Harvard School of Public Health: The wide angle lens of population research
Like a photographer who uses a wide-angle lens to capture a sweeping landscape, the Harvard School of Public Health applies biological, quantitative, and social sciences to analyze health problems across entire populations. This broad perspective provides a depth of insight into disease prevention – the school’s hallmark endeavor – while complementing basic and clinical research.
“The school is a unique institution that is, by nature, interdisciplinary,” says Nancy Mueller, SD, professor of epidemiology at HSPH and associate director of Population Science at DF/HCC. “We reach across disease programs to focus on research that we can apply to public health.” The school’s epidemiologists and biostatisticians work closely with laboratory and clinical investigators at other DF/HCC institutions to identify environmental, behavioral, and genetic factors that affect cancer risk and to develop interventions for prevention and treatment. Here is a snapshot of some of their cancer-related research.
Lessons from large cohorts
Epidemiology studies, which analyze the patterns and contributing factors of disease in groups of people, form the cornerstone of population research. These studies often involve large cohorts – such as the Nurses’ Health Study, Nurses’ Health Study II (NHS II), and the Health Professionals’ Follow-up Study – whose health and lifestyle are tracked over decades. In the center of this work is Walter Willett, MD, DrPH, professor of epidemiology and nutrition at HSPH and co-leader of the DF/HCC Cancer Epidemiology Program, who for 25 years has studied the nutritional and hormonal determinants of cancer risk.
To study diet and lifestyle factors during early adult life and their relationship to breast cancer risk, Willett and his colleagues established the NHS II, selecting a population of women (ages 25 - 42 ) younger than those in the original nurses’ cohort. Using detailed questionnaires, they collected data about the women’s eating habits during high school, the diets of their mothers during pregnancy, and other information that is helping investigators piece together the lifetime experiences of these women to see which factors relate to breast cancer risk.
In research published recently in JAMA, his group reported a 50% higher risk of breast cancer in women who consumed high amounts of red meat as compared with those who ate small amounts. “An interesting aspect of the study,” says Willett, “was that this relationship was found only in women with estrogen-receptor positive tumors, suggesting that a hormonal pathway may be involved.” Blood samples taken from the NHS and NHS II participants have allowed investigators to study further the relationship between hormone levels and breast cancer risk, while the women’s DNA has enabled genome-wide association studies (see feature story on CGEMS).
Another area of Willett’s research focuses on the impact of obesity on cancer risk. “Over the last decade,” he says, “we have realized that, after cigarette smoking, overweight is the most avoidable cause of cancer” – accounting for about 20% of cancer deaths in women and 15% of cancer deaths in men. Although the risk for postmenopausal breast cancer increases even with modest weight gain, the newest NHS findings show that shedding 10 pounds or more and keeping it off after menopause can reduce that risk by up to 60%. Knowledge of this kind, explains Willett, can provide strong motivation to lose weight.
Making sense of scientific data
Another core discipline at HSPH is biostatistics – the quantitative techniques for describing, analyzing, and interpreting health data. Rebecca Betensky, PhD, professor of biostatistics at HSPH and leader of the Biostatistics Program at DF/HCC, specializes in developing new methods for analyzing study data, such as genomic information from brain tumor assays. When a new technology becomes available, she says, often there is no optimal method for analyzing the data emerging from it.
One such technology is array-based comparative genomic hybridization (aCGH), a technique using microarrays for screening DNA sequences across the genome to detect copy number aberrations: losses, gains, or amplifications, which are often associated with the development and progression of cancer. Mapping these gains and losses accurately can help pinpoint the genes implicated in cancer and provide powerful tools in diagnosis and prognosis.
While writing the statistical section of a grant involving aCGH, Betensky wondered if there might be a better way to analyze the data. One of the drawbacks to earlier methods, she explains, is that they analyzed each chromosome and hybridization separately. The new method she developed – with the help of a biostatistics doctoral student – combined data across all chromosomes and hybridizations. “Our innovation was to use a model-based approach that allowed us to analyze data jointly, rather than piece by piece,” says Betensky, “and to quantitate our certainty about whether there was a loss, a gain, or no change at a given marker.” When compared to existing methods, Betensky’s approach proved better at recognizing true gains and losses and improved the certainty of copy number conclusions.
In other research, Betensky is devising new methods using survival data to guide the classification of patients based on their underlying genetics, and to apply this classification to help predict the prognosis of patients still living.
Molecular markers and DNA differences
At HSPH, a growing area of population science known as molecular epidemiology integrates the disciplines of biostatistics, molecular biology, and genetics to answer pressing public health questions. This is the domain of David Christiani, MD, MPH, professor of environmental health at HSPH and co-leader of the DF/HCC Lung Cancer Program.
In the Lung Cancer Susceptibility and Outcomes study, Christiani and colleagues are exploring how heritable risk factors interact with environmental exposures, particularly smoking, to alter the risk of lung cancer – a disease that kills more people every year than breast, colon, and prostate cancers combined. The goal of the study is to create a comprehensive risk model for lung cancer susceptibility, especially among patients who have stopped smoking. “We’d like to develop a profile for those who are at higher risk even after quitting that would be applicable to a whole population of current smokers and healthy ex-smokers and would help us identify interventions.”
The case-control study tracks single nucleotide polymorphisms (SNPs) – common gene variations, not mutations – in a number of molecular pathways implicated in cancer, including angiogenesis (new blood vessel development), metabolism, DNA repair, cell growth, and apoptosis (programmed cell death). To date, investigators have studied about 100 individual SNPs in these pathways; they will evaluate 1,536 more in the next two years.
Some of the most stunning gene-environment interactions have been discovered in DNA repair pathways. Polymorphisms that confer proficiency in DNA repair, says Christiani, also allow smoking-damaged cells to survive, though partially repaired, rather than undergo apoptosis. These polymorphisms were “certainly designed to preserve the species,” he adds, “but with exposure to smoking can lead to untoward effects.” This paradox is what he calls the “double-edged sword of DNA repair.”
In related research (the Molecular and Genetic Analysis of Lung Cancer Survival), Christiani and colleagues are developing simple predictive models of survival using pathways known to be important in therapy responsiveness. “We found that the same polymorphisms of DNA repair pathways could also predict outcome,” says Christiani. “We’ve been able to delineate several markers in DNA repair and cell cycle control that confer a three-fold advantage – or disadvantage – depending on which allele [version of the gene] the patient carries.” Investigators are now trying to determine which patients are likely to experience a significant toxic reaction to therapy, based on their tumor genetics.
“One aspect of this survival study highlights the importance of molecular epidemiology,” says Christiani, “and how it intersects with basic and clinical science.” Recently, the study became the basis for an important collaboration with other DF/HCC investigators from the Cancer Genetics and Lung Cancer Programs that led to one of the most significant advances in lung cancer treatment in five decades, explains Christiani. “It’s a very tangible example of the critical role the Harvard School of Public Health and population science play in cancer research.”