The science of large numbers
Cohort studies shed light on cancer
Every year, 50 modern-day gumshoes fan out across Shanghai, China, hot on the trail of some of the worst miscreants that afflict the human race.
Armed with survey forms, they question the inhabitants of thousands of homes. Their goal: to find out why some people develop cancer and other diseases, and others don’t.
Begun a decade ago by researchers now at Vanderbilt University Medical Center, the Shanghai Women’s Health Study has yielded important clues to the mysterious connections between environment, genetics and disease.
- Women who have never smoked but whose husbands are heavy smokers are at greatly increased risk of dying from stroke.
- High intake of soy foods lowers blood pressure and decreases the risk of both coronary heart disease and bone fractures.
“Sometimes the associations between lifestyle and disease are so striking it surprises us,” says Wei Zheng, who directs the Shanghai study with his wife and colleague, Xiao Ou Shu. “We are conducting additional studies to get more definitive answers.”
The Shanghai investigation is known as an epidemiological “cohort” study. It is designed to track the development of disease in a large group of people over an extended period of time—usually decades.
Cohort studies can help reveal the impact that diet, exercise and other lifestyle factors can have on health and longevity. More recently, they’ve been used to explore the disproportional impact of disease on different ethnic groups.
An example is the landmark Southern Community Cohort Study, which is attemptng to explain why African-Americans are more likely than other groups to develop and die from cancer.
The study is a collaboration of the Vanderbilt-Ingram Cancer Center, Meharry Medical College and the International Epidemiology Institute (IEI), a biomedical research firm based in Rockville, Md.
Since it was launched in 2001, about 63,000 adults aged 40 to 79 have been enrolled through community health centers in 12 Southeastern states, including Tennessee. The goal is to recruit 90,000 participants.
“Historically, the home of the African-American population in the United States has been in the South,” explains William Blot, Ph.D., IEI chief executive officer and the study’s principal investigator. “There had never been an investigation in the South on this order of magnitude.
“When you start getting up in numbers of people with a particular type of cancer that approaches 1,000, that gives you pretty good power to start looking at environmental and genetic factors,” says Blot, who also is a professor of Medicine at Vanderbilt.
Focus on the good
Zheng got the idea for the Shanghai Women’s Health Study in the early 1990s while working at the University of Minnesota.
He and his wife met at Shanghai Medical University, where they both earned medical degrees and master’s degrees in public health before coming to the United States in 1989 and 1990 for Ph.D. training at Johns Hopkins University and Columbia University in New York, respectively.
“I was involved in the Iowa Women’s Health Study and wrote a paper focusing on tea consumption and cancer risk. Most studies look at what is bad about diet. I thought, ‘We need to focus on what is good about diets to help protect against cancer.’”
“I thought, ‘We have done all these things to identify risk factors. How about identifying protective factors?’ So I came home and developed the proposal to focus on dietary protective factors, and the NIH funded it right away.”
So far the study has recruited approximately 75,000 women between the ages of 40 and 70 in seven typical communities in Shanghai.
“With an epidemiological study, we want to recruit a large number of participants in order to have an adequate power to evaluate study hypotheses,” he says. “In other words, the more participants we have, the more confident we have about our research findings.”
While working with her husband on this study, Shu realized that more could be gained than simply studying women. In 2001, she launched the Shanghai Men’s Health Study. To date, 60,000 men have been enrolled, half of whom are married to participants in the women’s cohort.
“First, we did a small pilot study and discovered that the husbands’ and wives’ dietary habits are very different, although they share same living environment,” Shu says. “For instance, men like to eat more meat compared to the women.”
Most women in China also work outside of the home. “So there is an opportunity to look at environmental and occupational factors as well.”
One goal of the Shanghai and Southern Community cohort studies is to determine whether differences in traditional Asian and Western diets account for widely varying incidences of different cancers among residents of China and the United States.
Researchers know that Asia and the United States have quite different cancer spectra. In China and Japan, stomach cancer used to be the number one culprit followed by cancer of the esophagus; whereas in the United States, lung, colon, and breast cancers dominate.
However, the cancer spectrum in some parts of China, such as Shanghai, is starting to more closely resemble that of the United States. For people who move from China to the United States, the risk of stomach and esophageal cancers decreases while the risk of lung, colon, and breast cancers dramatically increases.
The million-dollar question, of course, is “Why?”
One hypothesis is that lifestyle factors—including diet—account for these differences.
The Shanghai studies were designed to test the hypothesis that the traditional Asian diet, which includes soy foods, bok choy (Chinese cabbage), white radish, ginger root, tea and ginseng, may reduce the risk of diseases including some cancers.
To find out, the Shanghai studies rely on trained interviewers who go door-to-door. Because most Shanghai residents live in apartment towers, dozens of study participants can be found in one building. The interviews are later transferred from paper to an electronic form for data analysis.
“In-person interviews improve the quality of the data,” Zheng explains, “particularly across a large population with diverse educational and income backgrounds. If you asked someone to fill out a form, you may get different quantities and quality of responses. In-person interviews minimize the differences.
“Secondly, the response rate is high with in-person interviews. We have a 93 percent response rate. Mailed surveys typically get a 25 percent to 40 percent response.”
A higher response rate makes it easier to generalize findings across a population, he adds.
Empower the people
The Southern Community Cohort Study takes a different tack. Its researchers rely on community health centers to enroll study participants, most of whom are lower income.
Betty Scott, an interviewer at the West End Medical Centers in Atlanta, has enrolled over 2,600 people into the study. Her interest in the project is more than academic.
“I have a history of cancer in my family,” she explained in 2007, having lost her father, a brother, and a sister to cancer. Three other brothers have been diagnosed with cancer.
“Of course, I want to know what is causing so much disease,” Scott says. “We all have theories, but until research proves what is causing cancer, we will never know the answers.”
“A lot of folks in our area have been affected by cancer,” adds Emily Beauregard, who enrolled about 90 people a month in 2006 at the Family Health Centers’ Portland Clinic in Louisville, Ky.
“There’s not a lot one person can do to stop disease or make someone who has cancer better,” says Beauregard, who is currently enrolled in a master’s degree program in public health. “But if you can participate in something that may in the long run halt or decrease the rates of cancer, people feel that is empowering to be able to do something about it.”
Biological samples—urine, blood, cheek cells (for DNA)—are sent to Vanderbilt University Medical Center, where they are stored in freezers for future analysis.
“We get boxes from up to 30 different centers every day,” Blot says. “The blood is separated into 14 different tubes and stored in a freezer bank.”
Once the data and biological specimens have been collected, the real detective work begins.
“We will do that through a ‘case-control’ study,” he says. “… We will identify everybody, say 500 people… who (have) developed lung cancer, and get their blood specimens. Then we choose a control group of 500 or 1,000 people (without cancer) who are the same age, sex, race, etc., and pull their blood specimens.
“Then we look for differences between the cases versus the controls. Meanwhile, we have all this background information on everybody, their smoking history and other factors.
“If the cause of a disease in this lower income population proves to be genetic, we should be able to apply our findings to higher income and higher education populations,” Blot continues. “Even environmental associations seen in the study population may apply more broadly, but we will examine these closely before making any extrapolations to other segments of society.”
Typically, during the first five years of an epidemiological study, most of the effort is devoted to recruiting study participants and collecting survey data and biological samples. The value of the cohort study increases as it is followed over the years and cohort members begin to develop different diseases.
The Shanghai Women’s Study has already begun to shed light on a number of areas.
Among the findings: “Women who are non-smokers but who are exposed to the cigarette smoking of their husbands have an increased risk of dying of stroke,” Zheng says. “We also learned that soy food intake reduces the risk of fractures, hypertension, coronary heart disease and diabetes.”
Simply adopting Asian eating habits may not yield the same benefits in the United States, Shu cautions.
“Even though lots of people in the South eat rice and greens, as do people in Shanghai, the specific type of vegetables and the way the food is prepared is very different,” she says.
In addition, “the ways soy foods are consumed in the U.S. are quite different from how they are consumed in China,” Shu adds. “For example, many soy products in the U.S. contain a large amount of sugar, while most soy products are consumed in China without the addition of any sugar.”
As for the Shanghai Men’s Study, which recently completed recruitment, Shu notes that the smoking rate in this cohort is high—67 percent.
“We found that smokers have a lower body-mass index than their non-smoking counterparts,” she says, “but more centralized obesity or beer belly, meaning they gain more weight around their torso. This can more be harmful to health than less centralized obesity and may increase the risk of cardiovascular disease and cancer.”
Cohort studies can yield conflicting results. For example, a 2003 study of California residents that found no relationship between environmental tobacco smoke and tobacco-related deaths contrasts sharply with a more recent finding from the Shanghai Women’s Health Study, which linked second-hand smoke to an increased incidence of cardiovascular disease, stroke and lung cancer.
While conflicting results can sometimes delay effective prevention and treatment strategies, they can also lead to new insights, says Walter Willett, M.D., MPH, Dr.P.H., professor of Epidemiology and Nutrition at the Harvard School of Public Health who launched the second Nurses’ Health Study in 1989.
“Efforts to understand the differences (between studies) can result in new knowledge,” Willett explains. “Sometimes the questions being asked are really different questions.”
Margaret Hargreaves, Ph.D., co-principal investigator of the Southern Community Cohort Study and professor of Internal Medicine at Meharry, is all too familiar with conflicting study results.
“One reason you may see conflicting findings from studies is that people may not have large enough samples when they do their research,” she notes. “That’s why we’re going for large numbers of enrollees.”
Hargreaves recalls that for a long time researchers thought obesity was closely associated with breast cancer. “Then somebody came out and said no it wasn’t, based on their findings. Now, there’s another wave of studies saying that maybe it is,” she says.
“What that means is that you just have to keep doing different studies, studying different kinds of people and gathering more specific information and making sure you see as large a number of people as you can.”
Pooling results from cohort studies that involve different population groups also can yield valuable insights.
Researchers at the University of Texas M.D. Anderson Cancer Center in Houston, for example, are planning to compare findings from their Mexican-American Health Study with those from a Native American cohort study in South Dakota.
“We’re enrolling families, ages 5 to over 90,” says Melissa Bondy, Ph.D., professor of Epidemiology and principal investigator of the Mexican-American Health Study. “The onset of breast cancer occurs at much younger ages in Mexican-American women than in white or African-American women. And there is a high rate of diabetes in the population.
“The rate of smoking among Mexican-American women is very low, but we’re seeing it in girls,” Bondy says. “We’re seeing kids as young as 11 and 12 years old already experimenting with smoking tobacco.”
While the goal of this study is to identify risk factors associated with disease patterns within Mexican and Mexican-American populations, “comparisons across groups with ethnic and cultural differences may help explain determinants of the differences in disease rates across these groups,” Blot points out.
Blot hopes eventually the studies will identify certain subsets of people who are more likely than others, because of their genes, to respond to lifestyle and environmental exposures that put them at an increased risk.
“If you know that some people have an increased risk, then we can advise them to avoid certain exposures,” he says. “You could advise those at high risk to be under increased surveillance for early detection of a cancer. For, even if you can’t prevent the cancer, at least you may be able to catch it early when the disease is more amenable to successful treatment.”
View Related Article:
What’s Caterpillar got to do with it?: The role of the private sector in cancer research