Ten different “healthy diets” were all linked to slightly slower epigenetic aging in German adults

Nutrition advice comes in many brands: the Mediterranean diet, the Nordic diet, the DASH diet designed to lower blood pressure, the MIND diet aimed at the brain, plant-based diets. They overlap a lot (more vegetables, fruit and whole grains, less processed meat and sugar), but they also differ in the details. Does it matter which one a person follows?

A team from the German Center for Neurodegenerative Diseases (DZNE) in Bonn and colleagues looked at this question through DNA methylation: small chemical tags on DNA that help switch genes on and off and that change with age. Patterns of these tags can be turned into epigenetic clocks, which estimate how "old" a person's blood looks compared with their age in years.

What they did

The main group was 6,470 adults aged 30 to 95 (average 56, 57% women) from the Rhineland Study, a long-running study of people living in two districts of Bonn. They filled in a questionnaire about how often they had eaten a list of foods and drinks over the past year. From it, the team scored how closely each person's diet matched ten published diet patterns:

• eight "healthy" ones: Mediterranean, DASH, MIND, the Alternate Healthy Eating Index, Nordic, the EAT–Lancet "planetary health" diet, and an overall and a "healthful" plant-based diet index;
• and two that measure less healthy eating: an "unhealthful" plant-based index and a score of how much a diet promotes inflammation.

From a blood sample, they measured three epigenetic aging measures. Two (GrimAge and PhenoAge) say how much older or younger someone's blood looks than expected for their age. The third, DunedinPACE, estimates the current pace of aging, in "biological years per calendar year". They also scanned about 850,000 methylation sites one by one to see which were linked to each diet. The models took sex, smoking, total calories and, for some diets, alcohol into account.

Then they repeated the main analyses in 1,034 adults aged 35 to 64 from a second German study, EPIC-Potsdam.

What they found

• Few people are "top" on every diet. Of the 6,470 people, 1,212 were in the best quarter on at least one diet score, but only 18 were in the best quarter on all ten.
• Healthier eating, slightly younger clocks. People who ate closer to the healthy patterns tended to have lower values on all three aging measures; the two unhealthy scores went the other way. All ten diet scores were linked to DunedinPACE after correcting for the many tests.
• DASH and Nordic stood out. They were the two diets linked to all three measures. A one-step-up (one standard deviation) in the DASH score went with a pace-of-aging value 0.10 standard deviations lower, and with GrimAge and PhenoAge acceleration about 0.03 standard deviations lower. These are small differences.
• The exception. The EAT–Lancet score was linked to DunedinPACE, but not to the other two clocks.
• The second cohort agreed. In EPIC-Potsdam, the links between diet scores and the two clocks measured there were similar in direction and size.
• Different tags, shared pathways. Each diet was linked to its own set of methylation sites, from 54 (Alternate Healthy Eating Index) down to 3 (unhealthful plant-based), and there was little overlap between diets. But the genes near those sites pointed to largely the same biological processes, with more than 70% overlap, such as cell signalling and metabolism. Only 39 of the 240 sites found in Bonn were confirmed in Potsdam.
• Stronger in smokers. The diet links were clearer in current smokers than in former or never smokers, in both cohorts.

What to keep in mind

• A snapshot, not an experiment. Diet and blood were measured at about the same time. People who eat well may differ in many other ways, and the authors write that "residual confounding" can't be ruled out. The study shows associations, not that any diet slows aging.
• The authors' own limits. Both groups were mostly German people of European ancestry, so the results may not apply elsewhere. Food questionnaires rely on memory and miss details. The Potsdam diet data were collected in the 1990s, when people may have eaten differently.
• Small effects. The differences per step on a diet score were a few hundredths to a tenth of a standard deviation. What that means for health or lifespan isn't tested here.
• Clocks are estimates. Epigenetic clocks are research tools; a "younger" clock reading isn't the same thing as living longer or staying healthier.
• Weak replication of single sites. Most of the individual methylation sites from Bonn were not confirmed in the smaller Potsdam group.

Why it matters

The message the authors draw is that the label on a healthy diet may matter less than eating healthily in general: several quite different patterns were linked to similar, modestly slower epigenetic aging, and their molecular fingerprints converged on shared processes. They also see hints of diet-specific pathways, such as heart-valve development for DASH and memory-related processes for MIND, which they suggest could one day help tailor advice. Trials that change people's diets and track these clocks over time would be needed to show whether the link is cause and effect.