Before researchers can start mapping out patterns among the constellations of environmental exposures, however, they need to assemble a more comprehensive picture of the possible internal and external exposures.
"It's complicated," Rappaport concedes, of embarking on the massive endeavor. "But if you look at it from the perspective of the Human Genome Project that we tackled about 20 years ago, I don't think it's any more daunting."
The dearth of environmental data stems mainly from a preoccupation with the seemingly sexier field of genetic correlates, Rappaport says. "People have been spending all of their time and energy and money looking at the genetic factors," he notes. So in terms of understanding factors at work on the environmental—and arguably more potent—side, "they've hardly scratched the surface."
There has been some progress, however. Thousands of small-molecule metabolites have been profiled in the effort to develop a chemical signature profile, or metabolome. But the few studies that have been done have taken relatively smaller samples of available chemical readings to assess.
One study, published in May in PLoS ONE and co-authored by Butte, scanned blood and urine samples of thousands of people for the presence of different chemical compounds, looking for correlates with type 2 diabetes. "I think that's really a good example of what we should be able to do," Rappaport says.
The study, however, was not quite as strong as a contemporary genome-wide association study (GWAS) would be, Rappaport notes. He explains that a true GWAS surveys hundreds of thousands of genes and the diabetes study looked only at 266 environmental chemicals.
This reduced approach can lead to both false-positive associations and more robust correlations being missed. These smaller chemical sample studies are more "analogous to what they call a candidate gene study," in which researchers assess only a handful of likely genes," than to a GWAS, Rappaport notes. And to take a lesson from the genomics field, he says, after a GWAS follow-up original flagged genes from candidate studies "almost always turn out not to be important."
Much of the allure of the Human Genome Project as a model for other fields is its allegiance to the data. Its search-and-map mandate allowed for a largely unbiased survey of the genome. Such a clean plan, however, has thus far proved difficult in a field often tainted—and indeed driven—by sweeping chemical- and disease-specific hypotheses.
Rappaport and Smith argued in their essay that the familiar tactics of single-source and single-disease research are premature and should be put on hold in favor of more expansive investigation of all internal and external exposures. "We're at the point now where we don't really know what's important," Rappaport says. So surveying every possible exposure—and combination of exposures—is crucial, he notes.
A comparable environmental data set to the human genome, such as the exposome, however, is still a ways off, and its completion depends on support by major research funders, such as the National Institutes of Health. The NIH is in the midst of a $200-million, five-year program called the Genes, Environment and Health Initiative (GEI). Nevertheless, Rappaport says, such pairings are still challenging, given the lack of data about the environmental exposure side of the equation.
"Even if we keep making progress on the genetics of these diseases, we have to keep studying the environment," Butte says. "Maybe a variant of a gene only leads to a disease when an individual happens to be in an environment that we don't even know about," he says. Getting to the bottom of environmental exposure has the potential to clarify many of the diseases and genetic mutations that currently seem completely random.
The long-standing division between genetics and environment itself might need some blurring. "We set up these artificial constructs going back a hundred years," Butte says. "It makes us think it's either a genetic factor or it's an environmental factor, but in reality most of these are probably hybrid factors."