Skip to Content
Biotechnology and health

Three-parent baby technique could create babies at risk of severe disease

The approach was designed to prevent mitochondrial disease, but new evidence shows it might not work as planned.

petri dish with IVF procedure in progress
Getty Images

When the first baby born using a controversial procedure that meant he had three genetic parents was born back in 2016, it made headlines. The baby boy inherited most of his DNA from his mother and father, but he also had a tiny amount from a third person.

The idea was to avoid having the baby inherit a fatal illness. His mother carried genes for a disease in her mitochondria. Swapping these with genes from a donor—a third genetic parent—could prevent the baby from developing it. The strategy seemed to work. Now clinics in other countries, including the UK, Greece, and Ukraine, are offering the same treatment. It was made legal in Australia last year.

But it might not always be successful. MIT Technology Review can reveal two cases in which babies conceived with the procedure have shown what scientists call “reversion.” In both cases, the proportion of mitochondrial genes from the child’s mother has increased over time, from less than 1% in both embryos to around 50% in one baby and 72% in another.

Fortunately, both babies were born to parents without genes for mitochondrial disease; they were using the technique to treat infertility. But the scientists behind the work believe that around one in five babies born using the three-parent technique could eventually inherit high levels of their mothers’ mitochondrial genes. For babies born to people with disease-causing mutations, this could spell disaster—leaving them with devastating and potentially fatal illness.

The findings are making some clinics reconsider the use of the technology for mitochondrial diseases, at least until they understand why reversion is happening. “These mitochondrial diseases have devastating consequences,” says Björn Heindryckx at Ghent University in Belgium, who has been exploring the treatment for years. “We should not continue with this.”

“It’s dangerous to offer this procedure [for mitochondrial diseases],” says Pavlo Mazur, an embryologist based in Kyiv, Ukraine, who has seen one of these cases firsthand.

Three-parent babies

Mitochondria are little “energy factories” that float around in the cytoplasm of our cells. While most of our DNA is housed in the nucleus of a cell, a tiny fraction resides in mitochondria. This mitochondrial DNA, or mtDNA, is only passed down from mothers to their children.

This becomes a problem when the mtDNA carries a disease-causing mutation. Mitochondrial diseases are rare, affecting around 1 in 4,300 people in the US. And researchers are still working out how many of these cases are caused by mutations in mtDNA, as opposed to other genetic changes. But they can have serious effects, including blindness, anemia, heart problems, and deafness. Some are fatal. 

To avoid this, scientists have developed techniques that allow them to use mtDNA from a donor, along with DNA from a mother and father. These are generally called mitochondrial replacement therapies, or MRT.

There are a few different ways of doing this, but most teams use one of two approaches. Some scoop out the nuclei of two eggs, one from a prospective parent and one from a donor. Then they put the would-be parent’s nucleus into the egg of the donor, which still contains the cytoplasm, the fluid outside the nucleus that holds the mitochondria. The resulting egg can then be fertilized with sperm, creating an embryo that technically has three genetic parents.

Others first create a fertilized egg, called a zygote. Then they collect the DNA-containing nucleus of this zygote, which can be transferred to another fertilized egg that has had its own nucleus removed. The resulting zygote also has three genetic parents.

No one knows exactly how many babies have been born through MRT. Several clinics have described a handful of cases, mainly at conferences. An official trial at Newcastle Fertility Centre, in the UK, was launched in 2017. 

Since then, the Newcastle clinic has received regulatory approval to perform MRT for 30 couples with a risk of passing a mitochondrial disease to their children, according to published minutes of the statutory approvals committee of the UK’s regulatory body, the Human Fertilisation & Embryology Authority (HFEA). But the team has been extremely tight-lipped about the study and has avoided sharing any results with other researchers in the field.

A few other teams have been trying to learn whether the treatment works for infertility. Many couples struggle with unexplained infertility, and it is thought that the mix of proteins in the cytoplasm of an egg might somehow contribute to their inability to conceive. Because MRT essentially involves swapping the cytoplasm of one egg with that of another, some believe it might help treat some of these cases, and boost the success rates of IVF.

Dagan Wells, a reproductive biologist at the University of Oxford, is a member of one such team. Wells and his colleagues have also been trying to work out how safe the procedure is. Research in cells in a dish and in monkeys suggests there is a chance that MRT might not always prevent mitochondrial diseases. If this happens in people, it could have serious consequences.

Cell swaps

When you scoop out nuclear DNA, it is difficult to completely avoid taking some of the cytoplasm—including mtDNA—along with it. Embryologists have managed to limit the resulting so-called carryover to less than 1% of the embryo’s total mtDNA. “Usually that 1% ... shouldn’t be a concern, because the other 99% is healthy,” says Shoukhrat Mitalipov, an embryo biologist at Oregon Health & Science University, who is collaborating with Wells.

But research by Mitalipov and others has shown that this figure can increase over time. Scientists call the phenomenon reversion. This reversion could be a problem in couples where the mother carries a mitochondrial disease. If the percentage of “bad” mtDNA gets too high, it could cause disease in the child.  

To find out if this could occur in people, Wells, Mitalipov and their colleagues used MRT in 25 cisgender heterosexual couples, each of which had been through between three and 11 failed cycles of IVF. All of the women had been diagnosed with some form of infertility, and none had ever managed to become pregnant.

MRT is banned in the US, and the Newcastle clinic is the only one with approval to perform MRT in the UK, so the treatments were done at a clinic in Greece.

In each case, a woman with infertility first underwent standard IVF procedures that allowed doctors to collect a glut of her eggs. The “spindles” of these eggs, which contain the nuclear DNA, were then removed and put into eggs from a fertile donor that had already had their own nuclei removed. The resulting eggs were then fertilized with the male partner’s sperm to create embryos.

Once the embryos had started to develop, scientists took a couple of cells from them to look at their mitochondrial DNA. In all of the embryos, the vast majority of mtDNA came from the donor, with less than 1% from the infertile woman.

The team used a total of 122 maternal eggs and 122 donor eggs to generate 85 with donor mtDNA that were successfully fertilized with sperm. Twenty-four of these developed into healthy-looking embryos, and 19 of them were transferred to a woman’s uterus, resulting in seven pregnancies. One woman miscarried at nine weeks, but the other six pregnancies resulted in healthy babies, all of whom were born between the end of 2019 and 2020.

The team has also been checking the levels of mitochondrial DNA in the babies since they were born. The scientists have looked at DNA samples taken from swabs of the babies’ cheeks, as well as their urine, cord blood, and other blood samples. For five of the babies, the levels of their mother’s mtDNA has remained low, at less than 1%. But something strange has happened in one of the children.

At the embryo stage, less than 1% of this child’s mtDNA came from the woman with “bad” mtDNA, while over 99% came from the donor. But by the time the baby was born, the balance had shifted—with between 30% and 60% of the mtDNA coming from the mother. “It’s almost 50:50,” says Wells. “That’s a huge swing.” The results were published in the journal Fertility and Sterility in February.

“We were hoping we wouldn’t see [reversion] in babies,” says Mitalipov. “Now we have data to show that this is real—not just in monkeys … but in humans.”

“It’s the first time we’ve seen it in a person,” says Matthew Prior, the head of department at the Newcastle fertility center. He said his team has not seen reversion in any babies born following MRT—but he also won’t confirm if any MRT babies have been born there.

But while this is the first published report, a second case has been reported by doctors who performed the procedure at the Nadiya clinic in Kyiv, Ukraine. At an online meeting in 2020, Pavlo Mazur, then an embryologist at the clinic, told his colleagues about a baby boy who had also shown reversion.

The baby was one of 10 born in a pilot trial of MRT for infertility, says Mazur. He and his colleagues used a slightly different technique—the one that involves first creating an embryo and then removing its nucleus. This is also the approach used by the Newcastle team in the UK.

The baby, born in 2019, was the second child of a woman who had undergone MRT twice. Her first baby, a girl born in 2017, didn’t show any reversion, says Mazur—her levels of mtDNA from her mother remained below 1%. But despite the fact that the same team used eggs from the same woman, and performed the same procedure at the same clinic, her baby brother was born with around 72% of his mtDNA coming from his mother.

“We found it earlier [than Wells and his colleagues],” says Mazur. “We just never published it.”

Because the parents didn’t carry disease-causing genes in their mitochondria, these babies should be fine, says Wells. But, he says, “if this family were [carrying mtDNA mutations], this would be a big concern—60% is high, and it may cause disease.”

Risk of disease

Wells thinks it is difficult to predict how many babies might be affected by reversion. If his team did another 100 rounds of MRT, they might not see another case. Or they could see 90, he says: “The sample size is really too small to say anything about the frequency of this.” 

But Mitalipov is more confident. On the basis of the current study and his previous work in cells and monkeys, he believes there is around a 20% risk of reversion following MRT. In other words, if MRT is used to avoid passing on disease-causing mtDNA, there’s a one in five chance the baby will inherit potentially dangerous levels of that mtDNA anyway. “It’s not very rare,” he says.

The question is whether these odds are acceptable. For infertile couples without a history of mitochondrial diseases, the risks of using the technique appear to be low. But scientists using MRT in an effort to prevent mitochondrial diseases may be creating babies who could become severely unwell.

A 20% risk might be acceptable for some couples, says Prior. He says the results don’t change anything for the trial at Newcastle, which will continue as planned. “Obviously we will follow these results, and in due course we’ll publish our own results,” he says.

Heidi Mertes, a medical ethicist at Ghent University, says that it is important to think about what would-be parents would do if the technology were not available. If they would try for a baby regardless, then perhaps an 80% reduction in the risk of passing on disease-causing mtDNA is acceptable. But if they might otherwise consider using a donor egg, or adopting a child instead, then “those are better alternatives,” she says.

For Joanna Poulton, a mitochondrial geneticist at the University of Oxford, the 20% risk of reversion is “very concerning.” What’s more, the risk could end up being much greater than that. “There are mutations where quite low levels can cause problems,” she says. For some diseases, the level can be as low as 15%, she says.

And this is all complicated by the fact that mtDNA is messy. We can find different levels of mutations in different organs of a single person, and people with a mix of mtDNA can pass down either disease-causing or healthy genes in their eggs. A baby with low levels of “bad” mtDNA in the blood could still have high levels in the brain or muscles. This was also seen in the monkeys born using MRT, says Mitalipov. In a single animal, he says, the level of “bad” mtDNA could be “90% in the liver, and maybe 0% in the blood.” 

To complicate things even further, these levels can change over time. “A lot of these mutations progressively increase in life … so symptoms will happen much later,” says Heindryckx. Some mitochondrial diseases don’t make themselves apparent until people reach adolescence, for example. This all makes it very difficult to predict how many babies might be at risk of developing serious disease.

Problems with PGT

The finding also has implications for another, more established method of preventing mitochondrial diseases in babies.

Before MRT was developed, some clinics used a technique called preimplantation genetic testing (PGT) to screen embryos for disease. It is possible to pinch a couple of cells from an embryo created using IVF and check for disease-causing mutations. Prospective parents have the opportunity to avoid implanting any embryos that have high levels of “bad” mtDNA.

But the current findings suggest that PGT might not always work. If the levels of mtDNA can change as an embryo or fetus develops, there’s still a chance that the baby could be born with a disease. This might happen if disease-causing mtDNA replicates better than the healthy mtDNA. The balance between levels of “good” and “bad” mtDNA can change for the worse.

“We don’t know,” says Heindryckx. His is one of many centers that have performed PGT for couples with mitochondrial disease but didn’t follow up on the resulting children, he says. “It’s a wake-up call for us to do it more.”

We do know of one case in which it does not seem to have worked. A baby was born from an embryo that PGT revealed to have around 12% of the mother’s “bad” mtDNA. But by the time the baby was born, the proportion had shot up to around 50%. This baby had a plethora of symptoms, including atypical brain development, behavioral problems, and signs that he had experienced a brain hemorrhage.

Only a small number of babies have been born after using PGT to screen for mitochondrial disease, so again, it’s difficult to draw conclusions. The French center that pioneered the treatment, and has been offering it since 2006, recently reported that it has only had 29 babies born this way, says Heindryckx. His own center has only used it for the births of four or five babies in the last 10 years. And, as with MRT reversion, there’s a chance that babies who are disease free at birth might get sick as they get older.

“It’s alarming,” says Heindryckx. “We should also be following up the babies born after PGT, because it could be that this reversion is also happening there.”

A dangerous option?

What does this mean for MRT in the meantime? While the Newcastle team plans to proceed with its trial, others caution that, for the time being at least, we should pause the use of MRT for mitochondrial disease, and instead study it in people who don’t have these diseases, such as those with infertility.

Mazur himself refuses to use MRT for mitochondrial disease. And Heindryckx says the risk is too high for him—with a 20% risk of reversion, he says, there is no way the ethical committee at his institution would allow him to use MRT for mitochondrial disease.

Mertes says she has never been a fan of the MRT trials. Scientists knew beforehand that the trials were never going to be risk free, and that they involve a potential waste of perfectly good donor eggs and embryos. “In the end, you’re presenting an option to patients that is more dangerous than their alternative,” she says.

Experimental treatments like MRT also help to reinforce the idea that it’s very important for parents to have a genetic connection to their children, says Mertes. “Wouldn’t it be wiser to question whether it’s so important to have that genetic connection if the price you have to pay is a health risk for your child?” she asks. Parents can avoid all the risks that come with MRT by opting to use a donated egg in place of their own, or adopting a child, for example.

In the meantime, clinics that offer MRT need to update the information they provide “so that people know that this is a very real risk that they’re taking,” says Mertes. And both she and Prior think that the treatment should be restricted to those who “need it” or at least are adamant that they want a genetic link to their children.

Mitalipov is confident that scientists like himself will eventually come up with a solution to mitochondrial reversion. “We just need to figure out why it happens,” he says. “So far, no clue … but just give us time.”

Deep Dive

Biotechnology and health

What’s next for bird flu vaccines

If we want our vaccine production process to be more robust and faster, we’ll have to stop relying on chicken eggs.

The messy quest to replace drugs with electricity

“Electroceuticals” promised the post-pharma future for medicine. But the exclusive focus on the nervous system is seeming less and less warranted.

That viral video showing a head transplant is a fake. But it might be real someday. 

BrainBridge is best understood as the first public billboard for a hugely controversial scheme to defeat death.

People can move this bionic leg just by thinking about it

A mind-controlled prosthetic feels more like a part of the wearer’s body and promises to make walking easier.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.