were able to treat a young patient with life-threatening complications of neurofibromatosis type 1 in an exciting bench to bedside success story.

It has been shown previously that the formation of neurofibromas requires loss of Nf1 in Schwann cells of the peripheral nerve and heterozygosity for Nf1 in surrounding stroma (Zhu et al., 2002). It has long been observed that neurofibromas are often infiltrated by mast cells (Riccardi, 1990), and these authors have further shown signaling networks between Nf1 mutant Schwann cells and Nf1 heterozygous mast cells (Yang et al., 2003). They show here in this issue of Cell that Nf1 heterozygous bone marrow derived cells hone to developing neurofibromas in vivo and are required for neurofibroma growth, implicating c-kit-dependent mast cells as mediators of tumorigenesis. Based on the mast cell dependence on c-kit, the authors found that imatinib mesylate was able to shrink a neurofibroma in a critically ill three year-old patient. As the authors note, the therapy has the potential to target both the tumor cells and the microenvironment in this case, and subsequent clinical trials will determine how robust and stable the response to this therapy is.

It is interesting to note that the microenvironment is not “normal” in the case of this study. Wild-type mast cells do not support tumor growth. Only Nf1 heterozygous mast cells that have been shown to be hypersensitive to migratory stimulus can support tumor growth. Other studies have also demonstrated that tumor stroma is different than normal stroma, for example, in the case of vasculature in colorectal cancer (St Croix et al., 2000) or in the case of fibroblasts in basal cell carcinoma (Sneddon et al., 2006). However, it is not clear whether these stromal changes are the cause or result of tumor formation. The hypotheses regarding cancer evolution have themselves evolved from Nordling’s proposal that cancer required the accumulation of mutations over the decades of life (Nordling, 1953), to Knudson’s hypothesis that a second-hit in a tumor suppressor gene could explain the accelerated tumorigenesis in familial cancer syndromes (Knudson, 1971), to Fearon and Vogelstein’s model for the accumulation of oncogenic and tumor suppressive mutations in cancer (Fearon and Vogelstein, 1990), to Hanahan and Weinberg’s hypothesis of the six mutant traits that cancer cells must acquire for tumors to form (Hanahan and Weinberg, 2000). As cancer research moves away from what Hanahan and Weinberg described as “a reductionist focus,” there is a growing appreciation that the clonally evolving tumor is not the only cell type acquiring changes during the course of the individual’s life span, for example due to cellular senescence or environmentally induced epigenetic changes, and that these age-induced changes have the potential to alter the microenvironment to create more favorable conditions for tumor growth. This leads to a model of parallel tumor-causing changes, rather than a strict clonal evolution of the tumor cell (Figure 1). Clearly in the case of the study by Yang et al in this issue of Cell, the changes in the microenvironment were engineered in mice and are preexisting in neurofibromatosis patients; however, it will be interesting in the future to examine whether stromal changes are causal in sporadic tumorigenesis.

Figure 1.

These concepts have implications for developing therapeutic approaches beyond the elegant use of imatinib mesylate to target stromal cells in this study. Tumors co-evolve with their stroma within the individual, and the state of the stroma may be as important a factor in therapy development as the state of the tumor cells themselves. The current use of xenograft models to develop cancer therapy has not been highly predictive of efficacy in patients. This may be due in part to the fact that the stromal component of these models in naïve and has not undergone the same selective pressure as the tumor stroma in the patients. In effect, the xenograft models may only be addressing half of the cancer problem, targeting cancer cells that are out of their element and therefore already weakened in some way. Further study and better understanding of the tumor microenvironment and its relationship to normal stroma may help to develop better tools for testing experimental therapeutics. In this regard, genetically engineered mouse models of human cancer have the advantage that tumors in these models co-evolve with the tumor stroma, and therefore recapitulate the “highly integrated ecosystem” described by Folkman.

One of the difficulties in the widespread use of imatinib mesylate for the treatment of cancer has been the development of resistance. Imatinib mesylate targets a class of tyrosine kinases, including c-kit, abl, and PDGFR, and cancer cells have been shown to acquire resistance through mutations in the kinases. While cancer cells tend to be genetically unstable, continuing to develop new mutations over time, it is less clear whether benign tumors such as neurofibromas, or stromal cells such as mast cells, would share this capacity to mutate to a drug resistant form over the lifetime of the individual. The hope is that because the stromal cells do not demonstrate uncontrolled growth, they may be subject to greater control of the cell cycle and have intact repair responses to DNA damage. This would make them less likely to develop drug resistance. Although the study by Yang et al describes a single case report of successful treatment with imatinib mesylate, it will be very interesting to see how widely applicable this treatment is, particularly within the neurofibromatosis patient community, whether the drug must be given continuously to suppress the stroma contribution to tumorigenesis, and whether drug resistance becomes an issue after long-term use.

Studies of a rare genetic disease, retinoblastoma, led to a greatly expanded understanding of how tumors develop mutations and of the role of tumor suppressor genes in tumorigenesis (Knudson, 1971). Studies of another genetic disease, neurofibromatosis, are revealing the complexity of cellular interactions within tumors, and it will be exciting in the future to study whether and how the paradigms being developed in the field of neurofibromatosis research apply to sporadic human cancers.