Every randomised trial of cell therapy in stroke, by route and by phase — including the ones that reported nothing.
## ⚠️ If a stroke is happening now, stop reading and call emergency services
Stroke is time-critical. Clot-dissolving treatment (thrombolysis) and clot-removal (thrombectomy) work
within a narrow window measured in hours, and every minute of delay costs recoverable brain tissue. **They are
the treatments with by far the largest effect on stroke outcome.**
**Nothing on this page is an alternative to emergency stroke treatment, and nothing here is a reason to delay
seeking it.**
After a stroke, secondary prevention — antiplatelet or anticoagulant medication, blood-pressure control,
lipid management, treatment of atrial fibrillation — does more to protect your future than any investigational
therapy discussed here. Do not stop, reduce or change any of it on the basis of a web page. That decision
belongs to your stroke team.
What Happens After a Stroke
Stroke is among the leading causes of long-term disability worldwide [1]. Around 85% of strokes are ischaemic — a clot blocking an artery — and around 15% haemorrhagic, where a vessel ruptures.
The recovery curve, and why it matters for interpreting trials
Recovery after stroke follows a reasonably well-characterised pattern [2]:
- First weeks to 3 months — the fastest period. Swelling resolves, blood flow to threatened tissue re-establishes, and diaschisis (suppression of function in remote, undamaged regions) resolves. Most recovery happens here, and most of it happens regardless of treatment
- 3–6 months — continued gains, increasingly dependent on rehabilitation intensity
- 6–12 months — the rate of change slows and most people reach a functional plateau
- Beyond 12 months — traditionally considered near-static, though this is being re-examined
⚠️ This curve is the reason early-phase stroke trials are so hard to read. In the first weeks and months, people improve substantially on their own. An uncontrolled study that reports improvement in that window has measured spontaneous recovery plus whatever the treatment did, with no way to separate the two. This is why placebo-controlled trials matter here more than in almost any other field, and why single-arm results should be treated with particular caution.
How the brain rebuilds
Recovery involves several overlapping mechanisms [2]: resolution of diaschisis; perilesional reorganisation, where surviving neurons near the damage take over lost functions; contralesional compensation by the opposite hemisphere; limited neurogenesis; and angiogenesis supporting recovering tissue.
Cell therapy research targets the later mechanisms — not by replacing dead neurons, but through paracrine (trophic) signalling: neurotrophic factors such as BDNF and GDNF, anti-inflammatory mediators, and angiogenic factors such as VEGF. Almost no researcher now proposes that transplanted cells become functional neurons in useful numbers.
The Treatment That Actually Works
Before the investigational material, the established one — because it remains the core of stroke recovery and has the strongest evidence on this page.
Intensity and repetition are the active ingredients. Nothing described further down this page has evidence approaching this, and no investigational therapy should displace it.
The Cell Therapy Trials, by Route and Phase
Route matters enormously in neurology, and grouping trials by it is the only way to read them honestly. Intravenous infusion, intracerebral implantation and intra-arterial delivery are different interventions.
Acute and subacute stroke — intravenous
MASTERS (Hess et al., 2017) [3]— phase 2, randomised, double-blind, placebo-controlled, dose-escalation, 33 centres in the UK and USA.
- Cells: multipotent adult progenitor cells (MultiStem) — a bone-marrow-derived allogeneic product. These are not mesenchymal stromal cells
- 400 million or 1,200 million cells intravenously, between 24 and 48 hours after symptom onset
- Patients aged 18–83 with NIHSS 8–20; 129 randomised, ITT populations of 65 and 61
- Primary efficacy endpoint: global stroke recovery — a composite of modified Rankin scale, NIHSS change from baseline, and Barthel index at day 90
Result: "There was no difference between the multipotent adult progenitor cell group and placebo groups in global stroke recovery at day 90 (odds ratio 1·08 [95% CI 0·55–2·09], p=0·83)." [3]
No dose-limiting toxicity, no infusional or allergic reactions, and no difference in treatment-emergent adverse events — though it is worth noting that 99% of the treatment group and 97% of placebo experienced at least one, which is what a trial in acutely unwell stroke patients looks like. The authors' own reading was that efficacy should next be tested in an earlier time window [3].
TREASURE (Houkin et al., 2024) [5]— phase 2/3, double-blind, placebo-controlled, 44 centres in Japan.
- Cells: MultiStem (HLCM051) — the same multipotent adult progenitor cell product as MASTERS
- 1.2 billion cells intravenously, within 18 to 36 hours of onset
- 206 patients (104 treatment, 102 placebo), mean age 76.5 years
- Primary endpoints: safety, and excellent outcome at day 90 (composite of mRS ≤1, NIHSS ≤1, Barthel ≥95)
Result: "There were no between-group differences in primary and secondary end points. The proportion of excellent outcomes at day 90 did not differ significantly (12 [11.5%] vs 10 [9.8%], P = .90; adjusted risk difference, 0.5% [95% CI, −7.3% to 8.3%])."
Conclusion: "…was safe but did not improve short-term outcomes." [5]
A signal in patients with moderate-to-severe deficits was reported, but it comes from exploratory analyses and the authors label it as such.
AMASCIS (de Celis-Ruiz et al., 2022) [6]— phase IIa, randomised, double-blind, placebo-controlled, single-centre pilot.
- Cells: allogeneic adipose tissue-derived MSCs, intravenously, within the first 2 weeks of onset
- Patients aged ≥60 with NIHSS 8–20
- Recruitment stopped after 19 of 20 planned patients; six did not receive treatment
- Final sample: 13 patients — 4 treated, 9 placebo
- Primary outcome: safety
Result at 24 months: a non-significantly lower median NIHSS in the treated group (3 [3–5.5] vs 7 [0–8]), and "neither treatment group had differences in mRS scores throughout follow-up visits up to month 24."
Conclusion: intravenous AD-MSCs within two weeks of ischaemic stroke "was safe at 24 months of follow-up." [6]
Four treated patients. This is a safety pilot and reports itself as one. It is sometimes quoted as though it demonstrated efficacy; it did not, and was not designed to.
ISIS-HERMES (Jaillard et al., 2020) [7]— single-centre, open-label, randomised 2:1, analysed as-treated rather than by intention to treat.
- Cells: autologous bone-marrow MSCs, intravenously, within 2 weeks of moderate-severe carotid-territory stroke
- 31 enrolled, 16 received cells
Result: "We observed no treatment effects on the Barthel Index, NIHSS, and modified-Rankin scores, but significant improvements in motor-NIHSS (p = 0.004), motor-Fugl-Meyer scores (p = 0.028), and task-related fMRI activity in MI-4a (p = 0.031) and MI-4p (p = 0.002)." [7]
Global function unchanged; motor-specific measures improved. In an open-label, as-treated analysis of 16 treated patients, that is a hypothesis worth pursuing rather than a demonstrated effect — which is how the authors frame it.
The first controlled human study of MSCs in stroke was published two decades ago [8], and the field has been described since as struggling to convert preclinical promise into clinical results [9][12].
Chronic stroke — intracerebral implantation
This is a neurosurgical procedure: cells are injected into the brain through the skull under stereotactic guidance. Its risk profile is not comparable to an infusion.
PISCES-2 (Muir et al., 2020) [9]— prospective, multicentre, single-arm, open-label.
- Cells: CTX0E03 allogeneic human neural stem cells, 20 million, implanted stereotactically into the putamen
- 23 adults with significant upper-limb motor deficit 2–13 months after stroke (median 7 months), at eight UK hospitals
- Primary outcome: improvement of ≥2 points on the ARAT subtest 2 at 3 months
Result: "One of 23 participants improved by the prespecified ARAT subtest level at 3 months, and three participants at 6 and 12 months." [9]
The authors also report that improvement occurred only in those who had residual upper-limb movement at baseline — a substantial selection caveat — and that transient procedural adverse effects occurred, with two deaths unrelated to trial procedures. No cell-related adverse events were seen to 12 months.
Steinberg et al., 2019 (SB623) [10]— 2-year, open-label, single-arm, phase 1/2a.
- Cells: SB623, modified allogeneic bone-marrow MSCs, implanted stereotactically around the subcortical stroke region under MRI guidance
- 18 patients with stable motor deficits 6–60 months after non-haemorrhagic stroke
Improvements from baseline at 24 months, in the 16 patients who completed, included a European Stroke Scale gain of 5.7 points (95% CI 1.4–10.1).
⚠️ Two things must travel with those numbers. First, there is no control group. These are within-group changes from baseline, not comparisons against anything, and the study describes itself as open-label and single-arm. Second, the safety findings:
"All experienced at least 1 treatment-emergent adverse event… The most frequent TEAE was headache related to the surgical procedure (88.9%). Seven patients experienced 9 serious adverse events, which resolved without sequelae." [10]
No deaths and no dose-limiting toxicities. But every patient had an adverse event, and nearly 40% had a serious one. That is the trade for opening the skull, and it belongs beside the improvement figures.
What the systematic evidence concludes
A Cochrane review covering ischaemic stroke in all phases — acute, subacute and chronic — across all cell sources, all routes and all doses, with efficacy at a minimum of six months as its primary outcome, states the position plainly [4]:
"…to date, evidence for the benefits of stem cell transplantation in people with ischemic stroke is lacking."
Trials in progress
Phase 3 trials of both MultiStem (MASTERS-2) and SB623 (ACTIsSIMA) have been conducted. Results are not yet published in a form that can be verified here, and this page will not characterise them until they are.
Reading This Honestly
What is reasonably established: cell therapy has been delivered to stroke patients by several routes without a consistent safety signal attributable to the cells. Intravenous infusion in particular appears well tolerated across several hundred patients.
What is not established: that any of it improves recovery. Every placebo-controlled trial has missed its primary endpoint. The positive results that circulate come from secondary endpoints, exploratory subgroups, or single-arm studies without controls — and in early stroke, uncontrolled improvement is largely spontaneous recovery.
What would change this: an adequately powered randomised trial meeting a prespecified primary endpoint. None has.
The field has been explicit with itself about why that has not yet happened. The STEPS 3 consensus — a working group convened specifically to improve the design of stroke cell-therapy trials — set out recommendations on patient selection, timing, dose, delivery route and outcome measurement precisely because early trials had been too heterogeneous to interpret [11]. Reading the trials above together, those problems are visible: three different cell products, three different time windows, two different routes, and primary endpoints that are not the same measure. That is a field still establishing how to test itself, not one whose tests have come back positive.
On haemorrhagic stroke
Almost all of this work is in ischaemic stroke. The acute pathophysiology of haemorrhagic stroke differs substantially, and extrapolating from one to the other is not justified. Separate trials are needed.
On chronic stroke specifically
Most people reading a page like this are months or years past their stroke. That is the population where spontaneous recovery is least confounding — and it is also where the evidence is thinnest and the interventions studied are most invasive. The two chronic-stroke studies above are both single-arm, one met its primary endpoint in 1 of 23 patients, and the other required neurosurgery in which every patient had an adverse event.
Questions Worth Asking
- "Is this a registered clinical trial with ethical approval?" — for stroke, that is currently the only setting in which cell therapy is appropriate.
- "By what route, and what is the evidence for that route?" — intravenous, intra-arterial and intracerebral are different interventions.
- "How long after my stroke, and what does the evidence say for that window?"
- "Was the study that supports this controlled?" — in stroke, an uncontrolled result in the first months is close to uninterpretable.
- "What are the procedural risks?" — particularly for anything requiring intracerebral delivery.
- "Is my rehabilitation optimised, and is my secondary prevention in place?" — these have the strongest evidence, and they come first.
The Bottom Line
Cell therapy for stroke has been tested properly, and the results are known: MASTERS, TREASURE and AMASCIS were all randomised and placebo-controlled, and none showed a benefit on its primary endpoint [3][5][6]. The Cochrane assessment is that evidence of benefit is lacking [4].
Safety looks reasonable. Efficacy has not been demonstrated. The encouraging figures that circulate about this field almost all come from studies with no control group — and in stroke, that distinction is not a technicality.
None of this means the science is worthless. Motor-specific and imaging signals in ISIS-HERMES [7], and the possibility that an earlier treatment window matters [3], are legitimate reasons to keep testing. But testing is what is happening, and a patient's best-evidenced options today remain intensive rehabilitation and rigorous secondary prevention.