Designing disappointment: The uneasy promise of PGT-P

Cheryl leaned across her kitchen island and glared at her son through a large glass of pinot grigio. She had worked so hard and sacrificed so much for her child. And here he was, all grown up: short, lazy, blue-eyed like her ex-husband, and worst of all, Cheryl thought – dumb. Where had it all gone wrong? Her eyes narrowed. She knew exactly where it had gone wrong. Thirty-three years ago, in 2025, with a subway advertisement and an app. She drained her wine and called her lawyer. This was not the child she was promised. 

Cheryl's situation is not an imagined future; advertisements really were on the walls of the New York City subway in 2025 suggesting that parents could 'have a healthier baby' by genetically testing their embryos prior to IVF and that 'IQ is 50 percent genetic'. The companies behind these claims, including Nucleus and Orchid, are openly advertising embryo screening that purports to predict the risks of a future child developing an array of conditions including cancers, Alzheimer's disease, anxiety, autism, insomnia, diabetes and hypertension.

But the companies are also using the technology to broach new, ethically fraught areas, including the prediction of traits like eye colour, height, male-pattern baldness, BMI, left-handedness, and – perhaps most controversially – intelligence. These predicted risks and traits are presented together in a large, interactive matrix for the consumer to scroll through on a phone app. The interface applies equal weight to each item; it is up to the consumer to choose whether that extra inch of height is worth the added risk of Alzheimer's or acne.

The tests the companies are advertising are preimplantation genetic testing for polygenic disorders, or PGT-P for short. They use polygenic risk scores, increasingly developed using computational methods including artificial intelligence (AI), to aggregate thousands of genetic variations across the genome to estimate an embryo's relative genetic predisposition to certain medical conditions (and in some cases non-medical traits) during IVF.

The technology behind these predictions should not be confused with preimplantation genetic testing for monogenic disorders (PGT_M), which looks at single genes instead of multiple genes, and has been used successfully in fertility treatment for some time. PGT-M is used to test embryos created through IVF for specific inherited conditions, such as cystic fibrosis and Huntington's disease. Unlike PGT-M, which targets a known disease-causing genetic variant, PGT-P uses polygenic risk scores to estimate the relative genetic risk of developing conditions influenced by many genetic variants.

Unlike tests for specific monogenic conditions, polygenic risk scores have limited predictive accuracy for an individual's future disease risk. As a result, even if prospective parents can choose between the various genetic that risks a child may face, environmental and lifestyle factors also play a substantial role in shaping disease and health outcomes. Furthermore, a lower polygenic risk score in a selected embryo does not necessarily translate into a meaningful reduction in the child's lifetime risk of developing a disease.

Despite the sleek, deterministic presentation of results (and much to Cheryl's chagrin), survey research suggests that many US adults are supportive of polygenic embryo screening, particularly when it is used to assess health conditions rather than select for non-medical physical or behavioural traits. Many fertility clinics across the USA, India, Saudi Arabia and Mexico have partnered with PGT-P companies. 

UK law does not currently permit the use of PGT-P to select embryos in the UK, and the Human Fertilisation and Embryology Authority argues that there is no robust evidence to support its use (see BioNews 1319). Despite this, some UK-based patients have sought to access PGT-P via US-based companies (see BioNews 1325).

Though the tests are expensive, typically costing thousands of pounds for up to 20 embryos, they are expected to get cheaper (and perhaps more accurate?) as AI capabilities mature, potentially widening access and increasing pressure on UK clinics to accept their results.

The companies behind the tests are already harnessing AI to help them advertise. Nucleus Genomics has links in the footer of its website that direct the user to ChatGPT, Claude and Grok, to 'see why AI recommends Nucleus as the best choice for you and your future child's health'. These chatbots are primed to provide glowing overviews of the technology to consumers.

Beyond the arguments over clinical utility and accuracy, there are obvious well-documented ethical concerns and potential harms associated with making PGT-P widely available to patients. These fears often centre on embryo selection and the prospect of 'designer babies' – for example, whether tenuous test results could prompt parents to dispose of viable embryos, create stigma around specific diseases or exacerbate disparities in healthcare and health outcomes. Articles often only briefly mention, joke about or overlook the psychological impact that these tests could have on both parents and their children.

So who is responsible for Cheryl's disappointment? Should she blame the company and the AI that made the prediction? Should she blame herself, as she ultimately made the choice of which embryo to use? Should she blame her son? The scientific community has published and promoted work to help clinicians and patients understand the limitations of these tests and manage expectations. But there is an imminent need to develop guidance for parents who may not understand these limitations until it is too late, as well as for the children resulting from embryos selected in this way.

Past innovations in fertility services were developed alongside ethical frameworks and best practices to protect and support children. There are great resources available to parents who have 'chosen' some of their children's genetics through donor conception, helping them to maintain close relationships and mitigate potential psychological impact, but these resources take time and careful thought to develop properly.

NHS England is embracing genetic testing for newborns and the government's Ten Year Health Plan for England (see BioNews 1297, 1300a and 1300b) proposes introducing population-based polygenic risk scoring alongside other diagnostic tools, supported by AI and predictive analytics. Although these initiatives are distinct from PGT-P, they may help to normalise the use of genetic risk predictions in healthcare and among the public.  

As an increasing proportion of children are conceived via IVF (see BioNews 1295), companies will continue to sell a powerful message of divination and reassurance. This may be particularly persuasive for people who use donor gametes, and may have limited information about the donor's family history or health background. As the ethical issues associated with 'designer babies' enter the mainstream, and PGT-P becomes increasingly available, who will take responsibility for the designer disappointments?

---

The article above by Dr Kate Sheridan Clay won second prize in the Marcus Pembrey BioNews Writing Competition 2026. The competition is named in honour of Professor Marcus Pembrey (Patron of PET), and is generously supported by the London Women's Clinic.

The issues discussed in the article will be further explored at this year's PET Annual Conference, AI and Automation in Fertility and Genomics: Pipeline? Or Pipe Dream?, taking place in central London on Wednesday 9 December 2026. Find out more and register here.