Blood and urine tests reveal how much ultra-processed food you really eat, study finds

Scientists have identified unique metabolic fingerprints in blood and urine that can objectively track ultra-processed food intake, paving the way for more accurate diet studies and new tools for public health.

Study: 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. Image Credit: Rimma Bondarenko / Shutterstock

With ultra-processed foods (UPFs) now accounting for over 50% of daily calories for many Americans, researchers are searching for biological clues in blood and urine that can objectively measure UPF intake. In a recent study published in the journal PLOS Medicine, researchers from the United States (U.S.) and Brazil aimed to identify specific patterns of metabolites—tiny molecules in the body—that serve as reliable markers of UPF consumption.

The study focused on an older, predominantly White U.S. population, which may limit the generalizability of its findings to other demographic groups.

Background

Ultra-processed foods, such as packaged snacks, sugary drinks, and ready-to-eat meals, are widely consumed worldwide, especially in the U.S. These foods are mainly made from refined ingredients and additives and are linked to a growing number of health problems, including obesity, type 2 diabetes, heart disease, and certain cancers.

While dietary surveys help track eating habits, they often rely on memory and self-reporting, which can introduce inaccuracies. Furthermore, the Nova classification system, which categorizes foods by their level of processing, requires detailed data that is not always available in food tracking tools. As a result, assessing UPF consumption reliably has become challenging in large-scale studies.

Scientists are now turning to metabolomics, the study of metabolites or the small molecules present in blood and urine due to metabolic processes, to find biomarkers that reflect actual dietary intake.

The current study

To identify biological markers linked to UPF intake, the researchers used data from the IDATA Study, which followed 1,082 adults aged 50 to 74 years. Of these, 718 participants had both dietary and biological data and were included in the metabolomics analysis.

The participants were required to complete up to six web-based 24-hour dietary recalls over a year. Foods were classified according to the Nova system, which sorts food items into four groups based on their level of processing. The researchers then calculated the percentage of total energy intake from UPFs.

Blood samples and two types of urine samples (24-hour and first morning void) were collected at two different time points, six months apart. The samples underwent metabolomics analysis using advanced mass spectrometry techniques to identify over 1,000 compounds from various chemical groups such as lipids, amino acids, carbohydrates, and vitamins.

Statistical analyses were performed to determine correlations between UPF intake and the metabolites, adjusting for factors like age, sex, race, body mass index (BMI), and smoking status. The researchers then used statistical tools to identify a combination of metabolites (poly-metabolite scores) that could best predict UPF intake. These scores were built separately for blood, 24-hour urine, and first-morning urine samples.

To test whether these scores were reliable, the team then used data from a previous crossover feeding trial. In this controlled setting, 20 adults consumed diets with either 80% or 0% UPF for two weeks each. Blood and urine samples from this trial were analyzed using the same methods, allowing researchers to compare how the poly-metabolite scores changed between diets. This helped confirm whether the scores could detect actual dietary differences in real time.

It should be noted that dietary recalls and biospecimen collection were not always precisely paired in time, which may affect the interpretation of some results. Additionally, the feeding trial sample size was small, and the study was not powered for disease outcomes.

Key findings

The study found that blood and urine contain distinct metabolic signatures that reflect a person's UPF consumption. The researchers identified hundreds of compounds that were either more or less abundant depending on UPF intake. Using this information, they developed poly-metabolite scores—combinations of selected metabolites—that successfully predicted an individual’s UPF consumption.

Specifically, 191 metabolites in blood and 293 in 24-hour urine showed strong correlations with UPF intake. These included amino acids, lipids, carbohydrates, and compounds from food additives or packaging materials. Notably, four metabolites were robust indicators across both blood and urine: (S)C(S)S-S-methylcysteine sulfoxide, N2,N5-diacetylornithine, pentoic acid (all negatively associated with UPF), and N6-carboxymethyllysine (positively associated). The latter, which is linked to advanced glycation end-products, is also associated with diabetes and heart disease.

The identified metabolic signatures reflect not only high UPF intake but also a lower intake of whole foods, including fruits and vegetables. For example, lower β-cryptoxanthin, a marker of fruit and vegetable consumption, was observed in those with higher UPF intake.

When tested in a controlled trial where participants consumed high- and no-UPF diets, these poly-metabolite scores reliably distinguished between diet phases within individuals. This showed that the scores worked not just in observational settings but also under tightly regulated conditions. The discrimination ability of these scores was moderate (AUCs of 0.66–0.78), suggesting further refinement is needed for use in diverse populations.

Furthermore, the results confirmed that high UPF intake is associated with lower levels of beneficial compounds like β-cryptoxanthin, found in fruits and vegetables, and higher levels of certain xenobiotics — foreign substances likely from food additives or packaging. This supported the idea that UPF-heavy diets may lack essential nutrients while introducing potentially harmful compounds.

The metabolic pathways associated with UPF intake included not only xenobiotic metabolism, but also amino acid, lipid, carbohydrate, and energy metabolism, highlighting the broad biological impact of dietary patterns high in UPFs.

Conclusions

In summary, the study provided strong evidence that blood and urine metabolites can serve as objective indicators of UPF intake. The researchers developed and validated poly-metabolite scores for blood and urine samples, and the scores consistently reflected habitual dietary patterns.

The findings offer a promising tool for objectively measuring UPF intake in future health studies. However, the scores need further evaluation in more diverse and younger populations, and their predictive value for long-term health outcomes remains to be established. While more work is needed to adapt these scores to diverse populations, the results mark an important step toward better tools for nutrition research and public health monitoring.