Conclusions

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/).

Implications for health care

Although there was no difference in the ADRa across treatment groups in the seAFOod trial, the decrease in total colorectal adenoma number associated with aspirin and its selectivity for proximal colorectal adenomas and serrated polyps adds to a substantial body of evidence that, although aspirin use is associated with overall CRC risk reduction in unselected groups,46–48 there are patient groups for whom, and clinical situations in which, additional benefit from aspirin chemoprevention might be gained. For example, these data are the first randomised trial data to bolster the argument that individuals with the rare, but increasingly recognised (in screening programmes), hyperplastic-serrated polyposis syndrome might benefit from aspirin prophylaxis, as well as those with ‘sporadic’ serrated polyps.154 Moreover, further evidence that there is proximal colonic selectivity for aspirin CRC chemoprevention highlights potential uses of aspirin in concert with FS screening and for reduction of clinically important PCCRCs, which are more commonly proximal than screen-detected CRCs.15

In the absence of any effect of 2 g of EPA daily on the total colorectal ADRa in the seAFOod trial and numerically small effects on colorectal number (MAP) and incidence rate (ADRa) in the left/distal colorectum only, no firm guidance about ‘nutraceutical’ EPA use, dietary ω-3 PUFA intake or use of RBC ω-3 PUFA as a predictive risk biomarker, in the context of CRC risk reduction, can be made prior to further exploratory, hypothesis-generating analysis of the seAFOod trial data and biobank samples (see Recommendations for research). However, even small chemoprevention benefit from EPA is likely to be accepted and taken up by a significant proportion of the public, who might consider that ω-3 PUFAs have wider health benefits.

Recommendations for research

The seAFOod trial data available to date will stimulate a debate about how the colorectal adenoma is best used as an ‘efficacy’ RCT end point. The colorectal adenoma can be viewed solely as an endoscopic biomarker of carcinogenic potential, and hence CRC risk, but also as a clinically important lesion, the presence of which prompts endoscopic removal and stratification for future endoscopy, with implications for endoscopy provision. This distinction is key to how CRC chemoprevention is viewed: as a long-term strategy for reduction in CRC risk and/or as adjuvant therapy for those in endoscopic screening/surveillance programmes with a view to decreasing need for colonoscopy. The lack of effect on total ADRa of either EPA or aspirin (and absence of any short-term reduction in colonoscopy frequency with no shift from ‘high’ to ‘intermediate’ risk classification) argues against a role as adjuvant therapy for those in endoscopic screening/surveillance programmes. By contrast, a reduction in MAP suggests chemopreventive activity that may manifest eventually as reduced CRC incidence. It will be important to determine long-term colonoscopy and CRC outcomes in the seAFOod trial participants using routinely collected BCSP data and National Cancer Registration and Analysis Service (NCRAS) data. Trial participants have already provided consent for use of long-term, post-trial data. Although the small sample size of the seAFOod trial cohort is likely to provide insufficient CRC diagnoses to provide a meaningful comparison, analysis of subsequent colonoscopy frequency and colorectal adenoma outcomes should be able to test whether or not short-term treatment provides longer-term chemoprevention of colorectal adenomas (with the weakness that ongoing aspirin or ω-3 PUFA use will not be known). An Office for Data Release request to Public Health England will be made prior to gaining approval from the BCSP Research Advisory Committee for this follow-up study.

Future ‘effectiveness’ evaluation of the two seAFOod trial interventions will be dependent on whether or not colorectal adenoma number (MAP) is used as a CRC risk biomarker. Including the seAFOod trial, there are now four independent RCTs (Table 39) that have reported on colorectal adenoma number associated with aspirin use.115,117,118 A key piece of work will be to attempt a meta-analysis of the colorectal adenoma number data from these trials to derive an overall colorectal adenoma risk reduction ratio for aspirin use that can be aligned with the observational data on CRC risk reduction, to propose a clinically meaningful MAP reduction value.

The observation that both EPA and aspirin have selectivity for CRC chemopreventive activity based on colorectal adenoma type and location prompts the question as to whether or not a precision-stratified-medicine approach to CRC chemoprevention can be taken whereby baseline colorectal adenoma features can be used to predict the type and/or location of colorectal adenoma recurrence and, therefore, best ‘personalised’ chemoprevention with a given chemoprevention agent. The detailed trial data set on endoscopic and histological features of baseline colorectal adenomas can be used to investigate the relationship between baseline colorectal adenoma features (location, type, number) and recurrence in the placebo-only group of the seAFOod trial. Other post-polypectomy tissue biomarkers may have utility for prediction of aspirin and EPA efficacy, such as tissue COX-2 expression.83 A recent study has demonstrated that COX-2 and 15-prostaglandin dehydrogenase (PGDH) expression (combined to provide a surrogate marker of tissue PGE₂ content) in baseline colorectal adenomas predicts the chemopreventive efficacy of the selective COX-2 inhibitor celecoxib,155 building on the previous data from the APACC trial.83 To this end, formalin-fixed, paraffin-embedded (FFPE) polypectomy specimens from the screening colonoscopy from the seAFOod trial placebo group can be obtained, as detailed in the trial protocol.100 Other, novel, tissue biomarkers based on the proposed mechanism of action of the type and location specificity of EPA and aspirin can also be tested, for example FFAR2 protein expression in FFPE polypectomy specimens, but also in the non-neoplastic rectal mucosal samples available in the seAFOod trial biobank. Rectal mucosal samples will be invaluable for investigation of any ‘field effect’ mucosal biomarker with predictive value for colorectal adenoma type recurrence.

The intriguing observation that combined EPA and aspirin use in the seAFOod trial was associated with fewest total and left-sided colorectal adenomas, when comparing across the four treatment groups, suggests that the two agents might have an additive relationship for CRC chemoprevention (with the additional, beneficial feature that aspirin co-therapy may improve tolerability of EPA treatment). In the first instance, an exploratory analysis of the predictive value of the baseline/post-intervention RBC and/or post-intervention rectal mucosal EPA level for the reduction in total and left MAP associated with aspirin should be performed in order to strengthen the case for an ‘efficacy’ RCT of combined aspirin–EPA (or mixed ω-3 PUFA) treatment or stratified (by baseline EPA level) aspirin use.

Two large RCTs of mixed ω-3 PUFA treatment are due to report in the near future. The ASCEND (A Study of Cardiovascular Events iN Diabetes) trial (NCT00135226)156 is a 2 × 2 factorial study of long-term (median 7.5 years) ω-3 PUFAs (840 mg of EPA/DHA EEs daily) and aspirin (100 mg daily) treatment for prevention of cardiovascular and cerebrovascular events in patients with diabetes (n = 15,480). Cancer outcomes are a secondary end point with the ability to continue with post-trial follow-up. VITAL (VITamin D and OmegA 3 TriaL) (NCT01169259) is a 2 × 2 factorial study of the same dose and formulation of ω-3 PUFAs (840 mg of EPA/DHA EEs) and vitamin D₃ (2000 IU daily) in 25,871 participants.157 Colorectal cancer is a specified secondary outcome. It will be very interesting to interpret the results of the ASCEND trial and VITAL in the light of the seAFOod trial data, particularly in relation to CRC location. It should be noted that both trials tested a smaller dose of mixed ω-3 PUFAs than the daily dose of EPA employed in the seAFOod trial.

The data on recruitment performance and placebo group outcomes will be invaluable for any future CRC chemoprevention RCT set in the BCSP. A key lesson is that the ADRa seems to lack discriminatory power, at least in the ‘high-risk’ colorectal adenoma cohort investigated in the seAFOod trial, given that the intervention with known chemopreventive efficacy (aspirin) did not demonstrate an ADRa signal.

As discussed earlier, the comprehensive data set of PUFA levels in RBCs and rectal mucosa at baseline and after ‘nutraceutical’ EPA intervention should prompt an analysis of the predictive value of individual ‘EPA status’ at baseline and after intervention, independent of allocation to active EPA or placebo, for colorectal adenoma reduction. An understanding of whether or not ‘EPA high’ status is associated with reduced colorectal adenoma recurrence, regardless of whether or not EPA supplementation was provided, may have important consequences for dietary ω-3 PUFA guidelines and possible measurement of ω-3 PUFA status for CRC risk stratification.

The seAFOod trial biobank is also a rich resource of pre- and post-treatment plasma and urine samples, with which to extend and strengthen the tissue biomarker analysis. For example, parallel measurement of urinary PGE-M levels (the main metabolite of PGE₂) may provide further insight into the use of baseline colorectal adenoma COX-2/15-PGDH/PGE₂ status127 and was originally specified in the trial protocol.100 The possible additive relationship between EPA and aspirin observed for total and left MAP justifies the use of the biobank repository to address the hypothesis that RvE1 is measurable in plasma samples from individuals treated with EPA, with and without aspirin, as described in the trial protocol.100

The seAFOod trial data should drive several studies aimed at understanding the mechanism(s) of action of both interventions. Differential preventative activity against conventional and serrated pathway lesions by EPA and aspirin has highlighted the relatively poor characterisation of the molecular phenotype of colorectal adenomas compared with CRC,4,154 as well as the continuing lack of understanding of mechanism(s) of action of both agents. Further in vitro studies (using human CRC cells) of aspirin and EPA should use representative cellular models of CIN, MSI-H and CIMP+ phenotypes.158 The tumour immunology of the colorectal adenoma has been largely ignored, perhaps because of the relative difficulty in collecting sufficient, fresh human colorectal adenoma tissue and absence of a suitable animal model.159 However, regulation of host immune surveillance as a mechanism of primary CRC chemoprevention activity remains a valid hypothesis for both seAFOod trial interventions.160 We have recently reported that mixed ω-3 PUFA treatment is associated with an increase in abundance of SCFA-producing bacteria in the colon.129 The hypothesis that the propensity for EPA benefit in the distal colorectum relates to increased anticarcinogenic SCFA-FFAR2 signalling132 should now be tested by measuring stool SCFA concentrations in human studies, in parallel with FFAR2 expression studies using rectal mucosa from the seAFOod trial biobank.

The seAFOod trial biobank contains a blood leucocyte sample for genomic DNA extraction for nearly all participants, as well as individual participant consent to perform genetic single nucleotide polymorphism studies relevant to CRC risk and both interventions. The results of the ω-3 PUFA biomarker analyses above will generate hypotheses about which genetic polymorphisms that are known to control ω-3 PUFA levels,161 for example fatty acid desaturase (FADS) genes, should be characterised in parallel with the dietary analysis.