Carcinoma of the oesophagus (ICD code 150) was the eighth commonest cancer in 2008.³ Worldwide there were 481 000 new cases representing 7% of the total cases of cancer. Mortality was high, with 406 000 deaths or 84% of all registered cases. Incidence varies across the world, with the highest risk in the so-called Asian ‘oesophageal cancer belt’, which extends from Northern Iran through Central Asia to North Central China. SCC predominates in these less developed countries, reflecting low socio-economic status and poor diet. Overall, the male to female ratio is 2.1:1, although there are variations. In more developed countries the incidence of SCC has declined, with age-specific rates in white males in the USA at 2.2 per 100 000.⁴ However, there have been increasing trends in some regions; for example, in Scotland rates are increasing in women and decreasing in men, possibly reflecting the changing patterns of tobacco and alcohol consumption.

Oesophageal and oesophagogastric junctional adenocarcinoma

Adenocarcinoma of the oesophagus and junction accounts for variable proportions of oesophageal cancer across the world, ranging from 0% in parts of China to 10% in Northern Europe to 48% in the UK.⁴ Adenocarcinoma of the oesophagogastric junction includes cancer located in the distal third of the oesophagus and the cardia of the stomach. Although there is a male predominance there are differences according to organ of origin. For oesophageal tumours the male to female ratio is 2.6:1 and for gastric origin tumours 4:1. Recent data from England show an increase in incidence of lower third oesophageal cancer from 8.1 per 100 000 in 1998 to 10.1 per 100 000 in 2007, although the rate of increase has stabilised since 2002.⁵ Over the same period there has been a slight decrease in cardia cancer incidence. The peak age group affected is between 50 and 60 years of age.

Gastric cancer

Gastric cancer (ICD code 151) is the fourth most frequent cancer worldwide.³ In 2008 there were 988 000 new cases with 737 000 deaths. This represents 14% of all new cases of malignancy and 10.3% of all cancer deaths. Crude numbers are still increasing in relation to demographic changes of the ageing population throughout the world. The majority of cases occur in less developed countries, where the male to female ratio is 1.8:1. This contrasts with a ratio of 1.6:1 in more developed countries. Highest incidence rates are found in Japan (male 69.2 per 100 000 and female 28.6 per 100 000). Other countries with high incidence include East Asia, Korea, Eastern Europe, and Central and South America. Although distal cancers still predominate in countries with highest incidence, there has been a fall in mid and distal gastric cancer, with a progressive increase in cardia cancer.

Inter-country variations are well known between the Far East and the West. There are, however, significant intra-country variations. These largely reflect a north to south gradient, which is particularly apparent in the northern hemisphere. In both Japan and China mortality rates in the northern provinces are almost double those in the south. Similar differences are observed in the UK, with higher standardised mortality rates in north and northwestern regions. In the southern hemisphere, however, the gradient is reversed. Indeed, the higher geographical latitudes in both hemispheres are more temperate or colder and have a higher risk of gastric cancer, thus implicating environmental and particularly dietary factors in aetiology.

Aetiology

Squamous cell carcinoma of the oesophagus

Smoking and alcohol

Smoking and alcohol are established risk factors for SCC, particularly in the West. Smokers have a fivefold higher risk than non-smokers, which doubles again in heavy smokers. There is a positive dose–response effect for both duration and intensity of smoking, although long-term use of tobacco appears to be the stronger influence. The risk reduces after stopping. In the USA tobacco consumption decreased between the 1960s and 1990s, and this is reflected in the decrease in disease incidence. However, the reduction in smoking was less apparent in the black population, in which SCC rates have remained high.

The effect of alcohol is similar to smoking, although the amount consumed appears to be the greater risk factor. Several studies have demonstrated relative risks ranging from 2.9 to 7.4 for heavy drinkers. As with smoking, abstention from alcohol does appear to decrease the risk. Alcohol and smoking seem to have both synergistic and independent effects. The mechanism of action for the damaging effect of alcohol is unclear: it may directly damage the oesophageal mucosa or increase its susceptibility to other carcinogens, or may have its effect via the secondarily associated dietary deficiencies. There also seem to be individual influences reflecting variations on genes coding for enzymes involved in alcohol metabolism (see below).

Socio-economic and dietary influences

Areas of highest incidence are those countries of low socio-economic status where poverty and malnutrition predominate. The development of SCC appears to be related to a type of chronic oesophagitis that is different from that found in the West and is often complicated by atrophy and dysplasia (see Chapter 1). It is not usually associated with gastro-oesophageal reflux and is often asymptomatic.

SCC has been associated with ingestion of very hot beverages, a family history of oesophageal cancer, prevalence of oesophagitis among siblings, and a low intake of fresh fruits and wheat flour products.⁶ Furthermore, riboflavin deficiency and vitamin A and C deficiency⁷ have been identified as risk factors that are particularly important at a young age. By contrast, vitamin C intake confers a protective benefit; Hu et al.,⁸ in a case–control study, found that 100 mg of vitamin C per day decreased risk by 39%.

Associated conditions

SCC is associated with a variety of uncommon conditions that relate to some form of inflammatory injury. Oesophageal strictures developing after ingestion of corrosive agents, particularly in childhood, are associated with a 1000-fold increase in the risk of carcinoma. There is a time delay of 20–40 years after ingestion of the corrosive, and as a result tumours are seen at a younger age than normal.

Achalasia is associated with SCC, but the magnitude of the risk is unclear. Brucher et al.⁹ report from their single institution series that the risk of developing a carcinoma in long-standing achalasia is increased 140-fold when compared with the general population. The risk appears to relate to retention oesophagitis secondary to stasis and exposure to possible carcinogens in fermenting food residue. There is a lead time of approximately 15–20 years and these cases probably warrant long-term surveillance. Treatment of the achalasia does not seem to reduce the risk.

Tylosis palmarum is a rare inherited autosomal dominant condition in which there is a very high incidence of SCC. Perhaps of greater significance is the finding of the increased risk in low-risk areas for offspring of parents with oesophageal cancer.¹⁰ There are numerical and structural chromosomal aberrations in patients with a family history not seen in those without a family history (see below).

Adenocarcinoma of the oesophagus and junctional cancers

Gastro-oesophageal reflux disease (GORD)

Gastro-oesophageal reflux is now the most common symptomatic presentation of all conditions affecting the upper gastrointestinal tract. Estimates suggest that 4–9% of all adults experience daily heartburn and up to 20% experience symptoms on a weekly basis.¹¹ Of these, 60% have no endoscopic abnormality, 30% have oesophagitis and 10% have Barrett’s columnar lined oesophagus. Many are self-treated and do not attend for further investigation, yet 80% with Barrett’s are asymptomatic. The relationship of GORD and oesophageal ACA has been evaluated in case–control studies.¹² The individual cancer risk is small because of the high frequency of GORD. Lagergen et al.¹³ have estimated the risk of developing ACA of the oesophagus by scoring symptoms of heartburn and regurgitation (alone or in combination), timing of symptoms (particularly at night) and frequency of symptoms. Among those with recurrent symptoms of reflux, the odds ratio of developing cancer was 7.7 in comparison with those without symptoms. More frequent, more severe and longer-lasting symptoms of reflux were associated with a much greater risk (odds ratio 44). The risk associated with GORD is related to the development of Barrett’s metaplasia, which is greatest among Caucasian males with a history of alcohol consumption and continuous smoking. Further detailed discussion of the role of Barrett’s in the aetiology of ACA is presented in Chapter 15.

Obesity and dietary factors

In the last 20 years the incidence of junctional cancer has increased in parallel with the epidemic of obesity. There is a three- to sixfold excess risk among overweight individuals.¹⁴ Obesity predisposes to hiatus hernia and reflux, and hence contributes mechanically to increase risk. However, data from a number of studies demonstrate an effect independent of reflux. Lindblad et al.¹⁵ have reported a 67% increase in the risk of oesophageal ACA in patients with a body mass index (BMI) greater than 25, and this increases with increasing BMI. This effect was noted irrespective of the presence of reflux symptoms.

There appears to be a sex difference in that the effect was only found in women with a BMI greater than 30, whereas in men it was observed in both overweight and obese individuals. Recently this effect in women has been confirmed, with 50% of cases of oesophageal adenocarcinoma in postmenopausal women in the Million Women study being attributed to obesity.¹⁶

Evidence is accumulating to support different types of obesity. The distribution of abdominal fat tends to be central and retroperitoneal. This acts as a potent source of growth factors, hormones and regulators of the cell cycle. Such individuals develop the metabolic syndrome, which is linked to raised serum cholesterol and triglycerides, hypertension and hyperglycaemia. In the general population the metabolic syndrome occurs in 10–20%. Power et al.¹⁷ have demonstrated that 46% of those with Barrett’s oesophagus and 36% of those with GORD have features of the metabolic syndrome.

The factors released by centrally deposited fat may have an effect on the process of metaplasia transforming to dysplasia. Vaughan et al.¹⁸ have examined the potential relationship between a series of biological markers of progression from metaplasia to cancer in obese and overweight patients. There was little evidence of change in the biomarkers in association with increasing obesity. However, abnormalities in the biomarkers were observed in individuals with high anthropometric measures of abdominal fat. The study concluded that increased BMI contributed to reflux and development of metaplasia but it was the ‘male pattern’ of abdominal obesity that was actually associated with malignant transformation.

Helicobacter pylori

The role of Helicobacter pylori infection in the aetiology of junctional cancer is unclear but appears to be evolving. Gastric infection with H. pylori is characterised by gastric atrophy and hypochlorhydria. It has been suggested that this reduction in acid production could, in association with ammonia production from urea by the bacteria, protect the lower oesophagus by changing the content of the refluxing gastric juice. In countries with an increase in junctional cancer, there has been a corresponding decrease in incidence of H. pylori infection. Furthermore, community-based approaches to eradicate H. pylori infection in the treatment of ulcer and non-ulcer dyspepsia may be inadvertently contributing to the increase in these cancers.

There is accumulating evidence that there may be two distinct types of junctional cancer reflecting the two potential sites of origin. McColl and Going ¹⁹ have suggested that one is similar to oesophageal cancer and the other gastric cancer. In a series of studies of patients with junctional cancer, they evaluated H. pylori infection from serology, gastric atrophy from pepsinogen I and II ratios, symptoms of reflux and histological subtype of diffuse or intestinal type. They also included biopsies of the distal stomach to document atrophy associated with Helicobacter. Tumours of oesophageal origin are intestinal type with no evidence of gastric atrophy or Helicobacter infection and occur in the context of reflux. By contrast, tumours of gastric origin are diffuse type or intestinal but with evidence of atrophy and Helicobacter infection and without a history of reflux (Table 2.1). Such different characteristics would imply a different carcinogenic process at the two sites and should be considered in prognosis and patient management.

Table 2.1

Oesophageal and gastric origin of junctional adenocarcinoma

Socio-economic factors

Lifestyle has an effect on the risk for junctional cancers. There is an association with lower socio-economic class but this is not as strong as for SCC. Powell and McConkey ²⁰ demonstrated that the increase of ACA of the lower third of the oesophagus and the cardia was mainly in social classes I and II – that is, in professional and managerial occupations. In addition, in a large surgical series, Siewert and Ott²¹ reported that patients with ACA were more frequently from an educated background, a characteristic not present in the population with SCC. However, the effect of socio-economic class may not be independent as, when adjusted for GORD, BMI and smoking, Jansson et al.²² found the effect to be less apparent.

Gastric cancer

The Correa hypothesis ²³ (Fig. 2.1) describes the steps in the process of malignant transformation for gastric cancer. It highlights where environmental factors stimulate changes, particularly in the development of intestinal-type gastric cancer (see Chapter 1). These include socio-economic and dietary influences, as well as exposure to carcinogens.

Figure 2.1 Correa hypothesis for gastric carcinogenesis.

Socio-economic influences

Gastric cancer is a disease of lower socio-economic groups. The incidence of tobacco smoking tends to be higher in these groups and there is a 1.6-fold risk of developing stomach cancer for smokers in comparison to non-smokers.

An excess risk has been linked to certain occupations such as coalmining. Evidence for such a relationship is circumstantial, and as certain occupations reflect social background the risk may equally reflect lifestyle, particularly dietary habits, rather than actual occupational risk.

Exposure to potentially carcinogenic agents at an early age is clearly crucial to the risk of developing both precursor lesions and subsequent gastric cancer. Evidence for this risk is available from migrant studies. Japanese migrants to the USA had a far lower rate of intestinal-type cancers than the equivalent population who remained in Japan, indicating environmental or dietary aetiology, whereas the rates of diffuse-type cancer remained the same, suggesting a hereditary component.²⁴

Diet

The prevalence of gastric cancer in poor communities reflects both malnutrition and intake of a poor-quality diet. Foodstuffs that are cheap prevail, as well as low-cost methods of food preservation and preparation. High carbohydrate intake has been implicated. Areas with a high dietary carbohydrate content have a low protein intake. Protein deficiency will impair gastric mucosal repair and indeed high carbohydrate/low protein may impair defence mechanisms against injurious agents.

Salt preservation of food was common during the early years of the 20th century throughout the world; in some landlocked parts of the world this still occurs. In such areas and in those still using salt preservation there have been high rates of gastric cancer. The consumption of salted and pickled fish is high in Japanese and Colombians and correlates with their disease incidence. On the basis that salt induces injury to the gastric mucosa it may act like high carbohydrate intake, as an initiator to allow access for more potent carcinogens. By contrast, the rapid and widespread adoption of refrigerators in the 1950s and 1960s has significantly affected the preservation of fresh foods. The reduction in mortality observed in Japan shows an inverse relationship with the increase in ownership of domestic refrigerators.²⁵

Fresh vegetables and fruit theoretically act to protect against gastric carcinogenesis. Vitamin C inhibits intragastric formation of nitrosamines from nitrite and amino precursors. Both vitamins A and E act as antioxidants within cells, as well as regulating cell differentiation and protecting the gastric mucosal barrier. However, dietary studies have failed to confirm these proposed effects. An inter-country variation in fruit and vegetable intake has not paralleled differences in gastric cancer incidence. It is possible, however, that prolonged exposure is more relevant, again supporting the philosophy of a balanced diet rather than one supplemented with a potentially beneficial foodstuff.

Helicobacter pylori

In 1994 the International Agency for Research on Cancer designated H. pylori to be a type I carcinogen ²⁶ for gastric cancer. The initial effect of H. pylori is acute inflammation. Since the infection does not resolve spontaneously, an effect is likely to persist and may proceed to chronic gastritis and associated mucosal atrophy and intestinal metaplasia, dysplasia and eventually cancer. The evidence for its role is from a number of sources. Areas of high cancer incidence have a high rate of H. pylori infection. In a prospective population-based study in Japan, 2.9% of those infected developed gastric cancer compared with none from the uninfected population; 4.7% of those infected who had non-ulcer dyspepsia progressed to cancer.²⁷ In high incidence areas H. pylori infection tends to occur early in life. These early rates of infection are linked to low income, poor education, poor sanitation and overcrowding. There has, however, been a progressive fall in rates of H. pylori serology positivity in longitudinal studies, which have paralled the decline in gastric cancer incidence.

Although the evidence for H. pylori inducing gastric cancer is convincing, not all those infected develop the disease. The risk of malignant transformation appears to be enhanced by bacterial virulence and host factors (see below). Helicobacter pylori with cytotoxin-associated gene A (cagA) appears to be associated with the greatest risk.²⁸ In the West, 60% of H. pylori infections are cagA positive compared with 100% in Japan.^29,³⁰ It is likely that H. pylori induces an environment that is susceptible to malignant transformation. It induces tissue monocytes to produce reactive oxygen intermediates, which are potent carcinogens. Infection is associated with a significant reduction in gastric juice ascorbic acid,³¹ which acts to scavenge and suppress N-nitroso compounds and oxygen free radicals. It also facilitates the proliferation of nitrosating bacteria, which promote the development of N-nitroso compounds.

Precancerous conditions

Pernicious anaemia imposes a three- to fourfold increased risk of gastric cancer compared with the normal population. Patients who have undergone gastric resection for benign disease are considered to have a greater risk, possibly because of increased alkaline reflux.

Prevention of oesophageal and gastric cancer

Prevention strategies are either primary or secondary. Primary approaches aim to prevent cancer developing, whereas secondary prevention is intended to identify precancerous processes and conditions and to intervene to prevent progression to cancer.

In oesophageal and gastric cancer primary prevention approaches are currently limited to population education to alter social habits (such as decreasing or stopping tobacco or alcohol consumption) and dietary habits (such as maintaining a diet containing fresh fruit and vegetables with a low or minimal salt intake). In addition, the need to prevent obesity is now well established. The role of H. pylori eradication is important but programmes of eradication should only be considered according to the level of risk for oesophageal or gastric cancer in the population. In populations with a high risk of gastric cancer, eradication is indicated; however, in populations in which oesophageal ACA is common, eradication may have an adverse effect. The overall benefit of these approaches would be greatly enhanced if specific markers of risk could be identified to focus prevention strategies (see Chapter 15).

Secondary prevention depends upon understanding the natural history and detection of premalignant conditions. In SCC there is limited evidence that secondary measures could be effective because of lack of understanding of the histological changes leading to cancer. In oesophageal ACA, surveillance of Barrett’s metaplasia to identify progression to dysplasia is theoretically a positive approach (see Chapter 15). Identification of p53 expression and aneuploidy in biopsies of Barrett’s has been shown to predict the risk of progression.³² In both gastric and oesophageal cancer there is a potential role for chemoprevention. Increasing levels of cyclo-oxygenase-2 (COX-2) are present in the progression of atrophic gastritis to intestinal metaplasia and gastric cancer.³³ Smoking, acid and H. pylori are all associated with COX-2 expression. Aspirin and other non-steroidal agents inhibit COX-2 and their use may act as a chemopreventive for gastric cancer. Aspirin also seems to have an effect in Barrett’s metaplasia and in combination with acid suppression may minimise progression to dysplasia. The ASPECT trial in the UK is assessing whether such a strategy can have a secondary preventive effect.³⁴

Genetics of oesophageal and gastric cancer

The majority of the available evidence for the aetiology of oesophageal and gastric cancer implicates environmental factors. In oesophageal adenocarcinoma this is consistent with the sequence of Barrett’s metaplasia to dysplasia and on to cancer, and in gastric cancer with the Correa hypothesis of atrophic gastritis, intestinal metaplasia, dysplasia and cancer. There is, however, an increasing body of evidence supporting a genetic predisposition. In oesophageal cancer there is some evidence of genetic effects from study of rare coexistent conditions such as tylosis palmaris. Epidemiology studies have shown a familial clustering in approximately 10% of gastric cancers, with 1–3% related to a hereditary gastric cancer precancer syndrome (hereditary diffuse gastric cancer). Gastric cancer is also one of the cancers in hereditary tumour syndromes. In addition, certain genes have been implicated with a greater risk for oesophageal and gastric carcinogenesis from environmental factors, suggesting a link between host genetic make-up and established aetiological agents.

Oesophageal cancer

Evidence for an inherited type of oesophageal cancer is limited. However, the rare skin condition of tylosis palmaris and familial clustering for Barrett’s cancer raise the possibility of an hereditary risk. The relationship of tylosis palmaris with SCC had been recognised from epidemiological studies. Investigation of a group of families in Liverpool including several generations has identified a specific tylosis oesophageal cancer gene.³⁵ Subsequent studies have detected this gene in 69% of cases of sporadic SCC. Recently, more specific proteins coded for by this gene have been reported to be related to poorly differentiated SCC and potentially predict for those with a poorer prognosis.³⁶

There have been a number of reports of families with Barrett’s metaplasia who have developed adenocarcinoma. In these families the frequency of Barrett’s was more than 20% and the frequency of GORD was approximately 40%. In a case–control study, 24% of those with Barrett’s, oesophageal or junctional ACA had a family history compared to 5% in the control group.³⁷ Multivariate analysis confirmed that family history was an independent risk factor with equal weighting to age, male gender, obesity and alcohol consumption. Further analysis of similar families is required to evaluate possible genetic linkages.

Studies into sporadic SCC have identified gene polymorphisms in relation to genes involved in alcohol metabolism, detoxification of environmental carcinogens and folate metabolism. Some are protective and others promote malignant transformation. Most studies in SCC have been undertaken in Japanese and Chinese populations with a high incidence of alcohol consumption.

There are very few studies examining polymorphisms in sporadic adenocarcinoma and those that have been done show weak associations with risk of cancer development. In Barrett’s it is likely that there are host polymorphisms, which interact with the environmental factors to promote progression to malignant transformation. The largest body of evidence supports a role for p53 and aneuploidy as markers of risk (see Chapter 15).

Gastric cancer

Hereditary diffuse gastric carcinoma (HDGC)

The first description of a germ-line mutation was in 1998 in three New Zealand Maori families. Mutations in the CDH1 tumour suppressor gene (the E-cadherin gene) have since been described in several families of different ethnic backgrounds. The CDH1 mutation occurs along the gene in these families as opposed to clustering in one site as observed in sporadic cases. In order to develop a common approach for HDGC, the International Gastric Cancer Linkage Consortium defined HDGC as including families with more than two pathologically proven diffuse gastric cancers in individuals under 50 and families with more than three close relatives with pathologically proven diffuse gastric cancer at any age.³⁸

HDGC is an autosomal dominantly inherited syndrome. Carriers of the CDH1 mutation have in excess of 70% lifetime risk of developing diffuse gastric cancer. Female carriers have an additional risk of lobular breast cancer in about 40% of patients. Screening for CDH1 mutations in the research setting has shown its presence in 40% of families with multiple gastric cancers and at least one diffuse gastric cancer in a member under 50. The criteria for potential screening have been extended to include a number of other combinations of presentation (Box 2.1).

Box 2.1 Criteria for screening for CDH1 in HDGC

• Three or more cases of gastric cancer at any age with at least one case of diffuse gastric cancer

• Isolated individual with diagnosis of diffuse gastric cancer < 40

• Isolated individual with both diffuse gastric cancer and lobular breast cancer

• One family member with diffuse gastric cancer and another with lobular breast cancer

• One family member with diffuse gastric cancer and another with signet-ring colon cancer

The proportion of HDGC families that have the CDH1 germ-line mutation, however, is only approximately 30%. In the remaining two-thirds either the detection methods to identify the mutation have been insensitive or there are other, as yet unidentified, HDGC susceptibility genes. A number of genes have been proposed, including those coding for other cell adhesion molecules such as β-catenin and γ-catenin and those involved in other hereditary cancer syndromes. Although some data support a role, this remains investigational.

The clinical issue for these families is their optimal management. In view of the 70% or more chance of developing diffuse gastric cancer with its attendant poor long-term survival, the options are either prophylactic total gastrectomy or endoscopic surveillance. There have been a number of reports of the pathology of resected stomachs after prophylactic gastrectomy. In clinically, endoscopically ‘normal’ stomachs, supported by biopsy, diffuse multifocal intramucosal disease was identified in all specimens.^39,⁴⁰ Some studies have shown concentration of disease in the distal third, whereas others have more widespread involvement.⁴¹ The question arises, however, as to the appropriateness of total gastrectomy in essentially a young population for whom the nutritional sequelae will be lifelong. Furthermore, the morbidity and mortality of such a procedure must be minimal. Therefore the role for endoscopy must be explored. Standard surveillance is limited but the distribution of disease from the pathology studies indicates where biopsies should be concentrated. Advances in endoscopy, which include endoscopic autofluorescence spectroscopy and chromoendoscopy, have the potential to enhance accuracy. A further point is that 20–30% of CDH1 germ-line mutations do not progress to clinically diffuse gastric cancer. There are some data to suggest intramucosal disease may not progress and be of biological rather than clinical importance, analogous to prostatic cancer in elderly men. Thus, counselling of individuals from HDGC families produces very difficult questions, particularly as knowledge is incomplete as to risk, most appropriate management and the role of genetic intervention, as well as the sequelae of life following total gastrectomy.

Hereditary cancer syndromes

The development of molecular genetics has allowed confirmation of primary genetic aetiology for a spectrum of cancers which epidemiology studies had suggested were inherited (Table 2.2). Gastric cancer has been found to be coexistent in these syndromes, further supporting a genetic basis for its development.⁴² There are differences across the world, consistent with evidence that the gene pool varies within different populations. In patients with familial adenomatous polyposis (FAP) there is an excess of gastric cancer in Japanese families that is not observed in US non-oriental families. Similarly, in the Lynch syndrome, gastric cancer is more common in China and Korea yet rare in Caucasians. Thus, screening surveillance in such populations should be directed accordingly. Unless there is gastric cancer in the family then upper gastrointestinal endoscopy is not routinely required in those with the Lynch syndrome.

Table 2.2

Hereditary cancer syndromes

Syndrome	Main tumours	Associated tumours
Lynch syndrome (hereditary non-polyposis colorectal cancer)	Colon carcinoma	Endometrial, gastric, small bowel and urothelial cancer
Li–Fraumeni syndrome	Breast cancer, osteosarcoma, brain tumours, soft tissue sarcoma	Gastric and colon cancer, adrenocortical carcinoma, haematological and gynaecological
Familial adenomatous polyposis coli	Colon cancer	Gastric cancer, papillary thyroid cancer, desmoid tumours, medulloblastoma and hepatoblastoma
Peutz–Jeghers syndrome	Hamartomatous polyps of the small bowel, colon and stomach	Gastrointestinal carcinomas, breast, testicular and ovarian cancers
Juvenile polyposis	Hamartomatous polyps of the colon and occasionally stomach and small bowel	Gastrointestinal cancer

Moderate cancer risk

Worldwide studies have shown that approximately 5–10% of patients with gastric cancer have a family history but without other features to suggest an inherited aetiology. However, it is possible that in this population there is some hereditary predisposition to increased susceptibility to environmental factors such that their risk is increased. Studies have shown increased rates of H. pylori infection with atrophic gastritis and hypochlorhydria in first-degree relatives of gastric cancer patients compared with normal controls. This could of course be purely due to environmental factors. Alternatively, normal variations in the genetic coding sequence of multiple genes (polymorphisms), which are inheritable, may lead to differential inflammatory responses to agents such as H. pylori or tobacco. Thus, the combined effect of inflammation promoting host genetic polymorphisms and different microbiological genotypes such as CagA H. pylori may increase the risk in a particular population. Specific studies including p53 have shown certain polymorphisms to be associated with the production of variant proteins.⁴³ These have been identified more frequently in patients with diffuse gastric cancer than in matched controls. DNA polymorphism in the interleukin-1 gene cluster has been associated with a response to H. pylori infection. It is postulated that the polymorphism increases the production of interleukin-1β, a proinflammatory cytokine, which inhibits gastric acid secretion and hence achlorhydria and gastric atrophy.

Molecular genetics of oesophageal and gastric cancer

The development and progression of oesophageal and gastric cancer has been clearly demonstrated in numerous studies to have a genetic basis. Alterations in tumour suppressor genes and oncogenes have been identified in both cancers. Specifically genes, which have roles in diverse functions such as cell adhesion, signal transduction, differentiation, development, gene transcription or DNA repair, have been demonstrated in both oesophageal and gastric cancer. Figure 2.2 shows some of the changes described in oesophageal cancer arising in Barrett’s metaplasia, and Fig. 2.3 shows the changes in gastric cancer and highlights different mechanisms for the intestinal and diffuse types. Studies of cDNA microarrays for gastric cancer have reported characteristic patterns of gene expression in chronic gastritis, intestinal metaplasia, and intestinal and diffuse gastric cancer. These raise opportunities for identification of molecular markers and gene profiling in cancer progression and for the prediction of prognosis and treatment sensitivity.^42,⁴⁴ Studies of host genetic factors are likely to provide vital information to explain the diverse risks in differing populations. This may require whole genome sequencing studies but these may be limited by the capability of bioinformatics to cope with the associated vast amounts of data.

Figure 2.2 Genetic changes described in the progression from Barrett’s metaplasia to oesophageal adenocarcinoma. Reproduced from Lin J, Beer DG. Molecular biology of upper gastrointestinal malignancies. Semin Oncol 2004; 31:476–86. With permission from Elsevier.

Figure 2.3 Genetic alterations described in gastric cancer. Abbreviations: APC, adenomatous polyposis coli; Bcl-2, B-cell CLL/lymphoma 2; CD44, CD44 antigen; CDC25B, cell division cycle 25B; c-erbB2, v-erb-B2 erythroblastic leukaemia viral oncogene homologue 2; c-met, met proto-oncogene (hepatocyte growth factor receptor); DCC, deleted in colon cancer; K-ras, v-Ki-ras 2 Kirsten rat sarcoma viral oncogene homologue; K-sam, encodes fibroblast growth factor receptor 2; MSI-H, microsatellite instability – high; nm23, non-metastatic cells 1 (protein, NM23, expressed in); p53, tumour protein p53 (Li–Fraumeni syndrome); SIP-1, SMAD-interacting protein 1; TERT, telomerase reverse transcriptase; TWIST 1, twist homologue 1. Reproduced from Keller G, Hofler H, Becker K-F. Molecular medicine of gastric adenocarcinomas. Expert Rev Mol Med 2005; 7:1–13. With permission from Cambridge University Press.

Screening for oesophageal and gastric cancer

Screening programmes for any disease are dependent on a number of criteria. Firstly, the disease must be common in the target population. Secondly, a reliable and accurate test that is as sensitive and specific as possible is required, and the test should be acceptable to the screened population. There should be an effective treatment for the screened abnormality with minimum morbidity and mortality. Finally, not only does the treatment need to show an improvement in results, but implementation of the screening programme should also result in an overall benefit for the screened population.

The worldwide differences in incidence of oesophageal and gastric cancer allow the implementation of screening programmes for asymptomatic populations only in those areas where the incidence is high. However, lessons from these programmes have increased knowledge of natural history and have allowed high-risk groups to be targeted in low-risk areas in order to detect disease at an earlier stage.

Asymptomatic screening

Oesophageal cancer

Evaluation of asymptomatic screening for carcinomas of the oesophagus has centred on those parts of China with the highest incidence. The screening test involves swallowing a small deflated balloon, which is then inflated at the lower end of the oesophagus. The balloon surface is covered with a fine mesh; on withdrawal from the oesophagus, this scrapes the mucosa to collect cells. A cytological smear is then made from the scrapings for microscopic examination. Those individuals found to have abnormalities are then subjected to endoscopy and appropriate biopsy. In 132 subjects with early oesophageal cancer detected in this way, 26% had normal radiological appearances.⁴⁵

The efficiency of this technique has had varying reports. Reviewing data based on 500 000 examinations, Shu ⁴⁶ suggested an accuracy for the differentiation of benign from malignant of 90%. Mass surveys have shown that 73.8% of detected cancers were either in situ or minimally invasive. In a provincial review, Huang⁴⁷ reported on 17 000 examinations screened during a 1-year period. Abnormalities were found in 68% of the population, with low-grade dysplasia in 37%, high-grade in 26% and in situ cancer in 2%. A group with high-grade dysplasia were followed for up to 8 years. Regression to normal or low-grade change was observed in 40%, 20% remained as high grade, 20% fluctuated between high and low grade, and 20% developed cancer. In the absence of dysplasia, 0.12% developed cancer. Progression from dysplasia to in situ cancer occurred over 3–12 years and from in situ to invasive cancer over 3–7 years. Tumour risk was consistent, with a known distribution of middle-third chronic oesophagitis in 76%. It would seem that the duration of severe dysplasia is the greatest risk for malignant transformation. Follow-up by endoscopy is therefore important and in order to ensure biopsy of the same site vital stains have been used. Huang⁴⁸ reported that staining with toluidine blue was effective for identifying neoplastic epithelium; 84% of cancers were identified in positively staining areas.

The problem associated with this approach is the management of dysplasia. Oesophageal dysplasia is a dynamic process with both spontaneous regression and progression. Furthermore, even if in situ cancer develops, progress to advanced disease is often prolonged and may be associated with prolonged survival. In one series of 23 untreated patients, 11 developed late-stage disease at a mean of 55 months. In the remainder there was no change for over 6 years and the 5-year survival of the group was 78%.⁴⁹ Five-year survival needs to be considered with caution as detection of asymptomatic slowly progressive disease introduces lead-time bias and this can falsely give the impression that treatment results for screen-detected cases are better.

As a result an International Union Against Cancer (UICC) recommendation has been to limit oesophageal cancer screening to areas of high risk.⁵⁰ The aim is to identify the natural history of dysplasia more completely. Common standards are required for the classification of dysplasia to identify those changes with greatest risk. Once the assessment is more reliable, control studies should be developed to determine whether screening intervention could reduce mortality for oesophageal squamous cell cancer.

Gastric cancer

The prominence of gastric cancer as a public health problem in Japan led to the development during the 1960s of a mass screening programme for all men over the age of 40 years. The programme has been based on double-contrast radiology with endoscopic assessment of any abnormalities.⁵¹ Members of the public are invited to undergo radiology in mobile units at which seven films are taken after the ingestion of an effervescent contrast agent. Screening is undertaken annually or biannually depending on the area of Japan and the associated risk of disease. Government recommendations set a target of 30% for the annual examination rate. Despite the recognition of gastric cancer as a public health problem, attendance for screening is low. In 1985 over 5 million were examined, representing 13% of the at-risk population. Therein lies one of the problems with any screening programme, namely the cooperation of the public.

Approximately half the cases diagnosed are limited to the mucosa or submucosa (early gastric cancer). Half of those detected are symptomatic and an alternative approach could be envisaged. In keeping with the criteria for a screening programme there has been a highly significant decrease in mortality. However, as already discussed, there may be other reasons for the decline in mortality.

Oshima et al.⁵² compared screened and unscreened populations to determine whether screening was important over and above the other influences on the decrease in mortality. In a case-controlled study they found that the risk of dying from gastric cancer among screened cases was at least 50% less than that for non-screened cases. Other Japanese groups have reported similar results. However, the actual effect on mortality remains to be proven as none of the studies have been randomised or controlled. As a result the UICC recommended that studies should be continued in Japan to resolve the problem, but screening in this way should not be adopted as public health programmes in other parts of the world.⁵⁰

Symptomatic screening and early detection

The rate of dyspepsia and reflux in the general population and the non-specific nature of the symptoms do not justify endoscopic assessment for newly presenting patients of all ages. In the UK, in order to improve rates of early diagnosis a programme under the National Awareness and Early Diagnosis Initiative (NAEDI) is currently under way. This includes increasing patient awareness of symptoms and reporting them rather than self treating, improving primary care and pharmacist awareness by careful review of repeat prescriptions for acid-suppressing medication and ensuring adequate resources for evaluation of symptomatic patients.

Studies have evaluated methods of selecting those potentially at higher risk of having a significant diagnosis. Dyspepsia has been classified as uncomplicated or complicated by alarm symptoms including weight loss, anorexia, vomiting, dysphagia and signs of anaemia or an abdominal mass. Further classification according to age has also been studied as early gastric cancer tends to present approximately 10 years younger than advanced disease.⁵³ Although such studies have increased rates of detection of early gastric cancer to approximately 15–20%, many patients with uncomplicated dyspepsia have undergone normal examinations. In a series of 25 patients under 55 years with gastric cancer, 24 had complicated dyspepsia.⁵⁴ Furthermore, in a population database of 3293 oesophageal and gastric cancers, 290 were under 50 and 21 (7%) had uncomplicated dyspepsia.⁵⁵ The simple conclusion of this evidence is to restrict endoscopy for those under 55 years to complicated dyspepsia. However, the alarm symptoms used to define complicated dyspepsia are those of established locally advanced disease with the expected poor prognosis. This has been confirmed in a large case series of open access endoscopy from Newcastle upon Tyne.⁵⁶ It could be argued that the low index of suspicion for the significance of simple dyspepsia in younger patients had led to a delay in investigation until they developed more significant or alarm symptoms. The failure to diagnose earlier cancers in younger patients may be a result of a failure to initiate investigations until the cancer is advanced and raising the age threshold to 55 for uncomplicated dyspeptics would decrease the rate of diagnosis of upper gastrointestinal cancer. Indeed, the effect of early intervention in unselected dyspepsia not only increases the rate of earlier diagnoses of cancer, but this is also translated into a survival advantage (Fig. 2.4).⁵⁷

Figure 2.4 Survival after early detection of gastric cancer (study population) compared with historical control population (pre-study population).

A pragmatic approach has been adopted in the UK.⁵⁸ Urgent specialist referral or endoscopic investigation (within 2 weeks) is indicated for people with dyspepsia of any age when presenting with chronic gastrointestinal bleeding, progressive unintentional weight loss, iron-deficiency anaemia, progressive dysphagia, persistent vomiting, epigastric mass or suspicious barium meal. In addition to these alarm symptoms, similar referral is required for a dyspeptic patient over 55 years with onset of symptoms within the last year and/or continuous symptoms since onset. The advantage of referral within 2 weeks is largely procedural and has only limited support from the literature. For example, gastric cancers limited to the mucosa and submucosa have a doubling time of 1.5–10 years, whereas advanced disease has a doubling time of between 2 months and 1 year.^59,⁶⁰ Reducing symptomatic delay is unlikely to alter outcome for early disease significantly, but may render more advanced disease amenable to resection. In a comparative audit of 2-week referrals (TWRs) with conventional presentations, Radbourne et al.⁶¹ have found that although the TWR produced more cancers, the stage of disease was equivalent at diagnosis and survival was comparable between the two groups.

High-risk groups

GORD and Barrett’s oesophagus: A variety of approaches have been assessed for early diagnosis in patients with GORD and at risk of Barrett’s oesophagus. Small-calibre nasal endoscopy in unsedated patients has been evaluated for accuracy of detecting Barrett’s in patients with GORD. It was found to be well tolerated, technically feasible and accurate, despite producing smaller biopsy samples.⁶²

A modification of the Chinese wire mesh covered balloon used to screen for SCC has recently been reported as a non-endoscopic screening test for Barrett’s oesophagus in a population with a history of GORD.⁶³ The device, Cytosponge, has a similar design in which the patient swallows a gelatin capsule attached to a length of string. The capsule contains a compressed mesh, which expands to 3 cm in diameter once the outer gelatin coating dissolves in the proximal stomach. The string is withdrawn and the mesh takes samples of cells from the oesophageal mucosa. The mesh is processed to produce a cytological preparation, which is stained with a marker for Barrett’s oesophagus. Further evaluation is required of this novel technique to determine if this has a role in selecting those with GORD for surveillance.

Helicobacter pylori: The role of H. pylori as a marker for endoscopy has received considerable attention. Both serological estimation and breath tests depending on exhalation of urea have been investigated. Serology has been assessed for concordance with the underlying histological presence of H. pylori. Farinati et al.⁶⁴ found 82% agreement between a measurable antibody response and histological evidence of H. pylori infection. Urea breath tests are in routine use in Helicobacter eradication programmes for duodenal ulceration. Again, the problem is one of specificity and sensitivity.

Helicobacter pylori seropositivity does not necessarily imply active infection. Equally, seropositivity is a common finding and may not be specific for the at-risk population. It increases with age and to a certain extent parallels gastric atrophy, which is equally an age-related phenomenon and in the majority does not progress to cancer. There is also evidence that seroreversion may occur, with seropositivity frequently seen in early gastric cancer and seronegativity in more advanced disease.⁶⁵ Whiting et al.⁶⁶ reported a retrospective analysis of H. pylori seropositivity in cancer patients compared with a group of undiagnosed dyspeptics. Although the cancer patients were significantly more likely to be seropositive, this was very much site related; cardia cancers were not usually seropositive. Thus, any screening programme based on H. pylori serology could miss proximal tumours, which are currently the more common cancers. Further investigation is required and longitudinal studies may resolve the issue of whether patients with H. pylori seropositivity warrant close endoscopic follow-up.

Gastric atrophy and intestinal metaplasia: Those found at endoscopic biopsy to have gastric atrophy and columnar-type gastric intestinal metaplasia may also form a risk group. Whiting et al.⁶⁷ have followed a group of patients by annual endoscopy who were found to have chronic atrophic gastritis and intestinal metaplasia at diagnostic endoscopy for dyspepsia. This group was reported to have an 11% risk of developing gastric cancer and the authors suggest that such patients should be considered a high-risk group.