Research CommentariesONCOGENESIS

Landscaping the Cancer Terrain

See allHide authors and affiliations

Science  15 May 1998:
Vol. 280, Issue 5366, pp. 1036-1037
DOI: 10.1126/science.280.5366.1036

Few lines of investigation have taught us more about cancer than the study of inherited tumor susceptibility syndromes. Initially, the mutations responsible for these diseases were thought to promote malignancy in a straightforward manner, through inactivation of “tumor suppressor” genes, which directly modulate cell birth or cell death. More recently, however, susceptibility genes that work through less-direct mechanisms have come to light. The genes defective in patients with juvenile polyposis syndromes (JPSs), for example—one of which is described on page 1086 of this issue (1)—illuminate this principle and also raise fundamental questions about the relation between neoplastic cells and the “other cells” that together constitute a tumor mass.

A dozen tumor suppressor genes are known to prevent cancer through direct control of cell growth—including p53, Rb, VHL, and APC. Inactivation of these genes contributes directly to the neoplastic growth of the tumor; thus, they normally function as “gatekeepers” to prevent runaway growth (see the figure). Accordingly, restoration of the missing gatekeeper function to cancer cells leads to suppression of the neoplastic growth.

These traditional tumor suppressors are being joined by an ever-increasing number of susceptibility genes that indirectly suppress neoplasia (for example, XPB, ATM, MSH2, and MLH1). The prototypes for this class of genes encode DNA repair proteins that act as “caretakers” of the genome. Inactivation of a caretaker gene results in a greatly increased mutation rate and is equivalent to a constant exposure to mutagens. It is not surprising that such defects should lead to cancer, but restoration of caretaker function to a cancer cell will not affect its growth. As these indirectly acting genes are never required for neoplasia, most nonhereditary (sporadic) tumors will evolve without them.

Pathways to colorectal cancer.

Colorectal cancer can be caused by genetic defects that interrupt “gatekeeper,” “caretaker,” or “landscaper” functions in the colon. The bold arrows indicate the step accelerated in that class of tumor.

Sporadic colorectal cancer. The probability that a colorectal epithelial cell will acquire the genetic changes leading to a benign tumor is low, but still half the Western population develops such a tumor by age 70. A fraction of these benign tumors (adenomatous polyps) progress to cancer, yielding a lifetime cancer risk in the general population of 5%.

Gatekeeper defects. In patients with familial adenomatous polyposis (FAP), for example, an inherited defect in the APC gene leads to the development of hundreds of adenomatous polyps (10). Because of their great numbers, some polyps are virtually guaranteed to progress to cancer.

Caretaker defects. Patients with hereditary nonpolyposis colorectal cancer (HNPCC) develop adenomatous polyps at about the same rate as the general population, but these polyps progress to cancer much more often because of defective mismatch repair, which results in an increased mutation rate (10).

Landscaper defects. Patients with juvenile polyposis syndrome (JPS) and ulcerative colitis (UC) develop hamartomatous polyps in which the proliferating defective population of cells appears to be derived from the stroma. Consequently, the epithelial cells associated with the polyps are more likely to undergo neoplastic transformation, as a result of an abnormal microenvironment. Likewise, the initially normal epithelial cells associated with the inflammatory process of UC are at increased risk of neoplastic transformation.

Illustration: K. Sutliff

A second class of indirectly acting cancer susceptibility genes is suggested by recent studies on JPS. Individuals with JPS have an increased risk of colorectal cancer, but the primary manifestation of this syndrome is the development of multiple hamartomatous polyps of the colon at a young age. These polyps are markedly different from the epithelium-rich adenomatous polyps that give rise to most cases of colorectal cancer. Polyps from patients with JPS have a low potential to become malignant and are composed largely of stromal cells, comprising a mixture of mesenchymal and inflammatory elements in which epithelium is entrapped, often forming dilated cysts. The epithelial cells within and surrounding the polyp are initially devoid of neoplastic features but nevertheless are at increased risk of becoming malignant.

It would thus seem that the increased cancer susceptibility due to inherited mutations in JPS is the product of an abnormal stromal environment. That an abnormal stroma can affect the development of adjacent epithelial cells is not a new concept. Ulcerative colitis (UC) is an autoimmune disease that leads to inflammation and cystic epithelium in the mucosa of the colon. Initially the embedded epithelium shows no neoplastic changes, but foci of epithelial neoplasia and progression to cancer eventually develop in many cases. The regeneration that occurs to replace damaged epithelium may increase the probability of somatic mutations in this abnormal microenvironment. The increased risk of cancer in JPS and UC patients therefore seems primarily the result of an altered terrain for epithelial cell growth and can be thought of as a “landscaper” defect.

How can one test the landscaper hypothesis and demonstrate that the primary oncogenic effect of the mutation is on stromal rather than epithelial cells? This should be possible through careful genetic evaluation of the stromal and epithelial populations of the hamartomas. It is intriguing that the stromal cells, but not the epithelial cells, of most hamartomas from JPS patients contain a clonal genetic alteration (2). Similarly, clonal genetic changes have been demonstrated in the stroma, but not the epithelial cells, of endometrial polyps (3). In contrast, clonal genetic alterations have been demonstrated in epithelial cells, but not stromal cells, of the polyps arising in patients with familial adenomatous polyposis (FAP) (4) or Peutz-Jeghers syndromes (5)—which are morphologically distinct from those in JPS patients. Now that we know that inherited mutations in PTEN (6) or SMAD4 (1) can lead to the development of the hamartomatous polyps, it will be informative to determine whether the stromal or epithelial compartment of hamartomatous lesions (and the cancers that arise within them) show inactivation of the wild-type copy of PTEN or SMAD4 inherited from the unaffected parent.

Both PTEN and SMAD4 directly control cell growth in other tumor types. Accordingly, somatic mutations of PTEN (7) and SMAD4 (8) commonly occur in brain and pancreatic cancers, respectively, although mutations of these genes occur infrequently in colorectal cancer cells (<5% of colorectal cancers for PTEN and <15% for SMAD4). Could the same gene function as a gatekeeper in one tumor type or at one stage of tumor development and as a landscaper in another, violating the principle of Ockham's razor? Perhaps this is not unexpected given the functions of these genes. PTEN is a dual-specificity phosphatase that is likely to affect a plethora of processes in many cell types, and SMAD4 is a central player in the signal transduction pathway activated in response to the large family of TGF-β (transforming growth factor-β)-like ligands. The ligand that triggers the pathway containing SMAD4 in pancreatic epithelial cells may be entirely different from the one it responds to in JPS hamartomas.

These results add to the emerging realization that solid tumors are not simply composed of neoplastic epithelial cells. Historically, the search for drugs that can modulate neoplasia has focused on such epithelial cells. More recent results, however, have suggested that targeting specific stromal cells (such as those forming blood vessels) might be more valuable for therapeutic purposes (9). Could drug targeting of the paracrine factors and other features of the stromal-epithelial interaction be similarly useful? Although such drugs would be unlikely to affect advanced tumors, in which the neoplastic cells are largely autonomous, they might be efficacious in the early, benign stages of tumorigenesis, nipping them in the bud.

References and Notes

Navigate This Article