Telomere length and epigenetic clocks: two different aging signals
Prepared by: Research Desk
Prepared by: Research Desk. Sources are linked for verification.
Telomere length and epigenetic clocks are the two most widely known "biological age" measurements. Both shift with chronological age, and both appear in aging research. It would be natural to expect them to agree. In cohort data, they mostly don't. Once chronological age is accounted for, leukocyte telomere length and DNA methylation age tend to correlate only weakly. This explainer covers why, how each is measured, and what the methylation-based telomere estimator DNAmTL actually reports [1].
Two biomarkers, two kinds of biology
Telomeres are repetitive DNA sequences that cap the ends of chromosomes. In many dividing cell types, they shorten a little with each round of cell division. Telomere length is therefore partly a record of replicative history. It also reflects inherited starting length and other influences on telomere maintenance.
Epigenetic clocks do not measure chromosome ends at all. They read DNA methylation, the chemical tags on cytosines at CpG sites. Statistical models combine methylation levels at selected sites into an age estimate or a rate [2]. For a fuller walkthrough, see how epigenetic clocks work.
So one biomarker measures a physical structure. The other reads a pattern of gene-regulatory marks. They share a dependence on age, and their overlap is largely limited to that.
How telomere length is measured
Laboratories use three main methods for leukocyte telomere length (LTL). Each has different strengths and noise profiles:
qPCR (quantitative PCR). This method compares the amount of telomere repeat sequence with a single-copy gene and expresses the result as a T/S ratio. It is cheap, fast and suited to large cohorts. It is also relative and sensitive to DNA quality and run-to-run variation, so results are hard to compare across labs.
Southern blot TRF (terminal restriction fragment) analysis. Here, DNA is cut with enzymes that spare telomeric repeats. The fragment lengths are then sized on a gel. TRF reports an absolute length in kilobases and is often regarded as a reference method. It needs relatively large amounts of DNA and is labour-intensive.
Flow-FISH. Fluorescent probes bind telomere repeats in intact cells, and a flow cytometer reads the signal. It can report telomere length by leukocyte subtype. It requires fresh or carefully preserved cells and specialised equipment.
Because these methods differ in precision, the choice of assay alone can change how strongly telomere length appears to relate to any other biomarker.
How epigenetic clocks are built
The clocks most often discussed were trained against different targets:
Horvath (2013) was trained to predict chronological age across many tissues and cell types [2]. See the Horvath clock.
PhenoAge was trained against a composite of clinical blood markers linked to mortality [3].
GrimAge combines methylation-based surrogates of plasma proteins and smoking history and was built to predict lifespan and healthspan [4]. See GrimAge explained.
DunedinPACE estimates a pace of aging, derived from repeated physiological measurements over two decades in a single birth cohort [5].
None of these clocks was trained on telomere length. Their CpG sites were chosen to predict age, physiology or mortality, not chromosome-end biology. Our biological age test guide sets out how each clock's training target shapes what its number means.
Why the correlation is weak after age adjustment
Both biomarkers move with age. Telomeres tend to shorten, and methylation clocks tend to rise. A raw correlation between them therefore looks meaningful. The informative question is whether people whose telomeres are shorter than expected for their age also have methylation age higher than expected for their age. On that age-adjusted measure, cohort studies generally find only modest agreement. Several factors contribute:
Different mechanisms. Telomere attrition tracks cell division and telomere maintenance. Methylation clocks capture regulated epigenetic drift and programmed changes, which can occur without cell division.
Measurement noise. qPCR telomere measures carry substantial technical variability. Noise in either measure weakens any observed correlation.
Inherited starting length. Much of the variation in telomere length between adults reflects where a person started, not how fast their telomeres shortened. Clocks are less dominated by that inherited baseline.
Cell composition. Both blood measures shift with the mix of leukocyte subtypes, but not in the same way.
Large population analyses increasingly model the two as related but distinct variables. A 2025 additive Bayesian network analysis in US adult populations examined telomere length, epigenetic age acceleration and mortality risk together within one framework. It modelled them as separate variables rather than interchangeable readouts [6].
What DNAmTL actually estimates
DNAmTL, introduced by Lu and colleagues in 2019, is a methylation-based estimator of leukocyte telomere length [1]. The authors trained a model of 140 CpG sites to predict telomere length measured by the TRF method. The output is therefore a prediction of telomere length from methylation data. It is not a direct measurement of telomeres.
Key points from the original paper [1]:
DNAmTL correlates with measured LTL but does not reproduce it. Part of what it captures is methylation biology that tends to accompany telomere length.
The authors reported that DNAmTL showed a stronger negative association with age than measured LTL. They also reported that it outperformed measured LTL in predicting time-to-death and several age-related outcomes in their cohort data.
A 2026 study in people with HIV infection reported that DNAmTL outperformed qPCR-measured telomere length in reflecting clinical and sociodemographic factors [7]. One plausible reading is that DNAmTL sidesteps some qPCR noise. Another is that it captures aging-related methylation signal that is not strictly telomeric. These readings are not mutually exclusive, and the paper's title frames the finding as relative performance, not as proof of what DNAmTL measures. Commentators on HIV and epigenetic aging have also discussed how such markers might be interpreted in that population [8].
The practical upshot: a DNAmTL value and a qPCR or TRF telomere result are two different quantities that share a name. Disagreement between them is expected.
Tissue matters
Most telomere and clock data come from blood. Tissue-level studies show why blood findings do not automatically generalise. A 2026 study linked shorter kidney telomeres with nephrosclerosis through an epigenetic signature [9]. Brain tissue research has evaluated genomic signatures of aging in relation to Alzheimer's disease [10]. The Horvath clock was explicitly designed to work across tissues [2]. Most telomere-length assays used in large cohorts are blood-specific in practice.
How consumer telomere and methylation tests are regulated
Consumer telomere and epigenetic age tests are typically offered as laboratory-developed tests. They are not FDA-cleared devices [11]. CLIA certification relates to laboratory quality standards. It does not establish that a test's aging claims are valid [12]. Our page on whether biological age tests are FDA approved covers this in detail.
On neutral criteria, the useful facts to look for on any report are:
which telomere method was used (qPCR, TRF or Flow-FISH)
whether a telomere figure is measured or methylation-estimated
the reference population behind any percentile
published repeat-measure variability
Our page on biological age test accuracy discusses repeatability.
Further reading
Books on cellular aging and molecular biology can provide the background to read telomere and methylation studies more critically. Useful titles explain chromosome structure, DNA replication, epigenetic regulation and basic biostatistics, and cite primary literature. Look for a recent publication date, an author with a research background, and clear separation between established findings and speculation.
A good introductory text will cover telomere biology and DNA methylation in separate chapters. That separation mirrors how differently the two are measured.
Frequently asked questions
Is telomere length an epigenetic clock?
No. Telomere length is a physical measurement of chromosome-end DNA, while epigenetic clocks are models built from DNA methylation levels [2]. DNAmTL bridges the two by predicting telomere length from methylation, but it is still a methylation-based estimate [1].
Why can telomere and clock results point in different directions?
After chronological age is accounted for, the two biomarkers correlate only weakly in cohort data. They reflect different mechanisms, carry different measurement noise, and are influenced differently by inherited baseline and blood cell composition.
Is DNAmTL more accurate than measured telomere length?
It depends on what "accurate" means. In the original paper, DNAmTL associated more strongly with age and with several outcomes than measured LTL did [1]. It remains an estimate of telomere length, not a direct measure of it.
Does a lab's CLIA certification validate a telomere test?
CLIA concerns laboratory quality standards [12]. It is not a judgment that a test's biological-age interpretation is scientifically validated.
Related on this site
Biological age tests: how epigenetic clocks work and what they mean
The Horvath clock: the 2013 multi-tissue epigenetic age estimator
GrimAge explained: how the mortality-trained epigenetic clock works
First, second and third generation epigenetic clocks, explained
Details of how pages are sourced are on our about page.
Last updated: 23 October 2026.
Not medical advice. This page reports published research and regulatory records for information only.
Research-only notice: Content on this site is provided for research and educational purposes only and is not intended for human use. It is not medical advice.
Translation notice: Some content on this site may have been machine-translated using Google Translate and may contain errors.