Blood Test Score Predicts Heart Disease Years Ahead—and Improves with Lifestyle Change

A simple blood test that distills dozens of metabolic signals into a single number may soon give people in midlife something they have never had before: a personalized, clinically validated readout of their overall health—and evidence that the number can actually be moved in the right direction. In a study published in BMC Medicine, researchers at Leiden University Medical Center report that a revamped version of the MetaboHealth score, called Personal-MetaboHealth, predicts both death and the onset of cardiometabolic disease with striking accuracy, and that it responds to a three-month diet and exercise intervention in the people who need it most.

The original MetaboHealth score was developed as a cost-efficient indicator of overall health, built on high-throughput nuclear magnetic resonance (¹H-NMR) spectroscopy measurements from the Nightingale Health platform. Trained on prospective mortality data from 12 large European cohorts, the score combines 14 metabolic biomarkers—ranging from glucose and albumin to amino acids such as histidine, isoleucine, leucine, phenylalanine, and valine, plus lipoprotein measures—into a weighted sum that captures systemic changes preceding disease. Higher scores on the health-oriented version of the scale indicate better survival prospects, and previous work linked the score to frailty, cognitive decline, COVID-19 outcomes, and loss of functional independence.

But the original score had a practical flaw: it was scaled within individual study populations, meaning two people from different cohorts could not be directly compared. To fix this, the team re-anchored the biomarker distributions using BBMRI-NL, a nationwide Dutch biobank infrastructure encompassing 29,792 participants from 24 cohorts aged 16 to 103. This produced Rescaled-MetaboHealth, a version interpretable at the individual level against a defined population reference. The researchers then went a step further, restricting the score to the 10 biomarkers that have received CE marking as clinically validated components of an in vitro diagnostic device—yielding Personal-MetaboHealth, a score that meets regulatory requirements for real-world clinical use.

The predictive power of the new score was tested in the Leiden Longevity Study, a three-generation cohort of 421 long-lived families. The analysis focused on 2,404 second-generation participants—offspring and their partners—with a mean age of 59.2 years, followed for up to 22 years for mortality and 16 years for disease incidence. During follow-up, 489 participants died and 498 developed morbidity. All three versions of the score performed comparably for all-cause mortality: each standard deviation increase corresponded to a 23 to 25 percent lower risk of death, with hazard ratios of 0.77 for the original score, 0.76 for the rescaled version, and 0.75 for Personal-MetaboHealth.

Where Personal-MetaboHealth truly distinguished itself was in predicting the first onset of cardiometabolic disease, defined using ICD-10 codes for stroke, cerebrovascular accident, angina pectoris, acute myocardial infarction, hypertension, and diabetes. Among the 1,403 participants free of cardiometabolic disease at baseline, each standard deviation increase in Personal-MetaboHealth was associated with a 26 percent lower yearly risk of developing a first cardiometabolic disease—a 12 percent stronger effect than the other two scores. The researchers attribute this advantage partly to the removal of metabolites with higher technical variability, such as acetoacetate and small HDL lipids, which dropped out when the score was confined to the clinically validated subset.

Because the score is known to correlate with lifestyle factors, the team ran robustness checks adjusting for smoking quantity, alcohol consumption, and medication use. The mortality associations weakened slightly but remained significant. For cardiometabolic disease, the original and rescaled scores lost significance entirely after adjustment, largely because of smoking—but Personal-MetaboHealth held firm, with an adjusted hazard ratio of 0.78. This suggests the new score captures broader dimensions of health beyond those affected by smoking, an important property for a tool intended to stratify risk across whole populations.

To translate the statistics into something more intuitive, the researchers applied accelerated failure time models, which express effects in years rather than relative hazards. The result was remarkable: each unit increase in Personal-MetaboHealth corresponded to a 36 percent delay in the onset of first cardiometabolic disease, equivalent to approximately 13.69 additional years before the median participant would be expected to develop their first cardiometabolic condition. The magnitude of this effect is comparable to well-established risk factors such as smoking and socioeconomic status, underscoring just how much information is packed into this single metabolic composite.

Prediction alone, however, is not enough for a screening tool—it must be actionable. To test this, the team turned to the Growing Old Together (GOTO) study, a 13-week single-arm lifestyle intervention involving 164 healthy older adults with a mean age of 62.8 years. Participants achieved a 25 percent reduction in energy balance through a 12.5 percent decrease in caloric intake and a 12.5 percent increase in physical activity, guided by personalized plans from a dietician and physiotherapist aligned with the Dutch Guidelines for a Healthy Diet. Across all participants combined, the intervention produced no significant change in Personal-MetaboHealth. But when the researchers stratified participants by baseline score—using zero on the BBMRI-NL scale as the marker of the average Dutch adult—a clear pattern emerged.

Among the 52 participants whose baseline scores marked them as less healthy than the population average, the intervention produced a significant improvement of 0.26 units, driven by favorable changes in glycoprotein acetyls, glucose, phenylalanine, isoleucine, and leucine—biomarkers reflecting inflammation and metabolic function. These gains were accompanied by improvements in fasting triglyceride, C-reactive protein, and insulin levels. In contrast, the 83 participants who started out healthier showed a slight average decrease of 0.13 units, though 81 percent of them retained a healthy score and they continued to improve in blood pressure, body composition, and fat metabolism, with no significant worsening in any measured parameter—suggesting the dip reflects normal fluctuation within the healthy range rather than harm.

The findings position Personal-MetaboHealth as a candidate for what the authors call an actionable health check in middle age: a blood-based number that identifies who is at risk, quantifies how far ahead disease may lie, and demonstrates that it can be improved through evidence-based lifestyle change. The researchers note that the score captures different aspects of intervention response than epigenetic clocks such as GrimAge, which declined in all GOTO participants and tracked body composition changes, whereas the metabolic score’s improvement in at-risk individuals reflected broader immune-metabolic shifts—implying the two approaches could be complementary. Limitations remain, including the family-based cohort design, baseline-only metabolomic sampling, and a predominantly Caucasian reference population that may not generalize to other ethnic groups. The team plans to validate the score in underrepresented populations and to develop country-specific versions. If those efforts succeed, a routine blood draw in midlife could become the wake-up call that turns years of silent risk into decades of healthier life.

Subject of Research: A metabolomics-based health score for predicting mortality and cardiometabolic disease and its response to lifestyle intervention

Article Title: Personal-MetaboHealth, an actionable health check in middle age, is improved by an effective lifestyle intervention in those at risk

Article References: van den Berg, N., Natalle Lopes, G., Bogaards, F. A., Beekman, M., Amaro Junior, E., Slagboom, P. E., & Deelen, J. (2026). Personal-MetaboHealth, an actionable health check in middle age, is improved by an effective lifestyle intervention in those at risk. BMC Medicine, 24(1), Article 470. https://doi.org/10.1186/s12916-026-05178-z

Image Credits: AI Generated

DOI: 10.1186/s12916-026-05178-z

Keywords: Personal-MetaboHealth, metabolomics, biomarkers, cardiometabolic disease, all-cause mortality, lifestyle intervention, biological aging, NMR spectroscopy, preventive medicine, Leiden Longevity Study, healthy lifespan, risk stratification