The gastrointestinal tract runs from the mouth to the anus, and includes the oesophagus (gullet), stomach, small bowel or intestine, and the large bowel (colon and rectum). Cancer can affect any part of the gastrointestinal tract, although, curiously, it is rare in the small intestine where most digestion takes place.

• Bowel (colorectal) cancer is the second commonest cause of cancer-related death (after lung cancer), affecting 6 per cent of the population in Westernised (industrialised) countries and causing death in about 3 per cent. About 25 per cent of all deaths are caused by cancer in industrialised countries, and bowel cancer accounts for 12.5 per cent of those deaths.
• Stomach cancer accounts for about 8 per cent of cancer deaths and occurs in twice as many men as women. It is gradually becoming less common but, sadly, this fall has been balanced by an increase in oesophageal cancer (particularly a glandular type called adenocarcinoma) in men.
• Oesophageal cancer accounts for about 3 per cent of cancer deaths and the ratio of affected men to women is 1.8:1.
• Cancer of the pancreas accounts for about 4 per cent of cancer deaths and affects both sexes approximately equally.
• Cancer that develops first in the liver (primary liver cancer) is strongly linked with hepatitis virus infection. It is uncommon in Western countries where the rate of hepatitis is relatively low, but is often the most common cause of cancer-related death in developing countries where hepatitis is much more frequent. Overall it accounts for only about 0.7 per cent of cancer deaths in Western countries. However, the liver is a common site for other gastrointestinal cancers to spread to, particularly colorectal cancer, resulting in so-called secondary deposits or metastatic cancer.

Can screening tests detect gastrointestinal cancers before they cause symptoms?

Screening is the process of checking people who have no symptoms for unsuspected disease, which can then be treated more successfully than if the disease had been left until it showed itself. Screening programmes are already in place for breast cancer and cervical cancer in women. Screening has been proposed for gastrointestinal cancers. Each cancer needs to be considered separately but any screening programme should have certain common properties that are essential for success:

• the disease should be fairly common in the population or group of people that is considered to need screening (otherwise the benefit to any one individual will not be sufficient to counterbalance the risk and inconvenience of screening to the rest).
• a diagnostic test to detect the condition is needed that is simple, cheap and reliable.
• effective treatment should exist.

The last of these properties might seem straightforward – surely surgery is effective treatment for all gastrointestinal cancers if they are caught early? Although this is a reasonable generalisation this statement is not true in all cases. Many cancers shed small numbers of cells or 'micrometastases' into the blood or lymph ducts from a very early stage so removal of the original (primary) cancer may not cure the patient. In these cases, the body's own immune response to these tumour cells seems at least as important as the speed with which the primary cancer is identified and removed.

Screening for bowel (colorectal) cancer

This cancer is the best 'bet' to have a successful screening programme for several reasons: it is relatively common, cure rates for surgery are very high (>90 per cent) if it is caught at an early stage and in most cases a pre-cancerous stage called a polyp can be found and simply removed to prevent cancer developing.

Population screening tests

The identification of a simple test is less straightforward. The simplest test is faecal occult blood screening (checking for blood in the faeces that is not visible) although this involves collecting faeces, which is unpleasant. But the major problem is the inaccuracy of the test. A positive faecal occult blood test indicates approximately a 10 per cent chance of cancer or a 34 per cent chance of a polyp, but the test will be negative in up to 50 per cent of cancers (this figure falls to about 30 per cent if the test is repeated on three consecutive days). A positive test has to be followed by colonoscopy (an examination of the bowel with a thin telescope). So, the high rate at which faecal occult blood is found when cancer is not present (false-positive rate) means that considerable stress is caused to individuals who subsequently turn out to have a healthy colon.

Reports of the best results show that if yearly faecal occult blood testing is recommended to all individuals over the age of 50, screening would reduce the death rate from colorectal cancer by about one third. This result sounds well worthwhile and so this screening programme is now recommended as standard practice in the USA. It costs about US$ 45,000 to gain one year of life - a figure likely to be beyond the reach of most public healthcare systems. Also, only about 3 per cent of the population (excluding high-risk individuals) will die from colorectal cancer and they will be diagnosed at an average age of 65 with roughly a 10-year life expectancy. So, the average length of life gained per person if all deaths from colorectal cancer could be prevented would be 3 per cent of 10 years which is four months. If one third of deaths were prevented, the length of life gained would amount to about five weeks. Furthermore, this benefit might be lower still in someone who follows a healthy lifestyle with a high vegetable intake and plenty of exercise. As a consequence, the UK National Health Service and the UK medical profession quite separately decided that the advantage of such a screening programme is not yet proven for individuals who have a normal risk of colorectal cancer.

Trials are currently in progress in the UK to further study the effectiveness of faecal occult blood screening and also to assess flexible sigmoidoscopy (examination of the final 40cm or so of the bowel, where most colorectal cancer occurs, with a bendy telescope passed into the rectum). In theory, flexible sigmoidoscopy is a better screening test because it has a much better chance than occult blood screening of detecting pre-cancerous polyps. But it has the disadvantages of being more invasive for the patient and of inevitably missing cancers higher up in the bowel that are beyond the reach of the sigmoidoscope. So in the USA, pressure exists to include colonoscopy (examination of the whole large bowel with a flexible telescope) routinely as part of a screening programme that also includes faecal occult blood testing. UK experts view this proposal as premature and suggest that routine colonoscopy needs to be found worthwhile in future trials before being introduced for screening.

Screening for high-risk individuals

Screening tests are more likely to find disease in individuals who are at higher risk of the disease than the general population. So, screening of high-risk individuals will improve the overall benefit from a screening test. Two groups are generally considered to need screening:

• individuals who have a first-degree relative (a parent, brother, sister or child) who developed colorectal cancer before the age of 45, whose risk of developing colorectal cancer in their lifetime is 1 in 10.
• individuals who have two (or more) first-degree relatives with colorectal cancer, who have at least a one in six life-time risk.

Since colorectal cancer that runs in families (hereditary or genetically determined) tends to occur at an average age of about 50, the potential length of life gained for each screened individual is much greater than for a normal-risk individual. The benefit is probably not quite 10 per cent of 25 years (see calculations above) because some cancers found will be 'sporadic' (occur by chance) rather than genetically determined and so will have an older age of onset. Nevertheless, the average benefit per screened individual, assuming a 50 per cent reduction in cancer death rate, would certainly be greater than 12 months. The screening test has to be colonoscopy because of the tendency of certain hereditary colon cancers to be in the highest part of the colon (caecum). Screening should be started approximately 10 years earlier than the age when the youngest affected relative developed cancer and should be done every five years.

Hereditary non-polyposis colon cancer (HNPCC)

This is a type of colorectal cancer that runs in families and tends to cause cancer at a relatively young age - under 45 years. The majority of the abnormal genes responsible for HNPCC have now been identified, so laboratories are starting to offer DNA testing. This can be performed on tissue saved from colon cancer surgery done on an affected relative in the past, so that if an abnormal gene is found, the rest of the family can be offered blood screening for this gene. Only family members found to have this type of mutation would then need colonoscopic screening.

One major issue in screening patients with possible HNPCC is that individuals with a mutant HNPCC gene have an increased risk not only for colorectal cancer but also for cancers of the stomach, uterus, ovaries and breasts. Thus there are reasonable grounds for suggesting that, once identified, these individuals should undergo regular screening of all these organs, which would involve regular colonoscopy, gastroscopy (telescope examination of the stomach), mammography (breast X-rays), pelvic ultrasonography (ultrasound examination of the pelvic organs) and uterine cytology (examination of the lining of the womb). This is a huge undertaking and whether the risks and inconvenience of such a complex screening regimen are balanced by the benefits is not yet known.

Familial polyposis coli (FAP)

Familial polyposis coli is much rarer than HNPCC but tends to affect fewer organs. The condition gives rise to cancers of the bowel or duodenum (the first part of the small intestine below the stomach) and a rare but potentially life-threatening tumour called a desmoid that arises from abdominal connective tissue, which supports and surrounds internal organs. In FAP, bowel cancer is virtually inevitable without therapy, so it is essential to screen direct relatives of any individual with FAP. At-risk individuals (who can often be identified by DNA testing) will need regular flexible sigmoidoscopy or colonoscopy from about age 10 and, if FAP is confirmed, the adolescent will need their colon removed surgically to prevent subsequent cancer formation (prophylactic colectomy).

Inflammatory bowel disease

People who have either ulcerative colitis or Crohn's disease causing colon inflammation (colitis) affecting more than half of the colon, have an approximately 10-fold increased risk of colorectal cancer. The standard recommendation is that such individuals should have colonoscopy performed yearly once the colitis has been present for eight years.

The main aim of the screening is to detect pre-cancerous lesions. With colitis, these occur as flat lesions whose cells have an altered appearance (dysplasia) under the micrscope. If high-grade (severe) dysplasia is found and confirmed by at least two independent experts (because diagnosis is difficult and rather subjective) then prophylactic colectomy is needed. This screening programme, although now fairly standard in Western countries, has been justifiably criticised because it has not been studied adequately.

In addition, review of similar programmes from large hospitals has shown that only about 50 per cent of cancers that develop during the programme are actually detected by the colonoscopies.

Finally, about 200 colonoscopies are performed for every cancer detected and the cost per year of life saved is about US$ 250,000 if colonoscopy is performed yearly.

Fortunately, evidence suggests that the cancer rate in colitis is falling, possibly because of a cancer-preventing effect of mesalazine, a drug taken to treat inflammation of the bowel, so screening may eventually become unnecessary.

Oesophageal cancer

Oesophageal cancer is not sufficiently common to justify use of the screening procedure unless the individual is at high risk.

Screening for high-risk individuals

The screening procedure for high-risk individuals is endoscopy (internal examination of the gullet with a telescope) and biopsy (removal of a small amount of tissue for examination under the microscope). Individuals with longstanding acid reflux, the common cause of 'heartburn', are at risk of developing oesophageal cancer if the acid that spills up from the stomach damages the lining of the lower part of the gullet. Normally, the lining is made from squamous cells, which are very similar to skin cells. With acid reflux, these squamous cells can be replaced by mucus-secreting cells similar to those found in the intestine. This cellular change is commonly known as 'Barrett's oesophagus' (after the British surgeon who first described it) and unfortunately carries an increased risk of cancer of the lower oesophagus. It has been suggested that individuals with Barrett's oesophagus should undergo annual endoscopy to check lining cells for pre-malignant change. In some hospitals, laser therapy is being used experimentally to reverse the process.

One major problem with planning a screening programme is that about one third of the adult Western population have at least minor degrees of Barrett's oesophagus. Most experts agree that if this change extends over at least 5cm of the lower oesophagus then the life-time risk of oesophageal cancer is probably about 10 per cent and endoscopic screening is justified, but the whole area is still a subject of considerable controversy and change. Individuals who smoke are particularly at risk of developing this sort of cancer and some argue that the risk in non-smokers is too low to justify screening.

Stomach cancer

Stomach cancer is too rare in most Western countries to justify screening. The situation is different in Japan where stomach cancer accounts for 11 per cent of all deaths in men. Screening with gastroscopy has been carried out in the workplace and deaths rates from the cancer have fallen markedly, but whether the two are related is not known. Another screening technique is barium radiology in which the patient swallows barium liquid that shows the inside of the stomach on X-ray. This test is easier for the patient but less accurate than gastroscopy and does not allow a biopsy to be taken.

Risk of stomach cancer is strongly linked with stomach infection with Helicobacter pylori, a bacterium found in about 30 per cent of Westerners. Moreover, H. pylori causes stomach cancer in experiments on animals. The bacterium is also thought to cause duodenal ulcers, so widespread testing for H. pylori followed by appropriate treatment might substantially reduce rates of both stomach cancer and ulcers. An alternative to this strategy that is being developed is a vaccine against H. pylori. However, one anticipated difficulty with these approaches is that severe H. pylori infection can damage acid-producing cells and reduce stomach acid production. Eradication of the bacteria might then increase acid production and lead to an increased rate of heartburn and therefore possibly even to an increased rate of oesophageal cancer.

It is therefore far from certain that the overall effects of H. pylori eradication would be beneficial. In Japan, a blood test (serum pepsinogen) that indicates the level of gastric acid production has been assessed as a screening test based on the fact that stomach cancer is linked with a relative lack of acid. Serum pepsinogen may turn out to be a useful initial screening test to indicate which individuals should then have a gastroscopy but would only be useful in a country such as Japan with a very high rate of stomach cancer.

Pancreatic cancer

Pancreatic cancer only affects about 10 people per 100,000 of the adult Western population at any one time. Current tests for pancreatic cancer (blood tests or scans) have a high false-positive rate of about 15 per cent. So, widespread use would result in 15,000 individuals incorrectly given a positive result for every 10 patients who have pancreatic cancer.

A case has been made that screening may be justifiable in the very rare cases of familial pancreatic cancer. This would require the highly invasive endoscopic pancreatography test (which involves telescopic examination and X-rays of the pancreas) plus sampling of pancreatic juice, a test that would carry about a 3 per cent risk of causing acute pancreas inflammation (pancreatitis), itself a serious and very unpleasant condition. This screening programme is therefore very experimental at present.

Liver cancer

Examination of people after death shows that about 50 per cent of patients with cirrhosis have primary liver cancer (hepatocellular carcinoma) at the time of death although it caused the death in only about 10 per cent. The level of a protein called alpha fetoprotein increases in the blood of about 70 per cent of affected individuals. Common practice is to check the protein levels with a blood test every six months in patients with cirrhosis who are considered fit enough to withstand the major surgery (usually removal of part of the liver) that would be required for a cure.