Skip to main content

Verified by Psychology Today


Cell Therapy and Autism

A moment for re-evaluation.

There is a long history of abuse of autistic individuals by society in general, and the medical profession in particular. Steve Silberman, in his outstanding 2015 book Neurotribes, documents the inhuman and degrading conditions in which autists were held in the US, Germany, and throughout the Western world in the past. One might have hoped that merely documenting these facts would ensure that such abuse would cease in what we might consider to be caring societies.

So, what is particularly uncomfortable for biomedical scientists like me is that our abuse of the vulnerable continues. For example, recent reports in the UK­—not, I would suggest, the world"s least empathetic environment—indicate that some autists are still subjected to debasing treatment and unsuitable, uncaring therapies. One hesitates to consider how treatment plays out elsewhere.

I have written elsewhere of the treatment of the vulnerable with cell therapies. There is, of course, no more vulnerable group than children, and autistic children particularly so. How then should we consider the treatment of autistic children with stem cell therapies?

Two recent publications touch on this topic. My review of clinical trials of cell therapies for autism was published this week in Molecular Autism. Almost simultaneously, Geraldine Dawson and her colleagues reported on their most recent trial of a cell therapy for autistic children. These studies together allow us to draw some reasonably clear conclusions. Most importantly, in my opinion, the next steps are unambiguous.

Dawson and the team at Duke University is one of two groups in the US who have constructed controlled clinical trials in this area. The other is that of Michael Chez and colleagues in Sacramento. In both cases, the trials have been honest and accurate, following appropriate clinical practice as far as possible, and reporting the outcomes faithfully. Both teams have infused blood stem cells into their patients, either umbilical cord blood cells taken from the patients themselves, or similar cells donated from others. They have then followed the progress of the autistic children over ensuing months looking for improvements in behavioural symptoms.

Such studies generate lots of data, and the analysis is complex, but the fundamental outcomes are clear. The Duke team conclude that the stem cell treatment "was not associated with improved socialization skills or reduced autism symptoms." The Sacramento study was primarily concerned with safety, and their primary conclusion is that there were in fact no safety concerns. With regard to improvement in behavioural outcome, however, they concur with the Duke team. There were, they conclude, "no statistically significant changes" in behavioural outcomes.

It seems the therapy didn"t work, which is not to say that nothing was learned, a point to which I will return.

In addition to the work of these two teams, my review identified 11 other trials registered on the US National Institute of Health"s online database. Unfortunately, none of them were of the calibre of the two US studies. Some were clearly not trials at all. Rather they were "pay-to-participate" programs of a type that rather scandalously use the government website as an advertising vehicle. In contrast to the Dawson and Chez studies, all of these 11 reported successful outcomes, dramatically successful outcomes in some cases. One study reported improvement in 91% of patients. Boys, girls, children, or adults, almost everyone got better. What"s not to like? Well, almost everything as it turns out.

The first problem is that none of the 11 were adequately controlled. Most had no attempt at a control group at all, and in the one case that did, the control was poor being a totally separate patient cohort in a separate hospital.

We"ve long known the importance of proper controls in clinical trials. To properly assess a therapy, there has to be a group of patients equivalent in all regards, except for not receiving the therapy being tested. Importantly, no patient, nor carer, nor even clinician, must know to which group the patient belongs, because unsurprisingly expectation affects the outcome. We"ve known this forever, but one of the virtues of the Dawson study is it brings it sharply into focus in relation to the treatment of autism. They report a substantial "expectancy effect.' Part of the assessment was done by the patients" parents. Any small improvement that might have emerged as a consequence of the cell treatment was swamped by the improvement brought about by simply being in the trial.

How then are we to treat the 11 uncontrolled studies, where there was no control group for comparison, and where everyone, clinicians, parents, grandparents, even the hospital porter, knew that these kids had just gone through a painful (in some cases very painful) clinical procedure, hoping for a more promising future? This is quite apart from the cognitive dissonance that could be expected where substantial sums of money had been paid to take part. Are friends and relatives really going to be told that all was in vain, money wasted?

Let me expand on "very painful." The Dawson and Chez studies infused the cells into the bloodstream. This is not a trivial procedure, but has been used in other circumstances for years, and is relatively safe.

In two of the eleven other studies, the cells were infused directly into the child"s spinal column. I cannot begin to express how inexplicably irresponsible I find this. First, it is risky. This mode of administration of a drug is normally reserved for the relief of extreme pain, or for the administration of aggressive cancer therapy. It has never been validated for a use such as this.

Unsurprisingly, some of the patients experienced considerable discomfort with pain, vomiting, or spinal headaches. Some had seizures. How can this be acceptable to test a therapy of such uncertain outcome?

Second, it makes no sense. Where are the cells expected to go, and why? Such studies would normally be predicated with pre-clinical "biodistribution" data, but in this case, we have no idea where the cells ended up, or whether they even survived. Third, it is not obvious why this is even necessary. What hypothesis regarding the action of the cells requires this mode of administration?

Let"s go back to the beginning: Why do we imagine that a cell therapy might bring about an improvement in autism symptoms? The key mode of action is "immuno-modulation." There is a body of evidence that suggests imbalances of the immune system play a role in autism. Some autistic kids have disturbed levels of immune factors in their blood. Individuals with certain genetic variants associated with the immune response are more susceptible to autism. As I mentioned in a previous post, exposure to immune factors while in the womb increases the risk of a child subsequently being diagnosed with autism. This is all highly suggestive, but a clear hypothesis doesn"t really emerge. Something immunological seems to be out of kilter, but what?

This has led some researchers to propose cells as a therapy. Various cell types are "immunomodulatory": They regulate other immune cells and can act in various ways to adjust the immune system. Perhaps these cells might be able to correct the autism problem, whatever it is. And indeed, some cells do appear to have an effect when injected into experimental animals with behavioural deficits similar to those seen in autistic humans.

All this is fine as far as it goes, but you might sense a degree of vagueness. What exactly is the disturbance? Needless to say, the immune system is immensely complex. There are lots of ways in which it can get thrown out of balance. What do we think is actually going wrong? Is the problem transient, inducing autism in a fetus or newborn, then disappearing, or is it ongoing in autistic individuals throughout life? Are all autistic individuals effected, or just some? Given that there are lots of other drivers of autism — genes and environmental factors — we might imagine that it is the latter. In which case, which subset of autistic individuals has the problem, and how do we recognise them? Currently, none of these questions have answers, and until we do, cell therapy is a shot in the dark.

It need not be this way. All of these questions can be addressed by sound pre-clinical analysis. If there is an immune disturbance, we should be able to identify it precisely. This, in turn, should lead to a biomarker; some key immune molecule in the blood or the brain of autistic individuals that indicates something amiss. Our therapy then can be directed at rectifying this specific imbalance.

One of the problems with all of the studies cited above is that they don"t actually test the hypothesis. Since they have not identified a specific immune imbalance, they don"t actually know if any of their cohort of autistic individuals have it or not. They have no way, therefore, of knowing if the cells actually corrected the immune problem or not. If the cells did work to rectify the immune problem, but the patients still didn"t get better, then the scientists would know they are barking up the wrong tree. But in the absence of this information, the researchers don"t know if they failed because they failed to rectify the immune imbalance — in which case they need to seek out an alternative therapeutic — or if the imbalance was rectified but the autism was unaffected. A negative outcome tells them very little.

There is lots more wrong with these studies, as I describe in my review. The cells themselves are poorly characterised. The clinical protocols are hit-and-miss: How many cells should be injected? What time course should the study follow? But to cut the chase, we need to ask ourselves as a community: Are these studies ethical?

With regard to the 11, I have no doubt that they are not. The risk/benefit profile is unacceptably poor. These are children (mostly), unable to consent on their own behalf, being exposed to quite risky procedures, with very little chance of a successful outcome. The International Society for Stem Cell Research have proposed guidelines for proposed clinical studies, and this combination of risk, unverified protocols, unsure science, poor prognosis most definitely falls foul of these ISSCR recommendations.

My view of the recent Dawson study is that it decided to go for that shot in the dark. They came up short — predictably so in my opinion ­— but let"s move on. There should, I believe, be no more clinical trials for autism until some of the basics are sorted out: What, in biological terms, are we asking the cells to do, and how will we know if they have done it? We need to stratify patients so that only those with the appropriate pathophysiology go into the trial. If, say, just 20% of autistic patients have the "immune dysfunction," then a study that accepts all autistic individuals has no chance of producing a significant outcome.

Next, we need potency assays for the cells. Once we are clear on precisely what we are asking them to do, we need to test each batch of cells to be sure they are fit for purpose. This is currently standard practice for biological medicines, and there is no reason for us not to demand precisely this before treating our vulnerable autistic children.

Finally, when trials are fit to resume, let"s control them properly so we learn something whatever the outcome.

There were probably no Greta Thunbergs or Dara McAnultys in these trial cohorts, but we don"t need outstanding individuals to teach us that autistic lives are valuable. They should not be held lightly, least of all by the biomedical community.