Biological-age tests: the names behind the number

Imagine Eva, a fictional 52-year-old, opening a report that gives her a biological age of 47. It is a pleasing number. Then another section shows a pace-of-aging score rather than an age. A second laboratory offers a different result again.

The apparent contradiction becomes easier to understand once the tests have names. “Biological age” is an umbrella term for models built from different measurements and trained to predict different outcomes. An age-like number is the model’s summary, not a direct measurement of how many years are left in a person’s body.

Four names that explain much of the landscape

Many epigenetic tests measure DNA methylation, chemical marks at selected places on DNA. An algorithm converts that pattern into a score. The algorithm matters as much as the sample.

Steve Horvath’s 2013 paper introduced the widely used multi-tissue clock. Morgan Levine and colleagues’ PhenoAge work connected an age measure to health-related blood chemistry. Ake Lu and colleagues’ GrimAge used methylation-based estimates of seven plasma proteins and smoking history, rather than directly measuring all of those proteins in the test sample.

Daniel Belsky and colleagues developed DunedinPACE using a New Zealand birth cohort followed over two decades. Its reference scale corresponds to biological change per calendar year. A value of 1.0 represents the reference pace; a lower value indicates a slower estimated pace relative to that reference. It does not tell Eva that every organ is a particular number of years younger.

PhenoAge can mean two different calculations

The original clinical PhenoAge combines age with nine blood measurements: albumin, creatinine, glucose, C-reactive protein (CRP), lymphocyte percentage, mean red-cell volume (MCV), red-cell distribution width (RDW), alkaline phosphatase and white blood cell count. DNAm PhenoAge estimates that construct from methylation data. The original paper explains the distinction. Identifying the input prevents a misleading comparison between reports.

Commercial names add another layer. For example, TruDiagnostic’s TruAge product currently lists both OMICmAge and DunedinPACE among its report components. That is a description of the company’s listed product contents, not an independent endorsement. A brand can package several algorithms; evidence for one algorithm does not validate every score or recommendation in the package.

A real trial where the clocks disagreed

In the two-year CALERIE calorie-restriction trial, a DNA-methylation analysis of 197 participants found a small slowing on DunedinPACE, but no significant effect on the PC versions of PhenoAge and GrimAge. “PC” identifies versions designed to improve technical reliability. The same intervention therefore did not produce the same answer across all three measures.

That is a reason to keep the exact test and version visible when tracking a result. A switch between algorithms should not be presented as a before-and-after improvement. Our guide to why biological-age tests give different answers explores this further.

An “organ age” is also a model

A different research approach uses proteins in blood. In Hamilton Oh and colleagues’ study of 44,498 UK Biobank participants, protein-based estimates for 11 organs were associated with later health outcomes. The study used the Olink proteomics platform, which measures many proteins from a blood sample.

A protein-derived “heart age” is not an echocardiogram, and a “brain age” is not a diagnosis of dementia. The associations are valuable for research, while an individual symptom still requires clinical assessment.

Put Eva’s score next to her actual health

Suppose Eva also has repeatedly elevated blood pressure and an abnormal HbA1c result. A younger epigenetic age would not resolve either finding. HbA1c estimates average glucose over roughly three months; a diagnostic result may need confirmation. A lipid profile, including LDL cholesterol, can contribute to cardiovascular risk assessment. These tests address defined aspects of health. NIDDK explains the role and limits of HbA1c.

If the concern is reduced exercise capacity, a clinician might select a cardiopulmonary exercise test (CPET) after the initial assessment. It measures breathing and oxygen use during exertion, alongside heart and blood-pressure monitoring. That direct functional examination contributes different information from a molecular age score. These are examples, not a required test bundle for Eva or every reader.

The useful record keeps the algorithm name, laboratory, test date and clinical context together. Lowering an age score becomes a meaningful treatment target only if there is adequate evidence that changing it improves an outcome people value. The FDA’s explanation of surrogate endpoints describes why a biomarker cannot automatically stand in for clinical benefit.

Healthy Longevity Clinic describes its own integration of measurements and medical context on its methodology page. This remains an external-review draft; academic clock studies do not validate that separate approach. For the next step, read which aging tests can help a health decision: the aim is to turn an interesting result into an appropriate action, while retaining the information that explains it.