In the film Gattaca, some of the population have been enhanced by genetic selection and manipulation. Image: Pictorial Press Ltd / Alamy Stock Photo

Imagine you could select your future child based on likely intelligence. Would you?

The possibility of screening embryos for complex inherited traits will force society to profoundly rethink its reproductive ethics
March 6, 2023

Imagine you’re planning to have a baby and are told there’s a method that can select the embryo to increase, by 2 per cent, the chance of them getting into a top school. Would you use it? A new survey found that more than four in 10 Americans say they would. This study of attitudes towards a technique called preimplantation genetic testing for polygenic risk (PGT-P) shows that there could be a substantial market for it if it is made available for such applications. The technology would not, say bioethicist Michelle N Meyer of Geisinger Health System in Danville, Pennsylvania, and her co-authors, just be adopted by a few “idiosyncratic individuals”, as has sometimes been suggested previously.

Of course, if you’re sensible then you will ask about the small print. How much does the process cost? Is it risky? Since it requires IVF, so that the prospective embryos can be genetically tested before implantation, would you opt for that even if it would not otherwise be necessary to conceive? But the researchers intentionally set a low bar for the 6,823 people who took the survey: assume the service is free, that it’s safe and that you’re going to be using IVF anyway. You might be reassured to know that the basic technique of preimplantation genetic testing (PGT) is already routinely used for couples who know their child could inherit a rare but serious genetic disorder, such as cystic fibrosis or sickle cell anaemia. In the UK, more than 500 genetic disorders can be screened this way, and the procedure seems to create no adverse health consequences for the child. It involves no genetic manipulation; it merely identifies those embryos free from the risk of disease. 

But it is one thing to use technology to avoid serious and often life-threatening inheritable diseases. It’s quite another to use it to select embryos for characteristics deemed advantageous (such as intelligence), or merely cosmetic (such as eye colour). That prospect opens up profound social, ethical and philosophical questions. Could it deepen social inequalities? Does it turn babies into products to be optimised by the latest technologies? The new survey shows the urgent need to grapple with such questions—because both the technologies and the market for such services are here already.

“There is an important distinction between embryo selection to avoid serious harm and for so-called ‘enhancement’, like greater intelligence”, says Peter Thompson, chief executive of the Human Fertilisation and Embryology Authority, which oversees regulation of such technologies in the UK. Some argue that nonetheless this kind of selection should be permitted to those who want it. After all, we allow people to spend thousands of pounds sending their child to private schools to give them an educational advantage, so why should we not permit them to circumvent the genetic lottery involved in regular reproduction? It’s not as if anyone is “cheating” by tampering with a genome to boost a child’s prospects (although the new survey also asked about such gene editing and found that 34 per cent of respondents would be open to it too). PGT merely involves selecting an embryo from all those produced in a round of IVF (a distinction that the glib term “designer babies” rather glosses over). Why would you not want to select the best?

Yet others fear the kind of future portrayed in the 1997 movie Gattaca, in which the population is divided into the Valids, who have been enhanced by genetic selection and manipulation, and the In-Valids who have not—and who are excluded from the top jobs. Wouldn’t the new technology risk creating a wealthy genetic elite? Some advocates and prophets of genetic screening for enhancement (including Dominic Cummings while he was a government adviser) say that the technology must be made freely available to avoid precisely that danger. 

But before getting into sci-fi scenarios, there’s a more basic question: how effectively can we select embryos for “desirable” traits like intelligence anyway? Such a process requires the variety of PGT called PGT-P, which takes account of many genes at once. This method is controversial not just because of the societal implications of selection for enhancement but because scientists disagree about its potential to make much of a difference. It is, however, already being put to limited use for health reasons. And as the technique advances yet further, the questions over the proper boundaries of its application will become only more pronounced.

Scientists use the word “monogenic” to describe diseases linked to a single gene, such as cystic fibrosis and Huntington’s. These can be nasty but are rare. Most common ailments, such as heart disease, hypertension, diabetes and cancers, have a much more complicated genetic component—they are “polygenic”, meaning that many genes are associated with the condition. (“Associated” is, by the way, the right word—it is not clear the respective genes can be considered “causes”.) Even in the case of the variants of the BRCA1 gene linked to breast cancer (which are also currently screened with PGT), the association is mild—most women who develop breast cancer don’t carry these variants.

We allow people to send their child to private schools to give them an advantage. Why not permit them to circumvent the genetic lottery?

For polygenic traits, it is not possible to be sure from an analysis of your genome just how that trait will turn out. However, as ever-more genomic data becomes available for many individuals, we become ever-better at detecting correlations between a genetic profile—all the different gene variants an individual has—and that person’s traits. This kind of statistical analysis, performed with a method called a genome-wide association study (GWAS), enables even very weak associations between gene variants and traits to be identified. It means that not just disease risks but any traits with a genetic component (and in fact pretty much any human trait, from intelligence and height to musicality and athleticism, has some genetic component) can be assigned a statistical “polygenic score”. We can say that a person with this genetic profile, say, has a 20 per cent chance of growing to over 5’10” and a 30 per cent chance of developing heart problems by the time they are 65. “The recent emergence of polygenic scores—which sum up the tiny effects of thousands of genetic variants on complex clinical, behavioural and social outcomes—could expand PGT to virtually any trait,” says Meyer.

Any forecast of outcomes for an individual based on their polygenic scores is purely probabilistic. In the example above, three in 10 people with those gene variants will develop a heart condition by age 65. Will this embryo be one of them? No one can say. All the same, it would make sense to select an embryo for which that risk is only 30 per cent rather than 60 per cent, right? But what if the former embryo also has a higher polygenic score for diabetes—or a lower score for intelligence? We have no objective way of evaluating such trade-offs. Sure, we can cook up ways to combine all health risk factors into a single measure. But how then to weigh risks against “positive” traits like intelligence and athleticism? Do we want a child who will “burn twice as bright but half as long”, like replicant Roy Batty in Blade Runner? What’s more, we might not even know all of the gene-trait associations, so that by selecting for one trait we could inadvertently be selecting for something else (less desirable) too. 

In any case, doubts have been raised about whether embryo selection for polygenic influences spread so thinly through the genome will actually make much difference to the outcomes. “I don't consider the statistics to be sound for [this type of] embryo selection,” says Ewan Birney, joint director of the European Bioinformatics Institute near Cambridge. “I doubt the benefits would outweigh the risks.”

Intelligence is perhaps the most controversial trait that PGT-P could select for. GWASs show that there is a clear heritable component to it—even if we can argue about the meaning of intelligence measures such as IQ, there’s a genetic correlation with educational attainment that remains apparent even when socioeconomic factors are taken into account. But there are no “intelligence genes”—the associations are to a huge number of genes, each of which makes barely any difference on its own. What’s more, Birney points out, around half of that genetic component is thought to be due not to genes that make the child smarter but because the child shares gene variants with parents whose own genetically influenced traits make them more engaged with the child’s education—it’s an indirect measure of the home environment.  

Meyer and colleagues rightly figure, however, that intelligence is one of the traits most attractive to the prospective PGT-P market. In 2018, one US company announced plans to screen for gene variants associated with low intelligence, but has since shelved them on the grounds that the service was still too controversial.

A child selected because of their “high genetic intelligence” score could turn out to be wholly unexceptional academically

While acknowledging these uncertainties about effectiveness, Meyer and colleagues chose a scenario that does not seem wholly unreasonable given the present state of knowledge. They told participants that about 3 per cent of US high-school graduates attend the top-100 ranked colleges, and that polygenic screening could potentially increase the likelihood of their child getting into those colleges to 5 per cent. While 43 per cent overall said they would (given the other stipulations) use the service, that proportion was higher (48 per cent) among those with at least a college degree—suggesting the potential to widen existing inequalities of opportunity or attainment.

But given the large overlap in the probability distributions of “higher intelligence” and “lower intelligence” genetic profiles, there can be no guarantees that the parents’ selection will yield the desired results. A child selected because of their “high genetic intelligence” score could turn out to be wholly unexceptional academically. Even if they are told this, will ambitious parents really be able to keep their expectations realistic? 

Despite continuing debate about the value and social desirability of PGT-P, it is already here—and being used to screen for some polygenic health conditions. In May 2020, a girl was born in the US after her embryo was screened for risks of heart disease, diabetes and cancer. Screening embryos for schizophrenia and other mental illnesses with a polygenic inherited aspect is already on offer; the provision of such services to people undergoing IVF could become lucrative sidelines for companies that currently provide DNA sequencing services to adults for health reports and research into ancestry. And some countries already allow non-health-related PGT for sex selection of embryos. Fertility companies don’t currently appear to offer PGT-P for conditions other than illnesses or disabilities, says Meyer—“but they haven’t ruled it out”.  

Meyer and colleagues ask whether the complexities of PGT-P can be conveyed “to achieve appropriate consumer literacy”. Can people be given the understanding (of genetics, risks and statistics) they need to truly evaluate the potential and limitations of the technique? That is an enormous task for which both the scientific community and the educational systems seem woefully unprepared. For example, school teaching of genetics begins with the exceptional cases of monogenic traits, which tend to follow the simple inheritance patterns identified in the 1860s by the Austrian friar Gregor Mendel.

There is some evidence that this approach entrenches a belief that genes “cause” traits, and moreover in a one-gene-one-trait manner that is subsequently hard to dislodge even when students learn the full complexities. Education understandably looks for a tidy story, but genetics is anything but tidy. The problem is exacerbated by the insistence of public messaging from the scientific community that the genome is the “instruction manual” of the human body, despite this metaphor having been rendered obsolete and misleading by modern genomics itself. It’s not just that scientists are struggling to find a better way to convey what genes do and don’t do; they can seem actively resistant to the challenge.

In other words, right now public understanding of genetics is almost perfectly tailored for people to be misled by companies seeking to profit from PGT-P. It would not be difficult to engender inflated expectations. For example, rather than companies saying PGT-P will make it 2 per cent more likely that your child will get into a good college, it’s easy to imagine them saying that it will increase that chance by 66 per cent (that is, 5 per cent as opposed to 3 per cent). They will say that you surely owe it to your child to give them the best possible start in life. And the peer pressure could be irresistible: respondents in the survey were more likely to say they would use PGT-P for these purposes when they were told that 90 per cent of others in their position would do so. Prospective parents might even feel cajoled or obliged to opt for IVF so as to make genetic screening possible when they have no other reason to do so—especially if, as some experts forecast, techniques of egg harvesting or even of in vitro production of eggs from stem cells make it easier and less invasive. 

Right now, public understanding of genetics is almost perfectly tailored for people to be misled

Of course, we should always be wary of assuming that people will do what they say they will do—but that works in both directions. We should also, Meyer and colleagues warn, be wary of assuming that people will refrain from choices that they judge to be morally problematic. (36-46 per cent of the survey participants either thought PGT-P for this application was morally wrong or were unsure.) And social norms on new technologies can swing quickly. In 1969, a poll found that most Americans considered IVF objectionable, whereas shortly after the first IVF baby was born (in the UK) in 1978, over 60 per cent supported it and said they would consider it themselves.

In the UK, PGT-P for selecting “positive” traits like intelligence could not be introduced without approval from the Human Fertilisation and Embryology Authority (which it currently does not have). But in the US, regulation of private IVF practices and of genetic testing is very weak: just because a service is of marginal or even unproven benefit does not ensure companies cannot offer it. The new survey makes a strong case that we need an urgent public discussion about what kind of society we want and how these reproductive technologies might shape it.