Correlation coefficients between energy-adjusted glycemic load and major food and nutrient groups, tested with the Pearson's product moment correlation coefficients and Spearman's rank-order correlation coefficient for total alcohol intake.
1097-0215/asset/olbannerleft.jpg?v=1&s=45719cd7de57873027993264fcc568b335a8cd56)
1097-0215/asset/olbannerright.jpg?v=1&s=5e0fba63c1309b3036eb9215a0e1e83dd02efd19)
Epidemiology
You have full text access to this OnlineOpen article
Dietary glycemic load, glycemic index and colorectal cancer risk: Results from the Netherlands Cohort Study
Article first published online: 12 OCT 2007
DOI: 10.1002/ijc.23110
Copyright © 2007 Wiley-Liss, Inc.
Additional Information
How to Cite
Weijenberg, M. P., Mullie, P. F.F., Brants, H. A.M., Heinen, M. M., Goldbohm, R. A. and van den Brandt, P. A. (2008), Dietary glycemic load, glycemic index and colorectal cancer risk: Results from the Netherlands Cohort Study. Int. J. Cancer, 122: 620–629. doi: 10.1002/ijc.23110
Publication History
- Issue published online: 26 NOV 2007
- Article first published online: 12 OCT 2007
- Manuscript Accepted: 11 JUL 2007
- Manuscript Received: 23 JAN 2007
- Abstract
- Article
- References
- Cited By
Keywords:
- colorectal cancer;
- glycemic load;
- glycemic index;
- food frequency;
- prospective study;
- Netherlands Cohort Study
Abstract
Since hyperinsulinemia is implicated in the development of colorectal cancer, determinants of serum insulin levels, like the glycemic load and the glycemic index of the diet, could influence cancer risk. Our objective was to evaluate whether a diet with a high glycemic load or glycemic index is associated with increased colorectal cancer risk. In the Netherlands Cohort Study, 120,852 subjects completed a food frequency questionnaire in 1986. After 11.3 years of follow-up, 1,225 colon and 418 rectal cancer cases were available for analysis. A case–cohort approach was used to estimate multivariate adjusted rate ratios and 95% confidence intervals for quintiles of energy-adjusted glycemic load and glycemic index. The RR for colorectal cancer comparing the highest versus the lowest quintile levels of glycemic load and glycemic index were 0.83 (95% CI: 0.64–1.08) and 0.81 (95% CI: 0.61–1.08) for men and 1.00 (95% CI: 0.73–1.36) and 1.20 (95% CI: 0.85–1.67) for women. In general, no clear associations with cancer subsites were observed. Glycemic load and glycemic index were borderline significantly associated with an increased risk of proximal colon cancer in women (p-trend = 0.06 and 0.08, respectively), however, these associations were attenuated after exclusion of the first 2 years of follow-up (p-trend = 0.165 and 0.254, respectively). In men, glycemic index was associated with a reduced risk of distal colon cancer (p-trend = 0.03). Overall, our findings do not support the hypothesis that a diet with a high glycemic load or index is associated with a higher risk of colorectal cancer. © 2007 Wiley-Liss, Inc.
Colorectal cancer is one of the most important cancers in Western society.1 Although the 5-year survival rate for treatment of colorectal cancer is high, a preventive approach is always preferred. Dietary and other modifiable factors have been estimated to account for 90% of colorectal cancers.2, 3, 4
Scientific evidence indicates that insulin is implicated in colorectal carcinogenesis.3, 5 Risk factors for colorectal cancer, including being overweight or obese, visceral adiposity, absence of physical activity and diabetes are linked to hyperinsulinemia. Insulin has a role in energy metabolism, and can stimulate anabolic processes as a function of available energy. The anabolic signals can promote tumor development by inhibiting apoptosis and by stimulating cell proliferation.6 Giovannucci3 and McKeown-Eyssen5 proposed an insulin-hypothesis to unify many of the apparently independent risk factors for colorectal cancer. The suggestion is that diets high in energy and saturated fat and with high glycemic index carbohydrates and low levels of fiber and poly-unsaturated fatty acids lead to insulin resistance.7 Insulin resistance is defined as the resistance of cells to the action of insulin, so that the cellular uptake of glucose is slower. As a consequence blood glucose levels increase, leading to increased blood insulin levels.
If the insulin-hypothesis is correct, determinants of serum-insulin levels should influence colorectal cancer risk. The amount and composition of the diet, the rate of hydrolysis of nutrients in the gastrointestinal system and the rate of gastric emptying determine the absorption rate which, in turn, influences the extent and duration of the blood glucose rise after a meal. The dietary glycemic index provides an indication of the postprandial glycemic effect of foods (compared to the glucose response of a reference food, usually white bread or glucose) and of the postprandial blood insulin level.8, 9 Foods with high glycemic index produce a higher secretion of insulin than foods with a low glycemic index. Following the insulin-hypothesis, a regular consumption of foods with high glycemic index can increase the risk of developing hyperinsulinemia8 and associated diseases such as colorectal cancer. The glycemic index represents the quality of consumed carbohydrate, and not the quantity. The glycemic load, equal to the product of glycemic index of the diet and the consumed quantity of carbohydrates, estimates the total glycemic effect of the diet.10, 11, 12, 13, 14, 15, 16
In three case–control studies, the glycemic index10, 12, 15 and sometimes the glycemic load10 were reported to be associated with an increased risk of colorectal cancer and different sub-sites. In prospective cohort studies, associations were much more inconsistent with colorectal cancer11, 13, 16, 17, 18 and adenomas.14, 19
In the present article, the relation between glycemic load, glycemic index and the incidence of colorectal cancer is described within a large prospective cohort study.
Material and methods
Study population and follow up of cancer
The investigation was performed within the framework of the Netherlands Cohort Study on diet and cancer, that started in September 1986.20 The cohort consisted of 120,852 subjects (48% men and 52% women), aged 55–69 years at the beginning of the study, who completed a baseline questionnaire. This self-administered mailed questionnaire covered dietary habits, lifestyle, smoking, family history of cancer and demographic data. A case–cohort approach was used for data processing and analysis. Case subjects were enumerated from the entire cohort, whereas the person-years at risk were estimated from a random sample of 5,000 subjects, taken from the cohort at baseline. The study design was described in detail elsewhere.20
Follow-up of colon and rectal cancers was established by using a combination of a computerized linkage system to 9 cancer registries in the Netherlands and a nationwide pathology database.21 The completeness of cancer follow-up was estimated to be nearly 100%.21, 22 Colorectal cancers were classified as proximal colon cancer (International Classification of Diseases for Oncology, first edition codes 153.0, 153.1, 153.4, 153.5 and 153.6), distal colon cancer (codes 153.2, 153.3 and 153.7), rectosigmoid (code 154.0) and rectal cancer (code 154.1).
All prevalent cancer cases at baseline other than nonmelanoma skin cancer were excluded. A total of 1,361 colon (736 men and 625 women) and 450 rectal (295 men and 155 women) cancer cases were detected in the cohort after 11.3 years of follow-up.
Semiquantitative food-frequency questionnaire
The subjects completed a semiquantitative food-frequency questionnaire that included 150 food items and covered habitual food habits during the year before the start of the study. They could indicate their frequency of consumption by choosing predefined frequency categories and the portion size per consumption frequency in natural or household units or grams, depending on the type of food. The questionnaire data of all cases and subcohort members were processed in a manner blinded with respect to case/subcohort status to minimize observer bias in the coding and interpretation of data. The questionnaire was validated and tested for reproducibility.23, 24 Nutrient intakes were calculated from each food on the questionnaire as the frequency of consumption multiplied by the number of units, the size of a unit and the nutrient content of the food, using a computerized Dutch food composition table.25 Subjects with incomplete dietary data were excluded from the analyses (7.0%). Data were considered incomplete when 60 or more questionnaire items were blank and when fewer than 35 items at least were eaten once per month.
Glycemic index and glycemic load
The Foster–Powell glycemic index table26 was used as reference to calculate the glycemic index of foods consumed in the Netherlands Cohort Study on Diet and Cancer. The glycemic index ranks foods on how they affect blood glucose levels. This index measures how much blood glucose increases after eating foods with carbohydrates, using glucose or white bread as the reference. Per gram of carbohydrate, foods with a high glycemic index produce a higher peak in postprandial blood glucose than do foods with a low glycemic index.27 In the Foster–Powell table, glycemic index was expressed both with glucose and white bread as reference food. We chose the values with glucose as reference. When more than one glycemic index value was represented in the Foster–Powell glycemic index table, the average was used as reference glycemic index. There was a lack of information about glycemic index of vegetables, therefore a mean glycemic index was calculated for usually consumed vegetables in the Netherlands Cohort Study. Food items for which a glycemic index had not been determined were assigned the glycemic index of the nearest comparable food. A glycemic index for beer was not found, and therefore estimated using the type of carbohydrates (65% maltose, 35% glucose). The glycemic index of mixed meals or combinations of foods was calculated following the method of Wolever.27, 28, 29 For over 90% of the carbohydrate intake of each subject a glycemic index value was available. The overall dietary glycemic index (hereafter referred to as the glycemic index) was estimated for each participant by calculating the weighted average glycemic index of all food items eaten, using the carbohydrate intake from that item (gram/day) as weighting factor. The glycemic index represents the overall quality of carbohydrate intake for each participant. The glycemic load is the product of the dietary glycemic index and the amount of consumed dietary carbohydrate, divided by 100. Each unit of dietary glycemic load thus represents the equivalent of 1 g carbohydrate from pure glucose.30
Statistical analysis
Glycemic load and glycemic index were adjusted for energy intake using the residual method as described by Willett and Stampfer.31 Body mass index [BMI = weight (kg)/height2 (m2)] was calculated from self-reported height and weight of the individuals. Pearson's product moment correlation coefficients between glycemic load, glycemic index and selected nutrient intakes were calculated. Because of the non-normal distribution of alcohol intake, Spearman's rank-order correlation coefficient was used. χ2 tests were conducted to test the difference in distribution of categorical variables according to quintiles of energy-adjusted glycemic load and glycemic index. The aetiology of colorectal cancers may vary according to sex and sub-site of the cancer.32 Therefore, men and women were analyzed separately for colorectal (including rectosigmoid), colon, proximal colon, distal colon and rectal cancers. Since the number of patients with a rectosigmoid tumor was too small for adequate stratified analyses, and the rectosigmoid can be considered as a clinically applied term rather that an anatomically defined transitional zone between the colon and rectum, results for this specific subsite are not presented separately.
Cox proportional hazards analysis was used in the case–cohort analyses to obtain hazard rate ratios (RRs) and 95% confidence intervals (CIs) for the association between glycemic load, glycemic index and the incidence of overall colorectal, colon, proximal colon, distal colon or rectal cancers. We confirmed constancy of the baseline hazard visually by plotting the natural logarithm of the baseline survival function against failure time. The RRs were expressed with the lowest quintile of glycemic load or glycemic index as reference. The RRs were estimated in age-adjusted and in multivariate models. Potential confounders that were found to statistically significantly (p < 0.05) contribute to the multivariate model or that showed a more than10% influence on the RRs for colorectal cancer were included as covariates in the multivariate analyses. None of the studied factors changed the RRs with more than 10%. The possible confounders included in the final multivariate model were age (in years), body mass index (as continuous variable), smoking (never–ex–currently), alcohol use (in g/day), family history of colorectal cancer (yes or no), total energy intake (in kJ/day), calcium intake (in mg/day), intake of processed meat (in g/day), level of education (as categorical variable) and physical activity (less than 30 min a day, 30–60 min a day, 60–90 min a day and more than 90 min a day). Interactions were tested using cross-product terms between quintiles of glycemic load or glycemic index on one hand and body mass index (as a continuous variable), fiber from cereals (in g/day), fiber from fruit and vegetables (in g/day), intake of total fat (in g/day), family history of colorectal cancer (yes or no), physical activity and diabetes (yes or no) on the other. Those variables were selected based on previous published research. Additional analyses were conducted excluding the first 2 years of follow-up to account for potential underlying disease affecting dietary intake. Finally, RRs were also computed for total carbohydrate intake and mono- and di-saccharide intake. The analyses were performed using the Stata version 8 statistical software package (Stata Corporation, College Station, TX). Two-sided p-values are reported throughout the article.
Results
The mean (standard deviation) energy-adjusted glycemic load was 136.8 (23.9) g/day for men and 102.4 (17.1) g/day for women; for energy-adjusted glycemic index this was 60.7 (3.5) and 57.8 (3.3), respectively. Baseline characteristics according to energy-adjusted dietary glycemic load and glycemic index, grouped into quintiles are presented in Table I, respectively, for subcohort men, and Table II, respectively, for subcohort women. Total fat intake was inversely associated with the glycemic load (correlation coefficients were −0.32 for men and −0.30 for women), but there was no association between total fat intake and glycemic index. For men and women, the daily intake of alcohol in grams per day was much higher in the lowest quintile of glycemic load. The correlation coefficients were −0.38 for men and −0.37 for women, which was not the case for glycemic index. The prevalence of diabetes and ex-smoking was highest in the lowest quintile of glycemic load in men and women, although not statistically significant for diabetes in women.
| Characteristics | 1 (lowest) | 2 | 3 | 4 | 5 (highest) | |
|---|---|---|---|---|---|---|
| ||||||
| Quintiles of energy-adjusted dietary glycemic load | ||||||
| No. in subcohort | 416 | 415 | 407 | 417 | 417 | |
| Energy-adjusted glycemic load (range) | 18.6–118.7 | 118.7–130.9 | 130.9–141.4 | 141.5–155.6 | 155.6–240.3 | |
| Mean (standard deviation) | Correlation1 | |||||
| Energy-adjusted glycemic index (g/day) | 59.3 (3.5) | 59.9 (3.7) | 60.7 (3.3) | 61.4 (3.1) | 62.1 (3.1) | 0.30 |
| Body mass index (kg/m2) | 25.4 (2.6) | 25.1 (2.4) | 25.0 (2.5) | 24.6 (2.6) | 24.6 (2.7) | −0.15 |
| Energy intake (kJ/day) | 9,564 (1,966) | 8,760 (2,100) | 8,586 (2,054) | 8,957 (2,135) | 9,525 (2,244) | 0.00 |
| Total carbohydrates (g/day) | 190.7 (47.7) | 202.32 (54.6) | 214.1 (52.9) | 240.7 (55.9) | 287.4 (67.0) | 0.54 |
| Mono- and disaccharides (g/day) | 80.6 (29.9) | 89.4 (33.8) | 97.5 (36.2) | 113.8 (38.7) | 142.5 (51.1) | 0.50 |
| Total fat (g/day) | 110.7 (29.1) | 95.2 (27.4) | 90.3 (25.3) | 89.0 (26.1) | 85.2 (26.6) | −0.32 |
| Total saturated fat (g/day) | 43.0 (12.2) | 37.1 (11.4) | 35.6 (11.0) | 35.2 (11.6) | 33.4 (11.1) | −0.27 |
| Total fibers (g/day) | 27.3 (7.4) | 27.6 (8.1) | 27.5 (7.9) | 29.4 (8.9) | 31.8 (10.3) | 0.17 |
| Fibers from fruit (g/day) | 3.2 (2.2) | 3.4 (2.1) | 3.5 (2.6) | 3.6 (2.3) | 4.1 (3.1) | 0.13 |
| Fibers from vegetables (g/day) | 4.5 (1.8) | 4.3 (1.8) | 4.2 (1.9) | 4.3 (2.0) | 4.3 (2.4) | −0.05 |
| Total alcohol (g/day) | 26.8 (22.0) | 16.8 (16.9) | 12.3 (12.8) | 10.8 (12.7) | 8.4 (11.7) | −0.38 |
| Beer (n glasses/week) | 3.1 (5.9) | 3.1 (6.4) | 3.1 (6.9) | 3.3 (7.6) | 3.1 (8.1) | 0.01 |
| Potatoes (g/day) | 145 (77) | 144 (83) | 146 (82) | 154 (84) | 166 (94) | 0.10 |
| Milk and milk products (g/day) | 324 (233) | 331 (241) | 305 (217) | 305 (204) | 276 (192) | −0.07 |
| Cheese (g/day) | 27.9 (22.8) | 22.2 (19.3) | 22.8 (18.1) | 2,264 (18.8) | 19.9 (19.0) | −0.11 |
| Sweet complement on bread (g/day) | 5.5 (8.5) | 8.2 (10.0) | 8.3 (10.3) | 11.4 (12.3) | 13.8 (17.0) | 0.22 |
| Added sugar (g/day) | 12.2 (18.7) | 17.6 (21.9) | 24.4 (24.8) | 36.0 (28.1) | 58.2 (40.2) | 0.53 |
| Cake, biscuits (g/day) | 22.6 (17.2) | 24.7 (18.5) | 26.7 (18.8) | 27.0 (17.5) | 27.7 (20.9) | 0.10 |
| Candy (g/day) | 5.7 (8.4) | 4.5 (6.5) | 5.0 (7.6) | 5.1 (6.9) | 5.1 (8.0) | −0.00 |
| White bread (g/day) | 32.1 (47.5) | 33.3 (53.3) | 35.5 (54.3) | 35.5 (56.0) | 48.4 (70.1) | 0.10 |
| Meat (g/day) | 128 (45) | 112 (42) | 100 (38) | 100 (36) | 88 (41) | −0.32 |
| Processed meat (g/day) | 20.9 (20.0) | 16.7 (17.9) | 15.6 (15.9) | 16.0 (16.3) | 17.1 (17.6) | −0.07 |
| Calcium (mg/day) | 1,019 (363) | 955 (345) | 927 (330) | 939 (335) | 907 (325) | −0.29 |
| Percentage | p-value2 | |||||
| Family history of colorectal cancer | ||||||
| Yes | 3.9 | 4.6 | 6.9 | 7.0 | 4.6 | 0.144 |
| Prevalence of diabetes mellitus | ||||||
| Yes | 6.0 | 2.9 | 2.7 | 3.1 | 1.7 | 0.008 |
| Physical activity | ||||||
| less than 30 min/day | 20.0 | 17.8 | 17.4 | 17.0 | 16.8 | 0.770 |
| Cigarettes smoking | ||||||
| Currently | 36.1 | 32.8 | 33.2 | 35.0 | 41.3 | |
| Ex-smoking | 55.5 | 55.4 | 53.6 | 49.5 | 44.4 | |
| Never | 8.4 | 11.8 | 13.3 | 15.6 | 14.4 | 0.004 |
| Quintiles of energy-adjusted dietary glycemic index | ||||||
| No. in subcohort | 416 | 414 | 402 | 417 | 423 | |
| Energy-adjusted glycemic index (range) | 44.5–58.1 | 58.2–59.9 | 59.9–61.5 | 61.5–63.2 | 63.2–78.0 | |
| Mean (standard deviation) | Correlation3 | |||||
| Energy-adjusted glycemic load (g/day) | 124.7 (21.3) | 133.7 (22.1) | 137.0 (22.0) | 143.1 (23.7) | 145.0 (24.5) | 0.30 |
| Body mass index (kg/m2) | 25.0 (2.7) | 25.1 (2.3) | 24.8 (2.5) | 24.7 (2.4) | 25.1 (2.9) | −0.00 |
| Energy intake (kJ/day) | 8,857 (2,029) | 9,159 (2,057) | 9,285 (2,325) | 9,172 (2,006) | 8,941 (2,146) | 0.01 |
| Total carbohydrates (g/day) | 218.2 (60.9) | 229.4 (65.5) | 232.2 (71.4) | 233.9 (65.8) | 220.5 (63.0) | 0.00 |
| Mono- and disaccharides (g/day) | 108.8 (40.4) | 108.6 (43.4) | 108.4 (50.1) | 105.0 (42.2) | 94.6 (43.4) | −0.13 |
| Total fat (g/day) | 91.9 (27.7) | 95.0 (26.3) | 97.6 (30.3) | 95.4 (28.1) | 90.7 (28,8) | −0.03 |
| Total saturated fat (g/day) | 35.6 (11.3) | 37.5 (11.5) | 38.6 (13.2) | 37.5 (11.6) | 35.2 (11.6) | −0.03 |
| Total fibers (g/day) | 29.8 (8.5) | 29.8 (9.2) | 29.1 (8.5) | 28.6 (8.7) | 26.3 (8.3) | −0.16 |
| Fibers from fruit (g/day) | 5.3 (3.2) | 4.2 (2.5) | 3.5 (1.9) | 2.9 (1.6) | 1.9 (1.3) | −0.47 |
| Fibers from vegetables (g/day) | 4.6 (2.1) | 4.6 (2.4) | 4.3 (1.7) | 4.2 (1.8) | 4.0 (1.8) | −0.12 |
| Total alcohol (g/day) | 13.3 (17.5) | 13.8 (15.4) | 13.4 (14.3) | 13.5 (15.1) | 20.8 (20.5) | 0.15 |
| Beer (glasses/week) | 1.0 (2.2) | 1.4 (2.7) | 1.9 (3.3) | 2.4 (5.0) | 9.0 (12.4) | 0.52 |
| Potatoes (g/day) | 121 (68) | 149 (77) | 158 (89) | 168 (93) | 159 (87) | 0.14 |
| Milk and milk products (g/day) | 494 (279) | 360 (182) | 306 (172) | 227 (141) | 158 (110) | −0.55 |
| Cheese (g/day) | 22.0 (17.0) | 23.7 (20.0) | 26.7 (21.1) | 22.1 (19.9) | 21.2 (20.6) | −0.03 |
| Sweet complement on bread (g/day) | 9.4 (12.1) | 11.1 (14.9) | 10.2 (11.8) | 10.6 (12.7) | 6.0 (8.8) | −0.09 |
| Added sugar (g/day) | 13.6 (20.6) | 24.5 (27.4) | 33.2 (33.9) | 37.3 (31.9) | 39.8 (37.3) | 0.26 |
| Cake, biscuits (g/day) | 28.5 (18.9) | 29.7 (19.8) | 27.4 (19.6) | 25.4 (17.7) | 17.8 (14.6) | −0.20 |
| Candy (g/day) | 6.5 (9.1) | 6.0 (7.4) | 5.8 (8.6) | 4.5 (6.6) | 2.6 (4.4) | −0.17 |
| White bread (g/day) | 23.5 (36.0) | 38.3 (59.3) | 37.5 (55.3) | 43.4 (61.5) | 42.2 (65.9) | 0.11 |
| Meat (g/day) | 100 (45) | 106 (42) | 108 (42) | 106 (40) | 110 (44) | 0.09 |
| Processed meat (g/day) | 14.1 (16.3) | 16.9 (16.4) | 17.0 (17.0) | 18.4 (18.1) | 19.9 (19.8) | 0.09 |
| Calcium (mg/day) | 1,171 (368) | 1,027 (302) | 982 (304) | 837 (289) | 736 (259) | −0.46 |
| Percentage | p-value2 | |||||
| Family history of colorectal cancer | ||||||
| yes | 5.3 | 5.1 | 6.7 | 4.8 | 5.0 | 0.750 |
| Prevalence of diabetes mellitus | ||||||
| yes | 4.1 | 2.9 | 3.5 | 3.1 | 2.8 | 0.846 |
| Physical activity | ||||||
| less than 30 min/day | 18.3 | 14.3 | 15.9 | 17.0 | 23.4 | 0.008 |
| Cigarettes smoking | ||||||
| Currently | 24.5 | 27.8 | 39.1 | 38.4 | 48.5 | |
| Ex-smoking | 57.2 | 56.5 | 50.5 | 51.1 | 43.0 | |
| Never | 18.3 | 15.7 | 10.5 | 10.6 | 8.5 | 0.000 |
| Characteristics | 1 (lowest) | 2 | 3 | 4 | 5 (highest) | |
|---|---|---|---|---|---|---|
| ||||||
| Quintiles of energy-adjusted dietary glycemic load | ||||||
| No. in subcohort | 415 | 420 | 417 | 406 | 395 | |
| Energy-adjusted glycemic load (range) | 45.6–89.6 | 89.7–97.8 | 97.8–105.6 | 105.6–115.5 | 115.5–248.7 | |
| Mean (standard deviation) | Correlation1 | |||||
| Energy-adjusted glycemic index (g/day) | 56.2 (3.5) | 57.0 (3.2) | 57.6 (3.0) | 58.6 (2.7) | 59.6 (2.8) | 0.39 |
| Body mass index (kg/m2) | 25.0 (3.5) | 25.5 (3.6) | 25.2 (3.3) | 25.0 (3.6) | 24.5 (3.5) | −0.05 |
| Energy intake (kJ/day) | 7,458 (1,715) | 6,837 (1,529) | 6,800 (1,607) | 6,825 (1,581) | 7,416 (1,725) | −0.01 |
| Total carbohydrates (g/day) | 154.4 (39.8) | 161.1 (38.7) | 171.9 (40.3) | 184.6 (39.9) | 223.0 (49.8) | 0.51 |
| Mono- and disaccharides (g/day) | 71.2 (23.3) | 74.6 (25.7) | 79.7 (26.3) | 85.6 (27.6) | 114.7 (39.5) | 0.49 |
| Total fat (g/day) | 86.5 (25.2) | 75.5 (20.4) | 71.7 (21.2) | 68.4 (19.6) | 68.1 (21.4) | −0.30 |
| Total saturated fat (g/day) | 34.6 (10.1) | 30.5 (9.2) | 29.3 (9.5) | 27.3 (8.7) | 27.2 (9.1) | −0.28 |
| Total fibers (g/day) | 24.2 (6.8) | 24.7 (6.4) | 25.1 (6.9) | 25.9 (6.7) | 27.0 (8.1) | 0.12 |
| Fibers from fruit (g/day) | 3.9 (2.2) | 4.4 (2.5) | 4.5 (2.5) | 4.3 (2.6) | 4.7 (2.9) | 0.06 |
| Fibers from vegetables (g/day) | 4.6 (1.8) | 4.6 (2.0) | 4.3 (2.0) | 4.2 (1.8) | 4.1 (2.0) | −0.10 |
| Total alcohol (g/day) | 13.3 (14.6) | 5.9 (8.0) | 4.5 (6.8) | 3.2 (5.6) | 2.5 (5.3) | −0.37 |
| Beer (glasses/week) | 0.3 (1.7) | 0.1 (0.7) | 0.3 (1.8) | 0.1 (0.7) | 0.3 (1.9) | −0.01 |
| Potatoes (g/day) | 95 (57) | 95 (58) | 98 (57) | 104 (62) | 119 (65) | 0.15 |
| Milk and milk products (g/day) | 319 (195) | 307 (189) | 295 (187) | 291 (189) | 280 (176) | −0.09 |
| Cheese (g/day) | 28.6 (22.7) | 23.7 (17.1) | 21.7 (16.2) | 19.8 (15.7) | 18.5 (16.3) | −0.20 |
| Sweet complement on bread (g/day) | 4.5 (6.9) | 6.8 (8.0) | 7.7 (9.9) | 8.4 (8.5) | 11.7 (12.6) | 0.22 |
| Added sugar (g/day) | 2.9 (8.3) | 5.2 (11.2) | 6.9 (12.8) | 11.3 (16.4) | 31.3 (29.4) | 0.53 |
| Cake, biscuits (g/day) | 24.3 (17.0) | 24.9 (16.6) | 28.5 (18.9) | 28.9 (18.9) | 31.8 (22.8) | 0.14 |
| Candy (g/day) | 5.7 (8.2) | 4.9 (6.4) | 5.9 (7.6) | 6.0 (9.1) | 6.3 (8.1) | 0.05 |
| White bread (g/day) | 16.7 (25.7) | 20.2 (32.8) | 22.3 (35.9) | 23.4 (38.5) | 29.9 (43.9) | 0.13 |
| Meat (g/day) | 113 (41) | 100 (35) | 92 (34) | 85 (39) | 75 (38) | −0.34 |
| Processed meat (g/day) | 13.0 (13.9) | 11.1 (11.6) | 11.8 (12.1) | 10.1 (11.4) | 10.4 (13.0) | −0.06 |
| Calcium (g/day) | 983 (336) | 923 (307) | 883 (293) | 863 (309) | 853 (284) | −0.16 |
| Percentage | p-value2 | |||||
| Family history of colorectal cancer | ||||||
| yes | 4.8 | 6.4 | 7.9 | 4.7 | 5.8 | 0.269 |
| Prevalence of diabetes mellitus | ||||||
| yes | 4.6 | 3.6 | 4.6 | 2.7 | 1.8 | 0.129 |
| Physical activity | ||||||
| less than 30 min/day | 20.0 | 21.7 | 25.9 | 23.7 | 25.6 | 0.207 |
| Cigarettes smoking | ||||||
| Currently | 26.0 | 22.4 | 17.3 | 19.2 | 23.3 | |
| Ex-smoking | 30.4 | 20.0 | 18.2 | 20.9 | 15.2 | |
| Never | 43.6 | 57.6 | 64.5 | 59.9 | 61.5 | 0.000 |
| Quintiles of energy-adjusted dietary glycemic index | ||||||
| No. in subcohort | 410 | 414 | 420 | 405 | 404 | |
| Energy-adjusted glycemic index (range) | 36.7–55.1 | 55.1–57.0 | 57.0–58.7 | 58.7–60.6 | 60.6–75.2 | |
| Mean (standard deviation) | Correlation3 | |||||
| Energy-adjusted glycemic load (g/day) | 93.6 (13.8) | 99.2 (14.3) | 100.7 (15.2) | 106.7 (15.5) | 111.9 (20.1) | 0.39 |
| Body mass index (kg/m2) | 25.2 (3.3) | 25.0 (3.4) | 24.8 (3.6) | 25.0 (3.5) | 25.2 (3.9) | 0.00 |
| Energy intake (kJ/day) | 7,007 (1,633) | 7,134 (1,610) | 7,169 (1,577) | 7,089 (1,730) | 6,917 (1,734) | −0.01 |
| Total carbohydrates (g/day) | 175.6 (45.6) | 179.6 (45.6) | 177.9 (44.5) | 181.2 (50.7) | 178.1 (54.0) | 0.03 |
| Mono- and disaccharides (g/day) | 92.6 (27.6) | 87.3 (29.2) | 83.7 (29.6) | 82.1 (33.4) | 78.4 (40.9) | −0.14 |
| Total fat (g/day) | 71.6 (22.2) | 74.3 (22.1) | 76.2 (21.8) | 75.5 (23.9) | 73.0 (23.1) | 0.03 |
| Total saturated fat (g/day) | 29.1 (9.2) | 30.0 (9.8) | 30.7 (9.4) | 30.3 (10.2) | 29.0 (10.0) | 0.01 |
| Total fibers (g/day) | 27.0 (7.5) | 26.7 (6.7) | 25.1 (6.6) | 24.9 (6.7) | 23.0 (6.9) | −0.19 |
| Fibers from fruit (g/day) | 6.3 (3.0) | 5.3 (2.4) | 4.2 (1.9) | 3.5 (1.7) | 2.4 (1.4) | −0.53 |
| Fibers from vegetables (g/day) | 5.1 (2.2) | 4.5 (1.9) | 4.3 (1.9) | 4.1 (1.6) | 3.8 (1.8) | −0.22 |
| Total alcohol (g/day) | 6.0 (9.2) | 6.6 (10.4) | 6.9 (10.6) | 4.4 (7.5) | 5.8 (9.8) | −0.06 |
| Beer (glasses/week) | 0.1 (0.5) | 0.1 (0.7) | 0.2 (0.7) | 0.2 (0.8) | 0.7 (2.9) | 0.20 |
| Potatoes (g/day) | 78 (51) | 97 (52) | 104 (56) | 110 (60) | 122 (71) | 0.24 |
| Milk and milk products (g/day) | 459 (219) | 347 (156) | 296 (148) | 239 (133) | 150 (104) | −0.57 |
| Cheese (g/day) | 23.4 (18.4) | 24.6 (20.1) | 22.7 (17.0) | 22.8 (17.8) | 18.9 (17.0) | −0.09 |
| Sweet complement on bread (g/day) | 7.1 (8.8) | 8.3 (10.9) | 8.3 (9.1) | 8.6 (10.0) | 6.5 (9.1) | −0.01 |
| Added sugar (g/day) | 2.5 (7.1) | 5.7 (12.3) | 9.9 (15.9) | 15.2 (19.3) | 23.6 (29.7) | 0.37 |
| Cake, biscuits (g/day) | 27.4 (18.7) | 30.2 (20.2) | 28.6 (18.8) | 28.3 (19.4) | 23.6 (17.6) | −0.05 |
| Candy (g/day) | 6.7 (8.7) | 5.7 (8.4) | 5.7 (7.5) | 5.9 (7.9) | 4.7 (6.8) | −0.06 |
| White bread (g/day) | 15.4 (22.7) | 18.0 (28.1) | 21.9 (34.4) | 24.9 (38.2) | 32.0 (49.0) | 0.16 |
| Meat (g/day) | 91 (40) | 91 (38) | 96 (39) | 93 (38) | 96 (43) | 0.05 |
| Processed meat (g/day) | 10.2 (12.5) | 10.4 (11.0) | 10.6 (10.7) | 12.1 (13.3) | 13.1 (14.5) | 0.07 |
| Calcium (mg/day) | 1,118 (324) | 990 (277) | 901 (264) | 825 (255) | 669 (224) | −0.50 |
| Percentage | p-value2 | |||||
| Family history of colorectal cancer | ||||||
| yes | 6.1 | 7.5 | 3.8 | 5.7 | 6.7 | 0.227 |
| Prevalence of diabetes mellitus | ||||||
| yes | 4.2 | 3.4 | 3.3 | 2.5 | 4.0 | 0.717 |
| Physical activity | ||||||
| less than 30 min/day | 22.2 | 19.3 | 20.5 | 26.7 | 28.2 | 0.008 |
| Cigarettes smoking | ||||||
| Currently | 18.3 | 13.8 | 21.0 | 23.2 | 32.2 | |
| Ex-smoking | 25.1 | 22.7 | 20.5 | 17.5 | 19.1 | |
| Never | 56.6 | 63.5 | 58.6 | 59.3 | 48.8 | 0.000 |
Age-adjusted and multivariate RRs and 95% CIs for colorectal, proximal and distal colon and rectum cancers according to quintiles of energy-adjusted glycemic load and glycemic index are presented in Table III, respectively, for men and Table IV, respectively, for women. The age-adjusted risk estimates are generally comparable to the multivariate adjusted estimates.
| Q11 | Q2 | Q3 | Q4 | Q5 | p for trend | |
|---|---|---|---|---|---|---|
| ||||||
| Quintiles of energy-adjusted glycemic load | ||||||
| Median glycemic load (g/day) | 108.7 | 124.8 | 136.2 | 147.8 | 165.4 | |
| Person years in subcohort | 4,188 | 4,245 | 4,147 | 4,255 | 4,203 | |
| Colon-rectum | ||||||
| No. of cases | 253 | 216 | 193 | 223 | 197 | |
| RR2 (95% CI) | 1.00 | 0.81 (0.65–1.03) | 0.74 (0.59–0.94) | 0.86 (0.69–1.09) | 0.77 (0.61–0.97) | 0.08 |
| RR3 (95% CI) | 1.00 | 0.82 (0.64–1.04) | 0.75 (0.58–0.97) | 0.90 (0.70–1.16) | 0.83 (0.64–1.08) | 0.37 |
| Colon | ||||||
| No. of cases | 140 | 153 | 132 | 148 | 101 | |
| RR2 (95% CI) | 1.00 | 1.04 (0.79–1.36) | 0.91 (0.69–1.21) | 1.03 (0.79–1.36) | 0.71 (0.53–0.95) | 0.05 |
| RR3 (95% CI) | 1.00 | 0.99 (0.74–1.31) | 0.87 (0.64–1.17) | 1.01 (0.75–1.37) | 0.72 (0.51–1.00) | 0.10 |
| Proximal colon | ||||||
| No. of cases | 54 | 78 | 63 | 72 | 46 | |
| RR2 (95% CI) | 1.00 | 1.36 (0.93–1.98) | 1.12 (0.76–1.66) | 1.30 (0.89–1.91) | 0.83 (0.55–1.27) | 0.39 |
| RR3 (95% CI) | 1.00 | 1.28 (0.87–1.90) | 1.04 (0.69–1.57) | 1.23 (0.81–1.86) | 0.79 (0.49–1.26) | 0.30 |
| Distal colon | ||||||
| No. of cases | 86 | 75 | 69 | 76 | 55 | |
| RR2 (95% CI) | 1.00 | 0.83 (0.59–1.17) | 0.78 (0.55–1.11) | 0.87 (0.62–1.22) | 0.63 (0.44–0.91) | 0.04 |
| RR3 (95% CI) | 1.00 | 0.80 (0.56–1.15) | 0.76 (0.52–1.11) | 0.88 (0.60–1.28) | 0.67 (0.44–1.02) | 0.15 |
| Rectum | ||||||
| No. of cases | 78 | 37 | 37 | 59 | 69 | |
| RR2 (95% CI) | 1.00 | 0.46 (0.30–0.70) | 0.47 (0.31–0.71) | 0.74 (0.51–1.07) | 0.88 (0.62–1.25) | 0.89 |
| RR3 (95% CI) | 1.00 | 0.49 (0.32–0.76) | 0.52 (0.33–0.81) | 0.83 (0.56–1.25) | 1.01 (0.68–1.51) | 0.37 |
| Quintiles of energy-adjusted glycemic index | ||||||
| Median glycemic index | 56.6 | 59.1 | 60.6 | 62.2 | 64.5 | |
| Person years in subcohort | 4,263 | 4,160 | 4,099 | 4,212 | 4,305 | |
| Colon-rectum | ||||||
| No. of cases | 228 | 214 | 220 | 220 | 200 | |
| RR2 (95% CI) | 1.00 | 0.97 (0.77–1.23) | 1.04 (0.82–1.31) | 1.02 (0.81–1.29) | 0.92 (0.73–1.17) | 0.68 |
| RR3 (95% CI) | 1.00 | 0.93 (0.73–1.19) | 1.00 (0.78–1.28) | 0.98 (0.75–1.29) | 0.81 (0.61–1.08) | 0.27 |
| Colon | ||||||
| No. of cases | 150 | 140 | 149 | 127 | 108 | |
| RR2 (95% CI) | 1.00 | 0.97 (0.74–1.27) | 1.08 (0.82–1.41) | 0.90 (0.68–1.19) | 0.77 (0.57–1.02) | 0.07 |
| RR3 (95% CI) | 1.00 | 0.90 (0.67–1.19) | 0.99 (0.74–1.32) | 0.81 (0.59–1.11) | 0.64 (0.46–0.89) | 0.01 |
| Proximal colon | ||||||
| No. of cases | 70 | 67 | 67 | 56 | 53 | |
| RR2 (95% CI) | 1.00 | 0.99 (0.69–1.43) | 1.05 (0.73–1.51) | 0.86 (0.59–1.26) | 0.82 (0.57–1.20) | 0.22 |
| RR3 (95% CI) | 1.00 | 0.94 (0.64–1.38) | 0.97 (0.66–1.42) | 0.78 (0.51–1.21) | 0.71 (0.45–1.12) | 0.10 |
| Distal colon | ||||||
| No. of cases | 80 | 73 | 82 | 71 | 55 | |
| RR2 (95% CI) | 1.00 | 0.94 (0.67–1.34) | 1.10 (0.78–1.55) | 0.93 (0.66–1.33) | 0.72 (0.50–1.04) | 0.13 |
| RR3 (95% CI) | 1.00 | 0.86 (0.60–1.25) | 1.01 (0.70–1.45) | 0.83 (0.56–1.24) | 0.58 (0.38–0.89) | 0.03 |
| Rectum | ||||||
| No. of cases | 46 | 48 | 57 | 67 | 62 | |
| RR2 (95% CI) | 1.00 | 1.08 (0.70–1.65) | 1.31 (0.87–1.99) | 1.51 (1.01–2.25) | 1.38 (0.92–2.08) | 0.03 |
| RR3 (95% CI) | 1.00 | 1.11 (0.71–1.73) | 1.38 (0.89–2.15) | 1.64 (1.02–2.65) | 1.38 (0.85–2.23) | 0.08 |
There was no significant association between the glycemic load and risk of colorectal, colon, proximal and distal colon, and rectum cancer in men (Table III). There was no indication of a clear linear trend for any of the endpoints. However, some of the individual RRs in the second and third quintiles of glycemic load reached statistical significance and were smaller than one, suggesting a U-shaped association for overall colorectal cancer and specifically rectum cancer in men.
The glycemic index was significantly associated with a reduced risk of colon cancer, and specifically distal colon cancer, in men, namely in the highest glycemic index quintile (RR = 0.64; 95% CI: 0.46–0.89 for colon cancer, and RR = 0.58, 95% CI: 0.38–0.89 for distal colon cancer) (Table III). There was an indication for a positive association between glycemic index and rectal cancer in men (p-trend = 0.08), but the highest rate ratio was observed in the fourth quintile (RR = 1.65, 95% CI: 1.02–2.65) instead of the fifth quintile (RR = 1.38, 95% CI: 0.85–2.23).
For women, there was also no association between the glycemic load and colorectal cancer endpoints (Table IV), except for a borderline significant positive trend for proximal colon cancer (p-trend = 0.06). However, the highest and only significant risk estimate was observed in the fourth quintile (RR = 1.63, 95% CI: 1.06–2.51) instead of the fifth quintile (RR = 1.48; 95% CI: 0.93–2.35).
| Q11 | Q2 | Q3 | Q4 | Q5 | p for trend | |
|---|---|---|---|---|---|---|
| ||||||
| Quintiles of energy-adjusted glycemic load | ||||||
| Median glycemic load (g/day) | 82.5 | 94.0 | 101.7 | 107.9 | 123.6 | |
| Person years in subcohort | 4,423 | 4,490 | 4,510 | 4,352 | 4,200 | |
| Colon-rectum | ||||||
| No. of cases | 152 | 149 | 156 | 156 | 142 | |
| RR2 (95% CI) | 1.00 | 0.93 (0.71–1.22) | 1.00 (0.77–1.30) | 0.99 (0.76–1.29) | 0.94 (0.72–1.23) | 0.83 |
| RR3 (95% CI) | 1.00 | 0.96 (0.73–1.28) | 1.02 (0.77–1.37) | 1.05 (0.78–1.41) | 1.00 (0.73–1.36) | 0.81 |
| Colon | ||||||
| No. of cases | 105 | 112 | 106 | 123 | 105 | |
| RR2 (95% CI) | 1.00 | 1.02 (0.75–1.38) | 0.98 (0.72–1.33) | 1.13 (0.84–1.52) | 1.01 (0.74–1.37) | 0.72 |
| RR3 (95% CI) | 1.00 | 1.07 (0.77–1.47) | 1.02 (0.73–1.43) | 1.25 (0.89–1.74) | 1.13 (0.79–1.60) | 0.32 |
| Proximal colon | ||||||
| No. of cases | 51 | 63 | 63 | 72 | 61 | |
| RR2 (95% CI) | 1 | 1.17 (0.78–1.73) | 1.20 (0.81–1.78) | 1.34 (0.91–1.97) | 1.19 (0.80–1.77) | 0.28 |
| RR3 (95% CI) | 1 | 1.30 (0.85–1.99) | 1.36 (0.88–2.10) | 1.63 (1.06–2.51) | 1.48 (0.93–2.35) | 0.06 |
| Distal colon | ||||||
| No. of cases | 54 | 49 | 43 | 51 | 44 | |
| RR2 (95% CI) | 1 | 0.88 (0.58–1.32) | 0.78 (0.51–1.19) | 0.93 (0.62–1.39) | 0.84 (0.55–1.27) | 0.52 |
| RR3 (95% CI) | 1 | 0.86 (0.55–1.32) | 0.73 (0.46–1.16) | 0.91 (0.58–1.44) | 0.82 (0.51–1.33) | 0.58 |
| Rectum | ||||||
| No. of cases | 34 | 22 | 33 | 22 | 27 | |
| RR2 (95% CI) | 1 | 0.62 (0.35–1.08) | 0.94 (0.57–1.56) | 0.62 (0.36–1.08) | 0.80 (0.47–1.35) | 0.44 |
| RR3 (95% CI) | 1 | 0.61 (0.35–1.07) | 0.93 (0.55–1.57) | 0.62 (0.34–1.15) | 0.79 (0.43–1.43) | 0.55 |
| Quintiles of energy-adjusted glycemic index | ||||||
| Median glycemic index | 53.7 | 56.2 | 57.8 | 59.6 | 61.9 | |
| Person years in subcohort | 4,364 | 4,472 | 4,439 | 4,450 | 4,250 | |
| Colon-rectum | ||||||
| No. of cases | 132 | 159 | 173 | 144 | 147 | |
| RR2 (95% CI) | 1.00 | 1.17 (0.89–1.54) | 1.29 (0.99–1.69) | 1.07 (0.81–1.40) | 1.15 (0.88–1.52) | 0.56 |
| RR3 (95% CI) | 1.00 | 1.18 (0.89–1.56) | 1.32 (0.98–1.76) | 1.08 (0.80–1.47) | 1.20 (0.85–1.67) | 0.53 |
| Colon | ||||||
| No. of cases | 94 | 116 | 123 | 110 | 108 | |
| RR2 (95% CI) | 1.00 | 1.20 (0.88–1.63) | 1.29 (0.95–1.75) | 1.14 (0.84–1.56) | 1.19 (0.87–1.63) | 0.41 |
| RR3 (95% CI) | 1.00 | 1.23 (0.90–1.70) | 1.37 (0.98–1.91) | 1.23 (0.86–1.74) | 1.34 (0.91–1.96) | 0.22 |
| Proximal colon | ||||||
| No. of cases | 57 | 56 | 63 | 69 | 65 | |
| RR2 (95% CI) | 1 | 0.96 (0.64–1.42) | 1.09 (0.74–1.60) | 1.18 (0.81–1.73) | 1.18 (0.84–1.74) | 0.21 |
| RR3 (95% CI) | 1 | 1.00 (0.66–1.52) | 1.19 (0.78–1.81) | 1.32 (0.86–2.02) | 1.40 (0.87–2.24) | 0.08 |
| Distal colon | ||||||
| No. of cases | 37 | 60 | 60 | 41 | 43 | |
| RR2 (95% CI) | 1 | 1.58 (1.02–2.43) | 1.60 (1.04–2.46) | 1.08 (0.68–1.73) | 1.20 (0.76–1.91) | 0.87 |
| RR3 (95% CI) | 1 | 1.53 (0.98–2.37) | 1.56 (0.98–2.47) | 1.03 (0.63–1.71) | 1.18 (0.69–1.99) | 0.80 |
| Rectum | ||||||
| No. of cases | 29 | 26 | 36 | 21 | 26 | |
| RR2 (95% CI) | 1 | 0.87 (0.51–1.51) | 1.22 (0.73–2.03) | 0.71 (0.40–1.27) | 0.93 (0.54–1.61) | 0.60 |
| RR3 (95% CI) | 1 | 0.90 (0.51–1.59) | 1.28 (0.73–2.24) | 0.73 (0.39–1.40) | 1.01 (0.52–1.98) | 0.81 |
For the glycemic index, there was no significant association with any of the colorectal cancer endpoints in women (Table IV), except for a borderline significant positive trend with proximal colon cancer (p-trend = 0.08), as was also observed for glycemic load.
None of the interactions of glycemic load or glycemic index with fibers from cereals, fibers from fruit and vegetables, family history of colorectal cancer, diabetes, physical activity and body mass index reached statistical significance (p > 0.10). Additional analyses among overweight (body mass index ≥25 kg/m2) and obese (body mass index ≥30 kg/m2) individuals did not show an association between the glycemic load or glycemic index and colorectal cancer endpoints in men or women. The prevalence of diabetes was low (3.3% for men and 3.4% for women) and exclusion of diabetics from analyses did not alter results.
Analyses excluding the first 2 years of follow-up resulted in the attenuation of the positive associations between the glycemic index and rectum cancer in men (p-trend = 0.339) and between the glycemic load and glycemic index and proximal colon cancer in women (p-trend = 0.165 and 0.254, respectively). Inverse associations remained after excluding the first 2 years of follow-up, i.e., for glycemic load and rectum cancer in the second and third quintile of glycemic load in men (results not shown), and for glycemic index and distal colon cancer in men (p-trend = 0.023).
In additional analyses, total carbohydrate intake and mono- and di-saccharide intake were not associated with the risk of any of the colorectal cancer endpoints in men or women.
Discussion
In this large prospective cohort study of men and women, there was no statistically significant positive dose-response association between glycemic load and glycemic index and the risk of colorectal, colon, proximal and distal colon, and rectum cancer.
In Table V, our intake values for glycemic index and glycemic load can be compared with the values found in case–control studies10, 12, 15 and cohort studies.11, 13, 14, 16-19 The overall range of glycemic index for all the studies was 53–90 units and for glycemic load 80–285 g/day. The choice of reference food could have influenced these values. Glucose-based values are approximately a factor 0.7 lower than white bread-based values, therefore our glucose-based values are comparable to those of Flood et al.19 and generally lower than those from other studies indicating having used white bread based values.10, 12, 14, 17 Although the values may differ, this should not affect observed associations, since no threshold effect is to be expected. The relatively small range of glycemic load and glycemic index in our study could have resulted in smaller estimates of the RRs.
| Reference | Follow-up (years) | Sex | Controls/population | Cases | Reference table and reference food | Exposure | Method1 | Cancer subsite | Glycemic index (unit) | Glycemic load (g/day) | RR/OR (95% CI)1 | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Men | Women | Men | Women | ||||||||||
| |||||||||||||
| Case–control | |||||||||||||
| Slattery et al.15 | men | 1,290 | 1,120 | 1,099 | 894 | Wolever | GI1 | Diet his1 | proximal colon | 1.59 (1.10–2.29) | |||
| men | GI | distal colon | 1.23 (0.88–1.74) | ||||||||||
| women | GI | proximal colon | 1.39 (0.96–2.01) | ||||||||||
| women | GI | distal colon | 1.22 (0.84–1.78) | ||||||||||
| Franceschi et al.10 | both | 2,073 | 2,081 | 1,125 | 828 | Foster–Powell–white bread | GI | FFQ1 | colon | 71–802 | 1.9 (1.5–2.4) | ||
| both | GI | rectum | 1.4 (1.1–1.9) | ||||||||||
| both | GI | colorectum | 1.7 (1.4–2.0) | ||||||||||
| both | GL | colon | 151–2852 | 1.9 (1.5–2.4) | |||||||||
| both | GL | rectum | 1.5 (1.1–1.9) | ||||||||||
| both | GL | colorectum | 1.8 (1.5–2.2) | ||||||||||
| Levi et al.12 | both | 330 | 815 | 192 | 131 | Foster–Powell–white bread | GI | FFQ | colon | 74–862 | 2.35 (1.35–4.07) | ||
| both | GI | rectum | 1.31 (0.76–2.25) | ||||||||||
| both | GI | colorectum | 1.82 (1.19–2.78) | ||||||||||
| Cohort studies | |||||||||||||
| Terry et al.16 | 16.5 | women | 49,124 | 616 | Foster–Powell | GL | FFQ | colorectum | 82–2173 | 1.05 (0.73–1.53) | |||
| women | GL | colon | 0.95 (0.61–1.50) | ||||||||||
| women | GL | proximal colon | 0.70 (0.37–1.32) | ||||||||||
| women | GL | distal colon | 2.05 (0.97–4.29) | ||||||||||
| women | GL | rectum | 1.34 (0.70–2.58) | ||||||||||
| Higginbotham et al.11 | 7.9 | women | 38,451 | 174 | Foster–Powell | GI | FFQ | colorectum | 49–573 | 1.71 (0.98–2.98) | |||
| women | GL | colorectum | 92–1433 | 2.85 (1.40–5.80) | |||||||||
| Oh et al.14 | 18.0 | women | 34,428 | 1,715 | Foster–Powell–white bread | GI | FFQ | tot adenoma | 69–803 | 1.11 (0.94–1.32) | |||
| women | GI | colon adenoma | 1.20 (0.98–1.46) | ||||||||||
| women | GI | rectum adenoma | 0.90 (0.66–1.22) | ||||||||||
| women | GL | adenoma | 124–1673 | 0.92 (0.76–1.11) | |||||||||
| women | GL | colon adenoma | 0.95 (0.76–1.19) | ||||||||||
| women | GL | rectum adenoma | 0.85 (0.60–1.19) | ||||||||||
| McCarl et al.13 | 15 | women | 35,197 | 954 | Foster–Powell | GI | FFQ | colorectum | <81.0–>89.34 | 1.08 (0.88–1.32) | |||
| women | GI | colon – BMI <25 | 1.03 (0.70–1.51) | ||||||||||
| women | GI | colon – BMI >30 | 1.60 (1.02–2.51) | ||||||||||
| women | GI | rectum – BMI <25 | 0.52 (0.22–1.23) | ||||||||||
| women | GI | rectum – BMI >30 | 3.34 (1.09–10.2) | ||||||||||
| women | GL | colorectum | <146–>1934 | 1.09 (0.88–1.35) | |||||||||
| women | GL | colon – BMI <25 | 0.74 (0.47–1.14) | ||||||||||
| women | GL | colon – BMI >30 | 1.68 (1.06–2.67) | ||||||||||
| women | GL | rectum – BMI <25 | 0.88 (0.41–1.86) | ||||||||||
| women | GL | rectum – BMI >30 | 2.23 (0.91–5.45) | ||||||||||
| Michaud et al.18 | 20 | men | 47,422 | 83,927 | 696 | 1,113 | Foster–Powell | GI | FFQ | colorectum | 69–823 | 1.14 (0.88–1.48) | |
| men | GI | colon | 1.13 (0.84–1.51) | ||||||||||
| men | GI | proximal colon | 0.99 (0.63–1.57) | ||||||||||
| men | GI | distal colon | 1.06 (0.67–1.68) | ||||||||||
| men | GI | rectum | 1.21 (0.68–2.15) | ||||||||||
| men | GL | colorectum | 131–2233 | 1.32 (0.98–1.79) | |||||||||
| men | GL | colon | 1.25 (0.88–1.25) | ||||||||||
| men | GL | proximal colon | 1.10 (0.64–1.88) | ||||||||||
| men | GL | distal colon | 0.87 (0.51–1.49) | ||||||||||
| men | GL | rectum | 1.61 (0.82–3.17) | ||||||||||
| women | GI | FFQ | colorectum | 65–813 | 1.08 (0.87–1.34) | ||||||||
| women | GI | colon | 1.06 (0.83–1.36) | ||||||||||
| women | GI | proximal colon | 1.11 (0.77–1.58) | ||||||||||
| women | GI | distal colon | 0.91 (0.63–1.34) | ||||||||||
| women | GI | rectum | 1.14 (0.73–1.78) | ||||||||||
| women | GL | colorectum | 80–1673 | 0.89 (0.71–1.11) | |||||||||
| women | GL | colon | 1.89 (0.69–1.15) | ||||||||||
| women | GL | proximal colon | 0.77 (0.53–1.11) | ||||||||||
| women | GL | distal colon | 0.90 (0.60–1.36) | ||||||||||
| women | GL | rectum | 0.87 (0.52–1.44) | ||||||||||
| Larsson et al.17 | women | 61,433 | 870 | Foster–Powell–white bread | GI | FFQ | colorectum | <76–>834 | 1.00 (0.75–1.33) | ||||
| women | GI | colon | 0.84 (0.60–1.18) | ||||||||||
| women | GI | proximal colon | 0.97 (0.58–1.63) | ||||||||||
| women | GI | distal colon | 0.81 (0.46–1.40) | ||||||||||
| women | GI | rectum | 1.32 (0.80–2.17) | ||||||||||
| women | GL | colorectum | <164–>2004 | 1.06 (0.81–1.39) | |||||||||
| women | GL | colon | 0.97 (0.70–1.32) | ||||||||||
| women | GL | proximal colon | 1.00 (0.62–1.62) | ||||||||||
| women | GL | distal colon | 1.18 (0.69–2.00) | ||||||||||
| women | GL | rectum | 1.20 (0.74–1.95) | ||||||||||
| Flood et al.19 | men | Foster–Powell–glucose | GI | FFQ | all adenoma's | 53–583 | 1.05 (0.90–1.23) | ||||||
| men | GI | advanced adenoma's | 1.03 (0.81–1.31) | ||||||||||
| men | GL | all adenoma's | 106–1833 | 0.79 (0.68–0.93) | |||||||||
| men | GL | advanced adenoma's | 0.79 (0.62–1.00) | ||||||||||
| women | GI | all adenoma's | 53–573 | 1.00 (0.81–1.23) | |||||||||
| women | GI | advanced adenoma's | 1.02 (0.73–1.42) | ||||||||||
| women | GL | all adenoma's | 90–1433 | 0.98 (0.81–1.19) | |||||||||
| women | GL | advanced adenoma's | 0.94 (0.69–1.29) | ||||||||||
| Mullie et al. (present study) | 11.3 | men | 58,279 | 62,573 | 1,165 | 854 | Foster–Powell–glucose | GI | FFQ | colorectum | 58–634 | 0.81 (0.61–1.08) | |
| men | GI | colon | 0.64 (0.46–0.89) | ||||||||||
| men | GI | proximal colon | 0.71 (0.45–1.12) | ||||||||||
| men | GI | distal colon | 0.58 (0.38–0.89) | ||||||||||
| men | GI | rectum | 1.38 (0.85–2.23) | ||||||||||
| men | GL | colorectum | 119–1563 | 0.83 (0.64–1.08) | |||||||||
| men | GL | colon | 0.72 (0.51–1.00) | ||||||||||
| men | GL | proximal colon | 0.79 (0.49–1.26) | ||||||||||
| men | GL | distal colon | 0.67 (0.44–1.02) | ||||||||||
| men | GL | rectum | 1.01 (0.68–1.51) | ||||||||||
| women | GI | colorectum | 55–614 | 1.20 (0.85–1.67) | |||||||||
| women | GI | colon | 1.34 (0.91–1.96) | ||||||||||
| women | GI | proximal colon | 1.40 (0.87–2.24) | ||||||||||
| women | GI | distal colon | 1.18 (0.69–1.99) | ||||||||||
| women | GI | rectum | 1.01 (0.52–1.98) | ||||||||||
| women | GL | colorectum | 90–1163 | 1.00 (0.73–1.36) | |||||||||
| women | GL | colon | 1.13 (0.79–1.60) | ||||||||||
| women | GL | proximal colon | 1.48 (0.93–2.35) | ||||||||||
| women | GL | distal colon | 0.82 (0.51–1.33) | ||||||||||
| women | GL | rectum | 0.79 (0.43–1.43) | ||||||||||
Three case–control studies have suggested an increased risk between dietary glycemic index or glycemic load and colorectal cancer risk after adjustment for age, body mass index, physical activity and other known risk factors10, 12, 15 (Table V). The study by Higginbotham et al.11 was the only prospective study of a total of 511, 13, 17, 18, 19 reporting a positive association between dietary glycemic load and risk of colorectal cancer. They observed a statistically significant positive association of glycemic load and colorectal cancer in women. However, Higginbotham's study had a relatively short follow-up period of 7.9 years compared to the other prospective studies (Table V). Considering our observation that exclusion of the first 2 years of follow-up attenuated any positive associations of glycemic load or glycemic index with endpoints, underlying preclinical disease may have affected dietary intake levels. This may explain observed associations of glycemic load and glycemic index with increased colorectal cancer risk in Higginbotham's study, in our study as well as in almost all case–control studies.
In the Iowa Women's Health Study, McCarl et al.13 found a positive association between the highest quintile of glycemic index and glycemic load and colorectal cancer in obese women only. In our study, we did not observe significantly elevated risks for colorectal cancer associated with a high glycemic load or glycemic index diet in overweight or obese participants.
Although the prevalence of diabetes was associated with a lower glycemic load, probably reflecting adapted dietary habits as a result of the presence of diabetes, the effect of diabetes on our results is negligible. First in our cohort, the prevalence of diabetes at baseline was very low, second, adjusting for it or, third, excluding diabetics from the analyses did not alter the results.
In our cohort, the highest quintile of glycemic load was associated with a healthier lifestyle. The subgroup of men and women who never smoked was higher, and the consumption of meat, processed meat and saturated fatty acids was lower. A health-conscious diet and lifestyle would tend to negatively confound the associations. In our analyses we adjusted for life style factors, but still some residual confounding cannot be ruled out. This may explain the inverse association between glycemic index and distal colon cancer in men. However, a role for chance in observed significant RRs cannot entirely be ruled out.
The major strength of this study is the prospective design and the large number of male and female cases, i.e., 1,837 cases of colorectal cancer after 11.3 years of follow-up. The exposure was assessed with a validated food frequency questionnaire and dietary reporting was not influenced by knowledge of the existence of colorectal cancer because of the prospective design. Many known risk factors for colorectal cancer like age, body mass index, smoking, total energy intake, alcohol intake, family history of colon cancer, education level and physical activity were accounted for.
A possible limitation of the study is that dietary intake information was ascertained only once, and changes in diet during follow-up may have weakened the true relationship between glycemic load, glycemic index and colorectal cancer risk. The food frequency questionnaire was validated and tested for reproducibility.23, 24 Pearson correlation coefficients between a 9-day diet record and the questionnaire varied from 0.40 for vitamin B1 to 0.86 for alcohol intake (median 0.69). Although not available for the glycemic load and glycemic index, the energy and sex adjusted correlation coefficients were 0.71 for total carbohydrate and 0.79 for mono- and disaccharide. On average, the decline in correlation amounted to 0.07 after 5 years, indicating that the potential of a single food frequency questionnaire measurement to rank subjects according to nutrient intake dropped only slightly over time. The glycemic load showed a substantial correlation with total carbohydrates and mono- and di-saccharides (respectively 0.54 and 0.50 for men and 0.51 and 0.49 for women), but the glycemic index did not (respectively 0.00 and −0.12 for men and 0.03 and −0.14 for women). This is to be expected since the index is a measure of the quality of carbohydrate of the diet, while the glycemic load also assesses the quantity of carbohydrates. It could be argued that the glycemic load is then a surrogate measure of total carbohydrate intake, however no associations were observed between total carbohydrate intake or mono- and di-saccharide intake and colorectal cancer endpoints. Specifically, the U-shaped association of glycemic load with rectal cancer in men and the positive association with proximal colon cancer in women were not observed for total carbohydrate intake or mono- and di-saccharide intake.
Since its introduction in 1981 by Jenkins et al.,9 there is still a lot of controversy on the use of the glycemic index and the glycemic load in different areas of research,33 including its application in the management of diabetes, cancer, obesity and cardiovascular disease. Indeed, the intra-individual variability in the measurement of the glycemic index, used to calculate the glycemic load, is very high, and variability in a single food values often exceeds the range seen in prospective cohort studies.34, 35, 36 In addition, dietary glycemic load is only one element of the total diet. People do not eat single macronutrients; they eat meals or snacks. Dietary glycemic index and glycemic load as surrogate for insulin secretion, are indicators for the quality and quantity of the carbohydrates. Protein-rich foods are known to increase insulin secretion; therefore, as more protein is consumed in conjunction with carbohydrate, the insulin response will increase.36 Adding fat to carbohydrate also enhances insulin secretion. Thus, the insulin response to a carbohydrate food varies not only with the chemical property and amount of the carbohydrate, but also with the amount and possible insulinotropic effect of protein and/or fat.36
There is scientific evidence that hyperinsulinemia is a risk factor for colorectal cancer.37 The overall lack of association between glycemic load, glycemic index and colorectal cancer in our study and other prospective cohort studies raises the question whether the glycemic load and the glycemic index are adequate indicators of chronic hyperinsulinemia. In addition, we still do not know whether the total volume of glucose exposure (from a high glycemic load diet throughout the day) or the repeated high peaks of glucose (from a high glycemic index diet) are important in determining chronic effects on blood insulin levels.
In conclusion, a diet with a high glycemic load or glycemic index was not associated with a higher risk of colorectal cancer in men or women in the Netherlands Cohort Study on Diet and Cancer.
Acknowledgements
Dr. M.P.W. and Mr. P.M. carried out the statistical analysis, interpreted the data, and drafted the manuscript. Dr. M.P.W. and Dr. R.A.G. conceived the study, and participated in its design and coordination, and critically reviewed the manuscript. Prof. P.A.vd.B. conceived the study, and participated in its design and coordination and critically reviewed the manuscript. Mrs. H.A.M.B. developed the database of glycemic index and critically reviewed the manuscript. Mrs. M.M.H. critically reviewed the manuscript. We are indebted to the participants of this study and further wish to thank the cancer registries (I.K.A., I.K.L., I.K.M.N., I.K.N., I.K.O., I.K.R., I.K.S.T., I.K.W., I.K.Z. and V.I.K.C.), and the Netherlands nationwide registry of pathology (PALGA). We also thank Dr. A. Volovics and Dr. A. Kester for statistical advice; Dr. L. Schouten, Mrs. S. van de Crommert, Mrs. J. Nelissen, Mrs. C. de Zwart, Mrs. M. Moll, Mrs. W. van Dijk, Mrs. M. Jansen and Mrs. A. Pisters for assistance; and Mr. H. van Montfort, Mr. T. van Moergastel, Mrs. L. van den Bosch, Mr. R. Schmeitz for programming assistance.
References
- 1,,,,,. Estimates of the cancer incidence and mortality in Europe in 2006. Ann Oncol 2007; 18: 581–92.
- 2,. The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. J Natl Cancer Inst 1981; 66: 1191–308.
- 3. Insulin and colon cancer. Cancer Causes Control 1995; 6: 164–79.
- 4,,,,,,,,,. Genetic alterations during colorectal-tumor development. N Engl J Med 1988; 319: 525–32.
- 5. Epidemiology of colorectal cancer revisited: are serum triglycerides and/or plasma glucose associated with risk? Cancer Epidemiol Biomarkers Prev 1994; 3: 687–95.
- 6,. Energy balance and cancer: the role of insulin and insulin-like growth factor-I. Proc Nutr Soc 2001; 60: 91–106.
- 7,,. Mechanisms linking diet and colorectal cancer: the possible role of insulin resistance. Nutr Cancer 2000; 37: 19–26.
- 8,,,,,,,. Glycemic index: overview of implications in health and disease. Am J Clin Nutr 2002; 76: 266S–73S.
- 9,,,,,,,,,. Glycemic index of foods: a physiological basis for carbohydrate exchange. Am J Clin Nutr 1981; 34: 362–6.
- 10,,,,,,,. Dietary glycemic load and colorectal cancer risk. Ann Oncol 2001; 12: 173–8.
- 11,,,,,,. Dietary glycemic load and risk of colorectal cancer in the Women's Health Study. J Natl Cancer Inst 2004; 96: 229–33.
- 12,,,,. Glycaemic index, breast and colorectal cancer. Ann Oncol 2002; 13: 1688–9.
- 13,,,,. Incidence of colorectal cancer in relation to glycemic index and load in a cohort of women. Cancer Epidemiol Biomarkers Prev 2006; 15: 892–6.
- 14,,,. Glycemic index, glycemic load, and carbohydrate intake in relation to risk of distal colorectal adenoma in women. Cancer Epidemiol Biomarkers Prev 2004; 13: 1192–8.
- 15,,,,,,. Dietary sugar and colon cancer. Cancer Epidemiol Biomarkers Prev 1997; 6: 677–85.
- 16,,,,. Glycemic load, carbohydrate intake, and risk of colorectal cancer in women: a prospective cohort study. J Natl Cancer Inst 2003; 95: 914–16.
- 17,,. Dietary carbohydrate, glycemic index, and glycemic load in relation to risk of colorectal cancer in women. Am J Epidemiol 2007; 165: 256–61.
- 18,,,,,. Dietary glycemic load, carbohydrate, sugar, and colorectal cancer risk in men and women. Cancer Epidemiol Biomarkers Prev 2005; 14: 138–47.
- 19,,,,,,,,,. Carbohydrate, glycemic index, and glycemic load and colorectal adenomas in the prostate, lung, colorectal, and ovarian screening study. Am J Clin Nutr 2006; 84: 1184–92.
- 20,,,,,. A large-scale prospective cohort study on diet and cancer in The Netherlands. J Clin Epidemiol 1990; 43: 285–95.
- 21,,,,. Development of a record linkage protocol for use in the Dutch Cancer Registry for Epidemiological Research. Int J Epidemiol 1990; 19: 553–8.
- 22,,,,. The capture-recapture method for estimation of cancer registry completeness: a useful tool? Int J Epidemiol 1994; 23: 1111–16.
- 23,,,,,,. Validation of a dietary questionnaire used in a large-scale prospective cohort study on diet and cancer. Eur J Clin Nutr 1994; 48: 253–65.
- 24,,,,,,. Reproducibility of a food frequency questionnaire and stability of dietary habits determined from five annually repeated measurements. Eur J Clin Nutr 1995; 49: 420–9.
- 25Nevo table: Dutch food composition table 1986–1987, edn. The Hague: Voorlichtingsbureau voor de Voeding, 1986.
- 26,,. International table of glycemic index and glycemic load values: 2002. Am J Clin Nutr 2002; 76: 5–56.
- 27,. The use of the glycemic index in predicting the blood glucose response to mixed meals. Am J Clin Nutr 1986; 43: 167–72.
- 28,. Application of glycaemic index to mixed meals. Lancet 1985; 2: 944.
- 29,,,,,,. Prediction of the relative blood glucose response of mixed meals using the white bread glycemic index. Diabetes Care 1985; 8: 418–28.
- 30,,,,,,,. A prospective study of dietary glycemic load, carbohydrate intake, and risk of coronary heart disease in US women. Am J Clin Nutr 2000; 71: 1455–61.
- 31,. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986; 124: 17–27.
- 32. Causes, relevant mechanisms, and prevention of large bowel cancer. Semin Oncol 1991; 18: 316–36.
- 33,. Glycemic index in the treatment of diabetes: the debate continues. J Am Coll Nutr 2004; 23: 1–4.
- 34,. Much ado about (almost) nothing. Diabetes Care 1997; 20: 241–3.
- 35,. The clinical utility of the glycemic index and its application to mixed meals. Can J Physiol Pharmacol 1991; 69: 100–7.
- 36. Glycemic index and disease. Am J Clin Nutr 2002; 76: 290S-8S.
- 37. Insulin, insulin-like growth factors and colon cancer: a review of the evidence. J Nutr 2001; 131: 3109S–20S.