Prenatal Screening for Autism is an Ethical Dilemma

Autism is highly heritable—but is it a disease? Should it be cured? Could screening for it lead to eugenics? We explore the complicated landscape.

In 2021, the Autism Research Centre at the University of Cambridge announced Spectrum 10K, a plan to collect and analyze the genomes of 10,000 autistic people. The stated aim was to understand diversity among autistic people, including why some have conditions like epilepsy and ADHD, and to develop more personalized ways to help each autistic individual. Wary of potential criticism, members of the study wrote on their website, “The Spectrum 10K team views autism as an example of neurodiversity and is opposed to eugenics or looking for a cure for preventing or eradicating autism itself.” Nevertheless, criticism came. 

David Gray-Hammond, an autistic consultant and the author of The New Normal, heard about the project and helped assemble a group called Boycott Spectrum 10K. Together they wrote a statement outlining their concerns, including fears that the research could lead to prenatal testing and abortions. They gathered signatures and sent it to the Health Research Authority, the U.K.’s regulator for medical studies. They also demonstrated outside the Autism Research Centre, holding signs that said, “AUTISTIC PRIDE” and “HANDS OFF OUR DNA.” 

“The atmosphere was definitely one of anger that we were being ignored, but solidarity that we were there together defending ourselves and our community,” Gray-Hammond says. Amid the outcries, including from activists within the community known as “autistic self-advocates,” the Autism Research Centre stopped recruitment and data collection. Less than three weeks after announcing the initiative, its director posted an apologetic letter to the study’s site announcing a pause. 

When does parental choice become eugenics?

That does not mean research into the genetics of autism has stopped completely, or that universities and companies are no longer developing prenatal tests for autism. Both efforts are progressing apace. And as new services enter the market and the clinic, new questions will arise. They will join pressing ones already asked: What is a disease? When does parental choice become eugenics? How can autism researchers and autism self-advocates best work together? 

The power to decide who gets to be born and who doesn’t will always seem like a technological gift to some but a cruel curse to others. And if you make that choice, are you playing God or simply being a responsible parent? 

No single “autism gene” or profile

Autism is often used by the public as an “umbrella term” covering many types of developmental delays, learning disorders, and intellectual disabilities, says Robert Green, a medical geneticist at Mass General Brigham in Boston. The two most defining characteristics of autism spectrum disorder (ASD) are repetitive behaviors and problems with communication. Special education professor Stephen Shore famously said, “If you’ve met one person with autism, you’ve met one person with autism.” There is no objective diagnosis; a clinician must use judgment to evaluate each individual’s behavior and context, and autism is highly variable. Some autistic people require full-time care into adulthood while others live independently. Some have learning disabilities while others don’t. Some are savants and wildly successful tech entrepreneurs. Many suffer from things like digestive disorders or epilepsy while others don’t. 

ASD is highly genetically heritable—90 percent genetic by some accounts—but the picture is messy. Researchers have found hundreds of associated genes that control neurogenesis, synaptic function, cell structure, metabolism, and other functions, but even when combined they can account for only a small percent of all cases of autism.

There are a number of challenges to studying the genetics of autism, including the wide range of behaviors, the complexity of the brain, the huge number of genes that contribute, and the lack of a perfect animal model. 

Environmental factors—including pollution, pesticides, and maternal inflammation, obesity, or infection—also have the potential to play a smaller role in autism through gene-environment interactions. For example, a paper last year found that the NHIP gene, which protects the brain against oxidative stress, was less active in fetuses who developed into children diagnosed with ASD. Autism rates are climbing quickly in the United States, increasing roughly 5-fold since 2000, from one in 150 8-year-olds to about 1 in about 30 today. Much of that is due to increased awareness and screening, but some may be attributable to gene-environment interactions, researchers say. 

In some cases, clinicians can predict whether a pregnancy has a high probability of leading to an autistic child. (In medical parlance, they’re high-risk pregnancies, but some autistic self-advocates object to the negative connotations of the term “high-risk” as applied to autism.) 

Prenatal genetic screening for conditions other than autism has become fairly routine. U.S. ob-gyns might do a genetic screening if a parent or sibling has a genetic condition, the mother is over 35, or something looks odd on an ultrasound. They might insert a needle into the abdomen or a catheter through the cervix to sample the amniotic fluid (in amniocentesis) or placenta (in chorionic villus sampling, or CVS). Fetal DNA is contained in both. 

They might then sequence the DNA and look for any of hundreds of single-gene mutations that can cause disorders such as cystic fibrosis, Tay-Sachs disease, and sickle-cell anemia. They might also do a chromosomal microarray analysis (CMA) to look for copy number variants (CNVs), in which a section of DNA is duplicated or deleted. John Pappas, a clinical geneticist and pediatrician at New York University, says that among children he sees who develop autism, he finds a genetic predisposition—including gene mutations, duplications, and deletions—in thirty to forty percent. 

It’s not always clear what to make of genetic predispositions before traits develop, though. Some structural variations in chromosomes reliably cause problems. “If it’s a really obvious one that’s well recognized, that’s one thing,” Green says. “If you have a deletion or duplication that overlaps with a known region but hasn’t been seen before in that exact configuration, you’re not often sure what to tell the mother.” 

“I wouldn’t trust it at this moment. I certainly wouldn’t terminate a pregnancy based on it.”

According to Stephen Chasen, the director of obstetrical ultrasound at New York Weill Cornell Medical Center, “There’s no genotype that’s implicated in many, if not most, individuals who have ASD. And there’s not one single genotype. There are many, many different things that have been found.” Some clinics raise hope about autism detection, but “commercial labs have been overpromising a lot of things for many years,” Chasen says.

Fetal DNA also circulates in the mother’s blood, which can be sampled for noninvasive prenatal testing (NIPT). Labs have used it to detect extra copies of chromosome 21, which can lead to Down syndrome. “They are starting to do screening for deletions or duplications of much smaller quantities of DNA, for conditions like DiGeorge syndrome, cri du chat, Wolf-Hirschhorn, Williams syndrome,” Chasen says. Noninvasive prenatal testing could someday play a role in screening for autism, but collecting DNA that’s clean enough to sequence is still a technical challenge, Green says. “I wouldn’t trust it at this moment. I certainly wouldn’t terminate a pregnancy based on it.” 

Other noninvasive tests could be on the horizon. A study published in Brain last year found that 29 percent of children diagnosed with autism had shown “ultrasonography fetal anomalies,” versus 16 percent of their closest-age siblings and 9 percent of the rest of the population. Anomalies associated with autism occurred in the brain, head, heart, and urinary system. Another study published last year looked at MRIs of fetuses. The development of autism was associated with enlargements in several brain regions: the amygdala, hippocampal commissure, and insula. While these studies are preliminary, they suggest that even if we have no genetic tests for autism in utero, it may be possible to develop a screen during pregnancy based on imaging biomarkers. 

Detection might happen even earlier in development. Some fertility companies do pre-implantation genetic testing of IVF embryos to provide polygenic risk scores—combinations of many mutations, duplications, and deletions associated with developmental outcomes. Companies include Genomic Prediction, Orchid, and MyOme. These tests can detect predisposition to diseases like cystic fibrosis or traits like deafness and dwarfism, or risk of developing cancer, cardiovascular disease, or mental illness. (Based on draft materials, MyOme also appears to have considered offering scores for educational attainment, household income, cognitive ability, and subjective wellbeing.) 

But polygenic testing for embryo selection faces difficulties and criticism. There might not be much variance among embryos from two parents, for one thing. Increasing the chance of a desired trait may increase the chance of an undesired one. The studies on which scores are based might not translate well to different patient populations. And they might not translate from adults to fetuses. Further, Green says, cells can repair mutations or abandon clones with mutations. “Let’s say you find a mutation in a cell from an embryo. You can’t be sure that that mutation is going to persist throughout the life of that fetus.” 

Testing might occur even before the embryonic stage. A 2021 study identified DNA methylation regions—places where genes had been switched on or off—that were more prevalent in the sperm of fathers with autistic children than in other men. A startup called Inherent Biosciences is building on the work to develop sperm tests. On the maternal side, Judy Van de Water, an immunologist at the University of California, Davis, has identified a set of eight antibodies in mothers that increase the risk of maternal autoantibody related autism (MARA) in offspring, a form of autism that tends to come with intellectual disability. She founded a startup called MARAbio to bring clinical tests to market. 

Difficult decisions

What does such information enable hopeful parents to do? One outcome of prenatal screening could be to help parents prepare for an autistic child, and to meet their needs early, which can lead to better outcomes. A study published last year reported that of children diagnosed with autism before the age of 2.5, two in three showed reduced autism traits over the following year or two, while only one in four children diagnosed later showed such improvement. 

Other parents might decide to terminate their pregnancies if a prenatal test showed a high chance of autism. And this is where the genetics of autism become hugely controversial. Autism self-advocates have taken to calling examples of research aiming to cure autism eugenics, in reference to the long-discredited early 20th century notions of “genetic hygiene” and the heinous public health practices it informed like forced sterilization—or worse—largely associated with the Nazis but also practiced in the United States and throughout Europe in the 1920s. They also worry that uncovering ways to detect autism in utero or in an IVF embryo will lead to widespread screening, abandoned embryos, and terminated pregnancies.

6 out of 31 who said they’d consider prenatal genetic testing for autism also said they would abort a fetus destined to develop ASD.

Those worries are not unfounded.

In the case of Down syndrome, which is easier than autism to detect prenatally, parents abort nearly 100 percent of the time in Iceland and Denmark, as well as 90 percent of the time in the United Kingdom and an estimated 65 percent of the time in the United States. Would they handle certain autism diagnoses the same way? In one study of parents raising autistic children, 6 out of 31 who said they’d consider prenatal genetic testing for autism also said they would abort a fetus destined to develop ASD. (One of those was a mother who said her autistic son had punched out one of her molars when he was three.) In a study of 333 Taiwanese mothers of autistic children, half said they’d have an abortion if a test said they “might have a fetus with ASD.” 

NYU’s Pappas says that when it looks like a fetus will develop into a child with strong autistic traits, the parents he works with choose abortion about half the time, though that’s often because of other severe medical problems the child is also likely to have. He says parents wrestle with the decision. “I don’t know how good we are, but we strive to be nondirective and just provide information.” Chasen, of Weill Cornell, says that if parents already have one autistic child, he can look for genetic signatures in that child and see if they’re in the fetus. If they are, he says, most parents he sees typically end the pregnancy. “It’s not my business what the decision is,” he adds, “but it’s my business whether or not the decision is informed.”

Uncertainty makes the decision more difficult. Parents can rarely know for sure if their child will have autism, what the traits will be, or how strongly the child will be affected. “We’re really operating, if not blindly, then in the dark here,” Chasen says. In his practice, people tend to decide based on the worst-case scenario. “They’ll never know whether it was the right decision or not, but what I’ve had pregnant individuals tell me is, ‘Well, that’s true, but if I continue the pregnancy, and it’s the wrong decision, I’m going to know that and I’m going to know it forever,’” he says. “That’s something that they and the family are going to have to live with.” 

One small study found that some women had lingering discomfort after receiving uncertain information. “‘Watchful waiting’ became the norm,” the researchers reported, “and concerns weren’t totally alleviated by normal sonograms or by delivering a child who appeared normal at birth and during infancy.”

Abortion isn’t the only possible intervention. Chasen says that if he finds genetic markers of autism in a parent’s existing child, the parents sometimes use IVF and select an embryo without that genotype. Or if no marker is found, the parent might simply select a female embryo, as autism diagnosis is 3.8 times as prevalent in males. 

Andy Olson, the CEO and co-founder of Inherent Biosciences, says that diet, exercise, stress, and toxin exposure can affect sperm quality, so if one batch of sperm from a father shows differential DNA methylation in regions associated with autism, Olson might recommend that the father change his lifestyle for at least a few months before procreating. Michael Paul, the CEO of MARAbio, says that if a mother has maternal-autoantibody-related-autism antibodies but no known genetic markers of autism, the parents might decide to use a gestational surrogate, or begin therapy with their child as early as possible. The company hopes eventually to offer treatments that selectively degrade the MARA antibodies or prevent them from crossing the placenta. 

Research may lead to not only prevention of autism but targeted therapies. “Since autism is multifactorial,” Green says, “there will likely be specific treatments and possibly even cures for some forms, which you would not be able to differentiate without testing, including genetic testing.” 

Abortion, murder, and genocide

In 2005, CNBC published an article titled “Autism research focuses on early intervention,” in which a researcher said that a prenatal test could be available within a decade. Meg Evans, an autism self-advocate, saw the article and wrote a blog post critiquing the motivation for such a test. She added a ten-year timer to the page and called it the “The Autistic Genocide Clock.” Six years later, she took it down, relieved that prenatal tests seemed more than four years away and that society had become more accepting of autism. In 2005, “most people knew very little about autism, and the media were full of sensational horror stories,” Evans says. “That has changed, thankfully.” 

Not as much as she and others would like, however. In 2012, a mother in Sunnyvale, California, shot and killed her 22-year-old autistic son and then herself. Media reports appeared to show sympathy for the mother’s struggles to care for her son. “I felt the voice of people with disabilities was really missing from this conversation,” says Zoe Gross, director of operations at the Autistic Self Advocacy Network (ASAN), “because to us, it’s very obvious that the tragedy isn’t that George was born autistic,” she says, referring to the murdered 22-year-old by his first name. “The tragedy is that he was murdered.” She started the Day of Mourning, an annual vigil to remember people with disabilities murdered by caregivers. 

“ASAN is opposed to the development of genetic tests for autism. We think that they’re unethical.”

Last year, ASAN released a statement on genetic research and autism. “ASAN thinks that finding a ‘cure’ for autism is bad and probably impossible,” it reads. It notes concern about genetics research, saying, “We think this ‘cause research’ is just another kind of ‘cure’ research.” It goes on: “Instead of prenatal testing, we want to get rid of ableism.” It does not distinguish between people distant on the spectrum: “We do not believe that any autistic person needs to be ‘cured.’ This includes autistic people with the highest support needs.” The statement makes clear that “We support the legal right of any person to have an abortion for any reason,” but it warns against societal pressure to abort. 

“ASAN is opposed to the development of genetic tests for autism,” Gross says. “We think that they’re unethical. We think that the focus should be on helping autistic people live better lives.” 

In addition to difficulties with flexibility and social interaction, autism may co-occur with intellectual disabilities, mental health issues, neurological disorders, gastrointestinal problems, inability to communicate verbally or take care of oneself, and even violence toward oneself or others. (In high school, I worked with autistic people who wore bite guards on their arms.) Over a lifetime, caring for an autistic person can cost millions of dollars, which many families can’t afford. 

Yet some people take issue with calling autism a disease, disability, or disorder, instead calling it a healthy neurotype among humanity’s diversity. “In the autistic community, there’s a big push to have it recognized through the social model of disability,” Gray-Hammond, the autistic consultant, says. “Society presents obstacles that are difficult to circumvent.” According to Sam Farmer, a neurodiversity self-advocate and the author of A Long Walk Down a Winding Road, “We feel as though the rules in society that govern how we’re supposed to behave were not written with us in mind.”

Advocates point out many strengths associated with autism, such as attention to detail and comfort with repetition. Farmer notes creativity, loyalty, analytical skills, honesty, hyper-focus, and empathy, and he names celebrities on the spectrum—Dan Aykroyd, David Byrne, Greta Thunberg—and others believed to have been—Albert Einstein, Isaac Newton, Ludwig van Beethoven. In 2021, Elon Musk famously revealed in front of a live television audience while hosting Saturday Night Live that he has ASD. “What kind of world would we live in without diversity in all of its forms?” Farmer says. Gray-Hammond notes autistic people’s strong sense of justice and their contributions to science and art. And even those who don’t achieve greatness, he says, “have just as much right to exist as the rest of us do.” 

Autism researchers have gone out of their way to allay advocates’ concerns, or at least to not aggravate them. When I asked a spokesperson at MyOme if she thought prenatal autism detection might be possible in the next few years, she replied, “No comment.” Spectrum 10K declined to comment on their work as they continued to consult with the autistic community. The organization Autism Speaks, which had previously merged with a group called Cure Autism Now, last year clarified on their site that “Autism Speaks does not support eugenics.” They did not reply to emails. “Information is valuable, no matter what,” Van de Water of MARAbio says, but still they won’t offer test rest results to currently pregnant women, at least until there’s a way to remove the effects of the MARA antibodies.

Many proudly autistic people are on edge due to the trauma they’ve experienced through stigma. “When we look at a study like the one Harvard Medical School embarked on,” Farmer says, referring to the fetal MRI paper, “we get scared. Where is this headed? What’s the scientific community’s agenda? Is it an agenda of eugenics?” According to Gray-Hammond, “Knowing the genetic causes of autism almost always seems to segue into talks about prenatal testing. And we know what happened with prenatal testing for Down syndrome.” 

The advocates I spoke with are pro-choice, which puts them in a tricky spot. They support a woman’s right to choose, but seem to oppose science that informs that choice. Gray-Hammond acknowledges the conflict. “I think the contradiction comes from the fact that we’re a group of people who have spent our lives being told there is something wrong with us,” he says, “and so talk of prenatal testing—I think we internalize that. It almost feels like, Well, would we be here if our parents had the opportunity to abort us? It becomes personal.” 

“This puts me into an interesting dilemma,” Farmer admits. “As a pro-choice autistic, where do I go with this? What I see ahead is a struggle for neurodiversity-community advocates like me to make the argument to society.” He ponders a toned-down approach. Instead of telling society to stop studying the genetics of autism, maybe his aim should be merely to cultivate acceptance of autism. “Don’t be so quick to rush to judgment that this is a baby not worth having,” he says.

Gray-Hammond strikes a similar pose. “I would ask anyone considering aborting a child purely because they might be autistic to think about whether that child has a right to exist based on disability,” he says. As for researchers, he’d ask them to think about their motivations. Do they want to reduce humanity’s diversity? But moral tensions remain. According to Gray-Hammond, “It’s a very complicated field that I feel will be debated in ethics probably for a much longer time than I am alive.”

Editor’s note: This story was updated on 7.10.23 to correct the list of companies that provide polygenic risk scores of IVF embryos.

Go Deeper