Does a high glycemic index diet increase risk of colorectal cancer?

By Andrew Flood, Ph.D.
Assistant Professor
Division of Epidemiology and Community Health
University of Minnesota
School of Public Health

Insulin Resistance and Colorectal Cancer

In the mid 1990s, two investigators, Gail McKeown-Eyssen at the University of Toronto and Ed Giovannucci at Harvard, independently hypothesized that insulin may play a role in the cause of colorectal cancer.  This hypothesis was based on a number of observations, including that of the striking similarity in risk factors for both Type II diabetes and colorectal cancer.  For example, obesity is a strong predictor of both, with central adiposity (or excess abdominal fat) in particular showing an even more pronounced association with these outcomes.  Insulin, which is elevated in diabetics, is known to have important effects on cell growth and division. Disregulation of cell growth and division is a hallmark of cancer. Thus, it is the role insulin plays in cell activity that potentially linked the two diseases.

Supporting this role of insulin are reports from large cohort studies by investigators from the American Cancer Society and the University of Minnesota’s Iowa Women’s Health Study. Both studies show an increased risk of colorectal cancer with diabetes.  Perhaps more importantly, insulin resistance (a pre-diabetic condition characterized by chronically elevated levels of insulin in the blood), or factors linked to insulin resistance, have been associated with increased risk of colorectal cancer in a number of studies. These observations suggest that factors leading to diabetes or, more specifically, to insulin resistance should also increase risk of colorectal cancer.

What is the glycemic index?

The glycemic index was developed in the early 1980s by David Jenkins and Thomas Wolever at the University of Toronto as a tool to help diabetics manage their blood sugar.  The index is a measure of how much blood sugar rises after eating a particular food, compared to how much it rises after eating a similar amount of pure glucose.  A food with a high glycemic index is one where the carbohydrate it contains is absorbed very quickly and produces a spike of glucose in the blood.  The body uses insulin to transport the glucose from the blood into the tissues, so foods with a high glycemic index will demand a sudden increase in insulin to process the rise in glucose.  But in addition to the glycemic index of the food (how quickly the carbohydrate is absorbed) the insulin demand depends on how much carbohydrate is in the food.  A carbohydrate-rich food with a high glycemic index will place the greatest glycemic load on the body and will generate the greatest insulin demand.  Continual intake of high glycemic load foods is thought to lower the body’s sensitivity to insulin, and as insulin sensitivity decreases, more and more insulin is needed to process the same amount of glucose in the blood.  This state of insulin resistance will thus lead to chronically high levels of insulin, and in so doing will potentially put the body at increased risk of colorectal cancer.  Or so the theory would suggest.

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. This measure combines quantitative and qualitative indicators of carbohydrate intake.  Each has proven to be a useful measure of exposure in studies of risk for insulin resistance and a variety of outcomes related to insulin resistance, including diabetes and cardiovascular disease.

If insulin resistance and hyper-insulinemia are risk factors for colorectal cancer, and if a high glycemic index or glycemic load diet increases risk for insulin resistance, it should follow that such a diet also increases risk for colorectal cancer.  We evaluated the associations of carbohydrate intake, dietary glycemic index, and glycemic load with colorectal adenomas (pre-cancerous polyps) and cancer using data from two epidemiologic studies.

The PLCO Study

For the Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer study participants enrolled at ten screening centers throughout the United States (including the University of Minnesota).  Half of the PLCO study subjects were screened for colorectal cancer using flexible sigmoidoscopy (a procedure that is very effective in identifying cancer and pre-cancerous polyps in the last third of the colon).  These subjects also completed a questionnaire that assessed diet in the previous year.  Using the responses to the dietary questionnaire, we were able to estimate the typical daily glycemic load for each subject as well as the overall dietary glycemic index and total carbohydrate intake. 

A total of 38,513 men and women completed this protocol and were available for analysis.  Among these participants 3,696 had at least one adenoma identified by the sigmoidoscopy.  When we divided the study subjects into groups based on their glycemic load, we found a strong, statistically-significant association between glycemic load and presence of distal adenomas in men and suggestive evidence of an association in women.  But rather than increasing risk, we found that high glycemic load was associated with decreased risk of adenoma.  We found no association between dietary glycemic index and adenomas. Given the way we calculated glycemic load, the null result for glycemic index meant that the association we observed for glycemic load was in effect an association for total carbohydrate.  These results suggest, counter to our initial theory, that diets with a high glycemic load or, more specifically, that are high in carbohydrate will decrease risk of colorectal cancer, especially among men.  The glycemic index of the diet seems to be entirely uninvolved in risk for colorectal cancer.

The BCDDP Follow-Up Cohort Study

The surprising results from the PLCO study prompted us to consider the possibility that our theory was not valid but also to consider methodological limitations of the study that might have produced a seemingly contrary result.  For example, the PLCO analysis was cross sectional, meaning we could not tell if the subjects were consuming high glycemic index diets in the years prior to the actual finding of the adenomas we studied.  Also, adenomas are merely pre-cancerous lesions, only a small, unidentifiable fraction of which become invasive cancer, thus we could not make conclusions from PLCO about how glycemic load affects the development of the ultimate endpoint of interest, invasive cancer. 

Motivated by these concerns, and motivated by the concern that the PLCO results may simply represent a chance finding, we conducted a similar analysis in the Breast Cancer Detection Demonstration Project (BCDDP) Follow-up cohort study.  In this study, we assessed diet in roughly 45,000 women using similar methods as in PLCO but then followed the subjects for 8 years to identify which of them subsequently developed colorectal cancer.  Despite the stronger study design, we observed a nearly identical association for glycemic load and colorectal cancer as that we observed for adenomas in PLCO women.  As with PLCO, the association for carbohydrate and colorectal cancer indicated a marginally lower risk of cancer with the highest category of intake, and glycemic index showed a similar marginal inverse association.  Thus we repeated our PLCO results with the advantage in the BCDDP cohort of having a prospective design and a true cancer endpoint.

Conclusions

The results of these two studies (and of a now growing number of studies from other investigators showing similar results) have dampened the enthusiasm for what was once a promising hypothesis concerning the cause of colorectal cancer: that a diet producing regular, large glycemic responses would lead to insulin resistance, and that in turn would increase risk of colorectal cancer.  In fact, our results suggest something close to the opposite: the glycemic index of the diet is not strongly related to colorectal cancer risk one way or the other, and, if anything, diets high in carbohydrate are associated with decreased risk.  The curious issue with these results, however, is that although we see no evidence that high glycemic load or high glycemic index diets increase risk of colorectal cancer through direct observation of these exposures and outcomes, the indirect evidence linking glycemic load and glycemic index with colorectal cancer still holds.  Diets that are high in glycemic load or that have a high glycemic index are associated with increased risk of diabetes and insulin resistance-related outcomes.  Diabetes and insulin resistance have been linked in many studies to increased risk of colorectal cancer.  How dietary glycemic index and glycemic load can predict these known intermediate risk factors for colorectal cancer and yet not promote colorectal cancer itself remains a mystery.