Dr. Walter C. Willett is the Fredrick John Stare Professor of Epidemiology and Nutrition at Harvard School of Public Health and a founding investigator of the Nurses’ Health Study II. He spoke about diet and cancer at NCI on February 22-23 as a Division of Cancer Epidemiology and Genetics Visiting Scholar.

What have we learned from the completed long-term studies of diet and cancer risk?
What has become clear in the last 10 or so years is that by far the most important impact of diet on cancer is mediated through body weight, resulting from overweight and inactivity. I think it's important to keep in mind that it's not just being lean that's beneficial, it's also being physically active - staying lean and active is the most important thing one can do to prevent cancer, after not smoking.

There are still many specific components of diet that do seem to be important for cancer prevention, such as keeping red meat consumption, particularly processed meat, relatively low. The total percentage of calories in the diet from fat doesn't seem to be important. There may still be some modest benefit of higher fruit and vegetable intake for cancer prevention, but it's not the 'big bang' it was thought to be 15 or 20 years ago.

But it's also important that we don't adopt a lifestyle just for preventing one disease, and we particularly have to pay attention to cardiovascular disease, as that is still the number one cause of death. But in general, many of the same things we do for cardiovascular disease are going to have an impact on cancer.

During your visit, you mentioned that prospective observational studies might be
more appropriate than randomized trials for resolving questions about diet and cancer
prevention. What are the difficulties in doing randomized trials of dietary interventions?
The paradigm of randomized trials came from clinical medicine, where you have a sick patient and you know what the right time to start the intervention is - it's when they get sick. For prevention, particularly for prevention by lifestyle factors, the questions don't always fit the randomized trial paradigm, partly because very often we don't know when to start the intervention, and partly because there's a lot of evidence that, for many of these factors, the relevant period is early in life. It's impossible to imagine doing trials where you start in adolescence and follow people for 40 years.

Also, we're often looking at factors that are much more complicated to change than just taking a pill or placebo - to actually change human behavior is difficult. It's not impossible, but to keep the two trial groups separated is the real challenge. That can be a serious problem in a randomized trial, particularly when you're dealing with items that are on the shelves of your grocery store, in pill form or in food form.

What areas of research in the field of diet and cancer are investigators really excited about right now?
I think there are many different areas to investigate where there's promise, but the evidence is not clear, yet. Vitamin D is clearly a potentially important topic, where there are many suggestions that there will be important effects on cancer. Among the other aspects of diet, I think there may be a number of specific foods or nutrients that have modest beneficial effects, and when you 'package' them all together, there may be a very substantial effect.

Another critical area is milk consumption and calcium intake, because there are now quite a few studies showing that higher intakes of calcium or dairy products are related to increased risk of fatal prostate cancer. And that's important, because the national recommendation to drink three glasses of milk a day would double dairy consumption and production in the United States. As we have no good evidence suggesting it would reduce fractures and quite a few studies showing a relationship with fatal prostate cancer, that's an important area to resolve.