How do you counsel patients who ask about PGT-P testing?

PGT-P, or polygenic preimplantation genetic testing, is an extension of preimplantation genetic testing that is very interesting and exciting, but I’m not sure we’re necessarily ready for prime time yet. It’s important for us to know about, understand, and be able to address with patients because it’s already here, it’s commercially available, and it’s expanding every single day. Our guidelines and data are not going to catch up with this technology quickly. That’s true of many things in our field today: these tools will be in the hands of patients and providers long before we fully understand all the implications.

Right now a few companies are offering polygenic risk scores for conditions like thyroid disease, diabetes, and heart disease. At the same time, the science allows us to do things that raise greater concern. People start to mention the word eugenics—eye color, height, intelligence. We don’t yet know the implications of extreme genetic selection. If you do extreme selection for intelligence—something that would not occur in nature except at a very low rate—these genes have multiple functions. While we can sequence and study them, we don’t know how they interact. I often worry that extreme selection for intelligence could produce someone with extraordinary cognitive ability but major social deficits, similar to the stereotype of Rain Man.

Even though this isn’t genetic modification with CRISPR where we are cutting and editing the genome, we are selecting from what already exists. We don’t understand the clinical implications of extreme selection. So, PGT-P is very interesting, but officially it remains experimental. I advise my patients to use a company that is actively participating in and publishing research in this area. There are companies doing this work; full disclosure: I am a medical advisor for Genomic Prediction. Nathan Treff is the founder and lead scientist and is at the forefront of this type of testing. When he first sent me the company brochure, I told him he needed to remove certain examples—like including short stature and height—because I am sensitive about that. I’m 4 foot 11, and I know that if my mother had used PGT-P, she might have chosen a taller, quieter embryo instead of a short, loud one, and I might not be here. That anecdote highlights how complex and personal these decisions are.

PGT-P should ideally be done in an IRB-type setting. Patients need thorough counseling and must understand that it is very experimental. It’s also unrealistic for many patients because we are still limited by current reproductive technology. We cannot create eggs, so people can usually produce only a dozen or two eggs at most. Some people occasionally produce 30 to 50 eggs, but that’s not generally recommended. Many eggs will be chromosomally abnormal—at least half if the person is in their 30s. For patients in their 40s, only 10 to 20 percent of eggs may be chromosomally normal. Once you select for aneuploidy or euploidy, you often have only a small number of embryos left to screen with polygenic scoring, so you may not have much choice.

We’re not yet at the point of being able to manufacture eggs and create thousands of embryos for extreme selection, though we are edging toward more capability. You shouldn’t ignore the topic; acknowledge that it exists, listen to patients’ interest, and try to understand where it’s coming from. Some patients have traumatic family histories—cardiac disease, for example—so it’s reasonable to discuss options and, if appropriate, connect them with a reputable PGT-P company that provides extensive genetic counseling. Patients need realistic expectations: for example, a patient with an AMH of 0.3 is unlikely to benefit from PGT-P—she may be fortunate to have one good embryo. In short, patients deserve an informed conversation and access to information about these technologies.

PGT-P stands for polygenic preimplantation genetic testing for polygenic disorders. It is a very new technology. With PGT-P, the idea is to risk-stratify an embryo for one or several diseases or disorders that can manifest later in life. For example, PGT-P can estimate an embryo’s future risk of developing hypertension. It does not provide a simple yes-or-no answer. Like all tests, it is imperfect and can produce false positives and false negatives. Instead of a binary result, PGT-P gives a risk-stratified score—for example, an embryo might be estimated to have a 40–60% chance of developing hypertension later in life.

That approach raises challenges. When you consider many possible medical conditions, it is difficult to decide how to weigh one risk against another. People tend to think in black-and-white terms—yes or no—but risk stratification resists that framing. It is tempting to treat anything above 50% as a “yes” and anything below 50% as a “no,” but that oversimplifies the trade-offs and raises the question of which conditions deserve more weight.

There are also significant ethical concerns. Evaluating medical disorders like hypertension or diabetes is one thing and has clear potential value. But once you start including physical characteristics—hair color, eye color, height, weight—you open the door to choices that may not be in the embryo’s best health interest and that could reflect or amplify social biases rooted in racism, antisemitism, or other forms of discrimination. What is considered a desirable physical trait today may not be in 50 years, so it’s questionable whether we should be making selection decisions based on such attributes.

Before PGT-P becomes mainstream, we need to think critically and thoughtfully about how to use it. There is also a risk that this technology could further widen disparities in access to care based on wealth. Although the technology exists, how we apply it requires careful ethical consideration.

Many people do not yet know about PGT-P. Most of the discussion so far has come from ethicists and researchers, and there has been relatively little investigation into what patients want. I wrote a paper examining patient perspectives, but even asking patients about PGT-P is difficult because risk scores are hard to understand. People generally prefer binary outcomes—0% or 100%—which is why PGT-A and PGT-M, which are more categorical, are easier for many to grasp. PGT-P, however, produces probabilistic information that is challenging to explain.

Consider a hypothetical choice: an embryo estimated to have a 70% likelihood of developing hypertension but likely to have blue eyes versus an embryo likely to have brown eyes with a lower medical risk. That scenario raises ethically fraught decisions about whether physical traits should ever be prioritized over health risks. I worry about people selecting embryos for traits seen as desirable today, especially when those preferences can be tied to harmful social biases.

I strongly support patient autonomy, and at the end of the day I would allow a patient to choose. But we should still ask whether certain choices should be available simply because the technology exists. Right now PGT-P is not widely used. A few patients have requested it, but that is the minority; most patients seeking fertility care want a healthy baby.

Using IVF and genetic testing for trait selection in the absence of infertility is especially ethically charged. We already use PGT-M for patients with hereditary cancer predispositions, PGT-A for recurrent pregnancy loss or to select sex in some cases, and those applications are relatively mainstream. But when we move beyond health-related risks to non-health physical characteristics, the ethical questions multiply. If companies design scores that prioritize certain conditions or traits, how are those choices being made and by whom?

In summary, PGT-P offers the ability to estimate future disease risk in embryos, but it provides probabilistic scores rather than definitive answers. It raises complex challenges about how to weigh multiple risks, how to explain probabilistic information to patients, and how to avoid reinforcing social harms. The technology is promising but must be adopted carefully and thoughtfully, with attention to ethics, equity, and patient perspectives.

Genetic testing offers powerful capabilities, raises important questions, and has significant limitations with current platforms. Many people invest in IVF, and as providers we hope not only to improve the chance of having a child but also the chance of having a healthy child. I tell patients that although we don't have a perfect platform to guarantee a healthy baby, most babies born are healthy. Gestation itself is a tremendous test of our genome. Some babies are born with diseases, and some conditions manifest later in life; people understandably care about this, especially when a disease has affected them or a family member significantly.

I acknowledge that this is important and desirable. However, there are limitations to technologies like PGT when we try to assess broader risks beyond chromosomal abnormalities or single-gene disorders—for example, complex diseases that involve many genes or gene–environment interactions. It's challenging because we don't yet know how to test for those factors, and environmental variables outside the genome can have an impact later. In some cases, family history can be as predictive—or more predictive—than PGT algorithms.

This is a tricky area. I understand why companies are researching it and why patients ask for it, but we don't yet have enough outcomes data to recommend it routinely. What worries me is creating expectations that we can do more than we actually can, which leads to disappointment and undermines trust: 'Why did I do IVF?' IVF is a useful and important technology for fertility and for preventing monogenic disorders, and we must be careful not to create a perception we cannot meet. As IVF becomes more accepted, we need to be increasingly cautious about how we use these technologies.