HOUSTON – Polygenic scores derived from large genome-wide association studies can at least partially explain the variability seen in complex human traits, such as height or cognitive ability. Combined with the ability of preimplantation genetic testing to scan the entire genome of a human embryo, this has raised the specter of "designer babies" where prospective parents choose embryos based on traits they find desirable.
But, according to a new study presented here at the American Society of Human Genetics annual meeting, such fears may be premature because trait predictions based on polygenic scores are not very accurate, at least for now, and because of other limitations to embryo screening.
For their study, presented at the meeting yesterday by Shai Carmi, a researcher at the Hebrew University of Jerusalem in Israel, the researchers focused on polygenic scores for intelligence and height. The study was published as a BioRxiv preprint in May and has been accepted for publication in a peer-reviewed journal, he said.
For intelligence, a polygenic score based on a large GWAS with 300,000 individuals currently explains about 5 percent of the differences seen in this trait, and for height, a polygenic score based on a meta-analysis of data from 700,000 individuals accounts for about a quarter of the variability observed. However, larger GWAS and improved statistical prediction methods may in the future increase the proportion of the variance that can be explained, the researchers wrote.
To start, they simulated possible polygenic scores in virtual human embryos resulting from couples — both actual couples and randomly matched ones — from two cohorts for which they had actual genomic and phenotypic data available. One cohort consisted of 102 actual couples of Ashkenazi Jewish origin and came with both genome-wide genotypes and information on height, the other cohort comprised 919 Greek males with genome-wide genotypes and data on their general cognitive function.
For the 102 real couples as well as 500 simulated couples, the researchers generated 10 virtual embryo genomes by combining their genotypes. They then calculated the polygenic scores for these embryos to predict their future height and IQ. If the polygenic score was used to select an embryo for height, they found, the average gain was predicted to be about 2.5 cm, ranging from 1 cm to 6 cm, depending on the couple. If the polygenic score was used to select for cognitive function, the average gain was predicted to be about 2.5 IQ points, ranging from 1 to 7 points per couple. They also found no differences between embryos simulated from real couples and from randomly matched couples. The researchers cautioned, though, that environmental and unknown genetic factors might lead to differences in the actual gain in height or IQ from the predicted gain.
In order to assess how well actual height can be predicted from polygenic scores, the researchers next turned to real families. Specifically, they looked at 28 large families with grown children, with family size ranging from three to 20 children.
As it turned out, the children with the highest polygenic score for height were only the tallest child in seven of the families. In fact, they were shorter than the family average in five of the families, suggesting that selecting embryos based on polygenic score would not have resulted in the tallest possible child.
In addition to inaccurate predictions from polygenic scores, embryo selection remains impractical for other reasons, Carmi said. For example, IVF usually creates fewer than 10 embryos, and a proportion of these may be aneuploid, in particular if the woman is of advanced maternal age. Also, it is not guaranteed that the embryo selected will actually successfully implant and result in a live birth.
In addition, many traits are correlated with other traits, for example, a high IQ will increase the risk for autism or anorexia. It might also be difficult to select for several traits at once, as well as for a reduced risk for various diseases.
Finally, while the study showed that embryo selection for specific traits would not work very well at the moment, it did not address other reasons why it might be undesirable. "[W]e do not consider here the ethical, moral, and legal underpinnings and consequences of embryo selection," the researchers wrote. "We hope that this work will promote an open and evidence-based debate of these aspects among the public and policymakers."