A striking new transplantation study finds that old hearts acquire younger molecular signatures in young bodies. The experiment is fascinating — but “the heart became younger” is a much stronger conclusion than the data currently support.

A new bioRxiv preprint by Jesse Poganik, Vadim Gladyshev and colleagues has an irresistible premise. Take a heart from an old organism, transplant it into a young one, and ask: does the heart remain old, or does the young body make it young again?

The authors report the latter. In mice, old hearts transplanted into young recipients shifted toward younger patterns of DNA methylation and gene expression, while young hearts transplanted into old mice shifted in the opposite direction. In a small human dataset, methylation-based “epigenetic clocks” likewise appeared to make transplanted hearts resemble the age of the recipient more than the donor. The authors conclude that transplanted hearts “assimilate” the recipient’s biological age and even suggest that rejuvenation of old donor organs might eventually expand the transplantable organ pool.

There is almost certainly interesting biology here. In particular, the reciprocal molecular changes in the mouse experiment are difficult to dismiss as random noise.

But there is a large gap between:

“The age of the host changes molecular states inside a transplanted heart”

and:

“An old heart becomes biologically young.”

Several features of the study can make the second statement appear much stronger than the evidence warrants.

1. The human “rejuvenation” signal may be partly generated by the clocks themselves

Many epigenetic clock estimators systematically underestimate age of the old people and overestimate age of the young people. This is called prediction shrinkage. This is the most important problem because, in principle, it can create exactly the pattern the authors interpret as rejuvenation even if the recipient has no effect whatsoever. E. g. Khoury et al. showed that epigenetic clocks systematically underestimate the age of older subjects.

Suppose the true donor age is (A), but the clock behaves approximately like:

where (β<1).

This means the clock compresses the age range: very old samples are predicted too young and very young samples are predicted too old.

The transplant paper uses an age deviation concept of the form:

Substituting the first equation gives

Because (β-1) is negative, older donor hearts automatically get more negative AgeDev values.

In plain English:

the older the donor, the more “rejuvenated” the heart can appear simply because the clock underestimates extreme old ages.

This is closely related to regression toward the mean, although “prediction shrinkage” or “age compression” is the more precise term here.

Imagine a clock whose predictions are pulled 30% toward an average training age of 50. For an 80-year-old heart it predicts:

The heart now looks 9 years younger.

For a 20-year-old heart:

The heart now looks 9 years older. We have just generated an apparent 18-year difference between old and young hearts without rejuvenating or aging anything.

Now combine that with the transplant design

This is where things become particularly problematic. The interesting human cases are deliberately heterochronic:

old donor → young recipient

and

young donor → old recipient.

That means donor age and recipient age are strongly negatively correlated. Now take the clock artifact we just described. An old donor heart automatically receives a negative age deviation. But old donor hearts happen to be in the young recipient group. A young donor heart automatically receives a positive age deviation. But young donor hearts happen to be in the old recipient group.

The resulting pattern is:

young recipient → apparently rejuvenated heart

old recipient → apparently aged heart

which is exactly the biological effect the study is trying to demonstrate.

Importantly, the recipient need not have caused anything for this association to appear. It is just a specific interaction between:

heart-specific clock miscalibration + compression of extreme predictions + subtraction of donor age + anticorrelation between donor and recipient ages.

The human methylation arm is also extremely small — 11 recipients in the analysis summarized from the preprint—and the three clocks do not produce perfectly uniform evidence. That makes such calibration effects particularly consequential.

And there is no pre-transplant measurement

Even if an old donor heart looks young after transplantation, we do not know that it became young. Perhaps it already looked young before transplantation.

This matters especially because older donor hearts are not random old hearts. They are selected for transplantation precisely because they appear sufficiently healthy. Modern donor selection explicitly evaluates donor risk factors before accepting a heart.

An unusually healthy 60-year-old donor heart might already differ biologically from the average 60-year-old heart.

Without a baseline measurement, these two possibilities cannot be distinguished:

versus

How this could be fixed

For each epigenetic clock, researchers could take a large independent collection of non-transplanted hearts and estimate:

Instead of comparing clock age directly with donor chronological age, the relevant quantity would then be:

In other words:

Is this transplanted 70-year-old heart younger than an ordinary 70-year-old heart according to this particular clock?

They could also feed the actual donor and recipient ages into a null simulation using the observed heart calibration curve. That would show how much apparent “assimilation” should arise even if recipient age has zero biological effect.

Better still would be pre-transplant and serial post-transplant measurements from the same organ.

Until this is done, I would regard the human epigenetic evidence as intriguing but far from demonstrating rejuvenation.

2. A transplanted heart gradually contains cells from the recipient

There is a second problem that applies not only to clock age but to the study’s broader methylation and transcriptomic results: when researchers grind up a piece of heart and sequence it, they are not measuring a pure population of cardiomyocytes, they are measuring a mixture containing cardiomyocytes, fibroblasts, endothelial cells, smooth-muscle cells, immune cells and other cell types. And after transplantation, some of those cells belong to the recipient.

A 2024 single-nucleus study of human cardiac allografts used naturally occurring genetic variation to distinguish donor cells from recipient cells. It found that donor-resident macrophages were progressively lost while recipient-derived macrophages populated the transplanted heart.

Now consider an old heart transplanted into a young recipient.

Months later the biopsy may contain:

The bulk methylome will move toward a younger profile, the bulk transcriptome can move toward a younger profile too. Yet not a single old cardiomyocyte necessarily needs to have become younger. The reverse can happen when a young heart is transplanted into an old recipient.

This is especially important because the authors treat concordant methylation and RNA results as mutually reinforcing evidence. But if both are measured from the same changing mixture of cells, they are not completely independent confirmations.

The ideal experiment would perform single-cell or single-nucleus profiling, identify each nucleus as donor- or recipient-derived from its genotype, and then ask:

Do the original donor cardiomyocytes themselves change?

If old donor cardiomyocytes, old donor fibroblasts and old donor endothelial cells each shift toward young molecular states inside a young recipient, the rejuvenation argument becomes dramatically stronger.

3. A younger molecular signature is not the same thing as a younger heart

Suppose the first two objections are completely resolved.

Imagine that we isolate old donor cardiomyocytes and conclusively show that hundreds of age-associated CpGs and thousands of transcripts shift toward the pattern found in young cardiomyocytes.

That would be a fascinating result. But! It WOULD NOT demonstrate that the heart itself has become young.

An old heart contains many forms of accumulated change:

• fibrosis and remodeled extracellular matrix
• collagen crosslinks
• loss of cardiomyocytes
• coronary atherosclerosis
• microvascular damage
• somatic mutations
• mtDNA mutations and heteroplasmy
• altered conduction tissue
• accumulated protein damage
• changes in mechanical reserve
• etc.

A methylation clock measures methylation, RNA sequencing measures gene expression. Neither directly tells us whether all those other trajectories reversed.

This distinction becomes particularly important when looking at the mitochondrial result. The preprint reports a shift in expression of pathways related to mitochondrial translation, oxidative phosphorylation and electron transport in the expected direction. That is biologically interesting.

But increased expression of oxidative-phosphorylation genes is not the same as demonstrating:

• higher mitochondrial respiration
• more ATP production
• lower ROS
• fewer mtDNA mutations
• changes in fusion/fission dynamics
• restoration of youthful mitochondrial structure.

Gene expression is often adaptive. A cell can increase expression of a pathway precisely because that pathway is functioning poorly.

The recent hypoxia experiment is a useful illustration of the broader point. One month of intermittent hypoxia moved epigenetic clocks toward older values in old animals, and returning the animals to normal oxygen shifted them back.

It would be strange to conclude that one month of hypoxia literally created several years of comprehensive organ aging and that normoxia subsequently removed all that accumulated damage.

A more sensible conclusion is that at least part of what epigenetic clocks measure is a plastic physiological state associated with age.

4. The mouse heart is not doing the job of a normal heart

There is another major difference between the spectacular mouse experiment and a human heart transplant.

The mouse model uses heterotopic heart transplantation.

In the standard version of this model, the recipient keeps its native heart. The transplanted heart is vascularized and continues beating, but it does not pump the animal’s systemic circulation.

Indeed, this exact procedure is widely used experimentally as a model of mechanical unloading. The donor heart receives coronary perfusion but has dramatically reduced ventricular filling and essentially no normal systemic workload.

Mechanical unloading itself profoundly remodels the heart. Experimental heterotopic transplantation can produce rapid myocardial atrophy; previous work has documented very large reductions in ventricular mass within days to weeks. Other studies show changes in fibrosis, glucose metabolism and cardiac mechanics as a function of unloading.

So the transplanted mouse heart is simultaneously experiencing:

• recipient age
• transplantation
• ischemia/reperfusion
• denervation
• altered immune exposure
• and extreme mechanical unloading.

To their credit, the authors use age-matched transplant controls. That is an important strength and means that one cannot simply say “everything is caused by transplantation.”

But those controls do not necessarily remove interactions.

For example, unloading in an old host may not equal unloading in a young host. The same applies to inflammation, metabolism and denervation.

Therefore the mouse experiment convincingly asks:

How does recipient age affect an unloaded transplanted heart?

It does not yet perfectly reproduce:

What happens when an old heart becomes the sole load-bearing heart of a young organism?

5. The human functional result does not specifically measure the heart

The paper tries to go beyond molecular biomarkers by examining functional outcomes in human heart-transplant recipients.

This sounds like exactly the validation one would want. The problem is what “function” means.

Exercise capacity and maximal oxygen consumption are whole-body phenotypes:

An older recipient normally has lower exercise capacity than a younger recipient for many reasons even if both received identical young hearts.

This is particularly well established after heart transplantation. Exercise capacity reflects a combination of central cardiac factors and peripheral factors including skeletal-muscle mass and oxidative capacity, vascular function, pulmonary function, deconditioning, immunosuppressive therapy and other comorbidities.

So finding that a transplanted heart in a 70-year-old person is associated with poorer recipient exercise performance than a heart in a 25-year-old does not tell us that the first heart itself has become biologically 70.

Much stronger validation would involve heart-specific measurements such as myocardial strain, diastolic reserve, stress hemodynamics, coronary microvascular function, cardiac MRI fibrosis, graft vasculopathy or long-term graft performance.

Best of all would be to show that the same heart whose molecular age moved farther toward the recipient also underwent a corresponding heart-specific functional change.

The current functional analysis does not establish that link.

6. Transplanted tissues do not universally adopt the recipient’s age

There is good prior evidence that the environment of a host can influence aging-associated molecular signatures.

For example, a 2024 study of hematopoietic stem-cell transplantation found that some biological epigenetic clocks in donor-derived blood cells accelerated in older recipients and decelerated in younger recipients. Importantly, that study had longitudinal donor and recipient measurements and explicitly modeled chronological age, cell composition, batch and time after transplantation.

So the basic idea behind the heart study is biologically plausible.

But it is not universal.

A particularly relevant counterexample comes from lung transplantation.

Researchers studied airway epithelial cells from 28 lung-transplant recipients one year after transplantation. Epigenetic age strongly tracked the donor/allograft age:

while there was essentially no association with recipient age (p=0.96)

Interestingly, severe graft injury did increase apparent epigenetic age, showing that the clock could respond to the new environment without simply adopting the recipient’s chronological age.

This suggests a more nuanced picture:

some age-associated molecular states are cell-intrinsic; some respond to systemic environment; some respond to local injury; and the balance varies by tissue and cell type.

Conclusion

The age of the host appears capable of substantially remodeling age-associated molecular states in transplanted cardiac tissue. How much of this represents genuine rejuvenation of the original donor cells—and how much corresponds to reversal of durable structural and functional aging—remains unknown.

The biggest uncertainty lies in the human evidence. Because cardiac tissue is notoriously miscalibrated by many generic epigenetic clocks, and because donor and recipient ages are deliberately anticorrelated in heterochronic transplantation, the reported human “assimilation” can in principle be amplified—or partly generated—from clock behavior alone.

Before claiming that an old human heart actually becomes young, the decisive experiments would be heart-specific clock calibration, paired pre- and post-transplant sampling, separation of donor from recipient cells, single-cell or single-nucleus measurements, and direct measurements of cardiac structure and reserve.

If those experiments show that an old donor cardiomyocyte loses age-associated methylation, restores youthful transcription, improves mitochondrial function, reduces fibrosis and gains functional reserve inside a young host, then “heart rejuvenation” would be an appropriate description.

For now, “host-dependent molecular remodeling” fits the evidence considerably better.