Published: June 22, 2011
Updated: June 22, 2011
Two men walk into their doctor’s waiting room, and with them arrives the puzzle that plagues so many physicians.
The men are of similar build, age, and ethnic background. They have the same general diet and lifestyle habits. However, in two years, one of these men will die of a heart attack. The other will live for many years with no heart disease.
How does the doctor know which man is at immediate risk?
Two of our most intractable diseases -- heart disease and diabetes -- are widely accepted as having a “multifactorial” origin in most patients.
They can’t be explained by genetics alone, and the mishmash of contributing behavioral causes are not only maddeningly variable from patient to patient but also nearly impossible to measure accurately over the decades it takes for their impact to play out.
So far, the best laboratory tests still cannot identify with any reliability just who will get sick from these illnesses and who won’t.
But what if they could?
This is the goal of metabolomics: a relatively new field of studying the chemicals produced by the many metabolic processes in the body. “Your body is in a constant state of metabolism,” explains Duke cardiologist Svati Shah, MD.
“These are the processes that help regulate the sugars, protein, and fat that you eat, and the conversion of these fuels to energy.” The goal of metabolomics is to try to measure the byproducts of these processes and use those measurements as biomarkers for the health -- or illness -- of the body.
Metabolomics serves as the integrated readout for the other “-omics” sciences, such as genomics or proteomics, says Duke researcher Chris Newgard, PhD. “Genomics and proteomics have been big areas of research,” he says.
“But the metabolites are at the end of the funnel: mRNA is the product of the genes, the protein is the product of the mRNA, and the metabolic signature is the integrated readout of how all the individual genomic and proteomic variations affect a person’s physiology. So to me it is the most precise measurement of the phenotype of the individual.”
He skips a beat and says, “Of course, I’m biased.”
But the results of studies published in the last two years have so far backed his claims with hard data, showing that certain clusters of metabolites may be specific and reliable indicators of heart disease and impending diabetes at the individual level.
In fact, their levels in the bloodstream may even predict which heart disease patients will soon have a heart attack.
Measuring metabolites is hard to do, scientifically speaking. They are tiny and exist in the bloodstream in very low concentrations -- micromolar and nanomolar amounts. There are also lots of them -- an estimated 6,500 discrete metabolites in humans. And when you’re measuring anything in the blood, it’s like trying to categorize multicolored sand.
“Your blood is full of the coffee that you drank this morning, the cheeseburger you ate last night, the medicine you take,” says Shah. “So trying to isolate these metabolites among all the other molecules in the blood is a little like being in Times Square and trying to pick out one person.”
The techniques used to measure metabolites have taken years to develop. Newgard, who directs the Sarah W. Stedman Nutrition and Metabolism Center, came to Duke in 2002, after 15 years of working with metabolic technology at University of Texas Southwestern in Dallas.
Once at Duke he worked with David Millington, PhD, and Robert Stevens, PhD, both specialists in inborn errors of metabolism, and other key members of the Stedman Center team, including James Bain, PhD, Brett Wenner, PhD, Michael Muehlbauer, MD, PhD, and Olga Ilkayeva, PhD, to build what is today one of the world’s most sophisticated metabolic labs.
The Stedman laboratory collaborates with researchers from all over the world (including colleagues at Duke-NUS), who know the group for providing a level of data specificity that is not available in most labs.
Metabolomics researchers use mass spectrometry to identify and measure a wide array of small-molecule metabolites. “A lot of labs will do a mass spec analysis and get a pattern that shows metabolite levels, then compare the patterns among a group of patients,” Shah says.
But the Stedman lab goes further, adding standards to provide a truly quantitative measure of a sample’s components. So instead of patterns, what comes out of the Stedman lab are numbers -- measurements of metabolites in their exact amounts.
“The Stedman lab can tell you what’s in your blood, and exactly how much of it there is,” says Shah.
“Many metabolomics labs are built by either instrument jockeys or statisticians,” says Newgard. “We take the perspective of the biochemist, the molecular biologist, and the physiologist. To me it’s important to know what I’m measuring and in what concentrations. Our lab has been built so that we can say here’s exactly what’s in the sample -- so that we can also eventually say this is the significance of it, and this is what it portends.”
The data generated by the Stedman lab are beginning to paint very clear profiles of certain high-risk metabolic biomarkers for both heart disease and diabetes.
Part of the group’s success involves a home-field advantage: Duke’s CATHGEN biorepository, which holds health records and blood samples from nearly 10,000 patients who have come to Duke over the past eight years for heart catheterization.
This wealth of samples allowed Newgard and Shah to develop a first-of-its-kind investigation, which showed that the levels of certain clusters of metabolites could predict imminent cardiovascular events, including heart attack and even death.
Through a series of studies, the team has shown that metabolic profiles are heritable -- more heritable than other indicators of heart disease such as BMI, cholesterol, and C-reactive protein.
They compared the profiles of 174 patients who were diagnosed with heart disease and went on to have a cardiac event in the next two years with 174 other heart disease patients who were as closely matched as possible in terms of physical and demographic history, but who had no events over a 10-year period.
Among those groups, a specific metabolite cluster was elevated among patients who had heart events. Then they looked at a group of 2,000 patients who came to Duke for concerns about heart disease, and who have been followed every year regardless of diagnosis.
“We did analysis on the full 2,000, and the exact same metabolites were present in people who had heart events,” says Shah. “I was totally dumbfounded. I really didn’t think it would validate.”
“There was some element of gamble,” says Newgard. “It was a sort of dream, that we would set this toolbox up and it would reveal metabolic patterns that hadn’t been characterized and that were so telling. We couldn’t know whether the methods would be precise enough, or whether we’d be swamped out by human variance -- mood, behavior, all the things that go into being human, would those things make it impossible to see the chemistry of disease?”
But cardiologist William Kraus, MD, a co-investigator on these trials and a co-founder of CATHGEN, says that the experiments were hardly a shot in the dark. “We specifically selected clusters of metabolites that we know are involved in multiple pathways of lipid, protein, and glucose metabolism -- pathways that are often disrupted in heart disease -- and we found that they are indeed associated with heart disease and subsequent risk of cardiac events,” Kraus says.
“These metabolic profiles may be a long way from routine clinical use, but we feel they are a good first step in that direction.”
Newgard and Laura Svetkey, MD, director of clinical research at the Stedman Center, had previously identified a different discrete cluster of metabolites, dominated by the branched-chain amino acids (BCAA), as a player in insulin resistance.
“The data have shown very clearly that the higher the level of this factor, the higher the level of insulin resistance in the patient -- even after adjusting for factors including weight,” says Svetkey.
When the team looked at whether insulin resistance improved with weight loss, she says, it showed that the higher the BCAA factor in a patient, the more that patient’s insulin resistance improved with weight loss -- again, even after adjusting for weight.
Kraus says a forthcoming study shows that metabolites can help predict who will most benefit from exercise training, in terms of reducing insulin resistance.
The factors controlling levels of these metabolites in the body are downright mysterious: a recent study that Svetkey, Shah, and Newgard participated in showed that the levels of this cluster dropped significantly after gastric bypass surgery -- more so than after weight loss induced by dietary intervention.
“We don’t know some very fundamental things that will help us interpret our results,” Svetkey says. “For example, BCAA comes from meat, and we don’t know how meat consumption or the duration of fasting before we do the blood tests will affect the metabolomics profiling results. We also still need to understand the extent to which this factor differs by age, sex, race, and so forth. Some of these questions can be addressed with the data we already have, and some will require new research.”
And finding a reliable pattern isn’t enough, says Newgard. “When we see those signatures, we want to understand what they do mechanistically. What does this mean at the level of the cells, at the level of the pathways, and can we do anything to change these patterns,” he says. Newgard is continuing research in animal models to see how changing the diets of rats may affect their levels of BCAA metabolites -- and also their clinical outcomes.
Shah is working to identify the genetic makeup that may put people at risk for the metabolomic profiles that presage insulin resistance, type 2 diabetes, and cardiovascular disease.
“The ultimate utility of this kind of investigation remains to be seen,” says Svetkey. An analogy, she says, is the genomics revolution: while genetic discoveries may have had significant and direct impact on several diseases, none have yet made a dent in any of the major public health problems of our day -- hypertension, obesity, heart disease, diabetes.
“Chris would argue, and I’d agree, that metabolomics approach is more likely than the other ‘omics’ to lead to clinical impact. Because while genes are static, metabolites are mutable -- they can be altered by drugs, by behaviors. It’s something we may be able to affect, to use to help our patients.”