Genetic Screening: Risks and Benefits (Patreon)

[This is a transcript with references.] 

The American NBA player Shawn Bradley is 7 foot 6. A genetic analysis found that his height score is more than 4 standard deviations above the population mean. If you could have picked the genetic makeup of your child, would you have chosen one that tall? Before you answer, consider the cost it’d take to adjust your door frames.

Just what genes you pass on to your baby has so far been mostly up to chance, but this is no longer the case. We now have the possibility to choose what traits our children will inherit. In this video I will *not talk about gene editing, but I’ll talk about the less invasive method of genetic screening of human embryos, which allows parents to decide what embryos to keep or toss. But how? Right swipe? Well, not quite. How does genetic screening work? Is it legal? And is it a good idea? That’s what we’ll talk about today.

First things first, what is genetic screening? It’s more specifically called “preimplantation genetic testing”, PGT for short. PGT is typically performed on embryos that are three to five days old, when they consist of about eight cells. For the test, one takes several cells from the embryo and analyses their genetic makeup. Depending on the result, the embryo is either discarded or implanted into the uterus. In some cases, these tests are also done on eggs before fertilizing them in the first place.

This method can readily be applied to patients who have sought in-vitro fertilization, IVF for short, because they experienced difficulties conceiving. But of course, it can be used by any couple that’s willing to leave the implantation of an embryo to a doctor. And not only *can it be used, it’s being used already.

In the US, about 1 to 2 percent of all live births in 2020 were conceived using some sort of assisted reproduction, of which IVF is the most common one. About 40 percent of those were genetically screened, that’s about 30,000 babies a year, and the numbers are rising.

In Europe, the fraction of IVF with these scans is roughly the same. See, the US and Europe have something in common besides obesity. Data from other countries is hard to come by, but clearly genetic screening is no longer a rarity. It’s now done by health providers on a regular basis.

Genetic screening makes sense in particular for couples with hereditary genetic diseases in the family line who want to make sure their children aren’t affected. But it also makes it possible to select other traits, which brings up a lot of ethical concerns.

Should you let parents determine their children’s genetic traits? What if the parents make a decision that the child later doesn’t like? What if parents try to breed children that will bring in money on the expense of the child’s health? What if they want children that are ill, so they qualify for social help? Can the children sue their parents? The doctors? The government for passing the wrong regulations? That YouTuber who talked about it years ago?

The topic becomes even more controversial for genetic tests on embryos that are already growing in the womb, because then the mother needs an abortion to terminate the pregnancy. Under which circumstances are we to allow that? Should the decision be up to the mother alone? Should the father have a say? What genetic properties should make it permissible to abort a pregnancy?

You may say genetic screening makes sense in families with hereditary diseases. But what counts as disease? Female pattern baldness runs in my family and I’ll almost certainly get it too. If I’d had a chance, I would have liked to spare my daughters. But is it a disease? Will everyone end up looking like a fragrance model? That’d be terrible because I’m nearly faceblind and to me those models all look the same. Then again, maybe faceblindness will be gone too?

And there are good reasons besides hereditary diseases for parents to undergo genetic screening for their babies. For example, if they have a child with leukemia and no relative qualifies as bone marrow donor. In that case, parents might want to have a second child that can act as a donor to save the first. This is becoming increasingly common; there’s even a name for it, the “Savior Sibling.” If you had been brought into the world to save your brother’s life, how would you feel about that?

Those are all ethical questions that I don’t know how to answer, so let me instead give you more information about what tests are presently available, and how good they are. There are currently five types of those tests on the market: First, you can look for the sex chromosome to figure out whether it’s a boy or a girl. And then there are PGT-A, M, SR, and P tests. We’ll look at those one by one.

A particularly simple type of test is one that counts chromosomes, to check if there are some too many or too few. Those tests are called PGT-A, where the A stands for aneuploidy (‘a new ploy dee). Literally this means “not well folded” which makes it sound like a test for decent origami. But the technical term really just stands for an abnormal number of chromosomes, anything that deviates from the normal double set of 23.

At the moment, PGT-A tests are the most common genetic screening tests, making up about 80 percent. These tests can find conditions like the Turner syndrome, which is a missing or damaged second X chromosome, Patau Syndrome, the Edwards syndrome, and the Down syndrome, each of which is caused by a third copy of a chromosome.

What you do with this knowledge is another question entirely. Children with down syndrome look somewhat different, have a lower-than-normal IQ and will likely require help throughout their life. They have a high risk of heart conditions, but with good medical support their life expectancy is above 60 years. To make matters more complicated, it’s extremely rare, but sometimes trisomy 21 doesn’t lead to Down syndrome. So, yeah, genetic determinism isn’t as simple as it seems.

Still, most embryos with missing or additional chromosomes don’t successfully implant, or if they do, result in miscarriages. The risk of embryos with multiple or missing chromosomes increases with the age of the mother. According to estimates, at age 35, about half of the embryos have that problem. By age 40, about 80 percent. And most of those will result in miscarriages.

If you test the in-vitro created embryos before implantation, you can pick the ones most likely to result in a successful pregnancy. One may worry that removing cells for testing can damage the developing embryo and lead to pregnancy complications or developmental problems for the child, but this is exceedingly rare if the test is done properly. A number of studies have found that the risk for these complications isn’t any higher in the genetically screened group than in the control group.

All of this might make you think it’s a good way to increase the chances of having a baby, but unfortunately it isn’t so simple.

A 2018 study looked at half a thousand IVF candidates from clinics in several European countries and Israel. They found that the group with PGT-A screening was not any more likely to have a successful pregnancy outcome, though they did find a slightly decreased risk of miscarriage in the screened group.

A 2016 study in America found that there is a benefit, but only for women older than 37 years. For those, the chance to get pregnant increased by more than a factor 3. However, there were only 60 women in that group.

Another 2019 paper with participants in the US found that PGT-A screen doubled the chances of having a baby for women above to age of 38 but in the lower age group, the chance actually decreased. So that’s getting really confusing.

Another study from a 2020 used more than years’ worth of data from a clinic in California and compared the outcomes of PGT-A screened pregnancies to the US average. They found that the screened patients had a higher chance of having a child than the national average in all age groups but the difference was much smaller than the previous studies had found. This result is also difficult to interpret because the patients of this clinic might differ from the US average in several other ways.

What does this mean now? The large variation of the outcomes is probably due to small samples and different test providers. There are many different laboratory procedures that must work properly to get good test results and at least at the moment the outcomes depend sensitively on how good the lab is.

PGT-A tests themselves have a typical diagnostic efficiency of over 94% which is pretty good, but those are the numbers under ideal conditions. The best test won’t help if someone puts the wrong sticker on the test tube or sneezes into the petri dish. My reading of those results is that there’s tentatively a benefit of PGT-A screening for women who near the end of their fertile years, but you have to be careful with picking a good lab. Since the data is so inconclusive, the UK Authority on Human Fertilization & Embryology has for now rated the procedure with the color “red”, which means that there is insufficient evidence to demonstrate a benefit.

Let’s move on to the next test. The easiest hereditary diseases to identify are those in which only a single gene is affected. This includes diseases like cystic fibrosis and sickle cell anemia. Tests for those conditions are called PGT-M tests, where the m stands for “monogenic”. They are a little more complicated than the PGT-A tests because they actually have to look for the gene, but they are at the moment the second most commonly used ones.

There are more than four thousand known monogenic diseases. PGT-M can currently be used to avoid over six hundreds of them. The one that is most commonly tested for is Huntington’s disease followed by cystic fibrosis. Theoretically the risk of misdiagnosis can be as low as 0 point 1 percent, practically, the risk is almost certainly higher because those tests are often not done under ideal circumstances.

Is it a good idea to do such a test? I’m afraid that again there’s no simple answer to that question. That’s because in most monogenetic diseases, having a particular variant of a gene doesn’t mean you’ll necessarily get the diseases, it’s just that the risk is higher. An example is a group of mutations that increase breast cancer risk which is the mutation that Angelina Jolie has. Her risk for developing breast cancer during her lifetime was estimated with 87 percent. She had a preventative double mastectomy to reduce the risk.

In most cases, however, the risk associated with a mutation isn’t that high. That’s why there’s always that 90 year old uncle who smokes 50 cigarettes a day.

A third type of test that is currently available is PGT-SR, where SR stands for Structural Rearrangements. No, we’re not talking about home renovations, these tests look for uncommon chromosome rearrangements. Examples of diseases caused by this are intellectual disability, speech delay, or birth defects such as a cleft lip. These conditions can run in families, but they can also occur spontaneously.

Much like the PGT-A test, screening for those conditions can maximize successful implantation of the embryo, and reduce the risk of miscarriage, stillbirth, and babies with chromosomal abnormalities.

And then there are tests for polygenetic traits, called PGT-P, which are the most controversial ones. As the name suggests, the tests screen for traits that are affected by more than one gene.

Again, though, in most cases, genes alone don’t determine whether you’ll have a certain trait, it’s just that certain combinations are known to increase the likelihood of having it. Some diseases for which we know polygenetic risk factors are type 1 and 2 diabetes, breast cancer, heart attacks, hypertension... – it’s a really long list.

But it isn’t just diseases that you can test for this way. In principle, you can also test for genes that are known to be correlated with, say, height, or intelligence. This is totally possible with existing technology, though at the moment no one admits doing it.

These tests, like pretty much all DNA-tests that you can order online, do not decode the entire genome. This is possible, but it’s expensive and also impractical, because the effects of most of the genes in the DNA aren’t known anyway. Instead, with a PGT-P test, one looks for specific places of interest in the DNA, typically a few hundred or a few thousand, depending on the disease.

These places are called single nucleotide polymorphisms, SNP for short. These are particular expressions of genes that are more likely present in people with the given trait. One SNP alone might make a negligible difference to a person’s risk of developing a disease, but adding up thousands of them may lead to a significant increase in risk. It’s like with conspiracy theorists, one on their own isn’t too worrying but if you have a thousand of them, it’s a problem.

This screening procedure exists, and it’s already being used. The world’s first PGT-P baby was selected in 2019. The baby was born in summer 2020. It’s a girl, her name is Aurea.

Aurea’s genetic screening was done by the American company Genomic Predictions that was founded in 2017. Genomic Prediction now offers screening for schizophrenia, four heart conditions, five cancers, and type 1 and2 diabetes. These services can now be used by clinics and health providers all over the world. The company advertises its services with the slogan “Reduce disease risk for your future child”. Genomic Screening is no longer the only company offering polygenetic tests, there is now also Orchid and MyOme.

These companies all work similarly: They use their test results to estimate how much patients can reduce the risk of the child to develop a disease. They get these numbers by training artificial intelligence on large databases of sampled DNAs of people whose health problems are on record. They look for correlations between gene-expressions and health problems. And that can then be used to calculate the probability that genetic screening will reduce the risk.

For example, in 2020 a group of scientists used a computer simulation to evaluate how genetic screening affects the prevalence of certain diseases. They used DNA information from 12,000 pairs of siblings with data about their health status. Then they imagined they could either have randomly picked one of each pair or done genetic screening on their embryos and picked one based on the results.

They do this thing for several diseases and calculated how commonly the diseases would have occurred with genetic screening. You can see that the risk of all diseases is reduced though the amount of reduction differs with the disease. For example, the risk of heart attack went down from 3.8 percent to 2 percent. So, in that case genetic screening leads to a risk reduction of almost 50 percent which is pretty remarkable.

This computer simulation focused on the prevalence of diseases, but you can do the same thing to study what would happen to, say, height, or IQ. Both height and IQ are known to have a strong polygenetic component. This is rather uncontroversial among scientists, though it’s somewhat unclear how big the influence is.

Twin studies suggest that IQ in adults is 57 to 80 percent heritable. The estimate for height is that it’s 70-90 percent heritable. Those percentages are somewhat tricky to interpret, but loosely speaking they tell you how much of the variation in the population comes from genes. Basically, it tells you how much you can blame your parents.

What’d happen if you screened babies for height or IQ was subject of a 2019 paper published in the journal Cell. By using a computer simulation for polygenetic screening, they found that within one generation, the children gained on average about 2 point 5 centimeters in height and 2 point 5 IQ points. No one knows what happens if you repeat this for several generations. But if we look at how it went with dog breeding, well, weird things may happen.

But we live in the real world and not a computer simulation. So how do we know that the numbers which researchers put into their simulations are reliable in the first place? We don’t know for sure because to check the numbers, we’d have to follow a group of screened babies as they grow up and count how many develop the conditions they were screened for. It’d take decades for the results to come in.

We do know however, that there are some general problems with polygenetic screening. For one, the current DNA data banks are not representative of human ethnic diversity. Most of the data currently come from developed countries, and people with European ancestry are vastly overrepresented. This means that for people with a different genetic background, polygenetic risk estimates are extremely unreliable.

Then there is the issue that correlation doesn’t mean causation. People who are more likely to have gene A might be more likely to have disease X not because the gene causes the diseases but because the prevalence of both has a common origin, maybe they both originated in the same village near Chernobyl. And then there’s the issue that genetic variation explains only a proportion of the total risk in any case.

This is why, a year ago, a large team of international researchers warned that “to date, no clinical research has been performed to assess [the] diagnostic effectiveness [polygenic screening] in embryos.” And that “Patients need to be properly informed on the limitations.”

Okay, now that we have seen what’s scientifically possible, let’s have a look at what’s legally possible. Existing regulations differ from country to country. In the European Union most countries currently have laws saying that sex-selection is illegal and polygenetic screening is permissible only to prevent incurable hereditary diseases. There is however at the moment no common EU law. In the UK, genetic screening can also be used to select a match for a sick relative. So, as usual, it’s somewhat of a mess in Europe.

In Russia, genetic screening is approved for hereditary disorders but not for sex-selection. China has no laws. They have *guidelines against sex-selection but it’s widely practiced anyway. And in the US, embryo testing and selection is currently not legally regulated. They too have some guidelines from professional associations but they aren’t legally enforceable so it’s somewhat of a wild west there.

In particular, sex selection in the US is legal. What this’ll do to society remains to be seen. At the moment it doesn’t have a large effect because the total number of people doing it is small. But a fairly large study in 2020 found that genetically screened babies in the United States are significantly more likely to be male than the population average. This was just a data analysis so it doesn’t tell us why this is the case, could be because parents prefer male children or because something to do with the procedure.

How do we deal with the ethical implications of genetic screening? Researchers from Australia and the UK have proposed what they call it the “Welfarist Model”, that allows selection for any trait which is associated with well-being, or selection against a trait which reduces well-being. The authors admit that there’s a minor problem with this idea, which is that we don’t know what “well-being” means. They call for more ethical and philosophical research into that, including for example the question whether selecting for intelligence would bring more well being.

What do you think? Is it easier to be happy if you are dumb or smart? Let us know in the comments.