O-056 Evaluating the utility of screening human IVF embryos with polygenic risk scores for complex diseases
Abstract Study question It is now feasible to screen human IVF embryos with “polygenic risk scores” for predicting complex disease risk. What is the expected risk reduction? Summary answer Under some conditions, prioritizing embryos based on polygenic risk scores can lead to substantial disease risk reductions. However, only excluding high-risk embryos is less effective. What is known already Recent genetic studies have identified numerous mutations associated with complex diseases, leading to the development of accurate polygenic risk scores (PRSs) for disease risk prediction. Given that genomes of human IVF embryos can now be sequenced with relative ease, it has become technically feasible to use PRSs for prioritization of embryos for transfer. Clearly, such use is associated with ethical and social concerns, from inequality to eugenics. Nevertheless, polygenic embryo screening is already offered to consumers, with little research so far on expected outcomes. Our previous evaluation of screening IVF embryos for polygenic traits showed little current utility. Study design, size, duration This is a theoretical/computational study based on statistical genetics theory and simulations. Participants/materials, setting, methods We used the liability threshold model to estimate the disease risk given the PRS. We considered screening for a single disease (with known prevalence and PRS accuracy), and assumed that n viable embryos are available. We calculated the risk of the child given these parameters and the prioritization strategy, either when parents are random or when their disease status is known. We also used simulations based on genomic data from a schizophrenia case-control study. Main results and the role of chance We modeled the disease risk of a hypothetical future child when the PRSs of embryos are used for prioritization, relative to random selection. When selecting an embryo at random among those who do not have an extremely high risk (typically, top 2% of the PRS distribution), the relative risk reduction (RRR) is limited, and is under 10% for currently realistic scenarios. In contrast, selecting the lowest risk embryo for implantation results in substantial RRRs of ∼20-50% already with n = 5 embryos and realistic disease parameters. For example, the RRR for schizophrenia is > 40% with current PRSs, a result we validated with simulated genomes of parents and children based on genotypes from a schizophrenia study. When one of the parents is known to be affected, selecting the lowest risk embryo out of n = 5 may restore the risk of the future child to nearly normal levels. Limitations, reasons for caution Our analytical modeling is based on several simplifying assumptions regarding the dependence of the risk on the PRS and the accuracy of the PRS. Further, the estimated risk reductions depend on the availability of n = 5 embryos that could lead to a live birth. Wider implications of the findings Under some conditions, prioritizing embryos for transfer based on polygenic risk scores could lead to substantial disease risk reductions. However, predicted outcomes vary considerably with prioritization strategies and disease and PRS parameters. The emerging ethical and social concerns and the challenges in communicating these results need to be urgently discussed. Trial registration number Not applicable