Stanford University researchers have used statistical approaches employed in genome-wide association studies to link the effects of nearly 200 environmental factors with phenotypes of cardiovascular disease.
The approach, published in the International Journal of Epidemiology, provides a "systematic" way of associating multiple environmental factors with disease in a single epidemiological study — an area where the field is lacking, according to the team.
The IJE paper outlines how the authors used GWAS statistical tools to evaluate188 environmental factors associated with differing levels of triglycerides and low-density lipoprotein cholesterol in a cohort of over 3500 individuals as well as changing levels of high-density lipoprotein cholesterol in a second cohort of over 7,400 individuals all of whom participated in surveys conducted by the Centers for Disease Control and Prevention between 1999 and 2006.
The scientists reported that they used the method to identify 29 environmental factors correlated with changing levels of triglycerides, nine factors associated with changes in LDL-C levels, and 17 factors associated with differences in HDL-C levels.
The team also applied the same approach to environmental factors that were associated with type 2 diabetes in a pilot study published in PLoS One in 2010.
Chirag Patel, a postdoctoral research fellow in Stanford’s pediatrics department and one of the study’s authors, told BioInform this week that the researchers borrowed from GWAS because its toolkit was "simple and easy to use” and provided a means of “communicating results to the broader genetics community" in a way that could be easily understood.
Although several studies have linked environmental factors with disease risk, it hasn't been easy to get a holistic picture of the relationship between diseases and environmental factors because studies on the subject usually explore only a few factors at a time, the authors explain in the paper.
One study, for example, might explore the link between physical activity and serum lipid levels while another study might ask if dietary intake is associated with these phenotypes, Patel explained.
Part of the challenge is that environmental exposure is influenced by differences in time, place, and exposure to numerous other factors, the authors note in the paper.
Also, humans are often exposed to multiple "environmental adverse or protective factors simultaneously," and "due to this complexity, the net effects due to environmental factors on human health may be miscalculated when considering a few factors at a time," the paper states
Furthermore, looking at factors in isolation, with each study addressing issues such as the assessment of environmental effects in different ways, leads to conflicting study results, the researchers point out.
Patel et al.'s approach not only provides a more methodical approach to studying multiple environmental factors at the same time, but also offers "one type of analytic process to unify and standardize" the analyses of these factors, the paper states.
The researchers used statistical tools such as linear regression analysis and false discovery rates to correlate exposure to environmental chemical factors — such as hydrocarbons, polychlorinated biphenyls, dibenzofurans, and vitamins — with levels of triglycerides, LDL-C, and HDL-C, adjusting for factors such as age, sex, ethnicity, diabetes status, diet, medications, and blood pressure.
In the study, the researchers used data collected by the CDC's National Health and Nutrition Examination Surveys, including the results of tests on urine and blood samples that assessed levels environmental pollutants, contaminants, nutrient levels, and pesticides.
Among other results, the team found that higher levels of cotinine — a biomarker for exposure to tobacco smoke — were associated with lower levels of HDL-C — the form of cholesterol that is associated with good cardiovascular health at high levels.
The researchers also found that high levels of persistent organic pollutants such as organochlorine pesticides, dibenzofurans, and polychlorinated biphenyls were all unfavorably associated with both triglyceride and HDL-C levels.
For example, two PCBs were associated with between 15 percent and 19 percent higher triglyceride levels, which have been associated with an increased risk of heart failure; while heptachlor expoxide, an organochlorine insecticide, was associated with 3 percent lower HDL-C levels.
Meanwhile, gamma-tocopherol, or vitamin E, was associated with higher levels of triglycerides and LDL-C — a form of cholesterol associated with increased risk of cardiovascular disease.
On a more positive note, vitamins B, C, and D were all associated with higher HDL-C levels, the researchers reported.
Patel and colleagues hope to make their approach available as a software tool within the next year for both environmental epidemiologists and geneticists.
Patel believes that the availability of such software could increase the adoption of environmental data in health and disease research studies, and could also encourage the epidemiological community to create a public repository for environmental data derived from their experiments.
Although the NHANES surveys that provided the source material for the study are publicly available, most data from environmental studies isn't readily available for further study, Patel said.
"It would be great to have the method and to have a repository like the Gene Expression Omnibus come on board for some of these environmental datasets," he said.
Even though the researchers are planning to launch a dedicated software package for this application, Patel noted that researchers could easily apply the approach described in the paper since GWAS methodologies are well known in the community.
Right now, he said, “we are trying to crank out some more of these results to make a point about how valuable such a method and way of viewing results is.” Once that's done, “it would be a great time to move the method to being openly available.”
The team also hopes to study in more detail how interactions between environmental and genetic factors affect disease, Patel said, adding that the researchers have already begun to explore these interactions in diabetes.
Meanwhile, the researchers hope to launch a long-term study in which they would observe a longitudinal cohort of individuals for 10 or more years and measure levels of exposure to a number of environmental factors as well as track the health outcomes of participants.
The IJE study provides a single snapshot of the effects of environmental factors; however, the reality is that levels of exposure and the particular factors to which an individual is exposed vary greatly over the course of a lifetime, unlike genetic variants, which do not change, Patel noted.
For this study “we had to use summary measurements on folks coming from all different age groups and sexes” for the method to work effectively; and while the researchers were able to gain some valuable insights, a more detailed study could help “tease out the causality a little bit better,” he explained.
The researchers also plan to test the observed impacts of environmental factors in model systems such as mouse in order to validate their predicted associations.
Have topics you'd like to see covered in BioInform? Contact the editor at uthomas [at] genomeweb [.] com.