Article Text

Download PDFPDF
Open questions in oesophageal adenocarcinogenesis
  1. Carlo C Maley
  1. Correspondence to:
    Dr C C Maley
    Molecular and Cellular Oncogenesis Program, The Wistar Institute, 3601 Spruce Street, Philadelphia, PA 19104, USA; cmaley{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Transition from GORD to Barrett’s oesophagus, and possibility of a reversal mechanism

Barrett’s oesophagus develops as a complication of chronic gastro-oesophageal reflux disease (GORD), and is the only known precursor to oesophageal adenocarcinoma (EA). Barrett’s oesophagus is an ideal model for studying neoplastic progression, because it can be studied longitudinally in vivo, with protocols that allow for safe, reproducible sampling by endoscopy biopsies.1 Furthermore, the major genetic events that characterise progression in Barrett’s oesophagus are some of the most common genetic events across human neoplasms, including inactivation of the tumour suppressor genes p16 (CDKN2A/INK4A)2,3 and p53 (TP53),4,5 with the subsequent development of tetraploidy and aneuploidy.6,7,8,9,10

There are important open questions at all stages of progression from GORD to EA. Little is known about the transition from GORD to Barrett’s oesophagus. What is the cell of origin for Barrett’s oesophagus? And why do only a minority of patients with GORD undergo that transition? The presence of p16, p53 and ploidy lesions in a Barrett’s neoplasm makes it very likely to progress to cancer,11 but the complete set of necessary and sufficient molecular events for carcinogenesis are not yet defined. The difficulty in treating EA has long been a problem without any solution, and so attention has been turned to prevention of EA. Non-steroidal anti-inflammatory drugs11,12 and photodynamic therapy13 are showing promise for preventing EA in patients with Barrett’s oesophagus. However, the presence of genetic heterogeneity within Barrett’s oesophagus14 suggests that the evolution of resistance may plague our prevention efforts. For example, there is some evidence that p53 mutations provide some resistance to photodynamic therapy.15 Thus, preventing the transition from GORD to Barrett’s oesophagus might be even more effective than trying to prevent progression from Barrett’s oesophagus to EA.

The open questions around the mechanism and prevention of the transition from GORD to Barrett’s oesophagus are what make the current work of Chang et al16 (see page 905) so exciting. They have shown that exposure of squamous biopsy specimens to retinoic acid, ex vivo, caused columnar differentiation. Conversely, an ex vivo Barrett’s oesophagus biopsy specimen can be transformed into a squamous-appearing epithelium, through the inhibition of retinoic acid using citral. These results implicate retinoic acid in the transition from GORD to Barrett’s oesophagus. They also suggest a potential approach to cancer prevention through the use of retinoic acid inhibitors, which may be used either to cause regression of Barrett’s oesophagus or to prevent the transition from GORD to Barrett’s oesophagus. If retinoic acid inhibitors do cause regression of Barrett’s oesophagus in vivo, it will be important to determine whether genetically abnormal cells remain in the squamous epithelium and whether they continue to pose a threat of progression to malignancy.

Inhibition of retinoic acid will have to be followed up using in vivo models. One caution here is that the application of retinoic acid itself has been used as a chemopreventive approach in a variety of cancers including head and neck cancer,17–21 the genetics of which are similar to those of Barrett’s oesophagus.22 However, there are concerns over the toxicities of retinoids, including ulceration of the oesophagus,23 concordant with Chang et al’s findings.16 A better understanding of the role of the retinoic acid pathway in neoplastic progression to EA may also help to explain the varying success of retinoic acid in cancer prevention. Related questions include how suppression of retinoic acid affects an epithelium with Barrett’s oesophagus after p16 and/or p53 has been inactivated? Better (and better characterised) experimental models of Barrett’s oesophagus are needed to help develop these and other cancer prevention agents.

Both retinoic acid and lithocolic acid activate the retinoic acid pathway and upregulate p21. In fact, Chang et al16 show that the two acids act synergistically, activating the pathway threefold more than either acid alone. If activation of the retinoic acid pathway is key to the development of Barrett’s oesophagus, the fact that lithocolic acid is a component of gastro-oesophageal refluxate suggests that proton pump inhibitors alone may not be able to suppress the transition from GORD to Barrett’s oesophagus.

Chang et al also provide intriguing evidence that the cell of origin for Barrett’s oesophagus comes from the submucosal glands. In their ex vivo squamous biopsy specimens treated with retinoic acid, they observed sloughing of the squamous cells and subsequent merging of the submucosal glands with the surface epithelium. In addition, they showed that columnar epithelium could still be generated if the squamous cells were stripped off before application of retinoic acid. This transformation seems to be a matter of differentiation rather than proliferation of a pre-existing columnar cell type. No evidence of BrdU incorporation or Ki67 expression in the new columnar epithelium was found. This apparent transdifferentiation should be confirmed and investigated in future studies.

A hierarchy of stem cells might exist in the oesophagus if the submucosal glands contain multipotent stem cells that feed tissue-specific stem cells residing in the Barrett’s oesophagus and in the squamous epithelium (fig 1). Others have observed the emergence of both Barrett’s and neosquamous epithelia from submucosal glands.24,25 This, combined with the results of Chang et al’s, suggests that submucosal glands can produce both squamous and columnar epithelium. The appearance of genetically normal squamous islands within large fields of Barrett’s epithelium with p16 or p53 sequence mutations26 could also be explained by genetically normal, multipotent stem cells in the submucosal glands replacing the genetically abnormal Barrett’s oesophagus stem cells. Rare cases of neosquamous islands were found to be genetically abnormal,26 which might be explained by the presence of the abnormality in the multipotent stem cells.

Figure 1

 A hierarchical stem cell model for the oesophagus. Stem cells in the epithelium (grey cells) can replenish the Barrett’s or squamous epithelium in the absence of wounding. When the tissue-specific stem cell compartment is destroyed, it may be replenished by multipotent stem cells in the submucosal glands.

Chang et al have given us an important lead in understanding the transition from GORD to Barrett’s oesophagus, and a potential mechanism for reversing that transition. For this purpose, the role and manipulation of the retinoic acid pathway should be further studied in the oesophagus. In addition, they have given us a hint for the anatomical mechanism by which intestinal metaplasia emerges in the oesophagus. The role of tissue architecture in coordinating and constraining neoplastic progression deserves further study, and will probably have important connections with the role of the microenvironment in neoplastic progression, a topic that is currently of intense interest in the cancer research community.


I thank Rachelle Kosoff, Tom Paulson, Brian Reid and Tom Vaughan for their helpful comments.

Transition from GORD to Barrett’s oesophagus, and possibility of a reversal mechanism



  • Competing interests: None.

Linked Articles