PGT is four different tests hiding behind one acronym

PGT is a set of genetic tests run on embryos before transfer. Depending on which type, it can tell you whether an embryo has the right number of chromosomes, whether it carries a specific inherited mutation, or what its overall risk looks like for common conditions like heart disease, Alzheimer's, and schizophrenia.

But here's the problem: there are four types of PGT, and they've almost nothing in common besides the biopsy they start from. One type is accurate above 99%. Another produces results that wouldn't hold up two-thirds of the time on a second look. Lumping them together under one acronym does real damage to families trying to make decisions.

If you're interested in learning more about embryo screening or using Herasight's screening, you can reach out to us here.

The four types and what they're actually doing

All four types start the same way. On Day 5 or 6, an embryologist removes a few cells from the trophectoderm (the outer layer that becomes the placenta, not the baby). That biopsy goes to a lab. But what the lab does with it depends entirely on which test you ordered, and the differences are enormous.

PGT-M came first, and it's the simplest to understand. If you or your partner carry a mutation that causes a condition like cystic fibrosis, Huntington's disease, or BRCA1/2, the lab looks at each embryo for that specific variant. Did the embryo inherit it, or didn't it? That's it. One gene, one location, one answer. Its accuracy is above 99%. This has been around for decades, and it works.

PGT-A does something completely different. It counts chromosomes. An embryo should have 46. Sometimes it has 45 or 47, a condition called aneuploidy. PGT-A is the most common type, used in a rapidly growing share of US IVF cycles, and for good reason: aneuploidy rates climb steeply with age, reaching roughly 35% at age 35 and over 60% by age 40. Transferring an aneuploid embryo almost always ends in failed implantation or miscarriage. Identifying normal embryos first saves cycles, time, and heartbreak.

But PGT-A has a problem. About 85% of results come back clearly normal or clearly abnormal, and those calls are solid. The remaining 5-15% come back as "mosaic": the embryo appears to have a mix of normal and abnormal cells. And mosaicism calls aren't reliable. Only 35% of embryos diagnosed with whole-chromosome mosaicism remained mosaic when rebiopsied. Roughly two-thirds of those results wouldn't hold up on a second look.

Think about what that means for a parent whose only remaining embryo just came back "mosaic." They're agonizing over whether to transfer it or not, and the result itself may have been a measurement artifact.

PGT-SR detects structural rearrangements in chromosomes: translocations, inversions, deletions. It matters for parents who carry a balanced structural rearrangement that doesn't affect them personally but can produce embryos with serious problems.

PGT-P is the newest type, and it works on a fundamentally different principle. Instead of looking at one gene or counting chromosomes, PGT-P evaluates an embryo's genetic risk across many common diseases at once: type 2 diabetes, schizophrenia, heart disease, breast cancer, Alzheimer's. These conditions aren't caused by a single mutation. They're influenced by hundreds or thousands of genetic variants, each contributing a small amount of risk. PGT-P adds them up.

Why confusing the types costs families real decisions

So picture this. A couple carries BRCA2. They're considering PGT-M to screen their embryos for the variant. But they've read online that "PGT results are unreliable," that "two-thirds of results don't hold up." They almost skip it.

What they've actually read is a critique of PGT-A mosaicism calls. That has nothing to do with the single-gene mutation testing they need. PGT-M's accuracy for a known variant is above 99%. The acronym almost cost them the screening that mattered most.

It works the other direction too. Someone hears PGT-M is 99% accurate and assumes PGT-P must be similarly precise. But PGT-P produces risk probabilities, not binary answers. Saying "PGT is 99% accurate" and then applying that to polygenic screening is like saying "blood tests are 99% accurate" because one specific blood test is. The accuracy of each type depends entirely on what it's measuring and how it was validated.

How PGT-P validation actually works

This is the part that matters most if you're considering polygenic screening, and it's where most of the confusion lives.

Most polygenic scores are validated on large groups of unrelated people. A study takes 100,000 strangers, checks their genetic scores against their health outcomes, and reports how well the score predicted who got sick. That works fine for population-level research. But embryo screening is a different context entirely.

Your embryos from the same IVF cycle are siblings. They share the same parents, the same household, the same socioeconomic background. All the environmental factors that can inflate a score's apparent accuracy in a population study are held constant among siblings. If a score looks good in a population study but hasn't been tested within families, you don't actually know whether it's picking up real genetic signal or just correlating with things like ancestry and income.

This is exactly why we built within-family validation methods. Every time two siblings are created, the genes they inherit from each parent are scrambled and more or less randomly assigned to each. That natural randomization is powerful for testing whether genetic scores actually predict real differences. Our validation study (Moore et al. 2025) tested 17 disease scores on sibling pairs. Sixteen out of 17 showed no decrease in predictive performance within families. Our scores also achieved 28-193% higher explained variance than other available predictors.

We calibrate across 8+ ancestry groups because a score that only works for one population isn't useful for the families who actually walk through the door.

The right question about PGT-P isn't "is it accurate?" It's "were the scores validated in the same context where they'll be used?" If they weren't tested on siblings, the accuracy numbers don't mean what you think they mean.

If you're deciding whether embryo screening is right for you, start by figuring out which PGT type applies to your situation and evaluate it on its own terms. If you'd like to understand what Herasight's screening could show you about your embryos, reach out to us.