The topline. San Juan, October 23, 2026. Almost everything nutrition research knows about diet and longevity comes from food frequency questionnaires, which rely on people accurately remembering and reporting what they ate. A study published in The American Journal of Clinical Nutrition this month asks a better question: what if adherence to a healthy diet could be read directly from the blood? The answer is a 67-metabolite signature that predicted survival across four cohorts on two continents.

Key facts

The researchers used a previously developed plasma signature of 67 metabolites, selected from 302 compounds, that tracks Mediterranean diet adherence: olive oil, fruit, vegetables, legumes, nuts, fish.

The discovery cohort came from PREDIMED, the Spanish trial of 7,447 adults at high cardiovascular risk. Of those, 1,880 had baseline metabolomic data and 1,648 had one-year measurements, with roughly 13.2 years of follow-up from death registries and medical records. The highest third of signature scores had a 23 percent lower relative hazard of all-cause death than the lowest third; each standard deviation increase was tied to a 19 percent lower hazard. The one-year signature showed a similar pattern, with each standard deviation tied to a 26 percent lower hazard of cardiovascular death. The cancer association was not statistically significant.

Replication is what separates a signal from a finding. Applied to 12,704 participants in the Nurses’ Health Studies I and II and the Health Professionals Follow-up Study, with an average of 25 years of follow-up and 4,111 recorded deaths, the result held: the highest fifth of signature scores carried a 23 percent lower relative hazard of all-cause death than the lowest fifth.

Sensitivity analyses added sturdiness: excluding early deaths, limiting follow-up to ten years, and stratifying by age, sex, and diabetes produced consistent results. Crucially, the findings held after adjusting for self-reported dietary adherence. The blood signature predicted survival beyond what questionnaires captured, suggesting it reflects dietary and metabolic status that surveys miss.

Clinical relevance

The honesty clause matters. These are observational analyses, so they do not establish that changing diet to raise the score extends life. The platform could not capture certain phenolic compounds and antioxidants, and the U.S. analyses may carry selection bias.

But the direction of travel is clear. A day is coming when dietary counseling is checked against an objective metabolic readout rather than a remembered food diary, the way lipid panels already do. For now the takeaway is already usable: the people whose metabolism looks like the Mediterranean diet live longer. What their plates show and what their blood remembers are two different records, and the blood is winning.

23 percent

That is the lower relative hazard of all-cause death for the top of the signature score range versus the bottom, replicated across cohorts: 23 percent in PREDIMED’s top third, and 23 percent again in the top fifth of the U.S. cohorts after 25 years of follow-up. When the same number appears in independent populations on two continents, it stops being a curiosity and starts being a tool.

What we’re reading