Cell Therapy for Stroke

New data is challenging our optimism for brain cell therapy.

Posted Sep 10, 2020

Cell therapy always appeared as an unlikely strategy for stroke. Cell therapy was conceived as a means for tissue replacement. If tissue were lost through injury or disease, then new cells could be induced to replace the old. So new skin could grow to repair burns, new cartilage to replace tissue lost to arthritis, or new nerve cells in the brain to replace those lost to neurodegeneration. The attraction of stem cells arises precisely because of their ability to achieve exactly this: to generate new cells. Stroke seemed an unlikely candidate simply because of scale.

Stroke comes in two forms: ischaemic or hemorrhagic. Ischaemic stroke, the more common, is caused by a blockage to an artery supplying the brain. This results in the death of the area of the brain deprived of blood. The affected tissue could be the size of a pea or a golf ball, but either way, the tissue is lost, never to be replaced by any natural healing process.

Tissue replacement seems so unlikely because even a small stroke might involve the loss of billions of brain cells. Cell replacement, when applied to human patients, might involve a few million cells, falling orders of magnitude short of what is required. Why, then, would any serious scientist or clinician give this approach a second thought?

The answer is quite simply that it works in experimental animals, at least. Many research groups have administered stem cells of varying kinds to rats afflicted with experimentally induced stroke and discovered remarkable levels of functional improvement.

So, what’s going on? The answer is what we are now calling the "bystander effect." It transpires that transplanting stem cells into damaged brain tissue brings about a series of positive outcomes: New blood vessels are formed, immune cells redirected, new host-derived neurons produced. While it is still not clear specifically which of these impacts is important, the result is the remarkable functional recovery observed in experimental animals.

Biomedical scientists have pounced on this discovery: if experimental animals, why not people; if stroke, why not traumatic brain injury, multiple sclerosis, or dementia? Many of the clinical trials—and the wider application of unlicensed cell therapies—hang on the successful translation of this bystander effect from rats to humans. This has been quite controversial in the field.

Many have insisted the idea was too experimental. We don’t understand the mode of action, whether it works in humans, or how to administer cells to be most efficacious. Others (including me) argued that we have a potential medicine that works in appropriate pre-clinical models and meets regulatory criteria. It would be unethical not to construct proper controlled clinical trials and give patients a chance.

Such trials for stroke have now been ongoing for some years, and the picture is beginning to clear. Unfortunate, then, that on reviewing the situation in 2019, Krause and colleagues were forced to conclude that: "Most results reported up to date, demonstrate safety, but do not show sufficient data for clinical efficacy." (1) This conclusion was provisional, based mostly on the early-phase studies, where small numbers of patients would have been treated and their subsequent progress observed.

A true estimate of the effectiveness of the treatment requires a properly controlled study, in which a large group of patients is treated, usually across a range of clinical centers, and compared with an untreated group. This would be "blinded" so that patients were randomly allocated to either the treated or the control group, and neither patients nor their doctors know which group they were in. There are very few such trials, so each carries enormous weight as we consider the future of this approach.

In this context, the data released earlier this year from the trial sponsored by SanBio is significant. Though not yet peer-reviewed and published, the results were posted on the NIH Clinical Trials webpage earlier this year. Several measures of efficacy have been included in the study, but while the data are extensive, there is no accompanying statistical analysis. Consequently, any interpretation must be considered preliminary.

Nonetheless, a cursory look reveals clearly that patients receiving the treatment showed no improvement compared with the control group. A company press release noted only that the therapy "was not able to meet the primary endpoint regarding efficacy." And though that press release emerged in February 2019, there has been radio silence ever since, suggesting the company has no plans to wade back into this particular pool any time soon. A similar trial using the same cells for traumatic brain injury is still extant—complete but with no data yet reported.

In one sense, this failure shouldn’t surprise anyone. Nine of every 10 clinical trials fail, and even at the advanced Phase 3 stage, the success rate is only 1 in 4. And these are the data for conventional drug trials involving experienced pharmaceutical companies who know their business. How much more difficult to get things right with a totally novel therapy.

Nonetheless, context is important, and the earlier data for this therapy could not have been more positive. Videos were posted showing stroke patients rising from their wheelchairs, talking where they had previously lost the power of speech. The contrast between the first reports and the outcome of the pivotal control trial could not be greater.

The extreme reaction would be that we were all fooled into believing in the nebulous "bystander effect" and have been sold a bill of goods. More rationally, we can perhaps learn (or relearn) two lessons. First, stroke trials have a long history of failure. Therapies that work on rats mostly fail in humans. Through the 1990s, for example, almost 50 neuroprotective agents went into human trials for stroke (2). All had been tested on rats, and all failed in the clinic, causing drug companies to pull out of this field en masse, ruing the wasted millions. So, just a reminder: People aren’t rats.

Second, clinical medicine is hard. You don’t get lucky in science. Alexander Fleming and his bacteria-killing fungus is a great story that, unfortunately, has helped perpetuate the myth that if you keep your eyes open, a scientific miracle can fall into your lap. In truth, every single step of progress requires experiment upon experiment.

This is never more the case than when seeking to turn a simple observation—rat recovers from stroke—into clinical practice. There are a thousand variables that have to be nailed down: how many cells to inject; where precisely to put them; how soon after the stroke to treat; which patient cohorts respond best; which clinical parameters will show improvement and which not, etc. None of these questions have anything more than provisional answers at present.

Just to pick out one: The stroke expert, Cesario Borlongen, has pointed out that by extrapolation from the animal studies, the clinical trials are probably using at least an order of magnitude too few cells (3). In several of the rat studies, the cells only worked at the highest dose, yet we inject people with ten- to a hundred-fold fewer cells relatively and expect them to get better. Of course, early clinical trials by their nature are cautious, and injecting a slurry of cellular material into a person’s brain is not without its safety and logistical challenges. But this is just one parameter that remains to be optimized.

I’m struck by how similar in many regards the stroke field is now to the Parkinson’s Disease field a decade ago. At that time, stem cells for PD had shown promise in early uncontrolled trials. Optimism was high, but as now with stroke, none of the crucial parameters were well sorted.

In retrospect, it is now clear that among other problems, the cells for PD simply weren’t good enough. Newer studies are employing vastly improved cellular material, defined using clever molecular technologies such as single-cell transcriptomics. As a consequence, the current PD trials in the U.S., Europe, and Japan look much more promising. It is worth pointing out that the SanBio cells are not actually neural stem cells at all and date from an era when we understood far less about stem cell biology. There are certainly newer approaches to produce cells for stroke therapy ongoing.

That all said, a question mark undoubtedly hangs over the translation of the bystander effect from rodents to humans. I am not aware at this time of any clear demonstration of this mechanism in human subjects. Young, healthy rodents are quite different from the elderly victims of complex neurodegenerative conditions, and what works in the former may well fail in the latter.

Finally, it goes without saying that trusting to any therapy that has not been exhaustively tested in clinical trials in a fool’s game. Nothing could have looked more promising than the SanBio cells, but the hundred-plus individuals treated in that trial received nothing for their pains, except the gratitude of those of us who looked on in expectation. At least we are now a little wiser.


1. Krause M, Phan TG, Ma H, Sobey CG, Lim R. Cell-Based Therapies for Stroke: Are We There Yet? Front. Neurol. 2019;10:1–15. 

2. Gladstone DJ, Black SE, Hakim AM, Toward wisdom from failure: lessons from neuroprotective stroke trials and new therapeutic directions. Stroke. 2002;33:2123–36. 

3. Borlongan CV. Concise Review: Stem Cell Therapy for Stroke Patients: Are We There Yet? STEM CELLS Translational Medicine. 2019;69:904–6.