Polygenic Embryo Screening (PGT-P): What the Science Says vs. What Companies Are Selling

Last updated: March 2026

A company called Herasight sells a product called CogPGT 1.0. It claims to predict your embryo’s IQ. For context, the American Society for Reproductive Medicine, the American College of Medical Genetics, and every other professional body that has reviewed the evidence says polygenic embryo screening is not ready for clinical use (ASRM, Fertility and Sterility, 2025). Herasight has screened “hundreds of embryos” for private customers anyway.

This is the state of PGT-P in 2026: professional consensus on one side, subway ads on the other.

What PGT-P claims to do

Standard preimplantation genetic testing comes in two forms most patients encounter. PGT-A screens embryos for chromosomal abnormalities: extra or missing chromosomes, the kind that cause miscarriage or conditions like Down syndrome. PGT-M screens for specific single-gene disorders: cystic fibrosis, sickle cell disease, Huntington’s. Both test for things caused by clear, identifiable genetic errors.

PGT-P is different. The “P” stands for polygenic, meaning many genes. Instead of looking for a single mutation, PGT-P calculates a “polygenic risk score” by adding up the effects of thousands of genetic variants, each contributing a tiny amount to the probability of developing a condition. Type 2 diabetes, schizophrenia, coronary artery disease, breast cancer. Some companies also score for height, educational attainment, and intelligence.

The pitch: rank your IVF embryos by disease risk, transfer the one with the lowest score, and give your child better odds. Genomic Prediction, the first company to market (2019), calls their version the “Embryo Health Score.” It merges all disease risks into a single number. They operate through around 150 clinics across six continents.

What the science actually shows

The numbers look less impressive once you read past the marketing.

Within-family gains are small. The most-cited paper on PGT-P, by Turley and colleagues in the New England Journal of Medicine (2021), found that naive estimates dramatically overstate the benefit. The commonly cited scenario, selecting between an embryo in the top quintile and one in the bottom quintile for a trait like educational attainment, occurs in fewer than 3% of cases. When you account for sibling genetics (which is what embryo selection actually involves), the expected gain for educational attainment drops from 1.55 years to 0.53 years.

Half a year of schooling. That’s the ceiling, not the floor.

Absolute risk reductions are tiny. Capalbo and colleagues reviewed the full evidence base in Human Reproduction Update (2024), the most thorough assessment published to date. Their findings: relative risk reductions of 15-35% (median 22%) when selecting from 10 embryos. Sounds meaningful until you look at absolute numbers. For schizophrenia, which affects about 1% of the population, the absolute risk reduction is 0.5 percentage points. You’re going from a 1% chance to a 0.5% chance. For a condition that 99% of people never develop regardless.

Turley’s disease-specific numbers tell the same story. Type 1 diabetes: 0.12 percentage point reduction. Breast cancer: 1.9 percentage points. Type 2 diabetes showed the largest effect at 5.5 percentage points, which is the one number the companies prefer to quote.

The ancestry problem is severe. Polygenic risk scores are built overwhelmingly from studies of European-ancestry populations. Capalbo found that accuracy drops up to 75% for African-ancestry populations (Capalbo et al., Human Reproduction Update, 2024). Turley’s data showed accuracy drops of 25-50% for non-European ancestry (Turley et al., NEJM, 2021). A test that works poorly for most of the world’s population is, to put it gently, not ready for clinical deployment.

Selecting for one trait can make another worse. This is called pleiotropy, and it is not a theoretical concern. Turley’s team found that selecting embryos for higher educational attainment inadvertently increases bipolar disorder risk from 1.0% to 1.16%. Genes do more than one thing. Optimizing for trait A while ignoring what those same genes do to traits B through Z is a problem that no company has solved, because it may not be solvable with current methods.

Fewer embryos, less benefit. Capalbo’s review found that benefits diminish rapidly with fewer than five viable embryos per cycle. Most IVF patients do not produce ten embryos to choose from. The average is closer to four or five, and many patients have two or three. The marketing scenarios assume an abundance of options that most patients never have.

What the professional societies say

The ASRM Joint Report (December 2025) is unambiguous: PGT-P is not ready for clinical practice and should not be offered as a reproductive service. Their reasoning covers every major weakness: predictive models lack sufficient data, gene-environment interactions are poorly understood, diverse genomic representation is insufficient, and polygenic risk scores account for only a small percentage of clinical variation (ASRM, Fertility and Sterility, 2025). They recommend research-only use under IRB oversight.

ACMG, the American College of Medical Genetics (Genetics in Medicine, 2023), reached the same conclusion independently. Research settings only.

Two separate professional societies. Same answer. Not yet.

MIT Technology Review, meanwhile, named PGT-P a 2026 Breakthrough Technology (Julia Black, MIT Technology Review, January 2026). The same organization also published a companion piece titled “The race to make the perfect baby is creating an ethical mess.” The left hand and the right hand are both waving, just in different directions.

What the companies are selling

Four companies dominate the commercial PGT-P market. Their positioning varies from cautious to aggressive.

Genomic Prediction (LifeView) launched the first commercial PGT-P in 2019. Their Embryo Health Score collapses risk for type 2 diabetes, schizophrenia, coronary artery disease, breast cancer, Alzheimer’s, intellectual disability, and other conditions into a single number. Around 150 clinics across six continents offer it. They are the closest thing to an establishment player in a field that has no establishment.

Orchid Health sequences more than 99% of embryo DNA (compared to less than 1% by competitors) and screens for 1,200+ monogenic disorders alongside polygenic conditions. Their positioning emphasizes disease prevention, not trait selection. Investors include Anne Wojcicki (23andMe co-founder) and Dylan Field (Figma).

Herasight is the company your genetics professor loses sleep over. Their CogPGT 1.0 product is an explicit cognitive ability predictor. They claim an average 8.5 IQ point difference between just three embryos. They also screen for height, longevity, mental health, and 17 diseases. CBS News featured them claiming to predict lifespan, height, and IQ. Their founders include Jonathan Anomaly, whose name would be too on-the-nose for fiction.

Nucleus Genomics ran subway ads in New York City reading “Have Your Best Baby.” They screen for height, intelligence, disease risk, and behavioral traits. Backed by Reddit co-founder Alexis Ohanian. They open-sourced their models, which is either admirable transparency or a confidence play depending on your perspective.

The broader investor list reads like a who’s-who of Silicon Valley: Elon Musk and Peter Thiel have backed companies in this space. PGT-P is offered at more than 100 US fertility clinics. Testing costs up to $50,000. The global IVF market is worth $28 billion, and investment in women’s health and IVF tech hit $2 billion in 2024, a 55% increase over 2023.

What patients should know

The gap between what the science supports and what the marketing promises is wide enough to park a clinical trial in.

If a clinic or company offers you PGT-P, here are questions worth asking.

“What is the absolute risk reduction for the specific condition you’re screening?” Not relative risk. Absolute. A 30% relative reduction sounds impressive. A 0.5 percentage point absolute reduction, from 1% to 0.5%, may not be worth the cost, the additional biopsy, or the possibility of discarding an embryo that would have been perfectly healthy.

“How was the polygenic risk score validated for my ancestry?” If you are not of European descent, the score’s accuracy may be significantly lower. This is not a minor technical detail. It is a fundamental limitation of the underlying data.

“How many embryos do I need for this to make a meaningful difference?” With two or three embryos, the math rarely favors PGT-P over standard PGT-A screening alone. The studies showing benefit assume five or more viable embryos, which many patients do not have.

“Is this being offered under a research protocol with IRB oversight?” Both ASRM and ACMG recommend PGT-P in research settings only. If the clinic is offering it as a routine add-on, they are ahead of the professional consensus. That is not always wrong, but you should know it.

The desire for a healthy child is not the issue. Every parent wants that. The issue is whether a $50,000 test delivers what it implies when the professional societies reviewing the same evidence say it is not there yet. For most patients, standard PGT-A and PGT-M testing addresses the genetic questions that have clear, actionable answers. PGT-P may join them eventually. The science is not there today.

If you are weighing genetic testing options as part of your IVF cycle, get in touch. We can help you find clinics that offer evidence-based screening and explain what is included in their pricing.

Sources: ASRM Joint Report, Fertility and Sterility, 2025. Turley et al., NEJM, 2021. Capalbo et al., Human Reproduction Update, 2024. ACMG, Genetics in Medicine, 2023. Julia Black, MIT Technology Review, January 2026.