The NIH's New Autism Initiative Opens the Door to Research on Epigenetic Risk
A $50 million program is going beyond genes to find answers about what's causing rising prevalence of autism, emphasizing a need to look at "under-studied" factors
The traditional approach to environmental causation in autism is to examine how various toxicants or stressors affect fetal development. Under the approach I’ve been proposing for more than 12 years the focus should shift away from the baby and instead to the parents, and in particular the parental germline, that is sperm, egg, and their line of precursor germ cells. This paradigm is supported by research in animal models.
The NIH announced its new autism research initiative today.
It’s unnecessarily convoluted, it’s crazy fast (letters of intent are due June 6! applications due by June 27, 2025! what are they thinking, those are impossible deadlines which I assume will be extended), it’s all over the place.
But might it offer some hope for new answers about autism risk factors?
I have not exactly been a wallflower when it comes to questions of autism causation, and in fact have been stadium-screaming about the under-researched question of non-genetic, or what is called “epigenetic” inheritance in autism risk.
Epigenetics refers to the many layers of molecular factors that help control gene expression but do not alter the DNA sequence. Epigenetic programming is typically determined by genes, but it can also be susceptible to environmental insults.
This $50 million Autism Data Science Initiative (ADSI), while hardly a model of compositional clarity, seems reasonably calculated to address epigenetic risk in autism. In fact, it explicitly calls it out. The goal is dataset integration to foster “innovative analyses,” including “nongenetic factors” that may have a “role in increasing the prevalence of autism.” After an overview of some potential environmental risk factors such as chemicals and medications it says, “One mechanism by which these nongenetic factors may alter risk is by increasing or decreasing gene expression through epigenetic mechanisms.” It also calls for an “exposomics framework” to be applied to epidemiological and clinical data “to advance the study of GxE [gene x environment] interactions in autism.”
The request for applications hardly hides its interest in finding toxicants or other stressors that increase autism risk: “Given the significant increase in prevalence of autism over the past 25 years, enhanced, cutting-edge research is urgently needed to identify and understand the full complement of factors contributing to this rise.”
The ADSI offers “clues” as to what associations may be worth exploring. They include “altered prenatal and early-life brain development,” “dysregulation of neuronal and immune function,” and the “male bias and sex-related differences.” In my view, epigenetic mechanisms could explain all of those.
The funds — up to $5 million per award — will not support research in non-human animal models, which is a shame since mammalian models, like in mice and rats, is a very solid way to demonstrate how toxicants can cause heritable epigenetic shifts that perturb neurodevelopment.
The ADSI request for applications is, well, a bit impenetrable, but it seems the funds will support creation of new datasets, along with novel research on those datasets since the funds are intended to “support transformative and high impact research addressing a knowledge gap that deserves special emphasis and that requires collaborative and interdisciplinary data science approaches.”
This “knowledge gap” is a pretty strong theme. The request seeks “newly emerging and/or understudied factors that may contribute to autism” and its “increased prevalence over time.” Repetitive research need not apply.
I know I am the worst broken record in AutismLand. But I’ll repeat it here, as I have a zillion times before. The single most plausible yet under-studied factor in autism risk is the germline epigenome of the parents. See the graphic at the top of this post? It illustrates what I mean.
The traditional approach to “environmental” causation in autism is to look at prenatal exposures, i.e., exposures that could perturb the developing fetal brain. But let’s be honest. We’ve studied those exposures and with rare exception they are not exerting much autism risk at all. As the graphic suggests, instead we should be looking at exposures to the parents, and thus, to their germ cells.
Animal models have shown, for example, that exposure to modern anesthetic gases can change the epigenome, and thus transcriptional landscape, of egg and sperm, resulting in abnormal brain and behavior in offspring.
If this is happening in humans, it could help explain an element of autism risk, and the timing and scope of the mysteriously rising prevalence. Please see my 2024 Harvard presentation to get the details on this hypothesis or see my various scientific papers linked at jillescher.com.
Given the combined interest in epigenetics and novel hypotheses that could help explain increasing autism prevalence, the ADSI could potentially open up new avenues for this type of research. One can only hope that reasonable proposals with this under-researched intergenerational angle can be made, and accepted. Certainly I’ll be bugging my autism research friends to submit proposals along these lines (who will no doubt resist unless the insane deadlines are changed).
Finally, the ADSI also has intervention/treatment dimension that we can all agree is laudable. The announcement expresses frustration with the limited research on “how the complexity of genetic and nongenetic etiologies, together with variation in phenotype, is predictive of treatment response…. This ‘research to practice’ gap is a major impediment to improving the lives of individuals on the autism spectrum.” Without question, we need a finer-grained approach to treatment research so that we can tailor interventions to the type of autism in any given patient. Autism is not one thing.
That said, this intervention/treatment component of the grant program should have been a separate from the causation/risk program. I don’t see the utility of conflating the two realms of inquiry, since they address different scientific questions and would typically require different datasets.
I wrote this post very quickly and may have more to say about the ADSI after further info gathering and reflection.
The intervention aspect is very important and innovative. Intervention studies have already been done regarding early life screen exposure and the manifestation of autistic-like symptoms in toddlers. Parent training, stopping the affected child's screen time, and greatly increasing social interactions and play with non-electronic playthings has been shown to greatly lessen autistic-resembling symptoms in affected children. Some background: https://durablehuman.com/VirtualAutism
Very interesting. I don't think the effort will result in anything but some researchers getting paid, but maybe that's some benefit. Bobby Kennedy is an expert at doing this kind of flashy, political posturing that looks like something is getting accomplished even when it isn't.
As to your hypothesis that toxins cause epigenetic problems in parents' sperm and ova that results in autism, how would you ever be able to test that hypothesis? Since you can't do experiments on humans, the only option is causal inference, and it would be very hard to get the data to support that tool's use.
It does remind me, though, of when Richard Doll and AB Hill went out and interviewed so many lung cancer patients to see what caused their cancer. There were lots of possible suspects -- Richard Doll thought tar used to pave roads the most likely culprit. But as he said, as they talked to more and more people, 2/3 of the way into the study he gave up smoking.
Richard Doll and AB Hill had a debate with Ronald Fisher about just what was a valid test of the smoking causes lung cancer hypothesis. Even today it's unclear how to infer causes, but thanks to people like Judea Pearl, the causation toolkit has expanded.
As a side note, I was interested to see that sevoflurane may be causing epigenetic harm. Back in 1992 when I was working as a lawyer in Japan I did the legal work on the deal between Maruishi, Abbott, Baxter and Central Glass that gave worldwide rights to use the drug. It's a shame that it might have these terrible consequences.