Abstract: Extremely large genetic studies conducted in the past few years have identified numerous mutations associated with complex traits and diseases, and have been translated into increasingly accurate prediction models. Popular predictors are based on "polygenic risk scores" (PRSs), which are roughly counts of the number of risk (or trait) increasing mutations carried by an individual. One clinical application of PRSs, perhaps not originally envisioned by many researchers, is the selection of human IVF embryos for implantation based on their PRS for a complex disease or trait. Clearly, prioritizing human embryos based on genetic scores is fraught with ethical and social concerns, from stigmatization and inequality to eugenics. Nevertheless, embryo screening is already offered to consumers, with no oversight, and with barely any research to support its clinical utility. In my talk, I will describe statistical models and simulations that predict the utility of embryo screening. We show that when embryos are selected for implantation based on their predicted traits (e.g., height or intelligence), the expected increase in trait is relatively small and subject to wide uncertainty. In contrast, selecting embryos for reduced predicted disease risk may achieve extremely large relative risk reductions for complex diseases such as schizophrenia or diabetes.
Predicting the utility of screening human IVF embryos for complex traits and diseases