Bone marrow stromal cells (MSCs): what they are and what two decades of trials show

Spread a sample of bone marrow in a plastic dish, add nutrient medium and wait a week or two. Most of the blood cells float off and get rinsed away. What stays behind is a scattering of flat, spindle-shaped cells glued to the bottom of the dish, slowly forming small colonies. Those cells are the subject of this page.

They have gone by several names: marrow stromal cells, mesenchymal stem cells, mesenchymal stromal cells and even medicinal signaling cells. Most papers settle for MSC. Since the early 2000s they have been tested in hundreds of registered trials, for everything from transplant complications to heart failure to worn-out knees, and the results show how a promising cell can be both real and oversold.

Stem cell or stromal cell? The naming argument

In the 1960s and 1970s, Alexander Friedenstein and colleagues in Moscow showed that marrow contains fibroblast-like cells that stick to plastic, form colonies and, when transplanted into animals, can lay down bone. In 1991 the American biologist Arnold Caplan proposed calling them "mesenchymal stem cells", on the idea that they were the source of bone, cartilage and other connective tissue in adults.

The trouble is the word "stem". In the strict sense, a stem cell has to renew itself and rebuild a tissue inside a living body, the way a single blood stem cell can repopulate a mouse's entire blood system. That bar has been cleared for blood-forming cells many times. For cultured MSC populations the evidence is much thinner, and what grows in the dish is a mixed population rather than one uniform cell type. (Our stem cell technology page covers how cells are isolated and grown, and why culture changes them.)

The International Society for Cell and Gene Therapy (ISCT, called the International Society for Cellular Therapy at the time) tried to settle this twice. In 2005 it recommended "multipotent mesenchymal stromal cells" for the cultured, plastic-sticking cells, keeping the MSC abbreviation. In 2019 it went further, asking researchers to name the tissue source (bone marrow, adipose, umbilical cord) and to use "stem" only when they have shown true stem cell behavior. Caplan himself argued in 2017 that the cells seem to help mostly through the chemical signals they release, not by turning into new tissue, hence "medicinal signaling cells".

The name matters outside the lab too. "Stem cell" sells; "stromal cell" does not. An advertisement promising stem cells that regrow cartilage leans on a meaning the science has mostly moved away from.

The 2006 checklist

Because every lab grew its cells differently, results were hard to compare. In 2006 the ISCT published a short position paper (Dominici and colleagues, Cytotherapy) with three minimum criteria for calling a human cell population an MSC:

• Plastic adherence: the cells must stick to standard plastic culture vessels under normal culture conditions.
• Surface markers: at least 95 percent of the cells must carry CD105, CD73 and CD90, and no more than 2 percent may carry CD45, CD34, CD14 or CD11b, CD79a or CD19, and HLA-DR. In plain terms: the cells look like connective tissue cells and do not look like blood cells, blood vessel lining or activated immune cells.
• Trilineage differentiation: in the dish, under the right conditions, the cells must be able to become bone-forming cells (osteoblasts), fat cells (adipocytes) and cartilage-forming cells (chondroblasts), confirmed with standard stains.

It is an identity test, not a potency test. It says nothing about whether a batch will calm an immune reaction or do anything at all in a patient, and two batches can both pass yet behave very differently once infused.

Where the cells come from

Bone marrow was the original source and is still common. MSCs are a tiny fraction of marrow cells (a 1999 Science paper by Pittenger and colleagues put it at roughly 0.001 to 0.01 percent of nucleated cells), so a small aspirate from the hip bone is grown for weeks to reach a usable dose. The first MSC product approved in the US is made from the marrow of healthy adult donors.

Fat yields far more cells per gram, collected by liposuction. Be careful with the term "stromal vascular fraction", which clinics sometimes market: that is an uncultured mix of many cell types from digested fat, not the same thing as a cultured MSC population.

Umbilical cord tissue, especially the jelly-like Wharton's jelly, and the placenta are collected after birth with consent. Cells from these young donors tend to grow quickly, and many recent trials, particularly in Asia, use them.

Marrow, fat and cord cells all pass the 2006 checklist, yet they differ in growth rate and in the signals they release.

Why researchers got excited, roughly 2008 to 2012

Three findings drove the enthusiasm.

The first was immune modulation. In culture, MSCs damp down the proliferation of T cells. In 2004 Katarina Le Blanc's group at the Karolinska Institute in Sweden reported in The Lancet that a child with severe, steroid-resistant graft-versus-host disease (GVHD, where transplanted immune cells attack the recipient) improved after receiving MSCs. A multicenter phase 2 study from the same group, published in The Lancet in 2008, treated 55 patients; 30 had a complete response and 9 improved. It was an uncontrolled study, but in a condition with grim outcomes it looked striking.

The second was low immunogenicity. MSCs carry little of the HLA class II molecules that trigger rejection, so cells from one donor could, in principle, be given to many patients without matching. That made an "off the shelf" business model possible.

The third was homing and signaling. Animal studies showed infused MSCs collecting near injured tissue, and later work suggested the benefit came from secreted factors and tiny membrane vesicles rather than from the cells settling in and becoming new tissue.

Each point proved more complicated in people. When MSCs are infused into a vein, most are caught in the lungs within minutes (a 2009 study by Fischer and colleagues called this the pulmonary first-pass effect), and few survive long. Whatever they do, they seem to do it briefly and at a distance.

The trial record, condition by condition

Graft-versus-host disease

GVHD is where MSCs have the strongest regulatory record, and it is still a mixed one. A company phase 3 program for steroid-refractory acute GVHD missed its primary endpoint in 2009. Health Canada nonetheless granted a conditional approval for children in 2012, and Japan approved a marrow-derived MSC product for acute GVHD in 2015. In the US, the FDA turned down the application for remestemcel-L twice, in 2020 and 2023, asking for more evidence.

On December 18, 2024, the FDA approved that product as Ryoncil, the first mesenchymal stromal cell therapy licensed in the US, for steroid-refractory acute GVHD in children 2 months and older. The approval rested on a single-arm study of 54 children: at day 28, 30 percent had a complete response and 41 percent a partial response. There was no randomized control group, which is why the debate over its size of effect continues.

Heart disease

Cardiology ran some of the earliest and largest cell therapy programs, first with unsorted marrow cells and later with MSCs. Small early studies reported better pumping function; larger, better-controlled trials mostly found small or no improvements on their main measures. The detail is on our page on cell therapy after a heart attack.

Knees and other joints

Knee osteoarthritis is where MSC marketing is loudest and the evidence weakest. Trials are mostly small, use varied cell sources and doses, and measure pain, which responds strongly to placebo in knee studies. Convincing evidence that injected MSCs regrow cartilage in people is lacking. No MSC product is approved by the FDA for any orthopedic use.

Crohn's disease fistulas

This is the field's cautionary tale. In March 2018 the European Commission approved darvadstrocel (Alofisel), fat-derived donor MSCs injected around complex perianal fistulas in Crohn's disease, based on the phase 3 ADMIRE-CD trial. A larger confirmatory trial, ADMIRE-CD II, enrolled 568 patients and reported in October 2023 that it missed its primary endpoint: combined remission at 24 weeks was about 49 percent with the cells and about 46 percent with placebo. The marketing authorization was withdrawn in the EU, with no new patients treated after December 13, 2024.

Why the results keep coming out mixed

MSC trials disagree with each other for reasons that are now fairly well understood:

• Donor variability. Cells from different donors, of different ages and health, release different amounts of the signals that seem to matter. A product made from one donor's cells may behave unlike one made from another's.
• Culture conditions. The medium (fetal bovine serum, human platelet lysate or defined formulas), the number of times cells are split and how they are frozen and thawed all change their behavior. Cells given straight from culture and cells thawed at the bedside can differ in activity.
• Weak potency assays. Regulators want a lab test showing each batch has the activity that matters clinically. For MSCs, nobody is fully sure which activity that is, so potency tests are proxies, such as how well a batch suppresses T cells in a dish. A batch that passes may still underperform.
• Small, varied trials. Studies used different doses, routes and endpoints, so pooling them is hard and chance results can look meaningful.

Our page on how a cell therapy is made, from donor to dose walks through why manufacturing details carry this much weight.

What this means for "MSC injections" at clinics

Clinics in the US and abroad sell injections of fat, marrow or umbilical cord cells for joint pain, back pain, lung disease, autism, anti-aging and much else. Set against the record above, three points stand out.

The approved MSC uses are narrow. When this page was updated, Ryoncil, for a specific transplant complication in children, was the only mesenchymal stromal cell product on the FDA's list of approved cellular and gene therapy products. Anything else sold as a cell treatment in the US needs either an approval or an investigational new drug (IND) exemption, which applies to a registered trial, not to a cash-pay menu.

Even well-run company trials have failed. If a carefully manufactured product tested in hundreds of patients could not beat placebo for Crohn's fistulas, a clinic's in-house preparation without published potency testing carries far more uncertainty.

The risks are documented. The FDA has issued public warnings about unapproved regenerative medicine products, citing reports of serious harm including blindness, infections and tumors after unapproved stem cell procedures. The International Society for Stem Cell Research publishes similar guidance.

Our guide to warning signs at unproven stem cell clinics lists what to look for, and clinicians who field these questions will find practical framing on our page for health professionals. One concrete test applies to any offer: ask for the trial's ClinicalTrials.gov number (it starts with NCT) and the IND number. A legitimate US trial has both, and charging participants for an investigational product requires prior FDA authorization, which is uncommon. If neither number exists, the treatment is not part of a regulated study.