Biological-Age Tests: Why Two Clocks Can Tell Different Stories

Picture two reports arriving from the same blood sample. One suggests that biological aging has slowed. Another shows little change. It is tempting to decide that one test must be wrong—or to keep the more flattering result.

There is another possibility: the two algorithms are answering different questions.

DNA methylation clocks use patterns of chemical marks on DNA to estimate aspects of aging. Some generate age estimates; others are designed to capture a pace of change or a pattern associated with health outcomes. Their usefulness depends on the question, the population and the evidence behind the particular measure. Recent studies offer reasons to take these tools seriously while remaining careful about what an individual score proves. ^[1]^[2]^[3]

A clock is a model, not an extra birthday

Calendar age has a simple definition: time since birth. “Biological age” describes something more complicated. Researchers are trying to summarize molecular and physiological differences that may help explain why people of the same calendar age have different health trajectories.

A methylation clock translates measured patterns into a score. The training target and the way that score was validated matter. An age estimate, a rate-of-aging measure and a system-specific score need not move together. They are not automatically interchangeable units of a single quantity. The CALERIE analysis provides a concrete example of that distinction. ^[3]

This also changes how a headline should be read. A result described as slower aging by one algorithm should name that algorithm. Otherwise, a specific measurement can quietly become a much broader claim about the whole person.

Measurement or finding	Useful question it addresses	Conclusion it cannot establish on its own
A methylation-based age estimate	What age-related pattern does this model detect?	How many years a person will live
A pace-of-aging measure	Does the measured rate-related signal change?	That every other aging measure must change too
A system-specific model	Which modeled physiological patterns differ?	Direct examination or diagnosis of every named organ
An association with later mortality	Does the score help predict an outcome in a studied cohort?	That changing the score will change that outcome
A change after an intervention	Did this measured endpoint respond in this study?	Proven prevention of disease or extension of human lifespan

The last two distinctions are especially important when a test result becomes an advertisement for an intervention. Prediction and intervention are different scientific questions.

The encouraging finding: changes can carry information

A 2026 study followed 699 adults in the InCHIANTI cohort for up to 24 years. Researchers examined changes in several generations of epigenetic clocks and their relationship with mortality. Faster increases in several clocks were associated with higher risk of death, even after accounting for baseline epigenetic age and other confounders. ^[1]

That is a meaningful result. It suggests that repeated measurements can contain information beyond a single starting score. It also supports continued research into whether these measures can help describe changing health trajectories.

But the study observed relationships over time. It did not randomly assign participants to a product that changed a clock and then establish that the product extended life. A marker that predicts risk can be useful without yet being a validated substitute for a clinical outcome in an intervention trial. That is a limit on the inference, not a reason to dismiss the finding. ^[1]

The revealing experiment: one intervention, different clock results

The CALERIE trial offers a particularly useful reality check. In a post hoc analysis of a randomized trial, researchers examined blood-based methylation measures from 220 adults without obesity assigned to a controlled caloric-restriction intervention or a comparison diet over two years. ^[3]

The intervention slowed the measure called DunedinPACE. It did not produce significant changes in several biological-age estimates, including PhenoAge and GrimAge, and the reported treatment effects were small. The same experiment therefore yielded a positive result on one aging measure and no significant change on others. ^[3]

Those findings should travel together. Reporting only the positive clock suggests a uniform response the paper did not find. Reporting only the unchanged clocks erases a randomized finding that the authors considered important.

The authors also called for trials with long-term follow-up that assess chronic disease and mortality. A change in a promising molecular endpoint does not itself demonstrate those outcomes. Nor does this controlled study validate an unrelated supplement, research compound or commercial “age reversal” program. ^[3]

Eleven systems do not mean eleven organ biopsies

A 2025 paper introduced Systems Age, a framework that used a single blood methylation test to derive scores for 11 physiological systems. Its modeling integrated clinical biomarkers, functional assessments and mortality risk. The researchers reported improved prediction of relevant diseases and aging features compared with existing global clocks. ^[2]

The appeal is understandable: a single overall number can hide variation within a person. A system-oriented framework asks whether different patterns tell a more detailed story.

The output still needs to be understood as a model derived from blood methylation data. A score named for the brain, kidney or heart is not a direct examination of that organ, and the abstract does not establish that every commercial implementation can diagnose an individual patient's disease. The study supports research into more detailed aging measurement; the interpretation must stay tied to the validated model and task. ^[2]

A better score cannot certify a research material

Suppose someone uses an experimental product and later gets a lower clock score. Even if the measurement is accurate, several questions remain: was the comparison controlled, was this the planned endpoint, did other measures agree, and did the intervention improve outcomes that matter to people?

There is also a separate product question. A molecular score does not establish the identity, quality or suitability for human use of the material someone consumed. FDA's discussion of unapproved GLP-1 products offers a specific example of why this matters: encouraging clinical interest does not remove concerns about unapproved preparations and research-labeled products promoted for human use. That warning is about those products; it is not a safety study of biological-age tests. ^[4]

Research materials should not be consumed or self-administered in pursuit of a younger number. Neither reconstitution nor a favorable biomarker result turns a laboratory reagent into an approved medicine.

The more useful reading of a clock report begins with its name, its intended measurement and its validation. Two clocks can disagree because they were built to capture different features. The scientific opportunity lies in understanding those differences—and then testing whether changes predict better lives, rather than simply better-looking reports.

Source note: The study discussion uses the original published abstracts for the InCHIANTI, Systems Age and CALERIE papers. It does not present a full-text methods review, endorse a consumer test or claim that Chudik Scientific offers biological-age testing.