A slippery slope towards designer babies?

SOON TWO biotechnology firms will begin to offer couples undergoing in vitro fertilisation (IVF) the chance to screen embryos before they are implanted in the mother’s womb. This kind of preimplantation genetic diagnosis has, for nearly 30 years, been widely used to test for chromosomal abnormalities or specific genetic disorders that affect only a single gene, such as cystic fibrosis. MyOme, based in Menlo Park, California, and launching in 2019, and Genomic Prediction, based in North Brunswick, New Jersey, have something more revolutionary in mind.

The two companies hope to reconstruct the important parts of an embryo’s genome using just a few cells from a biopsy and genetic sequences of both parents. With the resulting genomic data in hand, the firms can then, in theory, calculate the risk of the embryo developing a wide range of different diseases in later life. Crucially, the ailments in question may be extraordinarily complicated, involving thousands of genetic variants (common mutations) in different parts of the genome.

Two advances underlie this feat. The first is in the method used to produce an accurate picture of the embryo’s genome. Genetic data gleaned by sequencing the tiny quantities of DNA available from the embryo are invariably noisy. The parents’ sequences, however, can be determined precisely. The companies take advantage of the second fact to overcome the difficulties associated with the first, by using powerful computer algorithms to find and stitch together the segments of each parent’s chromosomes that most closely match those of the embryo (and are therefore likely to have been inherited). MyOme’s near-complete genome sequence of the embryo is more than 99% accurate, according to research published by Akash Kumar and Matthew Rabinowitz, two of the firm’s founders, and their colleagues. (Genomic Prediction says its approach is related but not identical to that being used by MyOme.)

The second development has been the greater speed and lower cost of DNA sequencing, which has made an increasing amount of human genetic information available. By analysing the resulting data with machine-learning algorithms, researchers are producing risk profiles for heart disease, diabetes, breast cancer, auto­immune diseases and other common illnesses. The upshot is that, based on a person’s DNA data, a “polygenic risk score” for a disease can be calculated by adding contributions from hundreds or even millions of common genetic variants, each individually conferring only a small risk but which may, in sum, greatly increase the chances of contracting a specific disease.

Unnatural selection
For example, a conventional breast-cancer test may look for harmful mutations in two genes, BRCA1 and BRCA2. Such mutations occur in about 1 in 500 women (in America) but carrying one of them increases the chances of developing the disease by a factor of five or six. A polygenic risk score, calculated by considering some 5,000 genetic variants, can identify the 1.5% of women who are three times more likely than average to develop the disease, but who may not have any mutations in either BRCA1 or BRCA2.

By selecting between different embryos, a couple undergoing IVF (1-2% of all American births today) can optimise the health of their future progeny in a way that those who conceive naturally cannot. When the technique becomes widely available, as it no doubt will, those wealthy enough to do so may opt to undergo IVF even if they are able to conceive naturally. And although both firms say that, for ethical reasons, they will screen embryos only for disease risk, there is no reason why other traits such as height, or most controversially intelligence, might not be selected in the same way.

The technique has limitations. The most accurate polygenic risk scores can be calculated only for those of European ancestry, because fewer people from other populations have been sequenced so far. And although polygenic scores may hint that particular genetic variants increase IQ, they provide almost no information on how they do so. The causal connection between genes and traits remains murky.

All this means that to truly tailor a baby’s genome will require further advances in gene editing, and much more painstaking work. In 2019, however, those with the cash to do so will have an opportunity to give their offspring a greater chance of living a long and healthy life.

This article appears in "The World in 2019", our annual edition that looks at the year ahead, available on newsstands now