HAL Id: hal-01628671
https://hal.archives-ouvertes.fr/hal-01628671
Submitted on 26 May 2020
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Distributed under a Creative Commons Attribution| 4.0 International License
Do alcoholic beverages, obesity and other nutritional
factors modify the risk of familial colorectal cancer? A
systematic review.
Anthony Fardet, Nathalie Pecollo, Mathilde Touvier, Paule Latino-Martel
To cite this version:
Anthony Fardet, Nathalie Pecollo, Mathilde Touvier, Paule Latino-Martel.
Do alcoholic
bev-erages, obesity and other nutritional factors modify the risk of familial colorectal cancer?
A
systematic review..
Critical Reviews in Oncology/Hematology, Elsevier, 2017, 119, pp.94-112.
Contents lists available atScienceDirect
Critical Reviews in Oncology / Hematology
journal homepage:www.elsevier.com/locate/critrevoncDo alcoholic beverages, obesity and other nutritional factors modify the risk
of familial colorectal cancer? A systematic review
Anthony Fardet
a, Nathalie Druesne-Pecollo
b,c, Mathilde Touvier
b,c, Paule Latino-Martel
b,c,⁎aINRA, UMR 1019, UNH, CRNH Auvergne, F-63000 Clermont-Ferrand & Clermont University, University of Auvergne, Human Nutrition Unit, BP 10448, F-63000
Clermont-Ferrand, France
bSorbonne Paris Cité Epidemiology and Statistics Research Centre (CRESS), Inserm U1153, Inra U1125, Cnam, Nutritional Epidemiology Research Team (EREN),
Bobigny, France
cFrench Network for Nutrition and Cancer Research (NACRe Network), France
A R T I C L E I N F O
Keywords: Colorectal cancer Family history Lynch syndrome Foods Diet Alcoholic beverages Red meat Processed meat Physical activity Dietary patterns PreventionA B S T R A C T
Purpose: Individuals with family history of colorectal cancer are at higher risk of colorectal cancer than the general population. Until now, guidelines for familial colorectal cancer risk have only pointed at early diagnosis efforts via screening tests and surveillance, and payed scarce or no attention to lowering exposure to modifiable risk factors, notably nutritional factors.
Methods: We conducted a systematic review of epidemiological studies investigating the associations between nutritional factors, family history of colorectal cancer, and colorectal cancer risk. From the 5312 abstracts identified until December 2016, 184 full text articles were examined for eligibility. Finally, 31 articles (21 from case-control studies, 9 from cohort studies and 1 from an intervention study) met inclusion criteria and were analyzed.
Results: Mainly, the combinations of family history of colorectal cancer and higher consumptions of alcoholic beverages, red or processed meat, or overweight/obesity increase the risk of colorectal cancer. Consistently, a strong increase is observed with the combinations of family history of colorectal cancer and unhealthy dietary patterns/lifestyles. Statistically significant interactions between these nutritional factors, family history of col-orectal cancer and colcol-orectal cancer risk are reported. Other data are inconclusive and additional prospective studies are needed.
Conclusions: For thefirst time, our findings highlight that addressing high consumption of alcoholic beverages, red or processed meat, and overweight/obesity, and more largely the exposure to multiple unhealthy dietary/ nutritional behaviors could offer new perspectives of prevention to individuals with family history of colorectal cancer. A better information of these patients and of health professionals on these nutritional modifiable risk factors is recommended.
1. Introduction
It is well recognized that subjects with family history (FH) of col-orectal cancer (CRC) are at higher risk of CRC (Slattery and Kerber,
1994; Kerber et al., 1998). The overall risk is increased two-fold in
subjects with afirst-degree relative with CRC, the risk increasing with the number of relatives affected (Potter et al., 1993). Both inherited genetic alterations and acquired lifestyle factors are thought to be in-volved in such an increased risk but the interaction between them has been quite poorly studied (Keku et al., 2003).
Genetic factors associated with FH of CRC are mainly those
encountered in familial adenomatous polyposis (FAP) and Lynch Syndrome (LS), also known as hereditary nonpolyposis colon cancer (HNPCC). It is estimated that 1–3% of all CRC are due to LS (Aaltonen
et al., 1998). The high cancer risk in LS is caused by pathogenic
germline mutations in genes involved in or influencing DNA mismatch repair (MMR), i.e., hMLH1, hMSH2, hMSH6, PMS2, or EPCAM (
Abdel-Rahman et al., 2006; Ligtenberg et al., 2009). FAP is caused by APC
(Adenomatous Polyposis Coli) gene defects on chromosome 5q21, i.e., a flaw in the body's tumor suppressor genes that prevent development of tumors (Cruz-Correa and Giardiello, 2003). FAP accounts for approxi-mately 1% of cases of CRC (Cruz-Correa and Giardiello, 2003).
http://dx.doi.org/10.1016/j.critrevonc.2017.09.001
Received 10 March 2017; Received in revised form 28 July 2017; Accepted 6 September 2017
⁎Corresponding author at: Sorbonne Paris Cité Epidemiology and Statistics Research Centre (CRESS), Inserm U1153, Inra U1125, Cnam, Nutritional Epidemiology Research Team
(EREN), Bobigny, France.
E-mail address:paule.latino-martel@inra.fr(P. Latino-Martel).
Abbreviations: CC, Colon Cancer; CRC, Colorectal Cancer; FH, Family History; HNPCC, Hereditary Non-Polyposis Colorectal Cancer; LS, Lynch Syndrome
1040-8428/ © 2017 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
The Western-type dietary pattern has been shown as contributing to the development of CRC in the general population (Magalhães et al., 2012). It is characterized by high consumptions of animal-based, ultra-processed, refined and energy-dense foods. Based on the systematic review of the scientific literature and the evaluation of the level of evidence, dietary recommendations have been established to improve CRC prevention: higher physical activity, higher intake of foods con-taining dietary fiber, reduced consumption of red and/or processed meats and alcoholic drinks, reduced body and abdominal fatness
(WCRF/AICR, 2011; Latino-Martel et al., 2016; WCRF/AICR, 2017).
Interestingly, Western diet has also been reported to modulate the CRC risk associated with a FH of CRC (Slattery et al., 2000). However, compared to sporadic CRC, only very few studies have assessed the influence of lifestyle factors on familial CRC: as reviewed in 2013 by van Duijnhoven et al. in the case of LS,“most investigations have fo-cused on smoking habits and body fatness, which both seem to increase the risk of colorectal tumors in LS. Other lifestyle factors, such as physical activity, alcohol or diet have not or only scarcely been studied in relation to colorectal tumors in persons with LS. It is, therefore, difficult to draw firm conclusions from the current literature” (van
Duijnhoven et al., 2013).
A few studies compared health behaviors among subjects with or without FH of CRC. For example, Townsend et al. showed in a Californian population-based study that men and women with a FH of CRC are less likely to maintain a healthy weight and to consumefive or more servings of fruits and vegetables per day, than those without a FH of cancer (Townsend et al., 2013). In another Californian population-based study, it was shown that individuals with a FH have lower odds of adherence to lifestyle recommendations (Bostean et al., 2013). Authors suggested that this may reflect either “shared behavioral risks within families, or the lack of knowledge about how certain lifestyle behaviors impact personal cancer risk” (Bostean et al., 2013). In another recent study, Spanish individuals with a FH of CRC exhibited no better health-related behaviors than people without FH of CRC (Martinez-Ochoa
et al., 2012). Conversely, with data from the 2008 Oregon Behavioral
Risk Factor Surveillance System, a FH of CRC has been significantly associated with respondents reporting lifestyle changes (in eating habits or physical activity) to prevent CRC (OR, 2.6; 95% CI, 1.7–4.0) (Zlot
et al., 2012).
To better inform subjects with FH of CRC on factors that could modulate their risk of CRC and help them favor preventive behaviors, there is a need to better understand the role of lifestyle behaviors, especially dietary factors. Therefore, the objective of this systematic review was to provide an overview of the literature on human studies, both observational and interventional, that examined the association between nutritional factors and CRC risk in individuals with FH of CRC, and/or the joint effect of nutritional factors and FH of CRC on CRC risk as compared to non-exposed subjects without FH of CRC. Since FH of CRC and hereditary syndromes predisposing to CRC are not totally disjoint situations, we have evaluated studies on individuals with FH of CRC together with those on individuals with known or suspected her-editary syndromes.
2. Methods
The PRISMA protocol/checklist was followed for the systematic review (Moher et al., 2009).
2.1. Literature search strategy
We conducted afirst search in Medline and Embase databases (up to August 2013), without publication date or language restrictions, by combining the medical subject headings (Medline) or indexed terms (Embase) and corresponding entry terms for colorectal cancers, FH (or hereditary syndromes) and nutritional factors (Supplementary Material 1). After removal of duplicates, no additional eligible article has been
identified via the Embase database. Therefore an updated search (up to December 2016) was conducted in Medline only. We also hand-sear-ched reference lists from retrieved articles and reviews on the related topic.
2.2. Study selection
Abstracts or full-text manuscripts were identified and reviewed in-dependently by two investigators, with all discrepancies resolved through discussion.
Studies were included if they met the following inclusion criteria: original research article, case-control, cohort or intervention study design, conducted on subjects with FH of colorectal cancer, with col-orectal cancer as outcome and report of the odds ratio (OR), risk ratio (RR) or hazard ratio (HR) and 95% confidence interval (CI) for nutri-tional factors exposure. According to our inclusion criteria, studies whose outcomes were colorectal adenomas or polyps, and studies which did not provide any risk estimate were excluded. Only published peer-reviewed studies were included. No age restriction or minimum length of follow-up was required. In case of duplicate publications, the selection was based on completeness of the information.
2.3. Data extraction
Using a standardized data collection form, the following informa-tion was extracted for each article by one investigator and verified by a second investigator:first author’s last name, publication year, study characteristics (design, country, recruitment, and mean follow-up per-iods for prospective studies), participants characteristics (sample size, mean age, sex, cancer site, number of cases, FH of CRC), exposure (nutritional factor, exposure assessment), comparisons and corre-sponding ORs, RRs or HRs and 95% CIs, control for confounding (ad-justment for covariates), and potential bias. When available, data for subjects without FH within the same population sample was also re-corded.
3. Results
3.1. Characteristics of the studies
Fig. 1shows theflow chart of the selection process. From the 5312
abstracts provided by searches in Medline and Embase databases, 184 full text articles were identified and examined. Among the thirty-one potentially relevant full-text publications identified, we excluded one
Table 1 Characteristics of the studies examining the relationship between nutritional factors and colorectal cancer risk in populations with FH of colorec tal cancer. Location, recruitment period, follow-up a
Total cases/ total controls
b Sex, Mean age (y) Cancer site Family history of cancer (degree, number of cases/ controls) Nutritional factor Control for confounding factors (adjustment) Exposure assessment Comment First author, publication year (ref) Case-control studies USA, 1991 –1994 This data set is common to fi ve articles ( Slattery et al., 1997b , Slattery et al., 1997a , Slattery et al., 1997c , Caan et al., 1998 , Slattery et al., 2000 and Slattery et al., 2003 ): there is a potential overlap for two articles reporting results on Western diet, but in one article results concerned CRC ( Slattery et al., 2003 ) and in the other one they concerned CC and analyses were strati fi ed by age category ( Slattery et al., 2000 ); an overlap between results on physical activity from two articles ( Slattery et al., 2003; Slattery et al., 1997a ) cannot be excluded. 1993/ 2400 M + F, 30 –79 CC Yes (in fi rst degree relatives, 310/228)/ No Saturated, monounsaturated and polyunsaturated fatty acids – in combination with FH Age at diagnosis or selection, total energy intake, dietary fi ber, cholesterol, calcium, BMI, physical activity and use of NSAIDs Adaptation of the validated CARDIA diet history questionnaire Potential recall bias for exposures and FH of CRC ( Slattery et al., 1997b ) 1993/NA M + F, 30 –79 CC Yes (in fi rst degree relatives, NA/NA)/No Lifetime vigorous activity level Age at selection or diagnosis, BMI, energy intake, dietary fi ber, dietary calcium, FH of a fi rst-degree relative with colorectal cancer (for everyone combined only), and use of aspirin and/or NSAIDs Adaptation of the CARDIA physical activity history Potential recall bias for exposures and FH of CRC ( Slattery et al., 1997a ) 1993/ 2410 M + F, 30 –79 CC Yes (in fi rst degree relatives, 364/252)/ No Energy consumption − in combination with FH Age, body mass index, physical activity, use of NSAIDs/aspirin, dietary fi ber, cholesterol, and calcium Validated CARDIA diet history Potential recall bias for exposures and FH of CRC ( Slattery et al., 1997c ) 1983/ 2400 M + F, 30 –79 CC Yes (in fi rst degree relatives, NA/NA)/No BMI, WHR Age at selection, ever regularly used aspirin and/or NSAIDs, intake of dietary energy, fi ber and calcium and category of long-term vigorous leisure activity Adaptation of the validated CARDIA diet history questionnaire Potential recall bias for exposures and FH of CRC. OR and CI are not provided for WHR. ( Caan et al., 1998 ) 1624/ 1963 M + F, 30 –79 CC Yes (in fi rst degree relatives, 266/183)/ No Western diet (including 9 foods for women and 13 foods for men), eggs, re fi ned grains, sugar, red meat, processed meat, fast-food meat – in combination with FH Age, sex, calories, BMI, long-term vigorous physical activity, and usual number of cigarettes smoked per day Adaptation of the validated CARDIA diet history questionnaire Potential recall bias for exposures and FH of CRC. ( Slattery et al., 2000 ) 2260/ 2749 M + F, 30 –79 CRC Yes (in fi rst degree relatives, 299/230)/ No Prudent diet (including 5 foods), BMI, Western diet (including 9 foods for women and 13 foods for men), physical activity – in combination with FH All variables in same model and mutually adjusted for age, sex, and study Adaptation of the validated CARDIA diet history questionnaire Potential recall bias for exposures and FH of CRC. ( Slattery et al., 2003 ) USA (Hawaii), 1987 –1991 1192/ 1192 M + F, controls: 58, cases: 66 (medians) CRC Yes (in fi rst degree relatives, 169/75)/No Total calories, calories from fat, calories from other sources, saturated fat, P:S ratio, beef, processed meats, eggs, chicken without skin, margarine, total calcium, NSP, total carotenoids, methionine, legumes and soy, total vegetables, broccoli, ethanol, BMI, healthy and unhealthy lifestyles (including both dietary factors, smoking, BMI and physical activity) – in combination with FH or not Age, alcoholic drinks/week, pack-years of cigarette smoking, lifetime recreational physical activity, Quetelet index 5 years ago, dietary fi ber, calcium intake, egg intake, and total calories. Calories from fat and calories from other sources were additionally adjusted for each other. All of the nutrients were adjusted for calories by the method of residuals Validated food frequency questionnaire based on more than 282 food items. The reference period for the diet questionnaire was the 3-year period before the onset of symptoms for the cases and the 3-year period before interview for the controls Potential recall bias for exposures and FH of CRC. ( Le Marchand et al., 1999 ) Italy, 1985 –1992 The data set for subjects with FH is common to Refs. Fernandez et al. (1997) and Fernandez et al. (2002) : there is an overlap for some nutritional factors, i.e. red meat, seasoning fats, β -carotene and ascorbic acid. (continued on next page)
Table 1 (continued) Location, recruitment period, follow-up a
Total cases/ total controls
b Sex, Mean age (y) Cancer site Family history of cancer (degree, number of cases/ controls) Nutritional factor Control for confounding factors (adjustment) Exposure assessment Comment First author, publication year (ref) Case-control studies 112/108 M + F, controls: 56, cases: 61 (medians) CRC Yes (in fi rst degree relatives, 112/108) Alcohol, co ff ee, pasta, pastries, red meat, poultry, raw ham, ham, canned meat, cheese, cabbages, spinach, tomatoes, peppers, lettuce, total vegetables, citrus fruits, melon, butter, seasoning fats, retinol, β -carotene, ascorbic acid, calcium, vitamin D, vitamin E, folate and methionine Gender, age and area of residence. Allowance for other potential confounding variables (i.e. years of schooling, body mass index, total energy intake) did not substantially modify any of the estimates Weekly frequency of consumption Potential recall bias for exposures and FH of CRC. Small number of cases. Hospital-based controls. Limited number of food items. Over-representation of smokers in the control group. ( Fernandez etal., 1997 ) 1584/ 2879 M + F, controls: 55, cases: 62 (medians) CRC Yes (in fi rst degree relatives, 112/108)/ No Red meat, seasoning fats, daily meal frequency, β -carotene, ascorbic acid, adult life dietary risk score – in combination with FH or not Age, sex, area of residence, education, and total energy intake Food frequency questionnaire to collect information on the weekly frequency of consumption of 29 indicator foods Potential recall bias for exposures and FH of CRC. Limited number of food items. ( Fernandez etal., 2002 ) Italy, 1992 –1996 This data set is common to references ( Negri et al., 1998 , Tavani et al., 1998 , La Vecchia et al., 1999 , Fernandez et al., 2004 and Turati et al., 2011 ). There is no overlap, the factors studied in these articles being di ff erent. 1953/ 4154 M + F, controls: 58, cases: 62 (medians) CRC Yes (in fi rst degree relatives, 187/146)/ No Total fi ber Center, sex, age, education, physical activity and intake of proteins, fats, carbohydrates, and alcohol Interviewer-administered food frequency questionnaire (based on 78 foods, groups of foods, or dishes divided into 6 sections) to assess the usual diet during the 2 years preceding diagnosis (for cases) or hospital admission (for controls) Potential recall bias for exposures and FH of CRC. Hospital-based controls ( Negri et al., 1998 ) 1953/ 4154 M + F, controls: 56, cases: 61 –62 (medians) CRC Yes (NA,NA/NA)/No Alcohol Center, age, sex, education, physical activity, smoking status and intake of and intake of β -carotene, vitamin C and total energy Validated food-frequency questionnaire including 78 questions Potential recall bias for exposures and FH of CRC. Hospital-based controls ( Tavani et al., 1998 ) 1225/ 4154 M + F, controls: 58, cases: 62 (medians) CC Yes (in fi rst degree relatives, 134/146)/ No Physical activity, energy, vegetable, meal frequency Center, age, sex, education, level of occupational physical activity, total energy intake, meal frequency, and FH of colorectal cancer Validated food frequency section including 78 foods or food groups Potential recall bias for exposures and FH of CRC. Hospital-based controls ( La Vecchia et al., 1999 ) 1225/ 4154 M + F, controls: 58, cases: 62 (medians) CC Yes (in fi rst degree relatives, 134/146)/ No Risk factor score (including education, occupational physical activity, daily meal frequency, intake of fi ber, intake of calcium, and intake of β -carotene) – in combination with FH Age, sex, center, and total energy intake Validated questionnaire comprising 78 foods, food groups, or recipes and allowing the estimation of energy intake as well as of several micronutrients Potential recall bias for exposures and FH of CRC. Hospital-based controls ( Fernandez etal., 2004 ) 1953/ 4154 M + F, 20 –74 CRC, CC, RC Yes (in fi rst degree relatives (187/146)/ No ‘Starch-rich ’ dietary pattern (greatest loading on starch, vegetable protein and sodium), ‘vitamins and fi ber ’dietary pattern (greatest loadings on vitamin C, total fi ber, beta-carotene, total folate and soluble carbohydrates) – in combination with FH Age, sex, center, education, occupational physical activity, and tertiles of consumption of all the remaining dietary patterns [i.e. ‘animal products ’, ‘unsaturated fats (animal source) ’, ‘unsaturated fats (vegetable source) ’, and ‘starch-rich ’ or ‘vitamins and fi ber ’ when Validated food frequency questionnaire with satisfactory reproducibility and validity based on 78 foods, groups of foods, recipes, and 5 alcoholic beverages Potential recall bias for exposures and FH of CRC. Hospital-based controls ( Turati et al., 2011 ) (continued on next page)
Table 1 (continued) Location, recruitment period, follow-up a
Total cases/ total controls
b Sex, Mean age (y) Cancer site Family history of cancer (degree, number of cases/ controls) Nutritional factor Control for confounding factors (adjustment) Exposure assessment Comment First author, publication year (ref) Case-control studies appropriate] Japan, 1979 –1987 3327/ 14884 M, ≥ 20 CC, RC Yes (in parents, siblings and grandparents, NA/ NA)/No Beer – in combination with FH and occupation Age, residence, smoking habits The registry has routinely collected data on occupation, marital status, FH of cancer (parents, siblings and grandparents) and the questions about smoking and drinking have been included in the report format since 1980 Potential recall bias for exposures and FH of CRC. Controls with other sites of cancer. Limited number of confounding factors controlled ( Kato et al., 1990 ) Japan, 1988 –1998 1352/ 50706 M + F, 18 –79 CRC Yes (in fi rst degree relatives, 124/2456)/ No Pickled vegetables, fruits, raw vegetables, green tea, preference for co ff ee, miso soup, milk, eastern type breakfast, preference for oily foods, bean curd, carrots, pumpkin, cabbage, lettuce, potatoes, egg, chicken, beef, pork, sausage, instant foods, frozen foods, salted fi shes, cooked fi shes, physical exercise, alcohol Age and sex Self-administered questionnaire Including items on demography, medical history, FH of disease in parents and siblings, smoking and drinking habits, dietary habits, physical exercise, bowel habits, and reproductive history before symptoms appeared Potential recall bias for exposures and FH of CRC. Limited number of confounding factors controlled ( Huang et al., 2004 ) Canada, 1997 –2000, 2003 –2006 2696/ 2668 M + F, 20 –74 CRC Yes (high/ intermediate risk group according to Amsterdam-Bethesda criteria, 878/NA)/No BMI (2 y ago, or since age 20 y), weight gain since age 20 y Age, province, history of hypercholesterolemia/ hypertriglyceridemia, education, and history of colon screening endoscopy Self-administered personal history questionnaire Potential recall bias for exposures. Subjects recruited by the population-based Ontario cancer registry ( Campbell et al., 2007 ) Pays-Bas, 1999 –2002 145/103 M + F, 18 –75 CRC Yes (family ful fi lling the Amsterdam criteria for HNPCC, 145/103) Total energy intake, total vegetables and fruit, vegetables, fruits, cereals, total meat, red meat, poultry, fi sh, dairy products, alcohol, fat, protein, carbohydrates, dietary fi ber, calcium, vitamin C, total energy intake Age at last colonoscopy, sex, total energy intake, carrier status, and cigarette smoking. Alcohol and nutrients are also adjusted on total energy intake by the residual method Validated, semiquantitative food frequency questionnaire (based on 79 food items) that was originally developed for the Dutch cohort of the European Prospective Investigation into Cancer and Nutrition (EPIC) Potential recall bias for exposures. Small number of cases. Hospital-based controls. Subjects were either known or suspected MMR gene mutation carriers. ( Diergaarde etal., 2007 ) Australia, 1980 –1981 702/710 M + F, controls: 64.8, cases: 65.7 CRC Yes (in male siblings, 11/17)/No Beer – in combination with FH Age and sex Dietary questionnaire which included alcohol intake and tobacco consumption Potential recall bias for exposures and FH of CRC. Limited number of confounding factors controlled ( Kune et al., 1989 ) USA, Australia and Canada, 1997 –2007 188/274 M + F, controls: 55 (M) and 54 (F), cases: 56 (M) and 55 (F) CRC Yes (in fi rst degree relatives, 80/116)/No BMI (recent) Age, endoscopy screening and cigarette smoking. BMI < 18.5 excluded from the linear model Recent BMI was calculated from self-reported height and weight approximately 1 year before CRC diagnosis for cases subjects, or before Potential recall bias for exposures and FH of CRC. Small number of cases. Cases with high MSI (microsatellite instability) CRC, recruited and ( Campbell et al., 2010 ) (continued on next page)
Table 1 (continued) Location, recruitment period, follow-up a
Total cases/ total controls
b Sex, Mean age (y) Cancer site Family history of cancer (degree, number of cases/ controls) Nutritional factor Control for confounding factors (adjustment) Exposure assessment Comment First author, publication year (ref) Case-control studies enrollment for sibling control subjects genetically characterized by the Colon cancer family registry: potential overlap with the sample of another study ( Win et al., 2011 ) for Canadian subjects USA, Canada 3350/ 3504 M + F, controls: 57.8, cases: 59.4 CRC Yes (MMR gene mutation carriers, 243/350)/No Non processed red meat, processed red meat, total processed meats, total pan-fried meat, total oven-broiled meat, total grilled meat Age, BMI, gender, race, saturated fat, dietary fi ber, center, vegetables, physical activity, and total calorie intake Validated food frequency questionnaire for 200 food items and more than 100 nutrients Potential recall bias for exposures. Data on meat subtypes also provided. Potential overlap with the sample of another study for subjects from Hawaii ( Le Marchand et al., 1999 ), but analyses are di ff erent ( Joshi et al., 2015 ) Cohort studies USA (Iowa), 1986, 10 years 241/ 34975 F, 61.7 CC Yes (in fi rst degree relatives, 61/4178)/ No Total fruit & vegetable, total vegetable, total fruit, green leafy vegetables, cruciferous vegetables, legumes, fruit and vegetables high in vitamin C, dietary fi ber, garlic, total dairy, high fat dairy, low fat dairy, high sucrose foods, total meat, red meat, white meat, nitrate meat (bacon, hot dogs, processed meats), total calcium, dietary calcium, supplemental calcium, total vitamin C, dietary vitamin C, supplemental vitamin C, total vitamin A, dietary vitamin A, supplemental vitamin A, carotene, total vitamin E, dietary vitamin E, supplemental vitamin E, total vitamin D, dietary vitamin D, supplemental vitamin D and folate Age at baseline, total energy intake, and history of rectal colon polyps Semiquantitative 127-food-item frequency questionnaire Potential selection bias and recall bias for FH of CRC. Small numbers of cases limited the statistical power to detect associations (especially strati fi ed analyses and tests of interaction) ( Sellers et al., 1998 ) USA (NHS), 1980, 16 years 535/ 88223 F, 46.7 –46.8 CC Yes (in fi rst degree relatives, 107/6849)/ No Total folates, folates from food sources only, multivitamin supplement use in 1980, methionine, alcohol – in combination with FH or not Age, pack-years of smoking before age 30 years, body-mass index, regular vigorous exercise, regular aspirin use, screening endoscopy, beef, pork, or lamb as a main dish, alcohol consumption, and energy-adjusted levels of methionine 61-item semiquantitative food frequency questionnaire to assess diet as well as supplemental vitamin use Potential selection bias and recall bias for FH of CRC. Data on the duration of vitamin use or elevated folate intake also available ( Fuchs et al., 2002 ) USA (NHS and HPFS), 1976 (females) and 1986 (males), 26 years (females) 1801/ 133350 M + F, 52.5 –54.4 CC Yes (in parents and siblings, 484/NA)/No Alcohol – in combination with FH or not Pack-years of smoking before age 30 y, BMI, history of endoscopy, use of aspirin, and intakes of energy, calcium, folate, and red meat; in the NHS: information on ≈ 60 –130-item semiquantitative food frequency questionnaire each 4 years Potential selection bias and recall bias for FH of CRC. Data on women were likely to overlap with those from Fuchs et al. ( Cho et al., 2012 ) (continued on next page)
Table 1 (continued) Location, recruitment period, follow-up a
Total cases/ total controls
b Sex, Mean age (y) Cancer site Family history of cancer (degree, number of cases/ controls) Nutritional factor Control for confounding factors (adjustment) Exposure assessment Comment First author, publication year (ref) Case-control studies and 20 years (males) menopausal status and use of postmenopausal hormone therapy; FH of colorectal cancer in parents and siblings (2002) . Limited numbers of participants with heavy alcohol consumption USA, Australia, New-Zealand, Canada, 1997 –2007, 5 years 695/1848 M + F, MMR gene mutation carriers: 44.0 –44.9, non-carriers: 51.9 –56.8 CRC Yes (MMR gene mutation carriers, 659/665)/No BMI at age 20 years Sex, country, cigarette smoking and alcohol drinking in both carriers and non-carriers; and further adjusted for speci fi c MMR gene mutated in carriers Standardised personal interviews, telephone interviews or mailed questionnaires Retrospective study. Potential selection and recall bias for exposures. Subjects recruited and genetically characterized by the Colon cancer family registry: potential overlap with the sample of another article ( Campbell et al., 2010 ) for Canadian subjects ( Win et al., 2011 ) Netherlands, 2006 –2008, 28 months (median) 131/339 M + F, 39.1 –60.7 CRC Yes (inviduals with Lynch syndrome, 122/ 326) Folate, vitamins B2, B6, B12, methionine Age, sex, number of colonoscopies during person-time, NSAID use, physical activity, and mutually adjusted on other vitamins Validated 183-item semi-quantitative self-administered food frequency questionnaire Potential selection bias. Small number of cases. Analyses for MTHFR C677T genotype are also provided ( Jung et al., 2014 ) Taiwan, 2002 –2012, 12,529 total person-time 147/154 M + F, 41 (median) CRC Yes (individuals with MMR gene mutations ful fi lling the Amsterdam II criteria for Lynch syndrome, 147/154) Regular physical activity, alcohol drinking, tea, co ff ee, meat, vegetable, fruit, seafood and staple food intakes Mutated MMR genes, year of birth Interviews by trained professional nurses. Usual dietary intake of 14 food items, for 5 y preceding the date of study registry Retrospective study. Potential selection and recall bias for exposures. Limited number of food items. Separate analyses for MLH1 or MSH2 germline mutation carriers also provided ( Kamiza et al., 2015 ) International centers (northern Europe, UK, other regions), 1999 –2005, 55.7 months 55/882 M + F, 45.2 CRC Yes (individuals with Lynch syndrome, 54/ 842) BMI Age, starch, aspirin, geographic region, mismatch repair gene, and sex BMI was calculated from height and weight reported by participants at recruitment Potential selection bias. Self-reported height and weight. Small number of cases. Separate analyses for men and women also provided ( Movahedi etal., 2015 ) USA, Canada, Australia, New Zealand, 1997 –2012, up to 80,420 person-years 744/1222 M + F, 41.8 CRC Yes (individuals with Lynch syndrome, 744/ 1222) Multivitamin supplement, calcium supplement, acid folique supplement Ascertainment, education, country, sex, number of screening colonoscopies, regular physical activity, cigarette smoking and intake of aspirin and/or ibuprofen Data collection at the time of recruitment, using standardized questionnaires via personal interviews, telephone interviews or mails Retrospective study. Potential selection and recall bias for exposures. Separate analyses for men and women also provided. ( Chau et al., 2016 ) USA, Canada, Australia, New Zealand, 1997 –2012, 769/1156 M + F, 42.2 CRC Yes (MMR gene mutation carriers, 769/1156) Average daily ethanol intake from any alcoholic beverage Country, education, ascertainment, sex, BMI at age 20, diabetes status, regular physical activity, smoking status, and intakes of aspirin or ibuprofen, multivitamin, calcium and folic At the time of baseline recruitment, standardized questionnaires were used to collect self-reported information Potential selection bias. Analyses for ethanol intake from beer, wine or spirits, and for colon and rectum cancers separately, also provided. ( Dashti et al., 2016 ) (continued on next page)
duplicate publication. One supplementary article was identified from hand-searched reference lists. Finally, the results of thirty-one articles met the inclusion criteria and were included in the analysis.
The characteristics of the included articles are provided inTable 1. These articles were published between 1989 and 2016. Twenty-one articles corresponded to case-control studies (Slattery et al., 2000;
Fernandez et al., 1997;Le Marchand et al., 1999;Huang et al., 2004;
Diergaarde et al., 2007;Tavani et al., 1998;Kune et al., 1989;Kato
et al., 1990;Caan et al., 1998; Slattery et al., 2003;Campbell et al.,
2010; , 2007;Joshi et al., 2015;Fernandez et al., 2002;Slattery et al.,
1997a;La Vecchia et al., 1999;Negri et al., 1998;Turati et al., 2011;
Fernandez et al., 2004;Slattery et al., 1997b,c), nine to cohort studies
including six prospective studies (Dashti et al., 2016; Cho et al., 2012; Fuchs et al., 2002; Movahedi et al., 2015; Sellers et al., 1998; Jung
et al., 2014) and three retrospective studies whose recruitment of
subjects with FH of CRC occurred after CRC diagnosis (Kamiza et al.,
2015; Win et al., 2011; Chau et al., 2016), and one to an intervention
study (Mathers et al., 2012). In ten articles the samples were from USA
(Slattery et al., 2000; Le Marchand et al., 1999; Caan et al., 1998;
Slattery et al., 2003; Slattery et al., 1997a; Slattery et al., 1997b; Slattery et al., 1997c; Cho et al., 2012; Fuchs et al., 2002; Sellers et al., 1998); among them, six articles corresponded to the same US case-control study (Slattery et al., 2000; Caan et al., 1998; Slattery et al., 2003; Slattery et al., 1997a; Slattery et al., 1997b; Slattery et al.,
1997c): they included a number of cases varying between 1624 and
2260 and provided results for different factors, excepted two articles which both reported results for Western diet but from two different analyses (Slattery et al., 2000; Slattery et al., 2003); two articles in-cluding women from the Nurses’s Health Study are likely to present an overlap for results on alcoholic beverages (Cho et al., 2012; Fuchs et al., 2002). In nine articles the samples were from Europe (Fernandez et al.,
1997;Diergaarde et al., 2007;Tavani et al., 1998; Fernandez et al.,
2002;La Vecchia et al., 1999;Negri et al., 1998;Turati et al., 2011;
Fernandez et al., 2004;Jung et al., 2014); among them seven articles
were from Italy: for two articles (Fernandez et al., 1997; Fernandez
et al., 2002) sharing the same data set for subjects with FH of CRC, we
identified a potential overlap for a few factors including red meat (cf.
Table 1);five articles corresponded to the same Italian case-control
(Tavani et al., 1998; La Vecchia et al., 1999; Negri et al., 1998; Turati
et al., 2011; Fernandez et al., 2004): they included 1225 or 1953 cases
and provided results for different factors, without potential overlap. The samples were international in seven articles (Campbell et al., 2010; Joshi et al., 2015; Dashti et al., 2016; Movahedi et al., 2015; Win et al.,
2011; Chau et al., 2016; Mathers et al., 2012) and the others were from
Japan (Huang et al., 2004; Kato et al., 1990), Taiwan (Kamiza et al., 2015), Canada (Campbell et al., 2007) or Australia (Kune et al., 1989). Three articles potentially sharing data for Canadian subjects, reported results on BMI for men and women combined (Campbell et al., 2010;
Win et al., 2011) or not (Campbell et al., 2007): an overlap between
results of the two articles whose subjects were recruited by the same registry (Campbell et al., 2010; Win et al., 2011) cannot be excluded. Although two articles may present an overlap for subjects from Hawaii
(Le Marchand et al., 1999; Joshi et al., 2015), their analyses are
dif-ferent.
The risk of colon and rectum cancers combined was estimated in twenty articles, the risk of colon cancer in thirteen articles and the risk of rectal cancer in three articles.
Twenty-four articles provided results for men and women com-bined, six for men and women separately, two for women and one for men only. One article performed subgroup analyses per age category
(Slattery et al., 2000).
Whereas in most articles FH of CRC concernedfirst degree relatives, ten articles used other inclusion criteria, recorded as high/intermediate risk group according to Amsterdam/Bethesda criteria (Campbell et al., 2007), Amsterdam criteria for HNPCC (Diergaarde et al., 2007), car-riers of MMR gene mutations (Joshi et al., 2015; Dashti et al., 2016;
Table 1 (continued) Location, recruitment period, follow-up a
Total cases/ total controls
b Sex, Mean age (y) Cancer site Family history of cancer (degree, number of cases/ controls) Nutritional factor Control for confounding factors (adjustment) Exposure assessment Comment First author, publication year (ref) Case-control studies acid Intervention study International centers (northern Europe, UK, other regions), 1999 –2005, 52.7 months, 4699 –37960 person-years 53/865 M + F, controls: 44.2, intervention: 45.3 CRC Yes (individuals with Lynch syndrome, 53/ 865) Resistant starch Sex and duration of aspirin taken Intervention group (30 g/ day resistant starch) and placebo group Randomized and controlled trial. Small number of cases. CAPP2 study ( Mathers et al., 2012 ) NA: not available; CRC: colorectal cancer, CC: colon cancer, RC: rectal cancer; FH, Family History; NSP: non-starch polysaccharides; M: males, F: fe males; NHS: Nurses ’s Health Study; HFPS: Health Professionals Follow-up Study. a For prospective studies only. b Including subjects without family history, in some studies.
Kamiza et al., 2015; Win et al., 2011), or individuals with LS (Movahedi
et al., 2015; Jung et al., 2014; Chau et al., 2016; Mathers et al., 2012).
The number of cases in subjects with FH of CRC varied between 11–878 across articles, representing 1–100% of total cases. In four articles, this information was not available.
Nineteen articles analyzed the risk of cancer associated with the exposure to nutritional factors in subjects with FH of CRC (Fernandez
et al., 1997;Huang et al., 2004;Diergaarde et al., 2007;Tavani et al.,
1998;Caan et al., 1998;Campbell et al., 2010;Campbell et al., 2007;
Joshi et al., 2015; Fernandez et al., 2002; Slattery et al., 1997a; La
Vecchia et al., 1999;Negri et al., 1998;Dashti et al., 2016;Cho et al.,
2012;Movahedi et al., 2015;Sellers et al., 1998;Kamiza et al., 2015;
Win et al., 2011;Mathers et al., 2012), among which eleven articles
also compared the results with those obtained in subjects without FH of CRC. Twelve articles (Slattery et al., 2000; Le Marchand et al., 1999; Kune et al., 1989; Kato et al., 1990; Slattery et al., 2003; Fernandez et al., 2002; Turati et al., 2011; Fernandez et al., 2004; Slattery et al.,
1997b; Slattery et al., 1997c; Cho et al., 2012; Fuchs et al., 2002)
ex-amined the modification of cancer risk associated with the combination of nutritional factors and FH of CRC as compared to subjects without FH whose exposure is null or low, termed as “non-exposed” in the fol-lowing sections.
Since the limited number and heterogeneity of articles regarding study design, nutritional factors studied, exposure categories, reference group and data analysis did not allow a quantitative evaluation by meta-analysis, a qualitative synthesis was performed. Main results of the selected articles, for the nutritional factors already known to modify CRC risk in the general population, are presented inTable 2(nineteen articles corresponding to case-control studies, seven to cohort studies and one to an intervention study). Results for various other factors can be found in Supplemental Table 1.
3.2. Risk factors
The nutritional factors known to increase CRC risk in the general population with a level of evidence graded as convincing or probable are alcoholic beverages, red and processed meat, and overweight and obesity (WCRF/AICR, 2011; Latino-Martel et al., 2016; WCRF/AICR,
2017).
3.2.1. Alcoholic beverages
Eleven articles, seven from case-control studies (Fernandez et al.,
1997;Le Marchand et al., 1999;Huang et al., 2004;Diergaarde et al.,
2007;Tavani et al., 1998;Kune et al., 1989;Kato et al., 1990), one from
a retrospective cohort study on (Kamiza et al., 2015) and three from prospective cohort studies (Dashti et al., 2016; Cho et al., 2012; Fuchs
et al., 2002) provided results on alcoholic beverages (beer drinking),
alcohol drinking or alcohol intake (in g/day) and CRC risk. Two studies concerned MMR gene mutation carriers (Kamiza et al., 2015; Dashti
et al., 2016). As previously mentioned, a potential overlap between two
articles has been identified (Cho et al., 2012; Fuchs et al., 2002). Among the seven articles with analyses on subjects with FH of CRC, five (Fernandez et al., 1997;Huang et al., 2004;Diergaarde et al., 2007;
Tavani et al., 1998; Kamiza et al., 2015) reported no association
be-tween alcohol consumption and CRC risk and two (Dashti et al., 2016;
Cho et al., 2012) reported an increased risk. Three of them also
men-tioned results obtained in subjects without FH of CRC: no association
(Tavani et al., 1998; Cho et al., 2012) and an increase of risk (Huang
et al., 2004) for current drinkers (versus never drinkers).
The modification of cancer risk associated with the combination of alcohol and FH of CRC as compared to non-exposed subjects without FH, was examined infive articles (Le Marchand et al., 1999; Kune et al.,
1989; Kato et al., 1990; Cho et al., 2012; Fuchs et al., 2002): the
combination was associated with a significant increased risk for higher beer drinking (Kune et al., 1989; Kato et al., 1990). It was associated with a significant increased risk for higher alcohol intake in both gender
(Cho et al., 2012) and in men (Le Marchand et al., 1999); in women the
increase was significant in one study (Fuchs et al., 2002) and borderline significant in another one (Le Marchand et al., 1999). Three of these articles (Kune et al., 1989; Cho et al., 2012; Fuchs et al., 2002) men-tioned that the risk of CRC observed for the combination was higher than the one observed for FH alone. A significant interaction between FH of CRC, alcohol and risk of CRC was reported in men only (Le
Marchand et al., 1999) and in women (Fuchs et al., 2002). All five
articles also mentioned results obtained in subjects without FH of CRC: an increase of cancer risk in two articles (Kune et al., 1989; Kato et al., 1990), a borderline significant increase in men in one article (Huang
et al., 2004) and no association in two articles (Cho et al., 2012; Fuchs
et al., 2002).
3.2.2. Overweight and obesity
Eight articles,five from case-control studies (Le Marchand et al., 1999; Caan et al., 1998; Slattery et al., 2003; Campbell et al., 2010;
Campbell et al., 2007), one from a retrospective cohort study (Win
et al., 2011) and one from a prospective cohort study (Movahedi et al.,
2015) provided results on BMI (by category or dose-response at dif-ferent ages or periods) and CRC risk. One article concerned subjects with high/intermediate risk, according to Amsterdam/Besthesda cri-teria (Campbell et al., 2007), one included MMR gene mutation carriers
(Win et al., 2011) and one individuals with LS (Movahedi et al., 2015).
As previously mentioned, an overlap between samples of two articles
(Campbell et al., 2010; Win et al., 2011) cannot be excluded.
The analyses concerned subjects with FH of CRC infive articles: in three studies, higher BMI at age 20 y (Win et al., 2011) or 30–54 y
(Caan et al., 1998) or at recruitment (Movahedi et al., 2015) was
sig-nificantly associated with CRC risk; in one study the association was significant for three indicators (BMI 2 years ago, BMI at age 20 y or weight gain since age 20 y) in men but not in women (Campbell et al., 2007), whereas no association was observed for recent BMI with a smaller number of men and women combined (Campbell et al., 2010). Similar results were reported in subjects without FH (Caan et al., 1998;
Campbell et al., 2010; Win et al., 2011).
The modification of cancer risk associated with the combination of higher BMI and FH of CRC as compared to non-exposed subjects without FH, was examined in two studies (Le Marchand et al., 1999;
Slattery et al., 2003): the combination was associated with a significant
increased risk in one study (Slattery et al., 2003), and in men but not in women in the other study (Le Marchand et al., 1999). Both studies reported similar results in subjects without FH, the size of the effect being smaller.
3.2.3. Red and processed meat
Nine articles, seven from case-control studies (Slattery et al., 2000;
Fernandez et al., 1997;Le Marchand et al., 1999;Huang et al., 2004;
Diergaarde et al., 2007;Joshi et al., 2015;Fernandez et al., 2002), one
from a retrospective cohort study (Kamiza et al., 2015) and one from a women prospective cohort study (Sellers et al., 1998), provided results on meat, especially red meat and processed meat and CRC risk. One study concerned subjects fulfilling the Amsterdam criteria for HNPCC
(Diergaarde et al., 2007) and two MMR gene mutation carriers (Joshi
et al., 2015; Kamiza et al., 2015). An overlap has been identified
be-tween samples of two articles on red meat (Fernandez et al., 1997;
Fernandez et al., 2002): since the exposure is expressed in portions per
week in the more recent study (Fernandez et al., 2002) whereas ex-pressed as high versus low in the other one (Fernandez et al., 1997), only results from this recent study (Fernandez et al., 2002) are pre-sented.
The analyses concerned subjects with FH of CRC in six articles: in one case-control study (Fernandez et al., 2002), high consumption of red meat was significantly associated with an increased risk of CRC, whereas for ham and canned meat, the risk increase was borderline significant; in one retrospective cohort study (Joshi et al., 2015), a
Table 2 Odds ratios, relative risks or hazard ratios from studies reporting associations between nutritional factors and colorectal cancer risk in populat ions with FH of colorectal cancer. First author, publication year (ref) Family history of cancer (number of cases) Nutritional factor Comparison Sub-group OR, HR or RR (95% CI) Alcoholic beverages ( Kune et al., 1989 ) FH (11) combined Drink beer No (FH) vs No (no FH) 1.07 (0.4 –3.1) Yes (FH) vs No (no FH) 10.5 (2.0 –55) No FH (305) Drink beer Yes vs No 1.50 (1.1 –2.1) ( Kato et al., 1990 ) FH (NA) combined Drink beer + low/moderate occupation level Yes (FH) vs No (no FH) CC 10.07 (4.77 –21.29) RC 6.69 (3.12 –14.36) No FH (NA) Drink beer + low/moderate occupation level Yes vs No CC 2.44 (1.84 –3.23) RC 2.14 (1.65 –2.78) ( Fernandez et al., 1997 ) FH (112) Alcohol > 4 vs 0 drinks/day 0.8 (0.3 –2.0) ( Tavani et al., 1998 ) FH (NA) Alcohol High vs Never drinker 0.97 (0.33 –2.88) No FH (NA) Alcohol High vs Never drinker 0.95 (0.76 –1.18) ( Le Marchand et al., 1999 ) FH (M:90 –F:72) combined Ethanol > 8.7 (FH) vs < 0.4 (no FH) g/day M 5.7 (2.4 –13.2) > 0.13 (FH) vs < 0 (no FH) g/day F 2.7 (1.0 –6.9) Signi fi cant interaction between FH and ethanol (p = 0.03) No FH (M:603 –F:420) Ethanol > 8.7 vs < 0.4 g/day M 1.2 (0.9 –1.7) > 0.13 vs < 0 g/day F 1.3 (0.9 –2.0) ( Fuchs et al., 2002 ) FH (107) combined Alcohol 0 (FH) vs 0 (no FH) g/day 1.91 (1.32 –2.85) > 30 (FH) vs 0 (no FH) g/day 3.79 (2.13 –6.76) Signi fi cant interaction between FH and alcohol (p = 0.004) No FH (428) Alcohol > 30 vs 0 g/day 0.88 (0.56 –1.39) ( Huang et al., 2004 ) FH (124) Alcohol Current vs never drinker 0.89 (0.52 –1.52) No FH (1228) Alcohol Current vs never drinker 1.32 (1.11 –1.57) ( Diergaarde et al., 2007 ) FH (145) Alcohol ≥ 12.8 vs ≤ 2.6 g/day 1.0 (0.5 –2.0) ( Cho et al., 2012 ) FH (424) Alcohol ≥ 30 vs 0 g/day 2.02 (1.30 –3.13) FH (424) combined Alcohol 0 (FH) vs 0 g/day (no FH) 1.38 (1.06 –1.80) ≥ 30 (FH) vs 0 g/day (no FH) 2.80 (2.00 –3.91) No FH (1377) Alcohol ≥ 30 vs 0 g/day 1.23 (0.96 –1.57) ( Kamiza et al., 2015 ) FH (147) Alcohol drinking Ever vs Never 0.92 (0.62 –1.36) ( Dashti et al., 2016 ) FH (769) Average daily ethanol intake from any alcoholic beverage Ever user vs Abstainer 1.55 (1.11 –2.15) >0 to ≤ 14 g vs Abstainer 1.56 (1.11 –2.18) > 14grams to ≤ 28 g vs Abstainer 1.25 (0.77 –2.04) > 28 g vs Abstainer 1.79 (1.12 –2.87) Per 14 g/day 1.02 (0.95 –1.10) Overweight and obesity ( Caan et al., 1998 ) FH (NA) BMI (30 –54y) T3 vs T1 M 7.76 (2.60 –23.10) F 4.85 (2.33 –10.12) No FH (NA) BMI (30 –54y) T3 vs T1 M 1.70 (1.25 –2.32) F 1.53 (1.22 –1.92) FH (NA) WHR No signi fi cant interaction with FH was found for either age group of men or women ( Le Marchand et al., 1999 ) FH (M:90 –F:72) combined BMI 5 years ago > 25.4 vs < 23.0 M 3.6 (1.7 –7.6) > 24.6 vs < 21.3 F 1.7 (0.7 –4.6) No FH (M:603 –F:420) BMI 5 years ago > 25.4 vs < 23.0 M 2.2 (1.5 –3.1) > 24.6 vs < 21.3 F 1.3 (0.8 –1.9) ( Slattery et al., 2003 ) FH (299) combined BMI > 29 (FH) vs < 25 (no FH) 2.76 (1.94 –3.92) No FH (1978) BMI > 29 vs < 25 1.27 (1.09 –1.47) ( Campbell et al., 2007 ) FH (878) BMI 2 y ago ≥ 30 vs 18.5 –24.99 F 0.85 (0.62 –1.16) M 1.83 (1.33 –2.51) BMI at age 20 y ≥ 30 vs 18.5 –24.99 F 0.81 (0.38 –1.72) M 1.92 (1.02 –3.63) Weight gain since age 20 y ≥ 21 vs 1– 5 kg F 1.21 (0.85 –1.73) M 1.72 (1.22 –2.41) (continued on next page)
Table 2 (continued) First author, publication year (ref) Family history of cancer (number of cases) Nutritional factor Comparison Sub-group OR, HR or RR (95% CI) No FH (1141) BMI 2 y ago ≥ 30 vs 18.5 –24.99 F 0.93 (0.70 –1.24) M 1.74 (1.28 –2.36) BMI at age 20 y ≥ 30 vs 18.5 –24.99 F 1.04 (0.50 –2.15) M 1.78 (0.90 –3.53) Weight gain since age 20 y ≥ 21 vs 1– 5 kg F 1.05 (0.76 –1.44) M 1.65 (1.18 –2.29) ( Campbell et al., 2010 ) FH (80) BMI (recent) ≥ 30 vs 18.5 –24.99 kg/m 2 0.85 (0.29 –2.47) Per 5 kg/m 2 0.90 (0.60 –1.33) No FH (108) BMI (recent) ≥ 30 vs 18.5 –24.99 kg/m 2 0.68 (0.30 –1.55) Per 5 kg/m 2 0.95 (0.70 –1.31) ( Win et al., 2011 ) MMR gene mutation carriers (659) BMI at age 20 years Per 5 kg/m 2 1.30 (1.08 –1.58) Obese vs Normal 2.35 (1.30 –4.23) Non-carriers (36) BMI at age 20 years Per 5 kg/m 2 1.64 (1.02 –2.64) Obese vs Normal 3.00 (0.60 –14.97) ( Movahedi et al., 2015 ) FH (54) BMI Per unit of BMI 1.07 (1.02 –1.13) Overweight (25 –29.99) vs < 25 1.09 (0.57 –2.11) Obese (≥ 30) vs < 25 2.34 (1.17 –4.67) Red and processed meat ( Fernandez et al., 1997 ) FH (112) Red meat High vs Low 2.9 (1.4 –6.0) Raw ham High vs Low 2.1 (0.9 –4.9) Ham High vs Low 2.6 (1.0 –6.8) Canned meat Intermediate/high vs Low 1.9 (1.0 –3.3) ( Sellers et al., 1998 ) FH (61) Red meat > 7 vs ≤ 3.5 servings/week 1.0 (0.5 –2.1) Total meat > 15 vs ≤ 10 servings/week 0.8 (0.4 –1.8) Nitrate meat > 1.5 vs ≤ 0.5 servings/week 0.8 (0.4 –1.6) No FH (180) Red meat > 7 vs ≤ 3.5 servings/week 1.3 (0.8 –2.0) Total meat > 15 vs ≤ 10 servings/week 1.2 (0.8 –1.9) Nitrate meat > 1.5 vs 0.5 servings/week 1.0 (0.7 –1.4) ( Le Marchand et al., 1999 ) FH (M:90 –F:72) combined Beef > 32.3 (FH) vs < 13.8 (no FH) g/ day M 7.6 (3.0 –19.0) > 25.2 (FH) vs < 11.9 (no FH) g/ day F 2.6 (1.0 –6.8) Processed meats High (FH) vs < Low (no FH) M 4.5 (2.1 –9.4) High (FH) vs < Low (no FH) F 1.5 (0.6 –3.7) Signi fi cant interaction between FH and beef (p = 0.03) No FH (M:603 –F:420) Beef > 32.3 vs < 13.8 g/day M 1.4 (0.9 –2.1) > 25.2 vs < 11.9 g/day F 0.9 (0.6 –1.4) Processed meats High vs Low M 1.8 (1.2 –2.7) High vs Low F 1.3 (0.8 –2.1) ( Slattery et al., 2000 ) FH (266) combined Red meat High (FH) vs Low (no FH) Y 2.8 (1.0 –7.8) I 2.4 (0.9 –6.1) O 1.7 (0.9 –3.1) Processed meat High (FH) vs Low (no FH) Y 7.5 (2.0 –28.1) I 2.5 (1.2 –5.4) O 1.3 (0.9 –2.3) Fast-food meat High (FH) vs Low (no FH) Y 2.1 (0.7 –5.9) I 1.8 (0.8 –4.0) O 1.3 (0.7 –2.3) No FH (1358) Red meat High vs Low Y 0.7 (0.4 –1.3) I 1.1 (0.7 –1.7) O 1.2 (0.8 –1.6) Processed meat High vs Low Y 1.0 (0.6 –1.8) I 1.3 (0.9 –2.0) O 1.2 (0.9 –1.6) (continued on next page)
Table 2 (continued) First author, publication year (ref) Family history of cancer (number of cases) Nutritional factor Comparison Sub-group OR, HR or RR (95% CI) Fast-food meat High vs Low Y 0.9 (0.5 –1.5) I 1.0 (0.7 –1.5) O 1.4 (1.0 –1.9) ( Fernandez et al., 2002 ) FH (112) Red meat ≥ 5v s ≤ 2 portions/week 3.2 (1.4 –6.9) No FH (1472) Red meat ≥ 5v s ≤ 2 portions/week 1.8 (1.5 –2.2) ( Huang et al., 2004 ) FH (124) Beef ≥ 3 vs < 3 times/week 0.97 (0.51 –1.85) Sausage ≥ 3 vs < 3 times/week 0.86 (0.43 –1.72) Pork ≥ 3 vs < 3 times/week 1.15 (0.61 –2.15) No FH (1228) Beef ≥ 3 vs < 3 times/week 0.94 (0.76 –1.17) Sausage ≥ 3 vs < 3 times/week 1.11 (0.92 –1.34) Pork ≥ 3 vs < 3 times/week 1.07 (0.88 –1.31) ( Diergaarde et al., 2007 ) FH (145) Total meat ≥ 114.8 vs ≤ 73.3 g/day 1.0 (0.5 –2.1) Red meat ≥ 71.4 vs ≤ 46.2 g/day 0.8 (0.4 –1.6) ( Joshi et al., 2015 ) FH (243) Total non-processed red meat 10.81 –16.04 vs 0– 10.81 g/ 1000Kcal/day 0.9 (0.6 –1.4) 16.04 –21.11 vs 0– 10.81 g/ 1000Kcal/day 1.1 (0.7 –1.7) 21.12 –28.19 vs 0– 10.81 g/ 1000Kcal/day 1.0 (0.6 –1.5) 28.19 –102.43 vs 0– 10.81 g/ 1000Kcal/day 1.0 (0.6 –1.7) Processed red meat 3.97 –6.74 vs 0– 3.97 g/1000 kcal/ day 1.3 (0.9 –2.0) 6.75 –9.53 vs 0– 3.97 g/1000 kcal/ day 0.9 (0.5 –1.3) 9.53 –13.86 vs 0– 3.97 g/1000 kcal/ day 0.8 (0.5 –1.3) 13.87 –122.42 vs 0– 3.97 g/ 1000 kcal/day 1.1 (0.7 –1.8) Total processed meat (red meat + poultry) 4.43 –7.35 vs 0– 4.43 g/1000 kcal/ day 1.3 (0.8 –1.9) 7.36 –10.62 vs 0– 4.43 g/1000 kcal/ day 1.1 (0.7 –1.8) 10.63 –15.29 vs 0– 4.43 g/ 1000 kcal/day 0.8 (0.5 –1.3) 15.29 –152.04 vs 0– 4.43 g/ 1000 kcal/day 1.2 (0.7 –1.8) Total pan-fried meat intake 0.05 –0.12 vs 0– 0.05 g/1000 kcal/ day 1.3 (0.9 –2.0) 0.12 –0.21 vs 0– 0.05 g/1000 kcal/ day 1.3 (0.9 –2.0) 0.21 –5.96 vs 0– 0.05 g/1000 kcal/ day 1.5 (1.0 –2.3) Total oven-broiled meat intake 0.01 –0.05 vs 0– 0 g/1000 kcal/day 0.8 (0.5 –1.2) 0.05 –0.1 vs 0– 0 g/1000 kcal/day 0.8 (0.5 –1.2) 0.1 –3.97 vs 0– 0 g/1000 kcal/day 1.3 (0.9 –1.8) Total grilled meat intake 0.01 –0.07 vs 0– 0 g/1000Kcal/day 1.0 (0.7 –1.5) 0.07 –0.15 vs 0– 0 g/1000Kcal/day 1.1 (0.7 –1.6) 0.15 –4.97 vs 0– 0 g/1000Kcal/day 1.0 (0.6 –1.4) FH (876) Total non-processed red meat 10.81 –16.04 vs 0– 10.81 g/ 1000Kcal/day 0.8 (0.6 –1.0) 16.04 –21.11 vs 0– 10.81 g/ 1000Kcal/day 1.0 (0.8 –1.3) 21.12 –28.19 vs 0– 10.81 g/ 0.8 (0.7 –1.1) (continued on next page)
Table 2 (continued) First author, publication year (ref) Family history of cancer (number of cases) Nutritional factor Comparison Sub-group OR, HR or RR (95% CI) 1000Kcal/day 28.19 –102.43 vs 0– 10.81 g/ 1000Kcal/day 0.8 (0.6 –1.0) Processed red meat 3.97 –6.74 vs 0– 3.97 g/1000 kcal/ day 0.9 (0.7 –1.2) 6.75 –9.53 vs 0– 3.97 g/1000 kcal/ day 0.9 (0.7 –1.1) 9.53 –13.86 vs 0– 3.97 g/1000 kcal/ day 1.1 (0.8 –1.4) 13.87 –122.42 vs 0– 3.97 g/ 1000 kcal/day 1.1 (0.8 –1.4) Total processed meat (red meat + poultry) 4.43 –7.35 vs 0– 4.43 g/1000 kcal/ day 1.0 (0.8 –1.3) 7.36 –10.62 vs 0– 4.43 g/1000 kcal/ day 0.9 (0.7 –1.2) 10.63 –15.29 vs 0– 4.43 g/ 1000 kcal/day 1.1 (0.9 –1.4) 15.29 –152.04 vs 0– 4.43 g/ 1000 kcal/day 1.2 (0.9 –1.5) Total pan-fried meat intake 0.05 –0.12 vs 0– 0.05 g/1000 kcal/ day 1.0 (0.8 –1.3) 0.12 –0.21 vs 0– 0.05 g/1000 kcal/ day 1.0 (0.8 –1.3) 0.21 –5.96 vs 0– 0.05 g/1000 kcal/ day 1.1 (0.8 –1.3) Total oven-broiled meat intake 0.01 –0.05 vs 0– 0 g/1000 kcal/day 1.1 (0.9 –1.3) 0.05 –0.1 vs 0– 0 g/1000 kcal/day 1.0 (0.8 –1.2) 0.1 –3.97 vs 0– 0 g/1000 kcal/day 1.1 (0.9 –1.4) Total grilled meat intake 0.01 –0.07 vs 0– 0 g/1000Kcal/day 0.9 (0.8 –1.2) 0.07 –0.15 vs 0– 0 g/1000Kcal/day 0.9 (0.7 –1.1) 0.15 –4.97 vs 0– 0 g/1000Kcal/day 1.0 (0.8 –1.2) ( Kamiza et al., 2015 ) FH (147) Meat intake Medium vs Low 1.10 (0.75 –1.62) High vs Low 0.99 (0.65 –1.52) Physical activity ( Slattery et al., 1997a ) FH (NA) Lifetime vigorous activity level High vs None 0.89 (0.52 –1.53) No FH (NA) Lifetime vigorous activity level High vs None 0.59 (0.49 –0.72) ( Le Marchand et al., 1999 ) FH (M:90 –F:72) combined Lifetime recreational physical activity > 7632 (FH) vs < 1152 (no FH) h M 2.6 (1.2 –5.5) > 1152 (FH) vs 0 (no FH) h F 2.1 (0.9 –4.8) No FH (M:603 –F:420) Lifetime recreational physical activity > 7632 vs < 1152 h M 0.6 (0.4 –0.8) > 1152 vs 0 h F 0.7 (0.5 –1.1) ( La Vecchia et al., 1999 ) FH (134) Physical activity Low vs high 1.05 (0.53 –2.07) No FH (1091) Physical activity Low vs high 1.46 (1.20 –1.77) ( Slattery et al., 2003 ) FH (299) combined Physical activity Low (FH) vs High (no FH) 2.64 (1.78 –3.92) No FH (1978) Physical activity Low vs High 1.71 (1.42 –2.05) ( Huang et al., 2004 ) FH (124) Physical exercise ≥ 3 vs < 3 times/month 0.96 (0.64 –1.44) No FH (1228) Physical exercise ≥ 3 vs < 3 times/month 0.76 (0.66 –0.87) ( Kamiza et al., 2015 ) FH (147) Regular physical activity Yes vs No 0.58 (0.40 –0.86) Dietary fi ber ( Negri et al., 1998 ) FH (NA) Total fi ber 80th vs 20th percentiles 0.77 (0.43 –1.38) No FH (NA) Total fi ber 80th vs 20th percentiles 0.66 (0.58 –0.76) ( Sellers et al., 1998 ) FH (61) Dietary fi ber > 22.59 vs ≤ 16.17 g/day 1.2 (0.6 –2.6) No FH (180) Dietary fi ber > 22.59 vs ≤ 16.17 g/day 0.8 (0.5 –1.2) ( Le Marchand et al., 1999 ) FH (M:90 –F:72) combined NSP > 20.1 (FH) vs < 13.3 (no FH) g/ M 1.8 (0.8 –4.2) (continued on next page)
Table 2 (continued) First author, publication year (ref) Family history of cancer (number of cases) Nutritional factor Comparison Sub-group OR, HR or RR (95% CI) day > 17.5 (FH) vs < 13.0 (no FH) g/ day F 2.4 (0.9 –6.4) No FH (M:603 –F:420) NSP > 20.1 vs < 13.3 g/day M 0.6 (0.4 –0.9) > 17.5 vs < 13.0 g/day F 0.7 (0.5 –1.1) ( Slattery et al., 2000 ) FH (266) combined Re fi ned grains High (FH) vs Low (no FH) Y 3.8 (1.1 –12.7) I 7.1 (2.4 –20.6) O 1.6 (0.9 –3.2) No FH (1358) Re fi ned grains High vs Low Y 0.9 (0.5 –1.8) I 1.9 (1.2 –3.0) O 1.5 (1.0 –2.2) ( Diergaarde et al., 2007 ) FH (145) Cereals ≥ 73.6 vs ≤ 34.0 g/day 1.4 (0.6 –2.8) Dietary fi ber ≥ 27.5 vs ≤ 22.6 g/day 0.5 (0.2 –1.0) ( Mathers et al., 2012 ) Lynch syndrome (53) Resistant starch 30 g/day vs placebo 1.40 (0.78 –2.56) Dairy products ( Fernandez et al., 1997 ) FH (112) Calcium High vs Low 1.9 (0.9 –4.2) Cheese High vs Low 3.5 (1.3 –9.9) ( Sellers et al., 1998 ) FH (61) Total calcium > 1,296.6 vs ≤ 820.7 mg/day 1.2 (0.6 –2.2) Dietary calcium > 964.7 vs ≤ 615 mg/day 0.8 (0.4 –1.7) Total dairy > 20 vs ≤ 10 servings/week 0.7 (0.4 –1.4) High fat dairy > 9 vs ≤ 4.5 servings/week 0.7 (0.4 –1.3) Low fat dairy > 7 vs ≤ 2.5 servings/week 0.9 (0.5 –1.6) No FH (180) Total calcium > 1,296.6 vs ≤ 820.7 mg/day 0.5 (0.3 –0.7) Dietary calcium > 964.7 vs ≤ 615 mg/day 0.7 (0.4 –1.0) Total dairy > 20 vs ≤ 10 servings/week 0.7 (0.4 –1.0) High fat dairy > 9 vs ≤ 4.5 servings/week 0.9 (0.6 –1.3) Low fat dairy > 7 vs ≤ 2.5 servings/week 0.8 (0.5 –1.1) ( Le Marchand et al., 1999 ) FH (M:90 –F:72) combined Total calcium > 907 (FH) vs < 553 (no FH) mg/ day M 1.4 (0.6 –3.0) > 1055 (FH) vs < 521 (no FH) mg/ day F 1.2 (0.4 –3.2) No FH (M:603 –F:420) Total calcium > 907 vs < 553 mg/day M 0.7 (0.5 –1.0) > 1055 vs < 521 mg/day F 1.1 (0.7 –1.7) ( Huang et al., 2004 ) FH (124) Milk ≥ 1/day vs < 1/day 0.73 (0.50 –1.06) No FH (1228) Milk ≥ 1/day vs < 1/day 0.88 (0.78 –0.99) ( Diergaarde et al., 2007 ) FH (145) Calcium ≥ 1214.9 vs ≤ 972.1 mg/day 0.8 (0.4 –1.6) Dairy products ≥ 457.3 vs ≤ 245.7 g/day 1.6 (0.8 –3.4) Dietary patterns and dietary risk factor scores ( Le Marchand et al., 1999 ) FH (M:90 –F:72) combined Summary variable for lifestyle risk factors b Unhealthy (FH) vs Healthy (no FH) M 11.7 (5.8 –23.9) F 8.3 (4.1 –17.0) No FH (M:603 –F:420) Summary variable for lifestyle risk factors b Unhealthy vs Healthy M 4.8 (3.2 –7.2) F 3.1 (1.9 –5.2) ( Slattery et al., 2000 ) FH (266) combined Western diet High (FH) vs Low (no FH) Y 14.0 (3.9 –50.1) I 7.7 (2.0 –29.1) O 1.6 (0.8 –3.2) Signi fi cant interaction between FH, age and Western diet (p = 0.03) No FH (1358) Western diet High vs Low Y 2.8 (1.4 –5.7) I 1.4 (0.8 –2.3) O 1.5 (1.0 –2.3) ( Fernandez et al., 2002 ) FH (112) combined Adult Life Dietary Risk Score b High (FH) vs Low (no FH) 5.5 (3.5 –8.7) No FH (1472) Adult Life Dietary Risk Score b High vs Low 2.2 (1.9 –2.6) ( Slattery et al., 2003 ) FH (299) combined Prudent diet Low (FH) vs High (no FH) 2.65 (1.69 –4.16) (continued on next page)
Table 2 (continued) First author, publication year (ref) Family history of cancer (number of cases) Nutritional factor Comparison Sub-group OR, HR or RR (95% CI) Western diet High (FH) vs Low (no FH) 3.59 (2.29 –5.62) Signi fi cant interaction between FH and Western diet (p multiplicative = 0.03) No FH (1978) Prudent diet Low vs High 1.17 (0.97 –1.42) Western diet High vs Low 1.45 (1.18 –1.77) ( Fernandez et al., 2004 ) FH (134) combined Risk factor score b High (FH) vs Low (no FH) 7.08 (4.68 –10.71) No FH (1091) Risk factor score b High vs Low 2.27 (1.89 –2.73) ( Turati et al., 2011 ) FH (187) combined ‘Starch –rich ’ dietary pattern High (FH) vs Low (no FH) CRC 4.00 (3.03 –5.27) CC 4.63 (3.39 –6.31) RC 2.93 (1.96 –4.37) ‘Vitamins & fi ber ’ dietary pattern Low (FH) vs High (no FH) CRC 3.74 (2.85 –4.91) CC 3.89 (2.87 –5.27) RC 3.32 (2.24 –4.94) No FH (1766) ‘Starch-rich ’ dietary pattern High vs Low CRC 1.38 (1.19 –1.61) CC 1.42 (1.19 –1.70) RC 1.31 (1.05 –1.64) ‘Vitamins & fi ber ’ dietary pattern Low vs High CRC 1.29 (1.12 –1.48) CC 1.19 (1.01 –1.41) RC 1.49 (1.21 –1.82) Abbreviations;: NA: not available; M: malesF: females, T: tertiles; CC: colon cancerRC: rectal cancer; Y: younger age i.e. ≤ 55 yearsI: intermediate age i.e. 56 –66 years, O: older age i.e. ≥ 67 years; FH: Family History; NSAIDs: non-steroidal anti-in fl ammatory drugs; NSP: non-starch polysaccharides; WHR: waist-to-hip ratio. aFor case-control studies only. b The “Adult life dietary risk score ” included red meatseasoning fats, β -carotene and ascorbic acid consumption, and daily meal frequency ( Fernandez et al., 2002 ). The “Summary variable for lifestyle risk factors ” included 13 lifestyle variables related to nutrition, smoking and physical activity ( Le Marchand et al., 1999 ). The “Risk factor score ” included education level, number of meal/day, occupational physical activity, and fi ber, calcium and β -carotene intakes ( Fernandez et al., 2004 ).
borderline significant increase of CRC risk was associated with pan-fried meat but not with other meat categories (non-processed red meat, processed red meat, total processed meats, oven-broiled meat and grilled meat); in the four other studies, there was no association for beef, pork or sausage studied separately (Huang et al., 2004), red meat
(Diergaarde et al., 2007; Sellers et al., 1998),“nitrate meat” (Sellers
et al., 1998) or total meat (Kamiza et al., 2015). In the three studies
which also mentioned results obtained in subjects without FH of CRC
(Huang et al., 2004; Joshi et al., 2015; Sellers et al., 1998), there was no
association.
The modification of cancer risk associated with the combination of red meat or processed meat consumption and FH of CRC as compared to non-exposed subjects without FH, was examined in two studies
(Slattery et al., 2000; Le Marchand et al., 1999): the combination was
associated with a significant increased risk for beef (Le Marchand et al., 1999), processed meat in men (Le Marchand et al., 1999), and pro-cessed meat in young and intermediate age subjects (Slattery et al., 2000); it was associated with a borderline significant increased risk for red meat in younger subjects (Slattery et al., 2000) and beef in women
(Le Marchand et al., 1999). A significant interaction was reported
be-tween FH of CRC, beef and CRC risk in men (Le Marchand et al., 1999). These studies also mentioned results obtained in subjects without FH of CRC: an increased risk was observed only for processed meat in men (Le
Marchand et al., 1999) and for fast-food meat in older subjects (Slattery
et al., 2000).
3.3. Protective factors
The nutritional factors known to decrease CRC risk in the general population with a level of evidence graded as convincing or probable are physical activity, foods containing dietaryfiber, and milk and dairy products (WCRF/AICR, 2011;Latino-Martel et al., 2016; WCRF/AICR,
2017).
3.3.1. Physical activity
Five case-control articles (Le Marchand et al., 1999; Huang et al., 2004; Slattery et al., 2003; Slattery et al., 1997a; La Vecchia et al., 1999) and one retrospective cohort study on MMR gene mutation car-riers (Kamiza et al., 2015) provided results on physical activity (total, regular, recreational, vigorous activity level, frequency or duration) and CRC risk. An overlap between samples of two articles (Slattery
et al., 2003; Slattery et al., 1997a) could not be excluded (cf.Table 1).
The retrospective cohort study reported a significant decrease of CRC associated with regular physical activity (Kamiza et al., 2015). Three case-control studies (Huang et al., 2004; Slattery et al., 1997a; La
Vecchia et al., 1999) reported no association between physical activity
and CRC risk in subjects with FH of CRC, whereas they mentioned a significant inverse association in subjects without FH of CRC.
The modification of cancer risk associated with the combination of physical activity and FH of CRC as compared to non-exposed subjects without FH, was examined in two studies (Le Marchand et al., 1999;
Slattery et al., 2003): the combination was associated with a significant
decreased risk in one study (Slattery et al., 2003), and surprisingly with an increased risk in men, but not in women, in the other study (Le
Marchand et al., 1999). These studies also mentioned results obtained
in subjects without FH of CRC, i.e. an inverse association in the popu-lation of thefirst study with a lower size of effect (Slattery et al., 2003), and in men but not in women in the second study (Slattery et al., 2003). 3.3.2. Dietaryfiber
Five articles, three from case-control studies (Le Marchand et al.,
1999; Diergaarde et al., 2007; Negri et al., 1998) including one study on
subjects fulfilling the Amsterdam criteria for HNPCC (Diergaarde et al., 2007), one from a women cohort (Sellers et al., 1998) and one from an intervention study on individuals with LS (Mathers et al., 2012), pro-vided results on dietary fiber (total, non-starch polysaccharides or
resistant starch) and CRC risk.
The analyses concerned subjects with FH of CRC in three studies: no association between dietaryfiber and CRC risk was observed in one case-control study (Negri et al., 1998) and the cohort study (Sellers
et al., 1998), whereas a borderline significant decrease was reported in
one case-control study (Diergaarde et al., 2007). No association was observed with cereals (Diergaarde et al., 2007). In the intervention study, a high daily intake of resistant starch (30 g) for up to four years had no effect on CRC risk compared to placebo in subjects with LS
(Mathers et al., 2012).
In one study (Le Marchand et al., 1999) the combination of non-starch polysaccharide and FH of CRC as compared to non-exposed subjects without FH, was not associated with CRC risk, whereas a de-creased risk was observed in subjects without FH, in men but not in women.
Conversely, in one additional case-control study (Slattery et al., 2000), high intake of refined grains combined with FH of CRC increased CRC risk in young and intermediate age subjects.
3.3.3. Dairy products
Five articles, four from case-control studies (Fernandez et al., 1997;
Le Marchand et al., 1999;Huang et al., 2004;Diergaarde et al., 2007)
including one study on subjects fulfilling the Amsterdam criteria for HNPCC (Diergaarde et al., 2007), and one from a women cohort study
(Sellers et al., 1998), provided results on dairy products (total dairy,
milk, cheese, calcium) and CRC risk.
The analyses concerned subjects with FH of CRC in four studies: no association was observed for dairy products (total, high fat or low fat), milk, and total, dietary or supplemental calcium intake (Fernandez
et al., 1997;Huang et al., 2004;Diergaarde et al., 2007;Sellers et al.,
1998); a significant increased risk of CRC was reported in one study for high cheese consumption (Fernandez et al., 1997). Three studies men-tioned also results in subjects without FH: a decreased risk associated with milk (Huang et al., 2004) and total dairy products (Sellers et al., 1998); a borderline significant decreased risk (Sellers et al., 1998) or no association (Le Marchand et al., 1999) for calcium.
In one study (Le Marchand et al., 1999) the combination of total calcium intake and FH of CRC as compared to non-exposed subjects without FH, was not associated with CRC risk; no association with total calcium intake was observed in subjects without FH of CRC.
3.3.4. Dietary patterns and dietary risk factor scores
For the general population, the level of evidence of the association between dietary patterns and CRC risk has been judged as“limited-no conclusion” (WCRF/AICR, 2011).
The present systematic review permitted to identify six articles from case-cohort studies estimating the association between either dietary patterns (Slattery et al., 2000; Slattery et al., 2003; Turati et al., 2011), with a potential overlap between two articles (Slattery et al., 2000;
Slattery et al., 2003) (cf. Table 1), or integrative dietary variables or
scores (Le Marchand et al., 1999; Fernandez et al., 2002; Fernandez
et al., 2004) in combination with FH of CRC, as compared to
non-ex-posed subjects without FH (cf.Table 2): the combination was associated with a strong significant increase of CRC risk (risk estimate comprised between 2.7 and 14, according to studies), for subjects with high compliance with Western (Slattery et al., 2000; Slattery et al., 2003) or starch-rich (Turati et al., 2011) dietary patterns, or low compliance with prudent (Slattery et al., 2000) or vitamins andfiber rich (Turati
et al., 2011) dietary patterns, or for subjects with“unhealthy” dietary
lifestyles estimated by a higher summary variable of lifestyle risk (Le
Marchand et al., 1999), adult life dietary risk score (Fernandez et al.,
2002), or risk factor score (Fernandez et al., 2004). A significant in-teraction between FH of CRC, Western diet and risk of CRC was re-ported in one article (Slattery et al., 2003), and between FH of CRC, age Western diet and risk of colon cancer in another one (Slattery et al., 2000). All studies reported similar results in subjects without FH, the
