Barrett's adenocarcinoma (BA) has seen a rapid increase in incidence throughout the Western world. The diagnosis of BA is often at an advanced stage and is generally associated with a poor prognosis and a mean survival of less than one year. Adenocarcinomas however do not arise de novo but follow an established sequence from Barrett's metaplasia (BM) through dysplasia to neoplasia.

Efforts to intervene in the pathogenesis of oesophageal adenocarcinomas have so far been disappointing. Reduction of gastro-oesophageal reflux disease has led to minimal regression of BM and has yet to be shown to have any impact on cancer prevention. Surveillance programmes for patients with BM have had variable results and have raised important questions about their cost effectiveness and of better risk stratification of patients with BM. The prevalence of BM in the general population is approximately 1–3%, with only 0.5–1% of patients with BM converting to neoplasia each year.1 ,2 The reliable diagnoses of intestinal metaplasia and dysplasia have also been difficult to validate in each patient, mostly related to sampling errors due to the variable anatomy of the lower oesophagus and their patchy distribution within a segment of BM.

Our understanding of the molecular biology of BM has yielded many phenotypic and genetic changes within the epithelium that are associated with different stages along the metaplasia-dysplasia-neoplasia sequence.3 It has been suggested that some of these changes, such as cytokeratin subsets, might aid in the diagnosis and management of patients with BM and is the focus of the paper by Couvelard et alin this issue of Gut (See page761).4

Cytokeratins are highly conserved polypeptides that heterodimerise and form the building blocks for the intermediate filaments as part of the cell cytoskeleton. Intermediate filaments are anchored to desmosomes in the cell membrane and are of particular interest in epithelial systems in maintaining cell morphology, polarity, and intercellular adhesion.5 Cytokeratins are expressed in 20 distinct forms in epithelial cells but are absent from mesenchymal tissue where vimentin is used in the assembly of intermediate filaments.5

There are variable patterns of expression of cytokeratins in epithelial cells depending on the type, location, and differentiation of epithelium. Some cytokeratins have a broad range of expression in columnar epithelium, such as CK8 and CK19, while others such as CK7 and CK20 show highly restricted expression.6 CK20 is commonly used as a marker of intestinal differentiation. It is expressed on the surface and crypt epithelium of the normal colon and small intestine. In the stomach, expression is limited to the surface foveolar epithelium, with no gastric gland or pit staining. CK7 has been proposed as a marker of ductal differentiation. It is expressed in ductal breast carcinomas but not in the normal epithelium of the gastrointestinal tract. Cytokeratins are very stable proteins and are conserved in most epithelial cancers, even in the presence of gross phenotypic and genetic alterations. For this reason, detection of cytokeratin subsets such as CK7 and CK20 have become very useful in diagnosing the origins of occult or poorly differentiated cancers and are part of current clinical practice.

Differential patterns of cytokeratin expression have been demonstrated in the oesophagus and proposed as useful clinical markers. A phenotypic map of cytokeratin expression validates our current histopathological categorisation of the gastro-oesophageal junction (fig 1). However, a unique pattern of CK7 and CK20 immunohistochemical expression, designated the BM CK7/20 pattern, has been demonstrated to be both sensitive and specific to Barrett's oesphagus and may be used as an objective marker of BM.7 This pattern shows superficial CK20 staining and strong CK7 staining of both superficial and deep glands and can be used to distinguish BM from intestinal metaplasia of the stomach which, although can be histologically indistinguishable, has a different pattern of cytokeratin expression, is associated withHelicobacter pylori infection, and is not associated with an increased risk of malignancy.8

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Schematic illustration of cytokeratin (CK) 7 and 20 expression in the normal oesophagus, Barrett's metaplasia, squamocolumnar junction, and proximal stomach. CK 7 is only found in the superficial layers of the squamous epithelium in contrast with columnar lined tissue in the oesophagus that shows increased expression throughout the crypts and glands. CK20 is limited to superficial areas of Barrett's metaplasia and the gastric cardia but upregulated in intestinal metaplasia of the cardia.

The Barrett's CK7/20 pattern has been demonstrated in both long and short segment BM. The paper by Couvelard et al confirms the presence of a Barrett's CK7/20 pattern and demonstrates that the same CK7/20 pattern holds true for ultrashort segment BM.4 The authors further suggest that these findings may be useful to improve endoscopic surveillance strategies to specifically target those at increased risk of BM and its complications. Ultrashort BM is however an area surrounded by controversy. It is defined as intestinal metaplasia found at the gastro-oesophageal junction arising in either very short tongues of columnar mucosa or in eccentric or normal appearing squamocolumnar junctions (SCJs). It has been reported that the prevalence of this may be up to 43% in unselected patients undergoing routine endoscopy. The nature of metaplasia in eccentric or normal SCJs remains unknown. This has prompted much debate as to the risk (if any) of dysplasia or indeed cancer in this group and the appropriateness of endoscopic surveillance. Furthermore, if we are to regard this lesion in a similar way to short and long segment Barrett's and it is found in up to 43% of normal appearing SCJs, should we be performing screening biopsies of these patients? Certainly this does not seem to be a plausible option.

Perhaps it is in the understanding of the biology of BM where cytokeratins have already made a potentially useful contribution. Specifically, rearrangements in the composition of lateral cell-cell adhesion junctions, the desmosomes, have been recently reported in BM.9 These alterations in adhesion complexes while being associated with concomitant changes in cytokeratins are also important in releasing gamma catenin, a known oncogene involved in T cell factor/leucocyte enhancing factor transcription.10

In conclusion, greater understanding of the processes involved in the development and progression of BM is crucial if we are to develop strategies to intervene at an earlier stage in the metaplasia-dysplasia-carcinoma sequence and alter the outcome of oesophageal adenocarcinomas. Cytokeratins are undoubtedly of great scientific importance but caution should be observed in areas where clinical risk has not been already established.