| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
1 Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis and 2 Division of Gastroenterology and Hepatology, Mayo Clinic College of Medicine, Rochester, Minnesota
Requests for reprints: Aaron R. Folsom, Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Suite 300, 1300 South Second Street, Minneapolis, MN 55454-1015. Phone: 612-626-8862; Fax: 612-624-0315. E-mail: folsom{at}epi.umn.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: We examined associations of GI and GL with colorectal cancer (CRC) among participants in the Iowa Women's Health Study (n = 35,197; ages 55-69 years at baseline in 1986). Over 15 years of follow-up, we identified 757 cases of colon cancer and 209 cases of rectal cancer (954 CRC cases).
Results: Overall, neither GI nor GL were significantly associated with incident CRC. However, among obese women (baseline body mass index 
30 kg/m2) CRC incidence was increased in the highest versus lowest quintiles of GI (relative risk, 1.66; 95% confidence intervals, 1.13-2.43; P for trend = 0.02) and GL (relative risk, 1.79; 95% confidence intervals, 1.19-2.70; P for trend < 0.01). This pattern of increased risk for obese women with high GI or GL tended to hold for both colon cancer and rectal cancer, and for nondiabetic women as well. No statistically significant associations were observed between GI or GL and CRC among subjects whose baseline body mass index was <30 kg/m2.
Conclusion: Our findings suggest that high GI or GL are not major CRC risk factors among older women in general, but may increase CRC risk among women who are obese. (Cancer Epidemiol Biomarkers Prev 2006;15(5):392–6)
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Two indices of postprandial blood glucose response that are of epidemiologic interest are glycemic index (GI) and glycemic load (GL). GI is a ranking of carbohydrate-containing foods based on their postprandial glycemic effect (10). Foods with a high GI contain carbohydrates that are absorbed by the body quickly, which leads to a more rapid increase in blood glucose levels relative to foods with a lower GI. This is influenced by many things including the type of carbohydrates and other nutrients in the food and how the food has been processed. To calculate the GI of a specific food, the glycemic response is compared with the response of a control food (white bread or glucose) which is assigned a value of 100 (7). Whereas GI compares equal quantities of carbohydrate, GL is a measurement which also accounts for the quantity of carbohydrate consumed (7). Thus, GL quantifies the overall glycemic effect of a portion of food.
Because GI and GL have a direct effect on circulating insulin levels (7), we tested the hypothesis that they are positively associated with CRC incidence in a prospective cohort study of women. We also tested whether GL and GI interact with obesity, as previously suggested (4, 9).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Participants reported their current height and weight (12), which was used to calculate body mass index (BMI, kg/m2). A friend measured waist (1 inch above the umbilicus) and hip (widest point) circumferences. These were used to calculate the waist/hip ratio. Three levels of physical activity (low, medium, and high) were created by asking three questions about the frequencies of moderate and vigorous activities. Participants were identified as having prevalent diabetes if they reported ever being told by a doctor that they had sugar diabetes (diabetes mellitus) or if they indicated current use of insulin or "pills for sugar diabetes (or to lower blood sugar)."
A Harvard food frequency questionnaire (13) was included to determine each subject's usual dietary intake over the past year. Participants were asked to report their average consumption, over the past year, of 127 food items, including 29 vegetables; 15 fruits; 13 dairy foods; 14 meat, poultry, seafood, or egg items; 17 breads, cereals, or starches (including 8 whole-grain items); 14 beverages (including 4 alcoholic beverages); and 25 sweets, baked goods, and miscellaneous items. The food frequency questionnaire used in the present study was found to account for 93% of the total energy intake in a validation study with 194 female nurses (14). The average dietary GI was calculated in the following manner (15):
 |
The average GL was calculated as follows:
 |
In this sample from Iowa, the energy-adjusted reliability coefficients reported for total carbohydrate and crude fiber were 0.66 and 0.82, respectively, based on two administrations of the food frequency (16). The correlations of the carbohydrate and crude fiber values from the food frequency with five 24-hour dietary recalls were 0.45 and 0.24, respectively (16).
Cohort Follow-up
Cancer incidence and deaths in Iowa from 1986 to 2000 were ascertained by computer linkage with the State Health Registry of Iowa, which includes a Surveillance, Epidemiology, and End Results cancer registry. Colon and rectal cancers were identified using International Classification of Diseases for Oncology (second edition) codes 18.0 to 18.9 and 19.9 and 20.9, respectively. Address changes, deaths outside of Iowa, and nonfatal, noncancer end points were identified by follow-up questionnaires mailed to the participants in 1987, 1989, 1992, and 1997, and by the National Death Index.
Data Analysis
Our hypothesis was that baseline dietary GI and GL would be positively associated with incident CRC. We excluded women who had a prevalent cancer (n = 3,830); women who indicated implausible energy intake levels (<600 or >5,000 calories per day) or who left 30 missing items in the food frequency questionnaire (n = 3,102), and women who had a CRC of nontypical morphology (n = 19). These exclusions left 35,197 women for analysis.
Age- and energy-adjusted prevalences of risk factors were compared by quintiles of GI and GL using analysis of covariance. Age-adjusted incidence rates and relative risks (RR) of colorectal, colon, and rectal cancers in relation to baseline characteristics, along with their 95% confidence intervals (CI), were calculated using Poisson regression. Multivariately adjusted RRs and their 95% CI as well as P values for trend (using an ordinal variable to designate the quintiles) were computed using proportional hazards regression. Potential confounders were identified by first examining each characteristic's relation to the exposure (GI/GL) and outcome (colorectal, colon, or rectal cancer). Those that seemed to be related to both the exposure and the outcome were added to the proportional hazards model. They were retained in the model as confounders if they changed the GI/GL regression coefficient by >10%. Multiplicative interactions were tested using cross-product terms.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Over 15 years of follow-up, we identified 757 incident cases of colon cancer and 209 cases of rectal cancer (954 CRC cases). As shown in Web Table S2, age-adjusted incidence of CRC from 1986 to 2000 was associated positively with baseline BMI, waist/hip ratio, history of diabetes mellitus, pack-years of smoking, and height, and was associated inversely with estrogen use, physical activity level, fruit, dairy, whole-grain consumption, energy intake, fiber (insoluble and soluble), carbohydrates, calcium, folate, multivitamin use, vitamin D, vitamin A, and vitamin E. No association was seen with alcohol use, vegetable intake, starch, caffeine, ß-carotene, fructose, glucose, or sucrose intake.
Table 1 shows the adjusted RRs of CRC in relation to quintiles of GI and GL. When adjusted for age and energy, we found no statistically significant associations between either GI or GL and incident CRC. Adjustment for other risk factors did not appreciably alter the RR estimates. Adding other dietary variables (dairy, fruit, whole-grain, carbohydrates, soluble fiber, calcium, vitamin E, and vitamin D) to the risk model produced nearly identical results (data not shown). Separate analyses based on colon and rectal cancer subsites were similarly unremarkable. When we restricted these analyses to the 33,071 women free of diabetes at baseline, among whom 877 developed CRC, we still observed no association with GI or GL.
|
30 kg/m2) showed that GI and GL were positively associated with CRC in the highest BMI category (P for interaction = 0.04 for GI and > 0.05 for GL). Colon and rectal cancers were then analyzed individually (Table 2
). GL, but not GI, was positively associated with colon cancer in the highest BMI category, whereas GL and GI were both positively associated with rectal cancer in the highest BMI category. All of these associations in the high BMI stratum were somewhat stronger among nondiabetic women (Tables 1 and 2).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Three prior case-control studies (3-5) and two cohort studies (7, 9) have reported a positive association between GI/GL and CRC, but two cohort studies (6, 8) did not find an association. Similar to our finding, one of these previous studies found that elevated BMI (in men) or waist/hip ratio (in women) seemed to enhance CRC risk when the diet was high in GL (4). Another study found that the association between GL and CRC was slightly stronger among men with elevated BMI, but no such interaction existed among the women studied (9). Two previous studies have found that the associations between GL and CRC did not vary appreciably across strata of BMI (6, 8). Another study on colon cancer found a three-way interaction of colon cancer with physical activity, BMI and sucrose/dietary fiber ratio (3). Thus, several studies support a stronger association between GI/GL and CRC in obese subjects; however, this interaction is not universally seen, perhaps due to limits of statistical power, differences in study populations, or chance.
Obesity increases insulin resistance (17), which leads to high blood insulin levels. When compared with low GI foods, the consumption of foods with a high GI results in a higher increase in blood glucose levels leading to a higher increase in blood insulin levels (10). This combination could produce a condition of extremely high blood insulin levels; increased insulin has been shown to increase insulin-like growth factor-I, both of which could stimulate the proliferation of colorectal cells and inhibit apoptosis (18). The proliferation of colorectal cells in an environment with reduced apoptosis could potentially result in an accumulation of mutations leading to unregulated growth and cancer.
The main weakness of our analysis was that it was based on a single food frequency questionnaire. A single measure of diet is imprecise and a typically random error would lead to an underestimation of the true association. Although we examined a number of characteristics for potential confounding, additional confounding factors may have been missed (e.g., family history of CRC). Due to the small number of rectal cancer, separate analyses of rectal cancer and GI/GL had low power. Our study population comprised women who were 55 to 69 years old at baseline and were primarily Caucasian (99%); our findings may not hold true for men or other ethnic groups.
In summary, our analyses indicate that there is no overall association between diets with a high average GI/GL and CRC risk among nonobese, older women. However, women who are obese may experience increased CRC risks by consuming a high glycemic diet. Our data support recently revised diet and exercise guidelines (19), which recommend maintaining body weight within a healthy range, and consumption of whole-grains rather than refined grains as the preferred source of carbohydrates.
|
Footnotes |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Epidemiology Biomarkers and Prevention Online (http://cebp.aacrjournals.org/).
Received 9/ 6/05; revised 2/23/06; accepted 3/10/06.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  P. Gnagnarella, S. Gandini, C. La Vecchia, and P. Maisonneuve Glycemic index, glycemic load, and cancer risk: a meta-analysis Am. J. Clinical Nutrition, June 1, 2008; 87(6): 1793 - 1801. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Marsh and J. Brand-Miller State of the Art Reviews: Glycemic Index, Obesity, and Chronic Disease American Journal of Lifestyle Medicine, April 1, 2008; 2(2): 142 - 150. [Abstract] [PDF]
|
|
|
|
|
|
A. W Barclay, P. Petocz, J. McMillan-Price, V. M Flood, T. Prvan, P. Mitchell, and J. C Brand-Miller Glycemic index, glycemic load, and chronic disease risk--a meta-analysis of observational studies Am. J. Clinical Nutrition, March 1, 2008; 87(3): 627 - 637. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. C. Larsson, E. Giovannucci, and A. Wolk Dietary Carbohydrate, Glycemic Index, and Glycemic Load in Relation to Risk of Colorectal Cancer in Women Am. J. Epidemiol., February 1, 2007; 165(3): 256 - 261. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
