 n our Special Topics analysis of diabetes research, Dr. Walter
Willett ranks among the top 25 authors publishing in this field in the
past decade, with 46 papers cited a total of 1,624 times. Dr. Willett’s
work can be found in the ISI
Essential
Science Indicators
Web product in the field
of Clinical Medicine. Dr. Willett is the Fredrick John Stare Professor
of Epidemiology and Nutrition at the Harvard School of Public Health
in Boston, Massachusetts. Special Topics correspondent Gary Taubes
recently spoke with Dr. Willett about his highly cited work—both its
roots and its implications.
|
 Usually when people think of your research, they think of nutrition
and heart disease. How did you get into diabetes research?
Well, in some ways, when you have a large cohort, and we have three—the
Nurses' Health Study I and II and the Health Professionals Follow-up
study—it effectively pushes you in the direction of looking at the
most important issues. Diabetes is clearly among the most serious
diagnoses for women, second only to breast cancer in the cohort, and
we watched an epidemic unfold before our eyes. In the three cohorts
together, numbering almost 300,000 individuals, we probably have over
10,000 cases of type 2 diabetes. So the numbers are huge. Also, it’s
a disease that is serious; complications are major, and the treatment
is actually not very good. People have a misconception that diabetes
is just inconvenient, that if you have the disease, you have to take a
pill or an insulin injection and that's about it. Most people are not
aware how serious it is in terms of the cardiovascular implications.
This, in part, is because the risk of complications is directly
related to how long you have the disease. For the first five years,
your risk of complications is pretty low. But when you get up to 20
years, there's a 10- to 20-fold increase in heart attack risk, as well
as renal failure and blindness. That's a really serious problem. And
now we're seeing people getting it younger and younger—including
adolescents getting type 2 diabetes, which is a new phenomenon.
What's the difference between type 1 and type 2 diabetes?
Type 1 is the type that is usually called juvenile diabetes and is
almost always diagnosed in childhood. It appears to be an autoimmune
degradation or a dysfunction of the pancreas. So it really is more
purely a disease of insulin deficiency, and thus insulin treatment is
necessary. Type 2 diabetes, on the other hand, develops primarily out
of resistance to insulin and it eventually becomes insulin deficiency.
The pancreas becomes exhausted and develops relative insulin
deficiency. The epidemic is really in type 2 diabetes, and it's pretty
clear that the major factor underlying that epidemic is the rise in
obesity.
Take us through your three most-cited papers in diabetes and tell
us why each one had such impact.
I think the physical-activity paper ("Physical-Activity and
Incidence of Non-Insulin-Dependent Diabetes-Mellitus in Women," Lancet
338 [8770]: 774-8, 28 September 1991) was probably the clearest
demonstration that physical activity is a protective factor for type 2
diabetes. There had been some suggestions prior to this paper, but
this was the biggest and most detailed study on the association. I'm a
bit surprised that that second paper ("A Prospective-Study of
Maturity-Onset Diabetes-Mellitus and Risk of Coronary Heart-Disease
and Stroke in Women," Archives of Internal Medicine 151
[6]: 1141-7, June 1991) comes along with so many citations. We had
more detail than other studies, particularly in women, on the
magnitude of risk, but the finding wasn't surprising. We have updated
that in a recent paper, with many more cases and longer follow-up. Hu
is the first author. We looked more at the issue of how risk goes up
with duration of illness. The answer you get for how strong diabetes
is as a risk factor for cardiovascular disease is that it depends very
much on the age distribution over the population and when they come
down with type 2 diabetes.
As for the paper on weight gain ("Weight-Gain as a Risk Factor
for Clinical Diabetes-Mellitus in Women," Annals of Internal
Medicine 122 [7]: 481-6, 1 April 1995), that's interesting because
people knew that being overweight was a risk factor. But what wasn't
appreciated was the magnitude of the impact of excess weight, and we
even showed that people with what's called "healthy weight
gain," at the upper edge of the normal distribution, still have
three to five times the risk of diabetes compared to someone at the
lower range. If you get up to seriously obese, you can have 50 to 100
times the risk of getting type 2 diabetes. It's one of the strongest
associations we have ever seen.
What is the mechanism by which obesity increases diabetes risk?
It's not entirely understood. Clearly, obesity increases insulin
resistance, which is the hallmark of type 2 diabetes. There's quite a
bit of work going on now to unravel this connection. For instance,
Gokhan Hotamisligil, in our department, is doing a lot of work on the
molecular basis of insulin resistance. It looks like one of the major
factors is tumor necrosis factor, which does in part mediate the
relationship between obesity and type 2 diabetes. But there are some
other mechanisms that relate to fatty acid binding proteins, as well.
What have we learned in the last decade about the risk factors for
type 2 diabetes?
We have learned, number one, how tightly related diabetes is to
body weight and weight gain. That relationship is extremely strong. We
have learned that physical activity is an important protective factor,
in part by helping control weight. But even independent of body
weight, physical activity reduces insulin resistance and the risk of
type 2 diabetes. We learned again that diabetes is a strong risk
factor for coronary heart disease, and that the relationship is very
strongly dependent on the duration of diabetes. In some of the more
recent papers, we've learned various aspects of diet have important
influences as well—including glycemic load. In other words, large
amounts of rapidly absorbed carbohydrates are related to the risk of
type 2 diabetes, probably by accelerating exhaustion of the pancreas
through increasing the demand for insulin. The total amount of fat in
the diet is not related but the type of fat is. Trans fat is
positively related to risk, and polyunsaturated is inversely related
to the risk of type 2 diabetes. And cereal fiber is related to a lower
risk, and that's probably closely related to the glycemic load issue.
So people who eat modest amounts of whole grain fiber have a lower
risk than people who eat modest amounts of highly refined starches and
potatoes. Finally, we've learned that alcohol is moderately protective
for type 2 diabetes as well. To sum that up, by taking advantage of
these behavioral risk factors, by engaging in moderate exercise,
avoiding excess weight gain, eating more whole-grain carbohydrates and
avoiding partially hydrogenated fats and eating natural vegetable
oils, you can eliminate over 90% of type 2 diabetes.
That summation was published in the New England Journal of
Medicine in October of last year and it is likely to soon be among
the most-cited papers. The first author was Hu. We essentially looked
at all the behavioral risk factors together, and showed you could
avoid over 90% of type 2 diabetes. [Secretary of Health and Human
Services] Tommy Thompson called us personally after that came out.
That's the first time anything like that ever happened. That was one
that really synthesized a lot of different papers.
How do you decide where you're going to submit your papers?
First of all, we generally don’t send something to a journal that
will be covered by the general media—JAMA, for instance, or
the New England Journal of Medicine—unless we're highly
confident of the result. We can never be perfectly certain, but
usually if we're going to do that, we have multiple data sets showing
the same thing. We only send a paper to a major journal if we have a
high degree of confidence that the answer is right and that it's
novel. The New England Journal of Medicine, for example,
generally doesn't want to publish confirmatory studies. JAMA is
the same way. A lot of times, however, the confirmatory studies are
really important. That’s when we know we can really believe the
evidence. So sometimes we do send very good papers, for instance, to
the Annals of Internal Medicine, but they might not be quite as
novel. Sometimes it's important to publish recent findings, even if
we're not really sure of them, so we need to get them out there so
somebody else can confirm or refute the results. We might send those
to a nutrition journal or an epidemiology journal. In those cases, the
findings may not be quite as secure, but it’s useful to get the
information out. Those are the major factors. Another factor is simply
what readership would be particularly interested in the topic. If
we're looking just at gestational diabetes, for example, we might send
that to an ob/gyn journal .
Where are you going next with your diabetes research?
We are doing a lot of work on genetic factors that can influence
the risk of type 2 diabetes. Although it is very clear that the
dominant factors are not genetic, there is still interest in looking
at the risk in relation to various polymorphisms in genes that operate
on metabolic pathways related to type 2 diabetes. This may give us
some practical explanations as well as enhancing our mechanistic
understanding. We’re also looking at the issue of what you can do to
minimize the complications once you do have type 2 diabetes. With
10,000 people in our cohorts with type 2 diabetes, we can now study
how their dietary and behavioral risk factors relate to future risk of
complications.
Have there been particularly stubborn challenges or obstacles to
pursuing the diabetes research? If so, what have they been?
Oh, there are landmines in everything in this business. For a while
some people wanted to dismiss our data on obesity and type 2 diabetes.
They were willing to acknowledge that truly obese people had increased
risk; they were not willing to acknowledge that the risk is basically
linear and that even people who are, by definition, mildly overweight
are at substantially increased risk of diabetes. That stuff seems very
much accepted now, but we had trouble getting some of it published.
People had hypothetical biases. They didn’t want to believe it.
Another real controversial finding was that the type of
carbohydrate was related to the risk of type 2 diabetes. In part, that
involves around the glycemic index area, and that was a minefield well
before we ventured into it. We really brought data to it that didn’t
exist before. That field has seen 20 years of guerilla warfare among
people who thought the glycemic index was important and people who
wanted to totally dismiss it. We have had papers rejected in that area
because of the strong negative belief that it’s the total
carbohydrates and not the type of carbohydrates that is the problem.
What we've seen is that it’s both the type of carbohydrate and the
quantity that is important. Although for diabetes, it actually seems
like the glycemic index, the form of the carbohydrate, is contributing
more than the total amount of carbohydrate.
Dr. Walter Willett
Harvard School of Public Health
Departments of Nutrition and Epidemiology
Boston, Massachusetts, USA
Read
an interview with Dr. Walter Willet in Science
Watch®.
|
ESI Special Topics,
June 2002
Citing URL - http://www.esi-topics.com/diabetes/interviews/DrWalterWillett.html
|
|
