Blood-Based vs. Epigenetic Age Tests

Blood-based and epigenetic age tests can both return a number called biological age. They arrive there by different roads. A clinical blood composite combines measurements such as albumin, glucose, creatinine and blood-cell counts. An epigenetic clock reads chemical patterns on DNA, called methylation, and applies an algorithm trained to predict age, outcomes or pace of aging. Humanic Blue is comparing what goes into those scores and what comes out. Here, “blood-based” means routine blood chemistry and cell counts. An epigenetic test can also use a blood sample, but it analyzes DNA patterns instead.

Neither category directly measures how long one person will live. Blood composites use familiar, clinically interpretable inputs but the age calculation itself is not a diagnosis. Epigenetic clocks can capture molecular patterns associated with disease and mortality in cohorts, but interpretation is limited by model choice, technical error, tissue and population fit. For living longer and better, the useful comparison is what each result can change—not which one produces the more flattering birthday.

What does each kind of test measure?

	Clinical blood composite	Epigenetic clock
Input	Routine blood tests and sometimes body measurements	DNA methylation at selected CpG sites
Examples	Clinical PhenoAge, Klemera–Doubal physiological age	Horvath, DNAm PhenoAge, GrimAge, DunedinPACE
Output	Usually age-like years	Age-like years, age acceleration or pace
Main advantage	Components are familiar and often clinically interpretable	Molecular patterns can be trained against age, risk or longitudinal change
Main limit	Composite can move with illness and does not equal a diagnosis	Method, tissue, processing and training population can materially affect the result

Clinical PhenoAge illustrates the blood side. Its developers used about 10,000 US adults to combine chronological age with nine measurements: albumin, creatinine, glucose, C-reactive protein, lymphocyte percentage, mean cell volume, red-cell distribution width, alkaline phosphatase and white-cell count.

Epigenetic clocks measure methylation—chemical marks at DNA locations called CpG sites. The original Horvath clock used 353 sites to estimate chronological age across many tissues. Later clocks changed the target: GrimAge was trained around mortality-related surrogates, while DunedinPACE was trained on change in 19 organ-system indicators across two decades.

Calling both outputs “age” is convenient. It does not make their inputs, scales or intended uses interchangeable.

PhenoAge is the bridge between them

PhenoAge exists in two forms. The clinical version uses ordinary blood measurements. DNAm PhenoAge uses DNA methylation to estimate that combined clinical score.

After constructing the blood-based score, Morgan Levine and colleagues trained the methylation version in about 450 people, selecting 513 CpG sites. DNAm PhenoAge was therefore not an independent discovery of the same hidden age. It was designed to approximate a score first calculated from age and blood tests.

The development paper reported that the blood-based score predicted illness and mortality better than DNAm PhenoAge in its comparisons. That result does not prove that every blood model outperforms every methylation clock. It shows why “epigenetic” should not automatically be read as “more clinically informative.”

What difference does it make whether PhenoAge uses blood measurements or DNA methylation?

What do the outcome studies actually show?

In a later US analysis, Liu and colleagues evaluated clinical PhenoAge in about 11,400 adults ages 20–84 followed for a median of 12.6 years. Each year of acceleration beyond chronological age was associated with 9% higher mortality during follow-up. The observational design cannot show that lowering the score would reduce mortality.

Outcome-trained epigenetic clocks also predict risk in cohorts. GrimAge was developed in about 1,700 Framingham participants and tested in about 7,000 people across five cohorts; each year of GrimAge acceleration was associated with about 10% higher mortality during follow-up.

Those percentages look close, but they do not form a fair contest: the models used different populations, inputs and analyses. A 2020 eLife study compared nine aging measures in about 850 Swedish adults followed across 20 years. GrimAge, the Horvath clock and a clinical frailty index, which summarizes health problems and limitations, each added information about mortality when considered together. The physiological-age composite was easier to connect to specific components but had a weaker mortality association.

The current epigenetic evidence also resists a single winner. A 2025 Nature Communications analysis compared 14 clocks with 174 incident diseases and mortality in about 19,000 people in Scotland over ten years. Second- and third-generation clocks outperformed birthday-trained clocks, but no clock was clearly best across all diseases.

Reliability is not the same problem for both tests

Routine laboratory assays have established quality systems, but an age composite can still move because its inputs change. Infection, inflammation, fasting status, medication and chronic disease may affect individual components. That can be meaningful clinical information, temporary variation or both. Published test-retest data for the composite age calculations remain limited.

Epigenetic clocks add laboratory and algorithmic sources of variation. A 2022 Nature Aging study found differences of up to nine years when the same sample was measured again using six clocks. Redesigned versions used a statistical method called principal components to reduce noise and brought most repeat readings within 1.5 years. The authors disclosed Elysium licensing, supplied datasets and employment or advisory relationships.

Tissue also matters. A 2025 cross-tissue study analyzed samples from about 80 people ages 9–70 and found that many blood-trained clocks correlated poorly between oral and blood tissues. Some age estimates differed by almost 30 years.

Population fit is another constraint. A 2026 eLife study found that many clocks performed less accurately in populations of people with mixed genetic ancestry than in White comparison groups. A result can repeat consistently in the lab while predicting age less accurately in some groups.

Which result is clinically actionable?

The individual components of a blood composite can have established clinical meaning. Blood pressure, glucose and A1C, kidney measures and standard lipids are interpreted through current guidelines, history and repeat testing. The biological-age equation laid on top of them does not create a new diagnosis.

Epigenetic clocks are used widely in research, but no current guideline cited here recommends treating a person to a target GrimAge, PhenoAge or DunedinPACE. The FDA’s direct-to-consumer guidance notes that consumer tests have varying evidentiary support and review pathways, and advises against making dietary or health decisions without professional discussion.

This creates a practical split:

How would you separate the clinical meaning of a blood value from the meaning of the combined age score?

• If the purpose is a medical decision, discuss validated clinical measurements with a clinician.
• If the purpose is research curiosity or personal tracking, an aging clock can supply a model-based estimate, provided its error and limits are understood.
• In either case, do not treat the age output as a diagnosis or compare unlike methods as if they share a unit.