Your Lungs Are the First Checkpoint
When MSCs are injected into a vein in your arm, they do not travel directly to a knee, a hip, or anywhere else. They flow through the right side of the heart and into the dense capillary network of the lungs — the body's first filtration point for anything entering the venous circulation.
MSCs are large cells relative to the vessels they must cross. When Schrepfer and colleagues measured mouse MSCs held in suspension, they found a diameter of 15–19 micrometres [2]. To estimate how narrow the lung capillaries were, the same team injected fluorescent microspheres of graded sizes: 4-micrometre spheres passed through almost undetected, while 10- and 15-micrometre spheres were trapped in large numbers [2]. The cells are on the wrong side of that threshold.
Karp and Leng Teo's review of MSC trafficking sets out how much of the "homing" literature is complicated by exactly this — where infused cells actually end up depends heavily on how they were cultured, characterised and delivered [3].
The result is a genuine mechanical bottleneck. Understanding what happens there is essential for anyone considering intravenous cell therapy — and the answer is more interesting than "the cells get stuck".
What the Tracking Studies Found
A note on evidence, stated once and applying throughout this section: the biodistribution work below was done in rats and mice. There is no comparable study in humans, because tracking labelled cells through a living person's organs is not something that can be done in a clinical trial. This is the best evidence available on the question, and it is animal evidence. Where that matters for interpretation, it is said so.
In 2009, Fischer and colleagues at the University of Texas catheterised anaesthetised rats and infused four different labelled cell types, sampling arterial blood continuously and then imaging the harvested organs [1]. The finding that named the phenomenon:
The majority of MSCs were trapped inside the lungs following intravenous infusion.
Two details from that study are rarely quoted and are directly useful:
- Cell type mattered enormously. Neural stem cells and multipotent adult progenitor cells passed through the lungs about twice as readily as MSCs; bone-marrow mononuclear cells passed thirty times more readily [1]. Entrapment is a property of MSCs specifically, not of cell therapy in general.
- How the dose was given mattered. Splitting the infusion into two boluses increased pulmonary passage compared with a single bolus [1]. Blocking the adhesion molecule CD49d also increased passage [1].
Schrepfer's group found something similar from the vascular side: pre-treatment with the vasodilator sodium nitroprusside significantly reduced the number of trapped microspheres [2].
In other words, entrapment is not a fixed constant. It responds to how cells are prepared and how the infusion is run.
Entrapment Is Part of How It Works
This is the part of the story that is most often skipped, and it is the most important.
Cells lodged in the lung are not idle
In 2009, Lee and colleagues infused human MSCs into mice and used quantitative assays for human DNA and mRNA to work out what the cells were doing [4]. Most were trapped as emboli in the lung, and they disappeared with a half-life of about 24 hours — fewer than 1,000 cells turned up across six other tissues.
And yet the treatment worked. The trapped cells sharply upregulated TSG-6, an anti-inflammatory protein, which circulated onward. In a myocardial infarction model, intravenous MSCs reduced inflammation, reduced infarct size and improved cardiac function — but MSCs engineered to silence TSG-6 did not. Giving recombinant TSG-6 on its own reproduced much of the benefit [4].
The authors' conclusion is worth stating exactly: improvements after IV MSC infusion are "at least in part explained by activation of MSCs to secrete TSG-6" [4].
This reframes the whole question. The cells are not failing to arrive somewhere. Lodging in the lung is the step that activates them.
The cells are short-lived, and that appears to be by design
Eggenhofer and colleagues infused labelled MSCs into mice and recovered them from tissue over 72 hours [5]. One hour after infusion the label was mostly in the lungs; by 24 hours it was mostly in the liver. Viable donor cells could be re-cultured from the lungs for up to 24 hours and not after. No viable MSCs were found in any other organ at any timepoint — not even when liver injury was deliberately induced to attract them.
Their conclusion: MSCs are short-lived after IV infusion, viable cells do not pass the lungs, and long-term effects "must therefore be mediated via other cell types" [5].
That is a significant limitation on what IV cell therapy can be claimed to do — and it points directly at the mechanism that has since been worked out.
Clearance is the mechanism, not the failure
De Witte and colleagues tracked human umbilical cord MSCs infused into mice [6]. Within 24 hours most were dead and contained inside monocytes in the lungs and liver. The monocytes that had eaten them then migrated out of the lungs and were detectable systemically.
Critically, the monocytes were changed by the meal. After phagocytosing MSCs, human classical monocytes shifted to a non-classical CD14++CD16+CD206+ phenotype, expressed PD-L1 and IL-10, and produced less TNF-α. Those primed monocytes then induced Foxp3+ regulatory T cells [6].
Galleu and colleagues took this further in a murine graft-versus-host disease model, with a striking human parallel [7]. They showed that recipient cytotoxic cells actively kill infused MSCs by a perforin-dependent route, and that this killing is essential to initiate immunosuppression. When they looked at patients with GvHD who had received MSCs, only those with high cytotoxic activity against MSCs responded; those with low activity did not. Infusing MSCs that had been made apoptotic in advance reproduced the effect [7].
Their proposal was that patients might be selected by their ability to destroy MSCs — or simply given pre-apoptotic cells deliberately.
So the sequence that once looked like failure — trapped, killed, eaten, cleared — is, on the current evidence, the sequence by which the therapy acts.
What the Safety Record Actually Shows
Two independent systematic reviews have examined intravascular MSC administration specifically.
Lalu and colleagues (SafeCell, 2012) searched to June 2011 and included 36 studies covering 1,012 participants with ischaemic stroke, Crohn's disease, cardiomyopathy, myocardial infarction and graft-versus-host disease, plus healthy volunteers [8]. Meta-analysis of the randomised trials found no association between MSCs and acute infusional toxicity, organ-system complications, infection, death or malignancy. It did find a significant association with transient fever [8].
Thompson and colleagues (2020) updated this to 55 randomised controlled trials and 2,696 patients [9]. Their results, at full precision:
Heterogeneity was I [2]= 0% throughout, and no included trial was stopped early for safety concerns [9].
Two of these deserve to be read carefully rather than skimmed:
- Mortality was significantly lower in MSC-treated patients (RR 0.78, CI 0.65–0.94 — the interval excludes 1). These trials were not designed or powered as mortality trials and this should not be read as a survival claim for any individual indication. But it is the opposite of a safety signal.
- Thrombotic and embolic events showed no increase (RR 1.14, CI 0.67–1.95). This is the most direct reassurance available on the risk discussed in the next section.
Fever is the one consistent finding, in both reviews. It is transient, and it is what a patient should expect to be told about.
High intravenous doses in controlled trials
Weiss is the most demanding test on this list: 400 million cells in total, delivered intravenously to people whose lungs were already damaged, followed for two years, with no infusional toxicity [12].
There is no evidence-based hard ceiling at 50 million cells. That figure appears in marketing, not in the trial literature.
Two further points that cut against a simple "more is better" reading:
- CRATUS found the lower dose performed better. Physical performance improved preferentially in the 100M group: 6-minute walk, short physical performance battery and FEV₁ all improved in the 100M group (p = .01) and in neither the 200M nor the placebo group [10]. Immunological markers improved at both doses.
- Most of these trials used bone-marrow-derived cells. That matters, for the reason set out next.
The Question the Safety Data Does Not Yet Answer
The reassuring meta-analyses above are dominated by bone-marrow-derived products. Whether their conclusions transfer to newer products from other tissues is an open question, and the researchers who assembled much of that safety evidence are the ones raising it.
Moll and colleagues put it directly: meta-analysis has shown bone-marrow MSC infusion to be safe, and "it is not clear if this also applies to diverse new cell products derived from other sources, such as adipose and perinatal tissues" [14].
Tissue factor and IBMIR
The concern is coagulation, not viability. Different MSC products express different amounts of tissue factor (TF/CD142), the initiator of the extrinsic clotting pathway. On contact with blood, a high-TF product can trigger the instant blood-mediated inflammatory reaction (IBMIR) — a combined clotting and inflammatory response [14].
What the evidence shows:
- Source affects expression. In xeno- and serum-free culture, adipose- and umbilical-cord-derived MSCs expressed the highest tissue factor, dental pulp intermediate, and bone-marrow-derived MSCs the lowest [16].
- Route appears to matter. In the same study, patients receiving intravenous UC-MSC infusion showed a transient rise in D-dimer, while patients receiving the cells intrathecally did not [16].
- All MSC types are procoagulant in plasma. Adding clinical-grade bone-marrow, adipose or cord MSCs to normal plasma shortened clotting lag time and increased peak thrombin generation. Low-molecular-weight heparin reduced this dose-dependently, significantly at 0.2 U/mL and above [17].
- Case reports exist. Moll and colleagues note repeated reports of adverse events after intravascular MSC infusion "most commonly related to thromboembolic complications" with high-TF products, and the complications named in this literature include pulmonary embolism [15][16].
What is genuinely unresolved
This should not be overstated, and the same researchers are careful about it:
- Tissue factor is not a validated predictor. The study that found the source differences also found no correlation between TF expression and D-dimer or other coagulation measures, and concluded that TF "cannot be used as a solid biomarker to predict MSC-induced coagulant activity" [16].
- The rankings do not agree. Adipose MSCs showed the least thrombogenicity and the most sensitivity to heparin in the thrombin-generation assay [17]— which does not match the tissue-factor ordering [16].
- The meta-analytic data show no excess. Thrombotic and embolic events were not increased across 55 randomised trials (RR 1.14, CI 0.67–1.95) [9].
So the honest position is this: intravascular MSC products interact with the clotting system; source and manufacturing influence that interaction; no validated release test yet exists to measure it; and the randomised evidence to date shows no excess of clotting events. It is a question under active investigation rather than a demonstrated harm.
What the standards actually say
There is a common claim that the International Society for Cell & Gene Therapy sets a minimum viability threshold for clinical use. It does not. The ISCT position statement defines MSCs by three criteria: adherence to plastic in standard culture; expression of CD105, CD73 and CD90 with absence of CD45, CD34, CD14 or CD11b, CD79α or CD19 and HLA-DR; and differentiation to bone, fat and cartilage lineages in vitro [18]. There is no viability criterion and no potency criterion in it.
What exists instead is a live proposal to extend those criteria. Moll, Ankrum, Olson and Nolta have argued that because roughly half of clinical MSC applications are intravascular, the 2006 criteria are insufficient to establish whether a product is suitable for that route, and that tissue factor expression and hemocompatibility should be added as safety criteria before intravascular delivery [15].
That gap is the real state of the field, and it is more useful to a patient than an invented number.
Cell Viability: What It Tells You, and What It Does Not
Viability — the percentage of cells in a preparation that are alive — is a real and meaningful manufacturing measure. It varies with how cells are cultured, cryopreserved, thawed, shipped and stored.
What it legitimately tells you: how much of what you are being charged for is actually live cells. A dose labelled "100 million" at 60% viability contains 60 million live cells. Someone comparing two clinics' prices without comparing viability is not comparing like with like. Asking for the figure, and for the certificate of analysis that carries it, is entirely reasonable.
What it does not tell you: how you will do. There is no published evidence that higher-viability MSC preparations produce better clinical outcomes in patients, and no trial has randomised patients to high- versus low-viability product.
And the intuition that dead cells in the lung are the danger does not survive contact with the evidence. As set out above, the clearance of dead and dying MSCs by monocytes and macrophages is currently understood as the route by which the therapy exerts its immunomodulatory effect — producing IL-10, PD-L1 and regulatory T cells [6], with MSC apoptosis actually required for immunosuppression to begin [7].
There is a real distinction between apoptotic cell death (controlled, immunologically quiet or actively immunosuppressive) and necrotic cell death (membrane rupture, release of damage-associated molecular patterns that drive inflammation). That distinction is well established in general immunology. What has not been established is that clinical MSC preparations with lower viability contain predominantly necrotic rather than apoptotic cells, or that any such preparation has caused pulmonary harm in a patient.
So: ask about viability because it tells you what you are buying. Do not accept it as a safety guarantee, from us or from anyone else.
What Makes Intravenous Administration Safer
Infusion technique
How the dose is delivered demonstrably changes where cells go — splitting a dose into two boluses increased pulmonary passage in the rat model [1], and vasodilation reduced trapping in the mouse model [2]. Controlled, monitored infusion rather than rapid bolus is the sensible reading of this, and it is standard practice.
Aggregate control
Clumped cells present a larger effective particle than single cells. Verification of a single-cell suspension and appropriate filtration before infusion are basic quality-control steps for any intravascular cell product [14].
Product characterisation
Beyond the ISCT identity criteria [18], the assessment argued for by the field for intravascular products is hemocompatibility and tissue factor [15]. No validated release assay for this is yet in routine use — which is precisely why it is worth asking a provider what they measure.
Patient screening
Existing cardiac, pulmonary or coagulation disorders are the circumstances in which any of the above matters most, and they warrant specific medical assessment before intravenous cell therapy.
On anticoagulation: heparin has been shown in the laboratory to reduce MSC-driven thrombin generation [17], and co-administration has been proposed for intravascular protocols [14]. This is a decision for the treating physician, made for an individual patient. Nothing on this page is a reason to start, stop or alter any prescribed medication — that decision belongs to the doctor who prescribed it.
Questions to Ask Any Provider
- "What is the viability percentage, and how is it measured?" — flow cytometry (7-AAD or Annexin V) is more objective than manual trypan blue counting. Ask for the number, and be aware that no international body sets a threshold for it.
- "Will I receive a certificate of analysis for my specific dose?" — cell count, viability, sterility, mycoplasma, endotoxin, and identity markers (CD73+/CD90+/CD105+).
- "What tissue are the cells from, and do you assess the product for effects on coagulation?" — this is the question the current safety literature says matters for intravenous delivery [15][16].
- "What is the infusion protocol, and what monitoring happens during it?"
- "Are cells checked for aggregation and filtered before infusion?"
- "What evidence is there that intravenous cell therapy helps my specific condition?" — the first five questions are about product quality. Only this one is about whether the treatment works.
The Bottom Line
The pulmonary first-pass effect is real, well documented, and worth understanding before intravenous cell therapy. But the framing that treats it as an obstacle to be overcome has not aged well.
The current evidence says that MSCs given intravenously lodge in the lung, become activated there, secrete signalling proteins that travel onward, are killed and consumed by the recipient's own immune cells, and are cleared within about a day — and that this sequence is how the therapy appears to work, not a way in which it fails [4][5][6][7].
The safety record for the intravenous route is genuinely good, and can be stated with numbers: across 55 randomised trials and 2,696 patients, no increase in serious adverse events, infection, thrombotic or embolic events or malignancy; a significant excess of transient fever; and lower mortality [9].
Where real uncertainty remains, it is not about how many cells are alive in the vial. It is about how a given cell product interacts with the clotting system when it enters the bloodstream — a question that depends on tissue source and manufacturing, that has no validated release test yet, and that the field is actively working on [14][15][16][17].
A patient is well served by asking what is in the vial and how it was made. A patient is better served by also asking what evidence exists that it will help them.