By Kai Kupferschmidt
What if you could read much of your child's medical future while it was still in the womb? Taking a major step toward that goal, one fraught with therapeutic potential and ethical questions, scientists have now accurately predicted almost the whole genome of an unborn child by sequencing DNA from the mother's blood and DNA from the father's saliva.
At the moment, prenatal diagnosis for a small number of genetic conditions is usually done from fetal cells that doctors capture from fluid in the womb (amniocentesis) or a snippet of placental tissue (chorionic villus sampling). But these methods, which require the insertion of a needle or tube into the womb or placenta, can cause miscarriages in about 1% of all cases. In 1997, chemical pathologist Dennis Lo, now at the Chinese University of Hong Kong, discovered that roughly 10% of the cell-free DNA floating in a pregnant woman's blood stream stems from her fetus. In 2010 in a paper published inScience Translational Medicine, Lo's group showed that enough such fragments of fetal DNA are there to reconstruct the fetus's whole genome, and that it should be possible to use this DNA to test the unborn child for genetic diseases without exposing it to the risk of an invasive procedure. "The biggest advantage is that you are saving all those babies that would be lost," says Lo.
But it is tricky to distinguish fetal DNA in the blood from the mother's DNA. One strategy makes use of subtle genetic variations that exist between a mother's pairs of chromosomes. In most cases, for a particular genetic sequence on a specific chromosome, the variants from each pair should be represented equally in the woman's blood. But in an expectant woman, whose child has received only one variant as part of its genetic inheritance, her blood will contain a little more of that variant because of the free-floating fetal DNA. If the mother's patterns of genetic variants, or haplotypes, are known, statistics allow researchers to conclude what variants she passed on to her offspring. In 2010, Lo showed that with both parents' haplotypes known, it would be possible to predict the child's genome from the DNA in an expectant mom's blood.
"But there is a big difference between showing something can be done and doing it", says Jay Shendure, genome scientist at the University of Washington, Seattle. In the new study, he and his team sequenced DNA from the plasma—blood minus the cells—of a woman who was 18.5 weeks pregnant. Comparing that DNA with genome sequences obtained from the father's saliva and the mother's blood allowed the researchers to identify fetal DNA sequences that they could computationally piece together into the child's genome. Comparison with the baby's genome sequence determined after birth showed the team's predictions to be more than 98% accurate.
"This is the first time that a fetus has been sequenced noninvasively," says Shendure, whose team reports its findings online today in Science Translational Medicine. The researchers also successfully repeated the experiment on a second, younger fetus—it was 8.2 weeks after conception, a time when less fetal DNA is in the mother's blood.
The scientists also tried to find new mutations in the child that neither the father nor the mother carried but that arose during the reproductive process as sperm and egg were created, or as their DNA meshed during fertilization. Such de novo mutations are important because some of the most severe genetic diseases arise this way, and parents and physicians would typically have no hint that a child will be affected until after birth. For the initial child in the study, 44 de novo mutations were identified after birth—none of which were anticipated to cause severe disease. While the fetal DNA analysis had predicted 39 of those, it also wrongly predicted 25 million other potential de novo mutations. This huge false positive rate could suggest that a fetus has a genetic disorder it doesn't.
"You only worry the mother or the father a lot," says Lo. That part will need a lot of improvement, Shendure acknowledges. "The upside is: It shows that it is actually possible to predict these mutations."
The scientists also couldn't predict what the child's genome looks like at roughly 1 million points where both parents had differences between their chromosome pairs. That's quite important for prenatal diagnosis in marriages between "blood relatives" such as first or second cousins, common in many cultures, Lo says. The two parents could each be carriers of a disease-causing mutation at the same spot due to their shared inheritance, he notes, and any offspring inheriting both mutations could then develop the condition.
Nonetheless, other geneticists are impressed. Arthur Beaudet of the Baylor College of Medicine in Houston, Texas, calls the new study "outstanding work" and predicts that "the near future will involve the routine sequencing of the genomes of fetuses during the first trimester of pregnancy."
Improving the technique to make it a clinical reality will only take "a couple of years," says Shendure, who estimates that his study cost about $50,000 per child. But he says it would be naive to think that all the data could actually be useful medically at that point. "There will be many mutations whose impact we just don't know."
Hilger Ropers, a geneticist at the Max Planck Institute for Molecular Genetics in Berlin, agrees. The genetic causes of most developmental disorders haven't been found yet, he cautions. In the meantime, he advises scientists to put more effort into finding the causes of such conditions. "After all, we can only eliminate those genetic disorders that we know."
Even though it should be possible to sequence a fetus's whole genome, Lo says, it might be better to do it in a targeted fashion so that information doesn't overwhelm prospective parents. That might also avoid some difficult decisions, such as whether to abort a baby who has mutations that forecast a difficult future. For example, Lo says: "I don't think it would be ethical to use this to screen for late-onset diseases like Alzheimer's or cardiovascular diseases, for example."
But even if parents who learn a child will have a severe developmental disorder decide against an abortion, they—and their doctors—might be better prepared for a baby's needs. Beaudet argues that while some cultures or religious groups might object to a fetus's whole genome being sequenced, it will ultimately become a normal part of prenatal care. Indeed, Lo says he is optimistic that in 5 years, the technical problems of the process will be solved. "That is why we need to have a debate about this now."
ScienceNOW, the daily online news service of the journal Science