Molecular characterisation of serrated precursors in colorectal carcinoma

Abstract

Sporadic colorectal carcinoma (CRC) arises from two histologic pathways, the conventional and serrated pathways. Since their recognition, numerous histologic, molecular, and prognostic differences between the conventional and serrated pathways have been described.The serrated pathway has two recognised precursor lesions, sessile serrated adenoma (SSA) and traditional serrated adenoma (TSA). SSAs show a proximal predominance and are characterised by disordered crypt proliferation without dysplasia. TSAs, which are much less common, are predominantly distal and comprised of columnar cells with abundant apical eosinophilic cytoplasm. Ectopic crypt foci, which constitute abnormal proliferative areas away from the muscularis propria, are also common. These idealised descriptions have proved surprisingly controversial. SSAs can be difficult to separate from hyperplastic polyps, an important distinction as the latter does not have malignant potential. SSAs rarely progress to dysplasia (then termed SSA with dysplasia or SSAD), but once dysplasia develops progression to CRC is rapid. TSA-like features (e.g. eosinophilia) are common in conventional adenomas, with which TSA potentially forms a histologic spectrum; some of these hybrid lesions are termed “serrated tubulovillous adenomas” (sTVAs). TSAs are also commonly seen with a precursor lesion, either a hyperplastic polyp or SSA. These observations have led to the view that TSAs are “progressed” lesions and thus inherently dysplastic.Molecularly SSAs are initiated by BRAF mutation and accumulate methylation over time. After methylation reaches a threshold, tumour suppressor genes such as MLH1 are silenced and dysplasia supervenes. TSAs are initiated by either BRAF or KRAS mutation and become variably methylated. However, MLH1 is almost never silenced, and inactivation of other genes is believed to be responsible in TSA progression. The CRCs which arise from the serrated pathway carry these same molecular changes, and can have a superior prognosis (when microsatellite unstable due to MLH1 loss) or a poor prognosis (when microsatellite stable).To devise management strategies appropriate for serrated lesions, several points along the serrated pathway are examined in this thesis.Some gastrointestinal neoplasms harbour dual KRAS and GNAS mutations, and it is known a small subset of conventional adenomas and CRCs also have GNAS mutations. Motivated by this, Chapter 1 assesses whether GNAS mutations are present in a large series of conventional adenomas, serrated lesions and CRCs, and correlates the molecular findings with clinical data. Although no mutations are identified in conventional adenomas, a small proportion of sTVAs, TSAs and CRCs are found to be GNAS mutant. This lends support to separating out sTVAs as a distinct entity which shares molecular similarities with TSAs. Furthermore, GNAS-mutant CRCs are associated with poor prognostic features, including serrated morphology and high pathologic stage.The morphologic spectrum of SSADs has not been clearly defined. Although the 2010 World Health Organization classification and several publications have illustrated “conventional” and “serrated” forms of dysplasia, the reproducibility and relevance of this separation is uncertain. Chapter 2 evaluates the morphology and MLH1 status of a large series of SSADs, and correlates these with clinical features. Minimal deviation dysplasia carries only subtle cytologic abnormalities and is easily missed. Serrated dysplasia must be distinguished from TSAs. Adenomatous dysplasia is morphologically identical to conventional adenomas. Lastly, dysplasia not otherwise specified forms the largest group and contains the most typical SSADs.Endoscopically and histologically all SSAs appear identical, and it is difficult to identify lesions at risk of developing dysplasia. Since SSADs are almost uniformly methylated, highly methylated SSAs, if they can be identified in practice, will represent the set of lesions at highest risk. Chapter 3 correlates methylation status with clinical data in a large series of SSAs. SSA methylation is strongly associated with patient age and proximal colonic site, but not with patient gender or lesion size.Although age and site are powerful predictors of progression risk in SSAs, this still leaves an unacceptably large number of lesions. Which, then, are the lesions at highest risk? The highest risk SSA will be identified in a lesion which has already developed dysplasia, namely the remaining “background” of SSADs. Chapter 4 compares methylation status of this background with ordinary non-dysplastic SSAs. Although no novel hypermethylated tumour suppressor gene is identified, a large number of differential methylation events are nonetheless present in this background, which persist into the resulting CRC.In summary this thesis has clarified finer details of the serrated pathway, and the findings may translate into clinical practice. GNAS-mutant CRCs warrant closer surveillance due to aggressive behaviour. Subtle SSADs can now be recognised routinely. The close association between SSA progression risk, patient age and colonic site has implications for surveillance. Lastly, highest risk SSAs exhibit methylation changes similar to CRC, which may influence the development of diagnostic methylation panels.