Capacities and Limitations of Using Polygenic Risk Scores for Reproductive Decision Making

In their article “Implementing Expanded Prenatal Genetic Testing: Should Parents Have Access to Any and All Fetal Genetic Information?” Bayefsky and Berkman briefly mention that: “[s]ome are hopeful that polygenic risk scores [(PRS)] … will provide more accurate assessments of genetic risk” when conducting expanded prenatal genetic testing. Developed primarily for purposes of preventive medicine (Torkamani, Wineinger, and Topol 2018), PRS provide a quantitative metric of an individual’s predisposition to developing a polygenic condition, based on the cumulative effect of hundreds or thousands of gene variants, as derived from genome wide association studies (GWAS) (Khera et al. 2018). PRS have been studied mostly with respect to polygenic conditions (such as diabetes, different types of cancer, cardiovascular diseases, and psychiatric conditions) in pediatric and adult populations. Some commercial laboratories now offer PRS for preimplantation screening of IVF embryos to inform prospective parents’ decisions over which embryo(s) to transfer for implantation (Lázaro-Muñoz et al. 2021). If, indeed, PRS were to become part of expanded prenatal genetic testing, it is critical for stakeholders, such as clinicians and prospective parents, to understand the benefits, risks, and limitations of PRS for this use before deciding whether and how to use this information. Understanding all these aspects of PRS promotes reproductive autonomy and supports decision making that is based on prospective parents’ values, and not misinformation or misunderstanding of PRS.

The purpose of generating PRS of a fetus would be the ability to act on information about its risk of various polygenic conditions. Actions that prospective parents could take in response to such information include logistic and psychological preparation for the birth or termination of the pregnancy. Basing such actions on probabilities of polygenic conditions that tend to have lower penetrance compared to monogenic conditions may be considered more ethically contentious than doing so based upon results from prenatal genetic testing, such as chromosomal microarray analysis (Shkedi-Rafid, Horton, and Lucassen 2021). However, prospective parents may find PRS of significant importance for polygenic conditions that may have pediatric onset, such as type I diabetes or various psychiatric/neurodevelopmental conditions. As Bayefsky and Berkman recommend, it is essential for research to explore the specific conditions about which prospective parents want to have more information to facilitate individual reproductive decisions (Bayefsky and Berkman 2021).

The risks of PRS in the prenatal context may be physical or psychological. Potential physical risks include those associated with invasive prenatal tests, such as chorionic villus sampling or amniocentesis, including amniotic fluid leakage, slight bleeding, and a small (e.g., up to 0.30%) risk of miscarriage (Salomon et al. 2019). There are no serious physical risks if generation of PRS becomes possible with non-invasive prenatal genetic testing that evaluates the cell-free fetal DNA in maternal blood. Potential psychological risks of generating PRS with respect to a fetus include anxiety or emotional distress, especially due to the probabilistic nature of the information and uncertainty over condition severity and treatability.

In deciding whether the generation of PRS of a fetus will help inform their reproductive decisions, it is crucial for prospective parents to know its limitations. First, there are multiple issues concerning the metric’s accuracy. PRS provide probabilistic risk assessments, which typically have limited sensitivity (i.e., the proportion of those who actually develop a condition that are designated as high risk) and specificity (i.e., the proportion of those who do not develop a condition that are designated as low risk). Type 1 diabetes (T1D) provides an example of the current upper limit of predictive capability of PRS technology. Individuals with a PRS in the top 1% for T1D have a 30-fold increased chance of developing the condition compared to the general population; however, given that the population prevalence of T1D is approximately 0.3%, these “high-risk” individuals still have only approximately a 9% chance of developing T1D in their lifetimes (Sharp et al. 2019). In other words, such offspring would still have more than a 90% chance of not developing T1D. Yet, if the threshold for reporting a PRS for T1D as “high risk” is set to the top 1%, at which point PRS sensitivity is approximately 30%, then a large majority (approximately 70%) of individuals who will develop T1D would be missed (Sharp et al. 2019). Setting such reporting thresholds involves trade-offs; if it’s set to avoid many false-positives, then it also increases the potential to miss true-positives. Given the relatively simple and well-understood genetic architecture of T1D compared to other polygenic conditions, it is unlikely that PRS will deliver greater predictive accuracy for many common diseases in the foreseeable future.

Additionally, the predictive accuracy of PRS can vary depending on sex, age, socio-economic status (Mostafavi et al. 2020), or even due to chance alone (Ding et al. 2021). Furthermore, not all ethnicities have been equally studied with GWAS. PRS are most accurate for those with European ancestry – the population that comprises the majority of GWAS data (Martin et al. 2017). Finally, at this time, PRS do not capture rare variations (both inherited and de novo) that can contribute substantially to the probability of developing some conditions, like autism, that may be of interest to prospective parents (Weiner et al. 2017).

Another limitation is that many of the polygenic conditions for which PRS are currently available are adult-onset, such as heart disease and various types of cancer. This raises the issue of whether obtaining this information for a fetus would be accurate given that it is unknown whether current PRS will be accurate several decades from now when those conditions would typically present. It is worth noting that prenatally testing for adult-onset conditions may also be ethically contentious due to conflicting autonomies (i.e., those of the parent(s) vs. those of the future child), the potential for adverse effects on the parent-child relationship (e.g., treating a child as a “patient-in-waiting”), and the risk of discrimination (e.g., with respect to insurance coverage). Though the National Society of Genetic Counselors emphasizes the necessity of counseling prospective parents on the possible consequences of such prenatal testing, they ultimately recommend deferring to parents’ decisions (Hercher et al. 2016).

A third and final limitation of PRS is that many of the polygenic conditions subject to this metric have highly variable presentation in terms of chronicity and severity. For example, an individual diagnosed with depression may have a single brief episode with mild to moderate symptoms that spontaneously resolves; in other cases, the same diagnosis could lead to severe and treatment-resistant illness for years, with potentially fatal consequences (i.e., suicide). PRS are currently not and may never be able to provide more certainty about these variables at an individual level.

Given these uncertainties and the fact that PRS could be generated for multiple conditions, if PRS is ever used in conjunction with expanded prenatal genetic testing, the accompaniment of genetic counseling will be essential for assisting prospective parents in deciding whether to use this information and how to interpret and act upon it. Counseling is also important for helping prospective parents explore their values and goals with respect to parenting, as well as definitions of “health.” Such counseling should note that development of polygenic conditions does not necessarily preclude one from having a good quality of life.

Due to the aforementioned issues of accuracy that warrant further evaluation, we recommend caution in considering the use of PRS for expanded prenatal genetic testing. However, if it were to be used, we underscore the importance of counseling that specifies the technique’s benefits, risks, and limitations, as well as delves into parental goals, values, and conceptions of “health.”