Polygenic embryo screening: what it is and why the critics were wrong about it

Polygenic embryo screening is a type of genetic screening available to parents going through IVF that estimates each embryo's predisposition toward common diseases. It can be used to reduce a future child's risk of type 2 diabetes, breast cancer, Alzheimer's, heart disease, and more. It's a significantly more comprehensive version of the genetic testing that's been available for decades, which could only screen for conditions caused by a single gene, like cystic fibrosis.

For years, the main critique went like this: scores validated on large populations of unrelated people would collapse when applied to embryos from the same two parents. That was a real concern. It was also testable. We tested it. For 16 of 17 disease scores, the critique turned out to be wrong.

You can reach out to us here if you're interested in getting access.

How it works

When two parents go through IVF, each embryo they create inherits a unique random combination of their DNA. For any disease influenced by many genes, some embryos will happen to inherit more risk-increasing variants and others will inherit fewer. Nobody controls this. It's the genetic roll of the dice.

A few cells are taken from each embryo on day 5 or 6 of development (the same biopsy used for standard chromosome screening), and the DNA is sequenced. From that sequence, we can estimate each embryo's risk for conditions across the board.

That's a completely different kind of test from preimplantation genetic testing for single-gene disorders like Huntington's or BRCA mutations. Those tests look for one specific, known mutation. Polygenic screening looks at the combined effect of many common variants, each individually small, that together shape someone's risk for the diseases that actually fill hospital beds: diabetes, heart disease, cancer, Alzheimer's.

The critique that held things up

The concept has been around since the first clinical application in 2019, but it stayed controversial because of a real question about validation. Most polygenic risk scores are validated by testing whether they predict disease outcomes in large groups of unrelated people. But embryos from the same IVF cycle aren't unrelated. They're siblings, sharing roughly 50% of their DNA. Critics, most prominently Turley et al. in the New England Journal of Medicine, predicted that this shared genetic background would substantially reduce the scores' predictive power.

That prediction was directly tested. We built seventeen disease polygenic scores and validated them on sibling pairs. The result: 16 of 17 showed no meaningful decrease in predictive performance within families. The scores pick up real causal genetic signal, not statistical artifacts that vanish when you control for family background.

And there's a double standard here that we find hard to ignore. Preimplantation genetic testing for monogenic disorders has been used to screen embryos for BRCA1 and BRCA2 mutations since 2009. None of those children have reached breast cancer onset age. Nobody calls that testing "unvalidated." The underlying logic is the same: genetic variants that predict disease in adults are present in embryos from conception. But when it comes to polygenic predictions, the standard of proof suddenly gets much higher.

What the numbers look like

In Moore et al. (2025), our scores achieved a liability R2 of up to 0.21 for type 2 diabetes. In practical terms: for families where both parents have type 2 diabetes and there are 10 embryos to choose from, the absolute risk difference between the highest and lowest-risk embryo can reach 23.5%.

That's not a subtle statistical effect. That's the difference between a child who's likely to develop diabetes and one who probably won't.

Professional societies like ASRM have raised concerns about whether polygenic screening is ready for clinical use, and their primary worry has been exactly what we tested: do these scores hold up within families? The methodological standard they're calling for is the one we already practice.

We've also calibrated performance across 8+ ancestry groups, because genetic testing that only works for people of European descent isn't good enough. That's been true for decades and the field has been way too slow to fix it.

Genetics doesn't explain everything. It accounts for roughly 24% of disease susceptibility (varies by condition), and environment still matters. But the part that is genetic? You can now see it before transfer. And the more embryos you have to choose from, the more of a difference it makes.

What this means if you're going through IVF

If you've already done PGT-A, your existing data may already contain polygenic information. We built ImputePGTA to reconstruct polygenic risk scores from standard PGT-A sequencing data, with only about 5-10% less accuracy compared to dedicated sequencing. You don't need a new biopsy.

What you get is a report showing absolute risk estimates for each embryo across multiple conditions. Actual percentages, not vague categories. Our genetic counselors walk you through what those numbers mean for your specific situation, because a 15% risk of type 2 diabetes means something very different depending on whether your family has a long history of the disease.

Right now, tens of thousands of children are being born every year with higher genetic predispositions toward disease than they needed to have, because the standard approach to embryo screening mostly comes down to chromosome count and how the embryo looks under a microscope. That's not enough information to make the best decision. We think families deserve the full picture.

If you're curious about how polygenic screening applies to your family, you can explore our embryo screening guide or take the quiz to see if screening is right for you. If you're interested in learning more or using Herasight's screening, please reach out to us.