Back in July, Science ran an interesting news article about an on again, off again clinical trial of chelation therapy in the treatment of autistic children. I found the story fascinating because it highlights some of the challenges in setting up ethical research with human subjects — not to mention some of the challenges inherent in trying to help humans to make good decisions grounded in the best available scientific knowledge.
From the Science article:
Believing that mercury in vaccines triggers autism, thousands of parents, often at the advice of their physicians, have given their autistic children drugs to bind, or chelate, and remove heavy metals from the body. Some say the over-the-counter or off-label treatment can improve poor language skills, social problems, and other symptoms of the disorder. And yet the drugs are not risk-free, and the underlying rationale–that mercury from vaccines causes or worsens autism–has been roundly rejected by many scientific studies.
NIMH [the National Institute of Mental Health] has argued that the widespread use of the drugs creates a “public health imperative” to conduct a rigorous trial so that the institute can inform parents and physicians about any merits or dangers of the drugs. But some researchers and ethicists oppose studies that they say have no chance of working–and little chance of persuading the most zealous advocates–especially if the drug poses a substantial risk. “On balance, it’s not an ethical study,” says vaccine researcher Paul Offit of the University of Pennsylvania.
The reality on the ground is that many autistic children (2-8% of them, by some estimates) are receiving chelation therapy, sometimes with their doctors’ approval, sometimes not. Yet, in the absence of a carefully designed and controlled clinical study, we don’t have good scientific knowledge about how this treatment affects autistic patients — whether it has any effect on their autism, or what impact it has on their health more generally.
Given the availability of chelation therapies (either over the counter or through a physician willing to prescribe them for off-label use*) and the understandable urgency with which parents of children with autism are looking for treatments that might help their kids, establishing something like conclusive data on their effects in the treatment of autism seems like it would be a good thing. So, is there a way to set up a well-designed clinical trial that is also ethical?
Recall that the Belmont Report, the document that lays out the U.S. government’s guidelines for the protection of human research subjects, requires that research with human subjects maximize the possible benefits and minimize the possible harms of the particular intervention being studied. In a clinical trial, it is crucial that researchers consider the potential benefits and harms to the participants in the trial.
In a clinical trial on the effects of chelation therapy on kids with autism, the participants in the trial would be kids.
Here’s how the Science article describes the clinical trial in question:
NIMH wanted to conduct a study of the common chelator DMSA, which is approved by the Food and Drug Administration for treating lead poisoning. The idea was to give 120 children, aged 4 to 10, with a range of autism symptoms either DMSA or a placebo. After 12 weeks, NIMH researchers would evaluate the children to see if their social and language skills had improved. It would be the first controlled study of a chelator on autism.
As with any research on human subjects, the scientists who wanted to conduct this research first needed to get their protocol approved by the Institutional Review Board (IRB). In addition to presenting their experimental protocol to the IRB, they needed to make the case that the expected benefits of participating in the clinical trial outweighed the potential risks to trial participants. With chelation therapy, one of the risks is that the chelators will remove essential minerals (like calcium and iron) from the body; researchers addressed this potential risk by including multivitamin treatment in the protocol. The benefit, potentially, would be an improvement in social and language skills.
The crucial question for the IRB is whether the likely benefits outweigh the likely harms.
Given the body of scientific research that “roundly rejects” the hypothesis that autism is caused by heavy metal poisoning, an IRB might estimate the likelihood of any benefit as being very small. If it’s small enough, the risks from the chelator — even with multivitamin treatment — might be unacceptably high. But how do you estimate these risks before you conduct the relevant clinical trials?
“Clinical equipoise” is the tricky issue here. The reason to do a piece of research is to gain knowledge where there is currently uncertainty. Doing an experiment in conditions where you have a very clear reason to expect a particular result isn’t “research,” at least not research that can be expected to produce new knowledge. Rather, such an experiment is usually described as an unnecessary duplication (since it will almost certainly tell us something we already knew) — or a student lab.
In research with human subjects, though, if there is any kind of risk to the subjects from participating in the research, there must be some expected benefit to the research in the form of new (and useful) information generated in the study. If the study promises to tell us nothing but what we already knew, you’re exposing subjects to risk for no good reason — and that, clearly, would be unethical.
To have a reliable estimate of the benefits and harms of chelation therapy on kids with autism, you’d need to draw on the findings of something like a clinical trial. But it’s not ethical to expose human subjects to the risks of being in a clinical trial if you already have the knowledge such a clinical trial would produce.
In science, no matter how strong your hunches may be (and how much suggestive theory or data you can line up behind those hunches), you don’t know it until you have the experimental results to underwrite that knowledge. However, to the extent that getting that knowledge might subject human subjects to more harms than benefits, an IRB can withhold approval to conduct the trial that would answer the question.
In the case of the NIMH study of DMSA described above, the IRB did approve the protocol and the clinical trial began. However, within a few month of the launch of this study, research from animal models was reported that changed the best guess at the potential harms to trial participants:
An October 2006 online study in Environmental Health Perspectives examined the impact of DMSA on rodents. Although the drug helped rodents overcome lead poisoning, when it was given to rodents without lead it caused lasting cognitive and emotional problems. The finding “raises concerns about the use of chelating agents in treating autistic children without elevated levels of heavy metals,” says senior author Barbara Strupp of Cornell University, although she notes that it’s not known what the threshold might be for such adverse effects. The children in the autism trial would not have had elevated levels of mercury in their blood (otherwise, they could not ethically be given a placebo).
The animal study here suggested another potential risk of chelation therapy beyond the depletion of calcium and iron. Adding the potential for cognitive and emotional problems to the risks being undertaken by the human subjects in the trial shifted the balance of likely harms and likely benefits, and the researchers went back to the IRB with this new information. Based on the IRB’s review of this new information, the clinical trial was halted.
In some ways, the fact that this clinical trial did not run to completion will be a source of frustration — we don’t have the knowledge that this study would have produced. That knowledge might have been very helpful to physicians and parents trying to do their best to help kids with autism.
To the extent that building this knowledge would have exposed the kids in the clinical trials to certain harms, however, the well-being of those individual kids as best we can gauge it on the basis of the knowledge we do have trumps our desire to build the new knowledge. And arguably, while the available knowledge does not directly answer the question the NIMH clinical trial was designed to answer**, this knowledge does shift the weight of the evidence behind our hunches. Here, for example, the fact that neither the kids in the treatment group nor the kids in the control group (who got a placebo rather than a chelator) had elevated mercury levels in their blood seems to undercut the hypothesis that autism is caused by mercury poisoning.
But a blood test showing a normal level of mercury may not be persuasive to the parent of the autistic child, looking for some treatment that could make a difference.
Part of the motivation behind this clinical trial, it seems, was to be able to show such a parent solid scientific evidence that chelation therapy wouldn’t make any positive difference for his or her kid with autism. However, to the extent that chelation therapy seemed likely to make a negative difference, the trial would expose kids to harm just to prove their was no benefit. The results might be persuasive, but you’d run a real chance of hurting kids to get them.
This means you need to make the case against chelation therapy based on the preponderance of other evidence.
To the extent that significant numbers of parents and physicians seem already to be making decisions about chelation therapy as an autism treatment on the basis of anecdotes, rather than the clinical trials that haven’t been conducted or the other available evidence about the effects of chelators and the etiology of autism, it seems like there is a case to make here. Thousands of autistic kids may be exposed to harms that are not well-understood by the adults caring for them, in the hopes that they will receive a benefit that is not proven. How can you tell if a therapy will help or hurt your kid? Even if you’ve heard from another parent that a particular treatment “helped” his or her kid, what can you conclude about the likely effects on your own child?
Not everyone is inclined to demand a solid scientific basis for his or her medical treatments. Some people are happy just to follow the doctor’s orders, while others will follow the lead of friends, support groups, and websites. But every practitioner of evidence-based medicine ought to demand a solid scientific basis for his or her therapeutic recommendations. Physicians ought to be up on the research findings we have (and what they mean). It might even be good for physicians to be able to explain to their patients (and their patients’ guardians) how ethical commitments to patients account for some of the gaps in our knowledge.
*”Off-label” use of a drug refers to use to achieve a therapeutic goal for which it has not been officially approved, usually because there is not (yet) a sufficient body of research to support its use in achieving that therapeutic goal.
**And the available knowledge could not answer the same question without rendering the NIMH study unnecessarily duplicative — and thus, an unethical study from the point of view of exposing its human subjects to unnecessary risks.