No more cheat days

When heading up clinical trials of diets, scientists consistently run into the same problem: How can they know whether participants are actually sticking to their regimen? "It's a major issue in nutritional epidemiology as well as clinical practice," says epidemiologist Casey Rebholz. "We're trying to build evidence-based research and provide patient-centered medicine, and we're not able to do that if we don't have accurate information about what people are eating." There are typically two ways of determining this, neither of which is foolproof. Option 1: Ask them. That means requiring people to remember what they ate, sometimes within the past day, sometimes within the past year, Rebholz explains. Moreover, participants who slip and fail to follow directions tend to lie about it. Option 2: Sample their urine. The problem there? Participants have to collect a sample every time they use the restroom over a period of at least 24 hours. Rebholz knows from experience what a hassle this can be; she was a study participant in grad school. "I would be carrying these big jugs of urine around," she remembers. "It's very inconvenient for someone to go about their usual day."

Context

There had to be a better way, thought Rebholz, an assistant professor in the Department of Epidemiology at the Bloomberg School of Public Health. She recalled a collection of frozen blood samples being stored at the NIH from the 1997 landmark study of the DASH diet. The DASH diet—Dietary Approaches to Stop Hypertension—was designed to lower blood pressure, emphasizing a diet rich in fruits, vegetables, and low-fat dairy. For the 1997 study, participants were given all meals, ensuring that they were eating according to plan.

Rebholz wondered whether metabolomics, the study of small molecules called metabolites, would show any meaningful differences between the DASH diet eaters and the control group. "The actual method is a little bit complicated—it's gas chromatography–mass spectrometry, as well as liquid chromatography–mass spectrometry—but essentially we're taking a blood specimen and identifying all the metabolites that are found in that blood sample." Rather than revealing levels of potassium, vitamin A, or protein, metabolites point toward specific food groups, such as fruit or legumes.

Data

Rebholz and a team of Bloomberg School researchers analyzed blood samples from 329 DASH trial participants. They identified 97 metabolites whose levels differed significantly between the DASH diet group and the control group, as well as 67 metabolites whose levels differed significantly between the DASH diet group and a group assigned to eat a diet rich in fruits and vegetables (but not dairy). For the DASH versus control comparison, the 10 metabolites that differed most significantly were N-methylproline (associated with citrus fruit and juice), stachydrine (citrus fruit), tryptophan betaine (lentils and legumes), theobromine (chocolate), 7-methylurate (caffeine), chiroinositol (fruit, beans, grains, nuts, and seeds), 3-methylxanthine (desserts), methyl glucopyranoside (fruit), ß-cryptoxanthin (red peppers, corn, and citrus), and 7-methylxanthine (desserts). Their findings were published in August's American Journal of Clinical Nutrition. "The reason this is important is because people don't eat foods in isolation," Rebholz says. "You eat a meal, you eat a variety of different foods, and they don't act in isolation. So how you eat foods and what dietary pattern you're following influences how you digest, how you metabolize, and how the components of those foods are available in the body."

Conclusion

In part, Rebholz conducted her study to highlight the benefits of the DASH diet. But she says metabolomics could eventually be used to assess whether patients are following their diets—in the context of not only a dietary trial but also a routine doctor's visit. In theory, this method could be applied to assess patients' adherence to a range of diets beyond DASH.

One of the main limitations, she says, is similar to that of the urine samples. While metabolite levels tend to fluctuate more often in urine than in blood, Rebholz says, "we're measuring blood at a single time, so we aren't able to see, are these levels stable over time? Do they change to a considerable degree as the diet changes?" That's a topic for future research, she says.