Screening IVF embryos with polygenic risk scores: risk reduction estimates, population impact, and attitudes of patients and clinicians

Abstract Study question IVF embryos can now be screened for complex diseases. What is the expected risk reduction? What are the attitudes of patients and clinicians? Summary answer Substantial risk reductions can be achieved, but only in the best-case scenario. The views of patients and clinicians regarding the technology are not fully aligned. What is known already It recently became possible to screen IVF embryos for a variety of complex diseases using polygenic risk scores. Prioritization of embryos for transfer based on these scores may lead to reduced disease risk in offspring. Preliminary risk reduction modeling suggested that substantial risk reductions are possible, particularly when screening for a single disease. However, multiple factors remain unexplored, such as the effect of implantation failure and assortative mating on risk estimates and the impact of embryo prioritization on population allele frequencies. The attitudes of IVF patients and clinicians towards polygenic embryo screening are also unknown. Study design, size, duration We estimated the expected risk reduction under various settings using statistical modeling, based on the liability threshold model and other quantitative and population genetic models. To learn about attitudes of patients and clinicians, we performed in-depth, semi-structured interviews with 26 USA patients and 27 USA clinicians over a period of about one year. Participants/materials, setting, methods We used R to numerically estimate risk reductions and implemented a web app in Shiny. To recruit patients and clinicians for the interviews, we used flyers (for patients), emails on publicly available lists (for clinicians), snowball sampling, and convenience sampling. Interviews were conducted over Zoom for one hour each by a single interviewer (D.B.). The study was approved by Baylor College of Medicine's IRB committee. We applied deductive and inductive thematic analysis using a team-based approach. Main results and the role of chance Statistical genetic modeling suggested that with five viable embryos, risk reductions of up to 50% can be achieved when screening for a single common disease. We developed a risk reduction estimator web app, allowing users to examine the effect of the number of embryos, the accuracy of the score, the selection strategy, and the score percentile or disease status of the parents. Assortative mating and modeling of implantation failures are expected to only slightly decrease risk reductions. Risk allele frequencies are expected to change very little across generations even if embryo selection becomes widespread. In interviews, patients accepted screening for health conditions, particularly severe diseases. Many were concerned about costs, and some were interested in screening for informative purposes. Most patients did not oppose screening for traits. In contrast, about half of clinicians were unwilling to offer screening due to concerns over utility, implementation, harm to patients, and lack of scientific validity, although all were willing to reconsider it in the future for specific conditions. Many clinicians considered screening for traits a non-medical procedure. About half agreed that screening should be regulated. Limitations, reasons for caution The risk reduction estimates may depend on the assumptions of the statistical models. As with any qualitative research, interview findings may not be generalizable. Furthermore, the design of the semi-structured interview guide may have influenced participants’ expressed attitudes. Wider implications of the findings In the best-case scenario, prioritizing IVF embryos based on polygenic scores may achieve substantial risk reductions. Several initial concerns seem unfounded. IVF patients are broadly interested in the technology but may be unaware of its many limitations. Clinicians have major concerns regarding clinical implementation and harm to patients. Trial registration number not applicable