New blood clues may explain why ultra-processed foods are tied to heart disease

Blood-based patterns are offering fresh insight into why ultra-processed foods are associated with heart disease, beyond their sugar, salt and saturated fat content.

Those familiar ingredients might seem enough to explain the link between ultra-processed foods and heart disease, but a recent study suggests the story is more complicated.

Researchers who tracked British civil servants over many years found that a pattern in the blood helped explain about half of the connection between ultra-processed food intake and cardiovascular disease.

The results suggest that how a food is made may matter as well as what it contains, though that remains a possibility rather than a proven cause.

Looking beyond the label

Eunice Wang carried out the research while pursuing her PhD at University College London, working alongside colleagues at UCL and Imperial College London.

“We know there is a link between ultra-processed food and disease, but there is uncertainty over what is driving this connection – whether it is due to higher levels of sugar, salt and saturated fat, as well as dense calorie content, or whether the risks also come from food processing,” Wang said.

The researchers observed people’s health over time instead of assigning them to different diets. That allowed them to identify associations, but not to prove that processing caused illness.

“Our findings hint that processing itself may play a role, although more research is needed to support this,” Wang said.

Years of dietary records

The participants were not recruited for a new diet trial. They were 6,010 members of Whitehall II, a long-running study that began in 1985 among civil servants working in London government offices.

When blood samples were taken for this analysis in the late 1990s, none of the participants had heart disease or diabetes.

During that decade, they completed questionnaires covering 127 foods on two occasions.

Two researchers reviewed the list and used a system called Nova to classify each item.

Nova sorts foods by how much they have been processed. Ultra-processed products are mostly made from substances extracted from foods and combined with additives, leaving little or none of the original whole food behind.

At the opposite end are minimally processed foods such as fruit and vegetables, whole grains, beans, fish and meat. Water and unsweetened tea and coffee are included in that group too.

Measuring more than calories

For this study, the team measured food and drink by weight rather than calories, which allowed them to include non-caloric beverages.

By that measure, around 18% of a typical person’s daily intake came from ultra-processed foods, while minimally processed foods made up about 62%.

The researchers then examined 233 blood measures, ranging from sugars and amino acids to fats and the particles that transport them.

From those, they identified 34 measures closely associated with ultra-processed diets and 50 linked to minimally processed diets.

Those measures were combined into two blood patterns, or “signatures,” and each participant received two scores showing how closely their blood matched each one.

What the blood revealed

One of the key differences involved omega-3 fats, including DHA, which were lower in the blood of people who ate more ultra-processed food.

Levels of HDL particles, which carry so-called good cholesterol, were also lower, while particles associated with fatty deposits in arteries were more common.

The pattern was different among people who ate more minimally processed food. Their blood contained more omega-3 and other polyunsaturated fats, and saturated fat made up a smaller share.

Among the foods studied, fish stood out for its strong connection to omega-3 levels.

Within the ultra-processed category, meats, sweets and ready-to-eat meals showed the strongest and most consistent links to less healthy blood readings.

The team also had years of health follow-up data. Over a typical 20-year period, there were 1,094 cases of cardiovascular disease, meaning disease of the heart or blood vessels.

For type 2 diabetes, the typical follow-up was about 18 years, during which 853 cases were recorded.

To compare diets, the researchers divided people into thirds based on their ultra-processed food intake.

Those in the top third had a 23% higher risk of cardiovascular disease and an 18% higher risk of diabetes than those in the bottom third.

These comparisons accounted for age, smoking, exercise and body weight, as well as sugar, salt, fat and fiber intake.

The comparison for minimally processed food pointed the other way.

People in the highest third had an 18% lower risk of cardiovascular disease and a 15% lower risk of diabetes than those in the lowest third.

What the questionnaires may have missed

Blood scores were also tied to later disease. People whose blood most closely matched the minimally processed pattern had a 40% lower risk of type 2 diabetes than those with the weakest match.

A stronger match to the ultra-processed pattern was associated with about a 20% higher risk of both diseases.

Adjusting for reported diet weakened those links slightly, but did not remove them, suggesting that blood tests may capture information people miss when reporting what they eat.

The researchers estimated that the ultra-processed blood pattern explained 51% of the diet’s link with cardiovascular disease and 43% of its link with diabetes.

For the minimally processed pattern, the figures were 47% and 48%.

Salt and saturated fat did not explain any detectable part of the associations in these calculations.

Sugar also did not explain the connection between either diet and cardiovascular disease, though it accounted for about 27% of the ultra-processed food link with diabetes.

Roughly half of the associations remained unexplained. The authors suggested that gut bacteria, inflammation and changes in appetite or fullness may also be involved.

Not all processed foods behave the same

Processing can change a food’s physical structure and create new compounds.

Additives and substances that migrate from packaging into food may also play a part, though this study did not establish their role.

Not every ultra-processed food showed the same pattern either.

Ultra-processed drinks and meats were linked to higher risks of both diseases, while ultra-processed breads were linked to a lower risk of diabetes.

Senior author Clare Llewellyn is based in UCL’s department of Behavioural Science and Health.

“Current UK dietary guidance does not consider processing itself as a risk but recommends a reduced intake of foods high in ingredients such as sugar, salt and saturated fat,” Llewellyn said.

Questions left for future studies

Llewellyn said the findings could inform debate over whether nutrition advice should take processing into account as well as nutrient content.

The Scientific Advisory Committee on Nutrition, an independent adviser to the UK government, is reviewing evidence on ultra-processed food and health.

The study also has limitations. Participants had to remember what they ate, and the questionnaires did not capture every detail needed to classify foods by processing.

The blood tests also could not detect additives, chemicals from packaging or changes in gut bacteria.

Another limitation was that the same people provided the data used both to build and to test the blood patterns, so a pattern that works well in this group may be less useful elsewhere.

The calculations used to estimate how much of each link was explained also relied on assumptions the study could not verify.

Most participants were white male civil servants in midlife, and their food records came from the 1990s.

Because ultra-processed foods are now more common, the authors suggested the study may understate the strength of the association today.

Earlier research has reported similar blood patterns, and feeding trials have also found differences in blood and urine between diets with and without ultra-processed foods.

What still remains unclear is why about half of the links were not explained and whether processing itself is harmful. The authors want the work repeated in more diverse groups and with other blood-testing methods.

The study is published in the journal BMC Medicine.