Tracking diet from stool samples

Accurately tracking the foods people eat has been a serious challenge in nutrition research. Rawpixel.com / Adobe Stock

Diet influences many aspects of human health. Yet tracking dietary intake remains challenging. Typical approaches include questionnaires and food diaries. Questionnaires are limited by participants’ ability to accurately remember their food intake. Food diaries provide more detailed information but require more effort from participants.

To overcome these limitations, a team of researchers, led by Drs. Christian Diener and Sean Gibbons at the Institute for Systems Biology, developed a method to detect and measure food-derived DNA in human stool samples. This can be used to estimate a person’s dietary intake. Nutrient intake can then be estimated based on the nutrient content of the foods found. A description of the method, called Metagenomic Estimation of Dietary Intake (MEDI), and its use appeared in Nature Metabolism on February 18, 2025.

The team began by constructing a database of full or partial genomes for more than 400 organisms that people eat. These were linked to nutritional profiles for different food items and preparation types. Next, the team developed an algorithm to match DNA from stool samples to its source food in the database. Only a tiny fraction of the DNA in stool comes from food. Most comes from the person or their gut microbes. But the team demonstrated that MEDI could accurately detect and measure diet-derived DNA sequences, even when they made up as little as 0.001% of the total DNA.

To test whether MEDI results accurately reflected food intake patterns, the researchers analyzed genome sequences in stool samples from two controlled feeding studies. Participants in the studies were separated into two groups, with each being fed a controlled diet. In both studies, MEDI accurately reflected the differences in dietary intake between the groups. For one of the studies, participants also kept detailed daily food diaries, allowing estimates of detailed nutrient intake. MEDI estimates agreed with the food diary estimates for calorie, protein, carbohydrate, potassium, cholesterol, and vitamin B₁₂ intake.

In a dataset of infant stool samples, MEDI identified a steady increase in the prevalence of food-derived DNA beginning around 160 days of age, corresponding to the start of solid food intake. In adult samples, MEDI estimates of diet composition agreed with those obtained from food frequency questionnaires.

The team also used MEDI to identify dietary patterns associated with metabolic syndrome in study participants without dietary records. Metabolic syndrome is a group of conditions, including high blood pressure and high blood sugar, that increase the risk of coronary heart disease, diabetes, and stroke. Consistent with earlier findings, the team found that people with metabolic syndrome had higher intake of animal-based foods and lower intake of plant-based foods than those without metabolic syndrome. Metabolic syndrome was also associated with higher intake of specific nutrients, including lactose, cholesterol, and certain fatty acids.

“For decades, nutrition research has depended on self-reported diaries and questionnaires—approaches that require a high degree of effort and compliance from research participants,” Diener says. “How many strawberries did I eat two days ago? Did I have one glass of orange juice with breakfast, or two?”

With further development, food-derived DNA from stool samples could greatly improve how diets are tracked to improve the accuracy of nutrition studies. But MEDI can’t detect items that lack DNA, such as coffee and many other beverages, so it may still need to be supplemented with other approaches.

—by Brian Doctrow, Ph.D.