Abstract IA05: The serrated pathway of colorectal tumorigenesis

Abstract

Abstract Over last 15 years it has been recognized that a significant proportion of colorectal cancers arise from serrated polyps, previously called hyperplastic polyps that were thought to be of no malignant potential. Minimum estimates of the proportion are based on the presence of the BRAFV600E mutation since this only occurs in serrated polyps. In a series of cancers occurring between 1998 and 2007 13% had a BRAFV600E mutation whereas this proportion was 20% in our recent series from 2012 to 2013 (1). The proportion may be increasing since serrated pathway cancers are predominantly proximal and the rate of decline in incidence is less in proximal than distal cancers. The WHO classification of serrated lesions now recognizes hyperplastic polyps (HP), sessile serrated adenomas (SSA) and traditional serrated adenomas (TSA). SSAs and TSAs are premalignant but do not have conventional dysplasia. TSAs are rare but easily detected lesions mostly in the distal bowel with 20% having advanced dysplasia or invasive malignancy (2). By contrast, SSAs are common constituting 12% of all polyps (3) and are typically pale, sessile lesions with indistinct borders. More than 70% occur in the proximal colon and they are more common in females. Until very recently they were often missed or incompletely resected by colonoscopists and were underdiagnosed by pathologists. For these reasons, they have contributed disproportionately to the failure of colonoscopy to fully protect against cancer. The natural history of SSAs is only now beginning to be understood. Only a very small minority show dysplasia (SSAD) but these show an abrupt transition from bland SSA to dysplasia and often invasive malignancy. The average age of individuals with SSA is 59 years, with SSAD 75 years and with BRAF mutant cancer 75 years (4). This suggests SSAs have a long dwell time of about 15 years before a rapid and currently unpredictable transition to invasive malignancy. The gate keeper genetic alteration causing conventional tubular adenomas is increased WNT signaling usually due to APC mutation. The driver mutations in serrated polyps involve the MAPK pathway. Data from our laboratory show APC mutation in 80% of tubular adenomas, KRAS mutation in 5% rising to 21% if the histology is tubulovillous and no BRAF mutation. APC mutation is not present in HPs or SSAs and in only 15% of SSADs. Goblet cell HPs have a high rate of KRAS mutation whilst microvesicular HPs, SSAs and SSADs have a BRAF mutation rate of 70%, 78% and 93% respectively. TSAs have mutation in either BRAF (67%) or KRAS (22%) and APC mutation in 35%. Thus conventional colorectal tumorigenesis relies on Increased WNT signaling although some adenomas later acquire a KRAS mutation and villous histology which is known to indicate a high risk of malignant conversion. It is striking that conventional adenomas virtually never contain BRAF mutations. In contrast, serrated lesions are initiated by increased MAPK signaling. If due to a KRAS mutation, the polyp has little malignant potential although rarely it can acquire other changes to become a TSA. If due to BRAF mutation, the hyperplastic polyp may progress to SSA and SSAD. Some BRAF mutant SSAs develop into TSAs (2). When high risk dysplasia develops in a SSAD or TSA, an APC mutation is sometimes acquired. In colorectal cancers there is a strong association between the presence of BRAF mutation and CpG island methylation (CIMP) and CIMP can be shown to evolve within serrated polyps from <5% in HPs to >90% in SSADs (2,4,5). In TSAs, CIMP is more common if BRAF is mutated. CIMP is rare in tubular adenomas. The reason for the association of CIMP and BRAF mutation has been unclear. Work in progress in our laboratory is using a mouse model of BRAFV600E induced intestinal tumorigenesis to study DNA methylation. When the BRAFv600E mutation is induced at 2 weeks of age, the small intestine and colon become hyperplastic within days and remain so throughout life. Beginning at 10 weeks, dilated crypt bases resembling SSAs are observed and at 8 months dysplastic serrated adenomas develop. At 14 months invasive malignancy is present. DNA methylation was analyzed using EpiTect Methyl II PCR arrays (mouse colon cancer panel, Qiagen). DNA was extracted from small intestine and colon distant from any macroscopic lesions. There was no difference between the hyperplastic and control intestine until the age of 5 months after which there was progressive and significant methylation in the hyperplastic mucosa. This suggests that chronic oncogenic signaling from mutant BRAF in the intestinal cell in vivo results in CpG island methylation. This slow evolution of methylation accords with the natural history of SSAs observed in humans DNA methylation is functionally important to tumorigenesis. For example, methylation-induced silencing of p16 allows cells to escape oncogene induced senescence as SSAs become malignant. Probably the most important gene to be silenced by methylation is MLH1 since this results in mismatch repair deficiency and microsatellite instability (MSI). Methylation appears to reach a critical level leading to abrupt loss of immunohistochemical staining of MLH1 as SSAs develop dysplasia and malignancy. Once MSI is present, numerous mutations especially in repeat tracts drive malignant progression. However, some SSAs progress without methylation of MLH1 to BRAF mutant, microsatellite stable (MSS) cancers, a cancer subgroup associated with a particularly poor prognosis (1). BRAF mutant, MSS cancers also arise from TSAs as <1% of TSAs silence MLH1 during malignant progression (2). The pathway of conventional colorectal tumorigenesis proposed by Kinzler and Vogelstein suggests the progressive accumulation of mutations in the WNT, MAPK, TGBβ/BMP and p53 pathways. There is growing evidence the same pathways are targeted in serrated neoplasia but by different mutations and in a different order. In advanced serrated lesions the WNT pathway can be activated by methylation induced silencing of SFRPs and also by inactivating mutations of RNF43. RNF43 contains 2 coding region repeat tracts and was found to be mutated in 47/54 (87%) of BRAF mutant /MSI cancers but also had other inactivating mutations in 8/33 (24%) of BRAF mutant/MSS cancers. Only 3/79 (4%) of BRAF wild type cancers had RNF43 mutations (6). Mutations in RNF43 are present only infrequently in SSAs without dysplasia but are present in 25% of TSAs which will develop into MSS cancers (7,8). The coding region repeat tract mutations presumably develop when SSAs transition to MSI cancers. The TGBβ/BMP pathway which is sometimes disrupted by SMAD4 mutations in conventional neoplasia, is targeted by frequent mutations in coding region repeat tracts ( e.g., TGFβR2, ACVR2 and BMPR2). Aberrant epithelial expression of the BMP antagonist GREMLIN1 occurs in TSAs and may be the cause of their characteristic villiform architecture with ectopic crypts (9). p53 mutation frequently occurs as SSAs and TSAs transition to MSS cancers but is infrequent in MSI cancers (10). To further explore the concept that serrated neoplasia targets the same pathways as conventional neoplasia but by different mechanisms and in a different order, we are currently studying an organoid system.