Still doubts over embryo selection based on PGT for polygenic conditions

Published 19 April 2022

Advances in genomic technologies continue to expand the possibilities of PGT. In a preclinical research study recently published in Nature Medicine, scientists from Silicon Valley-based genetic testing companies have examined the potential of using PGT to predict complex polygenically inherited conditions in human embryos. While the authors present interesting data, Mina Popovic and Susana Chuva de Sousa Lopes from ESHRE’s SIG Stem Cells here report that the study raises profound medical, technical and ethical concerns.

PGT has principally focused on testing embryos for aneuploidies (PGT-A), structural rearrangements (PGT-SR) and monogenic disorders (PGT-M). Now, the ability to generate genome-wide genotypes from trophectoderm biopsies has expanded the scope of embryo testing, from which has recently emerged the development of ‘PGT-P’, a technique which classifies embryos based on their genetic susceptibility to a variety of complex diseases and traits. Such conditions result from the combined effects of multiple genes, often in concert with lifestyle and environmental factors. Until recently, the application of PGT to these multifactorial phenotypes was not possible, as the genetic contribution to such conditions was difficult to ascertain.

Coupled to technical developments in embryo genomic sequencing, advances in complex trait genetics and computational genomics have now enabled the introduction of PGT-P for embryo selection. Nevertheless, the predictive value, clinical utility and ethical implications of this test remain highly controversial. Given the lack of compelling evidence for its clinical application, several scientific societies, including ESHRE and the European Society of Human Genetics, have recently released statements against the use of PGT-P in clinical practice.(1,2)

Despite these concerns, genetic testing companies, including Genomic Prediction (, Orchid Health ( and MyOme (, appear to be showing interest in PGT-P to screen embryos for their genetic predisposition to cancers, diabetes, heart conditions, schizophrenia, Alzheimer’s, as well as other adult-onset diseases. Remarkably, the selection of embryos based on non-medical conditions - such as cognitive ability, education and household income - have also been proposed, with a tenuous health benefit already assigned to some of these traits.

The technology behind PGT-P draws on polygenic risk scores, derived from large-scale genome-wide association studies.(3) These studies employ high-throughput genomic technologies to scan thousands of genomes obtained from different individuals to uncover genetic variants, predominantly single nucleotide polymorphisms (SNPs), that are significantly associated with a particular complex trait. Such data can be used to generate polygenic risk scores which summarise a set of genetic variants in an individual to evaluate their risk of developing a certain condition. Polygenic risk scores have inherently gained intense interest for preventative medicine, to improve diagnoses and select optimal treatment for a variety of common diseases. Applying this concept in the context of preimplantation genetics has thus enabled PGT-P. Nevertheless, very few studies have examined the utility of polygenic risk scores for embryo selection and ultimately the extrapolation of such data to an IVF embryo cohort remains problematic.(4)

In their recent study, Kumar and colleagues from MyOme combined molecular and statistical techniques to infer the whole genome sequence of 110 embryos, using parental genome sequencing and embryo genotyping.(5) They suggest that this method, known as whole genome reconstruction, can be used to more accurately predict susceptibility to 12 common polygenic conditions in human embryos, including cancers, cardiometabolic and autoimmune diseases. The study used samples from couples who had previously had PGT-A or PGT-M, while DNA was also available from ten children who had prior PGT. The authors validated their technique by comparing the reconstructed genome and polygenic predictions for the embryos to those of the corresponding children. Genome-wide prediction accuracy ranged from 98-99%.

The findings suggest that whole genome reconstruction improves the accuracy of risk predictions by comprehensively profiling embryo genomes, whilst overcoming known technical limitations associated with the genetic analysis of small amounts of embryonic DNA. The paper touches on challenges regarding the statistical validity of polygenic risk scores for evaluating embryos in this setting, and discusses some ethical aspects of the procedure. Nevertheless, many questions remain unanswered.

Alongside the paper, Nature Medicine published two commentaries highlighting the technical, clinical and ethical dilemmas associated with PGT-P.(6,7) These questions are complex and wide-ranging, and while it is not possible to give them full justice in this report, they certainly deserve serious attention.(8)

Despite the growing interest in polygenic risk scores in biomedical research, limited guidelines for performing such analyses have ultimately led to inconsistent data and misinterpretations of results. Accordingly, several methodological limitations in prediction accuracy hinder the clinical utility of PGT-P. Ultimately, polygenic risk scores only capture parts of the relevant genetic component of a condition. Indeed, predicting the risk of adult-onset diseases cannot account for changes in environmental or lifestyle factors which might occur over time. Moreover, predictions only provide a relative risk compared to the specific study population they were generated in. As justly mentioned by Kumar and colleagues, polygenic risk scores exhibit limited predictive accuracy when extended to different populations, as they have been largely developed using genome-wide association studies of European ancestry.

Studies have further shown that the potential gain of quantitative traits, such as height or cognitive ability, remains relatively small.(9) In addition, a substantial number of embryos per cycle (>10) is required to increase the prediction accuracy of PGT-P. This is largely unfeasible for many couples undergoing IVF. Moreover, as PGT-P screens for multiple polygenic conditions simultaneously, it vastly complicates embryo selection.

Currently, the complex relationships between genetic variants and traits are not well understood, and it is uncertain whether a lower risk for one condition may in fact increase susceptibility to others. Indeed, PGT-P creates enormous challenges for clinical management and patient counselling. Prospective parents will face difficult decisions as they try to balance risks for multiple conditions. Complicated decision-making surrounding embryo selection may dimmish procreative autonomy, lead to additional cycles of ovarian stimulation, and ultimately elicit unwarranted disposal of viable embryos.

Adequate ethical oversight and societal discussion also remain imperative. Accessibility to the technology, the opportunity for using PGT-P for non-medical individual traits, discrimination and stigmatisation of certain conditions, and drawing attention away from individual responsibilities for managing disease risk, all present legitimate concerns. Yet, research on the perceptions and attitudes towards PGT-P, including those of its consumers, is sorely lacking. In this increasingly complex industry, clinical decision-making must rely on preclinical and clinical studies, controlled trials, as well as long‐term follow‐up. Such studies inherently demand time, and remain challenging to perform, particularly in the context of assisted reproduction. Nevertheless, balancing innovation with robust evidence surrounding the effectiveness, safety and ethical implications of newly emerging technologies is becoming more relevant than ever.

1. See
2. Forzano F, Antonova O, Clarke A, et al. The use of polygenic risk scores in pre-implantation genetic testing: an unproven, unethical practice. Eur J Hum Genet 2021;
3. Tam V, Patel N, Turcotte M, et al. Benefits and limitations of genome-wide association studies. Nat Rev Genet 2019; 20: 467–84.
4. Turley P, Meyer MN, Wang N, et al. Problems with using polygenic scores to select embryos. N Eng J Med 2021; 385: 78–86.
5. Kumar A, Im K, Banjevic M, et al. Whole-genome risk prediction of common diseases in human preimplantation embryos. Nat Med 2022; 28: 513–516.
6. Gleicher N, Albertini DF, Patrizio P, Orvieto R, Adashi EY. The uncertain science of preimplantation and prenatal genetic testing. Nat Med 2022; 28: 442–444.
7. Johnston J, Matthews LJ. Polygenic embryo testing: understated ethics, unclear utility. Nat Med 2022; 28(3): 446–448.
8. Siermann M, Tšuiko O, Vermeesch JR, et al. A review of normative documents on preimplantation genetic testing: Recommendations for PGT-P. Genet Med 2022: S1098-3600(22)00678-5.
9. Karavani E, Zuk O, Zeevi D, et al. Screening human embryos for polygenic traits has limited utility. Cell 2019; 179: 1424-1435.e8.

Get notified of new articles with our ESHRE newsletter.

Sign up and never miss an update