Skip to main content
Premium Trial:

Request an Annual Quote

ACMG Asserts Evidence Insufficient for Clinical Utility of Polygenic Risk Scores for Embryo Testing


NEW YORK – The clinical utility of polygenic risk scores (PRS) remains unproven and PRS should not be offered as a clinical service for preimplantation genetic testing at this time, according to a working group of the American College of Medical Genetics and Genomics (ACMG).

In a Points to Consider statement in Genetics in Medicine published on Friday, the Social, Ethical and Legal Issues Committee of the ACMG drew attention to the impact of in vitro fertilization (IVF) itself on the clinical utility of preimplantation genetic testing for polygenic disorders (PGT-P), questions concerning the clinical validity of PRS for embryo selection, ethical issues surrounding scenarios in which PGT-P could be requested, and challenges in providing informed consent and communicating disease risk information.

Although couples seeking a pregnancy through IVF have the option to perform PGT-P at that time, the resulting PRS may carry little practical information and the risks of performing IVF solely for the purpose of PGT-P may outweigh the potential benefits.

Jason Vassy, director of the Genomes2Veterans Research Program at the VA Boston Healthcare System and Harvard Medical School, concurred with the ACMG statement, noting in an email that "IVF should not be performed solely for the purpose of PGT-P."

IVF is a lengthy and invasive process, the committee wrote, and should be weighed against the potential risks of PGT-P, which include preeclampsia, abnormal placentation, cesarean section, prematurity, low birth weight, and miscarriage.

Additionally, PGT-P often necessitates additional rounds of IVF, leading to increased health risks and a higher financial burden.

"IVF can be a bewildering, heartbreaking, and beautiful process for patients," Vassy, himself the father of two children conceived through IVF, said, "and communication about the risks, benefits, and uncertainties is critical at every step."

One factor complicating the analytic validity of PGT-P, the committee wrote, is the need to amplify DNA from the limited number of cells that can be obtained through an embryo biopsy. The low-coverage sequence data typical of an embryo biopsy can be supplemented with parental genomic sequences to essentially reconstruct the embryo's full genome. However, the ACMG statement points out that very few published studies support the validity of this approach, and only one published study to date has shown a strong correlation between embryonic and postnatal PRS. 

Furthermore, a single biopsy may not accurately represent the genetic makeup of the embryo's cell mass and any PRS derived from embryos is likely to be affected by intrauterine environmental factors.

"The increased incidence of imprinting disorders among children born via IVF provides evidence of the epigenetic impact of the culture environment in the assisted reproductive technology process," the committee wrote.

Additionally, PRS derived from genome-wide association studies (GWAS) perform best for adult-onset disorders, typically involving decades of exposure to environmental variables.

As an example, the committee singled out coronary artery disease (CAD) for having one of the most predictive clinical PRS. Even among individuals in the top PRS decile for CAD, meaning those with scores most predictive of being diagnosed with the disorder, over half are never diagnosed with CAD. Conversely, 16 percent of people in the bottom decile still develop CAD.

The ACMG committee noted that in a public health setting, PRS does have the potential to identify groups who may be at risk for common disorders and who may meaningfully benefit from early intervention. However, a PRS cannot definitively tell a person whether or not they'll be diagnosed with a disorder, as they are not perfectly heritable and, in many cases, may play less of a role in disease occurrence than environmental and lifestyle factors.

Seamus Harrison, VP of global medical affairs and clinical development at Genomics Plc, which is developing a PRS for CAD, said in an email that he agrees with the ACMG statement, in that PRS and other genetic variants are not deterministic.

"Susceptibility to common diseases, such as coronary artery disease and type 2 diabetes is complicated," Harrison said.

As an example, Harrison mentioned that adherence to good cardiovascular health is effective at reducing risk of disease even in those with high PRS.

"For complex diseases, it seems there isn’t a single golden bullet for prevention," he said. "This is why PRS will become a powerful tool to improve health by allowing healthcare providers to direct valuable preventative resources to those most likely to benefit at the right time in their lives."

Harrison also noted that the ACMG statement often contrasts PRS to Mendelian conditions, which are typically dominated by the effect of a single gene variant and follow more straightforward inheritance patterns.

"As the field evolves," he said, "I hope that we'll see a more joined up vision, where both types of genetic information can be considered together, along with other nongenetic risk factors, in integrated risk tools that can predict risk over an individual's lifetime. With that said, this point doesn't change the overall conclusions of the statement, which I agree with."

Another significant challenge to proving the clinical utility of PGT-P that the ACMG statement highlights is that of conducting prospective clinical trials demonstrating that a child conceived using PGT-P is better off than a child would have been without having undergone that type of selection, as it may take decades to see whether or not a given condition develops.

The strongest potential means of demonstrating clinical utility that we are left with, the committee wrote, are studies comparing the predictive power of PRS between siblings in the same family. 

Related to the challenge of quantifying the predictive power of a PRS over a person's lifetime is that of pleiotropy, which describes that complex interplay of genetic interactions. Applied to PGT-P, the concern is that selecting against one undesirable condition may inadvertently select for another.

Harvard's Vassy stressed that it is currently feasible to conduct prospective trials to demonstrate the near-term safety of PGT-P compared with current standard of care, such as by assessing its impact on success of IVF –– as evidenced by live births –– and developmental outcomes in the first few years of life.

"The time is right for those trials to be conducted," he said. "Demonstrating benefit through prospective trials will be much more difficult and might first require an 'early win' with a PRS associated with disease outcomes early in the life course."

Otherwise, and absent a large, decades-long study, he said, the field might have to settle for trial evidence for short-term safety and theoretical evidence of potential long-term benefits.

Harrison, of Genomics Plc, agreed, saying that new technologies to predict an embryo's health outcomes remain many years in the future.

"Thankfully," he said, "for the more established use cases, which use PRS to predict disease onset in adults, there are now very large population-based studies that allow us to demonstrate, with high certainty, the impact of PRS on the health of the population."

Clinical utility in these cases, he said, will consist of demonstrating that the information derived from PRS leads to changes in how healthcare is delivered, such as by influencing clinical decisions around prevention and/or providing access to early or intensive screening measures, which have been proven clinically beneficial.

Harrison said that with respect to embryo selection, "it isn't obvious that this is an area where PRS alone will have the greatest clinical impact."

Nonetheless, a few companies have begun developing PRS tests for embryo selection.

Personalized genomics company MyOme, for example, develops PRS tests for conditions such as coronary artery disease and breast cancer, and is working toward a PGT-P offering.

Akash Kumar, MyOme's chief medical and scientific officer, said that his firm believes that PRS will transform healthcare and that its implementation must be done responsibly.

"Our approach to PGT-P is no different and has been in a stepwise fashion," he said by email, "starting with validating the accuracy of our measurements." 

In 2022, the company published a study in Nature Medicine demonstrating the feasibility of whole-genome reconstruction of an embryo for PRS application. In that study, MyOme researchers reported that combining rare variants with PRS meaningfully increased the risk difference across sibling embryos for important common diseases.

"Since that publication," Kumar said, "our team has focused on developing and validating cross-ancestry polygenic risk scores across multiple conditions … including breast cancer and coronary artery. We are excited to be at the forefront of developing clinical applications to support families of all ethnicities."

The applicability of PRS across populations with diverse ancestries has been a problem for some time, as most GWAS have been performed in populations of predominantly European ancestry. Many groups, however, are working to change this through initiatives such as the All of Us program, run by the US National Institutes of Health.

Last year, MyOme partnered with Natera to show that its cross-ancestry, integrated PRS improved breast cancer prediction over standard models.

Finally, the ACMG statement listed communicating polygenic test information as a key challenge in bringing PGT-P into the clinic.

"Informed consent is the foundation for ethically grounded initiation of any medical test or procedure," the committee wrote. In the context of an unproven diagnostic method, such consent is "problematic at best, if not impossible," until its clinical utility has been firmly established. Additionally, the committee commented that information regarding PGT-P is "extraordinarily complex" and care would need to be taken to effectively communicate the probabilistic nature of PRS test results, along with how they may be influenced by environmental factors.

Vassy, however, contested this to some extent, saying that the ACMG statement overstated the "impossibility" of patient informed consent for PGT-P.

"The risks and benefits of a medical test can be communicated, as can the uncertainty about those risks and benefits," he said. "In fact, a lay summary of this position statement would go a long way in helping clinicians communicate those very points."

The ACMG statement decisively concluded that PGT-P "should not be offered as a clinical service," and reaffirmed a recent statement by the ACMG board of directors on a related topic that prenatal testing for polygenic disorders "is not yet appropriate for clinical use and should not be offered as direct-to-consumer testing."

Vassy agreed, stating that the evidence base for PGT-P is "very thin," although technological and market forces are propelling the science forward.

"This position statement adds the voice of an important professional society, and calls for that of others, to fill this evidence gap with thoughtful expert opinion," Vassy said.

Similarly, Harrison said that evidence of safety and efficacy, as well as acceptability and feasibility for PRS usage in adults, is emerging rapidly.

"But for PGT-P," he said, "this is some way off."

While the current ACMG statement did not go into the ethical concerns and considerations surrounding PGT-P, Susan Klugman, president of the ACMG, said via email that those would be the focus of an upcoming paper.

"A second accompanying paper is planned discussing the ethical issues in great detail, including that balance between harm and benefit," Klugman said. "All four of the main tenets of bioethics including autonomy, beneficence, nonmaleficence and justice will be discussed in that paper."