Cancer of the colon and rectum is the second most common cancer after lung cancer in the Western world, though it is less common than gastric cancer worldwide. It therefore contributes considerably to morbidity and mortality in Western societies. Until the last decade treatment had been limited to excisional surgery; the generally poor outcome showed little sign of improving.

New information from epidemiological studies, molecular biology, and imaging, together with surgical innovations and trials of adjuvant therapy offer possibilities for preventing some cancers, diagnosing others earlier, and improving both quality and duration of survival for the majority of patients while avoiding unnecessary mutilation for those with no prospect of cure. A thorough understanding of the disease and the options for management are therefore more necessary than ever.

The incidence of large bowel cancer varies between and within countries, which strongly suggests an environmental cause. The incidence is almost equal between the sexes, with some differences in risk ratio for cancers of different parts of the large bowel. Rectal cancers are twice as common among men but for the rest of the large bowel the male : female ratio is about 0.8 : 1 with right-sided cancers even more common in women.

The incidence of bowel cancer is high in the urban ‘western’ world but is rare in Asia, Africa, and parts of South America (Table 1) 335. There are also apparent geographical differences in incidence within countries. For example the incidence is higher in Scotland than in England, and higher in northern Italy and the northern United States than in the southern parts of the same countries. This mirrors changes in incidence of coronary heart disease. This north/south divide is not a fundamental association with latitude: the incidence of bowel cancer among the Finns and the Eskimos is low, suggesting the differences are more likely to be due to differences in social and dietary behaviour, perhaps induced by climatic changes. The incidence is also different among different cultural groups within countries, for example, 18/100 000 white South Africans have colonic cancer while the incidence amongst their black compatriots is about 6/100 000. The proportions in New Zealand whites and Maoris are similarly 3 : 1, though the incidence is about twice that of South Africa.

These differences in incidence among racial groups and within different areas of countries suggest that genetic or cultural factors are dominant in the genesis of the tumours. However, several studies of different religious and cultural groups who live in the same place but who have different lifestyles, have shown that environmental and cultural differences are mainly responsible for the variable incidence of large bowel cancer. Furthermore, migrants, such as the Japanese moving to the United States via Hawaii, show the cancer incidence of their adopted country within a generation. Colon cancer is also more common in urban areas, for example in Hong Kong, Singapore, and urban Brazil. Lifestyle therefore appears a more important factor in causing colon cancer for whole populations than geography or genetic make-up. However, individuals or families may also have a strong genetic predisposition to develop colon cancer.

The peak incidence for the disease is in the seventh decade, some 5 years later than the corresponding peak for colonic adenomatous polyps, which suggests that prolonged exposure to weak environmental carcinogens is necessary to induce tumours and that most, possibly all, pass through the benign phase before turning malignant. The epidemiology of proximal and distal or rectal cancer is not the same worldwide. The ratio of rectal to colonic cancer varies from 1 : 3 in white South African males to 1 : 1 in Finns and Maoris. There is also some evidence that right-sided cancers are becoming more common in the United Kingdom and the United States of America which suggests that there are qualitative differences in the aetiology of cancers at each end of the colon.

The epidemiological evidence so far suggests that environmental factors predominate in the genesis of large bowel cancer within populations, but it would be surprising if inherited genetic factors did not play a numerically small yet important role in the formation of some colon cancers (see also Section 42.1 134).

Colonic cancers, in common with all other tumours, show qualitative and often quantitative changes in the chromosomes when compared to normal cells. Many colon cancer cell lines show chromosome distortion, with trisomy of some or many chromosomes, sometimes to the point of aneuploidy. There is evidence to suggest that tumours with a poor prognosis have a more bizarre chromosomal configuration with a greater proportion of aneuploid cells.

Many cancers also express oncogenes of various types, in particular the ras and myc classes. Oncogenes were first identified in RNA viruses capable of inducing cancers, but it is likely that these ‘viral oncogenes’ found in colon cancers are an expression of part of the normal cellular genetic material responsible for growth and not a new viral infection. Growth genes are counterbalanced by genes responsible for restricting growth and perhaps inducing differentiation. These apparently viral genomes were probably incorporated into the genomes of animals very early in evolution, in much the same way that symbiotic intracellular organisms became mitochondria. It is self-evident that very local environmental changes in the embryo induce expression or inhibition of the growth and regulator genes to produce the entire organism. Carcinogens may similarly enhance the ‘oncogene’ expression or inhibit the regulators when inducing carcinomas or their adenoma precursors.

Some genes or gene deletions may also provide some protection against the development of cancer, in much the same way that possession of the recessive sickle-cell or thalassaemia genes protect against malaria to some extent. It seems that the cystic fibrosis gene deletion, Delta F 508 codon, which is carried by about 1 in 27 people, may protect against colon cancer and some other tumours.

A small number of colonic cancers have a definite inherited genetic cause. These take the form of familial adenomatous polyposis syndromes and the familial cancer syndromes (or hereditary non-polyposis colonic cancer). Familial adenomatous polyposis syndromes account for as little as 0.5 per cent of all colonic cancers but they are important as a biological model for other cancers and as a model for constructing algorithms of therapeutic management based on sound biological principles (Table 2) 336.

Familial adenomatous polyposis is an inherited propensity to develop numerous adenomas throughout the colon, some of which become malignant in time. The condition is inherited as an autosomal dominant with high penetrance but variable expression, and approximately half the children of the index case inherit the condition. The polyps appear at around puberty, and nearly all patients manifest polyps by their early 30s. There are a few reports of patients presenting later, but the chance of being affected if a sigmoidoscopy is normal are 1 per cent or less after the age of 35 years.

Recent work has shown that the genetic abnormality in familial adenomatous polyposis is a deletion of genetic material on the long arm of chromosome 5. The absence of this gene or cluster of genes seems responsible not only for the colonic polyps but also neoplasms in other parts of the gastrointestinal tract, the mesenchyme, and occasionally the brain. There appears to be a general derangement of cell protein formation common to many tissues. Gardner's syndrome with osteomas, desmoids, gastric hamartomas, and a propensity for gastric, pancreatic, and small bowel tumours, is the most severe form of the syndrome. Patients with the more common familial adenomatous polyposis seldom have clinically obvious extraintestinal manifestations. The least severe manifestation, Turcot's syndrome, is characterized by fewer colonic polyps, but affected patients also have intracranial tumours, which are more lethal. While it is likely that Turcot's cases are part of the familial adenomatous polyposis spectrum and therefore exhibit the same DNA abnormality, it has yet to be proved.

Peutz-Jegher syndrome and juvenile polyposis are not strictly part of the familial adenomatous polyposis syndromes, but for convenience they may be considered alongside. These conditions are rarer and the polyps fewer in number, but the trait is inherited in a dominant fashion. The polyps are hamartomas, which often have adenomatous changes around them. This adenomatous change predisposes to malignant transformation and increases the risk of colon cancer. It is likely that the genetic abnormality of these two conditions is quite different from that of familial adenomatous polyposis.

Hereditary non-polyposis colonic cancer is more common but less obvious than familial adenomatous polyposis. The inheritance is again autosomal dominant, but the appearance of the cancer is clinically similar to sporadic cases of colorectal cancer, although it occurs at a younger age. In these cases, which account for some 5 per cent of all colorectal cancers, the genetic abnormality is usually on chromosome 17 or 18. Two subtypes of the syndrome have been recognized: site-specific colonic cancer, where individuals of a family are susceptible to colonic cancer but not cancers of other organs (Lynch type 1), and cancer family syndromes where female members of the family are prone to breast and uterine cancers as well as colonic tumours (Lynch type 2). The tumours may all develop in the same individual but, more commonly, the three types of cancer appear in different members of the family. Occasionally tumours of other organs are found (the Muir-Torre syndrome).

It is interesting that the cancers in patients with Lynch type 1 and 2 syndromes develop in early middle life rather than in the seventh and eighth decades, as do sporadic cases. The cancers occur predominately on the right side of the colon and are of low malignancy, again in contrast to the sporadic cases. While the two Lynch syndromes appear different there may be but a single genetic abnormality exhibiting pleiotropy.

The most important environmental factors in the aetiology of colon cancer are likely to exert their influence within the lumen of the colon. The strongest evidence is for a role of diet in the induction and growth of tumours. The significant association between diet and the prevalence of colonic cancer and heart disease suggests that a diet rich in fat and meat is a common factor in both diseases.

Epidemiological studies, case–control studies, cohort studies, and animal experiments provide strong evidence that a diet rich in fat, and particularly in animal fat, is associated with a high incidence of colon tumours and, in animals, with the promotion of malignancy. While fat is the major environmental factor so far identified, fat itself is not a carcinogen. Animal experiments suggest that all fats are mutagenic in very high concentrations, but unsaturated fats are more likely to induce progression from adenomas to cancer. Luminal fat, and in particular its oxidation products, promote colonic epithelial cell proliferation and an increase in crypt cell production rate, yet ketone bodies are also an important source of nutrition for the mucosa. Diets high in animal fats are rich in cholesterol, which is a major substrate for bacterial degradation, and there is a correlation between the incidence of colon cancer and the amount of cholesterol in the faeces. It is unlikely that cholesterol per se is a carcinogen, but its metabolites probably are. Some metabolites may also have an endocrine action which is an aetiological factor for some colonic cancers. For example: some HNPCC cases are associated with oestrogen dependent breast and uterine tumours: familial adenomatous polyposis patients develop their adenomas at or after puberty when sex hormone levels are high.

Probably the most important action of dietary fat is to stimulate choleresis, thereby increasing the levels of faecal bile acids which are, in turn, metabolized by bacteria to the secondary bile acids lithocholic and deoxycholic acid, which are powerful tumour promotors. Other situations which increase faecal bile acid concentrations, such as cholecystectomy, gastric surgery, and terminal ileal resection, are also associated with an increased risk of colon cancer.

Colon cancer is more common in human populations with a high total calorie intake or a high protein intake, though the two are usually associated. There appears to be a general promoting effect of calories in carcinogenesis but it is difficult, in man, to separate the effect of calories from that of fat and the lack of fibre or resistant starch. Some protein breakdown products and amino acids enter the colon to be degraded by bacteria into compounds such as the N-nitrosamines, which are powerful carcinogens. The relative load of protein products entering the colon is, however, small compared with that of cholesterol. Nevertheless it is increased in patients with intestinal hurry and may contribute to the carcinogenic effects of cholecystectomy, gastric surgery, and ileal resection.

Western food is notoriously low in dietary fibre and resistant starch, with a consequent increase in intestinal transit time. Epidemiological studies show a negative correlation between the amount of fibre in the diet and the prevalence of colonic and rectal cancer which, though it may be an epiphenomenon, suggests a causal relationship.

Dietary fibre consists of many different substances with different properties, from soluble compounds such as pectins and hemicellulose, to insoluble celluloses and lignins. Resistant starches may also be considered part of dietary fibre because they are indigestible. Indigestible fibre increases the bulk of stool, principally by retaining water, and decreases the colonic transit time in most people (but increases it in patients with colonic irritability and pathologically rapid transit). Fibre alters the colonic bacterial flora both qualitatively and quantitatively, and is partially degraded by these bacteria to produce numerous luminal products, such as flatus. Fibre also dilutes and absorbs luminal toxins. The combination of these factors may have conflicting effects upon the colonic epithelium, but the net result is to reduce the carcinogenic capacity of the luminal contents. Animal experiments provide conflicting data: some components of dietary fibre promote colonic carcinogenesis while others are protective. The combined effects of dietary fibre deficiency on the stool content, consistency, and transit may account for the well-known association of colonic carcinoma with diverticular disease, constipation, and haemorrhoids.

A number of dietary constituents have been found to inhibit carcinogenesis, including selenium, vitamins C and E, retinoids, &bgr;-carotene, and plant sterols. Diets low in vitamin C and &bgr;-carotene also tend to be low in dietary fibre. Selenium is an essential trace element which, experimentally, has been shown to inhibit carcinogenesis in rats. Endemic selenium deficiency is rare but does occur in New Zealand, which might account for the particularly high incidence of colonic and rectal cancer among those of European stock and the relatively high incidence among Maoris compared to other natives of the Pacific basin.

Bacteria are a major constituent of stool and are important contributors to the colonic luminal environment. The two bacterial enzymes so far shown to be important are &bgr;-dehydroxylase, which breaks down harmless primary bile acids into the mutagenic secondary bile acids lithocholic acid and chenodeoxycholic acid, and 4,5-nuclear dehydrogenase, which desaturates bile acids to produce substances that are both tumour initiators and growth promoters.

The close correlation between the incidence of colonic carcinoma and faecal secondary bile acid concentration is a clear indication of the importance of bacterial activity upon the colonic epithelial environment. Furthermore there is a correlation between the carriage of nuclear dehydrogenase-producing Clostridia in the colon and the incidence of colonic cancer. These bacteria also require a neutral pH and hydrogen acceptors for optimal activity and high-risk populations have been shown to have higher pH and hydrogen acceptor levels than controls. It is likely that further studies will yield more important information in the search for luminal carcinogens, growth promotors and growth inhibitors.

X-rays are important mutagens. It is therefore hardly surprising that intracavity irradiation in the treatment of carcinoma of the cervix uteri is associated with a small increased incidence of rectal cancer within the field of irradiation. The cancers appear some 5 to 15 years later. There is, however, no known increased risk of colonic cancer in people living in areas with higher background levels of radiation, nor after abdominal radiation for conditions such as testicular tumours or Hodgkin's disease.

Ulcerative colitis has, for a long time, been known to increase the risk of colonic cancer. The risk increases almost exponentially with time some 10 years after the onset of the disease, particularly in patients who have total colitis, those with a severe first attack, those who develop the disease in childhood, and those patients whose disease follows a relapsing course. Patients with mild distal colitis have no greater risk of developing rectal cancer than the normal population, while patients with severe long-standing disease have a 1 in 2 chance of developing colonic cancer after 30 years. Colonic epithelial dysplasia seems to be the precursor lesion, though progression of a particular area of dysplasia is not inevitable. Areas of dysplasia may be scattered throughout the colon, which explains the greater likelihood of multiple cancers among those with colitis. The tumours are seldom polypoid or exophitic but usually flat and infiltrating, which makes diagnosis more difficult. The prognosis of these colitic cancers was originally thought to be particularly poor but more recent data suggests that stage for stage there is no difference in prognosis compared to sporadic cancers.

Crohn's colitis is also associated with an increased risk of cancer in the diseased segment as well as in other areas of the digestive tract. The risk is, however, less than that associated with severe ulcerative colitis. It is uncertain whether the chronic inflammation is the main predisposing factor for carcinogenesis in these two conditions or whether genetic make-up or bile acids and their metabolites are more important.

Surgical injury in the form of an anastomosis increases the risk of a cancer developing at the site of injury both in experimental models and as a site of implantation in man after resection of a colonic cancer. There has been some debate about the effect of different suture materials upon the risk of implantation but the data are conflicting and the evidence inconclusive.

Cholecystectomy is also associated with an increased risk of colonic cancer, in particular right-sided lesions among women. The risk is similar after gastric surgery in man but not in experimental animals. Both operations may exert their effect by increasing the delivery of bile acids to the colon, thereby increasing the concentration of secondary bile acids within the colon.

Ureterosigmoidostomy, performed for urinary diversion, is particularly associated with the development of colonic cancer at, or near, the ureterocolic anastomosis. Why this should be so is not certain but it may be caused by the effect of phenols, cresols, and other compounds from cigarette smoke, excreted in the urine, upon the colonic epithelium (Table 3) 337.

The association between colonic polyps and colonic cancer has been known for a long time. The incidence of malignant change increases with both the size of the polyps and the degree of dysplasia. Tubular adenomas are less prone to malignancy than are villous adenomas, which may also be very large. It is now believed that most, if not all, colonic cancers originate within an adenoma, and while most polyps do not become malignant it is clear that a polyp which has been growing for some years to achieve considerable size is more likely to do so. Equally clearly some pass through the adenoma phase quickly, underlying the dual process of tumourigenesis and malignant transformation.

The prevalence of adenomas and the prevalence of colonic cancer run in parallel in epidemiological studies. Polyps are also frequently found in close proximity to cancers. The distribution of adenomas around the colon is the same as that of colonic cancers and some adenomas snared endoscopically disclose microscopic foci of cancer. The transformation of a benign adenoma into a cancer is thought to take, on average, about 5 years; the mean age of patients at diagnosis with polyps is 5 years less than that of patients presenting with a colonic cancer.

Long-term immunosuppression, either for the prevention of transplant rejection or as a consequence of HIV infection, predisposes to cancer. Lymphomas, which may affect the colon as well as the small bowel, are the most common abdominal neoplasm in immunosuppressed transplant patients. Squamous cell carcinoma of the anus is not uncommon among transplant patients and is usually associated with human papillomavirus infection. Small cell carcinomas of the colon, which are rare in the otherwise healthy population, have been reported in AIDS patients. However, there appears to be no association between adenocarcinomas of the colon and immunosuppression.

Neoplasms themselves induce some degree of immune suppression, either directly through products of the malignant cells or indirectly through malnutrition and cachexia. Some, but not all, colon cancers induce a vigorous local immune response with an intense cellular infiltrate within the tumour, or a histiocytic reaction in the draining lymph nodes. Suppression of this immune response is probably mediated by secretion of soluble suppressor factors such as the retroviral protein p15E, which is also responsible for the immunosuppression induced by retroviruses and is present in a number of tumours including colon cancer.

Colonic cancers, in common with most epithelial tumours, are polyclonal with clones of cells exhibiting differing degrees of ‘malignancy’. The more undifferentiated or more ‘malignant’ clones are more likely to spread and metastasize. A tumour's biological behaviour is the main determinant of the tumour's propensity to spread locally and to metastasize and, therefore, indicates the ultimate prognosis. This in turn is reflected to some extent by the histopathological features of a cancer. There is little doubt that, in the near future, other phenotypic characteristics, identified by molecular biological methods, will provide even more accurate prognostic information. For example, tumour cells that stimulate angiogenesis or have fewer surface adhesion molecules are more likely to metastasize and therefore to have a worse prognosis. Tumour secondaries, which are often derived from selected clones of a polyclonal primary, are unlikely to behave in the same way as the primary tumour and will generally be more malignant.

Tumour biology is very much reflected by the stage of the disease, and therefore the amount of spread, at presentation. It is no surprise that there is a significant inverse relation between the length of history and the stage of the disease at diagnosis. Quicker growing tumours usually present with a short history and are advanced at the time of treatment. Although the clinical and pathological stage is a ‘snapshot’ in the life of a tumour, it provides the most accurate prognostic index; this may be refined further by the histopathological features. Several staging methods are in use throughout the world, and each has strengths and weaknesses. The most commonly used are the Dukes' classification and derivations of it, or the Union Internationale Contre Cancer (UICC) TNM classification. The former has the advantage of great simplicity but considerable disadvantages from lack of precision: it does not reflect accurately the depth of tumour penetration, the extent of spread outside the bowel, the number of lymph nodes affected by tumour, or the presence or absence of metastasis, all of which have an important bearing upon prognosis. Derivations such as the Astler-Coller and Australian classifications refine the Dukes' staging but do not provide the flexibility of the TNM method which enables useful division into subsets without being unduly complex. It is therefore most appropriate that surgeons adopt the UICC TNM classification as a suitable international standard ( Table 4 338 and Fig. 1 1066).

The anatomical site of the cancer has an influence on the stage of the disease and an independent effect upon prognosis. Right-sided colonic cancers tend to be at a more advanced pathological stage at the time of presentation, but the prognosis is generally no worse than is that of left-sided tumours. Stage for stage, right-sided lesions have a better prognosis; this may reflect subtle differences in aetiology. Rectal tumours have a generally better prognosis, because of earlier presentation and easier accessibility.

Staging gives information about prognosis in general, but particularly indicates the probability of occult hepatic metastasis, which is the major factor affecting survival. Patients with Dukes' C tumours are more likely to have occult hepatic metastases than those with Dukes' B tumours, while patients with Dukes' A lesions are most unlikely to have hepatic metastases. Occult hepatic metastases account for the majority of deaths from colonic cancer: while only about 20 per cent of patients die from local spread of the disease, which is also reflected in the clinical stage.

Age generally has little effect on the behaviour or prognosis of colonic tumours, except for patients under the age of 40 years, who appear to have a particularly poor prognosis.

Grading depends upon subjective interpretation of the degree of tumour differentiation at histological examination. Various grading systems have been proposed, but grading into two broad groups, low or average grade tumours which are well to moderately differentiated and high grade or undifferentiated, reduces the variation between observers while at the same time providing useful prognostic information. Patients with high grade cancers fare worse than do those with well differentiated lesions after taking account of the tumour stage ( Table 4 338 and Fig. 2 1067).

Typing, on the other hand, reflects the cellular characteristics. Mucinous, signet cell, and small cell tumours are variants of the more common adenocarcinoma and the frankly undifferentiated cancers. Again typing may give some useful additional prognostic information. Signet cell and small cell tumours have a worse prognosis than adenocarcinoma, while mucinous lesions tend to recur locally. Occasionally rectal cancers turn out to be squamous cell types which are more responsive to chemotherapy and irradiation. Melanomas, which have a particularly poor prognosis, are found rarely in the rectum. Both of these tumours are, however, more common at the anus.

Histological features such as vascular, lymphatic, or perineural invasion are prognostically unfavourable. By contrast, lymphocytic infiltration of the tumour and a histiocytic reaction in the regional lymph nodes are minor favourable prognostic features.

Identification of surface tumour antigens, such as carcinoembryonic antigen, oncogene expression and DNA ploidy, add potential refinement, but these are not yet in routine use. Full characterization of the molecular biological features of cancer cells will in the future probably provide more important prognostic and therapeutic information than histological typing and grading. It may appear academic at present to stage patients and their tumours, but it will become clinically important as the place of adjuvant chemotherapy and irradiation becomes clearer, particularly when chemotherapy improves. In the mean time it is important to collect pathological data accurately to allow clinical studies and the audit of surgical performance. The latter cannot be judged in the absence of good histopathological information.

Stage remains the most important indicator of prognosis . The prognosis of patients with adequately treated Stage 1 cancers is little different from that of an otherwise healthy population of the same age; 95 to 100 per cent live 5 years or more after resection. Patients with cancer spread through the serosa only have a 40 to 60 per cent chance of living 5 years, although the prognosis is more favourable if the tumour is only just through the serosa and is correspondingly worse if adjacent structures are invaded. Lymph node metastasis further adversely affects prognosis, with only about 30 per cent of patients surviving 5 years. Subclassification is useful: 60 per cent of N1 patients may live 5 years while less than 30 per cent of N3 patients will survive. A few patients, some 5 to 10 per cent, live 5 years even with hepatic metastases, although 85 per cent of such patients die within 1 year of diagnosis.

The survival curve of patients with colonic and rectal cancer treated by resection is curvilinear, reaching a nearly flat plateau. Few patients develop metastatic disease 5 to 10 years after surgical treatment of the primary lesion. Many patients are therefore cured of their cancer. This suggests that large bowel cancer is somewhat unusual, because metastasis occurs relatively late in the lifespan of the tumour; many other cancers, such as cancer of the breast, lung, or melanoma, metastasize early. Natural survival with metastases then depends very much on the symbiotic relationship between the tumour and host.

Patients with colonic and rectal cancer have a broad range of clinical presentation which can be subclassified according to the anatomical site of the primary. Distribution of cancers and adenomas is uneven around the colon. Caecal and right-sided tumours account for about 20 per cent of large bowel cancers, while 70 per cent occur distal to the splenic flexure and about 45 per cent are at or below the rectosigmoid junction.

Right-sided tumours are often remarkably silent and many patients present with only the symptoms and signs of iron deficiency anaemia from protracted occult blood loss, a fact which is all too often forgotten or ignored. Rarely the blood loss is copious, particularly in patients who are receiving anticoagulants.

The faeces entering the caecum are liquid, and obstruction is a relatively late presentation. As the lumen becomes narrowed the patient complains of intermittent central or right iliac fossa colic, which is often postprandial, stimulated by the gastrocolic reflex. The pain is often followed by the onset of intermittent diarrhoea, possibly the consequence of faecal fermentation and the accumulation of bacterial toxins within the bowel lumen. Typical distal ileal obstruction occurs if the tumour obstructs the ileocaecal valve, or if the ileocaecal valve becomes incompetent in the face of complete caecal obstruction. Waves of central abdominal colic occur with progressive central abdominal distension and borborygmi. Visible peristalsis, faeculent vomit, and dehydration are late manifestations. Not infrequently a palpable mass is the presenting symptom, though more commonly the mass is felt during clinical examination.

Elderly patients occasionally present with acute appendicitis when the carcinoma occludes the appendicular orifice. It is therefore always wise to consider the diagnosis of caecal carcinoma in any older person who presents with acute appendicitis. The diagnosis may not be obvious at the time the appendix is removed and must be sought subsequently by barium enema examination. Quite commonly the tumour penetrates the bowel wall posteriorly, producing a sealed perforation and an abscess in the psoas muscle. Such patients present with the symptoms and signs of infection accompanied by a painful mass in the right iliac fossa. The pain may radiate down into the leg or hip, particularly if the femoral nerve is involved in the abscess, or if the abscess tracks down below the inguinal ligament to appear in the femoral triangle. Similarly the pain may radiate posteriorly if the abscess erodes the lumbar muscles to point in the loin. Occasionally an anterior tumour may produce a free perforation of the caecum or right colon to produce peritonitis with severe generalized abdominal pain, a silent abdomen, and percussion rebound tenderness.

Occasionally, right-sided colon cancers present with general symptoms of malaise and lack of well being, sometimes with a pyrexia of unknown origin. These symptoms are either the result of a small occult abscess or are due to tumour burden, usually from metastases. The symptoms and signs of metastases are legion, but are usually accompanied by pain and tenderness over the liver, which is the most common site of metastasis. These symptoms are usually produced by rapid growth of the secondaries distending the liver capsule. Metastases tend to grow much more rapidly than the primary because they are ‘aggressive’ selected clones from a polyclonal tumour. The metastases may also outgrow their blood supply, partially infarct and undergo necrosis. Haemorrhage into the necrotic metastasis may then occur as a secondary event. Either situation will produce pain, malaise, and perhaps a pyrexia which will resolve spontaneously. Many patients with hepatic metastases are quite asymptomatic in the early stages, when hepatomegaly may be the only physical sign.

Caecal tumours may spread locally and transperitoneally before causing bowel symptoms. As the tumour burden increases, cytokines released from leucocytes may cause the loss of well being, anorexia, and weight loss: these are common features of carcinomatosis. Cytokine release is thought to be the mediator that alters the patient's metabolism and ultimately causes cachexia. Local spread of tumour throughout the peritoneum and into the omentum may produce a protein-rich ascites. Infiltration of the small bowel mesentery sometimes produces pseudo-obstruction, or chylous ascites by obstruction of lacteals. Occasionally distension due to ascites is the presenting feature.

Rarely, patients present with signs or symptoms of widespread tumour dissemination such as leucoerythroblastic anaemia from marrow infiltration, a persistent cough from pulmonary secondaries, generalized lymphadenopathy or a hard tumour nodule at the umbilicus, the so-called Sister Marie Joseph sign.

The stool dehydrates and becomes harder as it reaches and passes through the left colon to be stored in the sigmoid before defaecation. Patients with a left-sided colonic cancer commonly present with a change in bowel habit, often constipation alternating with diarrhoea, usually accompanied by lower abdominal colic, possibly distension, and a desire to defecate. The symptoms tend to become progressively severe, and this may serve to distinguish cancer from diverticular disease or colonic irritability. The irritable colon syndrome is usually seen in younger adults; if a middle-aged or older patient presents with a change in bowel habit the symptoms should be assumed to be caused by a colon cancer until proved otherwise. Progressive constipation or diarrhoea are less common changes of bowel habit, either of which may end in complete large bowel obstruction.

Change in bowel function is often accompanied by the passage of altered blood, and sometimes mucus, in the stool or on its surface, particularly in the case of distal sigmoid lesions. Occasionally patients present with colonic bleeding as an isolated symptom. The loss is usually intermittent, with small amounts of dark blood but may be brisk, a symptom more usually associated with diverticular disease. Brisk bleeding from a colonic cancer is more likely to occur in a patient treated with anticoagulants.

A few patients present with a pain or mass in the left iliac fossa, but a mass is often palpable in the abdomen on physical examination. A palpable carcinoma of the splenic flexure must be distinguished from an enlarged spleen or kidney.

Some patients, surprisingly, have remarkably few symptoms until they present with massive abdominal distension due to complete large bowel obstruction. In these circumstances the caecum becomes very distended. Unless distension is recognized and treated rapidly, or unless the ileocaecal valve becomes incompetent, caecal perforation leads to faecal peritonitis. If the ileocaecal valve is incompetent, the obstructed large bowel decompresses into the ileum to produce a mixed clinical picture of large and small bowel obstruction. Occasionally the tumour itself will perforate, causing sudden acute abdominal pain and peritonitis. More commonly the tumour becomes attached to adjacent organs and may invade them. A sigmoid cancer may invade the lateral abdominal wall and form an abscess, or invade a loop of small bowel and either produce an ileocolic fistula with severe diarrhoea or small bowel obstruction. Neoplasms of the splenic flexure or descending colon invading the jejunum sometimes present with severe intestinal haemorrhage. Sigmoid cancers commonly invade either the uterus, ovaries, or the bladder. Colonic cancer is the second most common cause of colovesical fistulae after diverticular disease, and patients usually present with haematuria and recurrent urinary tract infections initially, and later with pneumaturia or faecaluria. Similarly a sigmoid cancer fixed in the pelvis may fistulate into the vagina to produce an offensive irritating discharge that changes to frank faecal incontinence per vaginam.

Left-sided lesions may sometimes present from the start with anaemia or the symptoms and signs of dissemination but this event is less common than with right-sided lesions.

Rectal cancers are a particular subset of left-sided lesions, important for two reasons. Firstly, the accessibility of the lesions should enable the diagnosis to be made at an early and favourable stage, and secondly the diagnosis is often needlessly delayed while the symptoms are attributed to haemorrhoids or an anal fissure.

Most patients with rectal cancer present with bleeding. While the blood is often dark and mixed with the stool or coating the surface, it may be bright and quite separate from the faeces. For this reason the symptoms are often attributed to haemorrhoids. Minor changes in bowel habit, such as increased frequency of defaecation, mucus with the stool, or mucus diarrhoea are also quite common. Mucus diarrhoea is particularly associated with large villous adenomas which have often become malignant. The mucus is rich in potassium and may be sufficiently profuse to produce dehydration and coma. Tenesmus, the continuous urge to defecate, is a grave symptom produced by an advanced rectal tumour inducing a permanent sense of fullness. Tenesmus may give way to continual sacral pain, sometimes radiating down the inner thighs, as the tumour invades the sacrum and the sacral plexus of nerves. Anal pain, initially on defecation and later continuous, may occur as a low rectal cancer invades the anal canal. Incontinence supervenes when the anal sphincters are destroyed. Proctalgia fugax (fleeting perineal pain) is a rare presenting symptom and therefore when it occurs de novo in later life, rectal cancer must be considered as a possible diagnosis.

Few patients present with disseminated disease though this may occur in younger people with rapidly progressive tumours or in those whose rectal symptoms are ignored for a long time. Similarly large bowel obstruction is a rare and late mode of presentation of rectal cancer.

Rectal cancers generally cause symptoms early in their course. The tumours are accessible and this translates into a better prognosis than that seen with cancers of the rest of the colon. Biological features may also contribute to the better prognosis but these have yet to be identified convincingly.

The importance of a good history and a careful physical examination cannot be overstressed. The completion of the physical examination should always be a digital rectal examination to feel for a mass, assess the mobility and position of the mass, and to feel for enlarged extra rectal lymph nodes. The examination is completed by an immediate study of the stool for occult blood using the Hemoccult® test or something similar. A proctosigmoidoscopic examination of the rectum and anus should follow.

Tests for occult blood in the faeces are still being improved, but in general the greater the specificity of the test the less its sensitivity. The most widely used test at present is the Hemoccult® , a guaiac test which depends on the peroxidase-like reaction of haematin, a degradation product of haemoglobin, to produce a blue colour on the test paper. The test is not particularly sensitive and so avoids detecting blood shed from the mouth or upper gastrointestinal tract, but it is quite specific for left-sided cancers. The overall accuracy is about 60 per cent in the presence of a colon cancer. Laboratory guaiac tests are usually more sensitive but less specific with more false-positive, but less false-negative results. Antibody tests for human haemoglobin in the stool are being developed but have yet to replace the simple and robust Haemoccult test as the most useful out-patient investigation.

There is a debate whether the proctosigmoidoscopic examination should be performed with a rigid or flexible fibreoptic instrument, in prepared or unprepared bowel, in the left lateral or prone jack-knife position. While it is largely a matter of tradition and individual preference, it is also to some extent governed by the information sought. A flexible sigmoidoscope can reach much higher up the colon, in comfort, than the rigid 30 cm instrument, but the examination requires a phosphate enema preparation. The diagnostic yield is also much better, mainly because more bowel is examined. The rigid instrument is nevertheless useful for inspecting the unprepared rectum. The mucosa, faeces, blood, and mucus can be examined in their natural state quickly and easily with the patient in the left lateral position. It is particularly useful to see whether blood is mixed with faeces and also if it is coming down from the sigmoid colon. An enema may wash away much of the ‘evidence’, and also makes the mucosa hyperaemic, stimulating it to secrete mucus as though inflamed.

Both the rigid and flexible instruments can be used easily in the left lateral position, most popular in Great Britain, or the prone jack-knife position using a special tilting table, which is the custom in the United States. Rigid sigmoidoscopy requires some care to avoid rectal perforation, but needs less experience than the fibreoptic method to obtain useful diagnostic information. Both instruments must only be advanced when the lumen is seen clearly. The investigation should be painless or at worst produce only mild discomfort. Blind intubation is a recipe for disaster ranging from pain to perforation.

It is possible to employ either method of sigmoidoscopy selectively to use the patient's and clinician's time most efficiently. When a patient presents with symptoms and signs of rectal pathology an unprepared rigid proctosigmoidoscopy is preferable. If, however, the history suggests that the problem lies more proximally, flexible sigmoidoscopy is more practical.

There is much futile debate about the relative merits of barium enema and colonoscopy as the better method of investigating the colon, for each investigation provides different but complementary information. The barium enema gives good anatomical and topographic information which not only may be quite sufficient to diagnose a polyp or carcinoma but demonstrates the site and configuration of the lesion. The anatomical position of a cancer is clearly of great importance to a surgeon planning an operation. The discrimination for small lesions and mucosal abnormality is considerably enhanced by the double contrast technique compared to the single contrast enema. The air insufflated after the barium shows clearly the mucosal destruction from an ulcerated carcinoma or the mucosal coating of both adenomatous polyps and polypoid carcinoma.

Colonoscopy requires the same meticulous bowel preparation as a double contrast barium enema, but the patients also need some sedation. Colonoscopy is therefore more invasive, and the patients need both time to recover and somebody to take them home. There is also a small risk, about 1 : 1500, of colonic perforation during endoscopy. Nevertheless colonoscopy enables a more detailed study of the mucosa, visualizing lesions of less than 0.5 cm. The principal advantage over radiology is that lesions can be biopsied, or removed by snare cautery if they prove to be adenomatous polyps or a small polypoid carcinoma (Fig. 18) 1083. The main disadvantages are that anatomical localization is too poor to plan an operation as to where a skin crease incision may be used; there is seldom a complete permanent record of the investigation to refer to, and it is quite possible to miss lesions behind haustra.

Both investigations may be necessary to reach a diagnosis and plan therapy, but it is reasonable to start with a barium enema when patients present with a change in bowel habit and abdominal pain, with or without abdominal distension. If the main symptom is bleeding or anaemia, then colonoscopy is the investigation of choice because it enables identification of other bleeding lesions such as angiodysplasia, which are impossible to identify radiologically. Treatment such as polypectomy and laser therapy is also possible. Biopsy specimens of a cancer or inflammatory bowel disease provides confirmation of the diagnosis and enables confident planning of treatment.

Small endoluminal ultrasound instruments have recently added refinement to the diagnosis and preoperative staging of tumours, particularly of the rectum. Ultrasound probes mounted on a colonoscope are expensive and seldom used, but rigid rectal ultrasonography is now an established means of assessing the depth of penetration of the bowel wall by the tumour (Fig. 19) 1084. Enlarged lymph nodes can also be identified, but confirmation of node metastasis is less reliable (Fig. 20) 1085. This improved diagnostic information is essential when considering local treatment for a rectal cancer, and sometimes when choosing between an abdominoperineal excision of the rectum or a low anterior resection.

Once the diagnosis of carcinoma of the colon or rectum is made secondary investigations are necessary to plan and organize treatment. Preoperative detection of metastasis is clearly crucial to allow treatment to be modified. It is quite possible to perform a whole battery of investigations which is both a burden for the patient and unnecessarily expensive. Tests should be efficient and yield maximal information at minimal cost. The liver is the most common site for metastasis followed by lung, retroperitoneum, ovary, peritoneal cavity, and, rarely, adrenal glands; abdominal ultrasonography and a plain chest radiograph are both useful and economic. However ultrasonography is very operator dependent and the investigation tends to vary in quality. The hard copy of the study is not so helpful as the real-time examination, and so review is not particularly useful. Transabdominal ultrasound can seldom detect metastases less than 1 cm in diameter. Abdominal computed tomography (CT) provides better discrimination for smaller hepatic lesions and can distinguish angiomata from metastases as small as 0.5 cm diameter, after contrast enhancement. CT also provides more information about lymphadenopathy and invasion of surrounding structures but at greater cost (Fig. 21) 1086. Either investigation is excellent for detecting small volumes of ascites, ovarian enlargement, and hydronephrosis from retroperitoneal spread to the ureters. CT or ultrasound-guided biopsy or fine-needle aspiration cytology may be particularly useful in confirming or excluding the presence of metastasis. Abdominal ultrasound is the investigation of first choice to screen for metastases and CT is then reserved for cases where more information is required, so justifying the additional costs. Magnetic resonance imaging (MRI) can also identify tumours and metastases but adds little to the investigation of most cases of colonic cancer. However it may provide useful additional information in patients with large sarcomas of the colon or rectum (Fig. 22) 1087.

Intraoperative hepatic ultrasound is a recent and particularly useful method of detecting hepatic secondary deposits, locating their anatomical site in the segments of the liver, and determining the number of metastases. Histological confirmation of metastasis can then be made by direct needle biopsy and frozen section histology. The investigation is easily performed using a flat ultrasound probe in the palm of the hand and systematically moving it over the liver surface. The anatomical detail of the liver is excellent, the discrimination is much better than transabdominal ultrasonography and probably better than CT, and allows sensible modification of the surgical procedure and assessment of resectability of hepatic secondaries.

Measurement of circulating antigens released by the tumour, such as carcinoembryonic antigen or, less commonly, &agr;-fetoprotein have frequently been advocated but they have proved to be of little diagnostic value and have not passed into general use. They are, however, of limited use in studies of tumour biology and in prospective studies, particularly of follow-up practice, and in the use of second look laparotomy to treat occult tumour recurrence.

Any patient being considered for an operation requires some preoperative investigations to confirm fitness for the procedure and to pre-empt postoperative problems. These investigations should include an electrocardiogram, simple lung function studies, particularly in people with a history of pulmonary problems, and a haemoglobin estimation. Renal function and state of hydration is satisfactorily monitored by estimation of blood urea, creatinine, and electrolyte concentrations. If an ultrasound examination suggests ureteric obstruction an excretion urogram should be performed to demonstrate its site. It is also most important to make a formal estimation of the nutritional state preoperatively, for postoperative morbidity is greatly increased in malnourished patients. The best estimation of nutritional state is by history of diet intake and weight loss, and physical examination, looking for the features of malnutrition. The level of serum albumin is the best single blood marker of malnutrition despite the many other causes of hypoalbuminaemia.

Treatment has two objectives: firstly to treat the patient's symptoms without causing more problems than the cancer, and secondly to cure the patient of his cancer without putting the patient at undue risk. These two aspects of therapy are not necessarily the same and may, in fact, be in conflict. The patient's best interests will be served by his wishes and expectations, the demands of therapy and the presence or absence of serious comorbidity. For example a frail, elderly, infirm patient with rectal cancer causing few symptoms, but who has severe cardiac disease, may require little or no treatment for the rectal lesion when otherwise major resectional surgery with or without adjuvant therapy would be appropriate for a younger, fitter patient.

Surgical cure of colonic cancer requires excision of all the cancer. The principle that the tumour should be excised with an adequate margin of surrounding tissue, and as much of the vascular supply as is practical to ensure generous clearance of the local and regional lymph nodes, has stood the test of time and is agreed upon by all surgeons. There is, however, a considerable debate about what constitutes adequate margins and sufficient lymph node clearance. It is rare for a tumour to spread up or down the bowel as much as 2 cm from the primary unless it is undifferentiated or a signet cell tumour. Little bowel therefore needs to be removed on either side of the cancer to provide an adequate clearance. For rectal tumours, a 5 cm margin of clearance is satisfactory, though this may be reduced safely to as little as 2 cm for small, well differentiated lesions. Usually more colon is resected but this is determined by the vascular and lymphatic anatomy. Spread is likely to be more extensive into contiguous tissues such as the mesorectum, the lateral rectal ligaments, the bladder, the abdominal wall, the omentum, or the small bowel. Lateral margins therefore need to be generous to obtain adequate clearance. Where a tumour is invading adjacent structures and is resectable, all involved structures should be resected en bloc to avoid tumour dissemination. The prognosis for patients where the tumour has been resected piecemeal is significantly poorer.

A no-touch technique was advocated by Turnbull of the Cleveland Clinic Foundation, Ohio, where the vascular pedicle was ligated and divided before the tumour was handled or mobilized on the theoretical assumption that mobilization would shed malignant cells into the portal vein and increase metastasis formation. Although the technique was widely adopted, only one objective study has lent any support to the practice and it is indeed surprising how few systemic metastases occur, even when malignant cells are infused directly into the circulation through peritoneovenous shunts used to treat malignant ascites.

It is customary to remove the entire mesentery, ligating the vessels supplying the involved colon at their origin in order to achieve a radical removal of the draining and potentially diseased lymph nodes. Lymph node metastases occur along the pericolic arcade as well as down the main vascular pedicle. Ligation of the vascular supply and the need to remove the territory of bowel supplied by those vessels principally determines the amount of colon removed on either side of the cancer. While the prognosis is appreciably worse if the apical lymph node is involved by the cancer this does not necessarily imply that removal of all the draining lymph nodes will improve the prognosis for any or all patients. Traditional oncological surgical thinking was essentially anatomical, namely that tumours spread sequentially down an anatomical pathway and therefore it was necessary to remove the pathway to ensure cure. In an anterior resection for a mid-rectal cancer, for example, it was considered essential to ligate the inferior mesenteric artery at its origin from the aorta and also remove the ascending left colic artery. Resection in turn necessitated removal of the left colon because the marginal arcade around the splenic flexure is frequently incomplete. It seems irrational, from a biological standpoint, to think that ligation of the inferior mesenteric artery 2 cm more distal, so sparing the ascending left colic artery, would make any difference to the prognosis, and there is now evidence to show that high ligation provides no significant survival advantage. The nodal involvement indicates the malignant potential of the tumour and hence the probability of occult distant metastases and has no real bearing on the surgery, which should be directed to controlling the local and regional disease. High ligation should be reserved for those cases where the proximal nodes are clinically involved with tumour. Accurate lymph node staging may become as important as it is for breast cancer when adjuvant chemotherapy becomes more effective.

Another important general principle is that the ends of colon to be anastomosed after resection should have a good blood supply and be seen to bleed arterial blood, for ischaemia is the main cause of anastomotic failure and subsequent leakage. Tension on the anastomosis and overtight sutures are other important avoidable technical errors which may lead to failure of the anastomosis.

The material with which the anastomosis is made does not seem to be of critical importance. Colocolic anastomoses were, until fairly recently, fashioned in two layers, an outer seromuscular layer with a non-absorbable suture such as linen or silk and an inner absorbable, all coats, layer of ‘catgut’. The outer layer was usually interrupted, while the inner layer was often a continuous suture. Randomized studies have demonstrated that a single layer interrupted, all coats, anastomosis performs better than the two-layer technique and most surgeons now use a single layer of slowly absorbable suture material, such as polyglactin or polyglycolic acid, when performing a hand-sewn anastomosis. Interrupted sutures are probably safer than a continuous suture in the left colon where the stool is solid, because they cause less stenosis before the sutures have disintegrated. If the mucosa is everted the anastomosis is more likely to leak and it is therefore important to ensure that the mucosa is inverted by the sutures. While it is quite easy to do this with an all layers interrupted suture Matheson has introduced the serosa/submucosal modification which avoids the mucosa, inverts satisfactorily, and has been shown to have a very low leakage rate (Fig. 23) 1088.

Stapling instruments, first developed in Russia, are now widely used to perform anastomoses. Construction of a side-to-side anastomosis using linear stapling devices is suitable if the bowel can be delivered on to the abdominal wall. End-to-end anastomosis using a circular stapler is most useful for low pelvic anastomoses. The instruments certainly shorten operating time, particularly when performing a low anterior resection of the rectum, but they are much more expensive than sutures. Controlled studies do not show any convincing differences in leakage rates between hand-sutured anastomoses and stapled anastomoses, provided the surgeon is equally competent with either technique. Selection of method is one of surgeon preference and economics. Concern has been expressed about steel staples potentiating anastomotic recurrence of the cancer, for which there is some experimental support. However, controlled clinical comparisons show that stapled anastomoses are significantly less likely to develop tumour recurrence.

Bowel preparation is generally accepted to be necessary before performing an anastomosis, though this surgical dogma was questioned in a recent prospective series of colon resections. However there is convincing evidence that reducing the faecal load, and with it the colonic bacterial load, reduces both wound and anastomotic problems. Mechanical preparation is the most important means of producing a ‘socially clean colon’. Earlier studies also suggested that further reduction in bacterial flora with preoperative oral non-absorbable antibiotics such as sulphonamides, neomycin, and erythromycin base, conferred additional reductions in septic complications. However, the therapeutic effect appears to have been systemic from small amounts of antibiotic absorbed, and greater efficacy is achieved by using therapeutic doses of systemic antibiotics. Luminal antibiotics offer no additional advantage to mechanical bowel preparation provided systemic antibiotic prophylaxis is used.

Numerous methods of preparation have been proposed, tried, and used with little evidence of superiority, as measured by anastomotic or wound problems. Saline or osmotic purges such as Picolax® or Golytely® are currently the most widely used, give an acceptably clean colon in most patients, and are reasonably tolerated by most patients. Picolax preparation involves taking two sachets of non-absorbable sodium and magnesium salts dissolved in water and 2 l or more of clear fluid through the day on the day before operation. Golytely produces purgation with polyethylene glycol in about 4 l of fluid. The large volume of fluid required to produce a good preparation nauseates some patients.

Preoperative preparation is impossible in some patients with obstruction or perforation; in these circumstances a clean colon can be produced by intraoperative antegrade lavage of the colon after resection of the cancer. A small Foley catheter is introduced through the appendix stump or into the distal ileum, which is occluded proximally with a soft intestinal clamp. Warmed saline is then irrigated through the colon until it is perfectly clean. The effluent is collected by tying a large bore plastic tube, such as anaesthetic gas scavenging tubing, into the cut end of the bowel and placing the distal end in a suitably large bucket. The proximal end of the colon is then freshened before creation of the anastomosis. Closed collecting systems have been devised which are more aesthetic to use but are neither necessary nor economic. There are clear clinical and economic advantages if a primary resection and anastomosis can be performed safely in the acute situation, rather than a preliminary stoma or a resection and stoma with subsequent re-anastomosis.

Sepsis bedevilled colonic surgery for many decades and was the cause of much morbidity and occasional mortality. Systemic broad-spectrum antibiotic prophylaxis aimed against aerobic and anaerobic bowel pathogens has revolutionized colorectal surgical practice. Wound infection rates have been reduced from as high as 70 per cent to less than 10 per cent, while serious intra-abdominal sepsis has been reduced from as much as 20 per cent to less than 5 per cent.

Second-generation cephalosporin drugs active against Gram-negative organisms, in combination with metronidazole or tinidazole, active against anaerobic bacteria, are effective and widely used. Different drugs used singly and in combination for prophylaxis, provide numerous study data but little therapeutic advantage. The essential for effective prophylaxis is a high tissue level of antibiotic at the time of operation and, therefore, of potential tissue contamination. The antibiotics should be administered intravenously at the induction of anaesthesia; if the operation lasts longer than 3 or 4 h a repeat dose should be given to maintain therapeutic tissue levels. A delay of four or more hours after operation before giving the antibiotic provides no prophylaxis against infective complications and is as good as no prophylaxis at all. Good tissue levels of metronidazole can be achieved from intrarectal metronidazole suppositories; this is acceptable for proximal colonic resections but not for rectal resections.

There is some debate over the number of doses of antibiotic required to provide effective prophylaxis, but there is definitely no advantage in giving more than three doses of most antibiotic regimens, while one good dose may well be as effective and more economical. The loading dose required may be higher than that used to treat an established infection. For example, cefuroxime 1.5 g and metronidazole 1 g is a very common regimen for prophylaxis. If the patient has an abscess or infection at the time of resection it is clearly prudent to treat the infection with a full course of antibiotic therapy. Unnecessary continuation of treatment or the use of inappropriate antibiotics, such as clindamycin, predisposes to Clostridium difficile infection and pseudomembranous colitis, a severe and sometimes fatal complication when not recognized early. Severe persistent postoperative diarrhoea is the commonest symptom and the diagnosis is confirmed by identifying the organism or toxin in the stool. Treatment with vancomycin is usually effective.

Elderly patients with malignancy are particularly liable to venous thrombosis and pulmonary embolism, a sometimes fatal and often preventable complication of surgery. Additional risk factors are obesity, varicose veins, and a previous history of thrombosis or embolism. It is therefore necessary to consider perioperative prophylaxis for patients undergoing resection for colonic cancer. Subcutaneous heparin, 5000 i.u. three times daily until the patient is fully mobile, is the most common method of prophylaxis and the one for which there is most supporting data. Blood loss may be greater than usual if the thrombin time is increased. Perioperative Dextran 70 infusion is an alternative method, but allergic reactions to the Dextran occasionally occur. Compression stockings and active calf exercises are also commonly used while the patient is bedbound, but evidence for their efficacy is scant. Intraoperative calf muscle stimulation is claimed to provide effective prophylaxis, but again the evidence is unconvincing.

Right hemicolectomy is the standard operation for cancers of the caecum and ascending colon, removing as little of the terminal ileum and as much of the ileocolic artery as is possible while still maintaining the blood supply to the terminal ileum. More terminal ileum may need to be removed for tumours near the ileocaecal valve, but vitamin B&sub1;&sub2; deficiency and bile salt diarrhoea are common complications if more than 50 cm is resected. The ascending right colic artery is also ligated at its origin from the ileocolic artery. The distal extent of colon removed depends upon the site of the tumour. The resection may need to extend to the mid-transverse colon for cancers near the hepatic flexure. The terminal ileum is then anastomosed to the proximal transverse colon in one or two layers (Fig. 24) 1089. On the other hand the caecum may be preserved when resecting tumours at the hepatic flexure and a colocolic anastomosis fashioned, thus preserving the reabsorptive capacity of the caecum and reducing the risk of postoperative diarrhoea (Fig. 25) 1090.

A midline incision is the most common approach, but small cancers are readily resected through horizontal or oblique (Rutherford-Morison) muscle cutting incisions. The latter approaches cause less pain because tension on the wound is less and the nerve supply to the wound is easily blocked with bupivicaine to provide immediate postoperative analgesia. Furthermore the wound heals with less spread or keloid scarring and is therefore more cosmetically acceptable. A midline incision is, however, easier to extend should more resection be necessary. Endoscopic techniques for colonic resection are already being developed in order to minimize the trauma and pain of the abdominal wall incision. The entire operation, including the anastomosis, can be performed endoscopically. Alternatively the colon can be mobilized endoscopically before delivering the bowel and tumour through a small abdominal incision where the resection and anastomosis are completed by hand. These new techniques look promising and may reduce hospital stay but are more time consuming and have still to be refined and tested rigorously against conventional operations.

The most usual approach is through an upper midline incision that may extend a little below the umbilicus, though a transverse muscle cutting incision provides excellent exposure for the transverse colon. The lymphatic drainage for cancers of the transverse colon is along the middle colic vessels to the root of the superior mesenteric artery. Resection therefore removes most of the colon supplied by the middle colic artery and in particular the splenic flexure where the pericolic vascular arcade is often incomplete (Fig. 26) 1091. If lymph nodes along the ascending right colic artery are enlarged and potentially cancerous, the resection may need to be extended to include part of the ascending colon. The omentum is removed en bloc with the tumour leaving the gastroepiploic vessels. Resection of part of the stomach or jejunum may be required if the colonic tumour is invading either organ. The resection is completed by an end-to-end colocolic anastomosis.

Generous exposure is necessary to provide good access to the splenic flexure, which is often high and attached to the spleen, and so again a midline incision, extending above and below the umbilicus, is most often used, though some surgeons prefer an oblique Rutherford-Morison incision. Careless traction on the colon while mobilizing the splenic flexure may tear the spleen and produce haemorrhage sufficient to require splenectomy which increases the risk of thromboembolism in the immediate postoperative period and the risk of overwhelming infection in the future, a complication therefore to be avoided. The proximal resection should include the splenic flexure because the ascending left colic artery is ligated at its origin and the pericolic anastomosis is often inadequate to ensure a good blood supply to the anastomosis. The transverse colon is mobilized sufficiently to perform a tension free anastomosis. The distal resection extends into the sigmoid colon preserving one of the distal sigmoid arteries and the pericolic vascular arcade. The inferior mesenteric artery is seldom ligated at its origin but should it be necessary to take this vessel because of node metastases, the distal resection should be extended to the proximal rectum which is adequately supplied from the middle rectal artery.

Three-quarters of all large bowel cancers lie distal to the splenic flexure and most are found in the sigmoid colon. This is readily approached through either a lower midline incision or a Rutherford-Morison incision. The sigmoid colon has the advantage of usually being mobile and on a generous mesentery and therefore the dissection seldom needs to be extensive. If the sigmoid colon is also affected by diverticular disease, mobilization needs to be more extensive to provide a tension-free anastomosis, particularly when there is inflammation and fibrosis. The sigmoid vessels are ligated as they arise form the inferior mesenteric (Fig. 28) 1093, preserving the ascending left colic and superior rectal vessels. If the proximal mesenteric nodes are involved then the inferior mesenteric artery may need to be taken at its origin and an extended left hemicolectomy performed to provide sufficient tumour clearance for adequate local clearance and control.

Sigmoid cancers often invade surrounding structures such as the abdominal wall, ovary, uterus, bladder, or small bowel. In such cases en bloc removal of the invaded organs is necessary. If the tumour is stuck to the abdominal wall, the peritoneum and muscle are excised, preferably using diathermy. Care must be taken to identify and preserve the ureter; preservation may be difficult, particularly when there is considerable pericolic inflammation. If the ureter is invaded then the involved segment may be removed and the ureter re-anastomosed obliquely with fine catgut or polyglactin, thereby reducing the risk of ureteric stricture. If too much ureter has been removed to restore continuity the proximal ureter may be joined end-to-side to the contralateral ureter producing a crossed ureteroureterostomy. Left oophorectomy or hysterectomy may also be required to give sufficient clearance for local control of the cancer. The bladder is more commonly invaded in males, and partial cystectomy may be necessary. The bladder is then closed in one or two layers of absorbable sutures and drained with a catheter for about 10 days to ensure adequate healing.

Rectal cancer presents a challenge for the surgeon both technically and in clinical judgement. About 25 per cent of large bowel cancers develop in the rectum; since the prognosis is generally more favourable than that for more proximal tumours, cure should more often be the rule. Some 30 years ago nearly all patients with rectal cancer were treated by excision of the rectum and anus and left with a permanent left iliac fossa end colostomy, because it was considered both dangerous and therapeutically inadequate to remove less and to attempt a restorative anastomosis. Improvements in operative technique and careful clinical studies have shown both premises to be incorrect, at least for most lesions of the midrectum and all in the proximal rectum. The skill and judgement comes in selecting patients for restorative resection, local excision, abdominoperineal excision or, sometimes, non-operative methods when palliation is the sole objective.

The distance of the tumour from the anal canal is central in deciding upon an anterior resection of the rectum with anastomosis, because sufficient clear margin is required beyond the cancer to minimize the chance of local recurrence, while satisfactory long-term function is better with a greater residual rectal reservoir. Radical resection close to the anus not only reduces the rectal reservoir but also impairs internal sphincter tone, thereby reducing anal control and continence. It used to be said that a clear margin of at least 5 cm distal to the tumour was essential to prevent anastomotic or local recurrence; more recent data suggest that cancers seldom infiltrate the bowel wall as much as 1 cm beyond the macroscopic margin, and a 2-cm margin is therefore both ample and safe, provided that the cancers are well or moderately well differentiated. Poorly differentiated or undifferentiated tumours may spread further along the bowel wall and a margin of at least 5 cm is advisable. What appears more important is that tumours that have extended through the bowel wall may spread more extensively within the mesorectum or laterally around the middle rectal vessels. All the mesorectum should therefore be removed, and the lateral clearance should be at least as far as the sacral parasympathetic nerves to prevent local recurrence. The obvious extension of this concept is that a complete pelvic node dissection, similar to that of a Wertheim's hysterectomy would improve survival, but it is disappointing that en bloc internal iliac lymph node removal adds nothing but morbidity and confers no survival advantage.

Tumour stage remains, at present, the most important factor in deciding upon the surgical approach for the patient; tumour differentiation is a useful subsidiary factor. For example a T3, N1 undifferentiated tumour is biologically less favourable, requiring greater margins of clearance to prevent local recurrence, than a T1, N0 well differentiated lesion. The latter may be quite suitable for local excision if the tumour is small enough.

Anterior resection is now the standard radical operation for cancers of the upper and mid-rectum and for the more favourable tumours of the distal rectum when a 2-cm margin of clearance is possible while still leaving the anal transition zone intact.

The usual approach is through a lower midline incision, which may extend above the umbilicus if the splenic flexure requires mobilization. A few surgeons advocate a subumbilical horizontal muscle cutting incision as an alternative. Early ligation of the vascular pedicle before the tumour is handled or dissected, originally advocated by Turnbull, may reduce the incidence of hepatic metastases, though the evidence is not yet conclusive. Flush ligation of the inferior mesenteric artery at its origin from the aorta is still practised by many surgeons for cancer clearance, although it is not based on a proper perspective of tumour biology and there are no satisfactory data to support the manoeuvre. Ligation just distal to the origin of the ascending left colic artery, which is my own preference, does not compromise survival and ensures a good blood supply to the left colon which may be lost by a flush ligation of the inferior mesenteric artery (Fig. 29) 1094. If the apical node is involved a flush ligation is necessary to control the local disease although the overall prognosis is poor.

Pelvic dissection requires accuracy firstly to avoid cutting too close to the tumour and secondly to spare the autonomic nerves to the bladder and penis. Posteriorly, the dissection is carried caudad behind the mesorectum but anterior to the sympathetic nervi erigentes. Laterally, the middle rectal vessels are secured on the side wall of the pelvis but medial to the parasympathetic plexus supplying the bladder. As the rectum and tumour are mobilized out of the pelvis, the surgeon must avoid dissecting too close to the cancer, thereby compromising the prospect of cure. Cutting close to the tumour is a common mistake particularly when traction is applied to the rectum and the surgeon is inexperienced. When the rectum is fully mobilized the mesorectum is cleaned from the bowel wall and a cross-clamp is applied just proximal to the proposed site of the anastomosis and the rectum is irrigated with 1 per cent chlorhexidine solution, povidone iodine solution, or perchlorate of mercury, to kill any viable tumour cells that have been shed into the bowel lumen and which could implant into the anastomosis. Shed neoplastic cells are a potential cause of local recurrence of cancer, particularly at the suture line, though most local pelvic recurrences arise from residual tumour in the mesorectum, in the lateral ligaments, or on the side wall of the pelvis. These recurrences may then invade the anastomosis to appear at the suture line.

The anastomosis is either fashioned manually with sutures or mechanically with stapling instruments. Each method is equally effective, provided the surgeon is adept at either technique although there is some evidence to suggest that tumour recurrence may be less in patients with staples. There are numerous minor technical modifications to suit individual surgical idiosyncrasies but the essentials common to all are a good blood supply to the cut ends of the bowel and an anastomosis free from tension. Poor blood supply and tension at the suture line are likely to cause anastomotic leakage, sepsis, and sometimes peritonitis. For this reason many surgeons insist that the splenic flexure should be mobilized fully, but there are no objective data to support this practice in all cases. Repeated taeniamyotomy proximal to the anastomosis, originally used to treat diverticular disease, will also increase the length of the colon, and reduce tension on the anastomosis and colonic contraction. However, the incidence of anastomotic leakage and local tumour recurrence after anterior resection has been shown to depend upon the surgeon's skills almost as much as the tumour biology and to vary four-fold from the best to the worst results. Each surgeon must audit his performance to ensure that his technique is unimpeachable.

The functional results of anterior resection are usually good, sometimes moderate, but occasionally poor in terms of stool frequency, urgency, and continence, particularly when the colon is anastomosed to the upper anal canal. Internal anal sphincter function is often impaired by the necessary cancer clearance, probably due to damage to the nerve supply. External anal sphincter function usually remains normal but is insufficient to maintain complete continence, particularly when the stool is loose and the proximal colon is not compliant. Bowel function continues to improve spontaneously for 12 to 18 months, but early functional results can be improved by fashioning a small colonic pouch to increase colonic compliance and then anastomosing the pouch to the anus.

This operation is now largely reserved for larger T2–3, N 0–3 tumours of the distal rectum and those which are poorly differentiated, when a safe anastomosis is not practicable. The original operation described by Miles is for surgeons operating single-handed. The rectum is mobilized down to the pelvic floor through a subumbilical midline incision. The bowel is divided and the distal sigmoid brought out as a left iliac fossa colostomy. The distal colon is oversewn and is folded down into the pelvis before closing the pelvic peritoneum over the rectum. The abdominal wound is then closed and the colostomy is matured using catgut or polyglactin sutures. A marginally better mucocutaneous junction is produced if the sutures are placed at a subcuticular location rather than through the skin, in the manner described by Turnbull. The patient is then turned on to the left lateral side for the perineal dissection. A perianal elliptical incision is made to mobilize and deliver the anus and distal rectum. The most common error in the perineal dissection is to dissect too close to the rectum, and therefore to the tumour, defeating the prime objective of controlling the local disease. To avoid this mistake the initial dissection is carried well into the ischiorectal fossa, dividing the pudendal vessels supplying the anus before dividing the levator ani muscles laterally. The tip of the coccyx may also have to be removed to give adequate clearance of posterior cancers. Anteriorly, the dissection in males is constrained by the urethra, prostate, and seminal vesicles, though the last can be removed if necessary. A Cluttons sound or urethral catheter provides a useful landmark to prevent damage to the urethral bulb. Mobilization is much easier in females, for the back wall of the vagina can be removed with the rectum, providing a particularly useful manoeuvre for clearing anterior tumours. The perineal tissues are then closed loosely over a drain. There is much discussion in the medical literature on the merits of different methods and types of drain with no clear consensus, which leaves surgeons to exercise their preferences: mine is to use an 8-mm tube drain through the perineal wound.

Abdominoperineal resection is now more commonly performed with the patient in the modified lithotomy Trendelenberg position and two surgeons operating simultaneously. The perineal part of the operation is often, wrongly, delegated to the less experienced operator, which may account for the unacceptably high rates of local recurrence in some series. It is my contention that the perineal operator should be the more experienced, or if a trainee is operating then it must be with direct supervision to avoid the mistake of cutting close to the tumour or damaging the urethra. The principles of the operation are otherwise similar to those of the Miles procedure.

Hartmann's operation is, in essence, an anterior resection of the rectum without an anastomosis. The proximal bowel end is brought out as a left iliac fossa colostomy and the rectal stump is oversewn deep in the pelvis. The operation is seldom performed electively or for cure and is usually reserved for palliation, or sometimes as a preliminary procedure for acute cases. The surgeon preparing to perform an anterior resection may be faced with a much more advanced tumour with pelvic peritoneal tumour seedlings and perhaps extension to the side wall of the pelvis or even more widespread metastases. Good palliation may be provided by removing the primary cancer to relieve intolerable bowel symptoms, but the certainty of early local recurrence and failure to palliate precludes creation of an anastomosis. Patients presenting with acute malignant colonic obstruction or peritonitis from perforation of a rectal cancer may also be unsuitable candidates for resection and primary anastomosis. Perforation of a cancer implies incurability because of pelvic seeding with malignant cells. Pelvic sepsis increases the chance of anastomotic breakdown and leakage, and may also encourage local recurrence. Safer palliation is therefore achieved by resection without anastomosis. However, for acute colonic obstruction the colon can be prepared satisfactorily by intraoperative antegrade colonic lavage to produce a clean bowel which may be anastomosed safely as an alternative to Hartmann's operation, provided the disparity between the ends to be joined is not large. A temporary defunctioning proximal stoma may give additional safety.

Some small early well differentiated rectal cancers (T1–2, N0) may be managed by local treatment with the expectation of cure. However, less than 10 per cent of patients are likely to be suitable for such treatment, although the precise proportion of patients with such lesions is unknown. Numerous techniques have been advocated: their proponents claim success but there are no satisfactory comparative studies to support such claims. Local recurrence of the cancer because of inadequate clearance or tumour implantation is the principal problem which may jeopardize alternative methods of treatment. Per anum disc excision of the rectal tumour, with a margin of more than 0.5 cm and suture of the defect in the rectum, is the technique most commonly used and one that provides an excellent specimen for histological assessment. This method, however, requires wide dilatation of the anus with the risk of subsequent faecal incontinence from damage to the anal sphincters, and is inadequate for lesions higher in the rectum because of inaccessibility. The method has been refined and the operating range extended by Buess using a specially modified operating proctoscope with binocular magnified vision and endoscopic instruments inserted through side ports in the proctoscope. The operating field is maintained by continuous CO&sub2; insufflation, magnification provides accuracy, and the anus is not forcibly dilated, but the technical skills require practice. Nevertheless the method looks promising, especially for small cancers and villous adenomas.

Alternatively lesions can be electrocoagulated, using conventional surgical diathermy apparatus, the charred tissue being curretted until soft healthy tissue is reached. Good results have been reported but again the anus requires forceful dilatation and no specimen is provided for histological examination. Endoscopic resection, using a modified urological instrument, glycine irrigation, and the same techniques used to resect bladder tumours or the prostate, overcomes these problems but the resected specimen is less easy to stage than a disc excision because of fragmentation. The results of endoscopic resection appear comparable to those of major resection for similar biologically favourable tumours. Both these local methods are simple and can be repeated whenever necessary. Larger cancers in frail elderly patients, unsuitable for major surgery, can be treated by endoscopic resection with the objective of symptom and tumour control rather than cure (Fig. 30) 1095. A few tumours regress and disappear with repeated resection while a similar proportion, particularly undifferentiated lesions, continue to grow and metastasize. The majority are contained.

Irradiation and, in particular, high-dose interstitial irradiation has also been claimed to give good results although there is an increased risk of carcinogenesis in the long term.

Laser therapy may also have a place in the treatment of smaller rectal and colonic tumours. The laser light is delivered through a colonoscope or flexible sigmoidoscope which allows treatment to be given to an outpatient under intravenous sedation. The most widely used laser is the neodymium YAG laser which vaporizes and coagulates the tissue to a depth of 5 mm. Small cancers can be destroyed but larger lesions are difficult to treat satisfactorily even with energies as high as 40 kJ. Photodynamic therapy also looks promising, but again for small lesions. This method requires pretreatment with a light-sensitive haematoporphyrin derivative which is selectively retained in tumours. Laser light, of the correct wavelength to be absorbed by the haematoporphyrin derivative, is delivered by a copper or gold vapour laser. The absorbed energy induces release of oxygen free radicals, which in turn produce local tissue damage and destruction within the tumour. The main complication of the treatment is a skin rash induced by exposure to sunlight. Patients therefore need to remain indoors and keep the skin covered for a few days after taking the haematoporphyrin derivative.

Transrectal ultrasonography appears an important method for selecting patients and tumours for local therapy and also for monitoring progress. Ultrasound provides good images of the rectal wall and therefore the depth of tumour invasion (Figs. 19, 20) 1084,1085. It is also able to image enlarged lymph nodes.

Nearly one-fifth of patients with colon cancer still present acutely with either obstruction or perforation and have a worse prognosis, stage for stage, than those presenting electively.

Approximately 15 per cent of patients in large unselected series present with acute large bowel obstruction. About half are left-sided cancers which suggests that proportionately more right-sided cancers present with obstruction since 75 per cent of colon cancers occur distal to the splenic flexure. This is not surprising because the stool is liquid in the caecum, which accounts for the later presentations with anaemia or obstruction; any stenosis in the left colon is likely to produce a change in bowel habit which prompts presentation to the doctor. In general, obstructing cancers are more advanced than those presenting electively which accounts for some, but not all, of the poorer prognosis.

The history of patients presenting with acute colonic obstruction is often surprisingly short, with rapidly progressive constipation, abdominal distension, abdominal pain, and the inability to pass flatus. Vomiting is a common consequence of the abdominal pain, but faeculent vomiting is a late symptom and only occurs when the ileocaecal valve is incompetent.

The physical signs depend on the site of the obstruction. Right-sided lesions usually present with what appears to be distal small bowel obstruction; patients suffer central abdominal colic, distension, borborygmi and later, faeculent vomit. Tenderness is not a common feature unless the obstruction is prolonged, but dehydration, hypotension, and tachycardia occur early because large volumes of fluid and electrolytes are sequestered in the ileum and are lost as vomit. A plain abdominal radiograph confirms the diagnosis of distal small bowel obstruction with gross gaseous distension of the small bowel and little or no gas in the colon. A contrast study of the colon is seldom required before surgical treatment.

Left colonic obstruction more commonly presents with gross distension limited to the colon: the ileocaecal valve remains competent in about 60 per cent of patients. Very high intraluminal pressures are generated, reducing circulation to the colon and, in particular to the thin-walled caecum, which is in danger of ischaemic rupture. Clinically, the distension is more in the flanks. Tenderness over the caecum is an early sign and signifies impending rupture, faecal peritonitis, and endotoxaemia, which often progresses to multiorgan failure and death. Obstructed bowel sounds are not a dominant feature: a secondary ileus with scant tinkling sounds is more usual. The plain abdominal radiograph shows gross colonic distension and there is often a sharp cut-off at the point of obstruction, with an airless distal colon. A contrast study of the colon is, however, essential to confirm the diagnosis, identify the site of obstruction, and exclude conditions such as pseudo-obstruction, acute myxoedematous colonic obstruction, and ischaemic colitis.

Acute colonic obstruction requires urgent surgical relief, but before going to the operating theatre the patient needs rehydration and re-expansion of the circulatory volume. If there are signs of peritonitis, antibiotics should also be given parenterally.

Obstructing right and transverse colon cancers are relatively easily treated by an immediate right hemicolectomy or extended right hemicolectomy, respectively, and an end-to-end ileocolic anastomosis which rids the patient of the primary and treats the obstruction at one operation. Left colonic and rectal obstruction, however, presents more of a problem and surgical strategies are still evolving. It used to be considered that emergency left colonic resection in obstructed, unprepared bowel carried too high a mortality rate and that the obstruction was best relieved by a caecostomy or transverse loop colostomy. The patient was then restored to better health, the bowel prepared by washouts, and the left colon resected and anastomosed at an elective procedure 2 or 3 weeks later. The colostomy was sometimes closed at the same time as the resection but more usually later when the patient had recovered fully from the resection, so completing a three-stage operation. The cumulative mortality and morbidity of the three stages is high, particularly in elderly patients. The length of hospital stay is also long, which substantially increases the costs of treatment. Such multistaged resections are now seldom performed in modern hospitals, but an emergency colostomy still has a place if the patient is far from good medical help.

Over the last 20 years evidence has accumulated which supports definitive single-stage treatment for the obstructed left colon. The present choice is between an extended or subtotal colectomy and ileosigmoid or ileorectal anastomosis, removing all the obstructed colon, and a more limited colonic resection, with on-table colonic lavage to decompress and clean the obstructed colon, and a colocolic, or colorectal anastomosis. There are no convincing comparative data to decide the issue. Early evidence suggested that limited colonic resection and anastomosis was liable to leak because the bowel was unprepared, there was great disparity between the ends to be joined, and the vascular supply to the obstructed colon was suspect; therefore resection of all the obstructed colon became fashionable. Mortality and morbidity was certainly lower but at the expense of poorer long-term, bowel function. With on-table colonic lavage, systemic antibiotics, and better suture material, anastomotic techniques, and postoperative support, the pendulum has swung back to more conservative resections with satisfactory results. The obstructed colon is more readily mobilized if it is first decompressed by inserting a 19 gauge needle into the lumen and attaching it to suction. The anastomosis may be protected by a caecostomy or loop ileostomy for added safety.

Rarely a Hartmann's operation is performed for an obstructing left colon cancer with the intention of subsequent re-anastomosis. The operation is usually reserved for palliation when pelvic peritoneal metastases would make an anastomosis hazardous or risk recurrent obstruction.

Perforation is less common than is obstruction, occurring in about 5 per cent of patients. The site of perforation is usually within the tumour and is not associated with obstruction but is the consequence of tumour necrosis. The patient presents with severe, often sudden, abdominal pain and signs of peritonitis. Rapid cardiovascular collapse and endotoxaemic shock usually signify a major leak and faecal peritonitis, rather than a small perforation with purulent peritonitis, but may also follow delay in diagnosis. Because the perforation occurs through the tumour, malignant cells are disseminated throughout the peritoneum, making cure unlikely. Good palliation is therefore the paramount aim and long-term survival is an unanticipated bonus.

The patient requires pain relief, rapid fluid and colloid resuscitation, and systemic antibiotic therapy before moving to the operating theatre. Delay in diagnosis and definitive treatment greatly increases the risk of subsequent multiorgan failure and death. The surgical judgements are similar to those required for obstruction, with the caveat that as few tissue planes as possible should be opened in order to reduce the risk of abscess formation. Resections should be conservative rather than radical, and the ends of the bowel should either be brought out as a stoma and mucous fistula, or, if a primary anastomosis is fashioned, a loop ileostomy should be made because an anastomosis is more likely to leak in the presence of severe sepsis. In addition the peritoneal cavity should be lavaged well with saline alone or containing antibiotics effective against enteric organisms. Tube drains should be inserted into the pelvis and subphrenic spaces. Systemic antibiotic therapy should be continued for at least 5 days. Occasionally it is wise to leave the abdominal wound completely open as a laparostomy to allow peritoneal exudates to drain freely and to enable easy, and if necessary repeated, access to spaces where pus may collect. Many patients will require inotropic support, visceral vasodilator drugs, and careful surveillance of cardiovascular function, and will therefore need to be in an intensive therapy unit postoperatively.

Evidence from treating other types of cancer suggests that adjuvant therapy should be successful, and is most likely to be effective against minimal disease and micrometastases. Such treatments should, in theory, be used postoperatively and in patients at greatest risk of having residual but microscopic disease (T3 and N2 lesions). Chemotherapy, immunotherapy, and irradiation have all been tried as adjuvant treatments before and, more often, after surgery without unequivocal success. Meta-analysis of many randomized trials of both irradiation and chemotherapy suggest some small survival advantage in some groups of patients for each method of treatment. Such evidence has been the stimulus for a large multicentre study in the United Kingdom (AXIS) to try to establish the roles of chemotherapy and irradiation. Very large numbers of patients are required to provide convincing proof because of the many clinical and pathological variables that affect outcome and must be accommodated.

Radiotherapy may be administered either before or after removal of the primary tumour with different aims in view. Patients with large T3 or T4 rectal cancers fixed to the pelvis on clinical and CT examination and therefore not usefully removable may have their tumours ‘downstaged’ sufficiently to convert an irremoveable lesion into one that is resectable some 6 weeks after a radiation dose of 50 to 60 cGy. Such pelvic irradiation slightly but significantly reduces the risk of local recurrence after resection, but this is not yet translated into a corresponding survival advantage. Postoperative irradiation of the tumour bed of less advanced resectable T3–4 cancers, particularly of the rectum and caecum, may also reduce the risks of local recurrence, which in turn could produce a 5 to 10 per cent improvement in 5-year survival rates. Lack of convincing evidence in favour of radiotherapy will, for the present, encourage surgeons to be selective and to restrict adjuvant irradiation to such patients judged clinically to be at particular risk of local recurrence where there may be a small but useful clinical yield.

Similarly, postoperative adjuvant chemotherapy, principally with 5-fluorouracil as a single agent, may provide a small survival advantage in selected patients, such as those with N2 tumours, who are at particular risk of occult hepatic metastases. The most promising results are from infusion of 5-fluorouracil into the portal vein to provide a high therapeutic dose directly to the micrometastases. Experimental and clinical studies suggest that the incidence of hepatic metastases is slightly less after portal vein infusion, but the differences are not yet sufficiently large or certain to warrant widespread use. Other agents such as Adriamycin, mitomycin C, doxorubicin, and cisplatin have been tried but are toxic and have, at present, no useful role as adjuvant chemotherapeutic drugs for colorectal cancer. Irradiation has been tried in combination with 5-fluorouracil and semustine as adjuvant therapy in patients with rectal cancer with apparent improvement in survival but at the cost of increased toxicity. The addition of folinic acid however improves the efficacy of 5-fluorouracil while reducing the toxicity. There is, however, little doubt that chemotherapy and irradiation, perhaps in conjunction with biological modulators such as interferons or cytokines, will in the future, play an increasingly important role in the treatment of patients with Dukes' B and C lesions.

Adjuvant immunotherapy with agents such as BCG and Corynebacterium parvum have been tried in patients with more advanced colon cancers with the objective of stimulating the immune system non-specifically, but sufficiently to produce a lethal effect on micrometastases. The theory is that immune surveillance can affect tumour survival and, indeed, those cancers with a marked lymphocytic and macrophage infiltration, and a histiocytic response in the adjacent lymph nodes have a better prognosis. Unfortunately the results of clinical trials have been disappointing. Levamisole, a non-specific immune stimulant, has also been tried both alone and in combination with 5-fluorouracil with mixed results. One study of 5-fluorouracil and levamisole appeared sufficiently encouraging for the combination to be widely used for adjuvant therapy in the United States although the control comparison was questionable.

Many patients present with incurable colonic cancer, or suffer a recurrence after surgery. This situation makes management difficult, for the burden of treatment must be weighed carefully against the potential gain, not only in length of life but, more importantly, in the relief of symptoms and the quality of the time remaining. The prospect of cure to some extent justifies radical therapy but the aims of palliation must be different and therefore the measures modified to achieve those aims. The options open to the surgeon are surgery, irradiation, chemotherapy, and drug management of symptoms.

Patients often present with symptoms from the primary colonic cancer and totally asymptomatic metastases. In these circumstances it is obvious that the patient is, in general, best served by removing the primary, provided that the burden for the patient is not too great compared to the expected survival. Reducing the tumour load appears to provide better palliation than leaving it in situ. If, for example, a patient has a readily removable caecal cancer causing obstruction and has hepatic metastases, the best procedure is to perform a right hemicolectomy. However, if the caecal cancer is invading the abdominal wall and there are also widespread peritoneal metastases then the obstruction is best relieved by performing an ileotransverse anastomosis. Resection is usually an easier choice for most colonic cancers, but palliative removal of rectal, and in particular low rectal cancers, poses problems: the risks of purely palliative surgery are greater and the functional results worse in the short term. In these cases alternatives to resection may be more appropriate and careful clinical judgement becomes crucial. Occasionally a palliative abdominoperineal resection or a low anterior resection with a coloanal anastomosis is justified provided that the primary is readily removable. If the tumour has infiltrated the pelvis widely, local recurrence is almost inevitable and after an abdominoperineal excision of the rectum may fungate through the perineal wound, or vagina, making the patient's position worse rather than better. Therefore a palliative endoscopic transanal resection (ETAR) or laser therapy is often more appropriate. Endoscopic resection debulks the tumour and provides good relief of symptoms such as diarrhoea, bleeding, urgency, incontinence, and tenesmus, but provides less relief of the pain produced by the tumour infiltrating the pelvic nerves. Similar results can be obtained with a neodymium YAG laser using 30 to 40 kJ of energy at each session. The main difference between endoscopic transanal resection and laser therapy is that laser treatment requires more frequent repetition at about 6-week intervals while the latter only needs repeating every 3 to 4 months. Occasionally, endoscopic transanal resection or laser therapy may be combined with an end left iliac fossa colostomy, when the cancer has invaded the anal sphincter or fistulated into the vagina producing faecal incontinence. The colon can easily be approached and divided through the small incision made for the stoma.

Cryotherapy using liquid nitrogen probes can produce tumour destruction but it appears slower and less effective than laser therapy. Photodynamic treatment using copper or gold vapour lasers with systemic haematoporphyrin derivative to concentrate the laser energy in the tumour, may destroy small anastomotic recurrences of cancer or cutaneous metastases. The laser is expensive and the haematoporphyrin derivative produces quite marked light sensitivity, so that treatment must be matched to the patients symptoms.

A very few patients present with an isolated secondary in one or other lobe of the liver which is suitable for removal either by excision of a hepatic segment or a hepatic lobectomy. Hepatic metastases draw most of their blood supply from the hepatic artery, in contrast to normal hepatocytes which receive 80 per cent from the portal vein. This physiological fact enables a selective attack to be made on the metastases by obliterating the arterial supply to the secondaries or delivering cytotoxic agents via the hepatic artery. Arterial blood flow can be stopped effectively by arterial embolization with fibrin clot or gelfoam. Alternatively the hepatic artery may be ligated at operation, although this is a much more invasive procedure and is therefore usually combined with resection of the primary. Symptomatic metastases can also be successfully treated with interstitial laser therapy. Several laser fibres are inserted into the metastases with ultrasound guidance under local anaesthesia and some sedation. Low energy Nd YAG laser light is then passed into the tumour for 2 to 3 min to produce gentle heat coagulation of tissue for a radius of 1 to 2 cm around the fibre tip.

Gastrointestinal tumours are generally considered to be radioresistant, but the symptoms produced by an unresectable rectal cancer can often be relieved by external beam irradiation. Useful palliation may be obtained at the relatively low dose of 35 cGy, but increasing the dose to 55 cGy improves the response without greatly increasing the morbidity, although some radiation cystitis may occur. Pain in particular, is better relieved than by either endoscopic transanal resection or laser therapy. Resection and radiation can be used in combination to produce rapid relief of symptoms but an indolent ulcer may follow endoscopic transanal resection performed some months after higher dose irradiation. Irradiation is seldom used for intra-abdominal colon cancer primaries or metastases and is ineffective for hepatic metastases.

Colonic cancers are relatively resistent to current chemotherapeutic agents, both singly and in combination. Nevertheless useful subjective and objective responses can be obtained even though the therapeutic margin is small. 5-Fluorouracil is the most widely used agent, either alone or in combination, and provides an objective response rate of about 15 per cent for metastatic disease when infused intravenously at a dose of 20 mg/kg day over 5 to 7 days, repeated every 4 to 6 weeks providing there is continued evidence of a reasonable clinical response. The response to 5-fluorouracil is enhanced by the addition of high dose folinic acid: partial response rates as high as 45 per cent have been claimed with no increase in toxicity. Semustine may also improve the response to 5-fluorourcil though at the expense of increased morbidity. The principal adverse reactions are nausea and bone marrow suppression, producing leucopenia and thrombocytopenia. However, these complications are time- and dose-dependent, and can be reduced easily by adjusting the delivery schedule. When palliation is the objective it would seem reasonable to withhold treatment until metastases cause symptoms, but there is evidence that treatment at the time of diagnosis, before symptoms appear, improves both survival and palliation.

Many other chemotherapeutic agents, such as the platinum compounds mitomycin C, and doxorubicin, have been used and useful responses claimed but usually at the cost of more adverse reactions and no convincing evidence of improved therapeutic efficacy.

Immunotherapy has a long but rather disappointing history in the treatment of advanced cancers, particularly with agents such as Corynebacterium parvum or BCG, but recent advances in molecular biology, recombinant technology, and immunology promise exciting therapeutic possibilities. For example, lymphokine-activated killer lymphocytes are to some extent tumouricidal in a non-specific manner, while specifically activated tumour infiltrating lymphocytes are many times more active against malignant cells. The cytokine interleukin-2, which activates lymphocytes, has been infused alone and in combination with other growth regulators such as the interferons or chemotherapeutic drugs in the treatment of cancers. Encouraging results have been obtained in some tumours, including colonic cancers, but with unpleasant side-effects. These cytokines, with the macrophage product tumour necrosis factor, are also responsible for many of the paraneoplastic phenomena such as anorexia, malaise, and cachexia, which are the main side-effects of treatment. Cytokines may also promote tumour angiogenesis, metastasis formation, and hypercalcaemia, so that therapeutic potential is mixed.

There is, however, little doubt that a therapeutic revolution which will substantially improve the outlook for patients with colorectal cancer and affect the way in which such patients are managed lies just round the corner.

The diagnosis of hepatic metastases usually signifies incurability but occasionally a metastasis is isolated and situated within the liver at a site suitable for resection with some expectation of prolonged survival and possibly cure. The results of liver resection are good where the metastasis is solitary, but poor when multiple, even when the metastases are confined to one lobe of the liver.

There is, at present, no convincing evidence that resection improves survival, even in the few patients with a single metastasis, for there are no controlled trials of hepatic resection. A possible explanation, if a little cynical, is that solitary slow growing metastases merely represent favourable biology with a naturally long survival while multiple metastases indicate the opposite. Hepatic resection is technically feasible with a low risk of morbidity and mortality in experienced units, providing the remaining liver has a good vascular supply and venous drainage as well as normal bile ducts. Ideally, the metastasis should be peripheral and confined to one hepatic segment, making segmental resection practical. Hepatic lobectomy may be necessary for large or multiple metastases affecting one lobe of the liver. Intraoperative ultrasonography is an essential prerequisite for resection to exclude metastases and to establish the hepatic anatomy. However, resection should probably be limited to patients with isolated secondaries until better systemic therapy becomes available, when surgical debulking of tumour may become a more worthwhile proposition.

Patients who have what appears to be a single small metastasis should be monitored carefully with CT or ultrasound examinations at approximately 3- or 4-monthly intervals to assess the growth rate of the metastasis and the development of other metastases. Most often patients who, at first, appear to have a solitary secondary tumour rapidly develop radiologically obvious multiple metastases and are therefore not suitable for hepatic resection. It is clearly important to allow time for occult metastases to become overt and so avoid subjecting patients to unnecessary operations. Fortunately there is no convincing evidence that hepatic metastases will generate further metastases within the liver, or elsewhere, which makes a reasonable delay both safe and practical.

It is a hallowed tradition in many units to review patients in the outpatient clinic at regular, if infrequent, intervals after colonic cancer resection. While these visits may provide a pastoral service to patients and an educational experience for the surgeon there is little evidence that they are therapeutically useful if the review is confined to a clinical examination and a proctosigmoidoscopy. The theoretical rationale for such clinical review, that recurrence or metastases will be detected early and that the results of subsequent treatment will be better, has not been borne out because most patients who develop recurrence present with symptoms in the intervals between clinic visits and the results of treatment have in the past not produced any survival advantage. The main reason for the lack of yield from routine clinical review is that the sensitivity for detecting small potentially treatable disease is too low. However, some recent studies that suggest improved survival and better palliation from treating patients with cancer recurrence before they become symptomatic give some hope that review may be useful if it is more directed.

Clinical review should be directed to detecting minimal disease recurrence in those patients at greatest risk of developing metastases (T3,4 N1–3; Dukes' B and C), and to diagnosing metachronous colonic polyps and cancers, which occur in about 10 per cent of patients (more if the first cancer develops at a young age). The latter is quite simply achieved by colonoscopy at 5-yearly intervals, up to the age of 75 years, once the colon has been shown to be clear of polyps or cancers. The detection of minimal disease requires more than clinical examination. Serum carcinoembryonic antigen is a sensitive indicator of recurrence or residual cancer in many, but not all patients. A postoperative rise in serum carcinoembryonic antigen has been used as a reason for a ‘second look’ laparotomy with the hope of removing small recurrences, but unfortunately with disappointing results. Hepatic ultrasound and CT scanning will also detect small metastases, particularly in the liver, while rectal endosonography will diagnose local extrarectal recurrence after anterior resection. Large controlled studies are, however, required to overcome the problems of lead time bias and length bias and to prove that the additional costs of such routine investigations can be reliably translated into useful benefit for patients. It is therefore reasonable to suggest that such detailed and structured review is confined to centres undertaking studies of treating minimal recurrent disease while others may reasonably restrict review to patients who become symptomatic.

Colonoscopy should be performed within a year of operation if the colon has not been visualized clearly preoperatively. If polyps are identified and removed, colonoscopy is repeated annually until the colon is clear of adenomas, before extending to 5-yearly. The improbability of developing a further colon cancer after the age of 75 years in a colon free of adenomas makes this a useful age to discontinue colonoscopy.

Screening has been introduced enthusiastically throughout the developed world for breast and cervical cancer in women, using mammography for the former and cervical cytology for the latter. By contrast routine chest radiographs have long since been abandoned as a means of screening for lung cancer which highlights the central issues for screening programmes.

For screening to be useful the disease must be fairly common, detectable fairly easily and economically by simple means, early treatment must carry a significant survival advantage, and the cost of each life saved must be a reasonable burden on the community.

A priori, colorectal cancer would appear to be an ideal disease for which to screen. It is the second most common cancer in the Western world; treatment is relatively straightforward and cheap in its early stages, and survival after treatment of early disease is excellent. Furthermore the colonic mucosa is accessible throughout its length both to inspection and sampling. Colonoscopy is clearly the best method of diagnosing and even treating the cancer precursors, adenomas, and small early cancers, but it is too invasive and expensive to be an acceptable means of screening the whole population at risk. Colonoscopy may reasonably be reserved to screen those people with a particularly high risk of developing colonic cancer (Table 9) 343 when the expense and unpleasantness become worthwhile.

Screening colonoscopy should be repeated every 4 to 5 years provided the patient remains free of adenomas or, in the case of colitis, dysplasia, when endoscopy should be repeated at least annually. Patients with colitis and colonic dysplasia are at such a high risk of developing invasive cancer or of having an occult cancer elsewhere in the colon that some clinicians advocate repeat colonoscopy at 6-month intervals, while others advise colectomy.

Populations with a lower risk of developing cancer require a simple, non-invasive, cheap and acceptable method of screening. Over the age of 50 years the incidence of colon cancer begins to rise quite sharply. Fortunately most polyps and early cancers shed blood into the lumen, which allows the blood itself, or degradation products such as haematin, to be detected in the stool. Testing the stool therefore provides a theoretically ideal means of screening for colonic cancers and polyps. However healthy individuals shed about 1 ml of blood into the gastrointestinal tract daily; patients with colon cancer may shed from 1 to 75 ml per day, and there are also numerous other sources of pathological bleeding into the gut.

Tests that are very sensitive lose specificity because they detect blood shed normally into the gut and blood ingested with food. By contrast, less sensitive tests may be highly specific but miss too many lesions to be useful for screening. The ideal test, which has yet to be produced, must be a compromise between sensitivity and specificity to be the highest predictor. The most widely used test at present which fulfils these criteria, Hemoccult® , detects haematin and may therefore miss very low lesions because insufficient blood has been degraded, or caecal cancers because the blood has been completely degraded beyond haematin. The sensitivity and positive predictive value of the test is increased if stools are sampled on three separate occasions. While Hemoccult® provides the best positive predictor currently available, other chemical and immunological tests are being developed which may supersede Hemoccult® in the near future. People who are positive for faecal occult blood are then selected for further investigation by flexible sigmoidoscopy or colonoscopy.

Screening not only presents profound logistical problems but also statistical problems in proving the efficacy of early treatment because of potential bias. Patients whose stools test positive may be from a biologically distinct population with an inherently favourable prognosis while people who refuse screening may bear unfavourable tumours. Interval cancers, which present and are diagnosed between screening, appear to be more aggressive and rapidly growing with a poorer prognosis. Selection bias is therefore important. Screening occurs at intervals and is therefore more likely to detect slower growing tumours producing a length bias on the outcome of screening. If early treatment does not cure the disease or affect the ultimate outcome—something that appears unlikely for colorectal cancer—then detection earlier in the natural lifespan will only give an apparent increase in survival time, producing the so-called lead time bias. Death occurs at the same time: it is merely the time of diagnosis which has moved. The expectation and hope is that time to death also moves significantly but this has yet to be proved for screen-detected colorectal cancer.

The Nottingham screening project is the largest and most comprehensive study of its kind addressing these difficult issues in relation to colonic and rectal cancer. The early results indicate that more and earlier cancers are being detected in the screened population. Allowing for non-responders and the interval cancers, cautious anticipation suggests that a useful survival advantage will be provided at an acceptable cost.

Sarcomas and lymphomas of the colon are rare in comparison to adenocarcinomas. Leiomyosarcomas are, however, more common in the colon than the small bowel, while the reverse is true for lymphomas, presumably because of the greater concentration of lymphoid tissue in the small intestine. Sarcomas usually present with a mass, which increases in size slowly if the lesion is low grade but quite rapidly if it is high grade. Surgical removal may be challenging and local recurrence is common (Figs. 21, 22) 1086,1087. The prognosis of high-grade tumours is particularly poor and not materially improved by adjuvant treatment.

Benign tumours can arise from each component of the colon but, with the exception of mucosal adenomas, they are rare. Adenomas share the same uneven distribution as cancers around the colon. Not only are they common, they are also important because of their association with, and their propensity to become, colon cancers. The aetiology and epidemiology of adenomas and their relationship to cancers has already been described.

Adenomas are usually polypoid and vary considerably in size, shape, and histological differentiation. All adenomas start life as a neoplastic change in a single crypt (microadenomas) which gradually increases in size. Well-differentiated tubular adenomas are the most common and are more often pedunculated, particularly when larger. The stalk may be 3 or 4 cm long and the shape of the polyp may sometimes be quite bizarre (Fig. 31) 1096. By contrast, the less differentiated villous adenomas are less common and invariably sessile, often extending as a carpet over many centimetres of colon. Tubulovillous adenomas lie between the two both in frequency and morphology. They are usually polypoid but sessile, with a fairly broad base or a short stalk. Why the different types of adenoma have such different morphology is not clear but it is possible that the tubular adenomas are sufficiently slow growing for the peristaltic action of the colon, pulling the polyp down the lumen, gradually to produce a mucosal stalk by traction, whereas the faster growing villous adenoma expands laterally faster than a stalk can develop and produces less luminal mass to induce intussusception. The incidence of malignant change in an adenoma increases with the size of the tumour and the histological dysplasia, such that foci of cancer are common in large villous adenomas (Table 10) 344.

Most adenomas are silent and are diagnosed during the investigation of other bowel symptoms. They may be detected by faecal occult blood screening, but when larger the blood in the stool may be clinically obvious. Bleeding is often episodic, small in volume, dark, and mixed in the faeces. However bleeding can occasionally be brisk, bright, and copious, particularly when a large polyp partially sloughs from its pedicle. Villous adenomas may also present with intermittent bleeding, but more often presents with the passage of mucous, which may be of sufficient volume to produce mucous diarrhoea. Large rectal lesions, replacing as much as 20 cm of mucosa, can cause enough mucus diarrhoea, rich in potassium, to induce dehydration, collapse, and hypokalaemic confusion. Urgency and urge incontinence are also common symptoms of these large villous adenomas.

There is much useless debate about the best method of diagnosing polyps. A good double contrast barium enema can reliably detect polyps 0.5 cm or more in diameter. Colonoscopy can detect smaller lesions and has the added advantage of allowing simultaneous treatment, but it is also only about 95 per cent accurate. Use of the two investigations together increases the overall accuracy to nearly 100 per cent. From an economic point of view however, colonoscopy is the method of first choice if a polyp is suspected. Treatment for very small adenomas is biopsy cautery. Bigger polyps are snared and those larger than 5 cm are snared piecemeal, using electrocautery to coagulate and destroy the base. Complications of colonoscopic polypectomy are uncommon with proper care, but the most serious are perforation and haemorrhage. The frequency of perforation in skilled hands is about 1 per 1000 polypectomies and the respective incidence of bleeding sufficient to require transfusion is about 1 per 300. Vapourization using a Nd-YAG laser has been used to destroy adenomas, but this method carries a greater risk of perforation and does not produce a specimen for histological examination. Photodynamic therapy has also been used effectively for small adenomas but has little apparent advantage over the simpler and less expensive snare polypectomy.

Patients require endoscopic monitoring after polypectomy, and most authorities advise repeat colonoscopy at annual intervals until the colon is clear of polyps, when the interval may be increased to 4 or 5 years. Faecal occult blood screening may be advised for the intervening years. The incidence of polyps at subsequent endoscopy is 25 per cent when one polyp was found initially and 50 per cent when two or more polyps were snared.

Large sessile rectal villous adenomas are not suitable for colonoscopic removal and sometimes present the surgeon with a formidable challenge particularly in the very elderly patients most likely to have such a tumour.

Small low lesions may be readily removed through the dilated anus by excision of a disc of rectal wall, or resection of mucosa alone if one can be sure there is no malignant change. The dissection of the mucosa is made much easier by infiltrating 1 in 200 000 noradrenaline solution into the submucosa. Small mucosal defects do not require closure and rapidly re-epithelialize, but defects resulting from a full thickness excision need to be sutured. The main advantage of this technique is that a good specimen is provided for easy histological examination. The Buess endoscopic technique of operating through a large proctoscope with the rectum insufflated with CO&sub2; is an extension of this method which provides better access to the proximal rectum but requires the skills of stereotactic operating. The instrumentation is expensive and the skills are not yet widely available. Both anterior resection and trans-sacral resection are major operations with major risks of morbidity, which may be justified for younger patients with large adenomas but not for the frail and debilitated. Parks devised the method of removing large villous adenomas per anum by infiltrating the submucosa with noradrenaline solution and excising a tube of rectal mucosa. The resulting large defect was then closed by plicating the rectum like a concertina. The technique is not easy and needs considerable dilatation of the anus, which is best avoided in elderly patients with already weakened sphincters. Deep electrocoagulation effectively destroys adenomas but produces no specimen and also requires considerable anal dilatation. Laser therapy avoids anal dilatation but again provides no specimen; however it is most useful for treating more proximal lesions, such as caecal villous adenomas, in patients unfit for resection. Endoscopic transanal resection, with continuous glycine irrigation, is in many ways ideally suited for the removal of small and large villous adenomas. The anus is not dilated, operation is quick with a low morbidity, can be repeated, and provides a reasonable, if fragmented, specimen for examination. However the operation does require the skills and experience of urological endoscopic surgery.

Patients with familial adenomatous polyposis syndrome require total colectomy and ileorectal anastomosis or a restorative proctocolectomy at some time in the later teens or early twenties because of the inevitable development of invasive cancer in one or more of the adenomatous polyps with the passage of time. The timing of surgery may reasonably be fitted around important social and educational events provided the colon is kept under annual endoscopic surveillance. The functional results of colectomy and ileorectal anastomosis are better than those of restorative proctocolectomy but the distal rectum of patients so treated must be inspected regularly to ensure that the residual polyps do not grow and become malignant. For reasons that are not clear the rectal polyps usually regress in size and number after colectomy and ileorectal anastomosis.

Two types of hamartoma are found in the colon: juvenile polyps (Fig. 32) 1097 and Peutz-Jegher's polyps which may be single or multiple (Fig. 33) 1098. There is a strong family history, particularly in patients with multiple hamartomas. Presentation is similar to that of adenomas but, like the familial polyposis coli syndromes, at a younger age than sporadic adenomas. The importance lies in the observation that the hamartomas are frequently associated with adenomatous change and, therefore, a risk of malignant transformation. Treatment is usually by colonoscopic polypectomy, unless the polyps are very numerous when colectomy and ileorectal or pouch-anal anastomosis is preferable.

Lipomas are not uncommon and are usually a chance finding at colonoscopy, although some may acquire a stalk and intussuscept, causing obstruction and pain. The appearances are obviously different from an adenoma. The polyp is smooth, usually dome shaped, and covered with normal looking mucosa. When large and intussuscepting the leading point may infarct (Fig. 34) 1099. Most lipomas need no treatment, but when large they are best removed. Colonoscopic snare removal is often possible but there is a real risk of perforation if the base is broad. Removal may be incomplete if the polyp extends into the colonic wall. In these cases the snare cuts slowly because of the insulating properties of fat.

Leiomyomas of the colon are uncommon and are usually asymptomatic chance findings. Recognition is, however, important. Like lipomas the polyp is dome-shaped and covered by normal mucosa. Unlike gastric leiomyomas there is little risk of haemorrhage. Haemangiomas also occur in the colon and present with intermittent and sometimes severe colonic haemorrhage. They are benign and are readily treated by segmental resection after diagnosis by colonoscopy or angiography.

Gut endocrine tumours occur in the colon but, though uncommon, are often multiple. Many are benign but larger lesions are usually malignant and present and metastasize in much the same way as an adenocarcinoma. The rectum is the more common site for malignant ‘carcinoids’. Treatment is similar to that of rectal cancer, although more bowel should be removed to encompass the multifocal nature of the disease. Colonic ‘carcinoids’ do not secrete 5-hydroxytryptamine and therefore do not produce the syndrome associated with ileal carcinoids. Immunological staining identifies the microscopic tumours and shows that colonic ‘carcinoids’ contain a wide variety of neuroendocrine substances rather than a single gut hormone. The prognosis for malignant carcinoids' is poor; less than 30 per cent of patients survive for more than 5 years.

Metaplastic polyps, which are synonymous with hyperplastic polyps, are the most common small polyps seen in the rectum and distal sigmoid colon. They are often multiple, usually about 5 mm or less in diameter, and are slightly raised above the surrounding normal mucosa. They are pale and uniform, which readily identifies them macroscopically from adenomas. The histological appearances are also distinctive (Fig. 35) 1100. Their importance lies in distinguishing them from adenomas for the former are quite innocent, with no malignant potential, and cause no symptoms. While their aetiology is unknown there is a statistical association between metaplastic polyps and adenomas, probably because both are more common in older patients. Some clinicians therefore suggest that all patients with metaplastic polyps should at least undergo a flexible sigmoidoscopy to exclude adenomas. If no adenomas are found then the patient needs no further endoscopic surveillance.

General reading

Deans GT, Parks TG, Rowlands BJ, Spence RAJ. Prognostic factors in colorectal cancer. Br J Surg 1992; 79: 608–13.

Goligher JC. Surgery of the Colon, Rectum and Anus. 5th Edn. London: Balliere Tindall, 1983.

Mortensen N, ed. Colorectal Cancer. Clinical Gastroenterology vol 3. No 3. Balliere Tindall, London, 1989.

Progress Symposium. Carcinoma of the large bowel. World J Surg 1991; 15: 561–622.

World Progress in Surgery—Management of Adenocarcinoma of the Low Rectum. World J Surg 1992; 16: 428–515.

Aetiology and epidemiology

Clarke PJ, Dehn TCB, Kettlewell MGW. Changing patterns of colorectal cancer in a regional teaching hospital. Ann R Coll Surg Eng 1992; 74: 291–3.

Kingsnorth AN, Lumsden AB, Wallace HM. Br J Surg 1984; 71: 791–4.

Stephenson BM, Finan PJ, Gascoyne J, Garbett F, Murday VA, Bishop DT. Frequency of familial colorectal cancer. Br J Surg 1991; 78: 1162–6.

Biology and pathology

Appleton GVN, Davies PW, Williamson RCN. Effect of defunction on cytokinetics and cancer at colonic suture lines. Br J Surg 1990; 77: 768–71.

Clausen MR, Bonnen H, Mortensen PB. Colonic fermentation of dietary fibre to short chain fatty acids in patients with adenomatous polyps and colonic cancer. Gut 1991; 32: 923–8.

Hermanek, P. Colorectal carcinoma: histopathological diagnosis and staging. In: Mortensen N, ed. Clinical Gastroenterology Vol. 3, No 3. Balliere Tindall, London. 1989.

Lofberg R, Brostrom O, Karlen P, Ost A, Tribukair B. Carcinoma and DNA aneuploidy in Crohn's colitis—a histological and flow cytometric study. Gut 1991; 32: 900–4.

Rew DA, Wilson GD, Taylor I, Weaver PC. Proliferation characteristics of human colorectal carcinomas measured in vivo. Br J Surg 1991; 78: 60–6.

Screening

Dunlop MG. Screening for large bowel neoplasms in individuals with a family history of colorectal cancer. Br J Surg 1992; 79: 488–94.

Hardcastle JD, Armitage NC, Chamberlain J, Amar SS, James PD, Balfour TW. Fecal occult blood screening for colorectal cancer in the general population. Cancer 1986; 58: 183–4.

Ultrasound

Charnley RM, Morris DL, Dennison AR, Amar SR, Hardcastle JD. Detection of colorectal liver metastases using intraoperative ultrasound. Br J Surg 1991; 78: 45–8.

Roubien LD, et al. Endoscopic ultrasonography in staging rectal cancer. Am J Gastroenterol 1990; 85: 1391–4.

Surgery and survival

Chapuis PH, Dent OF, Fisher R, et al. A multivariate analysis of clinical and pathological variaties on prognosis after resection of large bowel cancer. Br J Surg 1985; 72: 698–702.

Phillips RKS, Hittinger R, Blesovsky L, Fry JS, Fielding LP. Large bowel cancer: surgical pathology and its relationship to survival. Br J Surg 1984; 71: 604–10.

Surtees P, Richie JK, Phillips RKS. High versus low ligation of the inferior mesenteric artery in rectal cancer. Br J Surg 1990; 77: 618–21.

Liver resection

Doci R, Gennari L, Bignami P, Montalto F, Morabito A, Bozzetti F. One hundred patients with hepatic metastases from colorectal cancer treated by resection: analysis of prognostic determinants. Br J Surg 1991; 78: 797–801.

Scheele J, Stangl R, Altendorf-Hofmann A. Hepatic metastases from colorectal carcinoma: impact of surgical resection on the natural history. Br J Surg 1990; 77: 1241–6.

Chemotherapy

Goldberg JA, Kerr DJ, Wilmott N, McKillop JH, McArdle CS. Regional chemotherapy for colorectal liver metastases: a phase 2 evaluation of targeted hepatic arterial 5-fluorouracil for colorectal liver metastases. Br J Surg 1990; 77: 1238–40.

Hunt TM, Flowerdew ADS, Birch SJ, Williams JD, Mullee MA, Taylor I. Prospective randomized controlled trial of hepatic arterial embolization or infusion chemotherapy with 5-fluorouracil and degradable starch microspheres for colorectal liver metastases. Br J Surg 1990; 77: 774–82.

Scheithauer W, Rosen H, Kornek G-V, Sebasta C, Depisch D. Randomised comparison of combination chemotherapy plus supportive care with supportive care alone in patients with metastatic colorectal cancer. Br Med J 1993; 306: 752–5.

Obstruction and perforation

Runkel NS, Schlag P, Schwarz V, Herfarth C. Outcome after emergency surgery for cancer of the large intestine. Br J Surg 1991; 78: 183–8.

Stomas

Karanjia ND, Corder AP, Holdsworth PJ, Heald RJ. Risk of peritonitis and fatal septicaemia and need to defunction low anastomosis. Br J Surg 1991; 78: 196–8.

Polyps

Madden MV, et al. Comparison of morbidity and function after colectomy with ileorectal anastomosis or restorative proctocolectomy for familial adenomatous polyposis. Br J Surg 1991; 78: 789–92.

Minopoulos GJ, McIntyre RLE, Lee ECG, Kettlewell MGW. Colonoscopic polypectomy in a regional teaching hospital. Br J Surg 1983; 70: 51–3.

Shinya H, Wolff WI. Morphology, anatomic distribution and cancer potential of colonic polyps. Ann Surg 1979; 190: 679–83.

Palliation

Berry AR, Souter RG, Campbell WB, Mortensen NJMcC, Kettlewell MGW. Endoscopic transanal resection of rectal tumours—a preliminary report of its use. Br J Surg 1990; 77: 134–7.