Metabolic blood score tracks disease risk, with mixed lifestyle-intervention results

A metabolic blood score intended to help assess health in middle age has shown associations with later disease and death, alongside mixed results in a lifestyle intervention. The study, published September 7 in BMC Medicine, examines whether Personal-MetaboHealth could help identify people for preventive support and monitor their response.

The distinction matters for healthspan research: a measurement that forecasts illness must also demonstrate that using it improves decisions and outcomes. This study advances the assessment of a candidate marker, while leaving that clinical question open.

### What the new analysis found

Researchers assessed 2,404 Leiden Longevity Study participants, averaging about 59 years at baseline, with follow-up reaching 22 years for mortality and 16 years for cardiometabolic disease. The revised score combines 10 blood biomarkers and uses a Dutch reference population; higher scores indicate better health.

Higher scores were associated with lower mortality and later cardiometabolic disease. In a separate, three-month lifestyle analysis, however, the score did not improve significantly across all participants. It improved among 52 people starting below the reference threshold and declined among 83 starting above it.

That subgroup pattern warrants caution. Selecting groups by their initial measurements can produce movement toward the average on repeat testing, even without a treatment effect. The results cannot by themselves establish who benefits from an intervention.

### A research program with a longer history

The underlying approach predates this week's publication. A 2019 Nature Communications study examined 44,168 people across 12 cohorts and identified 14 circulating biomarkers associated with mortality. Those measurements reflected processes including fat metabolism, inflammation and glucose metabolism.

In that earlier work, a model combining the biomarkers with sex predicted five- and ten-year mortality more accurately than a comparison model using conventional risk factors. But the investigators explicitly identified further validation as necessary before clinical use. They also noted that calculating biomarker values separately within each cohort limited individual application.

This history helps explain the current research direction. A useful health check needs a consistent reference scale, as well as an association with future outcomes. A strong statistical relationship alone does not establish a screening program's value, or show that changing the measured molecules will change a person's prognosis.

### Why the intervention design matters

An earlier report on Growing Old Together, published in Nature Communications in August 2024, describes it as a nonrandomized lifestyle trial. It combined dietary changes and physical activity, with fasting insulin as its primary outcome and a wider range of metabolic, physical and molecular measurements.

That report documented changes in blood pressure, body composition and other health measures. These provide context for investigating blood-based monitoring, but the trial was not designed to establish that selecting care according to Personal-MetaboHealth prevents disease.

A decisive next test would compare a strategy using the score with an appropriate alternative and measure outcomes that matter to patients. Researchers would also need to distinguish a reproducible intervention response from ordinary measurement variability.

For now, the practical implication is a research opportunity rather than a new standard for preventive care. Healthspan tools must bridge three separate questions: whether a marker predicts risk, whether it responds reliably to intervention, and whether acting on it helps people remain healthy. Evidence for the first two does not automatically answer the third.