The use of polygenic risk scores for embryo selection has come up repeatedly over the past year so I was excited to attend a talk on the subject at the Fertility 2022 conference. Presented by Professor Daniel Benjamin, professor of UCLA Anderson School of Management and David Geffen School of Medicine, Los Angeles, California, his talk on the 'Challenges with embryo selection using polygenic scores' built on the scientific and communication challenges he and others had presented in a paper published in New England Journal of Medicine earlier in the year.
Professor Benjamin explained that embryo selection using polygenic scores is an emerging industry (see BioNews 1064) for IVF patients in the USA, and is different from current embryo selection options for monogenic disease or chromosome balance. He explained that polygenic scores, used to estimate an individual's lifetime genetic risk of disease, have been around for some time for conditions influenced by multiple genes such as schizophrenia, type 1 diabetes and certain types of heart disease.
To articulate these challenges, Professor Benjamin referred to an example he had seen presented in 2018 by a founder of one of the first companies to offer embryo selection using polygenic scores. He asked the audience to imagine two visually identical and equally viable embryos, which have different scores for a polygenic test designed to predict likelihood of high educational attainment; specifically the likelihood that it would give rise to a person who would complete college.
The polygenic risk score could place individuals into 20 percent segments, where for the lowest quintile, the prevalence of college completion is predicted to be ten percent and in the highest quintile 45-60 percent rate of completion. This would suggest the difference in likelihood of embryos in the lowest and highest groups for completing college is 35-50 percent. Taken at face value, this difference could influence embryo selection for the intended parents and also what information clinics give to patients.
Although this example is compelling, Professor Benjamin highlighted two problems. Firstly, this application of polygenic risk scoring vastly exaggerates the benefits and does not mention the limitations or risks. The paper published in the New England Journal of Medicine spells out these points, which he explained further. Keeping to the example of educational attainment (a measure of the highest level of education completed) where the gain from embryonic selection with polygenic risk scores is stated to be around an extra two years of education completed.
However, if the phenotype for two embryos selected from a set of ten, one selected using polygenic risk scores and the other not is simulated 100,000 times (including a variable to control for environmental factors), would the average gain in years spent in education for the embryo with the high score compared to an embryo which wasn't scored, be two years? The answer is no, as the non-selected embryo is unlikely to be positioned in the lowest quintile. Also, the gene pool for related embryos is not random, as genes may be shared between sibling embryos (genome wide association study (GWAS) data is compiled from unrelated individuals). When both embryos have the same genetic parents, the expected gain of years in education would be 1.55 years with a potential error margin of eight years!
Professor Benjamin also highlighted a wider issue across much of genomic data, in that it originates primarily from European populations. So the predictive power of these polygenic risk scores for individuals of other ancestries would be reduced. Further, using polygenic risk scores to select embryos has the potential to introduce risk as well as 'expected' gains. It was noted that genes tend to have more than one role in health (pleiotropy). We do not know what these relationships are, so selecting for desirable outcomes for one trait could select undesirable outcomes in other traits.
For example, there is a genetic correlation between higher educational attainment and the risk of bipolar disorder. So in the example above, if using a polygenic risk score to select an embryo for educational attainment has an expected gain of 0.53 years in education, this would increase the risk of developing bipolar disorder from one percent to 1.6 percent. Lastly polygenic risk scores are often advertised as reducing relative risk and not absolute risk. For example, a reduction in relative risk for type 1 diabetes for an embryo selected using a polygenic risk score could be given as 35 percent, which translates to a 0.12 percent reduction in absolute risk.
Professor Benjamin called for urgent society-wide conversation on this matter. As GWAS become more powerful, the insights we could gain from them are likely to grow. There is likely to be a desire to design ways to include this insight into fertility treatments. The problematic ethical issues were outside the scope of the talk; however, Professor Benjamin acknowledged these and believes the use of embryonic selection using polygenic risk scores could have profound social consequences over several generations.
Professor Benjamin gave a clear compact talk to explain the fundamental pitfalls of applying the principles of polygenic risk modelling to embryo selection. His call for an urgent wide conversation has since started with professional societies in the UK, Europe and the USA advising against clinical use (see BioNews 1137).
It is clear that fertility patients and providers can easily misunderstand what polygenic risk scores could offer them. This complicated scientific issue requires research on how to accurately convey risks and benefits together with regulation by governments and medical societies to curb misunderstanding, misuse and provide accurate marketing.
Leave a Reply
You must be logged in to post a comment.