The use of polygenic scores in healthcare is divisive and regularly contested. Some clinicians and researchers are exploring how PGS can be used in specific contexts and see real value in the added precision. Others remain unconvinced questioning whether the gains are large enough to matter compared to existing tools; whether performance holds up across diverse ancestries; and whether scarce public health and NHS resources might do more good spent on prevention approaches that already work at scale. Stuck in the middle ground are those open to the idea in principle, but wary of exactly how, where, and for whom these tools will actually be deployed.
The Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA) model underlies CanRisk, the web interface used to assess women for high risk of familial breast cancer syndromes, and is an example of well validated polygenic scores. In fact, the PHG Foundation used CanRisk and, by extension BOADICEA, in our own roundtable report as the reference point for thorough independent external validation, equity analysis across ancestry groups, iterative co-design with clinical services, and full regulatory scrutiny. If polygenic scores were going to prove their worth in clinical practice this was very likely to be the place.
This week, some of that proof arrived and it garnered interest well beyond academic journals. ITV News visited the study site to interview Dr. Juliet Usher-Smith, the research’s senior author, about the findings, signalling immense public interest in getting breast cancer risk assessment right. Whilst the genomic component of this tool was relatively underreported in the media this demonstrated a major step forward for polygenic scores, it bring the topics debated at our 2026 roundtable back into focus. We must ask not just whether the science holds up, but what it takes to turn a validated tool into care that reaches people well.
Identifying risk Under 50: BOADICEA vs. current criteria
Researchers from Cambridge and the Institute of Cancer Research compared two ways of identifying women under 50 at higher risk of breast cancer. They compared the BOADICEA risk model, which combines family history with lifestyle, reproductive history and genetic information from familial predisposition such as BRCA1/2 and polygenic risk, against current NICE referral criteria used by GPs. Current guidelines are based almost entirely on family history. This analysis was enabled by The Generations Study, a long-term UK cohort study to investigate breast cancer launched in 2003 and recruiting more than 112,000 people.
Offering the full BOADICEA assessment to all women under 50 would identify 26.5% as above-population risk and eligible for earlier screening. This would capture 34.8% of the women who go on to develop breast cancer within 10 years. In contrast, using the NICE criteria would result in just 1.4% of women being referred, only identifying 4.4% of those who actually develop the disease. The primary reason for this difference is that 73% of women under 50 who develop breast cancer have no family history, demonstrating current shortcomings of breast cancer screening in the UK .
Yet, while recent policy initiatives, such as the Genomic Population Health service, underscore a growing momentum toward implementation, the logistical complexities of scaling these tools present challenges as formidable as the underlying science.
Professor Montserrat García-Closas, a co-author from the Institute of Cancer Research, illustrates this tradeoff:
“The NICE criteria are much easier to implement, but miss a large proportion of women at elevated risk. Conversely, a full risk assessment including genetic testing places a heavier burden on resources.”
Key considerations for NHS implementation
Ultimately, health systems must weigh a myriad of logistical and financial costs against the benefits of accurate risk classification and the harms of misclassification, all key topics that were discussed at the roundtable. In light of this high impact work there are several factors brought to the fore once again:
- Better identification isn’t automatically translated to better care. The change from up to 1.4% to 26.5% of women being referred would be an almost twentyfold increase in assessment volume. Each referral needs a clinician’s time, an actionable pathway following the result, and a system that can cope with this increased workload. BOADICEA has successfully demonstrated the ability to identify risk more accurately but can the NHS can actually deliver the pathways and care that a positive result is supposed to trigger?
- Nearly 30% of women do not attend their screening invitation. Could a more precise, genetically-informed risk assessment provide an individual sense of risk that would give some of that group the confidence to attend? There’s also the question of what happens once someone is assessed as lower risk and moved off an enhanced pathway.
- A polygenic score doesn’t change, but the other factors BOADICEA draws on such as weight, hormone use, reproductive history, can shift considerably over time. Without a clear mechanism for reassessment, a one-off result risks becoming a fixed label rather than a living picture of risk. A tool that identifies risk more accurately only delivers benefit if the people it identifies actually engage with what follows.
- The new Genomic Population Health Service will need to address these aspects in more detail, and ongoing evaluation and scrutiny will be critical with growing momentum towards implementation
- Realising the full potential of risk profiling would require system transformation. Beyond raw referral numbers, achieving this level of early detection would require implementation of polygenic scores into GP risk assessments for women who are under 50 and therefore not eligible for the age-based NHS screening programme. Moving down this path represents a shift toward a paradigm of proactive, lifelong health surveillance in the NHS. To make this vision a reality, healthcare systems have to rethink how they combine technology with their daily operations and culture. The lack of interoperability between data systems needs to be addressed and importantly co-design with key stakeholders needs to be embedded at the earliest stage. Making that shift won’t happen overnight. It will take new support systems and serious funding to bridge the gap across primary and secondary care, a staggering undertaking considering the current state of the NHS, but one made more urgent by the fact that women under 50 currently have no systematic early-detection pathway at all. National screening doesn’t begin until age 50, so for younger women, catching disease early depends almost entirely on the family-history route; a route that the paper discussed here shows catches just 4.4% of cases.
- Equity goes both ways. A more accurate tool applied unevenly can widen the health disparities it is meant to close. Uptake of primary care risk assessments currently hovers around 16% and is far from evenly distributed. Risk models, particularly the PGS component, have been developed on European ancestry data and significantly more research is needed to understand uptake in different populations. Complexity and lack of trust risk undermining these efforts, deepening existing inequalities. This is particularly problematic if attention and funding shift away from simpler approaches that were at least reaching everyone. As Dr. Simon Vincent, Chief Scientific Officer at Breast Cancer Now, notes: “It’s equally important that any changes come with the needed investment in family history services, so they can be implemented effectively and fairly across the NHS.”
- More referrals means more false alarms. Of the 26.5% flagged as above-population risk under full BOADICEA assessment, most will not go on to develop breast cancer within the window studied. Whilst this is not a flaw in the model, and a feature of screening at this level of sensitivity, it has a real cost in anxiety, follow-up appointments, and instances of unnecessary investigation, all of which add to the burden on both the individual and the health system. For many women, being categorized as ‘high risk’ without an immediate diagnosis introduces a state of ongoing, unresolved anxiety that health services must be equipped to support. Is society ready to live under a state of perpetual risk surveillance?
Next steps for system integration
The translational bottleneck from research to clinical practice is a long standing problem. The creation of a tool that “works” does not mean that it will be adopted in practice. Encouragingly, the researchers aren’t stopping at “the tool works.” Dr Usher-Smith and colleagues are already running a study evaluating how multifactorial risk assessment could work safely, equitably and cost-effectively in routine general practice.
Breast cancer risk assessment has achieved a milestone that much of the polygenic score field is still striving toward: the clinical evidence is sufficiently robust, and now the core question is how it might be deployed to offer benefit to the population at risk. It serves as a clear, real-world reflection of the key themes highlighted during our roundtable discussions. Crucially, demonstrating clinical validity is a distinct step from establishing operational readiness. A validated tool does not automatically arrive with workforce capacity, equity frameworks, or a consensus on acceptable risk thresholds attached. These aspects require a different form of evidence, strategic planning, and health system preparation. At the PHG Foundation, we view addressing these practical building blocks prospectively and collaboratively alongside clinical and policy partners as key to ensuring smooth and successful integration into care.
