Predicting genetically complex traits from the DNA of cells taken from an IVF embryo (above) isn't easy.

NATIONAL GEOGRAPHIC IMAGE COLLECTION/ALAMY STOCK PHOTO

Screening embryos for IQ and other complex traits is premature, study concludes

HOUSTON, TEXAS—Two years ago, news headlines began to appear about a development that made many human geneticists uneasy. A U.S. company planned to offer a test for embryos created through in vitro fertilization (IVF) that screened the entire genome for DNA variants linked to cognitive ability, in order to help couples avoid having children with intellectual impairment. Many ethicists fear such multigene analyses could one day be used to screen embryos for desirable traits as well, such as tall stature or high IQ.

For those disturbed by the prospect, a study reported here last week at the annual meeting of the American Society of Human Genetics (ASHG) may come as a relief: For now, the strategy would not work very well.

Researchers, led by statistical geneticist Shai Carmi of the Hebrew University of Jerusalem, calculated exactly how much of a boost in IQ or height could be expected by scanning for relevant DNA markers in a batch of embryos and choosing those with the highest scores. The result: The gains would be slight, and prospective parents might even end up discarding their tallest or smartest potential offspring.

The work "is the first to empirically test the viability of screening embryos" for desirable traits such as height and high IQ that are influenced by many genes, says sociologist and demographer Melinda Mills of the University of Oxford in the United Kingdom. Such embryo screening goes beyond today's testing for single-gene disorders and currently "isn't plausible," she concludes.

Such tests are based on a polygenic risk score, a tool for evaluating a person's likelihood of a disease or trait that has emerged over the past decade from genomic studies combing through variable DNA markers in many thousands of people. Although having any one DNA variant may barely raise the risk of, say, heart disease, adding up the effects from hundreds or thousands of these markers can generate a score that helps identify people at relatively high risk of common diseases. Some direct-to-consumer DNA testing companies have begun to give customers polygenic risk scores for diseases such as heart disease, breast cancer, and diabetes.

Testing embryos, however, is hugely controversial, because of both the scientific limitations of such polygenic scores and the prospect of designer babies. Undeterred, a company called Genomic Prediction last year began to offer to test cells plucked from an IVF embryo for millions of DNA markers to produce risk scores for some common diseases and for "intellectual disability" or low IQ. Co-founder Stephen Hsu, a physicist at Michigan State University in East Lansing who has branched into genomics, says that for now, the company is not returning genetic scores predicting high IQ because "society is not ready for it."

Still, testing embryos for desirable traits could be coming soon. Most people agree it's not a good idea, but there are no data, Carmi said at the ASHG meeting. To find out whether the strategy could work, his team created virtual genomes for potential embryos by combining the DNA profiles of "parents." One parental group included actual and randomly chosen pairs of men and women from 102 Ashkenazi Jewish couples with recorded heights and the other 919 randomly paired Greek men who had cognitive test scores. The team then calculated polygenic scores for the synthetic genomes to predict height or cognitive ability. For the five embryos typically generated in an IVF cycle, the theoretical height gain from selecting an Ashkenazi couple's embryo with the highest "tallness" score was about 2.5 centimeters (with a range of 1 to 6 centimeters). Selecting for the Greek virtual embryo with the highest IQ-favorable score brought a similarly limited gain in cognitive ability, just 2.5 IQ points (in a range from one to seven points), Carmi said.

His team also looked at the actual genomes of adult offspring in 28 large families (about 10 children on average). They found that for height, unknown environmental influences, which could include factors such as diet, and genes not represented in the polygenic score apparently overpowered the assessed genetic markers: In only seven of the 28 families was the sibling with the top score for height the tallest; in five families, the best scoring child was shorter than all siblings' average height. Although Carmi's team didn't have similar real-life data for IQ, University of Edinburgh population geneticist Peter Joshi expects that any intelligence polygenic score would be even more unreliable. "You might be wrong almost as often as you're right," Joshi says. (Carmi declined to discuss his study, which is online as a preprint and in press at a journal.)

An embryo screening test that uses a polygenic score to predict low IQ is likely to face the same limitations, says Joshi, who views such testing as unethical. Hsu, however, emphasizes that Genomic Prediction doesn't use its risk scores to screen for subtle IQ distinctions, but rather to avoid embryos with rare "outlier" DNA profiles for which scores predict a high risk of an IQ below 75, indicating intellectual disability.

Carmi's embryo study comes on top of other problems with polygenic tests. Most such scores have been derived from scanning DNA of people of European ancestry, making them of limited use for other ancestry groups. And recent studies have found they often aren't as predictive for older people, men, or people living in a certain location. Scores predicting behavioral traits such as a person's level of education are even more problematic, because these traits are strongly shaped by the family environment in which a child is raised.

But such polygenic scores, for IQ as well as diseases, are sure to improve as researchers look at genetic markers in larger, more diverse groups of people. "What about when they do become predictive?" an audience member asked during an ASHG discussion, wondering whether that would justify use of the scores. "That equation can change," responded Oxford statistical geneticist Alexander Young. The ethical debate over this brave new expansion of embryo screening is just beginning.

*Clarification, 5 November, 10:05 a.m.: This story has been updated to clarify the context of a quote from Melinda Mills to emphasize that she was referring to screening for desirable traits, not disease risks.