• 1.a. The association of diet and cancer
  
• 1.b. Colorectal cancer
  
• 1.c. The aetiology of colorectal cancer and adenomatous polyps
  
• 1.d. Colorectal cancer and screening
  

Cancers are amongst the commonest causes of death in this country, accounting for about one in every four deaths - almost 130,000 per annum¹.  The majority of cancer deaths are from tumours found in four principal sites: lung, colorectal, breast and prostate (Figure 1).

Figure 1. Cancer deaths in England and Wales by sex, 1996.

These cancers are common in Western countries but there is a much lower incidence in third world countries (Figure 2).  It has been observed that immigrants moving from a low risk area to one of high risk acquire the same risk as the indigenous population within one or two generations suggesting that environmental factors are responsible².

Figure 2. Geographic variation in the incidence of colon and rectal cancer in men ².

Tobacco smoking, electromagnetic radiation, environmental chemicals, hormones, bacterial or viral infection, level of physical activity, reproductive and sexual behaviour are thought to be important in the aetiology of cancer at certain sites³.  However, it is thought that diet remains the most important factor and it has been estimated that dietary change could result in a reduction of fatal cancers of between 35 and 70%⁴.

1.b.i. Demographics

Colorectal cancer is the second most common cause of death from cancer in the Western world.  In 1985, there was estimated to be 677,500 new cases and 394,100 deaths worldwide^5,6.  In the United Kingdom there is approximately a 4% lifetime risk, 31,000 new cases annually and 16,000 deaths⁷.

The prevalence increases with age with over 90% of cases occurring after the age of 55 (Table 1).  The  incidence of colon cancer varies little between the sexes^8,9, whereas rectal cancer is twice as common in men as in women².

1.b.ii. The pre-malignant lesion: the adenomatous polyp

There is abundant evidence that virtually all colorectal carcinomas begin as adenomatous polyps. 

In a number of studies, colonic polyps have been left in situ and the follow up of these patients showed variable natural history from complete regression, increase in size or progression to carcinoma^11-13.

Histological studies have revealed a spectrum of dysplasia within adenomatous polyps up to carcinoma-in-situ and it is not uncommon for true invasive carcinomas to have associated adenomatous tissue¹⁴.  In a study of post mortem examinations of the colon, the population with the highest proportion of adenomas was observed in the area with the highest incidence of colonic cancer. Also, the segmental distribution of adenomas within the colon was found to be similar to the site distribution of cancer¹⁵.

The transition from benign adenoma to colorectal cancer is thought to have a long natural history of between 10 and 35 years ^13,16.  It is estimated that the annual conversion rate of a polyp to a cancer is approximately 0.25%¹⁷.

1.b.iii. The adenoma / carcinoma sequence

Colorectal carcinogenesis has long been thought to be due to a stepwise accumulation of cellular mutations¹⁸.  Studies of adenoma and carcinoma cell types have revealed that there is a monoclonal expansion of a single or small number of colonic epithelial cells^19,20.  Much research has been done to identify mutations in oncogenes, tumour suppressor genes or DNA repair genes that confer a growth advantage in these neoplastic cells²¹.

A genetic model for the sequence of adenoma transformation to carcinoma was proposed by Vogelstein²² which in light of recent advances in molecular genetics has become modified to include additional genes that contribute (Figure 3).

Figure 3. Genetic model of colorectal carcinogenesis

It is thought to be the accumulation rather than the order of genetic mutations that is critical to the development of colorectal cancer²³.

Cellular hyperproliferation is a preceding step to adenoma formation and is associated with loss or mutation of both alleles of the adenomatous polyposis coli (APC) gene on chromosome 5q²⁴.  Inherited mutation in this gene gives rise to familial adenomatous polyposis (FAP) which is typified by hundreds and occasionally thousands of adenomatous polyps²⁵.  The APC gene is a tumour suppressor gene which prevents uncontrolled epithelial cell proliferation.  Mutation results in a truncated APC protein that fails to bind to ß-catenin, which in turn fails to promote cell adhesion via the calcium dependent cell surface adhesion molecule E cadherin ²⁶.

The over expression of the cyclo-oxygenases (COX 1 and COX2) catalyses the conversion of arachidonic acid to prostaglandin H2 and other derivatives, e.g. malondialdehyde, which is itself mutagenic, promoting further polyp proliferation ²⁷.  This explains why the non-steroidal anti-inflammatory drugs, which inhibit the cyclo-oxygenases, have been postulated as possessing a potential therapeutic role in the prophylaxis of colorectal cancer ²⁸.

A significant loss of DNA methyl groups has been shown to occur early in colorectal tumourogenesis²⁹.  Hypomethylation interferes with chromosome precipitation and mitotic separation³⁰ resulting in increased frequency of  genetic alterations.

Approximately 50% of colorectal cancers^31,32 and adenomas larger than 1cm²² have been found to exhibit mutation in the K-ras oncogene.  This mutation is present from the intermediate stage (>1cm, mild & moderate dysplasia) of adenoma formation and therefore probably represents an initiating stage for adenoma progression²¹.

Allelic and chromosomal losses occur frequently in colorectal cancer^33,34.  This can result in loss of regions that code for tumour suppressor genes.  Allelic loss of chromosome 18q occurs in more than 70% of carcinomas²² and 50% of late adenomas.  In this area, the tumour suppression gene DCC (deleted in colorectal cancer) is located.  This gene encodes a cell surface-localized protein³⁵ and is thought to be involved in cell adhesion³⁶ and its preservation is associated with the majority of mucinous colorectal tumours³⁷ suggesting a function in cellular differentiation or phenotype modulation.

Similarly, the p53 gene located on chromosome 17p³⁸ frequently exhibits allelic loss (also termed loss of heterozygosity).  This occurs in approximately 75% of cases of colorectal cancers^22,39 but infrequently at the adenoma stage⁴⁰.  In addition, somatic mutations in the remaining p53 allele have been regularly observed in keeping with the proposed function of p53 as a tumour suppressor gene^38,41.  Thus, mutation in one allele coupled with loss of the other appears to be tumourogenic and an important step in the transformation of adenoma to carcinoma.

As can be deduced from the adenoma - carcinoma sequence postulated by Vogelstein (1.b.iii. The adenoma / carcinoma sequence), besides factors critical only to adenoma progression, the aetiology of colorectal adenomatous polyps should be the same as that for colorectal cancer - the adenoma just being an intermediate step in the development of cancer.

Indeed, there are well recognised inherited syndromes characterised by multiple colorectal adenomas that invariably give rise to colorectal cancer.  These cases are uncommon and only constitute about 5% of reported cancers.

There is increasing evidence however that many adenoma cases have some previously unrecognised genetic predisposition⁴² with absolute adenoma risk being modified by environmental factors.

1.c.i. Genetic Syndromes

Familial Adenomatous Polyposis

Familial adenomatous polyposis (FAP) is a rare autosomal dominant syndrome characterised by hundreds or even thousands of adenomatous colorectal polyps affecting approximately 1 in 10,000 individuals.  Untreated these polyps almost invariably progress to carcinoma by a mean age of 44⁴³.  Mutations are found in the APC (adenomatous polyposis  coli) gene which is located on chromosome 5q and penetrance is thought to approach 100%^44,45(Figure 4).  Mutation at the APC locus appears to be a common event in colorectal cancer induction, the only difference there being that in FAP-associated lesions there is an inherited genetic defect.

The designation Gardner Syndrome is used for phenotypic variants of FAP with additional extra colonic manifestations e.g. osteomas, epidermoid cysts and fibromas⁴⁶ and Turcot's syndrome for the association with medulloblastoma⁴⁷.

Figure 4. The APC gene on chromosome 5⁴⁸

Phenotypical variation in FAP is now known to depend on the location of the APC mutation.  Attenuated FAP, in which there are relatively few polyps, is associated with mutations in codons at the 5' end of the APC gene⁴⁹ whereas mutation between codon 1285 and 1465 result in profuse polyposis syndrome⁵⁰.  The best example of this variation giving rise to extra colonic features is the presence of Congenital Hypertrophy of the Retinal Pigment Epithelium (CHRPE) which is associated with mutations in codons 542 to 1309⁵¹.

Hereditary Non-Polyposis Colorectal Cancer (HNPCC)

HNPCC is an autosomal dominant syndrome that is estimated to be responsible for approximately 2% of colorectal cancers⁵².  The syndrome is characterised by familial clustering of colorectal cancers along with other cancers such as endometrium, ovary, gastric, hepatobiliary and renal tract.  In the absence of a clear phenotype an absolute definition of HNPCC is therefore obviously difficult - the Amsterdam Criteria⁵³ and Bethesda Criteria⁵⁴ attempt to classify on clinical grounds those families most at risk of having mutation in the mismatch repair genes that give rise to HNPCC (Table 2)

Non-coding regions of DNA exhibit variation which principally consists of highly repetitive segments of DNA consisting of several iterations of a specific sequence known at "DNA repeats".  Such sequences are unique to each person and are the basis for the precise DNA fingerprinting used in forensics.  These repeats of 2-5 nucleotide segments are known as microsatellite DNA.  A single pair of PCR oligonucleotide primers that surround such sequences produce variably-sized DNA fragments depending upon the number of repeats.

The mismatch repair pathway is responsible for detecting and repairing short segments of mismatched base pairs (such as a mutation from C to T on one strand T opposite G, or the addition of extra nucleotides, resulting in unpaired bases within the helix)⁵⁵. Since microsatellite repeats are susceptible to such mutation, disorders of the mismatch repair pathway lead to errors in these polymorphic segments - termed microsatellite instability (MSI).

In HNPCC, microsatellite instability was discovered to be the result of germline mutations in the genes that encode the components of the DNA proofreading complex. These genes are hMSH2, hMLH1, hPMS1, and hPMS2.  hMSH2 and hMLH1 are the most commonly mutated in HNPCC⁵⁶, whereas mutation in the genes hPMS1 and hPMS2 contribute only to a small proportion of cases.  The resultant colorectal cell lines show a higher accumulation of other mutations and deletions⁵⁷ presumably reflecting the in vivo accumulation of errors resulting in an increased risk of cancer.

1.c.ii. Genetic Predisposition

Excluding the specific genetic syndromes, the risks of colorectal cancer in first degree relatives of index patients is about twice that of the general population^58,59.  The genetic basis of this familial aggregation has not yet been characterised.  The prevalence of FAP and mismatch repair gene mutation is not sufficient to explain this family aggregation.

Metabolic Enzyme Polymorphism

The cellular response to environmental procarcinogens and carcinogens is a predominant factor in certain types of cancer. Cells are able to recruit a number of metabolic detoxifying mechanisms to ward off the threat of mutation. Inter-individual variation in the genes involved must therefore represent potential genetic susceptibility factors for disease⁶⁰.  Defective metabolism can lead to build up of carcinogens and result in increased damage to cells, with point mutations in specific genes, rearrangements and translocations, gene amplifications or deletions, and gross chromosomal aberrations.

Metabolism of ingested substances in man is thought to occur in two distinct phases. On entering the body, those substances are first subjected to phase I metabolism, where functionalisation reactions are performed, creating a 'reactive-centre' (for example, -OH, -NH2, -COOH) in the molecule.  The most ubiquitous of phase I catalysts are the cytochrome P450 mono-oxygenases, which are capable of metabolising a wide range of structurally diverse substrates.  Other enzymes involved in phases I metabolism are described in Table 3.  Following the creation of an electrophilic reactive centre, phase  II enzymes such as the glutathione S-transferases and UDP-glucuronyl transferases (Table 3) are responsible for conjugation reactions, involving the incorporation of, for example, a glutathione or glucuronic acid moiety into the molecule. Phase II reactions are, in most cases, responsible for the ultimate excretion of drug from the body.  The expression of many of the enzymes involved in drug metabolism in man is genetically polymorphic.

Model for carcinogenesis

The combination of genetic polymorphisms in these detoxifying enzymes can result in several 'high risk' phenotypes (Figure 5).

Figure 5. Model for carcinogenesis.

In the one step model above it can be seen that a defective enzyme would clear ingested carcinogens slowly and thus increase the risk of mutation.  Similarly, a functionally fast enzyme metabolising pro-carcinogens in to an intermediate carcinogen would result in carcinogen build up if the detoxifying step is slow.  It is thought that most carcinogens are ingested as pro-carcinogens so the latter model predominates.  For example, it has been demonstrated that the potent carcinogen NNK (4-(methylnitrosamino)-1 -(3-pyridyl)- 1 -butanone), a component of cured tobacco and tobacco smoke, is metabolically activated by the enzyme CYP2D6⁶¹.

Polymorphic Metabolic Enzymes

N-acetyltransferases

The N-acetyl transferases are the main focus of this thesis and are described in detail in chapter 1.c.iv. N-acetyltransferases.

Glutathione S-Transferases

The glutathione S-transferases (Table 3), GSTM1 (chromosome 1p13.3) and GSTT1 (chromosome 22q11.2), exhibit polymorphism in the genes coding these metabolic enzymes⁶² that are involved in phase II metabolism (Table 3) and both can be deleted producing a null genotype if present in the homozygous form.  Since these enzymes catalyse the conjugation of  a variety of compounds including a number of carcinogens there has been a great deal of interest in genotypic / phenotypic variation and the incidence of colorectal cancers and adenomas.

In 1993 Zhong et al. ⁶³ observed a significant increase in the GSTM1 null genotype in a group of 196 colorectal cancer patients when compared to 225 controls.  A subgroup analysis by Probst-Hensch et al.⁶⁴ reported a significantly raised incidence of colorectal adenomas for fast NAT2 and GSTM1 null compared to slow NAT2 GSTM1 non-null among current smokers (Relative risk = 10.3, 95% CI 1.94 - 55).  However, the majority of the studies investigating GSTM1 genotype have shown no evidence of an association^65-72.  Nevertheless, most of the odds ratios were higher than 1.0 suggesting there may be some correlation but that this is dependent on additional factors such as environmental exposure and/or gene - gene interaction.

Similarly, the majority of papers reporting GSTT1 interaction with colorectal neoplasia have reported no significant difference^67,68,70-72 except Deakin et al.⁷³ who found an odds ratio of 1.9 for GSTT1 null compared to non-null genotype in colorectal cancer cases.

P₄₅₀ Cytochrome Enzymes

The cytochromes P₄₅₀ in humans consist of at least 20 different proteins that are coded by multiple genes and act as Phase I metabolisers. The metabolic activation of food-borne heterocyclic amines to colonic carcinogens in humans is hypothesised to be enhanced by certain polymorphisms in the P₄₅₀ genes (CYP group).

CYP1A1

Sivaraman et al.⁷⁴ assessed the frequency of the MspI polymorphism in the 3'-end of CYP1A1 and another mutation in exon 7 of the gene (Ile-Val polymorphism) among 43 patients with in situ adenocarcinoma of the large bowel and 129 population controls. Homozygosity for the MspI mutant genotype was found to be positively associated with in situ colorectal cancer in Japanese (P = 0.008) and Hawaiians/part-Hawaiians (P < 0.001), whereas the study lacked power to detect a similar association in Caucasians. 

Kiss et al.⁷⁵ demonstrated the CYP1A1 Val allele was also overrepresented among colon cancer patients (OR = 1.57, 95% CI 0.90 - 2.74).

In contrast, Inoue et al.⁶⁵ in a more recent study investigating CYP1A1, colorectal adenomas and smoking found that smoking increased adenoma risk but that there was no influence from CYP1A1 MspI mutation.

CYP1A2

Lang et al.⁷⁶ considered the combined effects of fast NAT2 phenotype and fast CYP1A2 phenotype on the risk of colorectal neoplasia and found that 35% of cases had this phenotype compared to 16% of controls (p = 0.002).  Subsequent analysis using red meat as the measure of ingested substrate revealed an odds ratio of 6.5 for the fast / fast phenotype with a preference for well-cooked red meat compared to the slow / slow phenotype and a rare / medium cooked meat preference.

CYP2E1

CYP2E1 is expressed in colonic epithelial cells and also in the liver and has been implicated in the local activation of procarcinogens in the aetiology of colorectal neoplasia⁷⁷.  The c2 allele has been found to be phenotypically slow compared to the wild-type c1 allele⁷⁸.  Kiss et al.⁷⁵ investigated the CYP2E1 c2 allele and demonstrated a significant association with colorectal cancer (OR = 1.91, 95% CI 1.05 - 3.52).  Combined analysis with CYP1A1 val allele and GSTM1 null showed that individuals carrying all the three "high-risk" alleles have a strikingly increased risk for sporadic colorectal cancer (OR = 4.62, 95% CI 1.23 - 25.68) as compared to three "low-risk" alleles.

Epoxide Hydrolase

Microsomal epoxide hydrolase is an enzyme expressed in many tissues including liver and colon that contributes to the phase I metabolism of ingested mutagenic compounds (Table 3).  It is known to be polymorphic, its activity varying more than 50 fold in Caucasians⁷⁹.

Harrison et al.⁸⁰ investigated the relationship of the exon 3 and exon 4 mutations in the epoxide hydrolase gene in a case control study of colorectal cancer patients in Scotland.  There was a threefold increase in the exon 3 T to C (tyrosine residue 113 to histidine) mutation ("slow") in colon cancer cases compared to controls (OR = 3.84, 95% CI 1.83 - 8.04).  This suggests that putative slow epoxide hydrolase may be a risk factor for colon cancer.  There was however no association seen between cases and controls for the exon 4 A to G (histidine residue 139 to arginine) mutation which produces increased enzyme activity.

Cortessis et al.⁸¹ recently published a similar study but failed to show any difference in colorectal cancer incidence due to epoxide hydrolase alone.  However, subgroup analysis in current smokers (OR = 4.27, 95% CI 1.68-10.81) and regular well done red meat eaters (OR = 2.47, 95% CI 0.99 - 6.19, interaction p=0.03) gave significant results.

Methylenetetrahydrofolate Reductase

One postulated mechanism for the reduction seen in colorectal cancer incidence with high levels of vegetable intake is that folate influences the methylation of DNA.  It is thought that hypomethylation of colonic epithelial cell DNA is an early step in colorectal carcinogenesis (Figure 3).  Methylenetetrahydrofolate reductase is a polymorphic enzyme that effects folate distribution and the remethylation of homocysteine. 

Under low folate conditions, the homozygous TT (677 C to T mutation) "slow" genotype has been regarded as harmful because it is associated with a high concentration of plasma total homocysteine, in turn leading to DNA hypomethylation and an increased risk of colorectal neoplasia⁸².  But, in combination with adequate folate intake, the TT genotype is associated with decreased risk of colorectal neoplasia ⁸³.

This finding was reproduced by Ma et al.⁸⁴ who found that among men with adequate folate levels, there was a 3-fold decrease in risk (OR = 0.32, 95% CI 0.15 - 0.68) among men with the homozygous mutation compared with those with the homozygous normal or heterozygous genotypes.  However, the protection due to the mutation was absent in men with folate deficiency.

Similarly, Ulrich et al.⁸⁵ found that individuals with the TT genotype and intakes of folate, vitamin B12, vitamin B6, or methionine in the lowest tertile were at elevated risk for adenomas (about 2-3-fold when comparing with TT genotype with high intakes).

1.c.iii. Environmental Factors

The role of diet in tumourogenesis

Diet might influence the risk of cancer in a number of ways, such as the direct ingestion of pro-carcinogenic or carcinogenic compounds, induction of protective mechanisms such as apoptosis, suppression of DNA damage through antioxidants in food or modification of cell proliferation and methylation of DNA.

Dietary factors in colorectal neoplasia

The geographic variation in colorectal cancer (Figure 2) corresponds to a high per capita consumption of red meat and dietary fat and to a lesser degree is inversely associated with the amount of dietary fibre⁸⁶.  Epidemiological evidence has implicated these as principal risk factors and various other less consistent dietary factors with the risk of colorectal cancer⁸⁷.  Specific components within the diet are obviously hard to establish because of the heterogeneity of the food we eat.  The recognised risks in colorectal cancer are listed in Table 4.

Total Energy Intake

Total energy intake has been found consistently to correlate with an increase in the risk of colorectal cancer^{88,93,101,102}.  However, because energy-contributing nutrients such as dietary fat, protein and carbohydrate are highly correlated with total energy, the association with colon cancer from energy per se as opposed to other components of the energy-providing nutrients is often not clear.  Slattery et al.⁸⁸ demonstrated that once total energy and physical activity had been corrected for, there was no significant association seen for dietary fat, protein and carbohydrate.  This contradicts many hypotheses and studies that have shown that dietary fat correlates with the risk of colorectal cancer (below) and it may be that these factors are inextricable.

Dietary Fat

Dietary fat has been associated with cancers of the breast, colon, rectum, endometrium, ovary, prostate and gall bladder⁴.  Obviously, dietary fat correlates with total energy intake which has been shown to be a risk factor.

The risk of colorectal cancer correlates with the consumption of animal fat but not vegetable fat⁸⁶(Table 5). There is an inverse correlation with fish and fish oil consumption, when expressed as a proportion of total or animal fat, and this correlation was significant for both male and female colorectal cancer and for female breast cancer¹⁰³.

Prentice et al.¹⁰⁴ summarised the evidence from the literature in 1990 and presented 8 studies (Table 6).

Table 6. Relative risks for below- to above-median fat consumption

⁺males and females, colon and rectum

*males and females colon                          **males and females rectum

Most of the studies demonstrated a lower risk with a below-median fat consumption, although 2 studies^109,110 found relative risks above unity these were not statistically significant. 

Red Meat

Willet et al.⁸⁶ demonstrated in a cohort study of 88,751 women that the relative risk of colon cancer in women who ate beef, pork, or lamb as a main dish every day was 2.49 (95 percent confidence interval, 1.24 to 5.03), as compared with those reporting consumption less than once a month.  Similarly, Hsing et al.¹¹¹ in a cohort study of 17,633 men with 20 years follow up found an increased risk of colon cancer for those who consumed red meat more than twice a day (RR = 1.8, 95% CI 0.8 - 4.4).  There are now numerous studies that have demonstrated a similar effect in the aetiology of colorectal adenomas^112-116.

Exposure of meats to high temperatures can result in the formation of heterocyclic amines and aromatic hydrocarbons that are carcinogenic in animals^117-119.  Some studies have observed relatively strong associations of colorectal neoplasia with consumption of broiled or grilled meats and browning of the meat surface^120,121 and in a more recent study, Sinha et al. ¹¹³concluded that it was the high temperature cooking methods that contributed more to the increased risk than the absolute red meat intake itself.

Protein

The association of protein with colorectal cancer has been little investigated because the emphasis has been on red meat as the protein source most often associated.  Potter et al.⁹³ demonstrated that the most consistent risk factor for colorectal cancer was dietary protein, which was associated with two to three times a relative risk of colon and rectal cancer in women for all levels of consumption above the lowest quintile. For male colon cancer the corresponding relative risk was similar; but for male rectal cancer, risk was elevated only at old ages.  The actual protein source was not analysed and so how much the observation is due to meat proteins is uncertain.

Alcohol

Although ethanol has generally not been found to induce cancer in experimental animals, there is evidence that alcohol consumption increases the risk of cancer in humans¹²².  The exact mechanism for this increase risk is not clear but studies have shown a dose response relationship between the alcohol intake and relative risk of cancer (Table 7) ¹²³.

It is postulated that alcohol may:

• contain carcinogenic contaminants
• generate carcinogenic metabolites such as acetaldehyde
• act as a solvent to increase penetration of target tissues by carcinogens
• reduce nutrients necessary for health
• inhibit hepatic detoxification of carcinogens
• catalyse activation of pro-carcinogens
• affect levels of hormones e.g. oestrogens
• increase cellular exposure to oxidants
• suppress immune function

Alcohol and colorectal cancer

Methylation of DNA is thought to have a role in the regulation of gene expression. A high consumption of alcohol, an antagonist of methyl-group metabolism, therefore may give rise to an increase risk of colon neoplasia through DNA hypomethylation⁹⁵.

Furthermore, a diet high in folate and methionine can counteract the poor methyl group availability that alcohol causes¹⁰⁰.

The International Agency for Research on Cancer in a review in 1988 found that 4 of the 9 cohort studies and 6 of 9 case control studies demonstrated significant increase in rectal cancer particularly in beer drinkers¹²⁴.

Klatsky et al.¹²⁵ showed that when daily alcohol intake of three or more drinks was compared with abstainers, relative risk for rectal cancer was 3.17 (95% CI: 1.05 - 9.57) and relative risk for colon cancer was 1.71 (95% CI: 0.92 - 3.19).  This association was stronger in women who demonstrated a relative risk for colon cancer of 2.56 (95% CI: 1.03-6.40) compared with a relative risk of 1.16 (95% CI: 0.46-2.90) for men with colon cancer.

Sugar

In a small study by Bristol et al.¹²⁶, colorectal cancer patients were found to consume 16% more energy than controls mainly in the form of carbohydrate (21%) and fat (14%).  Furthermore, the carbohydrate was mainly ingested as refined sugars.  This study, however, was retrospective and therefore could be liable to recall bias, and the findings for carbohydrate intake were not corrected for total energy and fat intake.

Tuyns et al.⁹⁸ also showed a detrimental effect of high sugar (oligosaccharides) as opposed to polysaccharides but with no association with confounding risk factors such as fat and protein intake.

Dietary Fibre, Fruit & Vegetables

Dietary fibre, principally derived from cereals, fruit and vegetables, has been found consistently to be protective in colorectal cancer in many epidemiological studies^127-130.  By increasing faecal weight, diluting large intestinal contents, and speeding up transit time, fibre is thought to change the milieu within the colon to reduce interaction between faecal mutagens and the mucosa⁹².

However, a recent large prospective study of 88,757 women by Fuchs et al.¹³¹refuted this hypothesis following adjustment for age, established risk factors, and total energy intake.

Obviously, fruit and vegetables contribute to the total amount of dietary fibre we consume, but as a factor independent to total dietary fibre, cruciferous vegetables (such as broccoli) have been shown to be protective against colorectal neoplasia¹⁰⁹.  More specifically high carotenoid vegetables, cruciferae, high vitamin C fruits show the greatest reduction in incidence of colorectal adenomas⁹¹ suggesting that micronutrients within these food items play a part in addition to the fibre they contain.

Micronutrients

There are many papers written on micronutrients and colorectal neoplasia few have significant statistical power and, as with investigation of any dietary variable, there are many confounding genetic and co-environmental factors.  Certainly, the overall impression is that these factors may make a small difference in the aetiology of colorectal neoplasia and that much more work is required to define their role in carcinogenesis.

Calcium

In the United States in 1985, Garland et al.⁹⁴ investigated an observation that mortality rates from colorectal cancer were higher in populations exposed to the least amounts of natural sunlight, suggesting that differences in vitamin D production and calcium absorption could be responsible.  Risk of colorectal cancer was inversely correlated with dietary vitamin D and calcium and this remained true after adjustment for age, daily cigarette consumption, body mass index, ethanol consumption, and percentage of calories obtained from fat.

More recent studies have been divided as to whether low calcium intake was a risk factor^{88,128,132-135} for colorectal neoplasia or no association was seen.^136-141

It has been postulated that calcium acts to reduce the incidence of colorectal neoplasia by reducing lipid damage in the colon by complexing with fat to form mineral-fat complexes or soaps.

Vitamins

There has been great interest in the anti-oxidant vitamins A(and its provitamin beta-carotene), C and E as protective agents in colorectal carcinogenesis.  In addition to dietary intake, these compounds are common supplemental dietary products taken either separately or as "multivitamins" popularised by health promotion.

Vitamin A and its pro-vitamin beta-carotene act as anti-oxidants, are found in many fruits and vegetables and have been shown to be protective against colorectal neoplasia in the majority of studies^{114,137,142-145}, whilst a few have shown no significant effect^109,146.

Vitamin C has been found in humans to reduce colonic crypt cell proliferation¹⁴⁷ and therefore postulated to reduce neoplasia.  Again, some studies have demonstrated a preventative effect^{93,96,109,133,134,142,148} though a few showed no significant difference^{145,146,149,150}.

In 1980 Cook et al.¹⁵¹ investigated the effect of dietary vitamin E levels on colonic neoplasia in mice.  The results demonstrated a lower rate of adenoma and cancer formation in the high dose vitamin E group.  This result has been reproduced in several centres^{137 144,146} but most reports have shown no significant difference in relative risk^{143,145,150,152,153}.

Selenium

The trace element selenium has been associated with a decreased risk of colorectal neoplasia^154-158.  In a rat model, Feng et al.¹⁵⁹ demonstrated the ability of selenium to reduce aromatic amine-induced colon carcinogenesis.

Smoking

Tobacco smoke is a major source of a multitude of carcinogens including nitrosamines, polycyclic hydrocarbons and heterocyclic amines¹⁶⁰. Cigarette smoking has been strongly associated with colorectal adenomas^161-164 but this association has been less strong with colorectal cancer^111,165,166 although Giovannucci et al. in a large cohort study produce convincing data for a lead time between smoking and cancer formation of some 35 years^162,163. 

Terry et al.¹⁶⁵ postulated that this was because subjects with colorectal adenomas were included in the control group of cancer case-control studies.  Analysis of a "manufactured" control group compared to a pure adenoma-free control group showed a trend in keeping with this hypothesis but did not reach statistical significance.

In addition, it may be that the effect of cigarette smoking on the colorectal adenoma-carcinoma sequence occurs in the earlier stages of the formation of adenoma.

Physical Activity and Body Mass

Most studies that have investigated the relationship between physical exercise and colorectal neoplasia have found an inverse relationship^111,167,168.

One hypothesis for the role of physical activity in colorectal neoplasia is that exercise stimulates colonic peristalsis and thereby reduces colonic transit time^169,170.

  The combination of high physical activity and lower body mass effects many of the bodies homeostatic mechanisms - decreased insulin, glucose, triglycerides¹⁷¹, prostaglandins¹⁷² and possibly growth factors. 

Alternatively, there is an obvious link between physical activity and obesity.  However being overweight is likely to be a surrogate¹⁷³. As such, risk factors including a high-fat, high energy diet, with inadequate consumption of fruit and vegetables; and lack of physical activity are likely to contribute to a high incidence of colon cancer as well as obesity.  Therefore body mass per se has only been weakly associated with colorectal neoplasia¹⁶⁷.

Pharmaceuticals

There is a very wide variety of pharmaceutical products that are in use - probably in excess of 100,000¹⁷⁴.  It is hardly surprising then that some groups of products have been found to influence colorectal neoplasia.

Hormone Replacement Therapy

Hormone replacement therapy has been consistently shown to decrease the risk of colorectal neoplasia and there is an inverse relationship with duration of treatment (odds ratios compared with no hormone replacement therapy range from 0.39 to 0.74)^175-178.

It has been postulated that the reason for this is that with time the oestrogen receptor gene is silenced by methylation¹⁷⁹ and hormone replacement therapy reverses this trend.  Experimental data has shown that endogenous oestrogens protect against Apc-associated tumour formation and is associated with an increase in oestrogen receptor beta and a decrease in oestrogen receptor alpha expression in the target tissue¹⁸⁰.  Alternatively, oestrogens may act on the colonic vitamin D receptor to decrease neoplasia¹⁸¹.

Non-steroidal anti-inflammatory drugs (NSAIDs)

As can be seen from Vogelstein's genetic model for colorectal carcinogenesis (Figure 3), over expression of the cyclo-oxygenases COX1 and COX2 is thought to be an early step in the aetiology of colorectal adenomas and carcinomas²⁷.  The Non-Steroidal Anti-Inflammatory Drugs (NSAIDs) are potent inhibitors of cyclo-oxygenase¹⁸² and have been shown to inhibit polyp and cancer formation^127,183,184 even in patients with FAP (Page 36)^185-187.

1.c.iv. N-acetyltransferases

The N-acetyltransferases (NATs) are phase II cytosolic enzymes which catalyse the transfer of acetate from acetyl CoA to primary amine and hydrazine groups, forming acetamides and hydrazides¹⁸⁸.  Substrates for these enzymes include drugs such as isoniazid, procainamide, sulphamethazine, caffeine,  and  occupational  carcinogens,  such  as benzidine  and  4- aminobiphenyl, and amino acid pyrolysates, produced as a result of charring food^189,190.  The N-acetyltransferases therefore catalyse the formation of mutagenic products from heterocyclic amines, dietary carcinogens that are largely derived from cooked meat.  Enzyme activity in humans is coded for by 2 distinct genes, designated NAT1 and NAT2 which are located on chromosome 8 at 8p21.3-23.1 (Figure 6)⁴⁸.  In addition, there is a pseudogene termed NATP¹⁹¹ that is located on chromosome 8 at the p22 region but this does not encode a functional NAT protein¹⁹².

Both the functional NAT genes contain 870 base pair intronless protein coding regions encoding 290 amino acid proteins¹⁹³ and exhibit polymorphisms that give rise to differences in enzymatic function.  The nomenclature for these mutations is given in Appendix 1.

Figure 6. Locus of NAT1 and NAT2 on chromosome 8.

Windmill et al.¹⁹⁴ has demonstrated through use of hybridisation histochemistry that mRNA for both NAT1 and NAT2 are expressed in liver, gastrointestinal tract, urinary tract and lung - and that in the extrahepatic tissues the mRNA is seen within the epithelial cells of that organ.  Quantitative analysis has shown that NAT2 is primarily expressed in the liver, whereas NAT1 is primarily expressed in extrahepatic tissues including the colon^195-197.

N-acetyltransferase 1 (NAT1)

NAT1 was historically considered to be monomorphic in nature but reports of allelic variations at the NAT1 locus by Weber et al. suggested that it is a polymorphically expressed enzyme¹⁹⁸.  Currently, there are 26 polymorphisms described (Appendix 1: Arylamine N-Acetyltransferase Nomenclature).

The NAT1*3 (C1095A substitution), NAT1*20 (T402C substitution) and NAT1*23 (T777C substitution) alleles contain silent mutations and do not lead to any amino acid changes in the NAT1 protein.  Activity is thought to be the same as NAT1*4 (wildtype)¹⁹⁹.

There is some debate as to how the alleles NAT1*10 and NAT1*11 change the activity of the NAT1 protein.  The NAT1*10 allele (polyadenylation sequence) has been associated with twice the wild-type activity in human colonic tissue²⁰⁰ despite no amino acid change - the enhanced activity is thought to be because the T1088A substitution leads to a polyadenylation signal (TAATAA - TAAAAA) in the 3'-untranslated region that may enhance mRNA stability.  In keeping with the model for carcinogenesis, the NAT1*10 allele has been associated with an increased risk of bladder²⁰¹, breast²⁰², gastric²⁰³, lung¹⁶⁰ and oral cancer²⁰⁴.  Persons with the NAT1*11 alleles have also be considered as fast acetylators following the work of Doll et al.²⁰⁵.  The novel allele they identified and initially termed NAT1*17 was found to be a variant of NAT1*11 (now termed NAT1*11A) and demonstrated twice the activity compared to the wildtype.  In  a recent publication, however, by De Leon²⁰⁶ activity in the NAT1*10 and NAT1*11 alleles was unchanged.  Similarly, Bruhn et al.²⁰⁷ showed no difference in NAT1*10 activity and also showed NAT1*11 activity to be decreased in comparison to the wildtype.

It is agreed that the alleles NAT1*14, NAT1*15, NAT1*17, NAT1*19 and NAT1*22 reduce NAT1 protein activity^199,207-209. 

There is little data on the functional repercussions of the other NAT1 variants (Appendix 1: Arylamine N-Acetyltransferase Nomenclature).

Associations of NAT1 and Colorectal Neoplasia

NAT1 genotype and colorectal neoplasia

The studies that have analysed NAT1 genotype and colorectal neoplasia are listed in Table 8.

Of the six studies, only Bell et al.²¹⁰ showed a significant difference between NAT1 genotype and colorectal cancer.  The control group was taken from non-cancer hospitalised volunteers and so was probably not representative of the population at risk.  The NAT1*10 allele was compared to the others investigated (NAT1*3, NAT1*4 and NAT1*11) on the basis of putative fast phenotype for NAT1*10.  In addition, there was seen to be a stronger association in the presence of fast acetylator genotype for NAT2 (odds ratio, 2.8; 95% confidence interval, 1.4-5.7; P = 0.003), suggesting a possible gene-gene interaction between NAT1 and NAT2 (test for interaction; P = 0.12).

Of the remainder, 3 studies looked at the association of NAT1 with colorectal cancer.  Chen et al.²¹¹ analysed the same genotypes assigning them on the same basis as Bell et al.²¹⁰ with adequately matched population controls but failed to demonstrate any significant difference.  Lee et al.²¹² could be criticised for having a control group which was much younger than the cases (controls mean age = 27 years, case  mean age = 47) and Hubbard et al.²¹³ looked at 2 uncommon alleles and though they found that the *14 allele only occurred in the colorectal cancer group this did not reach statistical significance.

The 2 studies that looked at the relationship of adenomas and NAT1 genotype^199,214 benefited from having larger numbers. These studies however came out of the same institution and therefore there is considerable overlap in the cases and controls used.  The obvious control group was those attending for flexible sigmoidoscopy with no adenoma found - this was not in the context of a screening trial where one might expect the cases and controls to have the same risk of colorectal neoplasia.  Instead, the indications were "routine", "specific minor symptoms" or "not given".  There must be some question as to the comparability therefore as certainly for those who had symptoms and no polyp was found missed polyps must be a concern.  In neither study was there any association seen between colorectal adenomas and NAT1 genotype.

There are many more alleles that have been discovered since these studies (Appendix 1: Arylamine N-Acetyltransferase Nomenclature), and more is known about their putative phenotype, and this may introduce errors in analysis.  For instance, genotype misclassification (e.g. NAT1*10 not being distinguished from NAT1*14A, not described until 1998^209,215, which has an additional substitution G⁵⁶⁰A, the putative phenotype being fast and slow respectively) can produce substantial errors, requiring large sample sizes especially to investigate gene-environmental interactions^216,217.

N-acetyltransferase 2 (NAT2)

Three NAT2 phenotypes have been described dependent on the number of wild-type (NAT2*4) alleles.  Homozygous NAT2*4 is associated with fastest NAT activity, heterozygous with intermediate activity and homozygous mutant alleles with slow activity^218,219.  NAT2 phenotype can be readily measured in man by analysing the urinary metabolites after caffeine ingestion^220,221.

There are 10 recognised point mutations within the NAT2 gene (Table 9).

In European populations there is relatively high concordance between  acetylator phenotype and genotype^{218,219,222,223}.

Associations of NAT2 and colorectal neoplasia

NAT2 phenotype and colorectal neoplasia.

NAT2 phenotype has been assessed by looking at the metabolism of substances such as sulphamethazine which is a substrate specific to NAT2.  The principal studies looking at NAT2 phenotype and its relation to colorectal neoplasia (either cancer or adenoma formation) are summarised in Table 10.  Three of the four^226-228 studies show a positive association with colorectal cancer, and the fourth²²⁹ shows no association.  In addition, Robert-Thomson et al.²²⁸ showed no association for colorectal adenomas and NAT2 phenotype.

It can be seen that these studies were small and the control groups were often poorly defined - for instance Lang et al.²²⁶ chose hospital patients without malignant disease and Roberts-Thomson et al.²²⁸ used patients who had undergone colonoscopy or barium enema that revealed no neoplastic lesions.  As such, their interpretation could be open to bias as the control group is not representative of the population at risk.

NAT2 genotype and colorectal neoplasia

The hypothesis that polymorphisms in the NAT2 genotype modify cancer risk is supported by an increase in DNA adducts produced in the presence of fast acetylation^230,231 and an increase in urinary mutagenicity²³² indicating that there is a greater production of mutagenic compounds.

This however has not translated in to statistically significant differences in observational studies looking at colorectal cancer and adenomas (Table 11).  In all but one there was no significant difference seen between NAT2 fast/intermediate genotype and colorectal neoplasia.  Criticisms of this study  are that the numbers were relatively small (114 in the case group), the controls were significantly younger than the cancer group (mean age 46 vs. mean age 62 in case group) and the areas of residence of the cases and controls differed²³³.  These issues give rise to doubt as to how comparable the groups are for analysis.

Bias may have been introduced in to some of the analyses because the control groups were not representative of the population at risk - only 4 of the trials had reasonable control groups (Table 11)^211,234-236.  The remainder either had non-representative controls^{64,196,210,212,233,237-239} or the methods of control selection were not adequately described²⁴⁰.

Despite the lack of strong evidence for an association of NAT2 genotype and colorectal neoplasia alone, it must be remembered that N-acetyltransferase does not act alone in the proposed model of carcinogenesis.  Therefore, probably the more interesting analysis is the assessment of combinations of genotype and their relationship to ingestion of substrates such as the arylamine compounds in well-cooked red meat.

N-acetyltransferase gene - environment interactions

In order for the N-acetyltransferase enzymes to exhibit their effect in catalysing the formation of mutagens that may influence carcinogenesis, there has to be a substrate on which they act.

NAT genotype, colorectal neoplasia and red meat consumption

The most extensively studied interaction is between NAT2 and heterocyclic amines produced from well-cooked red meat.  Skog et al. ²⁴² demonstrated a link between temperature and charring and the amount of heterocyclic amines produced whilst cooking.  The studies that looked at subgroup analysis of colorectal neoplasia, NAT2 phenotype or genotype and well-cooked red meat intake are listed in Table 12.

Two studies investigated NAT2 phenotype compared with high red meat intake.  Lang et al. investigated fast NAT2 acetylator phenotype in the context of fast CYP1A2 phenotype with a high well-cooked red meat diet on the basis that the two enzymes both act to produce N-acetoxy arylamine that binds to DNA to give carcinogen-DNA adducts⁷⁶.  This showed an odds ratio of 2.79 for  the risk of developing colorectal cancer or adenomas.  Unfortunately, the numbers were small, only 75 in the group of cases and these were heterogeneous (made up of both cancer and adenoma cases), and the fast CYP1A2 / fast NAT2 subgroup at risk only consisted of 26 cases. 

Roberts-Thomson et al.²²⁸, looked at cancer and adenoma cases, NAT2 phenotype and divided red meat intake in to tertiles.  The covariate-adjusted odds of neoplasia over the three meat categories are shown in Table 12, these did not reach statistical significance.

Three studies investigated NAT2 genotype, colorectal cancer and meat intake^211,235,236.  Welfare et al.²³⁵ demonstrated that fried meat consumption of more than twice a week was found more in the fast acetylator cases (18%) than in their matched controls (4%) (OR = 6.0).

Chen et al.²¹¹ investigated both fast NAT1 and fast NAT2 as the 'highest risk phenotype'.  When taken in isolation, there was no significant increased risk for fast NAT2 alone with a high red meat diet, but the fast NAT1 / fast NAT2 phenotype with high red meat was associated with an almost six-fold increased risk.

The largest study however is that by Kampman et al.²³⁶, it was also the most detailed and each of the 3402 participants were interviewed to ascertain their dietary habits.  The interview questionnaire was validated at a previous study²⁴³ and included over 800 food items, data on portion sizes and "doneness" of red meat.  The participants however were asked to recall their diet for the 12 month period 2 years before they either developed colorectal cancer or were interviewed as a control.  Despite detailed analysis of many meat categories, the only significant results related to NAT2 genotype were for 'processed meat' (OR = 1.5, 95% CI 1.1 -2.0) and for total meat mutagen index (OR = 1.3, 95% CI 1.0 - 1.7).  There was however a suggestion of a trend with amount of red meat and "doneness"

There is no strong evidence for the postulated risk of fast NAT2 genotype alone, red meat diet and colorectal neoplasia.  Several of the studies above in addition to investigating NAT2 genotype have also looked at other polymorphic metabolic enzyme genes such as NAT1^76,211 and GSTM1²³⁶, and it is the combination of such gene-environment and gene-gene interactions that produces notable results.

There are no studies that have demonstrated a significant interaction between NAT1, colorectal neoplasia and red meat intake.  In a recent paper by Ishibe et al.²⁴⁴, it was demonstrated that there was an increased adenoma risk with increasing heterocyclic amine exposure (as measured by 2-amino-3,8-dimethylimidazo(4,5-f)quinoxaline (MeIQx) intake).  There was a six-fold increase in adenoma risk among rapid NAT1 acetylators who consumed high levels of MeIQx (OR = 6.50; 95% CI 2.16-19.7), whereas among slow NAT1 acetylators, the increase in risk was two-fold (OR = 2.32; 95% CI 1.12-4.81). While suggestive, these results however were not significantly different from each other.

NAT genotype, colorectal neoplasia and smoking

The other main postulated source of procarcinogens and carcinogens implicated in colorectal neoplasia is cigarette smoking.  Tobacco smoke contains high amounts of arylamines that are metabolised in part by NAT1 and NAT2.  The main studies investigating this association are in Table 13.

Probst-Hensch et al.⁶⁴ demonstrated an increased risk of adenoma formation in the NAT2 fast acetylator smoking group in comparison to the slow NAT2 non-smoking group.  Welfare et al.²³⁵ did not show any difference in the prevalence of colorectal cancer amongst the fast acetylators, but in the group of slow acetylators there was a significantly increased risk amongst smokers.

In three more recent larger studies, Slattery et al.²³⁴, Potter et al.²³⁹ and Tiemersma et al.²⁴⁵ found that the main determinant for risk of colorectal neoplasia was smoking and that NAT genotype modified this risk very little.

N-acetyltransferase gene - gene interactions

The model for carcinogenesis in Figure 5 (Page 44) is illustrated using a two step pathway with the intermediate compounds being carcinogenic.  This is obviously an over-simplification of what happens in vivo to the complex arylamines and heterocyclic aromatic amines exposed to through eating well cooked red meat and smoking.  These compounds may well go through several cycles in their detoxification.

NAT1 / NAT2 gene - gene interaction

Aromatic amines and their metabolites can be N- or O-acetylated by the NAT1 and NAT2 enzymes that have affinities for different substrates.  NAT1 / NAT2 interactions have been studied by Chen et al.²¹¹ above (Table 12), demonstrating a combined effect of fast acetylation status in both enzymes increased the risk of colorectal cancer in the presence of high red meat intake.  Similarly, Bell et al.²¹⁰ showed that the influence of the NAT1*10 allele on colorectal cancer was most apparent among NAT2 fast acetylators.  Lin et al.¹⁹⁹ however failed to show an association between NAT1 / NAT2 and the prevalence of colorectal adenomas.

NATs / GSTs gene - gene interaction

There have been few studies that have analysed the interaction of the N-acetyltransferase and glutathione S-transferase genotype (with or without dietary or smoking data).

Probst-Hensch et al.⁶⁴ reported a significantly higher risk of developing a colorectal adenoma in GSTM1 null / fast NAT2 genotype smokers compared to GSTM1 non-null / slow NAT2 genotype smokers (OR = 10.3, 95% CI 1.94 - 55.0).  There was no observed difference seen in the same genotype groups among the non-smokers.

Slattery et al.²³⁴ failed to demonstrate a convincing genotype combination of NAT2 and GSTM1 in a case control study of 1993 cases of colon cancer and 2410 age- and sex- matched controls.  Subgroup analysis did however reveal a dose response to smoking exposure amongst the GSTM1 null / NAT2 slow genotype - <20 cigarettes / day OR = 1.4 (95% CI 0.8 - 2.3), >20 cigarettes / day OR = 1.7 (95% CI 1.2 - 2.6).

Welfare et al.²⁴⁶ investigated the association of GSTM1, GSTT1, GSTP1 and NAT2 genotypes in a matched case-control study of 178 colorectal cancers.  There was no effect seen for GSTM1, GSTT1, or GSTP1 on colorectal cancer susceptibility.  However, individuals with both the GSTT1 null and NAT2 slow genotypes combined appeared to be at increased risk of colorectal cancer (OR = 2.33, 95% CI 1.1 - 5.0).

These observed difference in the genotypes associated with increased risk of colorectal neoplasia may represent that some of the subgroup analyses significance has arisen by chance, or possibly that geographic variations and hence dietary and ethnic variations can change the 'at risk' genotype combinations dependent on the substrates that are available.  This explanation was considered by Probst-Hensch et al.⁶⁴ having observed that white fast acetylators potentially had slightly increased risks for adenomas (OR = 1.29, 95% CI 0.90-1.84), whereas fast acetylation was potentially protective among blacks compared to slow(OR = 0.64, 95% CI 0.32-1.28).

NATs / CYPs gene - gene interaction

The association of a phase I metaboliser (CYPs) and a phase II metaboliser (NATs) (Table 3) giving rise to a high risk genotype combination would fit nicely with the model for carcinogenesis (Figure 5).

Lang et al.⁷⁶ analysed the combined NAT2 and CYP1A2 phenotypes.  The combined fast NAT2 / fast CYP1A2 phenotype was found in 35% of cases and only 16% of the controls, giving an odds ratio of 2.79 (P = 0.002).  Furthermore, fast NAT2 / fast CYP1A2 compared to slow NAT2 / slow CYP1A2 in a group that liked rare / medium cooked red meat had a relative risk of 3.1.  This contrasts with the same genotype analysis for those that liked well-cooked red meat in whom the relative risk was 6.5 for the fast / fast genotype.

1.d.i. Rationale for screening for colorectal cancer

In common with other cancers such as breast and cervix, colorectal cancer is suited to the institution of a screening programme:

·         it is a major health problem (1.b.i. Demographics)

·         there is a lengthy precancerous stage i.e. the adenomatous polyp.

·         intervention at an earlier stage affects morbidity and mortality²⁴⁷.

·         modalities are available to detect polyps.

·         the cost of screening is not prohibitive

1.d.ii. Screening methods for colorectal cancers and adenomas

At least 75% of colorectal cancers arise in persons who have no known family history⁴².  As such the screening of high risk families, while potentially valuable to the individuals, is unlikely to have a major impact on the overall incidence or mortality.  Therefore, population based screening is recommended as the way forward ¹⁰.

There are several methods by which the population might be screened:

·         Symptom questionnaire

·         Faecal Occult Blood Testing (FOBT)

·         Barium enema

·         Colonoscopy

·         Flexible sigmoidoscopy

·         New advances e.g. K-ras stool testing

Symptom Questionnaire

Bowel symptoms are very common and can be an indicator of many general medical conditions (e.g. thyrotoxicosis), large bowel pathology (e.g. colitis) or functional bowel disorders (e.g. irritable bowel syndrome).  As such, one cannot define a high risk group for further intervention through symptomatology alone and the addition of a symptom questionnaire to other screening modalities increases costs with little return²⁴⁸.

Faecal Occult Blood Testing (FOBT)

FOBT is mainly aimed at the detection of early asymptomatic cancers.  The premise is that such cancers bleed and the detection of these small amounts of blood define a high risk group who undergo further intervention (colonoscopy)^247,249,250.  The main advantages of FOBT screening is that it is non-invasive, easily performed without the need for bowel preparation, can be performed on transported specimens and of low cost.  The main disadvantages, however, are low sensitivity, because 40% of cancers and 80% of adenomas are missed by the test^251,252, and the late stage in the disease at which lesions bleed, leading to a short lead-time and a requirement for frequent testing.

Barium Enema

Traditionally before the advent of fibre-optic technology, imaging of the colon was performed using a barium enema.  It has been shown however that the sensitivity for even quite large colonic lesions can be quite low and the examination requires full bowel preparation and is poorly accepted by patients.

Colonoscopy

Initially, colonoscopy may appear to be the best method for screening for colorectal adenomas and cancers.  The sensitivity for even small polyps as small as 5mm is high so that neoplasia is detected at an early stage.  Also, lesions can be removed at the time of screening so colonoscopy can be both diagnostic and therapeutic. 

There are however disadvantages to colonoscopic screening:

·         full bowel preparation is required.  This normally involves stimulant or osmotic laxatives (e.g. Picolax or Klean prep).  Recommendations include a low fibre diet for several days prior to commencement of preparation which itself starts the day before the examination.

·         the majority of colonoscopic examinations are carried out with sedation.  This excludes the subject from driving to and from the hospital, prevents the adequate discussion of results on the day and can be distressing both from the anxiety of not being fully conscious but also the hangover effects that are reported.

·         complication rates including perforation are reported as being approximately 1 in 2,000 and a mortality of 1 in 5,000²⁵³.  Though this may be acceptable in clinical practice in symptomatic patients, it would represent an obstacle to the acceptance of the population to be screened.

·         the overall compliance is lower or comparable to flexible sigmoidoscopic screening^253,254.

·         colonoscopy requires considerable training, skill and experience.  At present, as colonoscopy becomes the investigation of choice for colonic disorders over barium enema, there is a relative shortage of colonoscopists.  A trend that seems set to continue and would increase in magnitude should screening be introduced.

·         colonoscopy is the most expensive of the potential screening tools.

Flexible Sigmoidoscopy

Flexible sigmoidoscopy has advantages over colonoscopy in that:

·         bowel preparation can be undertaken with a phosphate enema that gives good results, is quick and easy to be self-administered at home and is acceptable.

·         sedation is not normally required

·         there is a very low complication rate including perforation

·         it is associated with 70% acceptance rate²⁵⁵

·         it can be performed by non-medical personnel thus reducing the logistic problem of requiring more endoscopists

·         it is relatively cheap

However, flexible sigmoidoscopy does not allow visualisation of the proximal bowel.  A 60cm sigmoidoscope, however, can be passed in most cases to the junction of the sigmoid and descending colon below which 60% of colorectal cancers are located.

New Advances

Screening for stool markers that are more accurate than occult blood would substantially improve sensitivity.  There is great interest in looking for the DNA alterations that occur in the formation of polyps and cancers in cells exfoliated from neoplasms.  Early investigations targeting single mutations, usually K-ras that is present in less than half of all colorectal neoplasms, show that mutations in tumour can be detected in stools from the same patients^256-258.  Colorectal neoplasms however are genetically heterogeneous and no one mutation has been found to be universally expressed.  It is likely that an approach of investigating multiple mutations commonly expressed would improve diagnostic yield.  Ahlquist et al. demonstrated  sensitivities of 91% for colorectal cancer and 82% for adenomas >1cm using a multi-target assay that assessed 15 mutations commonly seen in colorectal neoplasia²⁵⁹.

1.d.iii. Rationale for flexible sigmoidoscopy screening - FlexiScope Trial

The FlexiScope trial is a multicentre randomised controlled trial looking at the efficacy of a single flexible sigmoidoscopy and polypectomy as required at between the ages of 55 and 64 years in preventing colorectal cancer.  It is funded by the Imperial Cancer Research Fund and the Medical Research Council and has completed the screening phase of 40,000 people.  The rationale for the protocol is detailed in the Lancet paper by Atkin et al.¹⁰ a brief summary is given below.

Polpypectomy

The benefit of polypectomy at reducing the incidence of colorectal cancer has been demonstrated^260-264  but until the FlexiScope trial, which will not report for at least 5 years, there has not been a well-designed randomised controlled trial looking at the efficacy of flexible sigmoidoscopy in the prevention of death from colorectal cancer.

Gilbertsen et al.²⁶⁰ followed up over 21,000 subjects after rigid sigmoidoscopy and polypectomy for a mean of 4 years with those having polyps removed undergoing annual surveillance.  In this time, 13 rectal cancers were diagnosed compared with 90 cancer that might have been expected concluding an 85% reduction in incidence by polypectomy.

A smaller case-control study by Newcomb et al.²⁶³ found a reduction in the incidence of colorectal cancer of 80% after examinations done mostly by flexible sigmoidoscopy.

Single Examination at age 55

Atkin et el.²⁶⁴ investigated the long term risks of rectal cancer following sigmoidoscopy in 1618 men and women who had adenomas removed via the sigmoidoscope up to 30 years previously.  Only 14 (0.9%) developed rectal cancers compared to the estimated 80 cancers expected and 11 of these had had incomplete removal of their adenomas.  So only 3 rectal cancers developed after complete adenoma removal and all of them were diagnosed 17 or more years after the procedure.

The conclusion therefore was that if all adenomas detected at sigmoidoscopy are removed completely the risk of rectal cancer may be very low for many years after²⁶⁵. 

There remains some uncertainty as to the ideal age at which screening should take place.  In keeping with the lead time for mutation of an adenoma to carcinoma^13,16, adenoma prevalence increases markedly after the age of 50 and appears to plateau at the age of 60.

Screening at an older age (>60) would have advantages in adenoma pick up rate and greater reassurance of a negative test, but would miss the 7% of cancers occurring in the 55 -59 age group.  It was decided to include the age range 55 - 64 to try and determine the best age for screening to take place.