What are Muse cells, and why the interest?
Muse stands for multilineage-differentiating stress-enduring. Researchers first described these cells in adult human mesenchymal cell populations in 2010. Selected cells survived laboratory stress, renewed themselves and produced cells with characteristics of all three embryonic tissue lineages. That range makes them interesting for studying tissue repair. It does not mean an infusion can rebuild any damaged organ. [1]
SSEA-3 is a surface marker used to identify the population, often together with CD105. Muse cells are a selected subset, not another name for every mesenchymal stromal cell (MSC). The preparation matters: the CL2020 product described in the neonatal trial contained at least 50% CD105/SSEA-3 double-positive cells, alongside other MSCs. A different source, culture method or cell mixture needs its own evidence. [2] [10]
What does the repair mechanism actually show?
In rabbits with a heart attack, transplanted Muse cells reached injured heart tissue, developed cardiac and vascular characteristics, and improved measures of heart function. Blocking the S1P-S1PR2 signaling pathway reduced homing, supporting a role for that pathway. Donor cells from other rabbits persisted for six months without immunosuppression. These were animal findings. [3]
Mouse stroke studies also found neural differentiation and functional improvement. One injected cells directly into the brain; another gave CL2020 intravenously. The route and animal model are part of the result. In people, a 29-patient stroke study found more Muse cells circulating after injury. Measuring cells in blood does not show that they entered the human brain, became working neurons or improved recovery. [4] [5] [6]
What happened in the randomized stroke trial?
The Japanese trial randomized 37 people after ischemic stroke. Thirty-five received an infusion: 25 received CL2020 and 10 placebo. Treatment was given about two to four weeks after stroke, alongside rehabilitation. The primary endpoint was safety through 12 weeks, with follow-up to one year. [7]
At 12 weeks, 10 of 25 CL2020 recipients (40%) and 1 of 10 placebo recipients (10%) reached a modified Rankin Scale score of 0 to 2, meaning no more than slight disability. This secondary outcome met a prespecified comparison with historical data. The direct difference between the randomized groups remained statistically uncertain: 30 percentage points, with a 95% confidence interval from -7.8 to 65.2 and P = .080. Some motor measures also favored CL2020, but baseline differences and the small groups limit confidence. [7]
That is a signal worth testing in a larger study. It does not establish that CL2020 caused recovery in 40% of patients or that it reliably outperformed placebo.
Did the heart studies confirm the animal results?
An initial uncontrolled study gave CL2020 to three people after a heart attack and coronary treatment. Heart-pumping measurements improved over 12 weeks. With no comparison group, recovery from the heart attack and standard care could not be separated from a cell effect. [8]
A subsequent Japanese phase 2/3 study randomized 86 people with ST-elevation myocardial infarction; 84 received CL2020 or placebo. Its registered results report no statistically significant difference in the primary endpoint, the change in left ventricular ejection fraction at 24 weeks. These are posted registry results, not a peer-reviewed trial report. They are still essential to understanding the evidence: the larger controlled study did not confirm its primary heart-function benefit. [9]
What have the other human studies found?
Several small, uncontrolled CL2020 studies add useful observations. They cannot establish how much improvement came from the cells rather than usual care, rehabilitation or the course of the condition:
- Newborn brain injury: SHIELD treated nine babies with hypoxic-ischemic encephalopathy after therapeutic cooling. At 78 weeks, six had normal developmental scores across all three tested domains. There was no randomized control group, and eight of the nine had moderate rather than severe injury. [10] [11]
- Spinal cord injury: Ten people with subacute cervical injury received one infusion. Six improved by at least one modified Frankel grade at 28 weeks, and five at 52 weeks. Some motor and daily-activity scores improved. The EQ-5D quality-of-life index did not significantly improve, and there was no control group. The published correction identifies Figure 2C as total motor score. [12] [13]
- ALS: Five people received six monthly infusions. Function continued to decline overall. A comparison of decline rates before and after treatment did not reach statistical significance (P = .096); the study cannot establish that CL2020 slows ALS. [14]
- Inherited skin fragility: In five people with dystrophic epidermolysis bullosa, selected ulcers and pain improved early. Overall ulcer size returned to baseline by 12 weeks. One skin biopsy did not show an increase in the missing structural protein, type VII collagen, or anchoring fibrils. [15]
What do we know about safety?
The original experiments found no teratomas, a type of tumor containing several tissue types, in immunodeficient mouse testes through six months. That is a useful finding under those experimental conditions. It is not proof that every Muse preparation is incapable of causing tumors or other serious harm in people. [1]
In the stroke trial, serious adverse events occurred in 5 of 25 CL2020 recipients and 1 of 10 placebo recipients over 52 weeks. A severe episode of status epilepticus, a prolonged seizure, was considered possibly treatment-related. The heart program also reported treatment-related ventricular tachycardia in one participant in its separate three-person safety lead-in; the randomized phase reported no serious adverse reactions attributed to CL2020. [7] [9]
In SHIELD, a transient rise in a liver enzyme was considered possibly related. A baby later developed a testicular yolk sac tumor. Genetic testing matched it to the child, not the donor cells, and investigators judged it unrelated to CL2020. Reporting that distinction is more accurate than saying no tumors occurred. [10]
The stroke trial used donor cells without HLA matching or immunosuppressive drugs. That observation does not prove universal immune tolerance or lifelong donor-cell survival. These small studies, including follow-up of one year or 78 weeks, cannot exclude uncommon or delayed complications. [7] [10]
Where does development stand now?
Mitsubishi Chemical Group announced that it was discontinuing CL2020 development in February 2023, citing the clinical-development situation, commercialization timeline and pharmaceutical business strategy. That statement should not be rewritten as proof that the treatment either worked or failed for every condition. [16]
A separate Muse Cell Patient Registry, sponsored by MuseCell Innovations, was listed as recruiting when checked on November 9, 2026. Its last posted update was June 12, 2026, with an estimated target of 5,000 people and no posted results. It is an observational registry, not a randomized CL2020 trial, and its target is not a count of treated participants. [17]
No Muse or CL2020 product appears on the FDA approved cell and gene therapy list or the PMDA regenerative-product list checked for this article. The PMDA list covers approvals through July 2026. A trial registration or a provider offering cells does not establish product approval. The exact product, indication and jurisdiction need to be checked separately. [18] [19]
Questions worth taking to a clinical team
What is supported
- Laboratory differentiation and animal repair findings.
- A promising but uncertain randomized stroke signal.
- Published human observations in several conditions.
What remains open
- Reliable benefit in larger controlled trials.
- Rare risks and safety over many years.
- Whether another product behaves like CL2020.
What to ask
- What is the exact product and its cell composition?
- Which controlled study matches my condition and route?
- What authorization, monitoring and follow-up apply?
What the evidence can answer
Laboratory
Selected cells show stress tolerance and broad differentiation capacity.
Small human studies
Signals in several conditions, with limited ability to separate treatment from recovery.
Controlled trials
Stroke findings remain uncertain; the heart study missed its primary endpoint.
Product status
No listing in the FDA or PMDA approval sources checked here.
The references below include original papers, registered results and regulator records. More cell-therapy research is available on our Science page.
Questions people ask
Are Muse cells the same as ordinary MSCs?
No. Muse cells are a selected subpopulation identified using markers including SSEA-3 and CD105. CL2020 was a particular manufactured cell preparation. Its findings do not establish the effects of every MSC product or every preparation called Muse. [2] [10]
Did the stroke trial prove that Muse cells repair the brain?
It found a promising disability signal, but the direct 40% versus 10% comparison was statistically uncertain. Safety was the primary endpoint. The trial did not directly demonstrate transplanted cells becoming functioning neurons in human brains. [7]
What happened in the larger heart trial?
The Japanese registry reports 86 randomized participants and 84 treated. CL2020 did not significantly outperform placebo on the primary endpoint, change in left ventricular ejection fraction at 24 weeks. [9]
Can Muse cells cause tumors?
Foundational mouse experiments did not produce teratomas under the tested conditions. That does not establish zero human risk. SHIELD reported a testicular tumor traced to the child rather than the donor cells; investigators judged it unrelated to CL2020. Longer and larger safety studies remain important. [1] [10]
Does treatment without immunosuppression prove immune privilege?
No. CL2020 was given without HLA matching or immunosuppressive drugs in the stroke trial. That is a clinical observation for that product and setting, not a guarantee that all Muse preparations avoid rejection or survive indefinitely. [7]
What to take away
- Muse research has reached human trials, with meaningful positive, negative and inconclusive findings.
- Animal homing and differentiation do not establish organ repair in people.
- The stroke signal needs confirmation; the controlled heart trial missed its primary endpoint.
- Cell identity, manufacturing, route and condition determine which evidence applies.
- Observed tolerability is useful, but it does not establish zero tumor risk or universal immune tolerance.
The evidence
Original studies and current source records checked for this article. Registry results and developer statements are labeled separately from peer-reviewed studies.
Sources checked November 9, 2026.
This article is for education, not medical advice or a treatment recommendation. Study findings apply to the particular cells, route and population tested. Any treatment decision should be made with a qualified physician. Individual results vary.