Issue Date: December 10, 2007
EAT A BALANCED diet, and you'll live a longer, healthier life. That refrain sounds simple enough.
The trouble is that the definition of a balanced diet is fuzzy. Does it mean the same thing for a 70-year-old woman and a 10-year-old boy? Or for an Olympic runner and a world-class couch potato?
"Would that it were so simple," sighs J. Bruce German, professor of chemistry and food science at the University of California, Davis. "If you're healthy, and I give you a food that lowers the risk of one thing, but in so doing it increases the risk of anything else, you may not be healthier at all. We're increasingly learning more about diet and health, enough to know that what makes some people better makes other people worse."
German and others believe that context-dependent nutrition might get a boost from metabolomics, the interdisciplinary study of metabolites, which are small molecules connected by the complex web of biochemical reactions called metabolism (C&EN, Dec. 2, 2002, page 66).
Metabolomics is a maturing science. Researchers are making decisions at every step: which biological samples to examine, how to obtain and analyze data, and most important, how to reliably link data output to the biological reality of how foods affect metabolic pathways. Nutrition experts are eyeing metabolomics as a potential tool for formulating diet guidelines. The field isn't ready for that yet, but its momentum is building toward making that possible.
Ultimately, metabolomics aims to make sense of the relationships between diet and health and then to transfer that understanding to caregivers who are making decisions for patients.
For example, the field might update the blood test given at a yearly checkup. Rather than just recording current numbers, this test would detect shifts in a healthy person's metabolic balance that predict potential future health problems. From there, a physician could intervene with a customized diet designed to work with that patient's metabolism. A follow-up blood test would measure biomarkers indicating whether the diet was helping the patient return to a healthy trajectory.
That kind of tailor-made nutrition requires a move beyond the traditional concept of using single biomarkers as health indicators. Think cholesterol for heart disease, or glucose for diabetes. In nutrition, "you aren't dealing with a target, but with an organization," says Jan van der Greef, director of personal health and systems biology at the Netherlands Organization for Applied Scientific Research. A person's diet is a complicated thing, a smorgasbord of cultural influences, ingrained habits, and personal preferences.
Understanding the effects of diet is harder than understanding what happens when someone takes a drug, an area in which metabolomics is a more established tool. "A drug has a limited number of active molecules, but foods have thousands of molecules that work together and have subtle effects," says Serge Rezzi, a senior scientist at Nestl?? Research Center (NRC).
Despite the challenges, German is enthusiastic about the potential of metabolomics in nutrition and health. He serves on the scientific advisory board of Lipomics, a company that already measures lipid harbingers of disease for clients. For him, now is "the most exciting time in nutrition since the early-20th century, when the essential nutrients were being discovered."
IF OTHER WORDS ending in "omics" sound familiar in the context of personalized health and nutrition, it's because genomics and proteomics are also tackling dietary issues. It could also be because there is some debate about what all-encompassing metabolite analyses should be called. By definition, metabolomics measures and catalogs thousands of small molecules present in fluids such as blood and urine under various conditions. The field's professional society and its eponymous journal adopt this term.
Another term, "metabonomics," coined by Jeremy K. Nicholson, professor of biomolecular medicine at Imperial College London, describes an alternative way of thinking about metabolic complexity. This other philosophy starts with the big picture of metabolism's multipronged responses to stress, dietary change, or some other life event. Metabonomics denotes this more top-down, biological approach and complements metabolomics' more bottom-up, analytical definition, according to Nicholson. For now, settling the terminology question is taking a backseat to advancing the science.
Genomics and proteomics have yet to deliver useful guidance on personalized nutrition, and it isn't guaranteed that metabolomics will prove more useful. However, researchers can think of several reasons why the metabolome, the body's sum total of metabolites, will be the key to demystifying diet.
For one thing, metabolites are a record of chemical processes that have already happened. This makes them better biomarkers than entities further upstream in the molecular biology of metabolism and that therefore require more inferential thinking to make sense of, according to Sunil Kochhar, manager of the metabonomics and biomarkers group at NRC.
In addition, the functions of many genes and proteins are unknown, while a host of metabolic pathways have been mapped out in undergraduate textbooks for decades. And the metabolome is highly sensitive to environmental changes, a trait that might facilitate detection of foods' subtle effects.
"If I asked someone to hold their breath for a while, and we were monitoring their genome or proteome, by those measurements we would think nothing happened. But if we took a look at their metabolome, we'd see all kinds of wild changes," says David S. Wishart, professor of biological science and computer science at the University of Alberta, Edmonton.
Sources emphasize that metabolomics alone cannot unravel the complexities of diet and the human organism. "Metabolomics research recruits information from other disciplines, and that's what gives it power," German says. For instance, by integrating metabonomics and genomics in mice, Nicholson's team recently learned more about how gut-dwelling microbes help absorb and store the energy harvested from the diet (Mol. Syst. Biol. 2007, 3, 112).
Robert N. McBurney, chief scientific officer of BG Medicine, a company that discovers biomarkers and was cofounded by van der Greef, is also a proponent of integrative science. He likens approaches integrating genes, proteins, and metabolites to a key indicator in the world of finance. "We measure the health of the stock market with the Dow Jones industrial index, a sampling of markers from across very different industries," he says.
The use of metabolomics to study nutrition is generating growing interest not only in academia but also in industry. Metabolomics is increasingly recognized as a key technology for nutrition at meetings in diverse fields, says Bruce S. Kristal of Brigham & Women's Hospital and secretary of the Metabolomics Society. For instance, the 2nd North American Congress of Epidemiology and the 2006 International Research Conference on Food, Nutrition & Cancer included metabolomics studies.
AND COMPANIES such as Nestlé have active programs in metabonomics research, both to improve the nutrient content of their food products and to gain a better understanding of how their products affect their consumers. A recent Nestl?? study that shed light on how calorie restriction extends dogs' life spans (J. Proteome Res. 2007, 6, 1846) was the most-accessed article for the Journal of Proteome Research between the months of April and June 2007.
Metabolomics technology takes routine biological fluid analysis to a more comprehensive level, which helps explain why it's of interest in basic, applied, and clinical research. An experiment usually begins with a noninvasive sample such as blood or urine. The next step is to detect and chemically characterize as many metabolites as possible.
Much like solving a protein X-ray crystal structure, collecting data isn't the last step. Somehow the numbers need to be linked back to the biology. However, there is a divergence in the community about this part of the process. Some researchers believe the best way to learn from the metabolism data is to let statistical methods uncover which metabolites are most significant for answering the question at hand. Others believe measuring every metabolite in the sample will generate the knowledge base that's needed to move forward.
STATISTICS ADVOCATES argue that statistical methods make the otherwise daunting task of discerning food-metabolism interactions more tractable. "Purely quantitative approaches are efficient once you have a good hypothesis and there's a metabolic pathway you know is involved," Nicholson says. His group has merged statistics and nuclear magnetic resonance (NMR) in a technique called statistical total correlation spectroscopy (STOCSY). STOCSY is useful for determining relationships between metabolites as well as for structural assignment of individual molecules in complex mixtures.
STOCSY's closest relative is a two-dimensional NMR technique called total correlation spectroscopy (TOCSY). TOCSY requires communication between nuclear spins in order to detect an association between two protons, but STOCSY works in a purely statistical way. When concentrations of one molecule differ from one NMR sample to the next, all the NMR signals from that molecule will increase or decrease in the same ratio. In a collection of hundreds of highly complex samples, such as blood or urine, all of the different metabolites will likely change by different amounts. STOCSY exploits this concept to partition NMR spectra of complex mixtures into sets of peaks, each corresponding to one molecule.
STOCSY analysis has already proven useful in studies of metabolism and diet. In one such example, Rezzi, Kochhar, and their Nestlé colleagues collaborated with Nicholson's group and uncovered intrinsic differences in metabolic profiles of self-proclaimed chocolate lovers and chocolate haters (J. Proteome Res. 2007, 6, 4469). The researchers found that varying responses to "a very abstract marker of behavior, a questionnaire, can translate into a significantly different metabolic profile," Nicholson says.
One variation the team found was in a special class of aromatic metabolites. For example, more 4-hydroxyphenyl acetate seemed to be converting to 4-cresol sulfate in chocolate haters than in chocolate lovers. STOCSY showed that the fluctuations of these two metabolites were interrelated, which allowed the team to infer that a conversion was happening. That process isn't part of any known human metabolic pathway, but it exists in the microbe Clostridium difficile. Because microbes such as C. difficile inhabit every person's gut, the result implies that chocolate haters have more C. difficile in residence. Microbes' influence on how foods affect people differently isn't yet clear, and more tests are necessary to confirm the ideas the study generated. However, Nicholson says, the study validates metabonomics and STOCSY for studying diet preferences.
But some researchers contend that statistics can take the field only so far and that metabolites must be quantified to build databases on which the field will move forward. German gives the example of cholesterol testing to illustrate that point of view. Just being able to compare cholesterol between people wasn't useful in clinical settings in the absence of a frame of reference for the data, he says. The key was to measure levels in many different people to figure out the normal range for cholesterol. Only then could physicians begin to associate numbers higher than that range with heart disease risk and recommend diet adjustments.
Parallel efforts are in place to compile that kind of data for the entire metabolome. One such inventory, a Canadian initiative known as the Human Metabolome Database, is freely available and contains links to other public resources. According to Wishart, the project's leader, it combines the project's own experimental data and information gained by text-mining decades worth of literature. Another effort, the U.K.-based Human Serum Metabolome Project (HUSERMET), is collecting samples from over 5,000 individuals and analyzing these by mass spectrometry and NMR. The goal is to establish normal ranges for a wide range of metabolite classes in an effort to discover genuine biomarkers of health and disease. Such biomarkers could help direct nutrition guidance when a patient is at risk of developing a specific condition.
Warwick B. Dunn, an experimental officer at the Manchester Centre for Integrative Systems Biology in England and who is involved in HUSERMET, cautions that the 5,000 individuals do not reflect the world's diversity. But he maintains that these types of projects are a key first step. The effort to inventory the human metabolome is like assembling "a big jigsaw puzzle, and we are slowly putting the metabolomics pieces together" he says.
REGARDLESS OF HOW researchers analyze data, however, it's agreed that metabolomics is only one step to understanding how nutritional and dietary effects can have so much variation among individuals. Multiple sources told C&EN that metabolism research is a hypothesis-generating technology. In other words, the results should be treated—as they were in the chocolate study—as pointers that should be tested extensively. According to John C. Lindon, a professor of chemistry at Imperial College and longtime collaborator of Nicholson's, the follow-up work could be anything from genomics to classical biochemistry, but the point of the exercise should be verifying hypotheses.
In fact, making sure that metabolomics experiments withstand scrutiny is a major objective for the field. "There is a need for rigorous proof that our work is valid," says Ian D. Wilson, senior principal scientist at AstraZeneca Pharmaceuticals, which has an active metabonomics program. The validation effort began with the 2003 and 2004 meetings of the Standard Metabolic Reporting Structures working group (Nature Biotech. 2005, 23, 833). The Metabolomics Society's Metabolomics Standards Initiative continues to work on that group's recommendations and is further refining guidelines for sound data reporting and analysis (OMICS 2006, 10, 158).
Metabolomics is not yet a framework for assessing diet. Although nutrition experts are eager to adopt its technology, they are waiting for further advances in standardization and data interpretation.
Metabolomics has potential as a chemical yardstick for personalized nutrition guidelines, but it's still a developing field, says Joanne R. Lupton, president of the American Society for Nutrition and a member of the U.S. Department of Agriculture's 2005 Dietary Guidelines Committee. "We should invest in this science and the infrastructure underlying it while keeping the basic principle in mind—not all people respond to diet in the same way."
- Chemical & Engineering News
- ISSN 0009-2347
- Copyright © American Chemical Society