Polygenic Embryo Screening

Scoring IVF embryos for traits and diseases governed by many genes (PGT-P), using polygenic risk scores derived from genome-wide association studies, and choosing which embryo to implant on that basis. Offered commercially since ~2025 by Herasight, Nucleus and Orchid (US), at up to $50,000, covering disease risk and, controversially, height, IQ, longevity, hair and eye colour.

Where it sits among embryo tests

PGT-A (aneuploidy / chromosome count): long established; claimed to cut miscarriage risk; the 2026 Lancet review of IVF add-ons found no effect on live-birth odds (IVF Add-ons).
PGT-M (monogenic): single-gene disorders such as cystic fibrosis or Huntington's; near-deterministic, uncontroversial in most jurisdictions.
PGT-P (polygenic): probabilistic risk estimates; the subject of this page.

What the evidence supports

Modelled gains are real but small. Karavani et al. (Cell, 2019, doi 10.1016/j.cell.2019.10.033): choosing among ten viable embryos yields a median gain of ~2.9 cm height or ~3 IQ points (the Economist Leader cites the same study as 1–6 cm and 1–7 points). Gains "drop precipitously" below five embryos. Larger GWAS databases could roughly double the IQ gain.
Demand exists. Survey in Science (2023): 43% of IVF participants would use polygenic screening to improve a child's odds of an elite university if safe and free; a third said the same for editing. ~Three-quarters of Americans would use it if already doing IVF; about a third support screening for non-disease traits.
Market base is small but growing. US IVF births 100,000 in 2024 (2.8% of births); fertility VC $254m → $496m (2019–22); the fertility market ~$25–30bn/yr, with a bull case of $400–500bn if a full IVF course fell to ~$15,000.

What is contested

Individual vs population prediction. Selection on polygenic scores works across many offspring (animal breeding) but is less reliable for one child (Kevin Mitchell, Trinity College Dublin). Herasight answers with sibling-based validation inside GWAS datasets.
Company claims are unverified. Herasight: ten embryos from two type-2-diabetic parents → child's absolute risk down 12–20 points from a 40–60% baseline. Not independently checked in the reporting.
Relative vs absolute risk. Large relative reductions can be tiny absolute ones, especially for rare diseases.
Ancestry bias. Scores are more accurate for European-ancestry embryos because databases over-represent them.
Pleiotropy. Selecting against one condition may raise another the provider does not score; some studies find most polygenic-disease correlations positive or insignificant, which would cap the downside.
Range limit. Even for strongly heritable traits, parents choose within the spread of their own embryos; environment dominates many conditions (Dagan Wells, Oxford).

Regulation and ethics

Illegal in Britain (selection on non-approved factors) and many jurisdictions; America has a de-facto ban on embryo editing since 2016 but not on screening.
Nearly every genetics and reproductive-medicine society in the rich world says the technology is not consumer-ready.
The Economist's position (Leader): don't ban; mandate disclosure of limits and independent scrutiny of tests. Objections it rejects: "unnatural" (applies to IVF too), rich-only (costs fall), eugenics (coercion vs parental choice).
Industry framing: "If this is eugenics, then so is Hinge" (Delian Asparouhov, Founders Fund). All three startups employ genetic counsellors.

Open questions

Will within-family predictive validity hold up in born children, not just models?
How fast does the ancestry gap close, and who pays to collect the data?
Does offering IQ and aesthetic scores, while describing customers as health-motivated, survive regulatory scrutiny once IVF is cheap enough for mass uptake?

Newborn Genome Sequencing — the post-birth counterpart; both are hostage to overestimated penetrance and risk-not-destiny communication.
IVF Add-ons — the evidence record of the procedures sold alongside screening.
Type 2 Diabetes Remission — the condition behind Herasight's flagship risk claim.
Sources: In Praise of Designer-ish Babies (Economist), How Much Would You Pay for a Smarter Baby (Economist).