Researchers raise concerns about using genetic risk scores to pick ‘healthier’ embryos

As more people turn to in vitro fertilization for help with conceiving, a host of companies is capitalizing on the opportunity by offering screening services that allow hopeful parents to select embryos least likely to result in a baby with genetic abnormalities or life-threatening diseases.

But in a report published Wednesday in the New England Journal of Medicine, a group of researchers is questioning the strength and ethical implications of such risk screening services, especially as some companies seem to be pushing beyond disease prediction to screen for traits such as cognitive ability.

The report underscores the concern among geneticists, legal scholars, and ethicists that applying genetic research to develop risk scores and select embryos might be imperfect and less robust than patients and companies think. The authors also outline strategies for companies and clinicians to use when communicating with parents seeking to use these tools.

“Risk scores being calculated should be taken with a very large grain of salt,” said I. Glenn Cohen, a professor at Harvard Law School who specializes in health law and bioethics and was not involved in the study.

Genetic testing has come a long way from the days of testing for a single disease-causing gene, such as for cystic fibrosis or Tay-Sachs disease. A surge of studies that associate multiple genetic variants with particular traits and diseases mean scientists can now analyze the many small genetic differences across a person’s genome to calculate a polygenic risk score, the probability that an individual will have a certain trait, condition, or disease.

These scores, provided often by 23andMe and other companies, have been used in adults to identify the risk of diseases such as type 2 diabetes and breast cancer. But genetic testing companies are now also providing them to individuals undergoing IVF to help them pick an embryo to implant.

Companies like Orchid, whose motto is “Have healthy babies,” provide genetic counseling that includes this type of analysis. Another, called Genomic Prediction, offers PGT-P, or preimplantation genetic testing for polygenic disorders, a kind of report card with risk probabilities for schizophrenia, cancer, and heart disease.

“I kind of felt like it happened pretty quickly. First one company and then two and then three. They all just kind of sprung up,” said Michelle Meyer, assistant professor at the Center for Translational Bioethics and Health Care Policy at Geisinger and an author of the study.

Beyond a boom in the number of such services, researchers worry that companies could make more ambitious claims about how they can help predict other aspects of a child’s future, including cognitive ability, education, and income. The authors of the report claim that as recently as December 2020, Genomic Prediction advertised screening for intellectual disability, though no such claim appears on its website now. “This is sort of science fiction hitting us right now,” said Anna Lewis, research associate at Edmond J. Safra Center for Ethics at Harvard University who was not involved in the report.

Genomic Prediction did not respond to a STAT request for confirmation it offered this service until recently. When the New Jersey-based company was founded in 2017, Patrick Turley, assistant research professor at the University of Southern California and first author of the new report, knew he had to look into what the authors call “Embryo Selection based on Polygenic Scores (ESPS).” The company was one of the first to offer screening for polygenic risk scores on embryos rather than in adults.

“That made us think that we should maybe get together to actually crunch some numbers and talk about what the actual expected potential benefits and risks are, about the things that we know and don’t know yet about the service, and the ethical questions surrounding it,” Turley recalled.

In the paper, Turley and his co-authors explore the risks and benefits and, using simulations, point out a number of factors that reduce the predictive capacity of using polygenic risk scores to select for embryos.

Chief among their concerns is that the vast majority of genetic research used to predict risk has been done in individuals with European ancestry, which makes risk assessments better at predicting outcomes in those with the same ancestry. In the author’s simulations, they found that the risk of conditions like hypertension and type 2 diabetes were different when considering diverse ancestries.

The research is also done using individuals from different families, which are more genetically varied than those from the same family. In contrast, for IVF, the process looks at the genes of embryos from the same parents, which would have much less variation between genes, and therefore between their risk scores.

Substantial interactions between genes and their environment may also influence an individual’s risk of a disease or certain trait. For example, one study found a relationship between genetic variants and the use of hormonal birth control in the development of breast cancer.  Due to these interactions, a risk score’s predictive power would be stronger for individuals in the same environment.

When choosing between two genetically similar embryos that would eventually be born into the same environment, the authors say, the probability that they will develop a specific trait may not be that different. One study cited in the report, for instance, found that the difference in height between selecting an embryo against short stature and a randomly chosen embryo would only be about 1 inch. “There are a lot of things that make it so that the benefits of something like embryo selection are smaller than you might think,” said Turley.

Both Orchid and Genomic Prediction told STAT they do not screen for educational attainment or IQ scores. Genomic Prediction emphasizes that even a small difference in risk between embryos could be meaningful. “In our opinion, these are an important benefit to IVF families. Families with a history of the disease would benefit more than this,” Laurent Christian Asker Melchior Tellier, CEO of Genomic Prediction, wrote in an email to STAT.

Another point of contention for the report’s authors is that of unintended consequences from trying to select for traits that are usually impacted by multiple genes. For example, using data from genetic research, the researchers performed a simulation in which they “selected” an embryo based on its score for educational attainment. Doing so, they found, increased the risk of developing bipolar disorder, possibly because the genetic signals for these outcomes overlap.

“Who knows how many other types of outcomes may also just get dragged along if people start adopting a service?” Turley said.

The unintended consequences could trickle beyond the individual level and into society. Selecting an embryo to avoid the development of particular traits or conditions, such as intellectual disability, sends a message, said Gabriel Lázaro-Muñoz, an assistant professor at the Center for Medical Ethics and Health Policy at Baylor College of Medicine, who was not involved in the report. “We have these companies telling us, in a way, these embryos are lives that are more worth living.”

Using risk scores for embryo selection in this way also has social justice implications because in vitro fertilization and genetic screening are costly and not covered by insurance, said Cohen. “You’re going to create haves and have nots.”

Experts say companies need to thoroughly communicate these nuances to their customers. The report authors lay out a list of recommendations for companies to communicate with customers about risk scores, such as emphasizing that the predictions have high variability and presenting the reduction in absolute risk instead of relative risk; the latter can exaggerate the significance of a difference. “Even scientists have a hard time understanding the nuances that surround the potential risks and benefits of a service like this,” said Turley.

Researchers also urged government regulation of these companies in the U.S. “This is essentially an unregulated market, it is quite remarkable,” said Lewis. The Federal Trade Commission mandates that companies avoid misleading consumers, and the authors called for the agency to confirm whether genetic testing companies are being transparent about the services they provide. This, according to the authors, could mean the FTC will have to establish what is considered enough scientific evidence to support a claim and make sure the companies are abiding by that.

But the more immediate concern is that “the science is at a premature stage, and we’ve got a lot more thinking and talking and gathering of evidence to do,” said Lewis. ”Because what the risks are, at least at the individual level, are not yet clear.”