NIH researchers develop biomarker score for predicting diets high in ultra-processed foods
Research news: NIH investigators reported blood- and urine-based poly-metabolite scores that aim to objectively estimate the percentage of dietary energy from ultra-processed foods. Using observational data from 718 older U.S. adults plus a small randomized crossover feeding trial (n=20), the scores differentiated high (80% energy) vs none (0%) ultra-processed diet phases. Authors note the need for replication in other populations and further study of associations with outcomes including cancer.
Type: Research news (not a regulatory approval)
NIH researchers (including the National Cancer Institute) reported metabolomics-derived biomarker scores intended to objectively measure dietary intake of ultra-processed foods, potentially complementing or reducing reliance on self-reported dietary questionnaires in large population studies. The findings were published May 20, 2025, in PLOS Medicine.
Ultra-processed foods were described as ready-to-eat or ready-to-heat, industrially manufactured products that are typically high in calories and low in essential nutrients; such diets have been linked to obesity and other chronic diseases, including some cancers. Because self-reported dietary assessment can be prone to reporting differences and may not capture changes in the food supply over time, the investigators sought objective measures.
The study integrated:
- Observational data: 718 older U.S. adults with biospecimens and dietary information over 12 months.
- Experimental data: a small NIH Clinical Center randomized crossover-feeding trial in 20 adults, who consumed a diet high in ultra-processed foods (80% of energy) and a diet with 0% ultra-processed foods for two weeks each.
Researchers identified hundreds of blood and urine metabolites correlated with the percentage of energy from ultra-processed foods. Machine learning was used to derive separate blood and urine poly-metabolite scores, which in additional testing could differentiate within individuals between the trial’s highly processed vs unprocessed diet phases.
The authors cautioned that findings may not generalize beyond older U.S. adults and recommended replication and refinement in other populations, including evaluating associations with risks of diseases such as cancer and type 2 diabetes.
Reference: Abar L et al. PLoS Med (2025). DOI: 10.1371/journal.pmed.1004560.