Cells that can renew themselves and give rise to specialised cell types. That definition is precise, and most of what follows depends on holding it precisely — including the fact that the cells sold as "stem cells" for most treatments may not really be stem cells at all.
The Definition
Two properties, both required:
- Self-renewal — dividing to produce more stem cells
- Differentiation — giving rise to specialised cell types
Everything else follows from how much a given cell can become.
⚠️ The gap between pluripotent and multipotent is the most important distinction on this page, and it is where most confusion about stem cell therapy begins. An embryonic or induced pluripotent cell can in principle become a neuron, a heart muscle cell or a liver cell. An adult MSC cannot. Claims made about the first category do not transfer to the second — and the treatments offered commercially are almost always the second.
The three types that matter clinically
Embryonic stem cells. Pluripotent, derived from the blastocyst [17]. Ethically contested and, for that reason among others, not used in commercial regenerative practice.
Adult (tissue-resident) stem cells. Multipotent. Found in bone marrow, adipose tissue, umbilical cord and elsewhere. This is the category used in essentially all regenerative treatments.
Induced pluripotent stem cells (iPSCs). Adult cells reprogrammed to a pluripotent state using defined factors — the discovery for which Yamanaka shared the Nobel Prize [3]. ⚠️ Still primarily a research tool. Extraordinarily valuable for disease modelling and drug testing; clinical applications are emerging and not yet routine.
⚠️ The Naming Problem, Stated Plainly
The cells in most regenerative treatments are called "mesenchymal stem cells" (MSCs). That name is disputed within the field, and a reader deserves to know why.
In 2006 the International Society for Cellular Therapy issued minimal defining criteria for these cells. The title of that position statement is "Minimal criteria for defining multipotent mesenchymal stromal cells" [2], and it sets three requirements: plastic adherence; a specific surface-marker profile; and tri-lineage differentiation in vitro into osteoblasts, adipocytes and chondroblasts.
Arnold Caplan — who coined the term "mesenchymal stem cell" — has since argued it should be abandoned, on the grounds that these cells do not meaningfully become new tissue in patients and that calling them stem cells misleads clinicians and the public. His proposed replacement: medicinal signalling cells [5].
⚠️ This matters commercially as well as scientifically. "Stem cell" implies replacement of damaged tissue. The evidence points to signalling rather than replacement — and the abbreviation MSC survives partly because it reads well either way.
This page uses "MSC" because it is the term you will meet everywhere. It is worth knowing that the "S" is contested.
How MSCs Actually Work
For years the assumption was simple: inject stem cells, they become the missing tissue, the problem is solved. The field no longer thinks that is the main mechanism.
Paracrine signalling — the dominant account
MSCs secrete a wide range of bioactive molecules: growth factors, cytokines, and extracellular vesicles carrying proteins and microRNA [5][6][7]. These modulate inflammation, influence resident cells, and support repair processes already present in the tissue.
⚠️ The observation that settled the argument: relatively few transplanted MSCs survive long-term, yet effects persist. A cell that has disappeared cannot be the new tissue. It can, however, have left signals behind — which is why "medicinal signalling cell" is the name its originator now prefers [5].
Immunomodulation
MSCs interact with both innate and adaptive immunity, generally dampening inflammatory responses [8]. This is the best-established property, and it is the basis of the one MSC therapy with full regulatory approval — for graft-versus-host disease.
⚠️ Note the direction: MSCs are immunosuppressive. Any claim that they "boost" the immune system runs against their established mechanism.
Differentiation — real in the laboratory, limited in the patient
MSCs do differentiate into bone, cartilage and fat cells in culture — the multilineage potential of adult human MSCs was the founding demonstration of the field [1], and the ISCT uses exactly this as an identity test [2]. ⚠️ What happens in a culture dish under directed conditions is not what happens after injection into a joint. The review literature is explicit that fundamental questions remain open, including "the best cell source, their nature, and the right amount" [4].
Direct tissue replacement is the weakest of the proposed mechanisms, not the strongest.
What the Clinical Evidence Shows
Knee osteoarthritis — the best-studied indication
⭐ The most thorough appraisal available [9]evaluated 15 randomised controlled trials and 11 non-randomised trials of culture-expanded MSCs in knee OA:
- Function improved in 12 of 15 RCTs versus baseline, and in 11 of 15 versus control groups at study endpoints
- Cartilage protection and/or repair reported in 18 of 21 clinical studies
- ⚠️ "The relatively small sample size of 610 patients limited the drawing of definitive conclusions" — the authors' own words
- Trends toward moderate-to-higher doses benefiting selected patient phenotypes
- ⚠️ Evidence for anti-inflammatory and immunomodulatory effects comes largely from preclinical studies
That is a genuinely encouraging body of evidence with an honestly stated ceiling. Both halves belong together.
A randomised phase II trial of intra-articular autologous bone-marrow MSCs adds controlled data to the same picture [18], and an earlier meta-analysis reached broadly consistent conclusions [14].
⚠️ Which cell source is best?
The only head-to-head comparison is a systematic review and network meta-analysis of 15 studies and 585 patients, comparing bone-marrow, adipose and umbilical-cord MSCs across pain, range of motion, MRI structural score, WOMAC and adverse events [16].
*Its conclusion: "MSCs transplantation effectively treats KOA patients, with autologous BM-MSC potentially offering more excellent benefits."*
⚠️ This points to autologous bone marrow, not umbilical cord. It is one network meta-analysis, and indirect comparisons across heterogeneous trials carry real uncertainty — but it is the best direct answer available, and it does not favour the source most commonly sold commercially.
What predicts response
Reviewing the trial evidence, the parameters identified as shaping MSC effectiveness are cell dose, tissue of origin, autologous versus allogeneic source, and patient clinical phenotype, endotype, age, sex and level of disease severity [9].
⚠️ Read that last one carefully, because it is often misstated. Disease severity predicts response: less advanced structural disease tends to respond better than end-stage disease. Everyone in that evidence has the disease. It is not a finding that healthy people respond better than sick people — that comparison has never been made, because no trial has enrolled healthy people.
Other indications
Chronic lung disease — a systematic review and meta-analysis of stem-cell-based therapy in COPD [11], with further reviews of mesenchymal stromal cell therapy across chronic lung conditions [12][20]. ⚠️ Much of the supporting work here is experimental rather than clinical, and the reviews say so.
Aging frailty — randomised placebo-controlled trials, including one using umbilical-cord MSCs [23], alongside earlier phase II work [21]and a mechanistic review [22]. ⚠️ All participants had diagnosed frailty; none were healthy.
Exosomes — MSC-derived vesicles are under active investigation [7], with reporting standards now formalised [6]. ⚠️ Much of the osteoarthritis work is in rat models [13].
Cell Sources
What is genuinely established about donor age: the proliferative capacity and secretory profile of MSCs do decline with donor age, and neonatal tissue does not carry that decline. Neonatal umbilical cord tissue is also a source of extracellular vesicles carrying anti-aging signals, shown to rejuvenate senescing adult bone-marrow MSCs [24].
⚠️ Two limits on that last finding, both material. It concerns extracellular vesicles, not cells; and its in-vivo work was conducted in aged mice [24].
⚠️ What is not established: "potency." Potency in cell therapy is a defined regulatory term requiring a validated assay tied to a product's mechanism of action. No such assay is cited for any product described here, and the word should not be used as though one existed. Higher proliferation in culture is a measurable laboratory attribute. It is not the same as a demonstrated clinical advantage — and the one head-to-head comparison points the other way [16].
Safety
A systematic review and meta-analysis of MSC clinical trials (SafeCell) found no association between MSC administration and acute infusional toxicity, organ system complications, infection, death or malignancy, with transient fever the one signal identified [10].
That is a reasonable safety record, and it is not the same as no risk. Long-term data remain limited, repeat-dosing safety is not well characterised, and safety in healthy people has not been studied because healthy people have not been enrolled.
Regulatory position. MSC therapy is not FDA-approved for orthopaedic, respiratory, autoimmune or anti-aging indications. The ISSCR guidelines are explicit that cell-based interventions require formal clinical evaluation rather than direct-to-consumer provision [19]. Availability in a permissive jurisdiction is not approval.
The State of the Field
Several hundred trials of mesenchymal stromal cells are recruiting worldwide across orthopaedic, autoimmune, respiratory, metabolic and neurological indications [15](ClinicalTrials.gov, retrieved 2026 — a live count, and a snapshot rather than a fixed figure). ⚠️ Trial counts measure activity, not evidence. A field with many recruiting trials is a field that has not yet answered its questions.
Reasonably established: the short-term safety profile; the immunomodulatory mechanism; symptomatic benefit in knee osteoarthritis across multiple randomised trials, on a small total evidence base.
Not established: durability; the optimal dose; the best cell source; whether structural disease modification occurs; and anything at all about healthy people.
Frequently Asked Questions
Are MSCs really stem cells?
*Formally they are classified as mesenchymal stromal cells [2], and the researcher who named them argues for medicinal signalling cells* [5]. They meet the laboratory definition of a multipotent progenitor. Whether they behave as stem cells in a patient is exactly what is disputed.
Will injected stem cells become my cartilage?
Probably not — and that is not how the benefit is currently thought to arise. Most transplanted MSCs do not persist; the prevailing account is that they signal to cells already present [5][7].
Which source should I choose?
The only head-to-head network meta-analysis points to autologous bone marrow [16]. Allogeneic umbilical cord cells have practical advantages — no harvest, immediate availability — and higher proliferative capacity in culture. Those are real advantages, and they are not the same as better outcomes.
Is this approved?
Not for the indications described here. One MSC product is approved for graft-versus-host disease. Everything else is investigational.
Does it work better if I'm healthier?
That has never been tested. What has been shown is that disease severity predicts response — less advanced disease tends to respond better than end-stage [9]— in populations where everyone has the disease. No trial has compared healthy recipients against unwell ones, and no trial has enrolled healthy people at all.