Systematic review: the chemoprevention of oesophageal adenocarcinoma


Dr M. Rhodes, Department of Upper Gastrointestinal Surgery, Norfolk and Norwich University Hospital, Colney Lane, Norwich, UK.


The incidence of oesophageal adenocarcinoma is increasing in the UK faster than any other malignancy. Despite its relatively poor prognosis and the limited success of existing treatments, there is enthusiasm that chemopreventive agents might be able to stem the transition from normal squamous epithelium to adenocarcinoma.

We discuss gastro-oesophageal reflux as the main risk factor for the development of Barrett's metaplasia, the only known precursor of oesophageal adenocarcinoma. Treatment options for reflux disease are considered with regard to their effects on cancer risk. Recent advances in the molecular and cell biology of Barrett's are outlined, and potential targets for chemoprevention examined.

Available treatments for reflux disease have not convincingly altered the likelihood of cancer development. Epidemiological and animal studies support the use of non-steroidal anti-inflammatory drugs as potential chemopreventive agents. Dietary agents, however, have a more favourable side-effect profile and may prove to be an attractive alternative, although more work is needed to fully explore this prospect.


The incidence of oesophageal adenocarcinoma (OAC) in Europe and the UK has increased considerably over the last 20 years,1 with a rise per annum greater than that of any other malignancy.2 Most patients with this condition are diagnosed at an advanced stage and have a poor prognosis with a mean survival of less than a year.3

Existing treatments for OAC have generally had limited success. Oesophagectomy can be curative for disease limited to the oesophageal wall, but is associated with a high morbidity and mortality. Chemotherapy and radiotherapy may improve survival4 but by a questionable degree, and trials are underway to find the most effective neoadjuvant regimen.5 Attention is therefore now focusing on identifying potential targets for therapy that might arrest or delay the progression to adenocarcinoma. The purpose of this article is to review our current understanding of the pathogenesis of OAC and to discuss potential strategies to prevent it.


A review of the medical literature was performed to identify studies evaluating the cellular mechanisms involved in the development of OAC. In addition, a Medline and PubMed search was performed for English language articles that specifically addressed the issue of chemoprevention of OAC. Terms used in this search included: chemoprevention, OAC, PPIs, NSAIDs. Cyclo-oxygenase (COX)-2 inhibitors, diet and n-3 fatty acids.

Epidemiology of oesophageal adenocarcinoma

Oesophageal adenocarcinoma rates vary considerably from country to country.6 Population-based studies from around the world have shown high prevalence areas in western Europe, parts of Africa and Asia.7 In addition, white males have a higher predominance8 compared with blacks. Many risk factors for OAC have been identified and these include smoking, alcohol consumption9 and medications that relax the lower oesophageal sphincter.10 Increasing obesity may be important in explaining the rise in prevalence of OAC in the developed world.11 Epidemiological studies in both the USA12 and Sweden13 have shown that patients in the heaviest body mass index (BMI) quartile have a significantly higher risk of developing OAC.

By far the most important risk factor for OAC is gastro-oesophageal reflux disease (GERD). Those with recurring symptoms of GER have an eightfold increase in the risk of OAC.14 Furthermore, the incidence of GERD appears to be increasing in the west.15

Gastro-oesophageal reflux disease and Barrett's metaplasia

Long-standing GERD can predispose the replacement of normal squamous epithelium in the lower oesophagus with columnar epithelium, a condition known as Barrett's metaplasia. Barrett's is now well-established as the premalignant condition from which OAC arises16 and the rate of development of OAC in patients with Barrett's is approximately 0.5% per annum.17 As with OAC the incidence of Barrett's has been increasing in the UK,18, 19 but fewer than 5% of those affected will go on to develop OAC.20 Furthermore, GER is not always symptomatic, and thus many Barrett's patients remain undiagnosed. It has been estimated from necropsy findings that for every one patient under surveillance there may be more than 20 who have Barrett's that are unrecognized.21

There are two favoured hypotheses for the origin of Barrett's. One is that metaplasia of pluripotent stem cells in the basal cell layer occurs as a result of repeated exposure to acid and bile reflux.22 The other is that columnar cells originate from some other source such as from the submucosal gland ducts.23 Metaplasia is followed by a series of histopathological changes, namely dysplasia, carcinoma in situ and finally culminating in the development of OAC.24 However, not all patients with persistent reflux go on to develop cancer, and the biological mechanism through which reflux constituents bring about the progression to OAC so far remains ill defined.

The components of reflux in the oesophagus have been widely studied. In vitro work has shown that it is the pattern of acid exposure that may be the important factor in driving carcinogenesis. In one model using cultured endoscopic biopsies from patients with Barrett's oesophagus continuous acid exposure blocked cellular proliferation whereas short acid pulses stimulated proliferation.25 These pulses may mimic physiological acid reflux, and would account for the variable risk of neoplastic progression between individuals. Bile salts also form part of the refluxate and Kaur et al.26 have shown that bile can enhance proliferation in Barrett's via protein kinase C. However, bile acids when mixed with acid have the opposite effect. Animal studies have additionally shown that bile constituents can activate the arachidonic acid pathway27 and enhance the formation of reactive nitrogen species, which may also be carcinogenic.28 These results suggest that specific patterns of acid and bile exposure can bring about the progression to neoplasia, and that the inflammatory response to GER may be unique in the context of Barrett's.29

Chemoprevention of OAC through pharmacological treatment of GERD

Proton-pump inhibitors (PPIs) are the cornerstone of medical treatment for symptoms of GERD. Although studies have shown that PPIs can cause regression of Barrett's30, 31 others have not conclusively shown that they do.32, 33 Discrepancies in the literature may be due to the lack of a standardized method for measuring the length and distribution of Barrett's.34 Furthermore, it is possible that the relatively insensitive nature of Barrett's oesophagus requires far higher doses of PPIs to achieve acid suppression than to relieve symptoms.35 This has been highlighted in a study demonstrating that standard doses of PPIs administered to Barrett's patients could relieve symptoms of GERD after a 6-month period, but many continued to have pathological acid reflux as measured by 24 h pH monitoring.36 In a randomized double-blind study addressing these issues, patients were given 40 mg omeprazole twice a day and underwent pH-metry to confirm adequate acid suppression.37 After 2 years, there was a statistically significant regression of Barrett's, although this was only 8% of the total surface area.

There have been relatively few studies so far investigating the cellular effects of PPI treatment. Complete elimination of acid reflux has been shown to reduce cell proliferation36, 38 and increase expression of the cyclin-dependent kinase inhibitors p16 and p21.39 These latter findings suggest that complete acid suppression may influence the alterations in cell cycle control that occur during carcinogenesis.

Whether such acid-lowering treatment can modify the risk of cancer development is as yet unknown. It may be that transformation of Barrett's to dysplasia is the important step that should be focused upon rather than regression of Barrett's per se. A prospective analysis of over 200 patients over a 20-year period has provided evidence that PPIs are associated with a significantly reduced risk of dysplasia in Barrett's oesophagus.40 However, more research needs to be performed in this area, despite the problems of recruiting large numbers of patients over a long period of time.

Genetic pathways

It is generally believed that progression through the metaplasia-dysplasia-adenocarcinoma sequence is characterized by a number of genetic and epigenetic alterations of normal cellular function. These include both somatic mutations, and gene-silencing events involving aberrant methylation of CpG-islands in the promoter regions of tumour-suppressor genes.41 Although a complete description of the state of present knowledge is outside the remit of this article, it is true to say that our understanding of the molecular alterations in Barrett's and OAC has so far lagged behind those of other gastrointestinal (GI) cancers.

The sequence of genetic alterations appears to occur in a stepwise manner and can result in the development of a clone of cells that have malignant potential (see Figure 1). Because of the low prevalence of OAC many studies have used small numbers and there has been disagreement between researchers regarding ‘when’ particular genetic abnormalities occur during the carcinogenic process. It would appear that multiple genetic events are required and that the interactions between genes are more complex than may be currently appreciated. Although these events could serve as potential targets for chemoprevention, at present delivery systems for gene therapy are not refined enough to be used in clinical practice, although there has been some success in inhibiting proliferation of OAC cells by gene therapy in the in vitro setting.42 The genetic changes could however, be used as potential tissue biomarkers that could stratify the risk of progression of Barrett's to OAC.

Figure 1.

Summary of the changes that occur during the progression from normal squamous tissue to adenocarcinoma in the lower oesophagus. Preventative strategies are aimed at events on the left-hand side of this diagram because genetic alterations may result in irreversible progression to malignancy.

Arachidonic acid metabolism in OAC

The correlation between arachidonic acid metabolism and carcinogenesis is suggested by population-based studies investigating non-steroidal anti-inflammatory drugs (NSAIDs), which target enzymes involved in arachidonic acid metabolism. For instance, long-term use of NSAIDs in rheumatic patients is related to a reduced risk of various human cancers43 including OAC. NSAIDs have also been shown to reduce the number and size of colonic polyps in Familial adenomatous polyposis (FAP).44

The NSAIDs act by interfering with the COX enzyme, which metabolizes arachidonic acid in order to synthesize prostaglandins such as PGE2, PGF2 and PGD2. Recent research has shown that there are two types of COX, denoted COX-1 and COX-2.45 COX-1 is a constitutively expressed enzyme involved in the maintenance of tissue integrity and homeostasis. COX-2 expression however, is restricted to certain tissues and may be induced in response to various stimuli including cytokines and growth factors.46 Elevated levels of COX-2 have been detected in both inflammatory47 and neoplastic48 conditions and in vitro studies have demonstrated that overexpression of COX-2 can reduce the rate of apoptosis, increase the invasiveness of malignant cells and promote angiogenesis.49, 50

Much of the evidence for the role of COX-2 in carcinogenesis is derived from work in the colon, from both analysis of neoplastic tissue in humans51 and also from animal models.52 Upregulation has also been described in Barrett's oesophagus,53 and levels of COX-2 mRNA and protein appear to increase as the metaplastic cells progress through to dysplasia and adenocarcinoma.54 Moreover, there is evidence that higher levels of COX-2 expression as measured by immunohistochemistry are associated with decreased survival in patients undergoing surgery for OAC.55

An important study by Shirvani et al. has shown a relationship between acid and bile exposure and increased expression of COX-2 in ex vivo cultures of biopsy specimens from Barrett's oesophagus.56 Although the molecular mechanisms responsible for this upregulation are yet to be elucidated, they may involve the mitogen-activated protein kinase (MAPK) signalling cascade, because the presence of MAPK inhibitors have been shown to attenuate the increased COX-2 expression as a result of acid exposure to cultured adenocarcinoma cells.57 Recently, it has been postulated that the Cl/HCO3 membrane exchange channel has a key role in activating this MAPK cascade through intracellular acidification.58 These studies lend further support to the notion that aggressive acid-suppressive treatment may reduce the carcinogenic process and that a reduction in COX-2 expression is a direct consequence.

NSAIDs and chemoprevention of OAC

There have been a number of in vitro and animal studies published supporting the possible chemopreventive effect of COX-2 inhibition in Barrett's. Buttar et al. have demonstrated that in cells cultured from endoscopic biopsies from Barrett's patients, treatment with a selective COX-2 inhibitor resulted in a significant decrease in COX-2 expression and a decrease in cell proliferation.59 Complementary work from our laboratory has shown that both synthetic and food-borne COX-2 inhibitors can suppress proliferation and induce cell cycle arrest in cultured OAC cells.60

Furthermore, NSAIDs that selectively inhibit COX-2 have been shown to reduce the formation of colorectal carcinomas in animal models. A number of researchers have carried out similar work in order to study neoplastic progression in the oesophagus. Invariably in these models surgery is used to simulate reflux and promote carcinogenesis in the lower oesophagus. In one such study oesophagojejunostomy was performed in a rat model; by adding COX-2 inhibitors to normal rat diet there was reduced progression to adenocarcinoma.61, 62

Recently, a human intervention study has demonstrated that rofecoxib, a COX-2 inhibitor, may reduce COX-2 expression and cell proliferation in Barrett's.63 This work together with the in vitro and animal studies outlined above advocate selective COX-2 inhibitors in the chemoprevention of OAC. However, there has been considerable media attention on the subject of commercially available COX-2 inhibitors. Rofecoxib (Vioxx) was withdrawn in September 2004 following findings from the Vioxx Gastrointestinal Outcomes Research (VIGOR) trial showing there to be a fivefold higher incidence of myocardial infarction in patients administered rofecoxib compared with those receiving a conventional NSAID (naproxen).64 Other more recent studies have not shown differences in cardiovascular events between other COX-2 inhibitors and conventional NSAIDs,65, 66 but there is a need for further large-scale comparative trials before any definite conclusions concerning selective COX-2 inhibitors can be reached.

Regular aspirin and other NSAIDs have also been shown to reduce the progression of Barrett's to OAC.67Table 1 is a summary of the available supportive epidemiological evidence. A recent meta-analysis has estimated a 43% risk reduction for patients on either NSAIDs or aspirin,68 and has also suggested the presence of a dose effect. In other words those taking more frequent aspirin or NSAIDs experienced the greatest protective effect. Many of the observational studies listed in Table 1 have inherent limitations, not least because not all confounding variables have been taken into account. For instance, use of aspirin and/or NSAIDs may be associated with certain patient-led behaviours that have an influence on risk. This would then lead to a false association being established between NSAIDs and cancer prevention. Such behaviours may include vitamin supplementation69 and dietary habits. Furthermore, those on aspirin may indeed be more health conscious and might be more likely to have their cancers detected than others. Finally, it is likely that those with upper GI symptoms such as heartburn and regurgitation, which are risk factors for OAC, are less likely to be prescribed NSAIDs or aspirin.

Table 1.  Available epidemiological evidence of benefit of non-steroidal anti-inflammatory drugs (NSAIDs) and aspirin in prevention of oesophageal adenocarcinoma
AuthorsType of studySizeEffect on cancer rateFollow-up (years)NSAID or aspirin
Thun et al.71Cohort635 03140% reduction6Aspirin
Funkhouser and Sharp43Cohort14 40790% reduction with occasional use12–16Aspirin
Farrow et al.72Case–control293 cancersOR 0.46 NSAID and aspirin
Coogan et al.73Case–control215 cancers (both squamous and adenocarcinoma combined)OR 1.0 for 1 year use NSAID
Langman et al.74Case–control550 cancers (both squamous and adenocarcinoma combined)OR 0.64 (at least seven prescriptions in last 3 years) NSAID
Bosetti et al.75Case–control965 casesOR 0.51 (<5 years); OR 0.33 (>5 years) Aspirin

As a direct result of both the scientific evidence borne from in vitro and in vivo animal work and in addition the abundant epidemiological evidence outlined above there are now chemoprevention trials underway to evaluate the efficacy of aspirin and NSAIDs. In the UK, the AspECT trial is currently evaluating the combination of high-dose PPI and aspirin in minimizing the risk of progression to cancer in 9000 Barrett's sufferers.70

Oxidative and nitrosative stress

Free radicals such as the hydroxyl radical (·OH), nitric oxide (NO·) and peroxynitrite (ONOO ) are short-lived but highly reactive species that can damage macromolecules including DNA. This oxidative damage to DNA is now well-established as playing an important role in the pathogenesis of many cancers.76 Acid and bile reflux directly cause inflammation, and inflammatory cells subsequently produce reactive oxygen species that can then play a part in the progression to Barrett's.77 Much of the work carried out in the oesophagus is derived from animal studies. Chen et al. have observed that supplementation with iron increases the prevalence of OAC in a surgical animal model of GER.78 This may be through catalysis of reactive oxygen species causing oxidative damage to cells at the squamo-columnar junction. The same researchers have also shown that in an equivalent model, supplementation with vitamin E inhibits development of cancer perhaps through its antioxidant properties.79

Nitric oxide can also form at high concentrations from chemical reactions within the lumen of the gut. Dietary nitrates derived mainly from nitrogenous fertilizers can be converted into nitrites by oral bacteria.80 These nitrites are present at high concentrations in saliva and are immediately converted into reactive nitrogen species by their interaction with acidic gastric juice.81 The site most favouring the generation of these nitrosating species has been identified as the cardia of the stomach,82, 83 although in those with acid reflux the two juices may meet more proximally in the lower oesophagus. McColl has recently forwarded a theory to explain the increase in OAC in those with GERD;84 these cancers are occurring at the optimal site where potentially mutagenic nitrosating species can damage surrounding epithelium and increase their carcinogenic potential. More work is needed to investigate this hypothesis, as it may reveal new ways in the chemoprevention of OAC either by reduction in dietary nitrates or by inhibition of the conversion by buccal bacteria of nitrates to nitrites.

Dietary chemoprevention of OAC

There is evidence that dietary supplementation with antioxidants may be beneficial in upper GI malignancy. Serafini et al.85 explored the importance of food-borne antioxidants in a case–control study using food intake data to calculate total antioxidant potential of foods consumed. This approach combines the effects of antioxidant vitamins with those of other biologically active vegetable constituents, including flavonoids and non-nutrient carotenoids. Whatever the source of the antioxidant activity, the researchers observed a strong, statistically significant protective effect of total antioxidant potential against gastric cancers of both the cardia and more distal sites (odds ratio: 0.65; 95% confidence interval: 0.48–0.89; for the highest quartile of antioxidant activity). Furthermore, a recent German case–controlled study has demonstrated that low intake of vitamins C and E correlated with the development of both squamous and adenocarcinoma of the oesophagus.86

Epidemiological evidence indicates that nutritional factors are strongly linked to the development of cancers of the alimentary tract,87 but there is only limited information regarding their effect in OAC. Obesity has already been mentioned as being a strong independent risk factor, and there is evidence that specific dietary components may also be correlated with the development of OAC. For example, in a population-based case–controlled study in the USA it was found that dietary fat was a strong risk factor,88 whereas dietary fibre, β-carotene, folate and vitamins C and B6 were all negatively correlated. However, a similar Swedish-based study concluded that although eating fresh fruit and vegetables was protective, the absolute risk reductions when applied to the population as a whole were relatively small.89

Despite these caveats, the conclusions from available epidemiological data are that a diet rich in green and yellow vegetables with high fruit intake is likely to be protective. This might be attributable to the effects of antioxidant vitamins as previously mentioned, to other micronutrients such as folate or selenium, or to the protective effects of anticarcinogenic secondary plant metabolites such as the flavonoids, some of which are naturally occurring COX-2 inhibitors103.

A recent study in Nebraska, which recruited 124 individuals with OAC, showed that the strongest inverse association with risk of OAC was found for the intake of fish.90 Fish has been previously shown to be protective against other GI cancers, including oesophageal, gastric and colorectal carcinomas,91 and this may be because they contain high levels of n-3 (omega-3) fatty acids.92

n-3 fatty acids

Fish oils are composed of the essential fatty acids eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA). Both EPA and DHA fall into an even larger category of polyunsaturated fatty acids (PUFAs), and are classified as n-3 fatty acids. The nomenclature of an n-3 fatty acid indicates that the first carbon–carbon double bond occurs at the third carbon atom from the methyl end of the molecule.

Both EPA and DHA can compete with arachidonic acid for the aforementioned COX enzyme and can inhibit its metabolism.93 Using an animal model, Singh et al.94 have demonstrated that a high-fat fish oil diet resulted in reduced expression of COX-2 in both colonic mucosa and colon tumours, following treatment with the carcinogen azoxymethane. Furthermore, there was a decrease in the development of colonic tumours compared with rats fed a high fat corn oil diet, which was rich in the n-6 PUFA linoleic acid. It has therefore been hypothesized that n-6 PUFAs may enhance colorectal carcinogenesis through increased production of the PGE2 and this can stimulate cellular proliferation in the colon. The inhibitory effects of n-3 PUFAs on colorectal tumours may similarly be related to inhibition of PGE2 synthesis.

In contrast to colorectal disease, little work has so far been performed on the influence of fatty acids in oesophageal adenocarcinogenesis. Sammon and Alderson95 have suggested that a high consumption of linoleic acid, leading to high levels of PGE2 in the gastric mucosa, may be a risk factor for squamous carcinoma of the oesophagus, but so far little attention has been paid to the possibility that high dietary levels of n-6 fatty acids play a role in the aetiology of OAC.

Recently, in our laboratory an in vitro study96 has shown a suppressive effect of EPA on PGE2 synthesis in a human OAC cell line (OE33). This study also demonstrated a significant suppression of COX-2 gene expression. Thus, decreased production of PGE2 may be due to reduced COX-2 synthesis in addition to competition between fatty acids at the enzymatic level. N-3 fatty acids have a much more favourable side-effect profile compared with NSAIDs, and might potentially be effective chemopreventive agents. As yet no work has been performed on the effects of omega-3 fatty acids in oesophageal neoplastic progression.

Future directions

The poor prognosis of patients diagnosed with OAC presents a challenge to the clinician, and consequently there is burgeoning interest in potential chemopreventive strategies. Considerable evidence is available of a protective effect of NSAIDs, yet because of their side-effect profile widespread use cannot be currently justified. The introduction of dietary modification or supplementation may offer an attractive alternative. Foodstuffs rich in naturally occurring COX-2 inhibitors offer an alternative to regular NSAIDs, whilst diets rich in n-3 fatty acids may prove equally effective. Crucially, there is a need for appropriately powered randomized-controlled studies to assess the long-term benefit of such strategies