Blood and Urine Biomarkers Reveal How Much Ultra-Processed Food We Eat
New research introduces poly-metabolite scores in blood and urine to measure ultra-processed food intake objectively.

Complete the form below to unlock access to ALL audio articles.
Measuring how much ultra-processed food (UPF) someone eats has always been challenging for nutrition researchers, often due to self-reporting bias.
Now, a new study from the US National Cancer Institute shows that blood and urine can reveal a person’s UPF intake using specific combinations of metabolites.
Published in PLOS Medicine, these “poly-metabolite scores” may offer a reliable alternative to self-reported diet data.
Why it’s hard to study UPF intake
UPFs make up over half the calories in the average American diet. These are products like packaged snacks, soft drinks and ready-to-eat meals – industrially formulated and often high in added sugars, fats and additives. A growing body of research has linked high UPF consumption with obesity, type 2 diabetes, heart disease and certain cancers.
However, studying UPFs concerning health is difficult. Most research relies on self-reported diet data, which can be patchy or inaccurate. People forget what they ate or misjudge portion sizes, making it hard to draw solid conclusions about how UPFs affect the body.
There is also no widely accepted objective way to measure UPF intake, such as an established biomarker, to give researchers a clearer picture.
“Global production and availability of UPF is high, but accurately measuring UPF consumption is challenging,” said the study authors.
Previous studies have used metabolomics, measuring small molecules in blood or urine, to reflect general dietary patterns. However, only a few have focused on UPFs and have been small or limited in scope.
The researchers set out to develop and test “poly-metabolite scores” – combinations of metabolites in blood or urine that reflect how much energy a person gets from UPFs. The goal was to find reliable, objective biomarkers that could be used in large studies to complement or reduce reliance on self-reported diet data.
Building a new biomarker score
The study was split into two parts: a large observational study and a smaller, tightly controlled feeding trial. The observational study used data from 718 older adults who took part in the AARP Diet and Health Study. They provided multiple 24-hour dietary recalls over 12 months, as well as blood and urine samples.
The second part of the study involved 20 healthy volunteers at the National Institutes of Health Clinical Center. Participants followed 2 diets, each for 2 weeks: the first obtained 80% of energy from UPFs, and the second with none, while living under observation. The two groups then switched diets. Blood and urine were collected at two time points, six months apart, helping account for day-to-day variation in diet and metabolism.
To measure UPF intake, researchers calculated the percentage of calories from UPFs using the NOVA classification system. At the same time, they carried out untargeted metabolomics profiling on all biospecimens. They looked at over 1,000 small molecules in blood and urine without pre-selecting targets.
They found over 200 metabolites that were significantly linked to UPF intake.
From this, the team built two “poly-metabolite scores”. The first was based on 28 blood metabolites, and the other used 33 metabolites from urine. These scores could estimate UPF consumption without needing self-reported food data.
In the observational group, the scores tracked well with what participants reported eating. In the feeding study, they clearly distinguished between the two diets, even within individuals.
The team used cross-validation to avoid overfitting and confirmed the findings across different types of samples and settings.
Overfitting
Overfitting is when a statistical model fits the training data too closely, capturing noise instead of the underlying pattern. As a result, the model performs well on that data but poorly on new, unseen data.
Objective UPF biomarkers could change nutrition research
The ability to measure how much UPF people actually eat, without asking them to fill out food diaries or remember what they ate, is a big step for nutrition research. These blood- and urine-based poly-metabolite scores could improve the accuracy of studies linking diet and disease and help move the field beyond self-reporting.
The scores could also help clarify how UPFs affect long-term health, including better tracking of diet-related disease risks and potentially tailoring nutrition advice to individuals.
“The identified poly-metabolite scores could serve as objective measures of UPF intake in large population studies to complement or reduce reliance on self-reported dietary data,” said the authors.
However, the participants in this study were mostly older US adults, so it’s not clear how well these scores would work in more diverse populations.
“Poly-metabolite scores should be evaluated and iteratively improved in populations with diverse diets and a wide range of UPF intake,” the authors added.
The researchers next steps will be to test and refine these scores in broader groups and see if the metabolite profiles also link to specific health outcomes, such as diabetes or cardiovascular disease.
As the team stated, these scores “could provide novel insight into the role of UPF in human health.”
Reference: Abar L, Steele EM, Lee SK, et al. 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. PLOS Medicine. 2025. doi: 10.1371/journal.pmed.1004560
This article is a rework of a press release issued by PLOS. Material has been edited for length and content.