Max Almond, Hugh Barr and Janusz Jankowski

Definition

Barrett’s oesophagus is a change in any portion of the normal squamous oesophageal epithelium to a metaplastic columnar epithelium that is visible endoscopically and can be confirmed or corroborated histologically.^1,2 There are three histologically distinct types of columnar metaplasia: intestinal (IM), cardiac (CM) and fundic. In the USA, unlike the UK and Japan, the diagnosis of Barrett’s oesophagus requires the identification of intestinalisation characterised by the presence of goblet cells. However, the UK definition considers that Barrett’s oesophagus is analogous to a ‘columnar lined oesophagus’ and does not require identification of goblet cells due to fears that sampling bias could lead to under-diagnosis and potentially exclude patients from surveillance programmes. It has been reported that a minimum of eight biopsies are required to confidently exclude intestinal metaplasia – if only four biopsies are taken the diagnostic yield is only 35%.³

Occasionally, biopsies will be histologically diagnostic for Barrett’s oesophagus in that they contain a native oesophageal gland or, more usually, a duct from these glands in close juxtaposition to metaplastic mucosa. However, the superficial nature of most biopsies makes this unusual. More typically, columnar epithelium is endoscopically recognisable but must be correlated with the location from which the biopsy is taken, as intestinal-type mucosa may also be found at the gastric cardia and fundus. Histologically, these biopsies can only be said to be corroborative of an endoscopic diagnosis of Barrett’s oesophagus. Thus Barrett’s oesophagus is a clinicopathological diagnosis.

Epidemiology

The exact population prevalence of Barrett’s oesophagus is unclear. Data described in post-mortem and endoscopic series range from 0.9% to 5.6% depending on the precise definition used and the type of study.4–7 It is likely that the true prevalence in the West is around 2%. When extrapolated to the UK and US populations, conservative estimates of prevalence are 1 million and 4 million affected individuals, respectively.⁸ There is also some evidence that the incidence of Barrett’s oesophagus in the West is increasing by up to 2% per year.^5,9–11 Data from the Netherlands demonstrated an increase in the number of cases of Barrett’s oesophagus despite a decrease in the number of endoscopies being performed over the same period, suggesting a true increase in incidence.¹¹

The incidence of Barrett’s oesophagus increases with age, the mean age at diagnosis being approximately 62 years for men and 68 years for women. It predominantly affects Caucasians¹² and is more common in men than women, with a ratio of approximately 1.7:1.¹³

The risk of developing Barrett’s is related to increased frequency and duration of reflux symptoms.¹⁴ This appears to correlate with the well-known association between increased frequency, duration and severity of reflux symptoms, and increased risk of adenocarcinoma of the oesophagus. The incidence of Barrett’s oesophagus in patients with symptomatic gastro-oesophageal reflux disease (GORD) is between 5% and 12%.^9,15 Evidence from one case series suggests that more than 60% of patients with Barrett’s oesophagus develop the condition secondary to chronic GORD, although other causes of oesophagitis, including non-steroidal anti-inflammatory drugs (NSAIDs), chemotherapy and viral infections are also associated with the disease. It does raise an intriguing possibility that a smaller proportion of patients can develop Barrett’s de novo in the absence of obvious symptomatic or perhaps even pathological reflux. Therefore, other factors that may catalyse changes at the oesophagogastric junction (OGJ) are obesity and cigarette smoking,which have been identified as risk factors for both Barrett’s oesophagus and progression to malignancy.¹⁶

A Swedish case–control study demonstrated that patients with recurrent reflux symptoms, when compared with asymptomatic patients, had an odds ratio of 7.7 for oesophageal adenocarcinoma and 2.0 for adenocarcinoma of the gastric cardia. Patients with severe long-standing symptoms had an odds ratio of 43.5 and 4.4 for oesophageal and cardia adenocarcinoma, respectively.¹⁷

Endoscopic assessment

Barrett’s oesophagus has a classical appearance at oesophagogastroduodenoscopy (OGD). There is proximal displacement of the squamocolumnar junction, which in normal circumstances lies at the proximal limit of the linear gastric mucosal folds. ‘Salmon pink’ columnar mucosa is seen in the distal oesophagus arising from the OGJ, often with characteristic tongue extensions and/or columnar islands.

Proximal extension above the OGJ should be measured and documented, taking care to accurately identify any sliding hiatus hernia that may confuse this measurement. It is crucial that biopsies originate from the oesophagus to prevent misclassification of cardiac intestinal metaplasia as Barrett’s oesophagus.

It is crucial to make a thorough and systematic inspection of the mucosa in order to identify any macroscopic neoplastic disease. Water or 1% acetylcysteine should be used to remove blood, saliva and refluxate from the oesophagus, and sufficient insufflation should be ensured to clearly visualise any mucosal abnormalities. Particular care must be taken to identify the OGJ in patients with a hiatus hernia as it is easy to miss the distal extent of a Barrett’s segment in these patients. Clinicians should be aware that at endoscopic inspection most areas of early neoplasia and cancer are detected in an area around the 2 to 4 o’clock position in the endoscopist’s view.¹⁹

This rigorous biopsy protocol, which is often poorly adhered to outside of specialist centres, samples less than 5% of the mucosa and may miss up to 57% of dysplasia.^20,21 Advanced endoscopic imaging techniques may allow targeted biopsies from high-risk areas, improving diagnostic yield (Table 15.1).^22–26 Potentially, these imaging tools may also facilitate targeted endoscopic resection of high-grade dysplasia (HGD) and intramucosal cancer.

Pathophysiology of Barrett’s oesophagus and progression to adenocarcinoma

It is currently believed that Barrett’s metaplasia develops as a mucosal ‘adaptive’ response to increased cell loss as a result of chronic inflammation, secondary to GORD. Oesophageal squamous epithelium is highly sensitive to acid, alkaline and biliary reflux, which all cause inflammation, with cell loss, necrosis and ulceration. There is strong evidence that the site of origin of Barrett’s metaplasia is a progenitor stem cell located in the submucosal oesophageal gland ducts, following demonstration that a p16 point mutation originating in microdissected squamous duct tissue was also present in adjoining metaplastic crypts.²⁷ Duodenal and gastric reflux-induced ulceration and inflammation is believed to induce tumour suppressor gene mutations, typically p53 and p16, in some of the stem cell populations located in oesophageal gland squamous ducts, which are present throughout the entire length of the oesophagus. Following this initiation phase, multiple distinct clones of metaplastic tissue compete to colonise the oesophagus, creating a mosaic pattern of clones across the segment. Clonal expansion of populations with greater selective advantage, such as ability to survive in a markedly acid- or bile-rich environment, leads to dominant and widespread clones. Once initiated, the promotion and propagation of metaplastic clones is dependent on the surrounding microenvironment, particularly the presence of a chronic inflammatory cell infiltrate, characterised by T lymphocytes, and cytokines such as interleukin-1, tumour necrosis factor-α and transforming growth factor-β. These lead to an increase in cyclo-oxygenase-2, c-myc and cyclin D1, which increase proliferation and decrease apoptosis, and a reduction in E-cadherin, with resultant loss of cell adhesion and localisation of β-catenin to the nucleus.²⁸ These molecular changes underlie the progression of Barrett’s oesophagus to cancer via the metaplasia–dysplasia–adenocarcinoma sequence (see Fig. 15.1).

The presence of dysplasia is regarded as the best marker for malignant transformation in the epithelium. Dysplasia is classified histologically: HGD is diagnosed when there are distinct cytological changes, particularly nuclear pleomorphism and loss of crypt architecture. Low-grade dysplasia (LGD) is more difficult to classify; there is loss of cellular differentiation and loss of goblet cells but with milder changes than those seen in HGD. Intramucosal cancer is said to have occurred when there is invasion through the basement membrane into the lamina propria. The term carcinoma in situ has been abandoned.

Although traditionally thought of as an acquired condition, genetic factors may play a part in a small proportion of patients with Barrett’s metaplasia, as family and twin studies suggest a subgroup of individuals with a strong familial tendency to Barrett’s oesophagus.^29,30 A family in the UK has been identified with a male index case with oesophageal adenocarcinoma, three brothers with Barrett’s-associated cancer or HGD, and six children with Barrett’s oesophagus.³¹ Linkage studies are being undertaken in order to further our understanding of this genetic inheritance. However, there are data to suggest Barrett’s is a polygenic disease with multiple contributing genes acting together.

Several studies have used a candidate gene approach to attempt to identify genetic variants in inflammatory and DNA repair pathways that could account for host susceptibility. Furthermore, results from ongoing genome-wide association studies of Barrett’s oesophagus and oesophageal adenocarcinoma will direct further study towards particular loci of interest in the future.

Risk of cancer and mortality in Barrett’s oesophagus

Barrett’s oesophagus is accepted as a significant risk factor for adenocarcinoma of the oesophagus, although the risk of progression to adenocarcinoma and the risk of disease-specific mortality is low. A large number of studies have estimated the risk of adenocarcinoma arising from Barrett’s oesophagus, with very variable results.32–42 Former studies included small numbers of patients and were likely subject to publication bias, with results only being published if they showed a high incidence of cancer, leading to an overestimate of risk.⁴³ Recently, two large population-based cohort studies have reported much lower annual rates of progression to adenocarcinoma (0.12–0.13% per year) than in former series (Table 15.2). It should be noted that these figures exclude carcinomas of the gastric cardia and also do not reflect progression to HGD. In addition, the two studies used different approaches to select patients with Barrett’s oesophagus. The study by Hvid-Jensen et al.³³ identified patients with intestinal metaplasia (IM) from the Danish National Pathology Registry without corroboration with endoscopic findings. Therefore, potentially, patients may have been included who had a diagnosis of cardiac IM rather than true Barrett’s oesophagus, producing an incorrect denominator and leading to a slight underestimate of the risk of disease progression. Bhat et al.³² included patients with columnar-lined oesophagus (CLO) at endoscopy (although the validated Prague system was not used), which was corroborated histologically, and demonstrated an increased risk of progression in patients who had IM confirmed histologically at index endoscopy, compared to those with CLO without IM. This finding is in keeping with previous studies demonstrating a higher risk of disease progression in patients with confirmed IM.^1,44,45

Previous studies have, however, suggested a significant geographical variation in the incidence of carcinoma arising in Barrett’s oesophagus in Western countries, with incidence rates in the UK almost double those in the USA.⁴⁶ It is also worth noting that the population demographic in Denmark differs somewhat from the USA and UK, where rates of obesity are significantly higher and where a greater proportion of men, who are at higher risk of malignant progression, develop Barrett’s oesophagus.

A recent meta-analysis reported a pooled estimate of the annual risk of cancer progression in non-dysplastic Barrett’s of 0.39% per year. Importantly, only eight of 47 studies that met all three quality criteria were included in this analysis; inclusion of the remaining studies significantly increased this figure.³⁵ The risk in the UK following a meta-analysis is indicated as closer to 1%, higher than is reported in the USA.⁴⁶

Natural history of dysplasia in Barrett’s oesophagus

When considering the natural history of dysplasia in Barrett’s oesophagus we must remember that in addition to potential problems with length of follow-up and sampling error at endoscopy, there is considerable inter- and intra-observer variation among experienced pathologists in the histological diagnosis of dysplasia. While pathologists can demonstrate acceptable levels of agreement in distinguishing HGD combined with carcinoma from no dysplasia combined with indefinite and low-grade dysplasia (kappa values of 0.8), there are much poorer levels of agreement in distinguishing between the four groups: indefinite for dysplasia, LGD, HGD and carcinoma (intra-observer kappa values of 0.43–0.64).⁴⁸ Pathologists find it particularly difficult to separate inflammation in Barrett’s oesophagus from LGD. In this situation pathologists should be encouraged to make use of the indefinite for dysplasia category: such a diagnosis does not mean that the pathologist is uncertain, but rather that it is not possible, with confidence, to exclude LGD in inflamed material. The diagnosis of HGD has serious implications for patient management and the diagnosis should be confirmed by two expert pathologists.

Risk factors for progression to cancer

The length of Barrett’s segment has been shown to be a significant risk factor for progression to cancer, a doubling of length increasing the risk 1.7-fold.⁷⁷ The extent of HGD and/or LGD also appears to be a risk factor for progression to adenocarcinoma.^55,78

Importantly, in a prospective longitudinal cohort study, individuals with Barrett’s oesophagus who were regularly taking aspirin or other NSAIDs were found to have a significantly lower 5-year cumulative incidence of adenocarcinoma compared with individuals not taking NSAIDs (6.6% and 14.3%, respectively), suggesting that this may be an effective chemotherapeutic intervention.⁷⁹ An ongoing phase III multicentre randomised controlled trial (RCT), the AspECT trial (Aspirin and Esomeprazole Chemoprevention in Barrett’s Metaplasia), designed to test this hypothesis is due to report in 2016. The primary aim of this study is to determine whether acid suppression with proton-pump inhibition (high dose vs. low dose) with or without aspirin can reduce mortality or the conversion from Barrett’s metaplasia to HGD or adenocarcinoma. Both high- and low-dose acid suppression are being investigated as there remains doubt about the optimal dose of proton-pump inhibitor (PPI) to use, especially given the fact that Barrett’s mucosa is relatively insensitive, thus rendering symptoms unreliable. There is an argument that incomplete acid suppression might increase the risk of cancer by exposing the mucosa to short pulses of acid, thus stimulating the proliferation of abnormal cells. In contrast, there is some epidemiological evidence that high-dose proton-pump inhibition might increase the risk of cancer as bile acid might become cytotoxic at neutral pH. In addition, there have been fears that PPI-induced hypergastrinaemia could stimulate hyperproliferation of Barrett’s epithelium.^80,81 Although this risk is yet to be evaluated in vivo, it appears more likely that gastrin induces epithelial restitution in Barrett’s oesophagus, without stimulation of clonal expansion or disease progression.⁸²

Screening for Barrett’s oesophagus and adenocarcinoma using molecular markers

It is accepted that GORD is a significant risk factor for the development of adenocarcinoma, with a well-known Swedish case–control study demonstrating a 44-fold increased relative risk in individuals with frequent heartburn of greater than 20 years’ duration.¹⁷ This has led to the suggestion that screening individuals with chronic reflux symptoms to detect Barrett’s oesophagus and cancer may be of benefit. However, it is important to appreciate two flaws in this concept: firstly, approximately 40% of individuals with cancer in the series mentioned above denied frequent heartburn; secondly, a significant proportion of individuals with Barrett’s oesophagus are asymptomatic. In addition, Barrett’s patients experience less heartburn and use PPIs less frequently compared with controls.^2,83,84

The endoscopic screening of individuals with chronic reflux symptoms to detect either Barrett’s or cancer is not currently recommended in the UK or USA.^1,2 This is because of the low absolute risk of developing adenocarcinoma in individuals with chronic reflux, combined with the knowledge that most individuals with Barrett’s oesophagus die from causes other than oesophageal cancer. There are also concerns about the cost-effectiveness and invasiveness of endoscopy as a screening tool.

Several attempts have been made to develop a scoring system using patient demographics and symptoms to predict the presence of Barrett’s oesophagus for screening purposes.^85,86 However, interest in these risk prediction strategies has declined due to inability to generate sufficient sensitivity and specificity.

It is hoped that future non-invasive molecular screening tests might be developed to detect patients with Barrett’s oesophagus who display phenotypes that could act as markers of disease progression. Promising techniques include DNA microarrays (measuring genome-wide alterations in DNA copy number), single nucleotide polymorphism (SNP) arrays (detecting allelic imbalances) and measurement of hyperproliferation, which occurs as a sequel of genetic mutation.

Mutations in the p53 tumour suppressor gene are widely found in dysplastic Barrett’s oesophagus and oesophageal cancer. Younes et al.⁸⁷ found p53 mutation in 9% of Barrett’s patients with LGD, 55% of patients with HGD and 87% of patients with carcinoma: no patients without dysplasia had a p53