Modeling polygenic embryo screening in real-world IVF patients demonstrates limitations on efficacy.

Background: Polygenic embryo screening (PES) has recently become available to in-vitro fertilization (IVF) patients, allowing them to evaluate the genetic risk of each of their embryos for polygenic conditions such as heart attack or diabetes. Initial modeling predicted that transferring the embryo with the lowest genetic risk for one or more diseases would substantially reduce prevalence in the next generation, with relative risk reductions up to 50%. However, these models assumed the availability of a prespecified number of embryos and that the embryo with the most favorable polygenic risk is born once transferred to the uterus. In reality, a large percentage of embryo transfers do not lead to live births, and IVF frequently results in no or only a single live birth. Methods: To quantify the expected risk reduction in the context of IVF, we used two datasets: 6944 ovarian stimulation cycles from 4452 Italian infertility patients and 2138 stimulation cycles of egg donors. In both datasets, we simulated the hypothetical application of PES in these cycles by assigning patients and their embryos randomly drawn polygenic risk scores for a given disease, assuming that embryos have been transferred in increasing order of their risk, and tracing their birth outcomes. We then compared the risk of the embryo born after hypothetical PES to the risk of an embryo born without PES. We either considered only completed cycles or integrated over possible birth outcomes of non-transferred embryos in incomplete cycles. Results: In stimulation cycles in infertility patients in which all embryos have been transferred and at least one child was born, we estimate that PES will result in relative risk reductions of just ≈1-3%. In an intention-to-screen analysis of all completed cycles (regardless of birth outcomes), relative risk reductions are under 0.5%. The risk reductions increase, as expected, with more euploid blastocysts and with younger maternal age. Including incomplete cycles (in which not all embryos have been transferred) increases risk reductions to ≈2-5%, due to the availability of more euploid blastocysts and a higher live birth rate per transfer in these cycles. Pooling all embryos from all cycles of the same patient increases risk reductions to ≈5-10%. Relative risk reductions in egg donor cycles reach ≈20% even with a single stimulation cycle per donor. Conclusions: With the exception of particularly good-prognosis patients or cycles, typical infertility patients would benefit little from PES. In fertile patients, as represented by egg donors, PES is predicted to achieve greater relative risk reductions. However, even though these reductions are still substantially lower than prior estimates that did not account for realistic live birth rates. Ethical, social, and clinical issues associated with offering PES in the general population should be prioritized in future research.