Clear Search

Investigating Breast Cancer: Dr. Walter Willett

By BCRF | August 19, 2019

Diet, Lifestyle and Breast Cancer Risk

Subscribe to Investigating Breast Cancer here:

It’s something we all think about every day: What should I eat? It’s a question that Dr. Walter Willett thinks about every day.

A BCRF investigator since 2001, Dr. Willett’s research has aimed to characterize the impact of diet and lifestyle on health outcomes, especially in relation to breast cancer risk. His current studies focus on the effects and protein sources on breast cancer risk and finessing the tools used to analyze dietary intake.

Dr. Willett, Harvard Medical School, is a global leader who focuses on the intersection of diet, lifestyle and health. As the most cited nutritionist worldwide, his work has influenced numerous health recommendations and continues to inform preventive strategies for breast cancer. Dr. Willett is Chair of the Department of Nutrition at Harvard School of Public Health.

Read the transcript below:

Chris Riback: Dr. Willet, thanks for joining me. I appreciate your time.

Dr. Walter Willett: Thank you, and glad to be with you.

Chris Riback: Before we get into the science, I’d love to ask you about timing, because you seem to have timed your career about as perfectly as anyone I’ve come across. Your initial studies into diet, lifestyle and disease began really in the 70s and early 80s, I guess, with the Nurses’ Health Study 1, and various follow-ups and work before that and obviously new studies in the decades since. But your work timed perfectly, as far as I can tell, with the explosion in our popular culture around those very topics: lifestyle, diet and disease. Should we just cut to the end and go ahead and claim that you are personally responsible for this cultural phenomenon, or did it just so happen that your area of scientific research intersected perfectly with the times?

Dr. Walter Willett: It would be overstating it to say I’m responsible for the current interest, but in fact I’ve been interested in food since I was under 10 years of age, and I learned how to milk a cow when I was four years of age. So, food has been part of my interest for a long time. I grew vegetables to put myself through college and studied food science … went to medical school. I was fortunate in the sense that all of these pieces of my background have proved to be very useful when the interest emerged in diet and health.

Chris Riback: Was it kind of emerging when you … I mean, you studied, as you said, food science. Was the … how strong was the research and kind of the general interest at that time as you were getting into it, into that intersection between what we eat and who we are, how we live and who we are? Bring me back to that time a little bit. I realize I was being facetious in hinting that perhaps you invented it, which I know, but what were those times like?

Dr. Walter Willett: This really goes back, I think, to the 1960s when there were some early, very simple, crude kind of studies, but that were important in stimulating research. They were what we call today ecological studies, looking at rates of major diseases like breast cancer and heart disease in various countries around the world, and what those studies showed was that there were huge differences that were first really well documented in the 1960s. Rates of heart disease varied tenfold across Northern Europe and Southern Europe and about eightfold for breast cancer between Japan and the United States.

And then some other very simple, but really critically important, studies showed that people moving from low incidence countries like Japan where breast cancer rates were very low to the United States eventually adopted … it took a generation or two, but eventually adopted rates of breast cancer and then heart disease that were really similar to European Americans living in the United States. So, those really profound basic observations fueled a lot of interest. People said, “Why? What is there about living in the United States or other western countries that leads to such high rates of heart disease and breast cancer and other conditions similar to those diseases?”

Dr. Walter Willett: And that really … those data were emerging in the 1960s, 1970s and then I went to medical school. I, during that time, got more interested. We were faced with people with cancer and heart disease and almost nobody was asking why someone had breast cancer, why someone had heart disease. That kind of question bothered me and got me interested in trying to understand the basic origins of these conditions. So, this was a good time, and I was fortunate that a lot of my background had prepared me to take on some of those very complicated, challenging questions.

Chris Riback: And what were the reactions in the scientific community when you started to ask why?

Dr. Walter Willett: Well, first of all, there were some indicators that diet might be important and that’s because there were correlations if we looked across countries with higher fat intake did have higher rates of cancer and higher rates of cardiovascular disease. But epidemiologists in general knew that there could be other factors that were correlated with, say, fat intake that were the real causes, other aspects of diet, smoking, physical activity. And so we really had to look more deeply. When we started to look at diet, the conventional wisdom was you can’t study that within the United States population because everybody eats the same. But it didn’t take us very long when we started collecting data, we realized that not everybody ate the same. There were huge differences in people’s diets and therefore we had an opportunity to identify the factors that might be important, or not important, for breast cancer and other conditions.

Chris Riback: Why are nutrition-based studies so challenging? You just mentioned or hinted at one aspect of doing a study, which is focusing on self-reporting of what people eat, and maybe you didn’t actually say self-reporting, you said when we discovered what folks eat. But there’s a self-reporting component, or at least there was historically, but that has evolved. Take me through the science behind nutrition-based studies. What made them historically challenging and what makes them potentially challenging today?

Dr. Walter Willett: Nutrition studies are challenging when we’re looking at long term consequences, like risk of breast cancer, and part of that is related to the origins of breast cancer and many other diseases themselves, because these are diseases that don’t just pop up overnight. As we dig more deeply, we see that the origins of these diseases … excuse me, one second … we see that the origins of these diseases often are many decades before the condition is actually diagnosed, so to understand the causes we’re going to have to do studies that last for many decades.

Dr. Walter Willett: Second, diet itself is very complicated, that probably no two people eat exactly the same diet. We can look at it on the basis of foods. We can look at it as nutrients, and these different dietary components are often correlated. They’re usually correlated with each other, so pulling them apart is challenging. There also is no simple biochemical test, blood test, for defining someone’s diet. For example, just to take something as extreme as sodium, the body regulates sodium intake very, very precisely so the blood test tells us almost nothing about sodium intake, even though we can measure sodium in the blood.

Dr. Walter Willett: So, for many aspects of diet we do need to rely, at least up until this point in time, primarily on individual reporting of what they ate. And one of the other challenges to doing these kinds of studies was that skeptics would say, “Well, I can’t remember what I ate for lunch yesterday. How can people possibly report what they ate?” But the fact is, we’re not really interested in what someone ate for lunch yesterday. We’re interested in what they usually eat over the longer term. And as we’ve studied this, we do see that people can report their intake reasonably well. Not perfectly, but reasonably well.

Dr. Walter Willett: For example, some people, if we’re studying milk, that’s been a great interest, there are many people that have three to four glasses of milk a day. Others have none at all, and a lot of people in between. And we can ask how often people have a glass of milk, and it’s not very hard to separate people. Those people who have three or four glasses a milk a day can easily identify themselves, especially when we contrast them to people who consume almost no milk in their diet. And with standardized questionnaires, we’ve found that people’s report actually does correlate quite well with biomarkers of intake, for example. We can see that blood levels of carotenoids do correlate quite well with people’s reported intake of fruits and vegetables that are high in carotenoids.

Dr. Walter Willett: So, there are definitely challenges. There will never be a perfect study, and we have to accept that. The perfect study would probably be too randomized. Children, when they’re born to diets high in carrots or high in milk compared to low in milk, and follow them for the rest of their life, and of course we can’t really do that. And now we’re even learning that the mother’s diet is likely to be important as well. So, we’re not going to be able to do a perfect study but we can look at pieces at a time and put the whole package together to see the picture, the total picture, even though we can’t do the perfect study.

Dr. Walter Willett: As we go on in time, that picture becomes clearer, with bigger studies, longer follow up, better measurements. The picture becomes sharper with time. So this is a long process. We’ve learned a lot, actually, in the last few decades, information that we didn’t have when we started off in the 1970s. But there’s still additional, many details to learn.

Chris Riback: It’s remarkable how many hundreds of thousands of people you and others have been able to … follow might be a little bit too specific, but have had as part of your studies for so many years, I guess beginning with the Nurses’ Health Study, the first one, which I think launched in 1980 or certainly the very early 1980s, that the range of inputs on data that you and other scientists and researchers must have just seems … it’s kind of incredible. I want to ask you about some of the specific work that you’re doing around not just breast cancer but specific types of breast cancer.

Chris Riback: To get into that, perhaps the broadest question that I’m going to ask and too broad so forgive me, but I’m hoping to use it as a launching point into the more specific studies. What would you characterize … what do we know about diet, lifestyle and cancer?

Dr. Walter Willett: We have learned, first of all, that this is a long process, that what you eat today doesn’t affect your cancer risk tomorrow. What you were eating as an adolescent probably does affect breast cancer risk many years later, so we have to have long term studies. We have … probably the single strongest claim to emerge is that overweight and especially weight gain during adult life is a major risk factor for many types of cancers, and I think a lot of people just take that as a matter of fact today. But 20 years ago, that actually wasn’t appreciated. Actually obesity and overweight are almost equal to smoking as a cause of cancer when we look at a total population on an individual basis. Smoking is definitely worse than being obese, but since we have many more people who are overweight and obese than we do people smoking today, the total number of cancers caused by overweight and obesity is actually about the same as the number caused by smoking.

Dr. Walter Willett: We’ve also found that some choice of foods does make some difference for cancer. The poster child of nutritionists for many years has been fruits and vegetables, and there was some clear overstatement about the potential benefit for cancer reduction of fruits and vegetables. But as data have come in, we have seen that particularly ER-negative breast is related to … associated with low intake of fruits and vegetables, so there is some payoff there, especially for some of the most aggressive forms of breast cancer.

Dr. Walter Willett: High consumption of red meat at various times of life is related to several cancers. And even moderate alcohol consumption is related to breast cancer. That was a finding that was very controversial when we reported it back in the 1980s, but that’s been confirmed in dozens and dozens of studies now, and is an accepted risk factor for cancer. So, that’s a quick overview of some of the key findings that have emerged for cancer risk, and overweight and obesity is clearly the biggest part of the picture.

Chris Riback: That’s a pretty good answer to about the broadest question I could possibly ask, so let me try to home in a little bit. As I understand it, your work now will examine the relation of dietary factors to the risk of specific types of breast cancer. You’ve talked about this a little bit so far in the conversation. Including tumors characterized by HER2 status, histology and stage, how does … describe for me if you would the work that you are doing now and how does this differ from your previous breast cancer work?

Dr. Walter Willett: Let me just take a step back there. For a long time, we’ve really considered breast cancer as one disease. Early on clinicians did learn that different forms of breast cancer mainly characterized by being estrogen receptor positive or negative responded differently to treatments and therefore the treatments were personalized, we might say in today’s language, to ER negative versus ER positive.

Dr. Walter Willett: Early on in our breast cancer studies, we didn’t really have a chance to separate these forms of cancers because we didn’t have enough cancers to look at those separately. But as time has gone on in our studies, we now have in fact in our large cohort studies almost 20000 women have developed breast cancer and we’ve been able to get medical records with the receptor status documented, or we’re now collecting tumor samples and actually analyzing the cells for receptor status and other tumor characteristics. And we do find that risk factors for estrogen receptor negative first, and estrogen receptor positive breast cancers are different, so it is really important to study these separately.

Dr. Walter Willett: And as time has gone on, of course we’ve learned about other characteristics of breast tumors and we’re essentially characterizing the cancers that develop in our large studies by these other features, and that provides quite a bit more power to identify risk factors. If we lump them all together we can miss some important relationships.

Chris Riback: How will this research occur? What’s the process? Do you leverage existing people and existing data that are already in your cohorts, and you will just look at that data differently or through different lenses? Or do you need to build a new group, or do you add a new group and compare that to a historical group? How will the research work for you? Or, how is it working for you?

Dr. Walter Willett: For the most part, we’re leveraging the information that’s already available, that we’ve been collecting since 1980. Since that time we’ve enrolled about 200,000 women. I should mention they’re all registered nurses and the success of these studies really has been dependent on the incredible commitment to being in research on the part of these nurses. Almost 90% who are still alive are still participating almost 40 years later. The information we have on diet and lifestyle and hormones and physical activity, other lifestyle factors that’s in our computer that has been accumulating since 1980 is really invaluable. You can’t buy that kind of information … when we want to look at types of breast cancer.

Dr. Walter Willett: So starting all over would be a very big step backwards. We’d have to wait decades before we had the answers. So, about 20 years ago we did ask … started asking women in our study for permission to get samples of their breast tumors, and we’ve been collecting those samples. In the early years we were just storing them. We now have tumor samples from about 9,000 participants. And more recently, we’ve been sampling each one of those tumors and making what we call micro arrays. So on a single slide, we can put samples of several hundred breast tissues, and then we work with our collaborating pathologists, who are experts in essentially identifying characteristics of these tumors using a variety of methodologies, some analyzing DNA, some analyzing different proteins in the tumors, and that allows us to identify the subtypes of breast cancer.

Dr. Walter Willett: We are also starting Nurses’ Health Study 3, but information from that won’t be available, results won’t be available for several decades down the road. So, we’re essentially piecemeal-ing various birth cohorts of women over time to learn about these important relationships. But the basic process here is using new technologies, applying the latest developments in genetics, metabolomics, microbiome and pathology to tumor tissues that have been accumulating in our cohorts’ first several decades.

Chris Riback: Wow. And Nurses’ 3, is that group the same initial age group as Nurses’ 2 and Nurses’ 1? I think that those were kind of in the age 20 to 32, was it? Or 35, something in that zone for when the nurses came into the studies? Is it the same for Nurses’ 3?

Dr. Walter Willett: In general, yes. One of the really important general findings we’ve found for breast cancer is that the origins are often in earlier life. And in the first Nurses’ Health Study in 1980 they were 34 years of age when we first collected dietary data, and we wanted to look at younger years so we started off as young as 25. That’s been very helpful because we could do two things, we could … enough of their mothers were alive we could actually ask their mothers, and about 40,000 mothers responded, described their pregnancy with the nurse who’s in our study and their breastfeeding characteristics, early life feeding characteristics, their weight gain during the pregnancy. So we do know a lot about our nurses even when they were in utero.

Dr. Walter Willett: And then we could also, when they were still young we could ask in retrospect about their diets during high school years, which we found to be particularly important. And one of the clues about that period of age was that in the American atomic bombing of Nagasaki and Hiroshima in Japan, if women experienced radiation exposure during those early years, a few decades later their breast cancer rates jumped up. But if they were over 40, there was almost no increase in breast cancer risk when they were … due to radiation exposure. So it really did point to early life being important, and it’s turning out that what people eat, what girls eat during high school, does make an important difference in their future breast cancer risk. So that’s why we’ve been going to earlier years.

Dr. Walter Willett: That’s not to say that everything is cast in stone after those years. We still see that changes in behavior, especially changes in weight even up into the 50s and 60s can make an important difference in breast cancer risks, so we really need to look at the whole life span.

Chris Riback: So I could imagine somebody listening to this right now and saying, “Okay, doc, I’ve got a daughter, she’s a teenager. You just said that early life and early diet and lifestyle matters. What should I tell her to do?”

Dr. Walter Willett: Well, having had a couple teenagers at one point in time, just telling them what to do maybe … doesn’t always produce the responses we would like, but-

Chris Riback: No, it doesn’t. In fact the recommendation might be not that this is what I say to do but this is what Dr. Willet says to do. So no one will … we’ll all blame it on you.

Dr. Walter Willett: Right, yes, blame your doc. Somebody else, that’s for sure. But still, we should convey information to our kids growing up and more importantly, we should provide to them healthy choices and encourage them in every possible way that we can, including making them attractive and interesting. For the specifics, we have seen that high consumption of red meat is related to higher risk and emphasizing more plant bases of protein sources is related to lower risk and that higher intake of fruits and vegetables and whole grains is related to lower risk as well.

Dr. Walter Willett: In general, this does fit a Mediterranean-type dietary pattern that’s got a lot of variety. It’s not vegan, necessarily, but emphasizes more plant-based protein sources than animal-sourced foods. So that’s the general picture of what we’re seeing, and as time goes on we hope fully we’ll be able to define some of those characteristics of a healthiest diet even better.

Chris Riback: What about the relation between diet and lifestyle and the survival from breast cancer?

Dr. Walter Willett: Until quite recently, we’ve been mainly focused on looking at aspects of diet and lifestyle that could prevent breast cancer. But now with long follow up after multiple decades in our cohorts, we’ve been able to look at aspects of diet after diagnosis of breast cancer and how that relates to survival from this serious disease. That’s taken a long time because we, in our research, first want to have a diet before breast cancer and someone develops breast cancer, then we collect the data after diagnosis of breast cancer, what people eat, the physical activity and other information.

Dr. Walter Willett: And then we start follow up, looking at survival. And we know that for breast cancer the risk remains elevated for two or three decades, at least, after the diagnosis, and that does offer an opportunity to potentially modify our diet and other lifestyle patterns. Our information is just starting to emerge from this follow up looking at survival. We do see that an overall healthy dietary pattern, we might call it a Mediterranean dietary pattern, is related to better overall survival, and right now we’re actively looking at more detail, at pieces of the diet, and how they can be important in improving survival. We’re hoping to have some pretty firm results over the next year or two in that regard. This is quite an exciting new area, and I think we’re going to have some answers.

Chris Riback: This is just coming to me as I’m listening to you, so this may be a wrong way to think about it, but is there any … so, when I think about smoking, I think okay, the correlation … as a lay person, the correlation between smoking and lung cancer has to be … that’s got to be really strong. And obesity, you mentioned, and so obesity with heart disease, or smoking with heart disease. I think I know, as a layperson, from having read popular culture that there are things that one can do in one’s life that really impacts those diseases. Colon cancer, perhaps, with what I’ve heard about red meat, and if any of this is wrong you’ll correct me on my facts, because I’m just going off of my general understanding.

Chris Riback: Where would breast cancer … is there any way to think about the strength, which may not be the right word, but the strength of the inputs into the negative effect of breast cancer, meaning I know that if I smoke the negative effects resulting in lung cancer has got to be pretty strong. I don’t know the science exactly, but I know from life that that’s got to be pretty strong. Are the inputs, dietary lifestyle, can it be correlated as strongly with breast cancer or is it more complicated, we don’t know as much? And if the question has just kind of confused the issue totally, then feel free to just set me straight.

Dr. Walter Willett: Breast cancer is definitely more complicated than diabetes and heart disease in terms of the causal factors, and how they operate over time. And for example, for smoking and lung cancer we can see if we eliminated smoking, we would reduce lung cancer by about 90% from what it was a few decades ago. In fact, we’ve already made a lot of that reduction, so further reductions are not going to be as great since most people are not smoking now. And for heart disease we can … if somebody’s smoking, they can cut their risk of heart disease by about two thirds by not smoking.

Dr. Walter Willett: The relationships between risk factors and breast cancer are not nearly as strong. Probably the strongest lifestyle factor is waking during adult life in relation to breast cancer after menopause. And breast cancer … one of the examples of why breast cancer is so complicated is that actually being overwinter as a young child or adolescent is related to lower breast cancer risk, not higher breast cancer risk. That is one of the reasons why a lot of women who are struck with breast cancer when they’re 35 or 40 say, “How can that be? I did everything right?” And they’re correct. This is an enigma. We have some clues about what might be explaining that, but we still don’t totally understand this really unusual finding of cancer rates being lower with higher obesity.

Dr. Walter Willett: But weight gain during adult life is related to breast cancer after menopause, so it’s that adult weight gain that’s really important. And that is moderately strong. We see that women who gain quite a bit of weight can about double their risk of breast cancer after menopause by doing that, or conversely, avoiding that weight gain can reduce the risk by about half, compared to what it would have been with gaining a lot of weight. That’s pretty typical in the United States.

Dr. Walter Willett: When we start putting together multiple risk factors, we can see, for example, that weight gain in combination with use of hormones explains about half of the breast cancer incidents, or mortality, in the United States. In other words, if almost no women use hormones after menopause and did not gain weight, breast cancer rates would probably be about half of what they are. And interestingly enough, in Japan, very few women have used hormones after menopause and quite amazingly Japanese women, on average, do not gain weight during adult life. If anything, they slightly reduce their weight during adult life. So those two variables explain about half the difference. So we’re seeing some important pieces, but they’re not as strong and they’re more complicated than lifestyle factors in relation to lung cancer or cardiovascular disease.

Chris Riback: What role has BCRF played in your research?

Dr. Walter Willett: BCRF funding allowed us to start the cohort of about 25,000 offspring of our Nurses’ Health Study 2 when these kids were 10 to 14 years of age, and that’s really still the largest cohort now of adolescents who were enrolled during that period of life and where we have a lot of dietary and other data. These children are now in their 30s and we’re already being able to look at diet and benign breast disease, and down the road we’ll be able to look at breast cancer itself.

Dr. Walter Willett: So the NIH funding was just not fit for that kind of study. They want answers within five years, but we know that long term investments in research are really critical if we’re going to understand the real origins of this disease. There are many other areas where BCRF funding has been critical. It allowed us to start collection of mammograms from participants in our study, and that’s a whole new direction that’s opened up. Actually, abnormal mammograms are the strongest risk factor we have, one of the strongest risk factors we have for breast cancer. That’s been an important aspect of our research.

Dr. Walter Willett: We’ve been able to do pilot studies that in the longer run gave us … allowed us to get funding for NIH research. It’s helped us develop the world’s, really, most comprehensive data base on the biochemical constituents of foods that we update every four years. There’s no other database in the world that has done that over a long period of time. So in many ways, BCRF funding has been critical. But it’s not just the funding that BCRF has brought us together … I’m speaking of our whole group of colleagues funded by the BCRF that come from many different fields, clinical areas, biochemists, pathologists and many other areas. Getting together and exchanging ideas has been an important part of the support BCRF has given as well.

Dr. Walter Willett: And then finally, just the fact that there are so many people, so many women out there who are actively contributing to research on breast cancer has been an inspiration for us that keeps us going, working into the night, working weekends to take on this really serious challenge.

Chris Riback: Well, that’s terrific. In that case, I ought to let you get back to work. Thank you, thank you for your time and thank you for the work that you have done over decades.

Dr. Walter Willett: Well, my pleasure and please give my thanks to everybody who has helped make this possible.

Chris Riback: That was my conversation with Dr. Willett. My thanks to Dr. Willett for joining and you for listening. To learn more about breast cancer research or to subscribe to our podcast, go to