(Photo by Nattakorn_Maneerat on Shutterstock)
In a nutshell
- Researchers identified specific metabolites in blood and urine that can accurately predict how much ultra-processed food (UPF) a person consumes, offering an objective alternative to self-reported dietary data.
- These “poly-metabolite scores” were validated in a clinical trial, where participants’ metabolic profiles shifted significantly when switching between diets high and void of UPF.
- The study found that higher UPF intake is associated with lower levels of beneficial nutrients and whole foods, and may increase exposure to potentially harmful compounds from food packaging.
ROCKVILLE, Md. — Your blood and urine reveal a lot about your diet, especially when it comes to those convenient, packaged foods that make up more than half of what many Americans eat. A new study from the National Cancer Institute reveals that researchers have identified chemical fingerprints left behind in our bodily fluids that can accurately predict how much ultra-processed food (UPF) someone consumes.
According to the research published in PLOS Medicine, ultra-processed foods account for a majority of calories consumed in the United States, yet their impact on human health remains unclear. The aim of this study was to identify biological markers in blood and urine that could reliably predict UPF intake.
America’s Ultra-Processed Diet
Americans currently get more than 50% of their daily calories from ultra-processed foods, those ready-to-eat or ready-to-heat industrial formulations made with ingredients rarely used in home kitchens, like high-fructose corn syrup, hydrogenated oils, and various additives for flavor and texture. Recent research has linked high UPF consumption to increased risks of obesity, heart disease, type 2 diabetes, and certain cancers.
The team analyzed blood and urine samples from 718 adults aged 50-74 who participated in the Interactive Diet and Activity Tracking in AARP (IDATA) study. Participants provided multiple 24-hour dietary recalls over 12 months, along with blood and urine samples collected six months apart.
The scientists measured over 1,000 different metabolites in each sample. They found that nearly 200 blood metabolites and almost 300 urine metabolites correlated significantly with UPF intake. These included compounds related to lipid metabolism, amino acids, carbohydrates, vitamins, and xenobiotics (foreign substances like food additives).
Measuring Ultra-Processed Food Consumption
The researchers then used statistical methods to develop “poly-metabolite scores,” combinations of specific metabolites that together could predict UPF consumption with remarkable accuracy. They validated these scores using data from a separate clinical trial where 20 participants were randomly assigned to eat either a diet consisting of 80% ultra-processed foods or one with no ultra-processed foods for two weeks before switching to the opposite diet.
The research team found that the scores changed substantially when individuals switched between high-UPF and no-UPF diets. This means that when the same person ate different diets, their metabolite patterns changed predictably based on how much ultra-processed food they consumed.
Some of the most telling markers of UPF consumption included a compound called N6-carboxymethyllysine, which has been linked to diabetes and heart disease. On the flip side, the researchers found lower levels of certain beneficial compounds in people who ate more UPFs, including one typically found in cruciferous vegetables like broccoli and Brussels sprouts.
One of the metabolites positively associated with UPF consumption was levoglucosan, a compound released during the combustion of cellulose. The researchers suggest this might come from food packaging materials, indicating that UPF consumers may be exposed to substances from packaging in addition to the food itself.
What This Reveals About Modern Diets
According to the researchers, these metabolite signatures reflect not just high UPF consumption but also typically lower intake of whole foods. Participants who consumed more ultra-processed foods had significantly lower intake of fiber, vitamins, and minerals compared to those who ate less UPF.
This doesn’t just point fingers at who eats more junk food. These biomarkers could change how we study diet and health by giving researchers a more accurate way to measure how much ultra-processed food people eat without relying on self-reporting, which is notoriously prone to error. This is especially important while we are experiencing the rapid transition to industrial food systems dominated by convenient, shelf-stable, highly palatable products engineered for maximum appeal rather than nutritional value.
Paper Summary
Methodology
Researchers conducted a two-part study to identify and validate metabolite markers of ultra-processed food consumption. First, they analyzed data from 718 participants in the Interactive Diet and Activity Tracking in AARP (IDATA) Study, who provided multiple 24-hour dietary recalls over 12 months as well as blood and urine samples collected 6 months apart. Using ultra-high performance liquid chromatography with tandem mass spectrometry, they measured over 1,000 metabolites in each sample. They then used statistical methods to identify correlations between metabolites and UPF intake (calculated as percentage of total energy) and developed predictive poly-metabolite scores using LASSO regression. In the second part, they validated these scores in a post-hoc analysis of a previously conducted randomized controlled crossover-feeding trial where 20 participants consumed diets with either 80% or 0% energy from UPF for two weeks each.
Results
The researchers found that 191 serum metabolites and 293 urine metabolites were significantly correlated with UPF intake. Using LASSO regression, they selected 28 serum and 33 24-hour urine metabolites as predictive markers and created poly-metabolite scores based on these. Key metabolites included (S)C(S)S-S-Methylcysteine sulfoxide, N2,N5-diacetylornithine, pentoic acid, and N6-carboxymethyllysine. When applied to the crossover-feeding trial data, these poly-metabolite scores successfully differentiated between the high-UPF and no-UPF diet phases within individual participants, confirming their validity as biomarkers of UPF consumption.
Limitations
The study population in the IDATA cohort consisted primarily of older US adults (aged 50-74), with 93% self-reporting as white/non-Hispanic, which may limit generalizability to more diverse populations. The researchers acknowledge that the participants’ diets and metabolic responses may not be representative of other demographic groups. However, the researchers suggest their findings likely apply more broadly, as the metabolite signatures successfully differentiated between diets in the validation trial, which included younger, more diverse participants. Additionally, dietary recalls could not be reliably matched with specific biospecimen measurements as recalls were, by design, unannounced. The crossover feeding trial used for validation had a small sample size (n=20), which may have contributed to some variation in the results.
Funding and Disclosures
The research was funded by the NIH Intramural Research Program at the National Cancer Institute (NCI) and the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). One of the authors, Lauren E. O’Connor, disclosed being the principal investigator on a grant administered by the USDA National Institute of Food and Agriculture to coordinate a workshop on food processing research, with collaborators including scientists from Archer Daniels Midland and General Mills. All other authors declared no competing interests.
Publication Information
The study titled “Identification and validation of poly-metabolite scores for diets high in ultra-processed food: An observational study and post-hoc randomized controlled crossover-feeding trial” was published in PLOS Medicine on May 20, 2025. The research was led by Leila Abar and Erikka Loftfield from the Division of Cancer Epidemiology and Genetics at the National Cancer Institute, along with collaborators from multiple institutions, including the University of São Paulo, Brazil.
