• colorectal neoplasms;
  • hereditary nonpolyposis;
  • nutrition;
  • epidemiology;
  • DNA mismatch repair;
  • diet


  1. Top of page
  2. Abstract
  7. Acknowledgements


Patients with Lynch syndrome (LS) have a high risk of developing colorectal cancer due to mutations in mismatch repair genes. Because dietary factors, alone and in combination, influence sporadic colorectal carcinogenesis, the association of dietary patterns with colorectal adenomas in LS patients was assessed.


In the GEOLynch cohort of 486 persons with LS, dietary information was collected, using a food frequency questionnaire. Dietary pattern scores were obtained by principal components analysis. Hazard ratios (HR) between dietary patterns and colorectal adenomas were calculated using Cox regression models. Robust sandwich variance estimates were used to control for dependency within families. Final models were adjusted for age, sex, smoking habits, colorectal adenoma history, and extent of colon resection.


During a median follow-up of 20 months, colorectal adenomas were detected in 58 persons. Four dietary patterns were identified: a “Prudent,” “Meat,” “Snack,” and “Cosmopolitan” pattern. Individuals within the highest tertile of the “Prudent” pattern had a HR of 0.73 (95% confidence interval [CI], 0.32-1.66) for colorectal adenomas, compared with the lowest tertile. Those with high “Meat” pattern scores had a HR of 1.70 (95% CI, 0.83-3.52). A high “Snack” pattern was associated with an increased risk of colorectal adenomas (HR, 2.16; 95% CI, 1.03-4.49). A HR of 1.25 (95% CI, 0.61-2.55) was observed for persons in the highest tertile of the “Cosmopolitan” pattern.


These findings suggest that dietary patterns may be associated with development of colorectal adenoma in patients with Lynch syndrome. The directions of these findings are corroborative with those observed in studies investigating sporadic colorectal cancer. Cancer 2013. © 2012 American Cancer Society.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Lynch syndrome (LS) is a dominantly inherited syndrome characterized by development of colorectal cancer (CRC), endometrial cancer and other cancers at an early age.1-5 The syndrome is caused by germline mutations in genes involved in or influencing DNA mismatch repair (MMR), MLH1, MSH2, MSH6, PMS2, or EPCAM.6, 7 Risk of developing CRC up to age 70 years in LS lies between 22% and 69%.1-4 In addition, MMR gene mutation carriers have an increased risk of developing colorectal adenomas and at a younger age compared with noncarriers from the same LS families.8 Clinical expression of LS varies between geographic regions.9 Moreover, risk of CRC varies in and between families.10 Possible explanations for these differences are influences of modifier genes, lifestyle, or dietary factors.

Numerous studies have investigated associations between single foods and sporadic CRC. There is general agreement that red and processed meat and alcohol increase risk of sporadic colorectal neoplasms.11 The influence of other foods or food groups is less convincing.11 Recognition of the interactive and synergistic effects between foods, explains the increased research focus on the effect of dietary patterns. Several epidemiological studies show that dietary patterns indeed influence risk of sporadic colorectal adenomas and cancer.12-16 Three prospective studies associate increasing consumption of a “Western” diet with an increased risk of colon adenomas, colon cancer,14, 15 or colon cancer recurrence.12 Another prospective study showed both an increased risk of CRC for persons with a higher intake of a “Meat and potato” pattern as well as a decreased risk for those with a higher consumption of a “Fruit and vegetable” pattern.13

Only few studies evaluated lifestyle factors and colorectal neoplasms in patients with LS.17-21 The only studies17, 19 reporting on dietary factors were from our group, and show that increased fruit and high fiber intakes possibly decrease the risk of colorectal neoplasms.17, 19 This case-control study included both MMR gene mutation carriers and untested individuals suspected of LS. The current study examined dietary patterns in a prospective cohort consisting only of MMR gene mutation carriers. The aim was to evaluate associations between dietary patterns, identified by using principal component analysis (PCA), and colorectal adenoma development.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Study Population

Eligible MMR gene mutation carriers for this prospective cohort study,21 the GEOLynch cohort study, were identified from families registered at the Netherlands Foundation for the Detection of Hereditary Tumors (NFDHT) in Leiden, Radboud University Nijmegen Medical Center (RUNMC) in Nijmegen, and University Medical Center Groningen (UMCG) in Groningen (all in the Netherlands). Eligible subjects were Dutch-speaking, Caucasian, mentally competent to participate, men and women between 18 and 80 years of age who were under regular surveillance by colonoscopy. Terminally ill patients and those with inflammatory bowel diseases, proctocolectomy, or colostomy were excluded. Between July 2006 and July 2008, 713 MMR gene mutation carriers were identified. The medical specialists gave approval to contact their patients, and 499 (73%) patients agreed to participate. Finally, 486 participants were included, because necessary questionnaire or medical data was incomplete for 13 participants. Approval for this study was obtained from the medical ethics committee of RUNMC. All participants provided written informed consent.

Dietary Assessment and Determination of Dietary Patterns

At baseline, we collected information on diet and lifestyle habits, medication use, physical activity,22 and relevant medical history. Dietary intake information was collected using a self-administered food frequency questionnaire (FFQ), ie, a 183-item questionnaire developed to assess habitual food intake during the previous month. This FFQ was an updated version of previously validated FFQs.23, 24 The questionnaire asked for frequency of use on a scale of frequency categories: not this month, once a month, 2 to 3 times per month, once a week, 2 to 3 times per week, 4 to 5 times per week, and 6 to 7 times per week. The number of servings per time point was asked in natural (eg, orange, slice) or household units (eg, glass, spoonful). Questions on vegetables and fruits were specified with respect to season. Frequencies per day and standard portion sizes were multiplied to obtain grams per day for each food item. Energy intake was calculated using the Dutch food composition table.25 When questionnaires were returned incomplete, participants were contacted by phone.

To identify dietary patterns, we used PCA to aggregate dietary variables. First, the 183 FFQ food items were grouped into 87 food groups. Foods were grouped according to type of food (eg, broccoli, cauliflower, and cabbage were combined into cruciferous vegetables). Per person, the intake of every food group (grams per day) was divided by total daily energy intake (kcal) and multiplied by 1000. This was done because we were interested in the dietary composition, independent of the kilocalories consumed. These intake variables (grams per day per 1000 kcal) were used in the PCA to construct dietary patterns. Eventually, we retained 4 dietary patterns. First, components with an eigenvalue > 1 (33 of 87) were selected. Second, inspection of the Scree plot, a plot of eigenvalues by number of components, indicated a final number of 4 or 6 dietary patterns, because it leveled off after the fourth and sixth component. Finally, we ran the PCA 3 times with a defined number of components, ie, 4, 5, or 6, and selected 4 patterns based on the interpretability of all components retained. To achieve a simpler structure and easier interpretability, components were rotated by orthogonal transformation. The 4 dietary patterns were labeled as “Prudent,” “Meat,” “Snack,” and “Cosmopolitan” patterns. We calculated dietary pattern scores by summing a persons' food group intake, multiplied by its component (dietary pattern) loading for each food group (ie, correlations with the patterns). The influence of food grouping on retained patterns was checked by repeating the PCA with all 183 original food items. The same patterns emerged.

Identification of Colorectal Adenoma Cases

Colonoscopy follow-up data was collected at the LS family registry at the NFDHT,8 and from medical records at the 2 hospitals up to at least January 31, 2009. Also, information about all previously performed colonoscopies, surgical interventions, and cancer and adenoma occurrences was gathered. For each colonoscopy, information on number of neoplasms, plus location, size, and histology of these was collected.

Statistical Analyses

Risk of developing colorectal adenomas was estimated by calculating hazard ratios (HR) and 95% confidence intervals (CI) using Cox regression. Because some participants were members of the same family, standard errors were calculated by computing robust sandwich estimates of the covariance matrix clustering on family membership to account for dependence of observations. For cases, person-time is time to adenoma diagnosis. It started at time of questionnaire completion and ended at the date of first adenomatous polyp diagnosis. Carriers who were diagnosed with colorectal cancer or metastases before an adenoma diagnosis were censored, and person-time ended at date of this diagnosis. For noncases, person-time started at date of questionnaire completion and ended at January 31, 2009, or at the date of their last known colonoscopy if later than January 31, 2009.

Dietary pattern scores were grouped into tertiles based on the total cohort, with the lowest tertile being the reference group. The following variables were evaluated for confounding, using backward selection: age (continuous), sex, smoking habits (current, former, never), regular use of nonsteroidal anti-inflammatory drugs (more or less than once per week), physical activity (tertiles), colorectal adenoma history (yes/no), extent of colorectal resection (none, partial, or subtotal colectomy), and number of endoscopies during follow-up (continuous). Variables remained in the model if removing them changed a HR with 10% or more. Extra adjustment for body mass index (BMI) was performed to see whether the influence of dietary patterns on colorectal adenomas was (partly) explained by BMI. Energy intake was not considered as a confounder, because the amount of energy consumed is interwoven within the dietary patterns. Part of the cohort did not yet receive a colonoscopy during study follow-up. These carriers differed from the 386 with a colonoscopy during follow-up with regard to history of CRC, percentage of carriers with colonic resections, and number of MSH6 gene mutation carriers. No differences in demographic or lifestyle factors between the groups were observed. To study the consequences of these differences, we studied extreme possibilities. In the main analyses, we assumed all 100 to be noncases, whereas in sensitivity analyses, we either assumed them to be all cases or restricted the cohort to persons who did have a colonoscopy. To test linear trends, we entered dietary pattern scores continuously in the models. All statistical tests were 2-sided. Cox regression models were tested for and met the proportional hazard assumption. All analyses were performed using SAS software, version 9.2 (SAS Institute, Cary, NC).


  1. Top of page
  2. Abstract
  7. Acknowledgements

Dietary Patterns

PCA identified 4 dietary patterns in this cohort of patients with LS. The component loadings, which are correlations between foods and dietary patterns, are shown in Table 1. The “Prudent” pattern heavily loaded (≥ 0.30) on several types of fruits and vegetables, whole grains, nonfat yoghurt and curd, low-fat cheese, poultry, fish, dressings, green and herbal tea, and added sweets. The “Meat” pattern heavily loaded on poultry, beef, pork, minced meat, processed meat, and coffee and negatively loaded on whole grains, peanut butter, cakes and cookies, vegetarian products, and soy-based desserts. The “Snack” pattern heavily loaded on chips, fried snacks, fast food snacks, spring rolls, mayonnaise-based sauces, cooking fat and butter, peanut sauce, ketchup, sweets, and diet sodas. The “Cosmopolitan” pattern heavily loaded on leafy vegetables, tomatoes, and allium vegetables, refined grains, fish, dressings, tomato sauce, cream, low-fat margarine, sweet sandwich spread, and wine.

Table 1. Rotated Component Loadings for the 4 Major Principal Components of 87 Food Items/Groups From the Food-Frequency Questionnaire of the Lynch Syndrome Cohort
FoodsFactor 1Factor 2Factor 3Factor 4
  1. Factor loading less than |0.15| were omitted for simplicity, loadings greater than |0.29| are bold.

  2. Foods for which all loadings were less than |0.15| were not shown: bread (multigrain), low-fat milk yoghurt and custard, dark chocolate, and fruit juice.

Cruciferous vegetables0.63
Leafy vegetables0.540.30
Allium vegetables0.37−0.210.34
Other vegetables0.210.17
Citrus and kiwi fruit0.48−0.25
Apples and pears0.56−0.16−0.18
Other fruits0.49−0.16
Potatoes, cooked0.28−0.25
Refined pasta, noodles, and rice0.260.30
Whole grain pasta, noodles, and rice0.32−0.38
Breakfast cereals, low/medium fiber0.17
Cereals high in fiber0.21
White rusk, matza, cream crackers−0.19
White bread−0.26
Whole wheat rusk, whole wheat cracker0.26−0.20
Brown bread−0.17−0.27
Whole wheat bread, rye bread0.19−0.24
Milk and fruit juice–based breakfast0.16
Fat milk−0.19
Nonfat milk0.28
Low fat yoghurt and curd−0.21
Nonfat yoghurt, custard, and curd0.54
Curd, pudding and mousse, ice cream0.21
Cream 0.32
Coffee milk (fat)0.20
Coffee milk (low fat)0.17−0.20
Low fat cheese0.40
Cheese (fat)−0.270.28
Cheese (luxury and fat)0.23
Organ meat0.17
Other meat0.16
Minced meat0.430.21
Processed meat0.40−0.17
Cooking fat and butter−0.16−0.340.21
Low-fat margarine0.20−0.37
Mayonnaise-based sauces0.46
Tomato sauce0.39
Peanut sauce−0.160.37
Mushroom cream sauce0.190.22
Fried snacks−
Fast food snacks−0.200.44
Spring rolls0.33
Kebab (snack)0.25
Cream cracker with spread0.17
Nuts and seeds0.24
Sandwich spread0.18
Peanut butter−0.44
Sweet sandwich spread−0.25−0.39
Cakes and cookies−0.32−0.27
Added sweet−0.45−0.19
Chocolates, milk and white−0.19−0.290.25
Black tea−0.16
Green and herbal tea0.39−0.25
Vegetable juice0.17−0.19−0.17
Nonsugar (diet) soda0.36
Vegetarian products0.22−0.54
Soy dessert−0.31

Baseline Characteristics

Cases were more often men, slightly older, and lower educated compared to the total cohort. There were more current smokers among the cases, and alcohol intake appeared to be slightly higher. In addition, cases were more likely to have had colorectal adenomas in the past (data not shown). Table 2 shows baseline characteristics of the cohort by tertiles of each dietary pattern score. Participants with higher “Prudent” pattern scores were older, more likely to be women, more physically active, less likely to be current smokers, and had lower energy intakes compared with those with low “Prudent” pattern scores. Participants with high “Meat” pattern scores were older, more likely to be men, less likely to have higher education, more often current smokers, had a higher BMI, slightly lower energy intakes, and slightly higher median alcohol intakes compared with participants in the lowest tertile of the “Meat” pattern scores. Participants with higher “Snack” pattern scores were more likely to be women, tended to be younger, had a higher BMI, and had slightly higher median alcohol intakes than those with low “Snack” pattern scores. Participants with higher “Cosmopolitan” pattern scores were more likely to be women, slightly younger, more likely to have college or university education, to be less physically active, had higher median alcohol intakes, lower BMIs, and used nonsteroidal anti-inflammatory drugs more regularly than those with low “Cosmopolitan” pattern scores.

Table 2. Baseline Characteristics of the Lynch Syndrome Cohort Stratified by Tertile of Dietary Pattern Scores
Dietary Pattern Tertile 1Tertile 2Tertile 3
  • Abbreviations: BMI, body mass index; MMR, DNA mismatch repair; NSAID, nonsteroidal anti-inflammatory drug.

  • a

    Higher education: a college or university education.

  • b

    High physical activity: highest tertile of the physical activity score.

  • c

    Regular NSAID use: one or more times per month.

  • d

    Time between last colonoscopy before and first colonoscopy after baseline, 7 carriers did not have a colonoscopy before baseline, 100 carriers did not have a colonoscopy during follow-up, for 23 of these, time since last colonoscopy was longer than 24 months.

Factor 1: “Prudent” pattern    
Total cohortn161165160
Adenomatous polyp casesn231916
Age, ymedian45.249.253.7
Sex, female%40.463.074.4
Education, highera%31.136.434.4
BMI, kg/m2median24.624.524.5
Energy intake, kcal/daymedian2423.12103.11786.1
Physical activity, highb%
Smoking status    
Alcohol intake, g/daymedian8.57.35.5
NSAID use, regularc%13.713.310.6
MMR gene mutation    
History of colorectal cancer%21.123.633.8
History of other cancer%8.721.824.7
History of colorectal adenoma%
Colon surgery    
 partial colon resection%14.917.025.0
 Subtotal colectomy%
No. of colonoscopies during follow-upmedian1.01.01.0
Time (months) between colonoscopiesd    
Factor 2: “Meat” pattern    
Total cohortn160165161
Adenomatous polyp casesn121630
Age, ymedian44.651.254.1
Sex, female%65.659.452.8
Education, highera%48.127.326.7
BMI, kg/m2median24.124.425.6
Energy intake, kcal/daymedian2201.42053.11909.9
Physical activity, highb%34.437.827.1
Smoking status    
Alcohol intake, g/daymedian4.58.19.2
NSAID use, regularc%13.113.910.6
MMR gene mutation    
History of colorectal cancer%29.324.824.2
History of other cancer%13.823.617.4
History of colorectal adenoma%
Colon surgery    
 partial colon resection%30.218.818.6
 Subtotal colectomy%
No. of colonoscopies during follow-upmedian1.01.01.0
Time (months) between colonoscopiesd    
 ≤ 24%60.659.454.0
Factor 3: “Snack” pattern    
Total cohortn160166160
Adenomatous polyp casesn172318
Age, ymedian57.350.041.9
Sex, female%57.553.067.5
Education, highera%38.128.935.0
BMI, kg/m2median24.024.524.9
Energy intake, kcal/daymedian2111.82079.92006.1
Physical activity, highb%34.436.029.1
Smoking status    
Alcohol intake, g/daymedian9.57.04.7
NSAID use, regularc%10.010.816.9
MMR gene mutation    
History of colorectal cancer%
History of other cancer%28.817.58.8
History of colorectal adenoma%37.534.324.4
Colon surgery    
 partial colon resection%29.416.910.6
 Subtotal colectomy%
No. of colonoscopies during follow-upmedian1.01.01.0
Time (months) between colonoscopiesd    
 ≤ 24%61.360.252.5
Factor 4: “Cosmopolitan” pattern    
Total cohortn160165161
Adenomatous polyp casesn201622
Age, ymedian52.849.548.1
Sex, female%57.559.460.9
Education, highera%21.335.844.7
BMI, kg/m2median25.124.424.2
Energy intake, kcal/daymedian2092.52026.12060.6
Physical activity, highb%43.030.926.0
Smoking status    
Alcohol intake, g/daymedian2.78.511.3
NSAID use, regularc%6.915.814.9
MMR gene mutation    
History of colorectal cancer%23.827.327.3
History of other cancer%18.119.428.0
History of colorectal adenoma%37.527.331.7
Colon surgery    
 Partial colon resection%19.43031
 Subtotal colectomy%7.510.39.9
No. of colonoscopies during follow-upmedian1.01.01.0
Time (months) between colonoscopiesd    

Influence of Dietary Patterns on Adenoma Development

During a median follow-up of 20 months, 58 of 486 (12%) carriers of the MMR gene mutation developed histologically confirmed colorectal adenomas. Thirteen of these adenomas had advanced adenoma pathology (ie, tumor size larger than 1 cm, with villous architecture, or high-grade dysplasia).

Associations between the 4 dietary patterns and colorectal adenomas are presented in Table 3. Persons within the highest “Prudent” pattern scores (third tertile) had a HR of 0.61 (95% CI, 0.28-1.32) of developing colorectal adenomas, compared with persons in the lowest tertile and adjusted for age and sex. With additional adjustment for smoking, colorectal adenoma history, and extent of colon resection, the HR of developing colorectal adenoma for the highest tertile of “Prudent” pattern scores was 0.73 (95% CI, 0.32-1.66) compared with the lowest tertile. For the “Meat” pattern, the HR for the highest tertile was 2.48 (95% CI, 1.22-5.02). After additional adjustment for smoking, colorectal adenoma history, and extent of colon resection, a statistically nonsignificant HR of 1.70 (95% CI, 0.83-3.52) was observed for the highest tertile of the “Meat” pattern scores versus the lowest tertile. Those within the highest tertile of “Snack” pattern scores had an increased risk of developing colorectal adenomas (“Snack” pattern: HR, 2.13; 95% CI, 0.99-4.60) compared with the lowest tertile, adjusted for age and sex. With additional adjustment for smoking, colorectal adenoma history, and extent of colon resection, the HR of developing colorectal adenoma for the highest tertile was 2.16 (95% CI, 1.03-4.49). The highest tertile of “Cosmopolitan” pattern scores had a higher HR (age- and sex-adjusted HR, 1.25; 95% CI, 0.64-2.43) of colorectal adenoma development than the lowest tertile. No change in HR was observed after adjustment for smoking, colorectal adenoma history, and extent of colon resection (“Cosmopolitan” pattern: HR, 1.25; 95% CI, 0.61-2.55).

Table 3. Hazard Ratios (95% Confidence Interval) of Colorectal Adenomas Occurrence According to Tertiles of Dietary Pattern Scores of the Lynch Syndrome Cohort
Dietary patternTertile 1Tertile 2Tertile 3P for Trend
(Low) (High)
HRHR (95%CI)HR (95%CI)
  • Abbreviation: HR, hazard ratio; CI, confidence interval.

  • a

    Adjusted for age, sex, smoking habits, colorectal adenoma history, and extent of colon resection.

Factor 1: “Prudent” pattern    
 Cases/total cohort23/16119/16516/160 
 Age- and sex-adjusted1.00.77 (0.41-1.45)0.61 (0.28-1.32).39
 Multivariate-adjusteda1.00.85 (0.47-1.54)0.73 (0.32-1.66).78
Factor 2: “Meat” pattern    
 Cases/total cohort12/16016/16530/161 
 Age- and sex-adjusted1.01.29 (0.61-2.75)2.48 (1.22-5.02).02
 Multivariate-adjusteda1.01.05 (0.49-2.28)1.70 (0.83-3.52).21
Factor 3: “Snack” pattern    
 Cases/total cohort17/16023/16618/160 
 Age- and sex-adjusted1.01.80 (0.96-3.40)2.13 (0.99-4.60).08
 Multivariate-adjusteda1.01.93 (1.04-3.60)2.16 (1.03-4.49).12
Factor 4: “Cosmopolitan” pattern    
 Cases/total cohort20/16016/16522/161 
 Age- and sex-adjusted1.00.74 (0.43-1.27)1.25 (0.64-2.43).49
 Multivariate-adjusteda1.00.79 (0.45-1.38)1.25 (0.61-2.55).56

Inclusion of BMI in the models did not substantially change HRs of all dietary patterns. Extra adjustment for energy intake, which might be considered as part of the dietary patterns or as intermediate variable of the associations between dietary patterns and colorectal adenoma development, changed HRs in 3 of the 4 dietary patterns with more than 10% (“Prudent” pattern: HR, 0.51; 95% CI, 0.21-1.20; “Meat” pattern: HR, 1.44; 95% CI, 0.66-3.11; “Snack” pattern: HR, 1.62; 95% CI, 0.79-3.76; “Cosmopolitan” pattern: HR, 1.35; 95% CI, 0.66-2.76).

Sensitivity analysis showed that, assuming that all persons without a colonoscopy would have colorectal adenomas, the association between the “Snack” pattern and colorectal adenomas also was statistically significant (HR, 2.02; 95% CI, 1.30-3.13). Restricting analyses to persons with at least 1 colonoscopy during follow-up (n = 386) did not markedly change associations (“Prudent” pattern: HR, 0.73; 95% CI, 0.33-1.64; “Meat” pattern: HR, 1.61; 95% CI, 0.77-3.33; “Snack” pattern: HR, 2.40; 95% CI, 1.15-5.03; “Cosmopolitan” pattern: HR, 1.35; 95% CI, 0.65-2.79).


  1. Top of page
  2. Abstract
  7. Acknowledgements

We identified 4 dietary patterns, referred to as “Prudent,” “Meat,” “Snack,” and “Cosmopolitan” pattern and observed a statistically significant increased risk of adenomas for individuals within the highest tertile of the “Snack” pattern as compared to the lowest tertile. For the “Prudent” pattern, a modest nonstatistically significant inverse association with colorectal adenomas was observed, comparing those in the highest tertile of intake with the lowest. The “Meat” and “Cosmopolitan” patterns showed nonstatistically significant positive associations for carriers within the highest as compared with the lowest tertiles.

Previous studies from our group observed that fruit and possibly dietary fiber influenced risk of developing colorectal neoplasms in LS families.17, 19 No other studies on LS and diet or dietary patterns have been conducted so far. Findings observed in the current study were consistent with associations between dietary patterns and colorectal adenomas in general population cohorts.14-16 In a cohort of US men,14 2 major dietary patterns were obtained, ie, “Prudent” and “Western,” comparable to the “Prudent” and the “Snack” and “Meat” patterns in our cohort. The study observed an increased risk of distal colorectal adenomas with higher “Western” pattern scores, whereas, as in our study, a substantial inverse association was not observed for the “Prudent” pattern.14 In a cohort of French women,15 4 patterns were identified, ie, “Healthy,” “Western,” “Drinker,” and “Meat eaters” pattern. Similar to our study, the “Healthy” pattern, largely comparable with our “Prudent” pattern, showed a statistically nonsignificant inverse association with colorectal adenomas. An increased colorectal cancer risk was seen with high “Meat” pattern scores, but no increased risk of adenomas. The “Western” pattern, a combination of our “Meat,” “Snack,” and “Cosmopolitan” patterns, was associated with an increased risk of colorectal adenomas. In a European study16 on adenoma recurrence, patterns were derived for men and women separately. High pattern scores of the “Mediterranean” pattern, mostly comparable to our “Prudent” pattern, were associated with a decreased risk of adenoma recurrence in women only. No associations were seen between the other patterns and recurrence of colorectal adenomas.

In our study, as well as in the studies mentioned above, PCA was used to identify dietary patterns. A criticism of this data-driven approach is that the component's validity is dependent on the study population. Identified patterns reflect actual existing dietary behavior within the studied population. In different populations, or in the same population at a different time, another set of components might have been observed.13 This limits interpretation of these dietary patterns and may explain differences between studies, especially differences between studies from different countries with different eating and lifestyle habits. All mentioned studies,14-16 including ours, identified a vegetable and fruit pattern, indicating that this pattern does exist in several populations. Furthermore, comparison of our dietary patterns with those from a general Dutch population cohort indicated that our “Meat,” “Snack,” and “Cosmopolitan” patterns were similar to the “Traditional,” “Refined foods,” and “Cosmopolitan” patterns of this other cohort,26 suggesting that our patterns reflect existing dietary patterns in the general Dutch population.

Using PCA requires subjective decisions about grouping of input variables, the number of retained components, method of rotation, and labeling of patterns. To study the influence of food grouping on the PCA results, we performed a PCA with the 183 original food items from the FFQ. This PCA gave us essentially the same dietary patterns, indicating minimal influence of grouping. To validate patterns, we should have performed PCAs in 2 random samples of the cohort. However, splitting the cohort made both groups too small (n = 243) to perform a PCA with 87 variables. To make all choices in retaining the number of components transparent, we described all steps in the Materials and Methods section.

The strength of the observed associations between dietary patterns and adenoma development seem relatively strong compared to those in sporadic adenoma studies,14-16 considering the short follow-up. In LS, only one hit is needed to disrupt functioning of the MMR system. The first hit, a germline mutation in 1 allele of a MMR gene, is already present from birth. Considering this, we might indeed expect to observe associations on a shorter term than in general population studies. Associations with risk factors capable of causing the hit might also be stronger than when 2 or more hits are needed. Two hits will not necessarily be the result of the same factor, and associations with an event might be weaker for a factor when there are multiple factors needed for this event to occur.

The dietary patterns were associated with other lifestyle factors; for example, smoking confounded associations between dietary patterns and colorectal adenoma development. To control for this, we performed a multivariate analysis with adjustments for lifestyle factors. Still, we cannot completely rule out residual confounding effects because of possible unmeasured confounding variables or variables measured with error. Although thus far this is the largest cohort of MMR gene mutation carriers, power to detect statistically significant associations may have been limited. In addition, because of insufficient power, it was not possible to perform subgroup analysis by sex, history of colorectal tumors, or MMR genes.

An important strength of the study is the dietary pattern approach, in which the whole diet is considered, not simply individual foods. This approach takes possible interactions between foods into account and reduces the number of dietary variables, using correlations between these variables, and as such diminishes problems of multicollinearity. Other strengths are the prospective design, the large cohort of MMR gene mutation carriers, and high percentage of carriers willing to participate. This makes these results generalizable to regularly screened LS patients in other clinical series.

Our study provides information on dietary risk factors for development of adenomas in patients with LS. The goal is to develop lifestyle and dietary recommendations in order to decrease risk of developing CRC in this group. However, such recommendations are only valid if CRC associated with LS develops via the adenoma-carcinoma sequence. The fact that risk of CRC substantially decreases by removal of adenomas in prospective surveillance studies suggests that the adenoma-carcinoma sequence is also applicable in LS, and that decreasing the risk of adenomas by adjusting dietary and lifestyle factors will also decrease risk of CRC.27

In conclusion, our findings suggest that dietary patterns may be associated with risk of colorectal adenomas in MMR gene mutation carriers. Directions of these findings were corroborative with those observed in cohorts investigating sporadic CRC. Although more research is needed to estimate the exact influence of dietary patterns on LS colorectal carcinogenesis, modifiable factors, such as diet, could influence development of colorectal neoplasms in LS.


  1. Top of page
  2. Abstract
  7. Acknowledgements

We are indebted to all study participants for their cooperation. We thank Mary Velthuizen and Alice Donselaar (Netherlands Foundation for Detection of Hereditary Tumors), Maria van Vugt (Radboud University Nijmegen Medical Center), and Leontien Witjes (Wageningen University) for assistance with participant recruitment and data collection. The medical specialists of the participants are gratefully acknowledged for their collaboration.


  1. Top of page
  2. Abstract
  7. Acknowledgements

This work was supported by the Dutch Cancer Society (grant UW-2005-3275), the Wereld Kanker Onderzoeksfonds Nederland (WCRF-NL), and the World Cancer Research Fund International.


F.M. Nagengast is member of the scientific board of Sensus, part of Royal Cosun (food industry). All other authors made no disclosure.


  1. Top of page
  2. Abstract
  7. Acknowledgements
  • 1
    Barrow E, Alduaij W, Robinson L, et al. Colorectal cancer in HNPCC: cumulative lifetime incidence, survival and tumour distribution. A report of 121 families with proven mutations. Clin Genet. 2008; 74: 233-242.
  • 2
    Jenkins MA, Baglietto L, Dowty JG, et al. Cancer risks for mismatch repair gene mutation carriers: a population-based early onset case-family study. Clin Gastroenterol Hepatol. 2006; 4: 489-498.
  • 3
    Quehenberger F, Vasen HF, van Houwelingen HC. Risk of colorectal and endometrial cancer for carriers of mutations of the hMLH1 and hMSH2 gene: correction for ascertainment. J Med Genet. 2005; 42: 491-496.
  • 4
    Hampel H, Stephens JA, Pukkala E, et al. Cancer risk in hereditary nonpolyposis colorectal cancer syndrome: later age of onset. Gastroenterology. 2005; 129: 415-421.
  • 5
    Watson P, Vasen HF, Mecklin JP, et al. The risk of extra-colonic, extra-endometrial cancer in the Lynch syndrome. Int J Cancer. 2008; 123: 444-449.
  • 6
    Abdel-Rahman WM, Mecklin JP, Peltomaki P. The genetics of HNPCC: application to diagnosis and screening. Crit Rev Oncol Hematol. 2006; 58: 208-220.
  • 7
    Ligtenberg MJL, Kuiper RP, Chan TL, et al. Heritable somatic methylation and inactivation of MSH2 in families with Lynch syndrome due to deletion of the 3′ exons of TACSTD1. Nat Genet. 2009; 41: 112-117.
  • 8
    de Jong AE, Morreau H, Van Puijenbroek M, et al. The role of mismatch repair gene defects in the development of adenomas in patients with HNPCC. Gastroenterology. 2004; 126: 42-48.
  • 9
    Park JG, Park YJ, Wijnen JT, Vasen HF. Gene-environment interaction in hereditary nonpolyposis colorectal cancer with implications for diagnosis and genetic testing. Int J Cancer. 1999; 82: 516-519.
  • 10
    Mitchell RJ, Farrington SM, Dunlop MG, Campbell H. Mismatch repair genes hMLH1 and hMSH2 and colorectal cancer: a HuGE review. Am J Epidemiol. 2002; 156: 885-902.
  • 11
    World Cancer Research Fund (WCRF)/American Institute for Cancer Research (AICR). Food, Nutrition, Physical Activity, and the Prevention of Cancer: A Global Perspective. 2nd ed. Washington, DC: AICR; 2007.
  • 12
    Meyerhardt JA, Niedzwiecki D, Hollis D, et al. Association of dietary patterns with cancer recurrence and survival in patients with stage III colon cancer. JAMA. 2007; 298: 754-764.
  • 13
    Flood A, Rastogi T, Wirfalt E, et al. Dietary patterns as identified by factor analysis and colorectal cancer among middle-aged Americans. Am J Clin Nutr. 2008; 88: 176-184.
  • 14
    Wu K, Hu FB, Fuchs CS, Rimm ER, Willett WC. Dietary patterns and risk of colon cancer and adenoma in a cohort of men (United States). Cancer Causes Control. 2004; 15: 853-862.
  • 15
    Kesse E, Clavel-Chapelon F, Boutron-Ruault MC. Dietary patterns and risk of colorectal tumors: a cohort of French women of the National Education System (E3N). Am J Epidemiol. 2006; 164: 1085-1093.
  • 16
    Cottet V, Bonithon-Kopp C, Kronborg O, et al; European Cancer Prevention Organisation Study Group. Dietary patterns and the risk of colorectal adenoma recurrence in a European intervention trial. Eur J Cancer Prev. 2005; 14: 21-29.
  • 17
    Voskuil DW, Kampman E, Grubben MJ, et al. Meat consumption and meat preparation in relation to colorectal adenomas among sporadic and HNPCC family patients in the Netherlands. Eur J Cancer. 2002; 38: 2300-2308.
  • 18
    Watson P, Ashwathnarayan R, Lynch HT, Roy HK. Tobacco use and increased colorectal cancer risk in patients with hereditary nonpolyposis colorectal cancer (Lynch syndrome). Arch Intern Med. 2004; 164: 2429-2431.
  • 19
    Diergaarde B, Braam H, Vasen HF, et al. Environmental factors and colorectal tumor risk in individuals with hereditary nonpolyposis colorectal cancer. Clin Gastroenterol Hepatol. 2007; 5: 736-742.
  • 20
    Pande M, Lynch PM, Hopper JL, et al. Smoking and colorectal cancer in Lynch syndrome: results from the Colon Cancer Family Registry and the University of Texas M.D. Anderson Cancer Center. Clin Cancer Res. 2010; 16: 1331-1339.
  • 21
    Botma A, Nagengast FM, Braem MGM, et al. Body mass index increases risk of colorectal adenomas in men with Lynch syndrome: the GEOLynch cohort study. J Clin Oncol. 2010; 28: 4346-4353.
  • 22
    Baecke J, Burema J, Frijters J. A short questionnaire for the measurement of habitual physical activity in epidemiological studies. Am J Clin Nutr. 1982; 36: 936-942.
  • 23
    Feunekes G, Van Staveren W, De Vries J, Burema J, Hautvast J. Relative and biomarker-based validity of a food-frequency questionnaire estimating intake of fats and cholesterol. Am J Clin Nutr. 1993; 58: 489-496.
  • 24
    Verkleij-Hagoort AC, de Vries JH, Stegers MP, Lindemans J, Ursem NT, Steegers-Theunissen RP. Validation of the assessment of folate and vitamin B12 intake in women of reproductive age: the method of triads. Eur J Clin Nutr. 2006; 61: 610-615.
  • 25
    Netherlands Nutrition Center. NEVO (Nederlands Voedingsstoffenbestand): Dutch Food Composition Table, 2001. The Hague, Netherlands: Netherlands Nutrition Center; 2001.
  • 26
    van Dam RM, Grievink L, Ocke MC, Feskens EJ. Patterns of food consumption and risk factors for cardiovascular disease in the general Dutch population. Am J Clin Nutr. 2003; 77: 1156-1163.
  • 27
    Järvinen HJ, Mecklin J-P, Sistonen P. Screening reduces colorectal cancer rate in families with hereditary nonpolyposis colorectal cancer. Gastroenterology. 1995; 108: 1405-1411.