Genetics: Predicting the risk of some common diseases in preimplantation fertilized embryos

The genome sequence of fertilized embryos during in vitro fertilization (IVF) — and the subsequent genetic risk of developing 12 common medical conditions — can be predicted before implantation with an accuracy of up to 99%, according to a study published in Nature Medicine. These initial findings, based on predictions for 110 fertilized, preimplanted human embryos which were compared to risk predictions based on the genomes of 10 children who were subsequently born, may have implications for the use of genome-based, preimplantation genetic testing (PGT) of fertilized embryos. However, the authors note that there are a number of ethical and practical considerations that will need to be addressed and incorporated into genetic counseling.

PGT of fertilized embryos during IVF treatment is currently used to prevent rare diseases caused by single-gene mutations, known as ‘Mendelian disorders’. However, assessment of a wider range of more common conditions — such as heart disease or cancer — is not yet an option.

Akash Kumar and colleagues sequenced the genomes of ten couples undergoing IVF treatment. Those data were then used to generate polygenic risk scores (PRSs) — a genetic tool used to predict risk of disease — for 110 fertilized embryos that had yet to be implanted, which allowed the authors to predict each embryo’s likelihood of developing 12 common medical conditions, such as certain cancers or cardio-metabolic diseases. Those predictions were then compared with risk predictions generated from the genomes of 10 children who were subsequently born. Overall, PRS predictions were between 99% and 99.4% accurate when generated with data from embryonic tissue at day 5 after fertilization, or between 97.2% and 99.1% accurate with data from tissue at day 3. Combining PRSs for each embryo with knowledge of rare variants in genes carried by the parents — for example, the BRCA1 variant — improved differences in risk prediction, a 15 times increase between ‘sibling embryos’.

The authors highlight several limitations to their study. For example, while models are validated against population data, patients need to be counseled that this validation is imperfect given generational and population differences. Also, this method considers only inherited genetic variation — as opposed to new variants or mutations that may emerge after conception. PRS predictions may be less effective in non-European populations, as research cohorts in this field, the authors state, have historically involved people of European ancestry. Furthermore, considering PRSs when making prenatal decisions presents ethical and practical questions that need to be addressed and incorporated into genetic counseling.

In an associated News & Views, bioethicists Josephine Johnston and Lucas Matthews caution that “PRS-informed PGT may further de-emphasize environmental and social determinants of common diseases,” essentially putting the “responsibility for managing disease risk” into the hands of the individual, rather than focusing attention on “structural solutions.”

Furthermore, Norbert Gleicher and colleagues discuss the limitations of prenatal genetic testing in an associated Comment, expressing that “although polygenic risk scores have become increasingly well-powered, they remain highly experimental”. Going forward, they emphasize how important it is that “genetic tests in reproductive medicine should be based on rigorous science, be transparent about efficacy, and be properly regulated.”