Summary: Different GEM models of intestinal cancer display characteristic disease manifestations. For example, lesions appear primarily in the cecum and colon in some mouse lines and in the small intestine in others. Gross morphology is also influenced by genotype; pedunculated polyps carry Apc mutations (regardless of the initial mutation) while sessile lesions do not. Disease severity also differs among the GEM models. For example, theApcmodels and some of the MMR GEM have a high incidence of adenoma but only rarely develop carcinomas. Others, such as the Smad3 and Tgfβ1xRag2 mutants exhibit adenoma and carcinomas with equal frequency. In these models carcinomas are observed more frequently in the large intestine than in the small intestine.

Mouse models of intestinal cancer exhibit some important differences with respect to human disease. In general, adenomas are the most common lesion in murine intestinal cancer models. However, unlike humans, adenomas occur throughout the intestine in mice, with no predilection for the large bowel. Another important difference is that metastasis is rare in mice. Only one mouse line,Smad3^-/-(83), exhibits metastasis of colonic adenocarcinomas to lymph nodes and liver, which is a common occurrence in human colorectal carcinomas. There is a critical need for a consistent model of metastasis.

Gut flora is an important modifier of intestinal tumor development. For example, when derived in a germ-free environment, Il-10^-/-, IL-2^-/-, and Tgfβ1^-/-XRag2^-/- mice do not develop colonic lesions (77,91,94). Despite evidence of its importance, the role of the gut flora is often overlooked in studies involving GEM. Understanding the role of bacterial flora in the intestine is necessary to understand disease initiation and progression in murine models of intestinal tumor formation. The precise role of bacteria in the development of lesions is not understood, but is an important issue that should be addressed in each model with an unexpected or invasive phenotype.

An important issue in interpreting the phenotypes of GEM is accurate discrimination between invasive cancer and herniation. In the mouse, the muscularis mucosa is very thin, thus, the mucosa is able to penetrate the muscle easily. Herniation can occur through weak points where blood vessels and lymphatics traverse the muscle. Herniation can also result from pressure exerted by the polyp that can push basal crypt cells through the muscularis mucosa. Both of these mechanisms are enhanced by inflammatory conditions.

The Consensus report on the intestinal pathology of GEM (58) includes a classification system (link to box: Features that distinguish?) to facilitate consistent and accurate distinction between carcinoma and mucosal herniation. At least one of the following features should be present to classify a lesion as a carcinoma: desmoplasia, loss of mucosal lining in invading glands, and the presence of irregular, sharp or angulated glands. Particular caution is necessary when evaluating inflamed areas because of the severe fibrosis and frequent disruption and herniation of glandular tissue that occurs in many chronic inflammatory bowel disease models.

Herniation of the intestinal mucosa of an Msh3-/- mouse.	The invading cells include multiple cellular types, and there is no desmoplastic response in the herniated intestinal lesion.	There is severe desmoplasia and loss of mucosa in the carcinoma.