AJCN Tufts Nutrition Symposium, Boston Sept 24-26
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Flood, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Flood, A.
Agricola
Right arrow Articles by Flood, A.
American Journal of Clinical Nutrition, Vol. 84, No. 5, 1184-1192, November 2006
© 2006 American Society for Nutrition

ORIGINAL RESEARCH COMMUNICATION

Carbohydrate, glycemic index, and glycemic load and colorectal adenomas in the Prostate, Lung, Colorectal, and Ovarian Screening Study1,2,3

Andrew Flood, Ulrike Peters, David JA Jenkins, Nilanjan Chatterjee, Amy F Subar, Timothy R Church, Robert Bresalier, Joel L Weissfeld, Richard B Hayes, Arthur Schatzkin for the Prostate, Lung, Colorectal, Ovarian (PLCO) Project Team

1 From the University of Minnesota, Minneapolis, MN (AF and TRC); the Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD (AF, NC, AS, and RBH); the Fred Hutchinson Cancer Research Center, Seattle, WA (UP); the University of Washington, Seattle, WA (UP); the University of Toronto, Toronto, Canada (DJAJ); the Division of Cancer Control and Population Sciences (AFS), Henry Ford Hospital, Detroit, MI (RB); and the University of Pittsburgh, Pittsburgh, PA (JLW)

2 Supported by grant K07-CA108910-01A1 (to AF) from the National Cancer Institute

3 Reprints not available. Address correspondence to A Flood, Division of Epidemiology, University of Minnesota, 1300 South Second Street, Suite 300, Minneapolis, MN 55454. E-mail: flood{at}epi.umn.edu.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Background: It is possible that high-glycemic-load diets, through their hyperinsulinemic effects, can increase the risk of colorectal cancer.

Objective: We analyzed data from a cancer screening study to determine whether persons with high-glycemic-load diets would be at an increased risk of distal adenomas.

Design: We included subjects with no prior adenoma or cancer from the Prostate, Lung, Colorectal, and Ovarian screening trial and whose results from flexible sigmoidoscopy exams indicated either no lesions (n = 34 817) or ≥1 distal adenoma (n = 3696). We used a 137-item food-frequency questionnaire to assess usual dietary intake over the preceding 12 mo. Using logistic regression analysis, we calculated, separately for men and women, prevalence odds ratios (ORs) and 95% CIs of sigmoidoscopy-detected, distal adenomas for quintiles of energy-adjusted dietary carbohydrate, glycemic index, and glycemic load.

Results: ORs decreased with increasing intakes of carbohydrate for both the men and the women in unadjusted models, but these associations were attenuated in multivariate-adjusted models. Among the men, the association remained significant after adjustment (OR: 0.71; 95% CI 0.60, 0.84; P for trend < 0.0001), but in the women it did not (OR: 0.89; 95% CI: 0.73, 1.10; P for trend = 0.30). The results for glycemic index showed no associations in either men or women. Results for glycemic load closely mirrored those for carbohydrate.

Conclusion: Despite expectations that increasing glycemic load and glycemic index would increase the risk of adenoma, we observed no association in women and even an inverse association in men.

Key Words: Colorectal adenoma • glycemic index • glycemic load • carbohydrate • insulin resistance • fiber


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Giovannucci and McKeown-Eyssen have independently hypothesized a role for insulin in the etiology of colorectal cancer (1, 2). Insulin receptors activate signaling pathways in the cell that are mitogenic, suggesting that chronically elevated concentrations of insulin, such as we would observe in insulin resistance or non-insulin-dependent diabetes before ß cell failure, would increase the risk of cancer (1-3). Insulin could also influence the cancer process through indirect effects on the insulin-like growth factor (IGF) family. Whether the mitogenic effects of insulin are mediated primarily through the insulin receptor, the IGF-I receptor, or both, remains unresolved (4). Insulin also increases the number of hepatic growth hormone receptors, which is potentially important because growth hormone stimulates hepatic production of IGF-I, the main source of circulating IGF-I (1, 5). Perhaps more importantly, insulin has a strong inverse association with IGF binding protein 1 (IGFBP-1) (3, 6, 7); thus, even if insulin had only a modest effect on circulating IGF-I concentrations, it could have important effects on bioavailable, or unbound, IGF-I through down regulation of this binding protein.

Obesity is a strong predictor of both colorectal cancer and diabetes (7-10), with central adiposity in particular showing an even more pronounced association with these outcomes (7, 10, 11). Recent reports from the American Cancer Society Cancer Prevention Study II cohort (12) and from the Iowa Women's Health Study (13) showed an increased risk of colorectal cancer with diabetes. Perhaps more importantly, insulin resistance, or factors linked to insulin resistance, have been associated with an increased risk of colorectal cancer (6, 7, 11, 14-18). All this suggests that factors leading to diabetes or, more specifically, insulin resistance or factors that simply have hyperinsulinemic or hyperglycemic effects should also increase the risk of colorectal cancer.

The glycemic index was developed in the early 1980s by Jenkins and Wolever (19). The index is a measure of the glycemic effect of a particular food compared with an equivalent amount of carbohydrate in the form of pure glucose or white bread. The glycemic load of a serving of a specific food is simply the product of its glycemic index and the grams of carbohydrate from a single serving of that food and in this way combines quantitative and qualitative indicators of carbohydrate intake. Each has proven to be a useful measure of exposure in studies of the risk of insulin resistance and a variety of outcomes related to insulin resistance, including diabetes (20, 21) and cardiovascular disease (22), as well as intermediate markers of risk, such as serum lipids (23-26), glycated hemoglobin (23), and high-sensitivity C-reactive protein (27).

If insulin resistance and hyperinsulinemia are risk factors for colorectal cancer, and if a high-glycemic-index or high-glycemic-load diet increases the risk for insulin resistance, it should follow that such a diet also increases the risk for colorectal cancer. We evaluated the associations of carbohydrate intake, dietary glycemic index, and glycemic load with data from a large, multicenter study of sigmoidoscopy-detected adenomatous polyps.


   SUBJECTS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
We undertook the present analysis as part of the National Cancer Institute's Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer study. The PLCO was designed for the investigation of early detection of cancer methods as well as for studies of cancer etiology. Participants enrolled at 10 screening centers throughout the United States (Birmingham, AL; Denver, CO; Detroit, MI; Honolulu, HI; Marshfield, WI; Minneapolis, MN; Pittsburgh, PA; Salt Lake City, UT; St Louis, MO; and Washington, DC) (28). The PLCO study randomly assigned subjects at entry into either a screening arm (where the subjects would undergo screening for each of these 4 cancer sites) or into a usual care arm. Because the sigmoidoscopic exam administered as part of the screening protocol provided the endpoint data (ie, presence of adenomatous polyps) for these analyses, the results presented here come from the subjects assigned to the screening arm.

Study population
Between September 1993 and September 2000, 77 470 men and women aged 55–74 y were randomly assigned to the screening arm of the PLCO Trial. All study participants provided informed consent to participate, and the study was approved by the Institutional Review Boards at all participating institutions. Successful sigmoidoscopic exams (insertion to ≥50 cm with >90% mucosa visible or suspect lesion found) were carried out for 57 569 participants, among whom 52 143 (90.6%) also provided risk factor and dietary information (as described below). The participants whose sigmoidoscopic exam was suspicious for neoplasia (ie, a polyp or mass) were referred to their primary care physician for further care and possible endoscopic follow-up. Medical-pathologic reports on the removed lesions were obtained and coded by trained medical abstractors.

Of the participants with both complete sigmoidoscopies and risk factor and dietary questionnaires, we excluded 7571 persons for the following reasons: self-reported history of cancer other than basal-cell skin cancer (n = 2363); self-reported history of ulcerative colitis, Crohn disease, familial polyposis, colorectal polyps, or Gardner syndrome (n = 4181); extreme values (ie, lowest and highest 1%) on sex-specific energy intake (n = 998); and missing >7 items in the food-frequency questionnaire (FFQ; n = 436). Some participants were excluded for more than one reason.

After these exclusions, data from 44 572 participants (24 017 men and 20 555 women) were available for analysis. We focused, in the present analysis, on the distal colon (descending colon, sigmoid colon, and rectum), because the sigmoidoscopic screening procedure examines the distal portion of the colon. No distal lesions suspicious for neoplasia were found in 34 817 persons (17 794 men and 17 023 women), and these participants formed the control group. We compared these controls to a case group composed of the 3696 persons (2378 men and 1318 women) with pathologically verified distal adenomatous polyps. Of the 3696 participants with adenomas, 1325 had advanced adenomas defined as large (>1 cm), high-grade dysplasia (including cancer in situ), or villous elements (including villous and tubulovillous adenomas). The following participants were not included in any analysis: 1574 participants with hyperplastic polyps only, 123 participants with benign lesions only, 183 participants with colorectal lesions of unknown location, 1530 participants with polyps of uncertain histology or cancer, 169 participants with indeterminate screening results, and 2480 participants with a positive screening but no follow-up endoscopy (most of whom had diminutive polypss). In a sensitivity analysis, including these participants, classified either as cases or noncases, made no substantive difference in the results (data not shown).

Dietary assessment
We used a 137-item FFQ to assess usual dietary intake for each participant over the 12 mo before enrollment. The FFQ provided information for the ascertainment of portion size for all food items except for fruit and vegetables. We calculated nutrient intake from diet by multiplying the daily frequency of each consumed food item by the nutrient value of the sex-specific portion size based on the method developed by Subar et al (29), which uses national dietary data and the nutrient database from the US Department of Agriculture's (USDA) 1994–96 Continuing Survey of Food Intakes by Individuals (CSFII).

Because values for glycemic index and glycemic load do not exist in standard nutrient databases, we developed values for inclusion in the PLCO database according to a method described in detail elsewhere (30). Briefly, the nutrient database for the PLCO FFQ is based on roughly 4200 individual foods reported by adults in the 1994–1996 CSFII. This list was condensed into 225 nutritionally similar groupings of individual foods. Using published glycemic index values compiled by Foster-Powell et al (31), we linked glycemic index values (using a scale in which the glycemic index for pure glucose = 100) to each of the individual CSFII foods in these food groups. The method of linkage was by manual review of the glycemic index table to identify those foods that, in the judgment of the investigators, were the best matches for each of the CSFII foods. In cases where CSFII foods did not correspond closely to foods with published glycemic index values, we used a series of decision criteria [previously described (30)] to assign glycemic index values. We then calculated sex- and serving size–specific glycemic loads for each of the 225 food groups using the weighted mean method as described by Subar et al (32). These glycemic load values can be used in the PLCO database to calculate overall daily glycemic load based on FFQ reported frequency and portion size by sex across all items on the questionnaire.

In the USDA food composition tables used to compute nutrient values for CSFII, the carbohydrate value includes both available (ie, digestible) carbohydrate and dietary fiber. Because glycemic load is meant as an indicator of the glycemic effect of food, and glycemic effect is inherently a function of the carbohydrate available for digestion and absorption, for the purposes of our glycemic load calculations, we defined carbohydrate to be the USDA-based value for grams of carbohydrate per serving minus the USDA value for grams of dietary fiber per serving. Strictly speaking, available carbohydrate excludes not just dietary fiber but also resistant starch, but the USDA tables include most resistant starch in their definition of fiber, so subtracting the USDA-based fiber value from total carbohydrate is a reasonable approach. Failure to remove fiber from the carbohydrate value used in these calculations would result in overestimation of the glycemic load from any food containing fiber or resistant starch.

Eighty-five percent of the subjects completed the FFQ on or before the day of their screening exam. Of the remaining 15%, 90% completed the FFQ within 1 mo of the exam. In stratified analyses, there was no material difference in the results among those who completed their FFQs before, on the day of, or after their screening exams.

Covariates
At the time of initial screening, the participants filled out a risk factor questionnaire about sociodemographic factors, smoking history, use of selected drugs, disease history, family history of cancer, recent history of screening examinations, height, weight, and physical activity. Responses to these questions were used in multivariate analyses as described below.

Statistical analysis
We calculated, separately for men and women, prevalence odds ratios (OR) and 95% CIs for sigmoidoscopy-detected, distal adenomas using logistic regression analysis (SAS version 8.2; SAS Institute Inc, Cary, NC) for energy-adjusted quintiles of dietary carbohydrate, glycemic index, and glycemic load based on the distribution in controls.

In calculating ORs and 95% CIs, we used 2 modeling approaches. In the minimally adjusted models, we included energy, age at randomization, body mass index (BMI; calculated in kg/m2), and study center as covariates. In the second approach, to control for potential confounders based on a priori hypotheses of risk factors for colorectal tumors, we used multivariate models that included family history of colorectal cancer, ethnicity (white or nonwhite), physical activity (30 min of moderate or vigorous physical activity 3 times/wk or > or < 3 times/wk), regular use of aspirin or ibuprofen in the preceding 12 mo (yes or no), smoking (categories of duration and intensity), education (some college or more, high school graduate, or less than high school), alcohol consumption, energy-adjusted dietary calcium, calcium from supplements, energy-adjusted average daily red meat consumption, and total folate intake (combining dietary folate and folate from supplements). For missing data points, we imputed the mean (for continuous variables) or mode (for categorical variables). In no case did any of the potential confounding variables have imputed values for >1% of the study population. To test for confounding, we used models that entered each of these variables one at a time and then models that entered all simultaneously.

We used the residual method to adjust glycemic load, glycemic index, and carbohydrate for energy intake as described by Willett (33). We also used this method to adjust dietary calcium, dietary folate, and all other dietary variables. We did not energy-adjust calcium and folate from supplements because these nutrient intakes are not fundamentally related to total energy intake as would be the case for a nutrient from food. Stratified analyses were conducted to explore effect modification for selected factors including BMI, history of diabetes, and physical activity. The P value for trend was estimated by using carbohydrate, glycemic index, and glycemic load as continuous variables. All P values were two-sided.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
A summary of descriptive characteristics for the study population is provided in Table 1Go, separately for men and women, by quintile of glycemic load. In both sexes, those with a higher glycemic load tended to be slightly older, more frequently of nonwhite ethnicity, more physically active, of somewhat lower BMI, and much less likely to have a history of smoking. Persons in the high-glycemic-load quintile, although not differing significantly from those in the low quintile in vegetable consumption, did consume much more fruit, more grains (especially whole grains), slightly more calcium (including supplements), less red meat, and less alcohol. The percentage of calories from fat decreased sharply as glycemic load increased. This is not surprising, given that fat and total carbohydrate are inversely correlated in this cohort (r = –0.46 for women and –0.28 for men; P < 0.0001 for both) and that total carbohydrate is in part what determines glycemic load.


View this table:
[in this window]
[in a new window]
  		TABLE 1 Baseline characteristics by quintile (Q) of energy-adjusted glycemic load score1

 
By definition, glycemic load can be increased either by increasing total carbohydrate consumption or by consuming carbohydrate with a higher glycemic index. Thus, it was not surprising that both carbohydrate and glycemic index were strongly correlated with glycemic load in this population. Carbohydrate, however, was much more strongly correlated with glycemic load than was glycemic index (r = 0.92 in men and 0.89 in women for carbohydrate, and r = 0.48 in men and 0.42 in women for glycemic index). Interestingly, whereas total dietary fiber was strongly correlated with carbohydrate in this cohort (r = 0.54 in men and 0.47 in women), the glycemic index showed only minimal correlation with carbohydrate (r = 0.11 in men and 0.04 in women). This suggests that the glycemic index of the diet in this population was less strongly correlated with its fiber content, even though people who ate more carbohydrate also tended to eat more fiber. In fact, the correlation between fiber and glycemic index was only –0.24 in men and –0.34 in women. The top 10 food sources of carbohydrate in this study population are listed separately for men and women in Table 2Go.


View this table:
[in this window]
[in a new window]
  		TABLE 2 Top 10 carbohydrate-contributing foods for men and women

 
Results of the logistic regression are shown in Table 3Go (for the men) and Table 4 (for the women). Models with interaction terms for sex indicated significant effect modification (P < 0.001) for both available carbohydrate and glycemic load, and thus we provide separate tables for men and women. For glycemic index, there was no significant interaction, but for clarity and consistency of presentation, we also provided these results stratified by sex. In the analyses of all adenomas (both advanced and nonadvanced), the ORs were dramatically reduced with increasing intakes of carbohydrate for both men and women in unadjusted models, although the association was somewhat less pronounced in the women (OR for quintile 5 compared with quintile 1 for the men: 0.52; 95% CI: 0.45, 0.60; P for trend < 0.0001; OR for the women: 0.74; 95% CI: 0.62, 0.89; P for trend = 0.0005). These associations were attenuated in multivariate-adjusted models such that they remained highly statistically significant in men (OR: 0.71; 95% CI: 0.60, 0.84; P for trend < 0.0001), but in women they did not (OR: 0.89; 95% CI: 0.73, 1.10; P for trend = 0.30). Of the variables in the fully adjusted model, smoking accounted for the greatest portion of the attenuation in men and women, and 4 variables together (smoking, alcohol, folate, and red meat) accounted for essentially all of the attenuation.


View this table:
[in this window]
[in a new window]
  		TABLE 3 Odds ratios (ORs) and 95% CIs for distal colorectal adenoma in men by quintile (Q) of energy-adjusted available carbohydrate, glycemic index, and glycemic load1

 
By contrast, the results for the glycemic index regressions showed almost no significant associations with all adenomas in either men or women regardless of whether we used models adjusted only for energy, age, and study center or if we adjusted for the full range of potential confounders. Only among women and only in the minimally adjusted model was there evidence of a linear trend between glycemic index and adenomas, but, even in this case, the risk estimates for each quintile of glycemic index had 95% CIs that included 1.0. Furthermore, the linear trend was greatly attenuated and no longer statistically significant in the multivariate model.

The estimated ORs for glycemic load and adenomas closely mirrored those of carbohydrate (OR for quintile 5 compared with quintile 1 in multivariate-adjusted models for men: 0.79; 95% CI: 0.68, 0.93; P for trend = 0.003; and OR for women: 0.98; 95% CI 0.81, 1.19; P for trend = 0.70). As is evident in Tables 3Go and 4Go, the results when considering only advanced adenomas were essentially identical to those using all adenomas as the outcome variable.


View this table:
[in this window]
[in a new window]
  		TABLE 4 Odds ratios (ORs) and 95% CIs for distal colorectal adenoma in women by quintile (Q) of energy-adjusted available carbohydrate, glycemic index, and glycemic load1

 
Given that the association we observed, most notably in men, was strictly a function of overall carbohydrate quantity rather than a function of its quality (as would have been the case had glycemic index been associated with adenomas), we determined that it would be important to consider if this was a beneficial effect of carbohydrate per se rather than merely the substitution of carbohydrate in place of some other presumably harmful dietary component such as fat. Thus, we ran the logistic regression models again using the addition method of energy adjustment because, unlike the residual method, it does not require that each positive increment in one macronutrient be balanced by a decrease in one or more of the others (33). Results from the addition models were analogous to those from models adjusting for energy using the residual method (data not shown) confirming that the apparent beneficial effect of increased carbohydrate consumption was not dependent on the decreased intake of any other macronutrient component of the diet.

Given that high carbohydrate and high-glycemic-load diets were associated with a number of health behaviors in this study population, that health conscious people would likely have higher screening rates, and that we excluded people with prior history of adenoma, it is possible that the hypothesized harmful effects of a high-glycemic-index or high-glycemic-load diet may have been missed if people who ate this type of diet were more likely to have polyps identified in prior screening exams and thus be excluded from these analyses. However, when we limited the analysis to subjects without a reported history of any colorectal cancer screening, the results remained unchanged (data not shown). For both men and women, adjusting for fiber produced little additional attenuation of the association between carbohydrate (and glycemic load) and distal adenomas (Tables 3Go and 4Go).

Because people with a history of diabetes are more likely both to have a history of prediabetic hyperinsulinemia secondary to insulin resistance (which is the hypothesized mechanism linking high-glycemic-load diets and colorectal neoplasia) and to have modified their diets to reduce consumption of refined carbohydrates (ie, high-glycemic-load foods), it is possible that including such individuals in our study population may have biased the results in such a way as to obscure a positive association. In models that excluded the subjects with a self-reported history of diabetes, however, the results remained unchanged (data not shown).

In some previous studies of glycemic index and cardiovascular-related outcomes, the effects of glycemic load were confined primarily to that subgroup of the population with a BMI of >25 (22, 26, 27). Tests of interaction, however, showed no significant effect modification by BMI above or below the 25 kg/m2 threshold.

Physical activity has well-established, direct effects on insulin sensitivity (34-36), and therefore those who engage in regular physical activity may have different responses to the carbohydrate in their diets compared with more sedentary persons. However, as with BMI, tests of interaction provided no evidence of a significant effect modification by level of physical activity.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
We began these investigations with the hypothesis that diets characterized by a high glycemic index or, perhaps more significantly, a high glycemic load would increase the risk of colorectal neoplasia due to their promotion of insulin resistance and, consequently, hyperinsulinemia. After analyzing the data, however, we observed either no significant association or even the opposite result. Dietary glycemic index was not significantly associated with distal adenomas; moreover, glycemic load, although not significantly associated with the risk of adenoma in women, had a significant inverse association with distal adenomas in men.

Because glycemic load is the product of carbohydrate content and glycemic index, and because glycemic index was not significantly associated with adenomas in this study population, the association of glycemic load with adenomas was reduced to identity with that for carbohydrate. Furthermore, the results of the models that used the addition method of energy adjustment supported the conclusion that the beneficial effect of carbohydrate itself, and not the substitution of carbohydrate for some other dietary component, was responsible for these reductions in risk.

As stated above, previous studies have shown more pronounced effects of glycemic index and glycemic load in subjects with BMI values >25 (27). The proposed explanatory hypothesis in this case is that subjects with a higher BMI, who may be entering a state of insulin resistance, would be especially sensitive to the rapid rises in blood glucose concentrations that we would expect from eating the foods typical of a high-glycemic-index or high-glycemic-load diet. In our analysis, however, we saw no significant effect modification by overweight or obesity for the associations between glycemic index, glycemic load, or carbohydrate and distal adenomas.

Previously we reported in the PLCO study that dietary fiber had a strong inverse association with distal adenomas (37). In the initial models for the present analysis, we did not adjust for dietary fiber because fiber itself is an intrinsic component of glycemic index (and, therefore, glycemic load as well). But given that glycemic index showed no significant association with distal adenomas, and given the above-mentioned high correlation between carbohydrate and fiber, we considered the possibility that the carbohydrate association was confounded by dietary fiber. However, the odds ratios were almost entirely unaffected by the additional adjustment for fiber.

Although the results were unaffected by adjustment for fiber, it is possible that the observed results were due to something else found in a diet high in carbohydrates (eg, vitamins or phytochemicals derived from fruit or whole grains, etc) and perhaps not carbohydrate per se. To the extent feasible given the data, we controlled for this possibility in our logistic regression models, but we cannot rule out confounding by such unmeasured or imperfectly measured covariates.

These results were surprising and contrary to the prevailing hypotheses regarding the relation of glycemic load to hyperinsulinemia and, in turn, the relation of hyperinsulinemia to colorectal neoplasia. There is substantial evidence in support of this hypothesis, at least with respect to the underlying mechanism linking insulin resistance and colorectal cancer (3, 6, 7, 19, 38), but very few studies have examined the relation of dietary glycemic index and glycemic load per se with colorectal neoplasia (39-45). The results to date have been generally supportive of the hypothesis, although they have not been entirely consistent. In prospective studies, Higginbotham et al (40) found a strong positive association between glycemic load and colorectal cancer, and Michaud et al (42) found no association among women but modest positive associations among men (RR for the highest compared with the lowest quintile of glycemic load: 1.32; 95% CI: 0.98, 1.79). In a third prospective study, Terry et al (45) found no association overall, but they did find a positive association with distal cancer (and a complementary inverse association for proximal cancers). In 2 case-control studies of cancer, positive associations were observed for colon cancer but not rectal cancer (39, 41), whereas in another, the association was limited mainly to the proximal colon (44). In the only prior study of glycemic index and glycemic load and colorectal adenomas, Oh et al (43) found no significant associations. Why our results stand in contrast to the evidence relating to the underlying hypothesis as well as the existing literature looking explicitly at the glycemic load–colorectal cancer (though not adenoma) association, limited as it may be, requires some discussion.

As stated above, glycemic load was, in effect, a surrogate for carbohydrate in the PLCO study population, which may be explained at least in part by the rather narrow range of dietary glycemic index values we observed. The cutoff between quintile 1 and 2 was 52.1, whereas the cutoff between quintile 4 and 5 was 58.0 in men (with similar values for women), a difference between the high and low quintiles of only 5.9. Not only was the range low, the distribution of values was centered in the middle of the theoretical range for glycemic index (0–100), meaning the whole population was narrowly clustered around the middle value for glycemic index. With this type of distribution, detecting the effects of different levels of glycemic index becomes difficult unless this variable is a powerful determinant of disease risk even at midlevel (ie, nonextreme) values.

In contrast to our results, however, Higginbotham et al (40) observed a statistically significant increase in the risk for colorectal cancer with both increased glycemic load and glycemic index in a prospective study of women with a highly similar distribution of glycemic index values, which suggests that this range of intake is, in fact, adequate to observe glycemic index effects on colorectal outcomes.

Unlike all but one of the previous reports, the PLCO examined adenomas and not cancer as the outcome variable. If glycemic index and glycemic load elevated the risk in the later stages of disease, we would not have been able to observe that effect. Interestingly, in the only other study of adenomas, the authors similarly found no increased risk with higher glycemic load (43). This result would be consistent with the idea that glycemic load acts at late stages in the cancer process.

The PLCO study examined prevalent adenomas rather than incident adenomas. Although it is possible to argue that preclinical disease altered dietary preferences and thus created a spurious inverse association between glycemic load and adenomas, this seems an unlikely explanation for our results because prevalent adenomas are almost always asymptomatic. The 3 case-control studies, each of which would be subject to the same potential reverse-causality bias as the PLCO, all showed positive associations (44), and there is no specific evidence that persons with undetected adenomas do (or do not) change their diets in any particular way, if at all (39, 41). Furthermore, because all subject in the present analysis, both cases and noncases, were in the screening arm of the PLCO Trial, our study had the distinct advantage of having equal opportunity to detect an occult adenoma for each participant (ie, there was no detection bias).

The PLCO analysis considered only distal adenomas. Some evidence exists to suggest that proximal adenomas have, in many ways, a distinct etiology from distal adenomas, and, in fact, several recent cohort studies have shown stronger associations for proximal colon compared with distal colon cancer or for colon compared with rectal cancer (12, 13, 46). Because the proposed mechanism by which high GL or GI is thought to act is through diabetes-related hyperinsulinemia, it is possible that high GI or GL diets may have an effect in only one subsite. In the 3 prior case-control studies, the increased risk of cancer was limited either to cancers in the proximal colon or to cancers of the colon but not rectum (39, 41, 47). In one previous prospective study, the investigators also observed differences between distal and proximal colorectal neoplasia. Unfortunately, in this case, high glycemic load increased the risk of distal cancer (45), and no other studies observed different effects of glycemic load in the distal compared with the proximal colon, making it difficult to conclude with certainty that the effects of glycemic load are likely to be subsite specific.

In conclusion, the data we presented showed no significant increased risk with higher glycemic load, carbohydrate intake, or glycemic index and, in fact, showed a reduction in the risk for distal adenomas among men with higher carbohydrate intake (and glycemic load). Control for fiber resulted in no significant change in these associations. Given the divergence of these results from previous reports linking glycemic load to indicators of insulin resistance and related factors to colorectal outcomes, and given the sparse and inconsistent evidence available from studies directly connecting glycemic load and colorectal neoplasia, there is a clear need for additional research in this area.


   ACKNOWLEDGMENTS
 
AF developed the method for estimating glycemic load using the PLCO food-frequency questionnaire, analyzed the results, and drafted the manuscript. UP prepared the dietary data for analysis, assisted with analyzing the results, and reviewed the manuscript. DJAJ provided consultation on glycemic index methodology and assisted in drafting the manuscript. NC provided statistical expertise and assisted in drafting the manuscript. AS provided consultation on glycemic index methodology and assisted in drafting the manuscript. TRC, RB, and JLW participated in the planning of the study, oversaw execution of the screening study at regional centers, and assisted in drafting the manuscript. RBH and AFS provided consultation in development of the methods, assisted in drafting the manuscript, and supervised the research. None of the authors has any material interest in the contents of this manuscript.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Giovannucci E. Insulin and colon cancer. Cancer Causes Control1995; 6 :164 –79.[Medline]
  2. McKeown-Eyssen G. Epidemiology of colorectal cancer revisited: are serum triglycerides and/or plasma glucose associated with risk? Cancer Epidemiol Biomarkers Prev1994; 3 :687 –95.[Abstract]
  3. Sandhu MS, Dunger DB, Giovannucci EL. Insulin, insulin-like growth factor-I (IGF-I), IGF binding proteins, their biologic interactions, and colorectal cancer. J Natl Cancer Inst2002; 94 :972 –80.[Abstract/Free Full Text]
  4. Ish-Shalom D, Christoffersen CT, Vorwerk P, et al. Mitogenic properties of insulin and insulin analogues mediated by the insulin receptor. Diabetologia1997; 40 (suppl):S25 –31.
  5. Thissen JP, Ketelslegers JM, Underwood LE. Nutritional regulation of the insulin-like growth factors. Endocr Rev1994; 15 :80 –101.[Abstract]
  6. Bruce WR, Wolever TM, Giacca A. Mechanisms linking diet and colorectal cancer: the possible role of insulin resistance. Nutr Cancer2000; 37 :19 –26.[Medline]
  7. Giovannucci E. Insulin, insulin-like growth factors and colon cancer: a review of the evidence. J Nutr2001; 131 :3109S –20S.[Abstract/Free Full Text]
  8. Bianchini F, Kaaks R, Vainio H. Overweight, obesity, and cancer risk. Lancet Oncol2002; 3 :565 –74.[Medline]
  9. Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med2003; 348 :1625 –38.[Abstract/Free Full Text]
  10. National Task Force on the Prevention and Treatment of Obesity. Overweight, obesity, and health risk. Arch Intern Med2000; 160 :898 –904.[Abstract/Free Full Text]
  11. Schoen RE, Tangen CM, Kuller LH, et al. Increased blood glucose and insulin, body size, and incident colorectal cancer. J Natl Cancer Inst1999; 91 :1147 –54.[Abstract/Free Full Text]
  12. Coughlin SS, Calle EE, Teras LR, Petrelli J, Thun MJ. Diabetes mellitus as a predictor of cancer mortality in a large cohort of US adults. Am J Epidemiol2004; 159 :1160 –7.[Abstract/Free Full Text]
  13. Limburg PJ, Anderson KE, Johnson TW, et al. Diabetes mellitus and subsite-specific colorectal cancer risks in the Iowa Women's Health Study. Cancer Epidemiol Biomarkers Prev2005; 14 :133 –7.[Abstract/Free Full Text]
  14. Trevisan M, Liu J, Muti P, Misciagna G, Menotti A, Fucci F. Markers of insulin resistance and colorectal cancer mortality. Cancer Epidemiol Biomarkers Prev2001; 10 :937 –41.[Abstract/Free Full Text]
  15. Colangelo LA, Gapstur SM, Gann PH, Dyer AR, Liu K. Colorectal cancer mortality and factors related to the insulin resistance syndrome. Cancer Epidemiol Biomarkers Prev2002; 11 :385 –91.[Abstract/Free Full Text]
  16. Levine W, Dyer AR, Shekelle RB, Schoenberger JA, Stamler J. Post-load plasma glucose and cancer mortality in middle-aged men and women. 12-year follow-up findings of the Chicago Heart Association Detection Project in Industry. Am J Epidemiol1990; 131 :254 –62.[Abstract/Free Full Text]
  17. Nilsen TI, Vatten LJ. Prospective study of colorectal cancer risk and physical activity, diabetes, blood glucose and BMI: exploring the hyperinsulinaemia hypothesis. Br J Cancer2001; 84 :417 –22.[Medline]
  18. Saydah SH, Platz EA, Rifai N, Pollak MN, Brancati FL, Helzlsouer KJ. Association of markers of insulin and glucose control with subsequent colorectal cancer risk. Cancer Epidemiol Biomarkers Prev2003; 12 :412 –8.[Abstract/Free Full Text]
  19. Jenkins DJ, Kendall CW, Augustin LS, et al. Glycemic index: overview of implications in health and disease. Am J Clin Nutr2002; 76 (suppl):266S –73S.[Abstract/Free Full Text]
  20. Salmeron J, Ascherio A, Rimm EB, et al. Dietary fiber, glycemic load, and risk of NIDDM in men. Diabetes Care1997; 20 :545 –50.[Abstract]
  21. Salmeron J, Manson JE, Stampfer MJ, Colditz GA, Wing AL, Willett WC. Dietary fiber, glycemic load, and risk of non-insulin-dependent diabetes mellitus in women. JAMA1997; 277 :472 –7.[Abstract]
  22. Liu S, Willett WC, Stampfer MJ, et al. A prospective study of dietary glycemic load, carbohydrate intake, and risk of coronary heart disease in US women. Am J Clin Nutr2000; 71 :1455 –61.[Abstract/Free Full Text]
  23. Buyken AE, Toeller M, Heitkamp G, et al. Glycemic index in the diet of European outpatients with type 1 diabetes: relations to glycated hemoglobin and serum lipids. Am J Clin Nutr2001; 73 :574 –81.[Abstract/Free Full Text]
  24. Ford ES, Liu S. Glycemic index and serum high-density lipoprotein cholesterol concentration among us adults. Arch Intern Med2001; 161 :572 –6.[Abstract/Free Full Text]
  25. Frost G, Leeds AA, Dore CJ, Madeiros S, Brading S, Dornhorst A. Glycaemic index as a determinant of serum HDL-cholesterol concentration. Lancet1999; 353 :1045 –8.[Medline]
  26. Liu S, Manson JE, Stampfer MJ, et al. Dietary glycemic load assessed by food-frequency questionnaire in relation to plasma high-density-lipoprotein cholesterol and fasting plasma triacylglycerols in postmenopausal women. Am J Clin Nutr2001; 73 :560 –6.[Abstract/Free Full Text]
  27. Liu S, Manson JE, Buring JE, Stampfer MJ, Willett WC, Ridker PM. Relation between a diet with a high glycemic load and plasma concentrations of high-sensitivity C-reactive protein in middle-aged women. Am J Clin Nutr2002; 75 :492 –8.[Abstract/Free Full Text]
  28. Gohagan JK, Prorok PC, Hayes RB, Kramer BS. The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial of the National Cancer Institute: history, organization, and status. Control Clin Trials2000; 21 :251S –72S.[Medline]
  29. Subar AF, Thompson FE, Kipnis V, et al. Comparative validation of the Block, Willett, and National Cancer Institute food frequency questionnaires: the Eating at America's Table Study. Am J Epidemiol2001; 154 :1089 –99.[Abstract/Free Full Text]
  30. Flood A, Subar AF, Hull SG, Zimmerman TP, Jenkins DJ, Schatzkin A. Methodology for adding glycemic load values to the national cancer institute diet history questionnaire database. J Am Diet Assoc2006; 106 :393 –402.[Medline]
  31. Foster-Powell K, Holt SH, Brand-Miller JC. International table of glycemic index and glycemic load values: 2002. Am J Clin Nutr2002; 76 :5 –56.[Abstract/Free Full Text]
  32. Subar AF, Midthune D, Kulldorff M, et al. Evaluation of alternative approaches to assign nutrient values to food groups in food frequency questionnaires. Am J Epidemiol2000; 152 :279 –86.[Abstract/Free Full Text]
  33. Willett W. Nutritional epidemiology. New York, NY: Oxford University Press,1990 .
  34. Kahn SE, Larson VG, Beard JC, et al. Effect of exercise on insulin action, glucose tolerance, and insulin secretion in aging. Am J Physiol1990; 258 :E937 –43.
  35. Koivisto VA, Yki-Jarvinen H, DeFronzo RA. Physical training and insulin sensitivity. Diabetes Metab Rev1986; 1 :445 –81.[Medline]
  36. Sato Y, Nagasaki M, Nakai N, Fushimi T. Physical exercise improves glucose metabolism in lifestyle-related diseases. Exp Biol Med (Maywood)2003; 228 :1208 –12.[Abstract/Free Full Text]
  37. Peters U, Sinha R, Chatterjee N, et al. Dietary fibre and colorectal adenoma in a colorectal cancer early detection programme. Lancet2003; 361 :1491 –5.[Medline]
  38. Kaaks R. Nutrition, energy balance and colon cancer risk: the role of insulin and insulin-like growth factor-I. IARC Sci Publ2002; 156 :289 –93.[Medline]
  39. Franceschi S, Dal Maso L, Augustin L, et al. Dietary glycemic load and colorectal cancer risk. Ann Oncol2001; 12 :173 –8.[Abstract/Free Full Text]
  40. Higginbotham S, Zhang ZF, Lee IM, et al. Dietary glycemic load and risk of colorectal cancer in the Women's Health Study. J Natl Cancer Inst2004; 96 :229 –33.[Abstract/Free Full Text]
  41. Levi F, Pasche C, Lucchini F, Bosetti C, La Vecchia C. Glycaemic index, breast and colorectal cancer. Ann Oncol2002; 13 :1688 –9.[Free Full Text]
  42. Michaud DS, Fuchs CS, Liu S, Willett WC, Colditz GA, Giovannucci E. Dietary glycemic load, carbohydrate, sugar, and colorectal cancer risk in men and women. Cancer Epidemiol Biomarkers Prev2005; 14 :138 –47.[Abstract/Free Full Text]
  43. Oh K, Willett WC, Fuchs CS, Giovannucci EL. Glycemic index, glycemic load, and carbohydrate intake in relation to risk of distal colorectal adenoma in women. Cancer Epidemiol Biomarkers Prev2004; 13 :1192 –8.[Abstract/Free Full Text]
  44. Slattery ML, Benson J, Berry TD, et al. Dietary sugar and colon cancer. Cancer Epidemiol Biomarkers Prev1997; 6 :677 –85.[Abstract]
  45. Terry PD, Jain M, Miller AB, Howe GR, Rohan TE. Glycemic load, carbohydrate intake, and risk of colorectal cancer in women: a prospective cohort study. J Natl Cancer Inst2003; 95 :914 –6.[Abstract/Free Full Text]
  46. Hu FB, Manson JE, Liu S, et al. Prospective study of adult onset diabetes mellitus (type 2) and risk of colorectal cancer in women. J Natl Cancer Inst1999; 91 :542 –7.[Abstract/Free Full Text]
  47. Slattery ML, Schaffer D, Edwards SL, Ma KN, Potter JD. Are dietary factors involved in DNA methylation associated with colon cancer? Nutr Cancer1997; 28 :52 –62.[Medline]
Received for publication October 17, 2005. Accepted for publication June 22, 2006.




This article has been cited by other articles:


Home page
 Am. J. Clin. Nutr.Home page
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]

Home page
 Am J EpidemiolHome page
A. E. Cust, N. Slimani, R. Kaaks, M. van Bakel, C. Biessy, P. Ferrari, M. Laville, A. Tjonneland, A. Olsen, K. Overvad, et al.
Dietary Carbohydrates, Glycemic Index, Glycemic Load, and Endometrial Cancer Risk within the European Prospective Investigation into Cancer and Nutrition Cohort
Am. J. Epidemiol., October 15, 2007; 166(8): 912 - 923.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Flood, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Flood, A.
Agricola
Right arrow Articles by Flood, A.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS