Post‐diagnosis dietary factors, supplement use and colorectal cancer prognosis: A Global Cancer Update Programme (CUP Global) systematic literature review and meta‐analysis

The role of diet in colorectal cancer prognosis is not well understood and specific lifestyle recommendations are lacking. We searched for randomised controlled trials (RCTs) and longitudinal observational studies on post‐diagnosis dietary factors, supplement use and colorectal cancer survival outcomes in PubMed and Embase from inception until 28th February 2022. Random‐effects dose–response meta‐analyses were conducted when at least three studies had sufficient information. The evidence was interpreted and graded by the CUP Global independent Expert Committee on Cancer Survivorship and Expert Panel. Five RCTs and 35 observational studies were included (30,242 cases, over 8700 all‐cause and 2100 colorectal cancer deaths, 3700 progression, recurrence, or disease‐free events). Meta‐analyses, including 3–10 observational studies each, were conducted for: whole grains, nuts/peanuts, red and processed meat, dairy products, sugary drinks, artificially sweetened beverages, coffee, alcohol, dietary glycaemic load/index, insulin load/index, marine omega‐3 polyunsaturated fatty acids, supplemental calcium, circulating 25‐hydroxyvitamin D (25[OH]D) and all‐cause mortality; for alcohol, supplemental calcium, circulating 25(OH)D and colorectal cancer‐specific mortality; and for circulating 25(OH)D and recurrence/disease‐free survival. The overall evidence was graded as ‘limited’. The inverse associations between healthy dietary and/or lifestyle patterns (including diets that comprised plant‐based foods), whole grains, total, caffeinated, or decaffeinated coffee and all‐cause mortality and the positive associations between unhealthy dietary patterns, sugary drinks and all‐cause mortality provided ‘limited—suggestive’ evidence. All other exposure‐outcome associations provided ‘limited—no conclusion’ evidence. Additional, well‐conducted cohort studies and carefully designed RCTs are needed to develop specific lifestyle recommendations for colorectal cancer survivors.

polyunsaturated fatty acids, supplemental calcium, circulating 25-hydroxyvitamin D (25[OH]D) and all-cause mortality; for alcohol, supplemental calcium, circulating 25(OH)D and colorectal cancer-specific mortality; and for circulating 25(OH)D and recurrence/disease-free survival.The overall evidence was graded as 'limited'.The inverse associations between healthy dietary and/or lifestyle patterns (including diets that comprised plant-based foods), whole grains, total, caffeinated, or decaffeinated coffee and all-cause mortality and the positive associations between unhealthy dietary patterns, sugary drinks and all-cause mortality provided 'limited-suggestive' evidence.All other exposure-outcome associations provided 'limited-no conclusion' evidence.Additional, well-conducted cohort studies and carefully designed RCTs are needed to develop specific lifestyle recommendations for colorectal cancer survivors.
colorectal cancer survival, diet, evidence grading, food, systematic review What's new?
The role of diet in colorectal cancer prognosis is not well understood.As part of CUP Global, here the authors systematically reviewed, meta-analysed, and independently graded the quality of evidence on the associations of post-diagnosis dietary intake, dietary patterns, and supplement use with colorectal cancer prognosis.Whilst the overall evidence was graded as 'limited', it suggested that consuming a healthy diet, including diet patterns with plant-based foods, and avoiding sugary drinks may be associated with improved overall survival after colorectal cancer diagnosis.The study calls for well-designed cohort and intervention studies to help develop lifestyle recommendations for colorectal cancer survivors.

| INTRODUCTION
Colorectal cancer is the third most diagnosed malignancy after lung and breast cancer, 1,2 and ranks second in terms of cancer-related deaths in men and women worldwide. 1Globally, in 2020, there were more than 1.9 million colorectal cancer cases and more than 0.9 million colorectal cancer deaths. 3By 2040, incident colorectal cancer cases and colorectal cancer-related deaths are expected to reach 3.2 and 1.6 million, respectively. 1 Possible reasons for the rise in cases include adoption of a westernised diet and lifestyle, population growth and extended lifespan. 4,5lorectal cancer has an overall 5-year relative survival rate of at least 60% in higher income countries, 6 and a lower rate, on average 40%, in lower-income settings. 7,8Advancements in detection 9,10 and treatment services 11 have led to greater numbers of colorectal cancer survivors, especially in most developed countries. 6,12ere were an estimated 5.2 million colorectal cancer survivors living within 5 years of diagnosis in 2020. 3Extended survival, however, co-exists with increased co-morbidities, 13 including cardiovascular disease, the most common cause of non-cancer deaths. 14Colorectal cancer survivors are at risk of recurrence, metastasis 15,16 or second primary cancers. 17,18Colorectal cancer will continue posing an enormous global health burden and financial challenge across health systems. 4,5evention strategies through lifestyle modifications may improve cancer survivorship.Numerous lifestyle factors have been identified for their involvement in colorectal cancer development.The World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) Third Expert Report identified 'strong-probable' evidence that consuming red meat increases, and consuming whole grains, dietary fibre, dairy products, and calcium supplements reduces the risk of colorectal cancer and 'strong-convincing' evidence that consumption of processed meat and alcohol increases risk.Evidence on other dietary factors was limited. 19A recent umbrella review also reported strong meta-analytic evidence for alcohol intake and higher risk of colorectal cancer and for calcium, whole grain, and dairy product intake and lower risk of colorectal cancer. 20However, evidence on how dietary factors could influence survival is currently limited.Such knowledge is essential to develop targeted dietary guidance/ recommendations for colorectal cancer survivors.Currently, cancer survivors are advised to follow general cancer and chronic disease prevention guidelines. 19,213][24][25] Categorical meta-analyses showed inverse associations between post-diagnosis whole grain 23 and coffee 23 intake, the American Cancer Society recommendations (ACS-score) 23 and all-cause mortality, between a prudent dietary pattern, 23 calcium supplementation 25 and colorectal cancer-specific mortality.Positive associations were observed for an unhealthy dietary pattern and all-cause and colorectal cancer-specific mortality as well as for the Dietary Inflammatory Index 23 and all-cause mortality.
The most recent meta-analysis 26 of circulating 25(OH)D concentrations and colorectal cancer outcomes reported inverse associations for all-cause and colorectal cancer-specific mortality but included a mixture of studies with pre-or post-diagnosis 25(OH)D assessment.
A meta-analysis of two randomised controlled trials (RCTs) on vitamin D3 supplementation reported a favourable effect on colorectal cancer progression or death. 27Most of the meta-analyses on dietary factors to date focused on categorical analyses and did not explore nonlinearity.As part of the WCRF Global Cancer Update Programme (CUP Global), formerly known as the WCRF/AICR Continuous Update Project, we conducted comprehensive systematic literature reviews (SLRs) and meta-analyses to evaluate the magnitude and the shape of the associations of interest.The evidence was subsequently independently interpreted and graded by the CUP Global Expert Committee and Expert Panel. 28This paper presents the evidence on postdiagnosis dietary factors, supplement use and colorectal cancer outcomes.[31]

| METHODS
The present systematic review was conducted following a standard prepublished CUP Global protocol. 32Details on the methods and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist are available in Supplementary Text S1 and Table S1.

| Search strategy, selection criteria and data extraction
We searched in PubMed and Embase from inception to 28th February 2022, and screened through relevant reference lists, for publications that met the following inclusion criteria: (1) RCTs with at least 6 months duration, longitudinal observational studies, or pooled analyses of the aforementioned designs that (2) included at least 100 participants; (3) examined dietary exposures which were assessed at/after diagnosis (in a few studies diet was recalled shortly after diagnosis in some of their participants and the assessment may have included the pre-diagnosis period [33][34][35][36][37] ) that is, dietary patterns, foods, circulating 25(OH)D, beverages, macro-and micronutrients and dietary supplements (any vitamins, minerals, herbs or other botanicals, amino acids, or other dietary substances or their constituents 38 ).All exposures are being referred to as 'post-diagnosis' in the current manuscript for brevity.The examined colorectal cancer outcomes were all-cause mortality, cause-specific mortality, progression/recurrence/disease-free survival, any second primary cancers.In the case of multiple publications by the same study on the same exposureoutcome association, the publication with the greater number of outcome events was used.Studies that examined solely nutrientbased patterns or were lacking information on the foods and beverages contributing to the dietary pattern were excluded because the information could not be used towards the development of foodbased dietary recommendations.Study and participants' characteristics and the results for each exposure-outcome association were extracted into the CUP Global database.

| Systematic literature review
RCTs and observational studies were reviewed separately.Relevant publications were meta-analysed when at least three studies were available for a given exposure, or descriptively synthesised for the studies that did not report sufficient information for a meta-analysis.
The RCTs and the publications on dietary exposures related to the WCRF/AICR Cancer Prevention Recommendations 19 were descriptively synthesised even when there were fewer than three available studies.For descriptive synthesis, the results for each exposure-outcome association reported in the individual studies were summarised in text and the relative risks (RRs) comparing extreme exposure categories were presented in forest plots.The various dietary and/or lifestyle patterns were grouped into 'healthy' and 'unhealthy' to explore if there was any tendency of associations.

| Statistical methods for meta-analysis
Linear dose-response meta-analyses were conducted to calculate summary RRs and 95% confidence intervals (CIs) from multivariable adjusted results for a continuous exposure or for exposures with at least three categories reported in the individual studies.A Der Simonian-Laird random-effects model was used. 39Heterogeneity was assessed using the estimate of between study variance (tau 2 ) 40 and reflected by the range of effect estimates provided in the forest plots.
The proportion of total variability in effect estimates due to betweenstudy heterogeneity was assessed using the I 2 statistic. 41The 95% predictive intervals (PIs) were used to estimate the range of values that may contain the value of a new study. 42Sources of heterogeneity were explored when there were at least three studies in pre-defined subgroups (cancer subsite and subtype, exposure assessment timeframe relative to cancer treatment, and risk of bias domains).Smallstudy effects, a reason of which is publication bias, were examined using the Egger's test 43 and by visually inspecting funnel plots with 10 or more studies.One-stage non-linear dose-response metaanalysis was conducted using restricted cubic splines with three knots placed at 10th, 50th, and 90th percentiles of the overall dose distribution (based on the estimated or reported category midpoints of the included studies). 44This was done when at least five studies with three or more exposure categories were available.The likelihood ratio test was used to compare the linear and non-linear models. 45Sensitivity meta-analyses, including leave-one-out analysis to examine the influence from each study on the summary estimate 46 and analyses excluding locally advanced and metastatic patients were conducted when appropriate.Hypothesis testing and p-values reported are twotailed, unless otherwise mentioned.

| Risk of bias assessment
Version 2 of the Cochrane risk-of-bias tool for randomised trials (RoB 2) was used to assess the risk of bias in individual RCTs, 47 49 The Imperial College London (ICL) review team further refined and tested the tool to ensure its suitability for investigating exposure-outcome associations in cancer survivorship studies.This involved adapting the tool's prompting questions and providing additional guidance to encompass adiposity, physical activity, and dietary/nutritional exposures (working document version dated 11/07/2023 can be found in Supplementary Table S2).The tool consists of seven domains, including confounding, participant selection, exposure classification, departures from intended exposures, missing data, outcome measurement, and selective reporting.The studies not included in the meta-analyses were assessed descriptively considering the most likely influential sources of potential bias in survival studies 50 (selection bias, information bias of exposure and outcome assessment, and residual confounding).

| Quality control
Study selection, data extraction, and risk of bias assessment were checked by a second reviewer.Any disagreements were resolved by consensus with a third reviewer.

| Evidence grading criteria
The findings from the systematic review were interpreted by the CUP Global independent Expert Committee on Cancer Survivorship and the Expert Panel convened by WCRF International.The Expert Committee provided preliminary conclusions on the evidence, and these were finalised by the Expert Panel independently of the ICL review team.Pre-defined evidence grading criteria, covering the quantity, consistency, magnitude and precision of the summary estimates, existence of a dose-response relationship, risk of bias, study design, generalisability and mechanistic plausibility of the results, were used to assess the quality of the evidence evaluating likelihood of causality into 'strong (subgrades: convincing, probable, substantial effect on risk unlikely)' or 'limited (subgrades: limited-suggestive or limited-no conclusion)' (Supplementary Table S3).

| Observational studies
Evidence on dietary and/or lifestyle patterns and colorectal cancer outcomes were grouped to observe any tendency of associations (Supplementary Tables S5 and S6).Linear dose-response meta-analyses were possible for 21 dietary exposures-colorectal cancer survival outcome associations investigated (Figure 2).All other associations identified were descriptively reviewed.Subgroup analysis, sensitivity analysis, and test of publication bias was not possible, except for the few occasions as presented.Study and participants' characteristics and the main results of the studies included are provided in Supplementary Tables S7-S41.An overview of the risk of bias assessment of publications from observational studies is provided in Supplementary Figures S3 and S4.A summary of the evidence grading conclusions is provided in Table 2.

| Post-diagnosis dietary and lifestyle patterns combined
Two pooled analyses of two cohorts 80,81 and two additional studies 86,90 (4 publications) (1187 deaths) investigated four a priori healthy dietary and lifestyle patterns, defined based on recommendations for cancer prevention or a healthy lifestyle, which included diet, physical activity, and adiposity as components (Supplementary Tables S7-S9, Figures S5 and S6).One pooled analysis investigated three patterns, 81 other studies 80,86,90 investigated one pattern only, thus the four independent study groups reported six dietary and lifestyle patterns and all-cause mortality point estimates (Supplementary Table S5).The evidence was descriptively reviewed.There was a consistent trend for an inverse association between healthy dietary and lifestyle patterns and all-cause mortality across the studies (HRs for high vs. low scores ranged from 0.58 to 0.80, 3 out of 5 CIs crossing the null), apart from one study that indicated a positive association (HR = 1.19, 95% CI = 0.59-2.43)but with wide CIs crossing the null 90 (Figure 3).
Post-diagnosis red and processed meat, red meat, and processed meat intake Three studies (4 publications) were identified investigating postdiagnosis intakes of red and processed meat, 86,89,98 (unprocessed) red meat 97 and processed meat 97 (Supplementary Table S24).Linear dose-response meta-analysis was possible for studies that investigated post-diagnosis red and processed meat intake.No association was observed with all-cause mortality (summary RR per 1 serving/day = 0.87, 95% CI = 0.67-1.12;S19).
Note: Potential sources of heterogeneity and publication bias were not assessed because of the low number of RCTs and observational studies, except for the analyses of circulating 25-hydroxyvitamin D and all-cause and colorectal cancer-specific mortality, where between-study heterogeneity was partly reduced when advanced/metastatic cancer survivors were excluded, when possible, and for the analysis of circulating 25-hydroxyvitamin D and all-cause mortality, where there was no indication of small-study effects such as publication bias.The observational studies were at risk of bias in various methodological quality, mainly related to residual confounding, differential selection of participants into the study, exposure misclassification, reverse causation, and potential departure from intended exposure.Main concerns that impact the evidence grading were highlighted in the table.Biological mechanisms were not the decisive factor contributing to the present evidence conclusions except otherwise shown.Abbreviations: ACM, all-cause mortality; CCSM, colorectal cancer-specific mortality; CI, confidence interval; CVDM, cardiovascular disease mortality; DFS, disease-free survival; HR, hazard ratio; RR, relative risk.
a One trial of each on nutritional behavioural intervention and high-protein supplementation, omega-3 fatty acid supplementation, and vitamin C supplementation were identified.The trials conducted planned or post-hoc analyses investigating colorectal outcomes as either primary or secondary outcomes.The trials were of some concerns to high risk of bias and were judged as limited-no conclusion evidence.

Post-diagnosis sugary drink, and artificially sweetened drink intake
The four identified studies on post-diagnosis sugary drinks (3 publications) 92,103,104 and the three on post-diagnosis artificially sweetened drinks (2 publications) 104,105 (Supplementary Table S26 and Figures S24-S27) were included in the linear dose-response meta-analyses.

Post-diagnosis alcohol intake
Ten studies (8 publications) 81,86,95,[110][111][112][113][114] were identified (Supplementary Table S28).Two studies lacked sufficient information for inclusion in the meta-analysis 112,114 (Supplementary Figure S32).Seven studies Relative Risk (RR) F I G U R E 4 Linear dose-response meta-analyses of post-diagnosis intake of whole grains, sugary drinks, and total, caffeinated, and decaffeinated coffee in relation to all-cause mortality (associations presented are those graded as 'limited-suggestive').Forest plot shows the linear dose-response results (per 1 serving/day intake of whole grains and sugary drinks; and per 1 cup/day intake of coffee) from the inverse variance DerSimonian-Laird random-effects model.Each diamond represents the summary relative risk (RR) estimate, the diamond's width is the 95% confidence interval (CI), and the diamond's horizontal line is the 95% prediction interval (PI).Each square and the horizontal line across the square represents the RR estimate of the individual study and its 95% CI. (5 publications) 81,86,95,110,111 (Supplementary Figures S33 and S34) were included in the linear and non-linear dose-response meta-analyses of post-diagnosis alcohol intake and all-cause mortality and three (2 publications) 110,111 in the linear dose-response meta-analysis of colorectal cancer-specific mortality.There was an indication of an inverse association between post-diagnosis alcohol and all-cause mortality (summary RR per 10 g/day = 0.95, 95% CI = 0.91-1.00;I 2 = 0%, tau 2 =0.00,RRs range = 0.91-1.06),and some indication of a lower risk of all-cause mortality with alcohol consumption up to 20 g/day (P non-linearity = 0.23) (2122 deaths), but with the 95% CIs crossing the null value.There was no association between post-diagnosis alcohol and colorectal cancer-specific mortality (summary RR per 10 g/day = 0.86, 95% CI = 0.57-1.30; (Supplementary Figures S35 and S36).

Post-diagnosis dietary glycaemic load, glycaemic index, insulin load, and insulin index
The pooled analysis of two cohorts 117 and one additional study (2 publications) 116,118 were included in the linear dose-response metaanalyses of dietary glycaemic load, glycaemic index, insulin load, and insulin index and all-cause mortality (Supplementary Table S29 and Figures S37-S40).

Post-diagnosis dietary fat intake
Five studies (7 publications) were identified investigating postdiagnosis intakes of total dietary fat, 115,119 specific dietary fatty acids, 89,99,100,115,119,120 ratio of polyunsaturated-to-saturated fat, 89 and fat from animal or plant sources 92,115,119 (Supplementary Table S30).
Between-study heterogeneity was partly reduced when advanced/ metastatic cancer survivors were excluded, when possible, from the analyses (Supplementary Figure S52).There was no indication of small-study effects such as publication bias in the analysis of all-cause mortality (Egger's test p = .76)(Supplementary Figure S53).Non-linear doseresponse meta-analysis indicated a higher risk of all-cause mortality at lower levels of circulating 25(OH)D and the magnitude of risk gradually decreased with higher levels of 25(OH)D up to 70 nmol/L and remained constant thereafter (p non-linearity = .001)(9 studies, 7 publications) (3556 deaths 35,125,131,[134][135][136][137] ).For recurrence/disease-free survival, there was little evidence of non-linearity, with wide 95% CI and limited data (p non-linearity = .14)(5 studies, 4 publications) (2442 events) 35,131,132,136 (Supplementary Figures S49 and S51).

| Other descriptive reviews
The

| Risk of bias assessment
Three 76,78,79 of the five identified RCTs were overall at high risk of bias, due to unblinded investigator and outcome assessor, 76,78 and the inability to fully blind for whole food regimens 79 (Supplementary Figures S1 and S2).Across the 46 observational study publications that investigated dietary and/or lifestyle patterns and other exposures included in the meta-analyses (Supplementary Figures S3 and S4), many had some degree of incomplete adjustment, mainly for critically important confounders such as stage and/or treatment (47% moderate, 9% serious, 45% critical risk of bias).Three publications (6% moderate risk of bias) 94,108,125 84,96 had critical risk of bias due to selective reporting.The few publications at critical risk of bias in any domain (Supplementary Figures S5, S7, S9, S12, S14, S17, S21, S24, S26, S29, S33, S37, S39, S43, S47, and S55) generally showed similar RR estimates and overlapping 95% CIs compared to publications with less than critical risk of bias.For circulating 25(OH)D, it was possible to stratify the studies by moderate or critical risk of bias from confounding.The summary RRs were similar in direction and magnitude (Supplementary Figure S79).

| Overall evidence grading
Taking together the evidence from RCTs and observational studies, eight associations received 'limited-suggestive' evidence grading in relation to all-cause mortality, namely inverse associations for healthy dietary and lifestyle patterns combined, healthy dietary patterns, whole grains, total, caffeinated, or decaffeinated coffee, and the positive associations for unhealthy dietary patterns, and sugary drinks.All other associations received 'limited-no conclusion' evidence grading.
More information on the associations and the reasoning behind the evidence grading is provided in Table 2. 'Unhealthy' dietary patterns comprise mostly high intakes of refined grains, red and processed meat, sugary drinks, and low coffee intakes.

| DISCUSSION
The beneficial or detrimental association with these respective patterns could be partly explained by the individual, cumulative or synergistic effect of these components, as for other chronic diseases. 138,139udies that included 'healthy body weight' and high physical activity as additional components of healthy dietary and lifestyle patterns, reported inverse associations with all-cause mortality but the point estimates were similar to patterns based solely on dietary components.Studies have suggested that the association between healthy patterns and lower risk of all-cause mortality could be attributed to either the dietary and other lifestyle components 81,86 or driven primarily by physical activity in the patterns. 80The cut-point and the corresponding scoring for 'healthy body weight' may affect the risk estimation. 80,86In the present series of reviews, we found an inverted J-shaped relationship between body mass index and all-cause mortality.The lowest risk was observed at 28 kg/m 2 , 29 which is different to the conventional healthy body weight of 18.5-24.9kg/m 2 .1][142][143][144][145] Individuals (healthy or with chronic diseases) consuming a 'Western' or unhealthy diet generally have higher levels of inflammation and hyperinsulinemia. 140,143,146,147Similarly, in colorectal cancer patients, a Westernised diet and high glycaemic load can lead to chronically elevated insulin levels that could facilitate tumour recurrence, micro-metastasis or development of co-morbidities (e.g., cardiovascular diseases) and higher risk of mortality. 148Molecular epidemiology studies are required to elucidate the relevant proposed mechanisms 148 and clarify the influence of what dietary factors or modifications on risk of cancer recurrence and survival. 149Future studies should examine if associations of diet vary according to disease stage or treatment phase 149 to establish actionable lifestyle factors that can impact treatment response and/or survival. 1501][152] Any relevant markers with prognostic value could then be incorporated in trials with surrogate disease end points. 138Evidence from Mendelian randomisation (MR) studies could assist in prioritising certain nutritional interventions that are more likely to reduce cancer progression. 153gher intakes of individual food items, such as whole grains and coffee, were associated with a lower risk of all-cause mortality in this review.Whole grain intake was found to lower the risk of colorectal cancer incidence and all-cause mortality in general population studies. 19,154Whole grains are high in dietary fibre, which increases faecal bulk and decreases transit time, minimising exposure to intestinal carcinogens. 155Whole grains may be fermented by gut microbes 156 into short-chain fatty acids that could facilitate normal colonocyte growth and induce tumour cell apoptosis. 157Clinical intervention studies in healthy individuals or with various chronic diseases found that substituting refined grains with whole grains reduced inflammation 158 and insulin resistance, 159 possibly due to beneficial phytochemical constituents.Future studies should evaluate whole grains as a proportion of total grain consumption or a ratio to refined grain intake, in addition to overall intake.Coffee contains various bioactive phytochemicals that have antioxidant, anti-inflammatory, insulin-sensitising, and anti-tumour properties. 160,1613][164] The association with coffee was unlikely attributed to its caffeine content nor the avoidance of consuming caffeinated drinks after diagnosis and any potential influence from reverse causation, since the present meta-analyses and sensitivity analyses showed consistent inverse associations with allcause mortality, overall and both for caffeinated and decaffeinated coffee.The pooled analysis of two cohorts observed an inverse association between coffee and all-cause mortality among stage III but not among stage I-II cancer survivors when the analysis was stratified by cancer stage. 108The authors could not rule out chance findings due to multiple comparisons for the stratified analysis, nevertheless, the result was in line with our findings contributed primarily by studies of advanced/metastatic/stage III disease survivors. 107,109Liver metastases are common in colorectal cancer. 165An inverse association between coffee consumption and liver cancer development has been observed. 19Coffee consumption could improve survival by reducing risk of liver metastases, but additional studies are required.
Higher sugary drink intake was associated with higher risk of allcause mortality.The finding is supported by plausible influences of dietary sugar on energy metabolism, insulin resistance, lipid metabolism, inflammation, and immune function that could drive cancer progression. 166The result was unlikely influenced by whether fruit juice was investigated as part of the sugary drinks 103,104 or not, 92 as the evidence of an association with fruit juice intake is limited. 92,104ution is needed when interpreting the inverse association observed for artificially sweetened beverages and all-cause mortality, as the result was based on few studies.1 The included studies were rated as having moderate or serious risk of bias from confounding, since they did not account for cancer treatment and other important variables (such as smoking, adiposity, physical activity).This led to little confidence in reaching a stronger conclusion for the observed associations, despite vitamin D has anticancer and antiproliferative effects. 168,169e present systematic review on post-diagnosis intakes of nuts and peanuts, red and processed meat, dairy products, marine n-3 PUFAs, dietary glycaemic load, glycaemic index, insulin load, and insulin index included only a small number of studies and for most showed null results.Findings on post-diagnosis dietary supplement use, included supplemental calcium use that could lower the risk of colorectal cancer development, 19,20 were null or inconsistent in the limited studies identified.Future investigations into the types, dosages, duration of use, and potential interactions with cancer treatment 170 are needed to provide more definitive conclusions for cancer survivors who often use dietary supplements. 171Few RCTs, primarily on dietary supplementation, were identified that did not provide substantial supporting data.
Cancer survival studies have inherent methodological limitations, 50,172  reported results that generally agreed with results of studies that performed single time-point assessments. 81,85,98,120,128Some studies reported that a substantial number of colorectal cancer survivors (53%-85%) reduced post-diagnosis red meat, hamburger, and other fast-food consumption. 174,175It is possible that red and processed meat consumption over time was misclassified, potentially resulting in the null associations with colorectal cancer outcomes observed in this review.A study on nut consumption reported somewhat stronger associations between cumulative nut intake (weighted nut exposure average between two assessments) and improved disease-free and recurrence-free survival, versus a single baseline assessment. 96en possible, investigators should capture repeated dietary assessments across the cancer survivorship trajectory and perform time-varying analyses.We were not able to conduct stratified analysis because we lacked information for cancer characteristics, precise exposure timeframe, geographic location, 176 race/ethnicity, sociodemographics. 177st meta-analyses performed in this review included studies that looked at all-cause or cancer-specific mortality as the main out- This systematic review has enhanced the evidence on postdiagnosis dietary factors and colorectal cancer outcomes through comprehensive collection, synthesis, and evaluation of findings, based to a large extend on observational studies.Such findings will inform the design and execution of carefully designed RCTs that are currently limited but also more challenging to perform particularly when investigating 'hard' endpoints including mortality. 138,179Certain limitations of such RCTs in nutritional epidemiology include the challenges of adhering to particular dietary interventions (inclusive of economic burden), and difficulties in recruiting participants for long-term follow-up and the potential recruitment bias and threat to generalise that results.Moreover, identification of appropriate control diets and blinding of dietary interventions is often challenging. 96,138,179More personalised, multicomponent interventions would be also necessary. 180 whereas a modified version of the Risk of Bias for Nutrition Observational Studies (RoB-Nobs) tool 48 was used to assess the risk of bias in observational studies of dietary and/or lifestyle patterns and in other observational studies included in the meta-analyses.The RoB-Nobs tool was originally developed by the U.S. Department of Agriculture (USDA) Nutrition Evidence Systematic Review after modifications to the Cochrane's collaboration Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I).
Use of omega-3 fatty acid enriched supplements (n = 74; 2 g eicosapentaenoic acid [EPA]/day and 1 g docosahexaenoic acid [DHA]/day) versus omega-3 fatty acid supplements without fish oil (n = 74) resulted in worse 5-year overall survival (HR = 1.73, Evidence from RCTs and longitudinal observational studies examining the associations between post-diagnosis dietary factors, supplement use and colorectal cancer survival outcomes were systematically reviewed.Twenty-one of the investigated associations had sufficient information for meta-analysis.Other associations, including those for dietary and/or lifestyle patterns, were descriptively reviewed.The quality of the evidence was then independently graded.Several studies examined post-diagnosis dietary and/or lifestyle patterns that varied in definitions and derivation methods that is, a priori (hypothesis-driven methods) such as the Alternative Mediterranean Diet (aMed), Dietary Approaches to Stop Hypertension (DASH), or a posteriori (data-driven methods) such as prudent diet, western diet or with hybrid methods such as the Empirical Dietary Inflammatory Pattern (EDIP).We grouped the data broadly into 'healthy' and 'unhealthy' patterns and found generally consistent respective inverse and positive associations with all-cause mortality.'Healthy' dietary patterns comprise mostly high intakes of fruits and vegetables, whole grains, nuts and legumes, and low intakes of red and processed meat.
attributed to individual dietary or lifestyle components within the pattern or potential interactions between dietary and other lifestyle components.Future research on dietary patterns should involve intervention studies that examine both survival outcomes and intermediate omics outcomes (e.g., gut microbiota) that could inform on potential mechanisms and potentially provide biomarkers of effect.Biological mechanisms underlying the associations between dietary patterns and cancer development, or progression, are poorly understood.Adherence to healthy dietary patterns (e.g., Mediterranean diet or predominantly plant-based diets) has been associated with reduced circulating markers of inflammation, oxidative stress, insulin come.In general, such studies are simpler to conduct because information on mortality can be easily captured through death certificates and/or registries of vital status.Information on recurrence is usually only captured in clinical trials or via time-consuming clinical record review.A further limitation is that studies with 'recurrence' as outcome have used heterogenous definitions (such as 'disease-free survival', 'event-free survival', etc.) making comparisons more difficult and potential errors in outcome assessment more likely.178Future studies should use standardised, cancer-specific recurrence definitions to allow more consistent evaluation of this body of evidence and subgroup analyses.Health related quality of life outcomes should also receive more attention in future studies to support the design of suitable survivorship care/plans.An overview of limitations of cancer survival studies and future research recommendations is presented in the summary manuscript, Box 131 of the current manuscript series on colorectal cancer survivors.

-182 5 |
CONCLUSIONSThere was 'limited-suggestive' evidence for the associations between post-diagnosis healthy dietary and/or lifestyle patterns, intake of whole grain, or coffee (total, caffeinated, decaffeinated) with lower risk of all-cause mortality, and for the associations between post-diagnosis unhealthy dietary patterns or intake of sugary drinks with higher risk of all-cause mortality.The evidence for other exposure-outcome associations received a 'limited-no conclusion' grading.Conclusions made by the CUP Global independent Expert Committee on Cancer Survivorship and the Expert Panel may contribute towards future formulation of lifestyle guidance/recommendations specific for colorectal cancer survivors.The current evidence is not strong enough for the development of recommendations for cancer survivors following the well-established CUP Global process but a new complementary process, considering evidence which may be more 'limited' alongside expert opinion would allow the development of guidance, to provide cancer survivors with sound information based on the best available evidence.To provide conclusions with a higher level of certainty and develop specific lifestyle recommendations, additional evidence is needed from larger, well-designed observational studies in well-characterised populations, with repeated exposure and confounder assessments.Mechanistic studies exploring the biological pathways that underpin potential associations between dietary exposures and colorectal cancer outcomes are crucial to inform recommendations.RCTs,180 that could possibly, evaluate the effects of specific dietary patterns, or coffee183 that have shown survival benefits in this SLR would be informative.MR studies using instrumental exposures to account for confounding and reverse causation153 could be used to clarify the results for circulating 25(OH)D or other biomarkers, and examine the role of diet and gut microbiome on colorectal cancer prognosis. 184Additional studies are also needed in socio-demographically and ethnically diverse survivors, of different cancer stages, and at different phases of the cancer continuum.Georgios Markozannes and Nerea Becerra-Tomás did the data extraction and checking.Georgios Markozannes, Margarita Cariolou and Nerea Becerra-Tomás did the risk of bias assessment.Doris S. M. Chan, Georgios Markozannes, Margarita Cariolou, Katia Balducci, Sonia Kiss and Rita Vieira analysed and interpreted the data.Konstantinos K. Tsilidis interpreted the data.Dagfinn Aune was a WCRF International CUP Global ICL team member who revised the manuscript.Darren C. Greenwood was a statistical adviser.Amanda J Cross was a CUP Global advisor at Imperial College London.Esther M. González-Gil was a CUP Global collaborator on biological processes and provided input into the biological mechanism citations in the manuscript.Ellen Copson was a PEG member, Chair of CUP Global Expert Committee on Cancer Survivorship, and Expert Panel member.Wendy Demark-Wahnefried and Galina Velikova were PEG, OACD, and CUP Global Expert Committee members.Andrew G. Renehan was a PEG member and Deputy Chair of CUP Global Expert Committee on Cancer Survivorship.John Krebs, Matty P, Weijenberg, Monica L. Baskin, Sarah J. Lewis, Jaap Seidell, Rajiv Chowdhury, and Lynette Hill were CUP Global Expert Panel members.Anne M. May, Anne Tjonneland, Karen Steindorf, Martijn Bours, Melissa M. Hudson, Roderick Skinner, and Folakemi T. Odedina were CUP Global Expert Committee members.All members of the CUP Global Expert Committee and Expert Panel provided input into the judgements on the evidence and advised on the interpretation of the review, the public representative (Lynette Hill) did not contribute to the final decisions made by the Panel.Doris S. M. Chan and Margarita Cariolou drafted the original manuscript.All authors reviewed and provided comments on the manuscript.Doris S. M Chan is the guarantor and has full access to all the data and takes responsibility for the integrity of the data and the accuracy of the data analysis.The work reported in the paper has been performed by the authors, unless clearly specified in the text.CUP Transition workstream (Anne McTiernan, Steven Clinton, Viv Lund) into the development of the original protocol for this work.
Descriptive table of dietary randomised controlled trials and colorectal cancer outcomes.
79% CI = 1.06-2.83)and3-yearrecurrence-freesurvival of similar size but with a wide 95% CI crossing the null (HR = 1.66; 95% CI = 0.65-4.26).74Colorectalcancersurvivors who received individualised nutritional counselling had the longest colorectal cancerspecific survival and lowest number of recurrences compared with those who received high-protein supplements (40 g/day) or usual diet (median 7.3, 6.5, 4.9 years, respectively, p < 0.05 and n = 7/34, 9/29, 15/26, respectively, p < 0.01).79 Name of dietary and/or lifestyle patterns: 2007/2018 WCRF/AICR, 2007/2018 World Cancer Research Fund/American Institute for Cancer Research Cancer Prevention Recommendations; ACS, American Cancer Society Nutrition and Physical Activity Guideline for Cancer Survivors; AHEI, Alternative Healthy Eating Index; aMED, Alternative Mediterranean Diet; DASH, Dietary Approaches to Stop Hypertension; EDIH, Empirical Dietary Index for Hyperinsulinemia; EDIP, Empirical Dietary Inflammatory Pattern; HNFI, Healthy Nordic Food Index; National score (based on Dutch Healthy Diet index); MMDS, Modified Mediterranean Diet Score; h/uPDI, Healthy/Unhealthy Plant-based Diet Index; Prudent dietary pattern; Western dietary pattern.Study abbreviations: CALGB 89803, Cancer And Leukemia Group B 89803; CALGB SWOG 80405, Cancer And Leukemia Group B (Alliance) Southwest Oncology Group 80405; COLON, COlorectal cancer: Longitudinal, Observational study on Nutritional and lifestyle factors that may influence colorectal tumour recurrence, survival and quality of life; CPS-II NC, Cancer Prevention Study-II Nutrition Cohort; EnCoRe, Energy for life after ColoRectal cancer; HPFS, Health Professionals Follow-Up Study; IWHS, Iowa Women's Health Study; NHS, Nurses' Health Study.
used techniques that partially accounted for selection bias, the rest (94%) had a serious risk of selection bias.
50currence-free individuals, but this information is rarely available.50Exposuremeasurement error and misclassification is possible, since dietary information is largely self-reported, once at-diagnosis or at a non-clearly specified time after diagnosis, during which diet might have been affected by disease progression and/or cancer treatment.Future studies should report results in a 173cer progression could be a confounder in studies of mortality outcomes.Associations could be biassed if undetected disease progression leads to diet alterations (e.g., malnutrition due to altered nutrient absorption), weight loss173and hence worse outcomes after diagnosis.Potential approaches to reduce such bias in future observational studies include performing lagged analyses or restricting study participation to