Biodistribution and clearance of infused stem cells

The original rationale for stem cell therapy was structural. Cells would be delivered to damaged tissue, take up residence there, differentiate into whatever the tissue was short of, and rebuild it. That model is intuitive, it is the one most patients are given, and the tracking data have not supported it for intravenously administered mesenchymal stromal cells.

The problem was stated precisely in a 2009 study that set out to examine what its authors called a paradox: intravenously infused human cells can enhance tissue repair without significant engraftment. Using quantitative assays for human DNA and messenger RNA in mice, the group found that after infusion of 2 million cells, most were trapped as emboli in the lung, the lung population disappeared with a half-life of about 24 hours, and fewer than 1,000 cells appeared across six other tissues combined.^[4] Benefit after myocardial infarction was still observed. Something was producing an effect, and it was not a population of cells living in the heart.

Resolving that paradox has occupied much of the field since, and the answer has turned out to be more interesting than the original model. The cells are not the therapy in the way a transplanted organ is the therapy. They are closer to a transient stimulus that provokes a durable response in the recipient's own immune system. Understanding this changes what a patient should reasonably expect, including how quickly an effect should appear and what it means that the cells cannot be found afterward.

In one sentence

Biodistribution describes where administered cells travel and accumulate in the body over time, and clearance describes how quickly they are removed; for infused mesenchymal stromal cells both are measured in hours and days, not months.

The first constraint is mechanical. Cultured mesenchymal stromal cells are large relative to the vessels they must pass through, and this determines their initial distribution regardless of the target. The point has been demonstrated directly by intravital microscopy of rat muscle microcirculation: when cells were delivered into an artery, bypassing the lungs entirely, 92 percent still arrested at the precapillary level on first pass and interrupted blood flow in the feeding arteriole. Measurement of cortical tension showed the cells were no stiffer than circulating mononuclear cells, so the arrest is a consequence of size rather than rigidity.^[2] For intravenous delivery the equivalent obstacle is the pulmonary capillary bed, which is described in more detail alongside the practical consequences for cell dose and route selection.^[1]

The second constraint is that most cells never leave that first site alive. In mice given labeled cells intravenously, isotope counting placed most of the label in the lungs at one hour and mainly in the liver at 24 hours, but re-culture of the tissue told a different story: viable donor cells were recoverable from the lungs for up to 24 hours and then disappeared, and no viable donor cells were found in any other organ at any timepoint. Deliberately injuring the liver did not draw viable cells to it.^[3] The label moved. The living cells did not.

That distinction matters for reading the human data, which is necessarily imaging-based. Four patients with advanced cirrhosis received cells labeled with indium-111 and were imaged repeatedly over ten days. Radioactivity accumulated first in the lungs, then shifted: lung signal fell from 27 to 33 percent immediately after infusion to 2 to 5 percent by day 10, liver signal rose from under 3 percent to 13 to 17 percent, and spleen signal rose from 2 to 10 percent to 30 to 42 percent, exceeding the liver in every patient.^[5] This is the clearest picture available of what happens in a person, and it is genuinely informative. It should be read with the caveat the animal work established: a radiolabel reports where the isotope is, not whether it is still inside a living cell. Signal arriving in the spleen and liver is at least as consistent with clearance of cellular debris by the reticuloendothelial system as with cells migrating there and surviving.^[6]

Three independent lines of evidence converge on an answer measured in hours to days.

Quantitative DNA assays in mice gave a lung half-life of roughly 24 hours after intravenous infusion.^[4] Re-culture experiments found viable donor cells in the lungs up to 24 hours and none thereafter.^[3] Intravital microscopy of cells that had lodged in muscle microvasculature found the population reduced to 14 percent of its initial number by day three, through cell death in place; the small surviving fraction was spread on the luminal surface of the vessel at one day and integrated into the microvascular wall by day three, occupying a perivascular position.^[2] That residual fraction is the only part of the original model that survives contact with the data, and it is a small one.

A study tracking viable and dead human umbilical cord cells separately in mice put numbers on the transition. Viable cells were present in the lungs immediately after infusion. Twenty-four hours later the majority were dead, and were found in the lungs and liver contained inside monocytic cells.^[7] By one day, in other words, the relevant biological object is no longer the administered cell. It is the recipient immune cell that has consumed it.

This is the honest answer to how long stem cells stay in the body: as living, identifiable donor cells, on the order of a day in the lung, with a small perivascular remnant detectable for a few days more. Any claim that infused cells persist for months, continue dividing, or remain available to be recruited later is not supported by the tracking literature.

If the cells are gone within a day, the durable effects reported in trials must be carried by something else. The current answer is that the recipient's own immune system is reprogrammed during the brief window in which the cells are present and dying, and that the reprogrammed host cells are what persist.

The clearest description of the handoff comes from the umbilical cord cell tracking study. Dead donor cells at 24 hours were found inside monocytes of a predominantly non-classical phenotype, in the lungs and liver, and monocytes containing donor material were also detectable systemically, having migrated out from the lung. Laboratory experiments confirmed the causal direction: human classical monocytes that phagocytosed the donor cells shifted to a non-classical phenotype, began expressing programmed death ligand-1 and interleukin-10, reduced their production of tumor necrosis factor alpha, and, once primed this way, induced the formation of Foxp3-positive regulatory T cells.^[7] That is a complete mechanistic chain running from an infused cell that dies within a day to a regulatory T cell population that can persist far longer.

A separate line of work made the stronger claim that the death of the cells is not merely incidental but required. In a mouse model of graft-versus-host disease, infused cells were actively driven into apoptosis by recipient cytotoxic cells through a perforin-dependent mechanism, and blocking that process prevented the immunosuppressive effect. The group then examined patients with graft-versus-host disease who had received cell infusions and found a striking parallel: only those whose immune cells showed high cytotoxic activity against the infused cells responded clinically, while those with low activity did not. Infusing cells that had already been made apoptotic in the laboratory substituted for the recipient's own killing activity, and recipient phagocytes engulfing them produced indoleamine 2,3-dioxygenase, which was necessary for the immunosuppressive effect.^[8]

The implication is counterintuitive and worth stating plainly: on this evidence, a cell therapy may work because the cells die, not despite it, and a patient whose immune system fails to attack the infused cells may be less likely to respond. It also suggests why results vary so widely between patients receiving nominally identical products, a variability discussed further in our article on the safety and evidence base for stem cell therapy.

Macrophage reprogramming through soluble signals is a parallel route to the same kind of outcome. In a mouse sepsis model, infused stromal cells reduced mortality by acting on host macrophages through prostaglandin E2, driving them to increase interleukin-10 production; the effect required the host macrophages, not the donor cells, to do the work.^[9]

Being trapped in the lung is not a passive state. The same study that established the 24 hour half-life found that the cells lodged there actively upregulated multiple genes, with a large increase in the anti-inflammatory protein TSG-6. The causal role was then tested in both directions: cells in which TSG-6 had been knocked down by silencing RNA lost the benefit after myocardial infarction, and administering recombinant TSG-6 protein alone, with no cells at all, reduced inflammation and infarct size.^[4] A protein reproduced the effect of the therapy.

Cells also release membrane-bound vesicles carrying protein and nucleic acid cargo. One study found that mesenchymal stromal cells use extracellular vesicles to offload depolarized mitochondria to macrophages, which enhances the macrophages' bioenergetics, while simultaneously shuttling regulatory microRNAs that suppress macrophage activation.^[10] The recipient cell receives both a metabolic and a regulatory instruction from material the donor cell exported.

What the cells secrete is not fixed. It depends on the inflammatory environment they encounter, a phenomenon described as licensing: exposure to inflammatory signals is a multistep process that switches the cells into an actively immunosuppressive state, and cells that have not been licensed behave differently.^[11] This is one reason the same product can produce different results in different patients, and it is consistent with the observation that response tracks the recipient's immune activity rather than the dose administered.

Taken together, these findings have prompted serious investigation of whether the vesicles alone could serve as the therapy, removing the living cell from the equation. That work is real but early, and an international position paper has set out the manufacturing, characterization, and regulatory requirements that would have to be met before vesicle-based products could be used routinely in patients.^[13] TrueCell's current use of this class of product is described on our exosome therapy page, and the underlying cell biology on what a mesenchymal stem cell is.

The mechanism above predicts a specific shape for the timeline, and it is not the shape most patients expect. The administered cells are cleared within days. The immune reprogramming they trigger, including monocyte polarization and regulatory T cell induction, develops over the following weeks. Any structural consequence of a quieter inflammatory environment, such as change in a joint, develops over months. The infusion is therefore the beginning of a process rather than the delivery of a result, and an absence of change in the first days after treatment carries no information about whether it will work.

Trial design reflects this. Controlled studies in knee osteoarthritis assess their primary outcomes at 6 and 12 months rather than at weeks, and a randomized trial followed for four years reported pain scores still improved against control in both cell-treated groups at that point, with function improved in the lower-dose group.^[15] Those data are discussed further in our article on dose and delivery route. In graft-versus-host disease, where the target is acute and immunological rather than structural, response is assessed at 28 days.

That last setting also supplies the most important cautionary result in this area. Remestemcel-L, a manufactured allogeneic product, was tested against placebo added to second-line therapy in 260 patients with steroid-refractory acute graft-versus-host disease, using eight infusions over four weeks. It did not meet its primary endpoint of durable complete response, at 35 percent against 30 percent for placebo. Post hoc analyses found higher response rates in patients with liver involvement, in those with the highest-grade disease, and in pediatric patients, and the product was safe and well tolerated, but the trial as designed was negative.^[12]

A coherent mechanism is not the same thing as a demonstrated clinical benefit. The biodistribution and handoff literature explains how an effect could occur despite the cells disappearing, and it is well supported. Whether that effect is large enough to change outcomes in a given disease is a separate question, answered only by controlled trials in that disease, and the answers so far are mixed. The translational obstacles specific to delivery have been reviewed in detail elsewhere.^[14]

How long do stem cells stay in your body?

As living donor cells, on the order of a day. Quantitative assays in mice found a lung half-life of about 24 hours after intravenous infusion, with fewer than 1,000 cells reaching six other tissues combined.^[4] Re-culture studies recovered viable cells from the lungs for up to 24 hours and from no other organ at any time.^[3] A small fraction that lodges in microvasculature survives longer, falling to about 14 percent of the initial number by day three and taking up a perivascular position.^[2] Radiolabel signal remains detectable in the spleen and liver for at least ten days in patients, but that signal does not establish that the cells are alive.^[5]

How long does stem cell therapy take to work?

Longer than the cells last, because the cells are not what produces the sustained effect. Immune reprogramming develops over weeks as monocytes that have consumed the infused cells polarize and induce regulatory T cells.^[7] Structural change, where it occurs, develops over months, which is why controlled trials in osteoarthritis assess outcomes at 6 and 12 months. No conclusion should be drawn from the first days after an infusion.

If the cells die, is the treatment doing anything?

The death of the cells appears to be part of the mechanism rather than a failure of it. In a mouse model, infused cells had to be driven into apoptosis by recipient cytotoxic cells for immunosuppression to occur, and in patients only those with high cytotoxic activity against the infused cells responded clinically.^[8] Separately, the anti-inflammatory protein TSG-6 secreted by cells trapped in the lung was shown to be necessary for benefit, and recombinant TSG-6 alone reproduced that benefit without any cells.^[4]

Do the cells travel to my injury?

Largely no, when administered intravenously. Cells arrest at the first capillary bed they meet because of their size, and this occurs even when the lungs are bypassed by arterial delivery.^[2] Inducing injury in an organ did not recruit viable cells to it in animal studies.^[3] Where the target is a specific structure, such as a joint, it is injected directly rather than relied upon to attract circulating cells, a decision covered in our article on dose and delivery route.

How long do the results last?

This depends on the indication and is a separate question from how long the cells last. Randomized data in knee osteoarthritis show pain scores still improved against control four years after a single injection, with function improved in the lower-dose group.^[15] In acute immunological disease the relevant window is weeks, and a phase 3 trial in steroid-refractory graft-versus-host disease did not meet its primary endpoint at all.^[12] Durability should be discussed for the specific condition being treated rather than as a general property of cell therapy.

What does TrueCell tell patients about this?

That the cells are cleared within days, that the intended effect is a change in the recipient's own inflammatory and immune activity rather than replacement tissue grown from donor cells, and that any benefit should be assessed over months rather than days. A board-certified physician reviews history, imaging, and laboratory results before recommending a protocol, and states what the evidence supports for that indication and what it does not. Programs by condition are listed on the treatments page.