IPM Take
The new EPIC study does not prove that ultra-processed food directly caused the observed blood changes, and it does not justify a commercial “UPF blood test”. What it does is make the debate harder to dismiss. Higher intake was associated with a reproducible pattern across lipid and energy-related metabolites, even after adjustment for multiple dietary and lifestyle factors. The political implication is uncomfortable: if industrial food exposure is biologically traceable, governments cannot keep treating diet-related disease as a private failure of discipline. The response must include healthier food environments, better product data, affordability measures, marketing rules and independent research, while avoiding simplistic claims that every processed product carries the same risk.
Executive Summary
A peer-reviewed study used targeted metabolomics within the European Prospective Investigation into Cancer and Nutrition, or EPIC, to examine whether ultra-processed food consumption was associated with identifiable patterns in blood. The researchers analysed 129 endogenous metabolites in 6,177 participants and plasma fatty-acid data in 9,029 participants.
Higher ultra-processed food intake was associated with 22 circulating metabolites and eight fatty acids. The pattern included higher levels of some long-chain acylcarnitines, stearic acid, several omega-6 fatty acids and the industrial trans fatty acid elaidic acid, alongside lower levels of several phosphatidylcholines, sphingomyelins, docosahexaenoic acid and an odd-chain saturated fatty acid. The authors interpreted the findings as evidence of a distinct metabolic and fatty-acid signature linked to ultra-processed food intake.
The study is cross-sectional and based on one blood measurement and dietary data collected at EPIC baseline. It therefore cannot establish that ultra-processed food caused the metabolic differences. Food classification also involved uncertainty because historical questionnaires did not contain brand-level processing details. The findings need replication, longitudinal testing and controlled intervention studies before they can inform clinical biomarkers or regulatory thresholds.
The policy signal is nevertheless important. Nutrition rules still focus mainly on individual nutrients and consumer choice. Metabolomics is beginning to test whether the wider industrial formulation and food matrix leave biological patterns that nutrient labels alone may not capture. Governments should not rush to a single UPF score, but they should stop pretending that education alone can correct a food system shaped by price, convenience, aggressive marketing and unequal access to minimally processed food.
Why it matters
- Policymakers and public authorities: The study supports stronger surveillance of ultra-processed food exposure and a policy approach that combines nutrient profiling, degree of processing, affordability, procurement and marketing regulation.
- Clinicians and public health leaders: The findings add mechanistic plausibility to existing epidemiological evidence, but they do not support clinical screening or individual diagnosis based on metabolomic signatures.
- Researchers and academia: External validation, repeated blood measures, prospective analyses and controlled feeding studies are needed to determine whether the identified signature predicts future disease or changes after dietary intervention.
- Industry and innovation partners: Reformulation claims will face greater scrutiny if products remain metabolically distinguishable despite improved front-of-pack nutrient profiles. Transparent ingredient, additive and processing data will become increasingly important.
- Patients and advocates: The message should not be that consumers have failed. Healthier choices require time, money, access, storage, transport and food environments that do not make the least healthy option the cheapest and easiest one.
For years, the argument over ultra-processed food has been trapped in a familiar loop.
Researchers point to associations with obesity, type 2 diabetes, cardiovascular disease, cancer and premature death. Industry points to calories, sugar, salt and fat, arguing that processing is too broad a category to regulate. Consumers are told to read the label, exercise restraint and make better choices.
A new EPIC study moves the argument into the bloodstream.
Published in Critical Reviews in Food Science and Nutrition, the study examined whether people reporting higher consumption of ultra-processed food also showed a distinct profile of circulating metabolites and fatty acids. The answer was yes, at least at the level of association.
This matters because metabolomics asks a different question from a food-frequency questionnaire. It does not only record what people say they ate. It measures small molecules circulating in the body and looks for patterns linked to exposure.
In the metabolite analysis, researchers used data from 6,177 EPIC participants and measured 129 endogenous metabolites. In a separate fatty-acid analysis, they used data from 9,029 participants. After adjustment for age, sex, body mass index, waist circumference, smoking, physical activity, alcohol, fasting status and consumption of other NOVA food groups, ultra-processed food intake remained associated with 22 metabolites and eight fatty acids.
The signature was not a single toxic molecule. It was a pattern.
Higher intake was linked to higher levels of several long-chain acylcarnitines and to a fatty-acid profile that included more stearic acid, several omega-6 fatty acids and elaidic acid, an industrial trans fatty acid. It was also linked to lower levels of several phosphatidylcholines and sphingomyelins involved in cell membranes and signalling, as well as lower docosahexaenoic acid, or DHA.
The authors argue that the pattern provides clues about biological pathways connecting ultra-processed food exposure with adverse health outcomes, particularly altered lipid and energy metabolism.
That is a meaningful signal. It is not a verdict.
The study measured diet and blood at one point in time. It cannot show which came first, and it cannot prove that processing caused the metabolic differences. People who consume more ultra-processed food may differ in ways that statistical adjustment cannot fully capture. The dietary information also came from EPIC baseline assessments conducted between 1992 and 2000, when products, formulations and consumption patterns were different from those on supermarket shelves today.
Classification is another fault line. NOVA groups foods according to the extent and purpose of industrial processing, but broad categories can place very different products together. A fortified wholegrain breakfast cereal, a sugar-sweetened drink and a processed meat product may all be classified as ultra-processed while carrying different nutrient profiles and likely different risks.
That is why the policy response cannot be “ban everything processed”. Processing improves safety, shelf life and access in many contexts. Some industrially produced foods are important for people with limited time, disability, low income or poor access to fresh food.
But the opposite simplification is equally indefensible: that only nutrients matter and the food system is neutral.
The 2019 NIH controlled feeding trial already showed that participants ate about 500 more calories per day and gained weight on an ultra-processed diet compared with a minimally processed diet, even when the meals presented were matched for calories, sugar, fibre, fat and carbohydrates. The trial was small, but it challenged the assumption that a nutrition label captures everything relevant about how food affects eating behaviour.
The wider epidemiological evidence has also become difficult to ignore. A 2024 BMJ umbrella review found that greater exposure to ultra-processed food was associated with a range of adverse outcomes, with particularly consistent signals across cardiometabolic health. An EPIC analysis published in 2023 linked higher consumption with increased risk of multimorbidity involving cancer and cardiometabolic disease.
The new metabolomics study does not settle the mechanism. It narrows the gap between population association and biological plausibility.
That has consequences for policy.
First, governments need better data. Food surveys should record brand, formulation, additives, processing methods and product changes over time. Without product-level information, regulators are trying to govern a rapidly changing market using blunt historical categories.
Second, nutrient profiling and processing classification should be treated as complementary, not competing, tools. A product can be high in salt, sugar or saturated fat and highly processed. Another can have a better nutrient profile but still be engineered for rapid consumption, convenience and intense marketing. Policy needs enough precision to distinguish these cases.
Third, the response must move beyond health education. Telling people to cook from scratch is not a serious strategy when minimally processed food is less accessible, more expensive, quicker to spoil and more demanding of time, transport, storage and cooking facilities.
School and hospital procurement, restrictions on marketing to children, clearer front-of-pack labelling, fiscal measures, subsidies for healthier staples and urban food access are not lifestyle interventions. They are infrastructure.
The equity stakes are obvious. The people most exposed to cheap, aggressively marketed products are often the same people with the least capacity to absorb higher food costs or spend more time preparing meals. A policy that tells them to exercise willpower while leaving the commercial environment untouched is not prevention. It is blame dressed as advice.
There is also a risk of premature commercialisation. Metabolomic signatures are attractive because they sound precise and personalised. But this study does not provide a validated clinical test, a diagnostic cut-off or a way to tell an individual patient that ultra-processed food caused their disease.
No health system should begin screening people for a “UPF signature” on the basis of this paper. The next steps are replication in contemporary cohorts, repeated measurements, prospective links to disease outcomes and intervention studies testing whether the signature changes when diets change.
WHO is already developing guidance on ultra-processed food consumption, and its 2026 One Health agenda explicitly places ultra-processed foods within agriculture, trade, environmental sustainability, food safety and commercial governance. That is the right level of debate.
The problem is not one molecule, one snack or one irresponsible consumer. It is a system that can produce cheap convenience at scale while transferring long-term health costs to patients, families and public budgets.
The new study suggests that this system may leave traces in the blood.
Policy should be careful about what those traces mean. It should be far less cautious about asking who profits from the exposure, who pays for the consequences and why healthier food remains the harder choice.
