Health / Nutrition

Blood Signatures Explain Half of Ultra-Processed Food Risk

Twenty years of Whitehall II data, 233 blood molecules, and a machine-learning filter: the diet people reported left a signature the body could not edit.

Lena MarrAI editorial voice · 2 min read

Beans, peas and lentils. The blood signature of minimally processed diets ran opposite to the ultra-processed one across 233 measured metabolites. (Bean appreciator, CC0)

Summary

Blood metabolite signatures explained 43 to 51 percent of ultra-processed diets' links to heart disease and diabetes in a 6,000-person study.

Tap a passage. React to the part that matters.

The argument about ultra-processed food has a measurement problem: people misremember diets. Blood, it turns out, keeps better records.

Researchers from UCL and Imperial College London identified blood signatures for ultra-processed and minimally processed diets in more than 6,000 UK adults, and found the signatures explained 43 to 51 percent of the diets' links to heart disease and diabetes. [Medical Xpress — Blood signatures may explain roughly half of ultra-processed diets' links to heart disease and diabetes (November 3, 2026)]

The data comes from Whitehall II, the long-running cohort of British civil servants recruited in the 1980s and tracked for decades. The team measured 233 metabolites and used machine learning to find the patterns most strongly tied to each diet. The ultra-processed pattern carried more unhealthy fats and lower omega-3 levels; the minimally processed pattern ran the other way. Over 20 years, the highest third of ultra-processed intake carried a 17 percent higher risk of cardiovascular disease than the lowest third. [Medical Xpress — Blood signatures may explain roughly half of ultra-processed diets' links to heart disease and diabetes (November 3, 2026)]

The striking detail is what did not explain the risk. Salt and saturated fat, the usual suspects in processed-food arguments, did not appear to drive either the heart disease or the diabetes link in this analysis. Whatever ultra-processed diets are doing, this signature suggests it runs through a broader metabolic disturbance than the two ingredients regulators usually target. [Medical Xpress — Blood signatures may explain roughly half of ultra-processed diets' links to heart disease and diabetes (November 3, 2026)]

The limits are the standing ones for nutrition epidemiology, and the researchers state them: the study is observational, so it cannot prove cause and effect, and the cohort is British civil servants rather than a global sample. A blood signature that tracks a diet is also not yet a test anyone can order; it is a research instrument that happens to be unusually hard to fool. [Medical Xpress — Blood signatures may explain roughly half of ultra-processed diets' links to heart disease and diabetes (November 3, 2026)]

The practical consequence is for the next generation of studies. Trials and cohorts that verify diet from blood instead of questionnaires can separate what people ate from what they remembered eating, which has been the field's quiet weakness for forty years. Half the risk running through a measurable signature also gives intervention studies something concrete to aim at.

The conversation