Energy Balance Related Lifestyle Factors and Risk of Endometrial and Colorectal Cancer among Individuals with Lynch Syndrome: A Systematic Review

Abstract

Lifestyle factors related to energy balance, such as excess body weight, poor diet, and physical inactivity, are associated with risk of sporadic endometrial cancer (EC) and colorectal cancer (CRC). There are limited data on energy balance-related lifestyle factors and EC or CRC risk among individuals with Lynch syndrome, who are at extraordinarily higher risk of developing EC or CRC. We conducted a systematic review of evidence related to weight status, weight change, dietary habits, and physical activity on EC and CRC risk among individuals with Lynch syndrome. Findings are reported narratively. We searched Medline, EMBASE, CENTRAL, PubMed, and clinicaltrials.gov up to June 14^th, 2018. In total, 1,060 studies were identified and 16 were included. Three studies were related to EC and 13 to CRC. Overall, evidence suggests that weight status/weight change may not be associated with EC risk and multivitamin and folic-acid supplementation may be associated with decreased EC risk. Early-adulthood overweight/obese weight-status and adulthood weight-gain may be associated with increased CRC risk, whereas multivitamin supplementation, tea and high fruit intake, and physical activity may be associated with decreased CRC risk. Current evidence proposes that recommendations related to weight, some dietary habits, and physical activity recommended for the general public are also relevant to individuals with Lynch syndrome. More research is needed, specifically prospective cohorts and randomized controlled trials, to determine if tailored recommendations are needed among individuals with Lynch syndrome.

Introduction

Lynch syndrome is caused by germline mutations in DNA mismatch repair (MMR) genes (MLH1, MSH2, MSH6, and PMS2) and EPCAM. The estimated prevalence of Lynch syndrome in the U.S. population is 1:280 or 1.1 million people [1]. Among individuals with Lynch syndrome, the lifetime risk of developing endometrial cancer (EC) or colorectal cancer (CRC) is high compared to the general population, specifically 40–60% [2–6] and 40–80% [3, 5–7] respectively.

According to the most recent World Cancer Research Fund/American Institute for Cancer Research Continuous Update Project Expert Report (CUP), lifestyle factors specifically related to energy balance, such as excess body weight/body fatness, poor dietary habits, and physical inactivity, are associated with risk of developing sporadic EC and CRC [8, 9]. Strong, convincing evidence indicates that risk of EC and CRC is positively associated with body fatness [8, 9]. Further, strong, probable evidence suggests an inverse relationship between coffee intake and EC risk, and the following dietary factors in relation to CRC risk: wholegrains, high-fiber foods, dairy products and calcium supplements [8, 9]. Additionally, other studies have demonstrated that a Mediterranean dietary pattern is associated with reduced risk of EC [10] and CRC [11]. With regards to increased risk of EC and CRC, the CUP report indicates strong, probable evidence that diets with a high glycemic load are associated with increased EC risk [9]. Regarding CRC risk, there is strong, convincing and strong, probable evidence that processed meat and alcohol intake, and red meat intake, respectively, are linked with increased CRC risk [8]. Furthermore, there is strong, probable and strong, convincing evidence that regular physical activity is associated with decreased risk of EC and CRC respectively [8, 9]. These data provide the foundation for energy balance-related lifestyle interventions to further inform recommendations for prevention of sporadic EC and CRC.

With regards to individuals with Lynch syndrome, there are limited data on whether these energy balance related-lifestyle factors are associated with EC and/or CRC risk. To date, there are no reports synthesizing these data in relation to EC risk. Regarding CRC risk, one literature review, published in 2013, has been conducted pertaining not only to energy balance related lifestyle factors (weight status, dietary intake of foods, and alcohol intake), but also medication/medical supplement use (i.e. aspirin and/or resistant starch) and tobacco use [12]. Since 2013, epidemiological evidence regarding energy balance related lifestyle factors and EC and CRC risk among individuals with Lynch syndrome has grown. Therefore, the purpose of the present study was to systematically review the current state of epidemiological evidence related to the relationship between weight status, weight change, dietary habits, and physical activity behaviors (energy balance related lifestyle factors) and EC and CRC risk among individuals with Lynch syndrome in order to determine if there is adequate support for tailored lifestyle recommendations for this population at genetically higher cancer risk.

Methods

Search Strategy

We identified, screened, and described reports using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [13]. Three authors (AMC, LG, KBE) developed a draft of search terms and then worked with a medical librarian (KK) to further develop the search terms, develop the search strategy, and conduct the search (see search strategy provided in Appendix). All publication types were included in the search in order to capture all of the available evidence related to energy balance related lifestyle factors and EC and CRC risk among individuals with Lynch syndrome. We searched Medline, EMBASE, CENTRAL, PubMed, and clinicaltrials.gov. For Medline and EMBASE we searched from 1946 to June 14, 2018. CENTRAL, PubMed and clinicaltrials.gov searches did not include a time constriction to be as inclusive as possible.

Review Process & Selection Criteria

Two authors (AMC and LG) conducted both round-one and -two reviews. During round-one review, both authors independently reviewed the titles and abstracts of all results to determine eligibility. A meeting was held to review the results. Discrepancies between reviews were discussed at the time of the meeting and based on this discussion both authors came to a resolution. A third author (KBE) was available in the event of no resolution; however, both authors were able to resolve conclusively after discussion. Next, a full-text review of the selected studies from round-one were reviewed independently for round-two. The authors met on a second occasion to review the results and followed the same procedures as described for round-one.

Studies were considered eligible for the following reasons: assessed individuals with Lynch syndrome; assessed individuals who were confirmed carriers of gene mutations related to Lynch syndrome; evaluated weight status, weight change, dietary habits (as defined by dietary patterns, food intake, multivitamin and mineral supplementation, alcohol consumption), or physical activity in relation to EC or CRC risk; published in English; and included human subjects only. Among the CRC studies, studies assessing adenomas as an outcome were also included considering adenomas are a precursor for CRC. Published abstracts and personal communications were included in this review in order to facilitate a comprehensive review of the current body of evidence and knowledge pertaining to the impact of energy balance related lifestyle factors on EC and CRC risk among individuals with Lynch syndrome.

Data Extraction

We extracted data in the coding form recommended by PRISMA [13]. The following information was collected from identified peer-reviewed manuscripts, personal communications, and published abstracts: author, year, publication type, patient characteristics, study design, data source (i.e. national or international registry), lifestyle factor assessed (i.e. weight status, weight change, dietary habits, or physical activity), measurement methods of the lifestyle factor assessed, and major findings.

Assessment of Risk of Bias

A risk of bias assessment was conducted based on a three-item checklist to identify studies as either low or high risk [14, 15]. Briefly, the first item assessed study setting, where cohort studies and population-based case-control studies were considered low risk of bias and hospital-based or cancer-registry based case-control studies were considered high risk [14, 15]. The second item assessed the validity and/or reliability of the instrument used to measure the energy balance factor assessed in the study [14]. Studies with low risk either reported the validity and/or reliability of the questionnaire/method used or provided a reference for the known valid and reliable instrument used, whereas studies with high risk did not provide any information to demonstrate validity and reliability. The third item assessed whether or not results were adjusted for potential confounding effects [14]. Studies that reported on rationale of adjusting for potential confounding effects were considered low risk and studies that did not consider adjustment for confounding effects or did not report the confounding effects used in adjustments were considered high risk.

Analysis

Data were synthesized qualitatively in narrative form. A meta-analysis was not conducted due to the heterogeneity in study design and outcome measures, and limited number of studies assessing each lifestyle factor in relation to EC and CRC risk. We limited our summary of results to the three main energy balance related lifestyle factors, weight status/weight gain, dietary habits, and physical activity, in relation to EC and CRC risk.

Results

Literature Search

We identified 1,060 records through our database search. After duplicates were removed, 584 records remained. A total of 585 records were screened via round-one review. A total of 539 were excluded during this process, leaving 46 records to assess for eligibility in round-two review. Of the 46 records assessed in round-two review, 31 records were excluded. Major reasons for exclusion include articles that did not address Lynch syndrome or genetic mutations associated with Lynch syndrome, and articles that did not include assessment of weight status, weight change, dietary habits, or physical activity in relation to EC or CRC risk. One record was identified by personal communication, that was not included in our database search results. In all, 16 records met the inclusion criteria and were included in this systematic review. Of note, our search strategy did not populate any randomized controlled trials that assessed our energy balance related lifestyle factors of interest (weight status/weight change, dietary habits, physical activity) in relation to EC and CRC risk. Figure 1 contains the PRISMA flow diagram.

Fig. 1.

PRISMA Flow Diagram

MMR = mismatch repair; CRC = colorectal cancer; EC = endometrial cancer

Study Characteristics for Studies Related to Endometrial Cancer

Three studies were related to EC (one peer-reviewed manuscript and two abstracts) and were published between 2011 and 2016 [16–18]. Three studies assessed weight status or weight change [16–18], and one of the studies that assessed weight status also assessed dietary habits [17]. Two studies were prospective [17, 18], and one was retrospective [16]. Additionally, two studies utilized data from the Colon Cancer Family Registry [17, 18], and one study did not specify the data source [16]. Table 1 summarizes characteristics of the three studies pertaining to energy balance related lifestyle factors and endometrial risk.

Table 1:

Energy balance related lifestyle factors and endometrial cancer risk among women with Lynch syndrome

Author, Year, Publication Type	Patient Characteristics	Study Design & Data Source	Lifestyle Factor	Measurement Method	Findings
Maenpaa JU et al, 2016 Abstract	Lynch syndrome (n=136) -cases with EC (n=50)	Retrospective cohort N/A	Weight change via BMI	Self-reported height and weight via mailed questionnaire -present weight -weight at age 18 and 40	BMI and EC incidence: --age 18: HR, 1.03 (0.91,1.17) --age 40: HR, 1.04 (0.98,1.11) --present: HR, 1.02 (0.97,1.08) **Statistical adjustment not specified
Win AK et al, 2011 Manuscript	Lynch syndrome (n=601) -MLH1 (n=245) -MSH2 (n=299) -MSH6 (n=38) -PMS2 (n=19) *126 cases of EC Without Lynch syndrome (n=533) *8 cases of EC	Prospective cohort Colon Cancer Family Registry	Weight in early adulthood via BMI	Self-reported height and weight at age 18–20 via in- person interviews, telephone interviews, or mailed study specific questionnaires Follow-up approximately 5 years after baseline data collected	BMI (per 5 kg/m2) at age 20 and EC risk bv MMR mutation: --MLH1, HR 0.83 (0.41,1.69) --MSH2, HR 0.74 (0.31,1.76) --MSH6, HR 0.68 (0.07,7.16) --PMS2, HR 0.22 (0.01,13.1) BMI (per 5 kg/m2) at age 20 and EC risk in individuals with LS: *Nomial weight as reference group -unadjusted HR, 0.67 (0.38,1.20) -adjusted HR, 0.73 (0.40,1.30) BMI (per 5 kg/m2) at age 20 and EC risk in individuals with LS: *Nomial weight as reference group -unadjusted HR, 1.58 (1.0,2.48) -adjusted HR, 1.74 (1.27,2.37) **Data adjusted for age at menarche, year of birth, hormonal contraceptive use, cigarette smoking at 20 years old; MMR gene in carriers
Win AK et al, 2014 Abstract	Lynch syndrome (n=976) -MLH1 (n=361) -MSH2 (n=472) -MSH6 (n=90) -PMS2 (n=53)	Prospective cohort Colon Cancer Family Registry	Dietary habits Current weight status via BMI	Not specified Time to follow-up: 41,710 person-years of observation from birth	LS mutation carriers diagnosed with EC (n=140) Calcium, Alcohol, BMI and EC risk: -No association observed Ever use of MVI or Folic acid supplementation: -MVI: HR 0.38 (0.22–0.65) -folic acid supplementation: HR 0.22 (0.08–0.58) **Data adjusted for “measured potential confounders”

LS = Lynch Syndrome, N/A = Not Applicable, MVI = Multivitamin

Study Characteristics for Studies Related to Colorectal Cancer

A total of 13 studies were related to CRC and were published between 2004 and 2018 [7, 19–30]. Three studies assessed weight status or weight change [19, 21, 27], nine studies assessed dietary habits [7, 20, 22, 24, 25, 28–31], and two studies assessed physical activity [23, 25]. Of note, one of the studies that assessed dietary habits also assessed physical activity [25]. Out of the 13 total studies, five were prospective [19, 20, 24, 27, 31], six were retrospective [22, 23, 25, 28–30], and two were case-control [7, 21]. Additionally, four studies utilized data from the GEOLynch Cohort [19, 20, 24, 31], four from the Colon Cancer Family Registry [22, 23, 27, 28, 31], one from Ontario Cancer Registry and Newfoundland Cancer Treatment and Research Foundation population tumor registry [21], one from Taiwan Hereditary Nonpolyposis Colorectal Cancer Consortium [25], one from the Dutch Case-Control Study [7], one from the Hereditary Nonpolyposis Colorectal Cancer Registry and Genetic Testing Project of the Japanese Society for Cancer of the Colon and Rectum [29], and one from the Hereditary Cancer Institute at Creighton University [30]. Table 2 summarizes characteristics of the 13 studies pertaining to energy balance related lifestyle factors and CRC risk.

Table 2:

Energy balance related lifestyle factors and colorectal cancer risk among individuals with Lynch syndrome

Author, Year, Publication Type	Patient Characteristics	Study Design, Data Source	Lifestyle Factor	Measurement Method	Findings
Botma A et al, 2010 Manuscript	Lynch syndrome without adenoma (n=243) -MLH1 (n=80) -MSH2 (n=102) -MSH6 (n=59) -PMS2 (n=l) Lynch syndrome with adenoma (n=243) -MLH1 (n=107) -MSH2 (n=92) -MSH6 (n=41) -PMS2 (n=2)	Prospective cohort GEOLynch Study	Weight change via BMI	Self-reported height and weight collected from selfadministered questionnaire -current weight -weight 2 years before study entry (2-year weight gain) -weight at 18 and 40 years of age (adult weight gain) Median follow-up of 20 months	With and without adenoma (n=486): --risk for adenomatous polyps increased in obese based on current weight HR 2.04 (1.08,3.86) --association driven by HR in men, HR 2.94 (1.11, 7.78) as opposed to women HR 0.99 (0.27, 3.57) Without adenoma: Men & Women- 22 individuals without adenoma developed colorectal adenomas --these patients had a higher median current BMI compared to total cohort Men- --overweight or obese current BMI, HR 8.72 (2.06, 36.96) --per 5 kg/m2 increase in BMI, HR 1.84 (1.13, 3.02) --per 5 kg increase in weight, HR 1.17 (1.00,1.37) --2-year weight gain > 2 kg, HR 1.71 (0.30, 9.78) -adult weight gain, HR 3.60 (0.38, 34.28) Women- --No significant associations between current weight and adenoma in women --2-year weight gain > 2 kg, HR 0.74 (0.10,5.64) --adult weight gain, HR 0.83 (0.20, 3.48) With adenoma: Men & Women- 36 individuals with adenoma developed more colorectal adenomas --no differences observed in current BMI between these patients and total prevalence cohort (p>0.05) Men- --no significant associations with current weight and adenomatous polyps -2-year 2 kg weight gain, HR 1.73 (0.67, 4.45) --adult weight gain, HR 2.2 (0.54, 9.67) Women- --no significant associations with current weight and adenomatous polyps --2-year weight gain > 2 kg, HR 4.09 (1.04,16.19) --adult weight gain, HR 1.53 (0.37, 6.29) **Data adjusted for age, sex, extent of colon surgery, smoking habits, alcohol intake
Botma A et al, 2013 Manuscript	Lynch syndrome (n=486)	Prospective cohort GEOLynch Cohort	Dietary habits	Self-reported dietary intake via 183-item FFQ Dietary patterns identified by using principal component analysis. Pattern scores were grouped into tertiles based on total cohort. Median follow-up of 20 months	58 out of 486 individuals developed colorectal adenomas Tertile 1 serves as reference group for all dietary patterns: Prudent Pattern: **Adjusted for age and sex --Tertile 2, HR 0.77 (0.41,1.45) --Tertile 3, HR 0.61 (0.28, 1.32) **Adjusted for age, sex, smoking habits, adenoma history’, extent of resection -Tertile 2, HR 0.85 (0.47,1.54) -Tertile 3, HR 0.73 (0.32, 1.66) Meat Pattern: **Adjusted for age and sex --Tertile 2, HR 1.29 (0.61, 2.75) --Tertile 3, HR 2.48 (1.22,5.02) **Adjusted for age, sex, smoking habits, adenoma history, extent of resection --Tertile 2, HR 1.05 (0.49, 2.28) --Tertile 3, HR 1.70 (0.83,3.52) Snack Pattern: **Adjusted for age and sex --Tertile 2, HR 1.80 (0.96,3.40) --Tertile 3, HR 2.13 (0.99,4.60) **Adjusted for age, sex, smoking habits, adenoma history, extent of resection --Tertile 2, HR 1.93 (1.04,3.60) --Tertile 3, HR 2.16 (1.03, 4.49) Cosmopolitan Pattern: **Adjusted for age and sex --Tertile 2, HR 0.74 (0.43,1.27) --Tertile 3, HR 1.25 (0.64,2.43) **Adjusted for age, sex, smoking habits, adenoma history, extent of resection --Tertile 2, HR 0.79 (0.45,1.38) --Tertile 3, HR 1.25 (0.61,2.55)
Campbell PT et al, 2007 Manuscript	Total cases and controls included in analyses: -Cases: --males (n=1.292) --females (n=1.404) -Controls: --males (n=1.465) --females (n=1.203) Cases included in analysis stratified by sex and family history of CRC (data presented in this table): -High or intermediate risk of LS per Amsterdam/Bethe sda criteria” --males (n=461) --females (n=466) -Sporadic familial risk strata: -male cases (n=535) -female cases (n=665)	Case-control Cases: Ontario Cancer Registry and Newfoundland Cancer T reatment and Research Foundation population tumor registry Controls: Random sample of residents in each province	Weight change via BMI	Self-reported height and weight via personal history questionnaire -current weight -weight 2 years prior to cancer diagnosis -weight at 20 years of age *adult weight gain calculated as weight (kg) at age 20 subtracted from recent weight (kg)	Results for individuals at high or intermediate risk of LS- BMI 2 years prior to diagnosis: Normal weight serves as reference group Men- --overweight, OR 1.25 (0.97,1.61) --obese, OR 1.83 (1.33,2.51) Women- --overweight, OR 1.03 (0.80,1.33) --obese, OR 0.85 (0.62,1.16) BMI at age 20: Normal weight serves as reference group Men- --overweight, OR 1.37 (1.04,1.79) --obese, OR 1.92 (1.02,3.63) Women- --overweight, OR 0.90 (0.59,1.37) --obese, OR 0.81 (0.38,1.72) Adult weight sain: 1–5 kg is reference group Men- --weight loss, OR 1.07 (0.63,1.82) --6–10 kg, OR 0.98 (0.67,1.43) --11–20 kg, OR 1.10 (0.79,1.53) --21+kg, OR 1.72 (1.22,2.41) Women- --weight loss, OR 1.23 (0.76,1.98) --6–10 kg, OR 1.31 (0.91, 1.88) --11–20 kg, OR 1.11 (0.79,1.55) --21+kg, OR 1.21 (0.85,1.73) **Data adjusted for age, province, history of abnormal blood lipids, education, history of colon screening endoscopy’ Paper also includes results for the following analyses not included in this table: individuals with sporadic risk, comparison of high or intermediate risk vs sporadic risk, for total sample- risk estimates among women stratified by estrogen status, for total sample- risk estimated stratified by sex, for total sample- risk estimates stratified by sex for incident colon and rectal cancer
Chau R et al, 2016 Manuscript	Lynch syndrome (n=1966) -MLHl (n=719) -MSH2 (n=931) -MSH6 (n=211) -PMS2 (n=105) 56.4% female	Retrospective cohort Colon Cancer Family Registry	Dietary habits -MVI -calcium supplement -folic acid supplement	Standardized questionnaires conducted via in-person interviews, telephone interviews, or self-administered from mail	Supplementation ≥2×/week for ≥1-month: Individuals diagnosed with CRC- --18%, and 5%, reported intake of MVI, calcium, and folic acid respectively Individuals not diagnosed with CRC- --27%, 11% and 10% reported intake of MVI, calcium, and folic acid respectively Compared to never users: MVI- ≥1 month, HR 0.55 (0.40,0.75) ≥3 yrs, HR 0.47 (0.32,0.69) Per year, HR 0.95 (0.91,0.98) Ever taken, HR 0.51 (0.35,0.74) Calcium- ≥1 month, HR 0.46 (0.30,0.71) ≥3 yrs, HR 0.42 (0.23,0.74) Per year, HR 0.90 (0.83,0.98) Ever taken, HR 0.22 (0.08,0.60) Ever taken MVI and Calcium, HR 0.36 (0.20,0.64) No association with folic acid supplementation and CRC risk between users and never users among entire sample Compared to never users, stratified bv sex: Men- --MVI-- ≥1 month, HR 0.45 (0.28,0.72) ≥3 yrs, HR 0.41 (0.23,0.75) Per year, HR 0.93 (0.88,1.00) --Calcium-- ≥1 month, HR 0.61 (0.25,1.51) ≥3 yrs, HR 0.55 (0.17,1.74) Per year, HR 0.89 (0.80,0.99) --folic acid-- ≥1 month, HR 2.06 (0.58,7.33) Per year, HR 0.94 (0.87,1.01) Women- --MVI-- ≥1 month, HR 0.64 (0.43,0.95) ≥3 yrs, HR 0.55 (0.33,0.93) Per year, HR 0.96 (0.92,1.01) --Calcium-- ≥1 month, HR 0.63 (0.38,1.04) ≥3 yrs, HR 0.53 (0.27,1.04) Per year, HR 0.95 (0.89,1.02) --folic acid-- ≥1 month, HR 0.83 (0.49,1.40) Per year, HR 0.99 (0.90,1.08) **Data adjusted for education, country, sex, number of screening colonoscopies, regular PA, cigarette smoking, NSA1D use When data was adjusted for recent BMI, red meat intake, and fruit and vegetables intake, results were comparable
Dashti SG et al, 2017 Manuscript	Lynch syndrome (n=1.925) -with history of CRC (n=769) -without history of CRC (n=1.156) 56% female	Retrospective cohort Colon Cancer Family Registry	Dietary habits -alcohol: grams of ethanol consumed per day from alcoholic beverages from 20 years to age at CRC diagnosis or censor	Self-reported via questionnaires conducted in person, by mail, or telephone -number of servings of alcohol consumed daily or weekly in the following age ranges: 20–29, 30–39. 40–49. 50–59. > 60 years -number of years alcoholic beverages were consumed once per week for ≥6 months during each time period	No evidence for interaction between alcohol consumption and sex, smoking status, BMI at age 20. history of sigmoidoscopy or colonoscopy, MMR mutation, country of recruitment, ascertainment methods, or folic acid supplementation (p > 0.05) Abstainer serves as reference group CRC- -ever use, HR 1.55 (1.11,2.15) 0-≤14 g/d, HR 1.56 (1.11,2.18) >14-≤28 g/d, HR 1.25 (0.77,2.04) >28 g/d, HR 1.79 (1.12,2.87) Per 14 g/d, HR 1.02 (0.95,1.10) **Data adjusted for country, education, ascertainment, sex, BMI at age 20, diabetes status, regular PA, smoking, aspirin intake, MM, calcium, folic acid intake Colon Cancer- -ever use, HR 1.76 (1.27,2.46) 0-≤14 g/d, HR 1.78 (1.27,2.49) >14-≤28 g/d, HR 1.51 (0.91,2.49) >28 g/d, HR 1.94 (1.19,3.18) Per 14 g/d, HR 1.01 (0.94,1.09) Rectal Cancer- -ever use, HR 0.79 (0.41,1.53) 0-≤14 g/d, HR 0.79 (0.39,1.59) >14-≤28 g/d, HR 0.51 (0.19,1.40) >28 g/d, HR 1.14 (0.45,2.89) Per 14 g/d, HR 1.08 (0.88,1.31) **Data adjusted for country, education, ascertainment, sex, BMI at age 20, diabetes status, regular PA, smoking
Dashti SG et al, 2018 Manuscript	Lynch syndrome (n=2.042) -MLH1 (n=744) -MSH2 (n=936) -MSH6 (n=234) -PMS2 (n=106) -EPCAM (n=22)	Retrospective cohort Colon Cancer Family Registry	Total physical activity	Self-reported via questionnaires conducted in person, by mail, or telephone -Total physical activity selfreported across three age periods --20–29 years of age --30–49 years of age -- ≥ 50 years of age Calculated Metabolic equivalent (MET) hours per week based on self-reported data Assessed physical activity “near-term” = during the age period of CRC diagnosis and “long-term” = average physical activity during the lifetime	807 individuals diagnosed with CRC -colon (n=673) -rectum (n=121) -colon and rectum (n=13) In adjusted models: −29% reduction in CRC risk with activity ≥ 35 MET-hours/week compared to < 3.5 MET-hours/week for near-term physical activity --HR 0.71 (0.53,0.96) -Per 20 MET-hours/week increase in near-term physical activity, 5% reduction in CRC risk --HR 0.95 (0.91,0.99) -No evidence of inverse association separately with colon and rectal cancer -No association observed for long-term PA and risk of CRC, colon or rectal cancer **Data adjusted for country, education, ascertainment, sex, BMI at age 20, diabetes status, smoking status, alcohol intake
Diergaarde B et al, 2007 Manuscript	Cases: Patients ever diagnosed with CRC (n=145) -known or suspected to have LS(n=45) Controls: Tumor-free patients (n=103) -known or suspected to have LS (n=28)	Case-control Dutch case-control study	Dietary habits (daily) -total energy intake -total F&V -fruit intake -vegetable intake -cereals -total meat -red meat -poultry -fish -dairy -alcohol -fat -protein -CHO -dietary fiber -calcium -vitamin C -beta-carotene -folate	Self-reported via validated semi-quantitative FFQ	No significant findings among any dietary habits and CRC risk with the exception of fruit intake Significant trend (p=0.03) for fruit intake, where increased fruit intake reduces risk of CRC or adenoma >242.5 g/day fruit, OR 0.40 (0.20,0.90) **Data adjusted for age at last colonoscopy, sex, total energy intake, carrier status, cigarette smoking
Heine-Broring RC et al, 2013 Manuscript	Lynch syndrome (n=470)	Prospective cohort GEOLynch Cohort	Dietary habits -dietary supplement -fruit and vegetable intake in relation to dietary supplement use	Self-reported via questionnaire -frequency of intake (never, once per month, 2–3 days per month, weekly, 2–3 days per week, 4–5 days per week, 6–7 days per week) -amount of intake (1,2,3,4, or ≥ 5 tablets, capsules, or droplets) -brand names of MVI, vitamin C, B-vitamins, folic acid, vitamin D, vitamin A, vitamin E, calcium, iron, fish oil 183-item selfadministered validated FFQ Median follow uptime of 39.1 months	User of dietary supplements (n=188) Individuals developing adenomatous polyps during follow-up (n=122) Dietary supplement use and colorectal adenoma risk: --any dietary supplement, HR 1.18 (0.80,1.73) --MVI, HR 1.15 (0.72,1.84) --Vitamin C, HR 1.57 (0.93,2.63) --Calcium, HR 0.69 (0.25,1.92) --fish oil, HR 1.60 (0.79,3.23) Dietary supplement use and colorectal adenoma risk bv total fruit and vegetable intake: -never used supplements <179 g/d is the reference group --179–283 g/d, HR 0.48 (0.23,1.03) --283–387 g/d, HR 0.88 (0.44,1.74) --≥387 g/d, HR 0.78 (0.45,1.35) -used supplements -- < 179 g/d, HR 1.25 (0.63,2.52) -- 179–283 g/d, HR 0.97 (0.45,2.08) -- 283–387 g/d, HR 0.96 (0.45,2.03) -- ≥387 g/d, HR 0.66 (0.29,1.50) **Data adjusted for age, sex, number of colonoscopies during follow-up, similar results were observed when adjusted for age and sex
Kamiza AB et al, 2015 Manuscript	Lynch syndrome, specifically MLH1 and MSH2 germline mutation carriers (n=301) -MLH1 (n=209) -MSH2 (n=92)	Retrospective cohort Taiwan Hereditary Nonpolyposis Colorectal Cancer Consortium	Dietary habits -alcohol -tea -coffee -meat intake -vegetable intake -fruit intake -seafood intake -staple foods within the culture (i.e. rice) Physical Activity -regular PA	Dietary and PA data collected form standardized interviews with structured questionnaires. Data related to 5 years preceding study registry Cronbach’s alpha of questionnaire = 0.92 based on lifestyle and dietary variables	Significant results for all patients (n=301). Diet and PA behaviors related to CRC risk: --Regular PA compared to no PA, HR 0.58 (0.40,0.86) --Drinking tea compared to never drinking tea, HR 0.68 (0.48,0.96) --High fruit intake (3rd tertile) compared to medium or low intake, HR 0.60 (0.38,0.94) Significant results for MLH1 mutation (n=209) Diet and PA behaviors related to CRC risk: --Regular PA compared to no PA, HR 0.55 (0.35,0.86) --Drinking tea compared to never drinking tea, HR 0.62 (0.42,0.91) --High fruit intake (3rd tertile) compared to medium or low intake, HR 0.58 (0.35,0.94) Significant results iorMSH2 mutation (n=92) Diet and PA behaviors related to CRC risk: --Drinking alcohol compared to never drinking alcohol, HR 2.33 (1.04,5.21) **Data adjusted for MMR mutations
Miguchi M et al, 2018 Manuscript	Lynch Syndrome with history of CRC (n=66)	Retrospective cohort HNPCC Registry and Genetic Testing Project of the Japanese Society for Cancer of the Colon and Rectum	Dietary factor -alcohol	Via medical record or notes from genetic counseling session -alcohol consumer defined as reporting part of current regular consumption of greater than one alcoholic beverage per week	Non-users of regular alcohol considered the reference group Univariate model- Users 1.81 HR (1.06,3.036) Multivariate model- Users 2.44 HR (1.13, 5.16) **Adjustment not specified
Watson P et al, 2004	Lynch syndrome (n=360) -MLH1 (n=175) -MSH2 (n=185)	Retrospective cohort Hereditary Cancer Institute at Creighton University	Dietary factor -alcohol	Self-reported, extracted from medical record. Considered alcohol user if reported past or current regular consumption of 1 or more drinks per week.	Alcohol information available for 271 carriers. −83 carriers (30.6%) were classified as nonusers Significant associations were not observed between alcohol use and CRC risk (p>040) Paper also assessed tobacco use. When the interaction between alcohol and tobacco use was assessed, there was no significant association observed between alcohol use or the interaction and CRC risk (p>0.05). **Adjustments not specified
Win AK et al, 2011 Manuscript	Lynch syndrome (n=1324) -females (n=737) -MLH1 (n=500) -MSH2 (n=648) -MSH6 (n=l 17) -PMS2 (n=59) Without Lynch syndrome (n=1219)	Prospective cohort Colon Cancer Family Registry	Weight in early adulthood via BMI	Self-reported height and weight at age 20 collected from standardized interviews via phone, in person, or mailed questionnaires Follow-up approximately 5 years after baseline data collected	Individuals diagnosed with CRC: --With LS (n=659) -MLH1 (n=273) -MSH2 (n=308) -MSH6 (n=40) -PMS2 (n=38) --Without LS (n=36) BMI (per 5 kg/m2) at age 20 and CRC risk bv MMR mutation: --MLH1, HR 1.36 (1.04,1.77) --MSH2, HR 1.28 (0.96,1.70) --MSH6, HR 0.84 (0.38,1.80) --PMS2, HR 1.52 (0.48,4.85) --MLH1 & PMS2, HR 1.36 (1.05,1.76) --MSH2 & MSH6, HR 1.26 (0.96,1.65) BMI at age 20 and CRC risk in individuals with LS: *Nomial weight as reference group --Per 5 kg/m2 increments, HR 1.30 (1.08,1.58) --Underweight, HR 1.04 (0.66,1.64) --Overweight, HR 1.12 (0.78,1.62) --Obese, HR 2.35 (1.30,4.23) BMI at age 20 and CRC risk in individuals without LS: *Normal weight as reference group --Per 5 kg/m2 increments, HR 1.64 (1.02,2.64) --Underweight, HR 1.01 (0.28,3.73) --Overweight, HR 3.62 (1.05,12.55) --Obese, HR 3.00 (0.60,14.97) **Data adjusted for sex, country, cigarette smoking, alcohol, MMR mutation in carriers
Winkels RM et al, 2012 Manuscript	Lynch syndrome (n=386) -MLH1 (n=151) -MSH2 (n=159) -MSH6 (n=71) -PMS2 (n=3)	Prospective cohort GEOLynch Cohort	Dietary habits -alcohol intake	Selfadministered validated FFQ Median follow-up time of 10 months	Individuals developing adenomatous polyps during follow-up (n=58) Alcohol intake of individuals who developed polyps (n=58) --Tertile 1: 0–3 g/d (n=ll) --Tertile 2: 3–13 g/d (n=24) --Tertile 3: 13–71 g/d (n=8) Alcohol intake and development of polyps **Unadjusted --Tertile 1: reference group --Tertile 2: HR 2.01 (0.96, 4.21) --Tertile 3: HR 2.24 (1.09, 4.60) Per 10g units: HR 1.20 (1.03, 1.40) **Adjusted for smoking status --Tertile 1: reference group --Tertile 2: HR 1.77 (0.82, 3.84) --Tertile 3: HR 1.87 (0.87, 3.99) Per 10g units: HR 1.14 (0.96, 1.36) **Adjusted for smoking status, age, sex number of colonoscopies, history of colonic resection (yes/no), BMI --Tertile 1: reference group --Tertile 2: HR 1.59 (0.70, 3.60) --Tertile 3: HR 1.56 (0.71, 3.43) Per 10g units: HR 1.04 (0.85, 1.28)

LS = Lynch Syndrome, FFQ = Food Frequency Questionnaire, hMLH1 = hyper-methylated MLH1, PA = Physical Activity, CHO = carbohydrates, MVI = Multivitamin, NSAID = Non-Steroidal Anti-Inflammatory, HNPCC = Hereditary Nonpolyposis Colorectal Cancer

Risk of Bias Summary

All 16 studies were assessed for Risk of Bias. Based on the predefined criteria from the three-item checklist as described above, two of the 16 studies were considered high risk of bias for their study setting while the others were considered low risk [7, 21]. One of the studies was considered high risk because the study setting consisted of a cancer registry-based case-control study [21], while the other study was a hospital-based cased-control study [7]. All other studies were considered low risk because they were cohort studies or population-based case-control studies. Five of the 16 studies were considered at high risk of bias for data collection method because they either did not use a valid and reliable measurement method/instrument, or failed to report the validity and reliability of the method used [16, 19, 21, 29, 30]. Finally, two of the studies were considered high risk of bias for confounders because they did not report whether or not they considered confounders [16, 29]. Table 3 depicts the risk of bias summary results.

Table 3:

Risk of bias summary

Cancer Type	Author, Year	Study Settinga	Data Collection Methodb	Confoundersc
Endometrial Cancer	Maenpaa JU et al, 2016	Low risk	High risk	High risk
	Win AK et al, 2011	Low risk	Low risk	Low risk
	Win AK et al, 2014	Low risk	Low risk	Low risk
Colorectal Cancer	Botma A et al, 2010	Low risk	High risk	Low risk
	Botma A et al, 2013	Low risk	Low risk	Low risk
	Campbell PT et al, 2007	High risk	High risk	Low risk
	Chau R et al, 2016	Low risk	Low risk	Low risk
	Dashti SG et al, 2017	Low risk	Low risk	Low risk
	Dashti SG et al, 2018	Low risk	Low risk	Low risk
	Diergaarde B et al, 2007	High risk	Low risk	Low risk
	Heine-Broring RC et al, 2013	Low risk	Low risk	Low risk
	Kamiza AB et al, 2015	Low risk	Low risk	Low risk
	Miguchi M et al, 2019	Low risk	High risk	High risk
	Watson P et al, 2004	Low risk	High risk	Low risk
	Win AK et al, 2011	Low risk	Low risk	Low risk
	Winkels RM et al, 2011	Low risk	Low risk	Low risk

^aLow risk = Cohort studies and population-based case-control studies, High risk = Hospital-based case-control or cancer registry-based case-control studies

^bLow risk = Report and use of valid and reliable tool to assess energy-balance factor; High risk = No report of validity or reliability of tool used

^cLow risk = Adjusted for and/or considered potential confounders when approrpriate; High risk = Did not consider potential confounders

Energy Balance Related Lifestyle Factors and EC Risk

Weight Status and Weight Change

Table 4 summarizes findings for each lifestyle factor by cancer type. Of the three studies that assessed weight status or weight change in relation to EC risk, one was retrospective [16], and two were prospective [17, 18]. In Win and colleagues’ [18] prospective study among 601 women with Lynch syndrome, no significant association was observed between EC risk and BMI in early adulthood in both an unadjusted model (per 5 kg/m²: HR 0.67, 95%CI 0.38,1.20) and when adjusted for age at menarche, year of birth, hormonal contraceptive use, and cigarette smoking at 20 years (per 5 kg/m²: HR 0.73, 0.40, 1.30). Win and colleagues [18] had a follow-up time of 5 years. Additionally, Win and colleagues [17] conducted a second prospective study among 976 women with Lynch syndrome and still observed no significant association with increased weight status in adulthood and endometrial cancer risk when adjusting for “measured potential confounders” per the authors. Consistent with these findings, Maenpaa and colleagues [16] retrospective study among 136 women did not observe a significant association between body mass index (BMI) at age 18 years, 40 years, or current age with EC risk (BMI at 18 years: HR 1.03, 95% CI 0.91,1.17; 40 years: HR 1.04, 95%CI 0.98, 1.11; current BMI: HR 1.02, 95%CI 0.97, 1.08). Statistical adjustments for this analysis were not specified by the authors.

Table 4:

Summary of findings of the association between lifestyle factors, colorectal cancer risk and endometrial cancer risk among individuals with Lynch Syndrome

Cancer Type	Lifestyle Factor	Major Findings
Endometrial Cancer (EC)	Weight Status/Weight Gain	3 studies No association of EC risk with current weight status, weight change over time, or weight status in early adulthood
	Dietary Habits	1 study Significant association between multivitamin and folic acid supplementation and decreased EC risk No association of EC risk observed with calcium supplementation No association between alcohol use and EC risk
	Physical Activity	No evidence available
Colorectal Cancer (CRC)	Weight Status/Weight Gain	3 studies Increased CRC Risk: Significantly associated with early adulthood overweight/obesity status and excessive adulthood weight gain (> 21 kg), primarily among men Significantly associated with 5 kg/m2 unit increase in BMI among individuals with MLHl mutation and MLHl & PMS2 heterodimer mutations regardless of sex Risk of Developing Colorectal Adenomas among Individuals without Adenoma: Significantly associated with obesity, primarily in men Risk of Developing Additional Colorectal Adenoma among Individuals with Adenoma: Significantly associated with 2-year weight gain > 2 kg, primarily in women
	Dietary Habits	9 studies Decreased CRC Risk: Significantly associated with multivitamin supplementation among men and women Mixed evidence regarding calcium supplementationa Significantly associated with tea intake and high fruit intake among men and women Increased CRC Risk: Significantly associated with alcohol intake in individuals with MSH2 gemiline mutation No association of CRC risk observed with vegetable intake, total meat, red meat, poultry, fish, dairy, fat, protein, carbohydrates, dietary fiber, vitamin C, beta-carotene, folate, and cereals Risk of Developing Colorectal Adenoma: Mixed evidence with dietary patterns high in fresh and processed meat among men and womenb Significantly associated with dietary patterns high in snack foods among men and women No associations with alcohol and dietary supplement use among men and womenc
	Physical Activity	2 studies Significant association between engagement in physical activity and decreased CRC risk

^aTwo studies observed no association with calcium supplementation and CRC risk among men and women, but one study observed a significant association with calcium supplementation and reduced CRC risk among women when analysis was stratified by sex

^bSignificant association observed in models adjusted for age and sex but not in fully adjusted models (adjusted for age, sex, smoking habits, adenoma history, extent of colorectal resection)

^cDietary supplement use includes use of multivitamin, vitamin C, calcium or fish oil supplementation

Dietary Habits

Within the prospective study conducted by Win and colleagues [17] among 976 women, dietary habits in relation to endometrial cancer risk was also assessed. When adjusting for “measured potential confounders,” a significant association between alcohol intake and EC risk was not observed [17]. However, significant reductions in EC risk were observed with ever use of multivitamin supplementation (HR 0.38, 95%CI 0.22, 0.65) and folic acid supplementation (HR 0.22, 95%CI 0.08, 0.58) [17]. There was no significant association with having ever used calcium supplements.

Energy Balance Related Lifestyle Factors and CRC Risk

Weight Status and Weight Change

Two studies assessed weight status and weight change in relation to CRC risk [21, 27], and one study assessed weight change in relation to colorectal adenoma risk [19]. Regarding CRC risk, both studies assessed weight status in early adulthood, as defined by BMI at age 20 years, in relation to risk. Campbell and colleagues [21] case-control study was conducted among 461 males and 466 females with CRC that were considered high or intermediate risk of Lynch syndrome based on Amsterdam/Bethesda criteria. In addition, Win and colleagues [27] prospective study was conducted among 1,324 individuals with Lynch syndrome and had a follow-up time of approximately 5 years. Both studies observed a significant association between an early adulthood weight classification of obesity and CRC risk in adjusted models. Specifically, Campbell and colleagues [21] observed this association among men (odds ratio [OR] 1.92, 95%CI 1.02, 3.63) but not women (OR 0.81, 95%CI 0.38, 1.72) when adjusted for age, province where data were collected (i.e. Ontario or Newfoundland), history of abnormal lipids, education, and history of colon screening endoscopy (results from unadjusted model not provided). Win and colleagues [27] observed these findings in both men and women in the unadjusted model and when adjusting for sex, country, cigarette smoking status, alcohol, MMR gene mutation (hazard ratio [HR] 2.35. 95%CI 1.30,4.23).

In addition to early adulthood weight classification of obesity, Campbell and colleagues [21] also observed a relationship between an early adulthood weight classification of overweight and CRC risk. Specifically, this relationship was only observed in overweight men (OR 1.37, 95%CI 1.04, 1.79) but not overweight women (OR 0.90, 95%CI 0.59, 1.37) in adjusted models. In contrast, while Win and colleagues [27] did not observe a relationship between overweight status and CRC risk, a 30% increase in CRC risk was observed among men and women with each 5 kg/m² unit increase in BMI (HR 1.30, 95%CI 1.08,1.58). When stratified by sex, significant findings remained in men (HR 1.33, 95%CI 1.01,1.76) but not women (HR 1.28, 95%CI 0.98, 1.69) [27].

Win and colleagues [27] also stratified their analysis by Lynch-associated MMR gene mutations. Both adjusted (sex, country, cigarette smoking status, alcohol intake) and unadjusted models demonstrate a 36% increased risk of CRC with every 5 kg/m² unit increase in BMI among individuals with MLH1 mutation (HR 1.36, 95%CI 1.04, 1.77) and both MLH1 and PMS2 mutation carriers (HR 1.36, 95%CI 1.05, 1.76). For MSH2 and MSH6 mutation carriers alone and combined, no statistically significant association was observed [27].

Along with early adulthood weight status, Campbell and colleagues [21] assessed adulthood weight gain, as defined by the difference between current weight and weight at age 20 years. A 72% increase in CRC risk (OR 1.72, 95%CI 1.22, 2.41) was observed in men who gained ≥ 21 kg. Lower categories of adulthood weight gain (+11–20 kg, +6–10 kg, +1–5 kg = reference group) were not significantly associated with CRC risk. Further, there were no significant associations with CRC risk for any adulthood weight gain categories in women [21].

Regarding colorectal adenoma risk, Botma and colleagues [19] prospective study assessed weight change over time and weight at study entry, as defined by BMI, among 486 individuals with Lynch Syndrome. The median follow-up time for the study was 20 months. When adjusted for age, sex, extent of colon surgery, smoking habits, and alcohol intake, evidence demonstrated that obesity was associated with a significant increase in colorectal adenoma risk (HR 2.04, 95%CI 1.08, 3.86), regardless of whether or not individuals had adenomas at study entry. When stratified by sex, there was no significant relationship between BMI and adenoma risk in women (HR 0.99, 95%CI 0.27, 3.57) but obesity was associated with an increased risk in men (HR 2.94, 95%CI 1.11, 7.78). In men without adenoma at study entry, a BMI classification of overweight or obese was associated with an 8.7-fold increased risk of colorectal adenoma, and a dose-response relationship was observed between 5 kg/m² unit increase in BMI and adenoma risk (HR 1.84, 95%CI1.13, 3.02). None of these significant associations were observed among women without adenomas at study entry. Among men and women with adenomas at study entry, current BMI was not related to the risk of newly developed adenomas since study entry. However, among women who had adenomas at study entry, a 2-year weight gain greater than 2 kg was associated with increased risk of newly developed adenomas (HR 4.09, 95%CI 1.04, 16.19) [19].

Dietary Habits

Of the nine studies that assessed dietary habits, three were prospective [20, 24, 31], five were retrospective [22, 25, 28–30], and one was a case-control study [7]. The three prospective studies assessed the association between dietary habits and colorectal adenoma risk [20, 24, 31]. While the sample sizes used in these prospective investigations are relatively similar, inconsistencies exist among the dietary factors assessed and follow-up times for the assessment. Botma and colleagues [20] assessed different eating patterns over a median follow-up time of 20 months among 486 individuals with Lynch syndrome, whereas Winkels and colleagues [31] assessed alcohol intake and utilized a follow-up time of 10 months among 386 individuals with Lynch syndrome, and Heine-Broring and colleagues [24] assessed dietary supplement use (multivitamin and mineral supplement and specific vitamins and minerals) over a median follow-up time of 39 months among 470 individuals with Lynch syndrome.

Botma and colleagues [20] observed a significant association between a dietary pattern high in fresh and processed meats and increased risk of developing colorectal adenoma in models adjusted for age and sex (tertile 3: HR 2.48, 95%CI 1.22, 5.02), but in fully adjusted models (age, sex, smoking habits, adenoma history, extent of colorectal resection) this association was not significant [20]. Additionally, a dietary pattern high in snack foods significantly increased colorectal adenoma risk (tertile 3: HR 2.16, 95%CI 1.03, 4.49) in the fully adjusted model [20]. Other types of dietary patterns, characterized by a cosmopolitan pattern or prudent pattern, were not significantly associated with adenoma risk, regardless of adjustment [20]. In Winkels and colleagues [31] study, a significant association between adenoma risk and alcohol intake was only observed at the highest tertile (13–71 g alcohol/d) within the unadjusted model (HR 2.24, 95%CI 1.09,4.60). When adjusted for smoking status alone and smoking status with other factors (age, sex, number of colonoscopies, history of colonic resection, and BMI) significant associations were not observed. Finally, Heine-Broring and colleagues [24] did not observe any significant associations between risk of developing colorectal adenoma and use of a multivitamin, vitamin C, calcium, or fish oil supplementation, when assessed separately and when assessed by total fruit and vegetable intake in adjusted models (age and sex, or age, sex, and number of colonoscopies during follow-up).

In addition to work assessing adenoma risk, five investigations assessed dietary habits in relation to CRC risk [7, 22, 25, 28, 29]. Study design, dietary habits and sample size varied among all studies as follows: one retrospective study conducted by Chau and colleagues [22] assessed dietary supplement use among 1,966 individuals with Lynch syndrome; one retrospective study, conducted by Kamiza and colleagues [25], assessed different types of foods among 209 MLH1 mutation carriers and 92 MSH2 mutation carriers; one case-control study, conducted by Diergaarde and colleagues [7], assessed different types of foods among 247 individuals (145 CRC cases of which 45 were known or suspected to have Lynch syndrome, and 103 controls of which 28 were known or suspected to have Lynch syndrome); and three retrospective studies, conducted by Dashti and colleagues [28], Miguchi and colleagues [29], and Watson and colleagues [30] assessed alcohol intake specifically among 1,925, 66 and 360 individuals with Lynch syndrome, respectively.

Chau and colleagues [22] assessed multivitamin, calcium and folic acid supplementation and adjusted for the following factors in their analyses: education, country, sex, number of screening colonoscopies, regular physical activity, cigarette smoking and non-steroidal anti-inflammatory use. Overall, they observed a 36% reduction in CRC risk with ever use of a multivitamin and calcium supplement (HR 0.36, 95%CI 0.20, 0.64). Consistent associations were observed when assessing multivitamin use and calcium use alone among varying durations (i.e. ≥ 1 month, ≥ 3 years, annually, or ever taken) [22]. Further, no associations were observed with folic acid supplementation. When stratified by sex, similar results were observed for multivitamin use. Among men, there were no significant associations observed between CRC risk and calcium or folic acid supplementation. Among women, calcium supplementation for ≥ 1 year and 3 years, and folic acid supplementation for 1–11.9 months (HR 0.39, 95%CI 0.17, 0.87) was significantly linked with CRC risk reduction [22]. When data was adjusted for recent BMI, red meat intake, and fruit and vegetable intake, results were similar. In contrast, Diergaarde and colleagues’ [7] case-control study assessed calcium, vitamin C, beta-carotene, and folic acid supplementation and did not observe any statistically significant findings related to dietary supplement use and CRC risk. Diergaarde and colleagues adjusted their analyses for age at last colonoscopy, sex, total energy intake, mutation carrier status, and cigarette smoking.

Regarding the studies that assessed types of foods, Kamiza and colleagues [25] conducted a retrospective study that assessed alcohol, tea, coffee, meat, vegetables, fruit, seafood and rice. Diergaarde and colleagues [7] assessed total energy intake, total fruit and vegetable intake, fruit intake, vegetable intake, cereals, total meat, red meat, poultry, fish, dairy, alcohol, fat, protein, carbohydrates, dietary fiber. Both studies observed significant inverse associations with high fruit intake and CRC risk in their adjusted models (HR 0.60, 95%CI 0.38, 0.94 [25]; OR 0.40, 95%CI 0.20, 0.90; P_trend = 0.03 [7]). While Diergaarde and colleagues [7] did not observe any other statistically significant associations between dietary habits and CRC risk, Kamiza and colleagues [25] observed a significant inverse association between CRC risk and tea intake among MLH1 and MSH2 carriers specifically (HR 0.68, 95%CI 0.48, 0.96). In contrast to Diergaarde and colleagues [7], Kamiza and colleagues [25] only adjusted their analyses for MMR mutation.

Regarding alcohol intake, Diergaarde and colleagues [7] and Watson and colleagues [30] did not observe a significant relationship between alcohol intake and CRC risk. In contrast, Kamiza and colleagues [25] observed an association among MSH2 carriers, where MSH2 carriers had a two-fold increase in CRC risk with alcohol consumption (HR 2.33, 95%CI 1.04, 5.21). Regardless of MMR gene mutation, Dashti and colleagues [28] observed nearly a two-fold increase in CRC risk (HR 1.79, 95%CI 1.12,2.87) when individuals with Lynch syndrome consumed > 28 grams of alcohol per day (~ 2 alcoholic beverages per day). There was no evidence of interaction when assessed by sex (p>0.05). These analyses were adjusted for country, education, ascertainment, sex, BMI at age 20, diabetes status, smoking status, alcohol intake, and regular physical activity. Consistent with these findings, Miguchi and colleagues [29] observed a two-fold increase in CRC risk (HR 2.44, 95%CI 1.13,5.16) among Japanese individuals with Lynch syndrome who were previously or currently consuming at least one alcoholic beverage per week (models were adjusted but the adjustment was not specified).

Physical Activity

Two studies assessed physical activity in relation to CRC risk retrospectively [23, 25]. Kamiza and colleagues [25] observed a 58% reduction in CRC risk among individuals who engaged in regular physical activity, compared to those who did not (HR 0.58, 95%CI 0.40, 0.86). Dashti and colleagues [23] observed a 29% reduction in CRC risk for the highest vs. lowest level of recent physical activity, i.e. during the age period of CRC diagnosis or censoring [≥ 35 vs. <3.5 MET (metabolic equivalent) hours per week (HR 0.71, 95%CI 0.53,0.96)], among 2,042 individuals with Lynch syndrome. Furthermore, a dose-dependent relationship was observed with weekly physical activity and CRC risk during this age period such that each 20 MET hour increase in weekly physical activity was associated with a 5% reduction in CRC risk (HR 0.95; 95%CI 0.91, 0.99). These analyses were adjusted for country, education, ascertainment, sex, BMI at age 20, diabetes status, smoking status and alcohol intake.

Discussion

Energy Balance Related Lifestyle Factors and EC Risk among Individuals with Lynch Syndrome Compared to the General Population

Current evidence is insufficient to demonstrate obesity as an independent additional risk factor for developing EC in women with Lynch syndrome [16–18]. This contrasts with evidence available from the general population, where obesity is associated with a 50% increase in risk of EC per 5 kg/m² unit increase in BMI (RR 1.50, 95%CI 1.42, 1.59) [32]. Furthermore, descriptive statistics from studies comparing EC survivors with Lynch syndrome to EC survivors from the general population demonstrate more cases of overweight/obesity among EC survivors from the general population [2, 33]. In terms of diet, within the single study that assessed dietary habits among women with Lynch syndrome, ever use of a multivitamin or folic acid supplementation was associated with a reduction in EC risk [17]. These findings contrast with evidence available in the general population, which has demonstrated no association between EC risk and multivitamin use [34–36]; however evidence pertaining to folic acid supplementation or dietary intake of folate-rich foods and EC risk is mixed among women in the general population [37–40]. Thus, more research is needed in order to establish whether recommendations for dietary habits, including both food and beverage intake and multivitamin use relative to Lynch syndrome status should differ in terms of EC risk. Research is also needed to examine the relationship between physical activity and EC risk among women with Lynch syndrome. This review revealed that there is no literature available assessing this relationship.

Energy Balance Related Lifestyle Factors and CRC Risk among Individuals with Lynch Syndrome Compared to the General Population

Overall, evidence supports a relationship between overweight/obesity and increased risk of developing CRC among individuals with Lynch syndrome, including risk of colorectal adenomas. Specifically overweight/obesity status in early adulthood and adulthood weight gain of > 21 kg increases risk of CRC among individuals with Lynch syndrome [21, 27]. When stratified by sex, effects were primarily present in men [21, 27]. These findings are consistent with observations from the general population, such that overweight/obesity status is associated with CRC primarily among men, with non-significant positive associations observed among women [41, 42].

Colorectal adenomas are premalignant lesions and therefore linked with increased risk of CRC [43, 44]. In the one study that assessed adenoma risk in relation to weight among individuals with Lynch syndrome, obesity and adulthood weight gain was linked with increased risk of developing colorectal adenoma primarily among men [19]. Among women, only a two-year weight gain of > 2 kg was associated with increased risk of developing more adenomas, under the condition that colorectal adenomas were already present [19]. These findings are consistent with the general population such that a recent systematic review and meta-analysis of 23 studies (168,201 individuals without Lynch Syndrome) revealed a significant association between overweight and obesity status and colorectal adenoma risk (OR 1.24, 95%CI 1.16, 1.33) [45]. Interestingly among the 11 studies that reported findings by sex, overweight and obesity was significantly associated with adenoma risk among women (OR 1.19, 95%CI 1.01, 1.33) but not men (OR 1.16, 95%CI 0.94, 1.45) [45]. When stratified by menopausal status, while not statistically significant, the link between pre-menopausal status and adenoma risk was greater than post-menopausal status and adenoma risk (OR 2.48, 95%CI 0.56, 11.05; OR 1.06, 95%CI0.77,1.45; respectively) [45]. The authors suggest that these findings may be attributed to protective effects of exogenous estrogens and colorectal cancer risk [45]. Missing from the study conducted among individuals with Lynch Syndrome is stratification of women by menopausal status. More research is needed stratifying individuals with Lynch syndrome by gender and women by menopausal status in order to determine whether there are sex differences in the context of weight status and adenoma risk among this population.

Regarding diet, among the two studies that included assessment of multivitamin and mineral supplementation use and CRC risk among individuals with Lynch syndrome [7, 22], one study observed a significant relationship between supplementation and CRC risk [22]. Ever use of calcium and multivitamin supplementation was linked with a reduction in CRC risk primarily in women with Lynch syndrome [22]. Findings pertaining to calcium supplementation are consistent with the American Institute of Cancer Research’s Continuous Update Project, which states that there is probable evidence to support calcium supplementation for CRC risk reduction among the general population [41]. In terms of multivitamin supplementation, there is no evidence to support supplementation for CRC risk reduction among the general population. The recommendation for the general public is to consume a diet high in fruits, vegetables, and whole grains, as this should cover all of the major vitamins and minerals [24]. Consistent with these recommendations, two studies among individuals with Lynch syndrome observed a reduction in CRC risk with increased fruit intake [7, 25]. Regarding alcohol intake, for the general population there is strong evidence that consuming greater than approximately 30 grams per day of alcohol (~ 2 alcoholic beverages) increases risk of CRC [41]. Based on the studies reviewed that included assessment of alcohol intake, the relationship between alcohol intake and CRC risk among individuals with Lynch syndrome is consistent with recommendations for the general population [25, 28]. Further, one study suggested that the threshold for alcohol intake with regards to CRC risk may be lower specifically among Japanese individuals with Lynch syndrome [29].

In the context of adenoma risk and dietary habits, a significant association was only observed with dietary patterns high in meat and snack foods among individuals with Lynch syndrome [20]. These findings related to meat-heavy dietary patterns are consistent with those observed among the general population [46, 47]. To our knowledge only one investigation has assessed dietary patterns high in snack foods in relation to adenoma risk among men and women in the general population [48], and this study showed no association. Taken together, the evidence pertaining to dietary habits and adenoma risk among individuals with Lynch syndrome is limited and more research is needed in this area in order to draw conclusions to impact recommendations for cancer prevention.

Evidence from the two studies that assessed physical activity in relation to CRC risk support a protective effect of regular physical activity for individuals with Lynch syndrome [23, 25]. For example, Dashti and colleagues [23] reported a significant reduction in CRC risk among men and women when their total physical activity for the week was ≥ 35 MET hours per week. These findings are consistent with findings from the World Cancer Research Fund/American Institute of Cancer Research Continuous Update Project, where a significant inverse relationship was observed for total physical activity and CRC risk [8, 49, 50]. Additionally, among men in the general population, higher doses of recreational physical activity, ≥ 46 MET hours per week, were linked with a significant reduction in CRC risk [51]. Thus evidence suggests an inverse association between physical activity or total MET-hours of weekly physical activity and CRC risk among individuals with Lynch syndrome.

Limitations, Future Research and Practical Implications

The assessment of energy balance related lifestyle factors in relation to EC and CRC risk among individuals with Lynch syndrome is still in the early stages. Considering this, there are several limitations to this systematic review including: small number of studies assessing each energy balance related lifestyle factor (i.e. weight status, weight change, dietary habits, and physical activity); small number of studies assessing each MMR mutation individually, and in some studies MMR mutation results were pooled; retrospective design of the majority of studies, which may be subject to biases such as participant recall bias or the fact that the use of non-steroidal anti-inflammatory medications was not controlled for in all studies related to CRC; inconsistencies in confounding factors considered in statistical adjustments; no published manuscripts prospectively assessing dietary habits in relation to EC risk among women with Lynch syndrome.

More research is needed to determine whether energy balance-related lifestyle factors known to contribute to EC and CRC risk among the general population exhibit the same effects among individuals with Lynch syndrome. Further, more research is needed assessing the relationship between weight status and adenoma risk stratified by gender and among women, stratified by menopausal status. Additionally, more prospective studies and subsequently randomized controlled trials are needed within this population and should include the following: assessment of these lifestyle factors in relation to EC and CRC incidence, assessment of MMR mutations individually as opposed to pooled together, assessments stratified by gender and in women stratified by menopausal status, consistency among adjustments made in analyses, and in the context of EC, assessment of specific cancer sub-type.

Overall, current evidence suggests that use of multivitamin and mineral supplementation and folic acid supplementation may reduce risk of EC in women with Lynch syndrome. Regarding CRC risk, many health recommendations for the general public, such as maintaining a healthy weight, eating a diet high in fruit, and being physically active are also relevant to individuals with Lynch syndrome. However, more evidence, specifically prospective studies and subsequently randomized controlled trials, as indicated above, are necessary in order to determine whether there is a need to tailor energy balance related lifestyle recommendations among these individuals who are at much higher risk of developing endometrial and colorectal cancer.