Intended for healthcare professionals


Gastro-oesophageal cancer: death at the junction

BMJ 2000; 321 doi: (Published 19 August 2000) Cite this as: BMJ 2000;321:463

Understanding changes at molecular level could lead to screening opportunities

  1. Janusz Antonio Jankowski, consultant gastroenterologist (j.jankowski{at},
  2. Ian Perry, clinical fellow,
  3. Rebecca Faith Harrison, consultant pathologist
  1. Epithelial Laboratory, Department of Medicine, University of Birmingham, Birmingham B15 2TH

    The death rates from cancers of the oesophagus and gastro-oesophageal junction, adjusted for age, have risen steadily since the early 1970s (from 3 to 6 per 100 000 and from 1.5 to 3 per 100 000 population in the United Kingdom respectively).1 These figures are comparable to those in northern Europe and the United States. The incidence of Barrett's adenocarcinoma in the United States has increased from 0.3 per 100 000 to 2.3 per 100 000 over the past three decades.

    Despite improvements in multimodality therapy, especially chemotherapy regimens of combined epirubicin, cisplatin, and fluorouracil combined with surgery, survival has not improved significantly, suggesting that alternative strategies for identifying and treating these conditions are needed.

    The incidence of intestinal metaplasia of both the oesophagus (Barrett's oesophagus) and the gastro-oesophageal junction are also increasing. This metaplastic tissue is believed to have a premalignant potential, and Barrett's oesophagus is related to bile and acid reflux disease.2 About 8% of patients undergoing routine endoscopy and 3% of the adult population have at least 1 cm of Barrett's oesophagus.3 Furthermore, 17% of patients undergoing routine endoscopy and 6% of the adult population may have intestinal metaplasia of the gastro-oesophageal junction. 3 4 These metaplastic lesions are characterised by mucin-secreting epithelium, containing goblet cells, that replaces the native stratified squamous or transitional zone epithelium.

    Metaplastic changes may progress from dysplasia to adenocarcinoma.2 About 5-15% of people with Barrett's oesophagus and 2-5% of those with intestinal metaplasia of the gastro-oesophageal junction also have dysplasia, which in the case of Barrett's oesophagus increases the risk of cancer between 30-fold and 150-fold. The risk of cancer for people with metaplasia of the gastro-oesophageal junction has so far not been quantified.

    This has led many centres to establish surveillance programmes to identify dysplastic changes or early adenocarcinomas.5 However, although these programmes detect cancers earlier, there is controversy about their cost effectiveness. 6 7 Interest has therefore been rekindled in strategies to prevent the onset of Barrett's oesophagus or intestinal metaplasia of the gastro-oesophageal junction and to find other risk factors that more accurately detect the subgroups of patients who will progress to malignancy.

    Rare inherited syndromes of colorectal cancer have given valuable information about tumour initiation. Syndromes of familial gastro-oesophageal cancer are rare and heterogeneous and account for only 1-5% of cases, but they have also provided valuable information.8 In particular, inherited germline mutations of the E cadherin gene, involved in cell adhesion, leads to loss of E cadherin expression.8 E cadherin is not only a cell adhesion molecule but also a tumour suppressor gene. Reduced expression of adhesion molecules on the surface membranes of cancer cells makes them far more likely to have invasive properties. Furthermore, analysis of sporadic gastric cancer shows that the stage and invasiveness of gastric tumour is also associated with reduced expression of E cadherin. E cadherin binds with an intracellular protein called β catenin, to form adhesion complexes. β catenin levels are tightly regulated within the cell, and free, unbound β catenin is normally completely degraded. If any free, unbound β catenin accumulates it can enter the nucleus and bind with certain transcription factors that help activate target oncogenes such as COX-2 and c-myc that may induce proliferation.9 The amount of β catenin and transcription complexes in the nucleus is dramatically increased by the release of unbound β catenin in situations where E cadherin expression is reduced. This situation seems to occur during the progression from metaplasia to adenocarcinoma.10

    A second inherited predisposition to gastric cancer has also been reported. Infection of the gastric body with Helicobacter pylori can cause hypochlorhydria, atrophy, and malignancy, whereas infection of the antrum is related to the development of peptic ulcer disease. Evidence now suggests that these different outcomes are related to the host response. Abnormal variants of the interleukin 1β gene (genetic polymorphisms that enhance activity) are associated with an increased risk of developing gastric cancer.11 Patients possessing such a polymorphism produce higher levels of interleukin 1β in response to H pylori infection, and interleukin 1β increases the risk of developing atrophy and malignancy. Furthermore, interleukin 1β can reduce the expression of adhesion molecules, such as E cadherin, further accentuating the tendency to malignancy.12

    The role of mucosal inflammation

    Gastro-oesophageal metaplasia seems to be induced or potentiated by mucosal inflammation. Understanding the molecular changes in this process may mean that we can identify people at risk of developing malignancies. Identifying E cadherin mutations and interleukin 1β polymorphisms may make it possible to screen people who have known risk factors such as a strong family history or metaplasia or dysplasia. Now that there is evidence to implicate chronic inflammation in cancer development, the role of anti-inflammatory drugs such as cyclo-oxygenase-2 inhibitors or more specific cytokine inhibitors may provide a new impetus to medical intervention.2


    We thank Mr John Fielding, chairman of the NHS/Department of Health Upper GI Cancer Group, for his helpful comments during the writing of this manuscript.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    View Abstract