Cell dose and delivery route in stem cell therapy

A cell therapy protocol is not fully described by naming the cell type. Two further parameters determine what a patient is actually exposed to. The first is dose: the number of viable cells administered in a single session. The second is route: the anatomical compartment into which those cells are introduced, whether a peripheral vein, a joint cavity, or the cerebrospinal fluid. A protocol described only as "mesenchymal stem cell therapy" specifies neither, and the two protocols that share that name may differ from one another by an order of magnitude in cell number and by an entire physiological compartment in delivery.

The two parameters are not independent. Route sets a practical ceiling on dose, because the receiving compartment has a fixed tolerance for injected volume. A knee joint accommodates a few milliliters. The lumbar cerebrospinal space accommodates less. The venous circulation accommodates a much larger volume, which is one reason the largest reported cell numbers appear in intravenous protocols. Route also determines where the cells are first held, which turns out to matter more than the intuitive picture of an infusion suggests.^[1]

In one sentence

Dose is the count of viable cells administered per session, reported either as a fixed number per patient or adjusted to body weight in cells per kilogram; route is the compartment those cells enter, which determines where they are retained and how many reach the intended target.

Two conventions appear in the literature. The first is a fixed dose per patient, for example 100 million cells, used most often when the target is a defined anatomical structure whose size does not scale with body mass. The second is a weight-adjusted dose in cells per kilogram, inherited from conventional pharmacology, used most often in systemic indications where the relevant exposure is thought to be a whole-body one.

The scale of the difference between these conventions is easy to underestimate. In the phase II study of mesenchymal stromal cells for steroid-resistant acute graft-versus-host disease, 55 patients received a median of 1.4 million cells per kilogram, which for a 70 kilogram adult is on the order of 100 million cells in total.^[7] A fixed-dose intra-articular protocol for knee osteoarthritis may use 10 million cells in a single joint.^[9] Both are legitimate protocols. They are not comparable quantities.

The most systematic accounting of what has actually been administered comes from an analysis of 914 mesenchymal stem cell trials registered through 2018. Route and dosing information had to be extracted from individual registry records rather than downloaded in bulk, which is itself an indication of how inconsistently these parameters are reported. Intravenous injection was the most common route, accounting for 43 percent of all trials, and the median intravenous dose was 100 million cells per patient per dose. Among intravenous trials that reported positive outcomes, minimum effective doses fell between 70 and 190 million cells per patient in 14 of 16 trials, with the remaining two administering at least 900 million.^[5]

That range is the honest answer to the question of how many cells a treatment contains. There is no single figure, and the figure that applies depends on the route, the indication, and the manufacturing convention of the group reporting it.

Intravenous infusion is the most common route in registered trials because it is the least invasive and the most reproducible to administer.^[5] It is often described to patients as a way of sending cells throughout the body to find damaged tissue. The circulatory anatomy does not support that description, and this is one of the better-documented findings in the field.

Blood returning from a peripheral vein passes to the right side of the heart and from there into the pulmonary circulation before it reaches the systemic arteries. The pulmonary capillary bed is a fine filter, and a cultured mesenchymal stromal cell is comparatively large. In an anesthetized rat model with catheters in the jugular vein and carotid artery, allowing labeled cells to be counted as they reached the arterial circulation, the majority of infused mesenchymal stromal cells were trapped in the lungs. Neural stem cells and multipotent adult progenitor cells passed at roughly twice the rate, and bone marrow mononuclear cells, which are smaller, at roughly thirty times the rate. Infrared imaging showed the retained cells distributed evenly across all lung fields. The authors concluded that a therapeutically questionable number of cells reach the arterial system acutely after intravenous administration.^[2]

The same study identified two variables that modestly changed the outcome. Blocking the adhesion molecule CD49d significantly increased pulmonary passage, and administering the infusion as two boluses rather than one also increased passage.^[2] Neither eliminated the effect.

A separate study followed the fate of the retained cells rather than the fraction that escaped. Labeled mesenchymal stromal cells were infused into mice and organs were harvested at intervals for both isotope counting and re-culture, which distinguishes intact viable cells from residual label. One hour after infusion the majority of label was in the lungs. At 24 hours the label was found mainly in the liver, but re-culture 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. Inducing ischemia-reperfusion injury in the liver did not draw viable cells to it. The authors suggested that the label reaching the liver represented cell debris, and concluded that long-term regenerative and immunomodulatory effects of infused cells must therefore be mediated through other cell types.^[3]

This has a direct interpretive consequence. Whatever intravenous mesenchymal cell therapy does, it is unlikely to work by the infused cells arriving at a distant injury and rebuilding it, because in these models the cells neither arrive nor persist. It is more consistent with a transient signaling event: cells lodge in the lung, interact with host immune cells, release soluble factors and vesicles, and are cleared.^[4][15] That is the paracrine mechanism described in more detail in our article on what a mesenchymal stem cell is, and the fate of the cells after that first pass, including what the recipient's immune system does with them, is the subject of our article on biodistribution and clearance.

If the target is a defined anatomical structure, delivering cells directly into that structure bypasses the pulmonary filter entirely. Two regional routes account for most of this work.

Intra-articular injection places cells into a joint cavity, and is the standard route for osteoarthritis. A dose-escalation trial in 18 patients with knee osteoarthritis administered autologous adipose-derived cells at 10 million, 50 million, or 100 million per knee, with three patients in each phase I cohort and nine further patients receiving the highest dose in phase II. There were no treatment-related adverse events. In the high-dose group the WOMAC score improved at six months, the size of the cartilage defect decreased and cartilage volume increased in the medial femoral and tibial condyles on imaging, arthroscopy confirmed the reduction in defect size, and histology showed hyaline-like cartilage.^[8] Structural change confirmed on three independent modalities is a stronger form of evidence than symptom scores alone, though the cohorts were very small. The clinical context for this route is covered on our arthritis and orthopedic pages.

Intrathecal injection places cells into the cerebrospinal fluid through a lumbar puncture, and is used when the target is the central nervous system, which the pulmonary filter and the blood-brain barrier otherwise make difficult to reach. In a phase 1/2 open-safety study, 15 patients with multiple sclerosis and 19 with amyotrophic lateral sclerosis received a mean of 63.2 million autologous mesenchymal stromal cells, administered intrathecally in 34 instances and intravenously in 14. Adverse effects were transient fever in 21 patients and headache in 15, with no major adverse effects across follow-up of up to 25 months. In nine cases the cells were magnetically labeled, and imaging visualized them in the occipital horns of the ventricles, indicating migration through the meninges and subarachnoid space.^[11] A later trial in amyotrophic lateral sclerosis combined intrathecal and intramuscular administration of a modified cell product.^[13]

Route selection is therefore a question of target anatomy rather than of one route being inherently superior. A joint is reachable by needle and a systemic immune process is not, so the joint is injected and the immune process is approached intravenously. It is worth noting that reaching the target does not by itself establish benefit: the largest randomized trial of intravenous mesenchymal cells in multiple sclerosis, 144 patients in a placebo-controlled crossover design, was safe and well tolerated but did not meet its primary endpoint of reducing gadolinium-enhancing lesions at week 24, and its authors stated plainly that the study does not support the use of bone marrow-derived cells to treat active multiple sclerosis.^[12] That result is discussed in the disease context on our multiple sclerosis page.

The intuition is that more cells should mean more effect, and cell counts are frequently promoted on that assumption. The controlled evidence does not support a simple linear relationship, and on the question of what it does support the literature is genuinely divided. The disagreement is worth presenting rather than resolving artificially.

The clearest direct comparison comes from a randomized phase I/II trial in 30 patients with knee osteoarthritis, followed for four years. Patients received hyaluronic acid alone, hyaluronic acid with 10 million autologous bone marrow cells, or hyaluronic acid with 100 million. Both cell groups improved on pain scores against control, and the low-dose group improved on WOMAC against control, with no adverse effects reported in either. The finding relevant here is the comparison between the two active arms: no clinical differences between the cell-receiving groups were found.^[9] A tenfold difference in cell number produced no measurable clinical separation over four years.

A meta-analysis restricted to this question pooled eight treatment arms from six randomized controlled trials, covering 300 patients with WOMAC scores at 12 months. The pooled standardized mean difference was −1.35 (95 percent CI −1.97 to −0.74), a moderate to large treatment effect overall. Doses of 25 million cells or fewer were associated with statistically significant improvement, higher doses did not demonstrate additional benefit, and meta-regression found no significant dose-response relationship.^[10]

The registry analysis of intravenous trials points at something more specific than a plateau. Among the four trials that reported actual differential efficacy across doses, the effective range was narrow, roughly 100 to 150 million cells per patient, with both lower and higher doses appearing less effective.^[5] A dose window, if real, is a different proposition from a ceiling, and it implies that escalating beyond the window is not merely wasteful but potentially counterproductive.

Against these, the knee dose-escalation trial described above found its structural benefit specifically in the 100 million cell group and not in the lower cohorts.^[8] With three patients per arm in the escalation phase, that observation cannot carry much weight on its own, but it is a real result pointing the other way.

Three statements can be made responsibly. Evidence for a monotonic relationship in which more cells reliably produce more benefit is weak. Several controlled comparisons have found no separation between doses differing by a factor of ten. And a marketing emphasis on very large cell counts, which is common in this industry, is not supported by the controlled data and should not on its own be read as an indicator of a better protocol. The related question of how to evaluate a clinic's claims is addressed in our article on stem cell therapy safety.

A cell count describes quantity. It does not describe biological activity, and two preparations with the same nominal count can differ substantially in what they do. This is a central and well-documented problem in the field rather than a technicality.

A widely cited perspective on this problem observes that the outcomes of advanced clinical trials have fallen short of the expectations raised by preclinical animal data, and attributes much of that gap to biological and pharmacological disparities between laboratory models and human use.^[6] Potency is the practical example. An international consensus workshop convened by the International Society for Cellular Therapy concluded that identity markers alone do not establish immunological function, and that release testing for advanced clinical studies should include a matrix of functional potency assays measuring gene expression, functionally relevant surface markers, and secreted protein.^[16] No such assay is standardized across the industry. Two protocols quoting the same number of cells may therefore be administering materially different products.

The state of the cells at the moment of administration is a further variable, and one specific to the intravenous route. Freeze-thawed cells have been shown to have impaired immunomodulatory and blood-regulatory properties compared with cells harvested from continuous culture, including increased triggering of the instant blood-mediated inflammatory reaction and stronger complement activation on contact with blood.^[14] A dose of 100 million thawed cells and a dose of 100 million freshly harvested cells are not interchangeable descriptions, and viability at administration, which is what determines the count that matters, is not always the number quoted.

A complete protocol description therefore includes more than a figure. It states the cell type and tissue source, whether the cells are autologous or allogeneic, the passage number, the measured viability at administration, the total viable cell count, the route, the injected volume and rate, and whether the product was administered fresh or after cryopreservation. Sourcing and batch testing are covered in our article on autologous and allogeneic sourcing, and the delivery-related translational obstacles have been reviewed in detail.^[15]

How many stem cells are in a typical treatment?

There is no single figure, and any source quoting one without stating the route and indication is oversimplifying. Across 914 registered mesenchymal stem cell trials, the median intravenous dose was 100 million cells per patient per dose, with minimum effective doses in positive trials generally falling between 70 and 190 million.^[5] Intra-articular protocols for knee osteoarthritis have used 10 million to 100 million cells per joint.^[8][9] The weight-adjusted graft-versus-host disease protocol cited here used a median of 1.4 million cells per kilogram, across a range of 0.4 to 9 million.^[7]

Is intravenous infusion or direct injection better?

Neither is better in general, because they are answers to different anatomical questions. Direct injection into a joint or the cerebrospinal fluid places cells at a defined target and bypasses the lungs. Intravenous infusion cannot deliver most cells past the pulmonary circulation, and in animal models the infused cells are cleared within about a day, so it is appropriate where the intended mechanism is a systemic or immunomodulatory signal rather than local tissue repair.^[2][3] The right question is which compartment the intended mechanism acts in.

Do more cells produce better results?

Not reliably. A four-year randomized comparison of 10 million against 100 million cells in knee osteoarthritis found no clinical difference between the two doses.^[9] A dose-focused meta-analysis of six randomized trials found that doses of 25 million or fewer produced significant improvement while higher doses added no further benefit, with no significant dose-response relationship on meta-regression.^[10] An analysis of intravenous trials reporting dose-response data found an effective window of roughly 100 to 150 million cells, with both lower and higher doses less effective.^[5] One small dose-escalation trial did find its best structural result at the highest dose.^[8]

Why do the cells not travel to the injured area after an infusion?

Most of them do not get the chance. Cultured mesenchymal stromal cells are large relative to pulmonary capillaries, and blood from a peripheral vein reaches the lungs before it reaches the systemic arteries, so the majority of infused cells are retained on that first pass.^[2] Those retained cells remain viable for around a day and are then cleared, and viable donor cells have not been recovered from other organs even when injury was deliberately induced there.^[3] The prevailing interpretation is that the therapeutic signal, where one exists, is transmitted by soluble factors and vesicles rather than by the cells themselves relocating.^[4]

How does TrueCell determine dose and route?

Both are set by a board-certified physician after reviewing history, imaging, and laboratory results, and both follow from the indication rather than from a fixed package. A joint-localized problem is treated by intra-articular injection; a systemic or immune-mediated one is approached intravenously. The consultation includes the specific cell count, the tissue source, the route, and a plain statement of what the evidence supports at that dose and route and what it does not. Programs by indication are listed on the treatments page.