According to current NICE criteria, most young women at risk of breast cancer are not screened

The comprehensive BOADICEA risk assessment model was able to identify significantly more women under 50 who would develop breast cancer in the subsequent decade than current UK referral criteria. Using the full version of the model, 34.8 percent of prospective patients were classified as high-risk, while the UK National Institute for Health and Care Excellence criteria would have referred only 4.4 percent of them for further assessment. The difference between the approaches was nearly eightfold.

The cost of higher detection rates was significant. According to the researchers' calculations, applying the full model to all women under 50 would have resulted in referrals for specialized assessment for 26.5 percent of those screened. Using the current criteria, only 1.4 percent would have been referred. In other words, a more accurate search for potentially vulnerable women would have required a significant expansion of genetic, diagnostic, and counseling services.

One reason for this difference is that 73 percent of participants diagnosed with breast cancer within ten years had no family history of the disease. Meanwhile, the initial referral decision under current criteria largely depends on the number of affected relatives and the age at diagnosis. BOADICEA additionally takes into account reproductive, hormonal, behavioral, and genetic factors.

However, the statement that the guidelines "miss 95 percent of cases" should be interpreted with caution. The NICE guidelines were created primarily for the care of people seeking medical attention due to a family history of cancer, not as a tool for mass screening of all young women. The new study effectively compares a passive, family-focused approach with the proposed proactive population-based risk assessment system.

Why is risk in young women difficult to determine in advance?

In the UK, women with average population risk are generally not included in national mammography screening until they reach the age specified by the program. However, some women under 50 may require earlier screening if their risk of developing the disease is significantly higher than average. This requires identifying the increased risk before symptoms develop.

The existing pathway largely begins with the patient herself. A woman must be aware of any family history of cancer, assess its potential significance, and consult a general practitioner. The specialist collects information about first- and second-degree relatives, age at diagnosis, and history of ovarian cancer and certain other tumors. If the established criteria are met, the woman is referred to a specialized service.

This approach is well suited for identifying families in which rare inherited variants with a strong effect may circulate. For example, multiple cases of breast cancer at a young age or a combination of breast and ovarian tumors may indeed indicate a hereditary predisposition. However, the absence of affected relatives does not mean the absence of an increased risk.

The family may be small, female relatives may not have lived to the age of highest incidence, and information on paternal diseases is sometimes incomplete. Furthermore, a significant portion of the risk is determined not by a single high-risk genetic variant, but by a combination of numerous common genetic traits, reproductive history, hormonal influences, and lifestyle. It is precisely this shortcoming of the family-centered approach that the authors of the new study attempted to quantify.

Approach	Where does the assessment begin?	What data is taken into account?
Current NICE primary route	A woman comes in because of a family history or concerns	Mainly the number of sick relatives, the degree of kinship, types of tumors and age of diagnosis
Specialized assessment	After referral from primary care	More detailed pedigree, calculation models, sometimes genetic testing
Full BOADICEA	Proactive data entry and genetic evaluation	Family history, personal factors, reproductive and hormonal data, lifestyle, rare variants and polygenic risk
Potential population program	Inviting women of a certain age	Standardized multifactor calculation for all invitees

How the study was organized

The authors analyzed data from 1,258 women under 50 years of age from the British Breast Cancer Now Generations Study. Participants were included in a larger cohort between 2004 and 2011. They were free of breast cancer at the time of recruitment, and their health status was then tracked for the following decade.

The study was not a clinical trial of a new test or preventative treatment. The researchers used previously collected questionnaires, family history information, and stored biological samples to calculate baseline risk as if different assessment methods had been used at the time the woman entered the cohort. These predictions were then compared with actual diagnoses recorded over a ten-year period.

The analytical sample was a specially selected subset of the cohort, including women who later developed breast cancer. Therefore, simply dividing the number of cases by 1258 would not reflect the true incidence of the disease in the population. The authors used statistical weighting to ensure that the results corresponded to the expected risk distribution among British women of the corresponding age. Independent experts noted that this design is suitable for comparing models but introduces additional uncertainty when generalizing percentages to the entire population.

The researchers tested several versions of BOADICEA, from simpler calculations to a full assessment with genetic data. The main analysis compared the current referral criteria with the most detailed model, which included family history, questionnaire risk factors, and a polygenic score. The key question was not how accurately the model would diagnose, but rather how many future cases would be among women who would be recommended for further monitoring.

Characteristic	Data
Type of work	Analysis of predictive models in a prospectively observed cohort
Data source	Breast Cancer Now Generations Study
Analytical sample	1258 women
Age	Under 50 years of age at the time of initial assessment
Recruitment period	2004-2011
Forecast horizon	10 years
Compared approaches	NICE referral criteria and several BOADICEA options
Main outcome	Diagnosis of breast cancer within the next 10 years
Randomization or treatment	Not conducted

Main results of the comparison

According to the authors' calculations, the current NICE criteria would result in approximately 1.4 percent of women under 50 years of age being referred for additional assessment. Among those who actually developed breast cancer over the next ten years, 4.4 percent would meet the established criteria.

When applying the full BOADICEA to all women, the proportion of women who were referred increased to 26.5 percent. This group accounted for 34.8 percent of future breast cancer cases. The ratio of 34.8 to 4.4 means that the multifactorial approach identified approximately 7.9 times more women who were subsequently diagnosed with the disease.

This difference does not mean that BOADICEA could predict all future cancers. Even with the most detailed assessment, approximately two-thirds of women who subsequently developed the disease would not have been classified as above the population risk group by the chosen threshold. Any model operates on probabilities and cannot precisely predict the fate of an individual.

The results also show that increased sensitivity comes with a sharp increase in the number of screenings. For every 100 women, the current approach would have referred approximately one or two, while the full multifactorial strategy would have referred approximately 27. Most women identified by the model as candidates for further evaluation would not have received a diagnosis during the ten years studied. The authors explicitly identify this as a key tradeoff between risk identification, cost, anxiety, and potential oversurveillance.

Result	NICE criteria	Full BOADICEA
Proportion of all women referred for further assessment	1.4%	26.5%
Proportion of future cancer cases falling into the target group	4.4%	34.8%
Estimated relative increase in future case detection	Reference indicator	7.9 times
The main source of information	Family history	Family, personal, reproductive and genetic data
The need for genetic research	Not necessarily at the initial referral stage	Required for the most complete version
Expected load on health services	Relatively low	Significantly higher

Why family history was insufficient

Among the participants who developed breast cancer over a ten-year period, 73 percent did not report a family history of the disease. Therefore, they could not be identified using an approach that uses relatives with a history of the disease as the primary referral signal.

The absence of a family history should not be confused with the absence of a genetic predisposition. Some women may carry a rare pathogenic variant inherited from their father, even if no close relatives have ever had the disease. In other cases, the risk is determined by a combination of hundreds of common genetic variants, each of which individually only slightly alters the likelihood of developing the disease.

Family history also reflects not only genetics but also the structure of a particular family. In a large family with several older female relatives, hereditary predispositions are more pronounced than in a small family with few female relatives or most of the female relatives are still young. Therefore, the same biological predisposition can appear completely different in a family tree.

This doesn't render family history useless. It remains an important and relatively inexpensive source of information, especially for rare hereditary syndromes. The study shows otherwise: family history is poorly suited as a single input filter for identifying all women with increased multifactorial risk.

Why family history may not reveal risk	Example
A small number of female relatives	There are few women in the family, so the predisposition did not have time to manifest itself
Young age of the family	Relatives have not yet reached the age when the disease most often occurs
Inheritance through the paternal line	The variant was transmitted by the father, but his closest relatives did not have the diagnosis.
Incomplete information	The woman does not know the exact diagnoses or age of her relatives' illnesses
Polygenic risk	There is no single strong family variant, but a combination of common variants increases the likelihood
Non-genetic factors	Risk is influenced by reproductive history, hormonal and behavioral characteristics

What data does BOADICEA use?

BOADICEA is a mathematical model implemented in the CanRisk clinical tool. It can integrate detailed family structure, history of breast, ovarian, pancreatic, and prostate cancer, the ages of relatives, and genetic testing results. Based on this information, it calculates the individual probability of developing the disease at a specific age or time interval.

A separate part of the model is related to rare pathogenic gene variants that significantly increase risk. These include variants in BRCA1, BRCA2, PALB2, CHEK2, ATM, and several other predisposition genes. This genetic testing result could significantly change recommendations for monitoring and prevention.

A polygenic score has a different meaning. It summarizes the effects of a large number of common single-nucleotide variants. Each has a small effect, but their combined effect helps divide the population into groups with relatively lower and higher genetic predisposition. The expanded BOADICEA used a score based on 313 variants.

The model can account for age at first menstruation and menopause, number of pregnancies and age at first birth, use of hormonal medications, height, weight, alcohol consumption, and other factors. The general version of BOADICEA can also use mammographic density, but this important indicator was not included in the analysis of women under 50. Independent experts identified this as a limitation of the study.

Model component	What does it reflect?
Detailed pedigree	Diseases and age of relatives on the maternal and paternal lines
Rare pathogenic variants	Strong hereditary predisposition associated with individual genes
Polygenic assessment	The combined effect of many common genetic variants
Reproductive history	Age of menarche, childbirth, menopause and other hormonally related factors
Lifestyle and anthropometry	Body weight, height, alcohol and other characteristics
Hormonal drugs	Contraceptives and hormone replacement therapy
Mammographic density	An additional strong risk indicator not included in the main analysis of this paper

What you have to pay for higher sensitivity

The main practical obstacle is scale. A comprehensive assessment for all women of a certain age would require invitations, detailed questionnaires, quality control of family information, biological sampling, genotyping, risk calculation, and interpretation of results. This is a fundamentally more complex process than referring a small number of women after a routine consultation.

According to the modeling, the number of referrals would increase almost 19-fold: from 1.4 to 26.5 percent. Family oncology and clinical genetic services would need to see significantly more patients. Additional specialists, laboratory capacity, imaging equipment, and a follow-up system would be required.

Extended assessment inevitably produces false-positive results from a practical standpoint. A woman may be correctly classified as being at high statistical risk but never develop the disease. This is not a model error: risk denotes probability, not certainty. However, for a specific individual, such a result can trigger anxiety, repeated testing, and a sense of constant threat.

The opposite effect—false reassurance—is also possible. Approximately 65 percent of future cases were not included in the referral group, even with the full model. Therefore, a low or average score does not negate routine attention to breast changes and does not guarantee the absence of disease. Any future program must account for both sides of this uncertainty.

Possible consequence of extended assessment	Potential benefits	Potential harm or burden
Genetic testing	Identification of a previously unknown predisposition	Cost, privacy issues, and chance discoveries
More directions	More women are receiving individual counseling	Overload of specialized services
Early observation	Possibility of detecting tumors at an earlier stage	Additional radiation exposure and false positive results
Preventive treatment	Possible reduction in risk in some women	Side effects and the need for difficult choices
Risk communication	Informed health decisions	Anxiety and the misperception of probability as a diagnosis
Personalization	The intensity of surveillance is proportional to the risk	Risk of unequal access between regions and social groups

Why the "95 percent missed cases" claim is controversial

The authors and the university press release summarize the result as follows: the current criteria could have missed up to 95 percent of young women who subsequently developed the disease. Arithmetically, this follows from the fact that the criteria identified 4.4 percent of future cases. However, independent experts noted the purpose of the compared tools.

Guideline CG164 addresses familial breast cancer. It is used when a person presents with concerns about a family history of the disease and explicitly states that, in most cases, healthcare providers should not actively seek out individuals with a family history. Therefore, the criteria were not designed as a population-based test intended to detect the majority of future cases among all women.

Cancer epidemiology professor Paul Pharoah described the scientific portion of the study as high-quality, but found the primary media comparison misleading. He believes that family-focused guidelines are predictably poor at providing comprehensive prognostic information, as most young patients have no known family members affected by the disease.

Biostatistician Adam Brentnall assessed the study more positively, calling it important evidence of the potential benefits of comprehensive calculations. However, he noted that, given statistical uncertainty, the upper estimate of the proportion of missed future cases is closer to approximately 92 percent, not necessarily exactly 95 percent. Thus, the underlying difference is indeed large, but the headline figure is an extreme estimate.

What might change in prevention and screening?

NICE guidelines already recommend annual mammograms for women aged 40-49 years who are considered moderate-risk. The moderate category typically corresponds to a risk of developing breast cancer of 3 to 8 percent over the decade between ages 40 and 50; a risk above 8 percent places a woman in the high category. The problem is that many women don't reach the stage at which this risk is calculated.

A proactive program could invite women, for example, aged 30 to 49, to complete a standardized questionnaire and, if they consent, provide a sample for polygenic assessment. After assessment, some women would continue routine monitoring, while those at higher risk would receive counseling on earlier screening, preventative medications, and manageable risk factors.

This proposal is still a concept, not a new clinical recommendation. The current study showed how many future cases would statistically fall into the selected categories, but it did not test whether the program itself would reduce mortality, the incidence of advanced tumors, or the need for complex treatment. This would require practical studies with long-term follow-up.

There is data from a separate British randomized trial showing that annual mammography, starting at age 40, reduced breast cancer mortality. However, this does not mean that mass genetic screening and a risk-based program will automatically yield the same results. Optimal thresholds, screening intervals, and the benefit-to-risk ratio need to be determined separately.

NICE risk category	Estimated 10-year risk between ages 40 and 50	General approach
Close to population	Less than 3%	Routine observation
Moderate	3-8%	Annual mammography may be recommended between the ages of 40 and 49.
High	More than 8%	Specialized observation; tactics depend on genetics and other factors
Very high hereditary predisposition	Determined by gene variants and advanced calculation	MRI, medications, or discussion of risk-reducing surgery may be possible.

A risk category is not a diagnosis in itself and does not mean that the disease will definitely occur.

Main limitations of the study

The first issue is the sample size. All women included in the analysis were white. Polygenic assessments often perform less well in populations underrepresented in genomic studies. Therefore, the accuracy and thresholds of the full BOADICEA assay need to be separately tested in women of different ethnicities.

Participants in the large cohort were volunteers who agreed to complete questionnaires and provide biological samples. They may have differed from the general population in terms of education, health, family history, and attitudes toward preventive care. The authors used statistical adjustments, but it is impossible to completely eliminate systematic differences with this design.

The model did not include mammographic density, which has significant prognostic value. Artificial intelligence algorithms capable of extracting future risk information directly from mammograms are also rapidly developing. Therefore, the comparison does not cover all the tools that could potentially be included in a future personalized screening program.

Finally, the study assessed classification, not clinical outcome. It is unknown how many additional tumors would have been detected early, how many deaths would have been prevented, and how many women would have experienced unnecessary biopsies, anxiety, or side effects from prophylaxis. Without these data, the economic and medical feasibility of national implementation cannot be determined.

Limitation	Why is it important?
Only white participants	The accuracy of the genetic estimate for the entire ethnic diversity of the population is unknown.
Volunteer cohort	There may be systematic differences between participants and the general population
Specially formed analytical sample	The proportion of cases cannot be interpreted as a normal incidence
No mammographic density	The model did not use all known strong predictors
No data visualization and artificial intelligence	The comparison does not include new risk assessment methods.
Observation of classification, not intervention	No evidence of reduction in mortality or severity of treatment
A significant number of additional directions	Cost, anxiety and overdiagnosis are not fully appreciated
Results for women under 50	The findings cannot be automatically transferred to other age groups.

What the study showed—and didn't show

The study showed that family history, as a primary screen, only identifies a small proportion of women under 50 who may have an increased multifactorial risk. The full BOADICEA significantly expands the identified group and includes nearly eight times more future cases.

The study also showed that genetic information adds prognostic value to the questionnaire and family history. Polygenic assessment proved particularly valuable, allowing for the identification of predisposition in women without a strong family history. However, the results of a polygenic test do not identify a single "cancer gene" and do not determine a person's fate.

The study did not prove that all women under 50 should immediately undergo genetic testing. It did not compare actual screening programs, prescribe preventive medications, or assess mortality. The data obtained should be viewed as a model for a potential entry point into a prevention system.

The most definitive conclusion is that proactive multifactorial assessment is better at stratifying young women by risk than a passive approach based almost exclusively on family referrals. The next step is to determine whether this approach can be implemented safely, fairly, and cost-effectively.

The study showed	The study did not prove
BOADICEA identified more future cases than NICE referral criteria	What the model predicts for each future case
73% of those infected had no family history	That family history no longer matters
A full assessment would refer 26.5% of women	That the medical system is capable of handling such a volume
Genetic data improves classification	That all women need immediate genetic testing
A group can be identified for early observation	That such surveillance will definitely reduce mortality
The existing route is poorly suited for population search	That NICE guidance is wrong in its original family-focused objective
The approach is promising for personalization	That the benefits already outweigh the cost and possible harm

Results of the study

The new study demonstrates a fundamental difference between the two strategies. The current approach waits for a woman to disclose a significant family history and then refers her to specialists. BOADICEA proposes assessing risk proactively by integrating multiple independent sources of information.

The full model does indeed capture significantly more future cases: 34.8 percent versus 4.4 percent. However, it also increases the proportion of referrals from 1.4 to 26.5 percent. Therefore, the main question is not only about accuracy, but also about the implications of expanding the surveillance group.

The authors believe the findings warrant a review of the existing entry route and further exploration of proactive programs. Representatives of Cancer Research UK and Breast Cancer Now emphasize that changes must be accompanied by investment in staff, equipment, and family genetic services, and must address concerns and potential inequalities in access.

Independent experts agree that the study provides high-quality evidence of the effectiveness of multifactorial assessment. The disagreement centers on the next step: whether there is sufficient data for mass risk prediction or whether it is first necessary to demonstrate that the proposed interventions actually improve long-term outcomes.

News source

Frost R, Ficorella L, Berrington de Gonzalez A, García-Closas M, Usher-Smith JA, et al. Comparison of NICE criteria with the BOADICEA multifactorial risk model to guide breast cancer risk assessment and referral among women under age 50 within primary care. British Journal of Cancer. Published online August 4, 2026.

DOI: 10.1038/s41416-026-03547-2

The study is based on data from the long-term Breast Cancer Now Generations Study cohort. The authors used the BOADICEA tool, developed by the University of Cambridge and implemented in the CanRisk platform.