This story is part of a week-long series on reproduction, from preventing premature births to childfree Reddit.
Most nights, talk around Kristin Hamann’s dinner table eventually turns to genetics. Her husband, Rodrigo Martinez, works for George Church’s whole genome sequencing company outside of Boston, so that’s hardly a surprise. When their 4-month-old daughter Julia started getting a rash on her face recently, Hamann wondered if maybe something in her own diet was triggering it. That’s when her husband, who had recently had all 6 billion base pairs in his genome sequenced, told her that he had a genetic variation that made him sensitive to gluten. Hamann stopped eating wheat and her daughter’s rash seemed to clear up. “For a first-time parent I’m fretting all the time and it would be so helpful to know these little things instead of just guessing,” she says.
But back when Julia was still just a 10-week-old fetus inside her womb, Hamann wasn’t so keen on having a giant mountain of information. Not that prenatal whole-genome sequencing is commercially available yet (though it’s definitely coming). But what is available is something called noninvasive prenatal genetic screening (NIPT). Based on a blood sample from mom, they have been used for several years to tell expectant parents if their baby might have, say, a chromosomal abnormality. Then the parents might make the choice to terminate the pregnancy—or to prepare for a child with disabilities. Makers of those tests, though, are already pushing the technology beyond its recommended uses to flag a rapidly expanding list of the unborn’s potential genetic flaws. But these bigger and bigger menus of genetic testing also come with less and less information about how predictive the data they reveal actually is. And as these types of tests become routine, women like Hamann have to figure out what they want to know, and what they’ll do with the information they receive.
Here’s an example of some of their options, from a New York–based company called Sema4, which launched its NIPT line in August of last year. There’s a standard screen that tells you about three disorders that result from having three copies of chromosomes 21, 18, or 13—Down syndrome, Edwards syndrome, and Patau syndrome. The next step up is a screen that also detects any missing or extra Xs and Ys, associated with common intersex conditions. These are not usually life-threatening, and in fact many people don’t know they have one until they hit puberty; the main features include infertility, altered sexual development, and, in some cases, intellectual disabilities. Some can be treated with hormone therapies during childhood. The American College of Obstetricians and Gynecologists doesn’t think NIPT screens should be used for anything beyond these conditions (though it is currently reviewing that position). But Sema4 also offers a test that looks for trisomies in chromosomes 15, 16, and 22. And finally there’s the Expanded package—Sema4’s most frequently ordered test. It also detects microdeletions—chunks of missing DNA—that cause rare but serious developmental disorders like Prader-Willi and Cri-du-chat syndromes. Symptoms for some of these microdeletions vary widely, from potentially fatal to quite mild. A few are commonly associated with attention deficit disorders and certain forms of autism.
Sema4 is not alone in offering screens for microdeletions, which, while relatively rare, show up equally in pregnancies for women of all ages. Leading sellers of NIPT screens, Natera and Sequenom, both introduced similar products back in 2014. Sequenom's latest version of the test also goes one step further, picking up any large missing or extra chunks of DNA, even if those sections are not known to be associated with any medical problems. In 2017, Natera launched a new test that reports on 30 rare, single-gene disorders, marketed to "women who want to know 'everything.'" Hamann and her husband didn’t opt for any of these; they were only interested in a potential Down syndrome diagnosis, so went with the standard NIPT trisomy screen. “I was really pretty choosy about only taking a test that would give us the information we needed to know at the time, to help us prepare for the baby’s arrival," she says.
Because NIPT screens can be performed earlier than other kinds of prenatal genetic tests—as soon as 9 weeks into a pregnancy—it gives women and their partners more of a head start for coping with the implications of any abnormalities that might turn up. But those decisions get more complicated as NIPTs start being used to identify genetic conditions that are not as severe as, say, Trisomy 13. (Survival rates for those babies are very, very low.) Many people want as much information about their babies as they can possibly get, but what do you do if you discover your kid might have not a major chromosomal defect but a more minor one? Few people know as much about genetics as Hamann to know that more information doesn't come with more wisdom about how to wield it. As companies amass these valuable stockpiles of prenatal DNA, should there be limits to how much data they can report back to prospective parents? And what about people without the means to join in the genetic data sprint to the womb? Noninvasive prenatal testing was supposed to make the modern pregnancy easier, safer, and less anxiety-ridden. But what if all it did was move the age of information overload in utero?
Before 2011, the only way to learn anything about an unborn child’s DNA was to stick a three-inch needle into the uterus and retrieve fetal cells from the placenta or amniotic fluid, procedures that carry a small risk of miscarriage. But then scientists discovered that a fetus sheds small fragments of DNA through the placenta and into the mother’s bloodstream. And they figured out how to collect and interpret those bits of fetal genetic material floating in just two 10-milliliter tubes of the mother’s blood.
While considered only reliable enough to function as a screening tool (a definitive diagnosis still requires the big needle), this noninvasive prenatal screening method has exploded through the medical community since its introduction in 2011. When WIRED wrote about NIPT about a year after the first test hit the market, four companies offered the service. Today there are more than 40, and global revenues for the industry are expected to exceed $2.5 billion by 2025. Earlier this month, at an international meeting of reproductive scientists in Paris, the inventor of NIPT, Dennis Lo, told the audience that 6 million women from 90 countries have so far been screened. Some experts say it is the most rapidly adopted test ever. “It’s moving faster than maybe anything before it in medical history,” says Glenn Palomaki, a researcher of preventive medicine and medical screening at Women and Infant's Hospital of Rhode Island.
But the rise of NIPT has been both incendiary and chaotic, in part because there’s so much money to be made, and in part because the technology innovations that make the increasingly sophisticated tests possible have far outpaced the research community’s ability to assess both their clinical utility and their impacts on society.
“If you look historically at prenatal genetic testing, there’s a very clear ratcheting effect,” says Ben Berkman, a bioethicist who studies NIPT at the National Human Genome Research Institute. In the US, no single legal authority dictates requirements or limitations on prenatal genetic testing. Professional societies of genetic counselors, medical geneticists, and ob-gyns provide guidelines for what should or shouldn’t be included in the screens and who should be eligible to take them. But, says Berkman, no matter what those groups actually recommend, there’s always some company offering a little more, marketing to families who want to be just “extra secure.” And, as his own research shows, wherever the upper limit of information is, most people want that.
In a study published earlier this month, Berkman surveyed more than 500 women who were eight or more weeks pregnant, and asked them to envision undergoing prenatal whole genome sequencing—what they would want to know if they could read their unborn child’s entire genetic sequence, not just look for evidence of a few conditions. (Such technology isn’t available commercially yet, but it has been done in the laboratory and the National Institutes of Health is currently assessing its clinical utility.)
Then the women selected their level of interest in receiving results from eight different categories, ranging from recreational traits (eye color, height, athletic ability) to fatal conditions resulting in death shortly after birth. The majority of women opted to have information from every category. “The assumption has been that people are more likely to want information that’s actionable,” says Berkman. That’s why the recommendations for NIPT currently limit tests to just a small number of serious disorders. But Berkman’s study showed that women were just as likely to want to know about their child’s genetic risk of autism, or common, treatable adult conditions like breast cancer or heart disease.
Even more surprising was what women intended to do with that information. In every category at least some respondents said the primary reason they wanted the data was to make a decision about potentially terminating the pregnancy. While that has been the underlying purpose of prenatal genetic testing since the beginning, the results—while hypothetical—indicate a shifting set of criteria for what might inform such a serious decision.
And with that comes profound ethical implications. “Everyone thinks they’re going to have a perfect baby but if you test any genome you’re going to find a large number of things that each confer a little bit of risk,” says Berkman. On average, each individual carries several hundred potentially harmful genetic variants, and for the vast majority of those variants, it’s unknown exactly how severely they impact human health. “My worry is that people will panic and make these reproductive decisions that maybe aren’t in line with their values based on information that’s not quite ready for prime time.”
More research is needed to investigate how the availability of new kinds of genetic information will influence people’s choices, to understand to what degree societies might slip into a eugenics mindset. Another recent survey of 1,000 Australian women found that only one respondent would consider terminating a pregnancy based on non-medical traits such as intelligence, and 80 percent specifically indicated that they would “definitely not” do so. But the majority of women were interested in using NIPT for autism and 16 percent said they would consider terminating a pregnancy if they got a result that indicated their child would have a low-functioning autism. The only conditions that garnered more support for termination were the three major trisomies.
Autism spectrum disorders are still poorly understood, particularly to what extent they are defined by one’s genes. And so no one is explicitly testing for them in prenatal screens at the present time. But some of the newer NIPT products do scan for a microdeletion—a missing chunk of DNA—on chromosome 22 that occurs in one in every 2,000 babies born. Approximately a third of children with the deletion will develop normally (or at most suffer a mild delay in speech)—the majority will have heart defects and more severe learning disabilities. People with the deletion are also at a higher risk for developing schizophrenia or bipolar disorder in adulthood, and 10 to 40 times more likely to have an autism spectrum disorder. Though these individuals represent only a small proportion of overall autism cases, as genetic links continue to crystalize, future tests will likely detect larger and larger numbers of people on the spectrum.
Today’s microdeletion screens also rely on a newer method of analysing fetal DNA, and how it works in real-world conditions is still not entirely clear. Data from large studies has been slow to appear, but the sparse evidence available suggests that microdeletion screens are much less reliable than similar tests for Downs syndrome. Marketing materials usually draw attention to a test’s detection rates—how accurately it picks up a genetic alteration. But the metric that matters most is actually something called the positive predictive value, or PPV, says Palomaki. That’s the number that tells you if you get a positive result, how likely is it to be correct. By that measure, NIPT tests for anything besides the three major trisomies simply don’t measure up. “Maybe only half the time they’ll be correct,” says Palomaki. In 2016, when researchers appraised the first six months of performance for commercially available NIPT screens that pick up the chromosome 22 microdeletion, they found the tests correctly identified the genetic defect only 40 to 44 percent of the time in the general population.
For that reason, Palomaki doesn’t believe that more is always better when it comes to prenatal genetic screens.“Once you get the information you can’t take it back,” he says. Rather than vying to provide more information of more dubious quality, Palomaki wishes NIPT companies would instead compete on providing the information that really matters to as many people as possible. ”It’d be better to have a test that everyone can have for all the basic trisomies than adding additional disorders that are uncommon or less severe and raise the price of testing such that 70% of the population can't afford it."
While insurance coverage in the US is expanding for NIPT screens, to include all high-risk and most regular pregnancies (the risk of trisomies goes up after age 35), it’s still not available for everyone—the tests run $800 to $3,000 each for patients who pay out of pocket. At those prices, some of the 40-plus companies making NIPT screens are just trying to grab a piece of a lucrative 130-million-births-per-year global marketplace. But all those reproductive testing products are also a means of amassing valuable genetic data troves, a resource pharmaceutical companies are increasingly interested in signing lucrative deals to mine.
For Sema4, reproductive genetic tests are the foundation of an even more grandiose undertaking—collecting a universe of biodata so big and so powerful that it will blot out the one-size-fits-all model of care that has for so long dictated how doctors treat their patients. Sema4 was founded in 2017 by Eric Schadt, dean for precision medicine at the Icahn School of Medicine at Mount Sinai in New York and a well-documented big dataphile. Like Craig Venter and Patrick Soon-Shiong before him, he hopes to remake the US health care system in the image of personalized medicine. But where they tried (and failed) to reach exabyte scale by selling rich folks super expensive, super data-rich health screening services, Schadt is betting he can get there by relying on average people just trying to have healthy babies to be Sema4’s “growth hack engine.”
“Pregnancy provides for a really sticky patient,” says Schadt. Sema4 offers tests for women at each step of that journey—carrier screens that assesses up to 283 genes in each parent associated with diseases that range from mild to severe, four varieties of NIPT, and cheek-swab tests for after the baby is born. “All those provide for natural ways of creating more engagement with a patient,” says Schadt.
When a woman takes any of Sema4’s tests she is walked through a multistep consent process that make it clear how the company would like to use the genetic information from both her and her unborn child for research and provides opportunities for opting out. In return, Sema4 says it will provide her with more personalized recommendations for accessing care tailored just to her. “You can think of it like the Amazon model” says Schadt. “If we understand you better than the health system, we’re going to help match you with better tests and medical products.”
The consenting process is much more involved than a few simple click-throughs, but by being transparent about how Sema4 plans to use their data, Schadt says more patients are willing to share it. In addition to consenting to have their deidentified genetic information used for research, 50 percent of Sema4 customers also authorize the company to retrieve every piece of medical information on them from every health care system they’ve been a part of. That includes electronic health records, medical imaging, lab reports, prescriptions.
Combined with similar data from a research partnership with Mount Sinai and information scraped from public biomedical repositories, Schadt says Sema4’s database today exceeds 10 million patients. Hundreds of thousands of them are women who’ve had babies, a patient population that has not been historically well-studied (pregnant women are still excluded from clinical trials, for example). Now Schadt is using Sema4’s data to try to understand and predict what makes some women more susceptible to some of the dangerous conditions that can unexpectedly co-occur with pregnancy, including preeclampsia, gestational diabetes, and preterm birth. Mount Sinai is embarking on a prospective clinical study to see if Sema4’s models can accurately identify women who will develop such conditions and improve both mom and baby’s chances of staying healthy through the pregnancy.
In the meantime, Schadt will keep collecting as much data as women will give his company. Sema4 is currently changing the methods on all their reproductive testing products to an increasingly common so-called low-pass whole genome sequencing approach, which will create a very rough sketch of the fetus’s genome, including which genes came from which parents. They’ll only report back to parents on the disorders covered in whichever test they order. But all the data will feed back into Sema4’s genetic research bank. Data from patients who’ve consented to sharing for research purposes can be accessed by Sema4’s industry partners, including pharmaceutical companies, for a fee; academics can collaborate free of charge.
Does Schadt worry that his competitors will start offering screens for less and less serious diseases, potentially siphoning off this valuable datastream? “Yeah. If I could generate every molecule of data on everybody on Earth I’d do it.” But he’s not yet ready to start selling frivolous genetic forecasts just to elbow out adversaries. He says for now Sema4 is sticking to only reporting on genetic diseases that parents can do something about—whether that’s early treatment or termination of a pregnancy.
Still, he’s got an eye on what else is out there, including companies developing genetic screening tests for intelligence and inherited forms of deafness, and IVF clinics where couples can choose which sex they want their baby to be. “People are already making selections on embryos and pregnancies far outside the deleterious stuff," says Schadt. "That experiment’s going and we’re waiting to see how societies will respond. And then as attitudes change and studies get done on whether this information reduces health burdens or just makes people more anxious, then we’ll expand. But not before then.” A prenatal test available today might not tell you what your baby will look like or how likely they are to develop cancer or baldness as an adult, but data from that scan will almost certainly power tests that someday will.
How We Reproduce
- It's Either the Best Time or the Worst Time to Have a Baby
- How Big Data Could Help Prevent Premature Births
- Childfree on Reddit—Because Apocalypse, That's Why
- You Know What Else Can Reproduce Now? Robots
- Gadgets and Gear for Making (or Not Making) Babies
- Male Birth Control Could Actually Happen. But Do Men Want It?