Introduction

Parts of this chapter have been reproduced from Hull et al.1 © 2018 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/).

Background

The health burden of colorectal cancer

Colorectal cancer (CRC) continues to represent a huge health burden in the UK. There were approximately 41,300 new cases of CRC (also known as bowel cancer) in the UK in 2014, making it the fourth most common cancer.2 In the UK, 1 in 14 men and 1 in 19 women will be diagnosed with CRC during their lifetime. CRC is the second most common cause of cancer death in the UK, causing 16,000 deaths in 2014. Worldwide, it has been estimated that nearly 1.4 million cases were diagnosed in 2012.2 Despite significant advances in the diagnosis and treatment of CRC, overall 5-year survival is currently only 59% (survival figure is for England and Wales for 2010 and 2011).2

Prevention of colorectal cancer

One strategy to reduce CRC incidence and mortality is prevention. The scientific and clinical rationale for prevention of CRC is well established and is based on the following:

• Knowledge of several environmental and behavioural factors that increase CRC risk, including dietary factors (e.g. red and processed meat intake), excess body weight (i.e. obesity), lack of physical activity, tobacco smoking and excess alcohol consumption. The World Cancer Research Fund has estimated that approximately 45% of CRCs are preventable based on modification of these lifestyle factors.3
• The long natural history of colorectal carcinogenesis, during which a benign, precursor lesion termed a colorectal adenoma (or polyp) develops and transforms into a malignant neoplasm over a period of years (estimated to be approximately 5–10 years).4,5
• Improved outcomes for CRC treatment after diagnosis at earlier stages of CRC (98% 1-year overall survival for stage I CRC compared with 40% for stage IV disease; 95% 5-year survival for stage I compared with 7% for stage IV disease).6,7

Colorectal cancer prevention strategies that are currently used, or are under evaluation, include:

• population screening –
  • early CRC diagnosis by a guaiac faecal occult blood test (FOBt) or a faecal immunochemical test (FIT), conferring secondary benefit from colorectal adenoma identification and removal at colonoscopy
  • detection and removal (by polypectomy) of colorectal adenomas by primary screening endoscopy [colonoscopy or flexible sigmoidoscopy (FS)]
• endoscopic surveillance of high-risk groups, for example individuals with long-standing colitis or previous colorectal adenoma(s)
• chemoprevention (a term first coined in 1976 to describe the use of drugs, vitamins or other nutritional agents to try to reduce the risk of, or delay the development or recurrence of, cancer)
• health education, leading to beneficial lifestyle modification and screening uptake
• promotion of awareness and earlier diagnosis of CRC by education aimed at the public and health-care professionals.

Chemoprevention of colorectal cancer

Despite the undoubted clinical effectiveness of endoscopic polypectomy,8,9 CRC remains a significant problem in screened populations and high-risk surveillance cohorts because of a combination of factors that include suboptimal screening uptake, poor acceptability of endoscopic procedures and ‘interval’ CRC (i.e. those cancers that are diagnosed despite FOBt/FIT and/or endoscopy).10–13 In a Dutch biennial FIT-based CRC screening programme,14 23% of individuals developed a FIT-interval CRC. The corresponding interval CRC rate for guaiac FOBt programmes is approximately 50%.15 Moreover, 18 years’ follow-up of the Minnesota guaiac FOBt trial (in which the colonoscopy rate was nearly 40%) found only a 20% reduction in CRC incidence.13 The UK once-only FS trial10 demonstrated only a 23% reduction in CRC incidence in the intervention group compared with the control group at 10 years. It is also clear that CRC occurs even in patients under close colonoscopic surveillance (1.7 CRCs per 1000 person-years),16 with an estimated post-colonoscopy CRC (PCCRC) rate of between 2% and 9%.17 Overall, only 10% of CRCs in the UK in 2013 were diagnosed within the UK screening programmes.18 Therefore, there is still an unmet clinical need for safe and effective primary CRC chemoprevention in combination with existing screening and surveillance programmes.

The natural history and molecular pathogenesis of colorectal carcinogenesis

The molecular pathogenesis of CRC (histopathological term: colorectal adenocarcinoma) has been the subject of several recent reviews.4,19 In recent years, the original multistage model of cumulative genetic mutations, proceeding through a benign adenoma stage (the so-called adenoma–carcinoma sequence), based on loss of function of the ‘gatekeeper’ tumour suppressor gene adenomatous polyposis coli (APC), which was proposed by Fearon and Vogelstein,20 has been superseded by widespread acceptance that ‘sporadic’ (i.e. not occurring on a background of a distinct genetic predisposition syndrome or inflammatory bowel disease) CRC is not ‘one disease’, but occurs via several pathogenic pathways, which are not mutually exclusive.21,22

The chromosomal instability (CIN) pathway is characterised by chromosomal abnormalities, including aneuploidy, usually associated with loss-of-function APC mutation and later-stage gain-of-function KRAS mutation.23 It is exemplified by the rare genetic predisposition syndrome familial adenomatous polyposis (FAP), an autosomal dominant condition that occurs in carriers of a heterozygous germline APC mutation.24 The microsatellite instability (MSI) pathway is driven by defective deoxyribonucleic acid (DNA) mismatch repair (MMR), leading to accumulation of further somatic mutations (termed MSI-high) including BRAF.23 It is exemplified by Lynch syndrome, in which carriers of mutations in MMR genes (most commonly MLH1 and PMS2) exhibit increased cancer (including colorectal) risk.25 Furthermore, a CpG island methylator phenotype (CIMP) pathway is recognised in colorectal adenomas and adenocarcinomas, by which epigenetic changes in DNA methylation lead to altered gene function, in particular silencing of the MMR gene MLH1 leading to defective MMR, which is associated strongly with BRAF mutation.23 Epigenetic silencing of MLH1 explains the substantial overlap between MSI and CIMP pathways.4

There are limited data on how early during colorectal carcinogenesis the above phenotypes manifest themselves.4 There are some data to suggest that CIN features are present in adenomas.4,26 The CIMP pathway is linked strongly to benign serrated lesions (see below).

Colorectal adenoma

The importance of the benign precursor lesion, which exhibits epithelial cell dysplasia but not invasion of the epithelial basement membrane (termed the adenoma, or adenomatous polyp), as a risk stratification biomarker of future CRC risk but also as a clinically significant lesion (the removal of which is associated with reduced CRC incidence and mortality), has been reviewed in detail.5

In parallel with more nuanced understanding of the diverse molecular pathogenesis of CRC, the histopathological classification and terminology of the colorectal adenoma has been revised.5 In particular, hyperplastic-serrated pathway lesions are now acknowledged as separate entities from the more common (conventional) dysplastic adenoma (which can be tubular, villous or mixed tubulo-villous in morphology). Serrated lesions are recognised to have malignant potential per se and may account for 20–30% of CRCs.27 The World Health Organization’s (WHO’s) WHO Classification of Tumours of the Digestive System28 in 2010 included the term sessile (a term used to recognise that the vast majority of these lesions are ‘flat’ and not polypoid when viewed endoscopically) serrated adenoma. However, the term sessile serrated polyp is now preferred on the basis that the majority of sessile serrated lesions do not display any dysplasia, which is a prerequisite for pathological classification as an adenoma.5 The traditional serrated adenoma is a separate, rare, fully dysplastic lesion with classical serrated appearances.27 For the purposes of this report, the term serrated adenoma will continue to be used as the terminology employed continuously throughout the Systematic Evaluation of Aspirin and Fish Oil (seAFOod) trial from 2009 onwards.

Sessile serrated adenomas are more prevalent in the proximal (also known as right) colon (most commonly defined as proximal to the splenic flexure) than the distal (also known as left) colon.4 Conventional tubular/tubulo-villous adenomas are more uniformly distributed throughout the right and left colon.4 Molecular features also distinguish between conventional adenomas and serrated adenomas, with a high prevalence of CIN features in conventional adenomas and serrated adenomas commonly displaying a CIMP-high, BRAF mutation-positive, MSI-high phenotype.4,21,27 Results from a study29 of CRCs suggest that, in reality, there is likely to be a continuous positive gradient of CIMP-high, MSI-high and BRAF mutation frequency in tumours along the distal to proximal colon, rather than an anatomical dichotomy in the distal transverse colon. There is conflicting evidence that a given tumour genotype/phenotype predicts that of synchronous/metachronous lesions, but the majority of data pertain to CRC, not to colorectal adenomas.4

As direct precursor lesions of CRC, the removal of which is unequivocally associated with decreased future CRC risk,8,9 the colorectal adenoma is a clinically important lesion in its own right.5 It has been estimated (based on cohort prevalence studies) that approximately 1 in 10–20 colorectal adenomas may eventually acquire a malignant phenotype.5 The features associated with malignant progression are size, grade of dysplasia and ‘villousness’, namely the degree of villous histological architecture in an individual lesion.5 On the other hand, the colorectal adenoma can also be considered a biomarker of future CRC risk, regardless of its individual malignant potential.5 Both colorectal adenoma number and colorectal adenoma size are widely used as the basis for future CRC risk stratification for surveillance after colonoscopy in the UK and elsewhere in the world.30–33 A common feature of guidelines is the definition of the ‘advanced’ colorectal adenoma based on size (≥ 10 mm), with or without additional histological (e.g. grade of dysplasia, ‘villousness’) features.30,31,33

Colorectal adenoma measures

Based on widespread acceptance of the number and size of colorectal adenomas as a CRC risk biomarker, the colorectal adenoma has been used as a surrogate colonoscopic end point of reduced CRC risk in chemoprevention trials following polyp clearance at an index procedure (the ‘polyp-prevention trial’). Historically, the presence or absence of any colorectal adenoma [the so-called adenoma detection rate (ADRa)] has been employed as the primary end point in chemoprevention trials, with reliance on this binary end point reflecting the varying quality of colonoscopy between different endoscopists and susceptibility of colorectal adenoma detection to observer variation.5,34–36 However, this percentage value does not take into account any change in colorectal adenoma number (or size), unlike pre-clinical rodent studies and proof-of-concept clinical studies in FAP patients, in which lesion number and size are routinely measured.37 More recently, driven by the dramatic improvement in colonoscopy quality and quality assurance (QA) reporting,38 colorectal adenoma number has begun to be reported as a primary outcome in ‘sporadic’ polyp-prevention trials as the mean adenomas per participant (MAP).39 Population-based studies have consistently demonstrated that colorectal adenoma multiplicity predicts future CRC incidence and mortality.40–42

Candidate colorectal cancer chemoprevention agents

The existing literature on several potential CRC chemoprevention agents, including non-steroidal anti-inflammatory drugs (NSAIDs), hormone replacement therapy and micronutrients (e.g. folic acid, vitamin D), is well summarised in published reviews.34,36 The largest body of evidence supports the use of the NSAID aspirin for CRC chemoprevention.43,44

Aspirin

Observational and randomised controlled trial (RCT) data44,45 suggest primary prevention efficacy of low-dose (usually defined as < 325 mg daily) aspirin; these data are summarised in comprehensive reviews. In brief, observational follow-up studies of historical RCTs of aspirin (variable dose: 75–1000 mg daily) for antithrombotic indications have reported that aspirin reduces the risk of CRC incidence [hazard ratio (HR) 0.75, 95% confidence interval (CI) 0.56 to 0.97] and mortality, with a lag period of approximately 8 years, compatible with the long natural history of ‘sporadic’ colorectal carcinogenesis.46,47 A similar phenomenon was also observed in the Women’s Health Study,48 which is the only placebo-controlled RCT of aspirin (i.e. 100 mg of aspirin taken on alternate days) with a primary CRC prevention end point. Original 10-year follow-up did not reveal any effect on CRC incidence, but reduced CRC incidence (HR 0.80, 95% CI 0.67 to 0.97) emerged after longer follow-up.48 A consistent finding from these RCTs has been the differential effect of aspirin on proximal, as opposed to distal, CRC, with the risk reduction associated with aspirin being primarily for proximal CRC.47,48

Consistent with its role as an established biomarker of CRC risk, a random-effects meta-analysis49 of four previous polyp-prevention RCTs of aspirin (using daily doses varying from 81 mg to 325 mg), using ADRa as the primary outcome measure, reported a pooled risk ratio for ‘advanced’ colorectal neoplasm or any size of colorectal adenoma in aspirin users of 0.72 (95% CI 0.57 to 0.90) and 0.83 (95% CI 0.72 to 0.96), respectively, an effect that was already apparent at colonoscopy in the first year of follow-up in these studies (risk ratio 0.62, 95% CI 0.48 to 0.81). Three of these aspirin RCTs reported MAP data as a secondary outcome, and all demonstrated a consistent reduction in the MAP value associated with aspirin use.49

Aspirin (900 mg) has also been shown to reduce CRC (and other cancer) risk in Lynch syndrome.50 Chemopreventive efficacy of aspirin in Lynch syndrome was not associated with reduced colorectal adenoma risk during routine surveillance colonoscopy follow-up in a non-screening programme setting.51

However, despite the strength of the evidence that regular, long-term aspirin use prevents CRC, aspirin has not yet been widely adopted for primary or secondary CRC chemoprevention because of continuing uncertainty about the optimal daily dose [different trials have reported efficacy of either high- (> 300 mg) or low-dose (< 100 mg) aspirin49] and the absence of a clearly defined at-risk population in whom benefit would outweigh the small risk of gastrointestinal (GI) and intracerebral bleeding associated with aspirin.44,52,53 Nevertheless, the US Preventive Services Task Force52 has recommended low-dose aspirin use for the primary prevention of cardiovascular disease (CVD) and CRC in some adults aged 50–59 years who have a 10-year CVD risk of ≥ 10%.

Selective cyclooxygenase-2 inhibitors

Consistent with the role of cyclooxygenase (COX)-2-dependent prostaglandin (PG) E₂ in the early stages of colorectal carcinogenesis,54 selective COX-2 inhibitors (i.e. celecoxib and rofecoxib) displayed significant chemopreventive efficacy in RCTs in FAP patients (20–30% reduction in polyp number and size)55 and in individuals with previous ‘sporadic’ colorectal adenoma (risk reduction of approximately 20%).34 However, the unexpected CVD toxicity associated with prolonged selective COX-2 inhibition, which became apparent in the polyp-prevention trials, precludes a role for selective COX-2 inhibitors in primary ‘sporadic’ CRC chemoprevention.56

Omega-3 polyunsaturated fatty acids

Polyunsaturated fatty acids (PUFAs) are important components of a normal diet. Two classes of PUFAs, ω-6 and ω-3, are classified as essential in that they cannot be readily synthesised in the human body and so must be obtained from dietary sources.57 The principal bioactive ω-3 PUFAs are C20:5ω-3 [in CX:Yω-Z, X denotes the number of carbon atoms; Y denotes the number of carbon–carbon double bonds and Z denotes the carbon atom from which the first double bond starts from the ω (methyl) end], eicosapentaenoic acid (EPA) and C22:6ω-3 docosahexaenoic acid (DHA), which are found predominantly in oily, cold-water fish such as mackerel, having entered the food chain following synthesis by plankton and algae.57 In addition, C18:3ω-3 alpha-linolenic acid (ALA) is found in vegetables, but can be converted to EPA and then DHA (by a series of elongases and desaturases58) only very inefficiently (approximately 5%) in humans.59 In ‘Western’ diets, ω-6 PUFAs dominate, including C20:4ω-6 arachidonic acid (AA), which is the predominant substrate for the COX enzymes in humans.57

Anticolorectal cancer activity of eicosapentaenoic acid

Eicosapentaenoic acid is an attractive candidate as a ‘natural’ CRC chemoprevention agent based on several strands of evidence.60 There is strong pre-clinical evidence that ω-3 PUFAs have anti-CRC activity.61 However, a systematic review of epidemiological studies has not demonstrated unequivocal benefit from dietary ω-3 PUFA intake on CRC risk.62 This may be related to the methodological difficulties of measuring ω-3 PUFA or fish intake retrospectively. Alternatively, ω-3 PUFA exposure may not be sufficient for consistent anti-CRC activity in individuals consuming moderate amounts of fish (a portion of oily fish two or three times per week provides only the equivalent of approximately 500 mg per day of EPA and DHA combined). Omega-3 PUFA intake can be increased by ‘over-the-counter’ fish oil supplements, which contain a complex mix of ω-3 and ω-6 PUFAs.63 However, many of these supplements are associated with a range of minor, troublesome side effects [e.g. eructation (burping), halitosis]. In the prospective VITAL (VITamin And Lifestyle) cohort study, which has uniquely collected data on fish oil supplement use, as well as dietary fish intake, fish oil supplement users were shown to have a 49% reduced CRC risk compared with non-users, an effect primarily observed in men.64

Purified and concentrated EPA is available in several forms and pharmaceutical formulations.65,66 EPA alone (without DHA) is available as the free fatty acid (FFA), as a triglyceride (TG) conjugate (the predominant natural form of EPA) or as an ethyl ester (EE) conjugate.65 Dietary EPA-TG is converted to EPA-FFA in the small intestine by the action of pancreatic lipase, which is released in response to (particularly fatty) food intake. It is unclear which form of EPA is absorbed best from the small intestine and has maximal bioavailability, especially during prolonged use.65,66 Administration of EPA with food maximises absorption of all forms of EPA.65 A 500-mg gastroresistant capsule formulation of 99% pure EPA as the FFA has been produced (SLA Pharma AG, Liestal, Switzerland). This formulation was used for the administration of 2 g of EPA-FFA daily as four capsules in the RCT in FAP patients described below.67 Alternative formulations of purified EPA exist, including a 574-mg formulation of 90% EPA-TG (equivalent to 400 mg of EPA-FFA) in a soft gelatin capsule (Igennus Healthcare Nutrition, Cambridge, UK) that can be used to provide the equivalent 2-g daily dose of EPA-FFA in five capsules.

Eicosapentaenoic acid in all three forms (i.e. FFA, TG and EE) has been demonstrated to have chemopreventive activity in several rodent models of colorectal carcinogenesis, including azoxymethane-induced intestinal tumorigenesis and the Apc^Min/+ mouse model of FAP.61,68,69 Preliminary evidence that EPA has chemopreventive efficacy in humans was provided by two separate Phase II studies of 2 g of EPA-FFA daily in patients with previous colorectal adenoma(s), which demonstrated a significant reduction in rectal epithelial cell mitosis frequency (not observed with a 1-g daily dose), which was associated with a fivefold increase in rectal mucosal EPA content.70,71 These studies led to a Phase III RCT of the effect of 2 g of EPA-FFA daily for 6 months on rectal polyposis in patients with FAP (n = 58).67 This RCT provided the first definitive evidence of chemopreventive efficacy of EPA in humans, with a net decrease in rectal adenoma number and cumulative rectal adenoma size of 22.4% and 29.8%, respectively, between the EPA and placebo groups.67 The percentage reduction in adenomatous polyp burden was similar to that observed in FAP patients treated with celecoxib,55 a drug that was subsequently demonstrated to prevent ‘sporadic’ colorectal adenomas.34 In 2012, high dietary intake of marine-derived ω-3 PUFAs was associated with reduced colorectal adenoma risk.72 The protocol for a RCT of 2.7 g of EPA daily for prevention of rectal aberrant crypt foci was published in 2012 (UMIN000008172), but the trial has yet to report results.73

Mechanisms of the antineoplastic activity of eicosapentaenoic acid and aspirin

The precise mechanism(s) by which aspirin and EPA have anti-CRC activity is not fully understood.43,45,61 However, it is currently accepted that, even though these agents are likely to act via multiple COX-dependent and COX-independent mechanisms, modulation of COX activity plays an important role in their antineoplastic effects. EPA and, particularly, aspirin are both potent inhibitors of COX-1, but they alter COX-2 activity in different ways, leading to the production of different bioactive lipid mediators, including PGE₃ (EPA) and 15R-HETE (hydroxyeicosatetraenoic acid) (aspirin).57 There is some evidence that PGE₃ (unlike pro-tumorigenic PGE₂) has antitumorigenic activity74 and it is known that aspirin-triggered lipoxins derived from 15R-HETE have antiangiogenic properties.75

Aspirin irreversibly acetylates the COX enzymes.75 When EPA acts as a substrate for aspirin-acetylated COX-2, it leads to synthesis of 18R-hydroxyeicosapentaenoic acid (18R-HEPE), which can be converted in a 5-lipoxygenase-dependent manner to resolvin (Rv) E1.75,76 RvE1 has potent anti-inflammatory activity,76 but it is currently not known whether or not RvE1 has direct antineoplastic activity.77 Specialised pro-resolving (lipid) mediators, such as resolvins and lipoxins, including RvE1, are technically difficult to measure in biological samples and are likely to exert any biological activity at trace concentrations;77 therefore, it remains unclear whether or not sufficient quantities are generated in humans to have meaningful antineoplastic activity.78

Although RvE1 synthesis provides a hypothesis for a potential interaction between EPA and aspirin, the available clinical evidence suggests that the antiplatelet (COX-1-dependent) effects of EPA and aspirin are simply additive based on the accumulated evidence of extensive use of dual therapy in cardiology patients79 and the effects of the two agents in ex vivo human platelet aggregation studies.80,81 Colorectal carcinogenesis and atherosclerosis share common pathophysiological mechanisms and clinical risk factors, including obesity.82 As a consequence, ischaemic heart disease and stroke are common in elderly populations with colorectal neoplasia.82 Therefore, an attractive feature of CRC chemoprevention using EPA and/or aspirin is the potential for additional vascular benefit in elderly colorectal adenoma ‘formers’ at simultaneous risk of occlusive vascular events.43,79

A precision-medicine approach to colorectal cancer chemoprevention

A precision or stratified medicine approach to chemoprevention, whereby the need for chemoprevention and the use of a specific agent is determined based on an individual benefit–risk assessment, has yet to be realised.

The preliminary finding of the APACC polyp-prevention trial,83 that the pattern of COX-2 expression in an index colorectal adenoma predicted the preventative efficacy of aspirin, suggests that baseline colorectal adenoma characteristics have potential as predictive biomarkers of individual chemoprevention efficacy.

Red blood cell (RBC) membrane ω-3 PUFA levels (as a validated surrogate biomarker of ω-3 PUFA tissue exposure84,85) have been long established as a biomarker of dietary ω-3 PUFA exposure in cancer epidemiological studies.86 Between 2014 and 2016, ω-3 PUFA levels were used in RCTs of ω-3 PUFAs as a biomarker of target tissue ω-3 PUFA exposure (termed ‘bioavailability’ here), but also as a possible indicator of compliance and/or placebo group ‘contamination’ by over-the-counter (OTC) ω-3 PUFA use.67,87,88

In a RCT of EPA-FFA in patients with CRC liver metastasis, tumour EPA content predicted exploratory survival outcomes.88 However, there was no relationship between the individual rectal mucosal EPA content and the reduction in rectal polyp number in the small RCT of EPA in FAP patients.67 Therefore, there is a need for further evaluation of RBC and colorectal mucosal ω-3 PUFA levels, as well as novel biomarkers based on the mechanism of action of EPA, as predictors of individual therapeutic response.

There are no validated biomarkers of aspirin anti-CRC activity. However, all the COX-dependent lipid mediators described earlier are measurable by liquid chromatography–mass spectrometry (LC-MS)77,89 and may find utility as therapeutic biomarkers.77

Safety and tolerability of eicosapentaenoic acid and aspirin

Aspirin and ω-3 PUFAs are already used widely in patient populations, that are relevant to ‘sporadic’ CRC prevention, for prophylaxis following myocardial infarction (aspirin and ω-3 PUFAs), hypertriglyceridaemia (ω-3 PUFAs) and stroke (aspirin).79

The safety and tolerability of aspirin (≤ 325 mg daily) in previous polyp-prevention trials has been excellent.34,49 Aspirin use is associated with a dose- and age-dependent increased risk of upper GI and intracranial bleeding.45,90 Cuzick et al.53 have put forward the case for a favourable benefit–risk profile for aspirin dosing of ≤ 325 mg daily for 10 years for primary CRC (and other adenocarcinoma) prevention in average-risk individuals aged 50–65 years.

There is little doubt about the safety and tolerability of ‘nutraceutical’ forms of EPA, confirmed by vast experience of intake in healthy human populations.91–93 Gastroresistant EPA-FFA of 2 g daily has been compared with placebo for up to 6 months in four RCTs, in which tolerability has been excellent.67,70,71,87 In the RCT involving FAP patients, there was no significant excess of adverse events (AEs) in the EPA-FFA group compared with the placebo group, with only one patient withdrawing from the EPA-FFA group as a result of nausea and epigastric pain.67 In two Phase II studies of colorectal adenoma patients, there was a slight excess of mild to moderate AEs in the EPA-FFA group compared with the no-treatment70 and placebo groups.71 In the latter study,71 the GI AEs observed in the EPA-FFA 2-g daily group were not apparent in those taking 1 g of EPA-FFA daily.71 EPA-TG may be associated with fewer GI AEs, particularly diarrhoea, than EPA-FFA.66 However, a formal comparison of tolerability between different EPA formulations in a RCT has not yet been undertaken.

Although aspirin and ω-3 PUFAs share antiplatelet activity and both agents prolong bleeding time, excess bleeding episodes with their combined use have not been observed in cardiological practice, in which they are widely used together following myocardial infarction.79,94 However, clinically significant bleeding events associated with treatment with EPA either alone or in combination with aspirin have, to date, not been monitored in a RCT.

The NHS Bowel Cancer Screening Programme

The NHS Bowel Cancer Screening Programme (BCSP) in England began in 2006. It is currently based on a biennial guaiac FOBt targeted at all individuals aged 60–74 years who are covered by NHS registration data in England (the uptake, based on a returned FOBt kit, is approximately 50–60%).95 Individuals with an abnormal FOBt (≈2%) are invited for colonoscopy via a specialist screening practitioner (SSPr)-run clinic. All colonoscopy is undertaken by screening-accredited colonoscopists working within a continuous QA framework based on multiple measures, including individual caecal intubation rate, withdrawal time and ADRa.38,95 Recording of endoscopic findings and subsequent histopathological assessment is also directed by BCSP guidelines and a QA reporting system.96,97 Any abnormality detected is discussed with the patient at a SSPr follow-up clinic. Detection of a CRC (≈10%, but variable dependent on the number of prevalent vs. incident screening investigations) prompts further management by the local multidisciplinary team (MDT) for CRC. Detection of one or more colorectal adenomas prompts surveillance colonoscopy within the BCSP, as per British Society of Gastroenterology guidelines.30 Individuals classified as being at ‘low risk’ (i.e. those having one or two subcentimetre colorectal adenomas) are not offered colonoscopy, but remain in the biennial guaiac FOBt programme. Those individuals with three or four small colorectal adenomas (i.e. < 10 mm in size) or one colorectal adenoma of ≥ 10 mm in diameter are classified as being at ‘intermediate risk’ and are offered another colonoscopy at 3 years from the index procedure. Individuals with five or more subcentimetre colorectal adenomas, or three colorectal adenomas with at least one colorectal adenoma of ≥ 10 mm in diameter (i.e. 12% of men and 6.2% of women who undergo screening colonoscopy), are recommended to undergo surveillance colonoscopy 12 months from the screening colonoscopy.98

Since 2013, the bowel scope programme has been rolled out across England, whereby, in addition to the biennial guaiac FOBt invitation, a single FS is offered to all individuals aged 55 years.99 The presence of a colorectal adenoma of ≥ 10 mm in diameter, three or more small (i.e. < 10 mm) colorectal adenomas or any adenoma with ‘advanced’ features prompts an invitation for full colonoscopic evaluation, with the combined colorectal adenoma findings from the FS and colonoscopy directing the subsequent surveillance strategy within the BCSP, as described above.95

The seAFOod polyp-prevention trial

Based on strong proof of concept for primary CRC chemoprevention activity of EPA67 and aspirin,49 the National Institute for Health Research (NIHR)/Medical Research Council (MRC) Efficacy and Mechanism Evaluation (EME) programme funded a 2 × 2 factorial RCT of 2 g of EPA-FFA daily and/or 300 mg of aspirin in ‘high-risk’ individuals identified in the English BCSP. The trial was termed the seAFOod polyp-prevention trial.100

Main research question

Does the ω-3 PUFA EPA prevent colorectal adenomas, either alone or in combination with aspirin?

Objectives

Primary objective

The primary objective was to determine whether or not EPA prevents colorectal adenomas, either alone or in combination with aspirin. This was addressed by testing the following hypotheses:

• 2 g of EPA-FFA daily is more effective than placebo for reduction in colorectal adenoma recurrence.
• 300 mg of aspirin daily is more effective than placebo for reduction in colorectal adenoma recurrence.

Secondary objectives

The secondary objectives were to assess the tolerability and safety of EPA-FFA and EPA-TG alone and in combination with aspirin.