Nutrition and Cancer: An Update

Kristine M. Conner

University of Pennsylvania Cancer
Last Modified: May 13, 2001

What definite positive associations have researchers found between certain nutrients or dietary components and the development of cancer? Very few, according to three researchers who attempted to summarize the state of nutritional research at ASCO this afternoon. Arthur Schatzkin, MD, of the National Cancer Institute, Maria Elana Martinez, PhD, of the University of Arizona, and Lawrence Kushi, ScD, of Columbia University, pointed to the special challenges of nutritional research and provided a general overview of what nutritional studies of colorectal cancer and breast cancer have shown. In short, these studies have not confirmed conventional wisdom about fat and fiber-not yet, anyway.

That might be because trying to trace a cause- effect relationship between nutrition and cancer is tricky. Dr. Schatzkin outlined the main challenges in observational studies of nutrition and cancer in certain populations, including "exposure assessment error, inadequate range in diet for a given population, and confounding" of various factors. Schatzkin went on to explain each and propose possible solutions that may help in each area:

  • exposure assessment error has to do with the fact that nutritional studies often rely on questionnaires and journals, and people tend to misremember or underreport their intake of certain foods in their diets. It's also difficult to separate out what the key exposure may be when testing a category of food, such as vegetables or red meat. For instance, it may be cruciferous vegetables, not just vegetables in general, that make the difference, but it's difficult to measure that intake specifically. Better targeted and more detailed questionnaires may help, as might the development of biological markers of dietary intake, which would reduce the reliance on participants' error-prone recollections.
  • inadequate range results from the fact that people in the same culture or region may very likely to eat in the same ways, which makes it difficult to trace real variations that could be associated with changes in cancer risk. One possible solution is to pool results from different geographic areas, and he noted that this is already going on in Europe.
  • "confounding" means that other socioeconomic factors are likely to come into play when observing dietary habits. People who share these habits also share other socioeconomic traits, and it is often hard to tease out a clear cause- effect relationship between nutrition and the development of cancer. Randomized clinical trials-that is, trials that test a nutritional intervention by comparing a group that receives it with a control group that does not-are better in this regard, but it's not always possible to develop a randomized trial for a suspected dietary cause of cancer. Schatzkin cited the example of alcohol and breast cancer.

"We have incontrovertible evidence which shows that other environmental factors, such as chemicals, radiation, and infectious agents, really do cause cancer, " Dr. Schatzkin said, "but the scientific evidence for nutrition is much softer."

Dr. Martinez and Dr. Kushi proceeded to review the evidence that is available from studies of colorectal cancer and breast cancer respectively. The recurring theme of both presentations was that there really are no proven statistically significant relationships between the traditionally suspected dietary factors and the development of these types of cancer.

Conventional wisdom has long held that a diet high in fiber and low in fact reduces the risk of colorectal cancer. Dr. Martinez reviewed a number of prospective cohort studies (studies that follow large groups of people) that have not shown this relationship to be statistically significant. For example, results from a study of participants in the Nurses Health Study, published in the New England Journal of Medicine in 1999, showed "no sign of an inverse relationship" between cereal, fruit, and vegetable intake and colorectal cancer risk.

"The larger prospective studies just aren't showing a compelling relationship between total fiber intake and risk for colorectal cancer," Dr. Martinez noted.

Furthermore, studies haven't shown that certain dietary factors can help to prevent the recurrence of adenomas in people who have already been diagnosed with colorectal cancer. Dr. Martinez reviewed a number of factors that have been evaluated, including vitamins, high fiber/low fat intake, and high fruit and vegetable/low fat intake, and noted that none of these agents has been shown to affect adenoma recurrence. For example, the Wheat Bran Fiber Trial randomized participants to high wheat bran fiber and low wheat bran fiber intakes and then followed them for five years. There appeared to be no statistically significant difference in the risk for recurrence.

The only slightly encouraging evidence that has come from clinical trials, Dr. Martinez noted, is about the possible preventive effects of calcium. Prospective studies of calcium and colon cancer have shown a "weak relationship" between high calcium intake and lowered risk of recurrence. She cited the example of the Calcium Polyp Prevention Study, in which nearly 1,000 individuals were randomized to either calcium or placebo. The results, published in the New England Journal of Medicine in 1999, showed a "positive preventive effect," but it was weak. While 38 percent of patients on the placebo experienced a recurrence, 31 percent of those on calcium did. Even though these results are not overwhelmingly positive, they do suggest that calcium may be a candidate for further testing.

Just as studies haven't confirmed the suspected connection between high-fiber/low-fat diets and colorectal cancer risk, they haven't confirmed an association between dietary fat intake and breast cancer, noted Dr. Kusci of Columbia University. The wide variation in breast cancer rates in different countries does seem to point to dietary habits, particularly the high intake of fat, as a factor. But large prospective studies have not shown that women who report higher intake of dietary fat also have a higher risk of developing breast cancer.

Most of the evidence Dr. Kushi cited was from the Iowa Women's Health Study, which he and his colleagues started with 42,000 women in Iowa fifteen years ago. They have been following the women closely during that time, and they have found no statistically significant relationship between dietary fat intake and breast cancer risk, nor between high intake of fruits and vegetables and level of risk.

Thus far, Dr. Kusci noted, there are only two factors that seem to have some associated with elevated risk: consumption of well-done red meat, and low levels of dietary folate- particularly when that low level is combined with alcohol consumption. The results are preliminary, but Dr. Kusci noted that there appears to be a correlation that simply wasn't observed with dietary fat intake.

"The only thing that's emerging as a possible factor is alcohol, especially when in the presence of decreased folate intake," he said. He and his colleagues currently have a paper in progress regarding their observations about low folate/higher alcohol consumption and breast .

Soy has been of interest, too, but too small a sample of Iowa women consume soy to make it possible to study this effect. The Shanghai Breast Cancer Study, Kusci noted, has shown a possible association between soy intake and decreased breast cancer risk, especially when a woman consumed soy as an adolescent.

Not surprisingly, members of the audience asked the panel for their advice about what to tell patients who ask questions about what's known about diet and these forms of cancer. The general consensus among the panel was that conventional wisdom should not necessarily be abandoned in the absence of data, but that more nutritional research is needed before physicians can answer patients' questions with any certainty.

OncoLink ASCO 2001 coverage is provided by an unrestricted educational grant from Amgen