NICE Breast Cancer Referral Criteria Miss 95% of At-Risk Women Under 50, Study Finds
核心洞察
Current NICE family history-based referral criteria fail to identify 95% of women under 50 at above-population-level breast cancer (搜索) risk and 95% of those who develop BC within 10 years.
The BOADICEA (搜索) multifactorial risk model incorporating polygenic scores, lifestyle, and family history identified 34.8% of women who developed BC within a decade versus 4.4% with NICE criteria.
73% of women under 50 who develop breast cancer (搜索) have no family history of the disease, explaining the poor sensitivity of family history-only approaches.
Researchers from the University of Cambridge and The Institute of Cancer Research, London have demonstrated that the current NICE family history-based referral criteria for breast cancer (搜索) risk assessment miss up to 95% of women under 50 who are at above-population-level risk, and a similar proportion of those who go on to develop breast cancer within 10 years. The findings, published in a study comparing NICE criteria with the BOADICEA (搜索) multifactorial risk model, suggest that adopting comprehensive risk assessment tools could dramatically improve early identification of at-risk women.
The study analysed data from 1,258 women under age 50 participating in the Breast Cancer Now (搜索) Generations Study, a prospective UK cohort recruited between 2004 and 2011. Among these participants, 392 developed breast cancer (搜索) (338 invasive and 54 ductal carcinoma in situ) within the 10-year follow-up period.
The BOADICEA (搜索) advantage
BOADICEA (搜索) (Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm) integrates multiple risk factors including family history, lifestyle factors, reproductive history, and a 313-single nucleotide polymorphism polygenic score (PGS-313). The model demonstrated good discrimination across all tested configurations, with AUCs ranging from 0.74 (95% CI: 0.70–0.78) for the model without unaffected family members and without PGS, to 0.76 (95% CI: 0.72–0.79) for the full model including both unaffected family members and PGS.
When the full BOADICEA (搜索) model was applied to the entire population of women under 50 in primary care, 26.5% met referral criteria and were categorised as at above-population-level risk. This group captured 34.8% of women who would develop breast cancer (搜索) within 10 years. In stark contrast, applying the current NICE criteria to the same population resulted in only 1.4% meeting referral criteria, capturing just 4.4% of women who subsequently developed breast cancer.
"We need to get better at identifying women at highest risk of breast cancer (搜索) so that we can intervene early, when there are more options for treating, or even preventing, their disease," said senior author Dr Juliet Usher-Smith. "The current NICE criteria used in general practice are missing up to 95% of women under 50 who will go on to develop breast cancer. It's time to look again at these criteria in the light of our findings."
Why family history alone falls short
A critical finding explains the poor performance of the NICE criteria: 73% of women under 50 years of age estimated to be at above-population-level risk based on the full BOADICEA (搜索) model have no family history of breast cancer (搜索). Women who developed breast cancer but were not identified by either approach were less likely to have a first- or second-degree family history of breast cancer (7% vs 21%) and had a lower mean standardised PGS (0.12, SD 0.83 vs 0.46, SD 0.97).
Staged approaches and resource considerations
The researchers evaluated several staged strategies to balance sensitivity with healthcare resource demands. Offering the full BOADICEA (搜索) risk assessment only to women with any first- or second-degree family history of breast cancer (搜索) (approximately 11% of the population) would identify 15.7% of women who develop breast cancer within 10 years, with only 45 women needing to be risk assessed to identify one case — compared with 161 under current NICE criteria.
Restricting assessment further to only those with a significant family history (1.5% of the population) reduced sensitivity to 3.7% but required just 26 women assessed per case detected, still outperforming the NICE criteria's efficiency.
The role of polygenic scores and unaffected relatives
Removing PGS from the full BOADICEA (搜索) model when assessing the entire population substantially reduced sensitivity — from 34.8% to 18.4% of future breast cancer (搜索) cases identified — and increased the number needed to assess to 350 per case detected. However, when risk assessment was restricted to women with a family history, the impact of removing PGS was much smaller. Removal of unaffected relatives from the model had minimal effect across all strategies.
Dr Simon Vincent, Chief Scientific Officer at Breast Cancer Now (搜索), commented: "These findings highlight the limitations of NICE's current referral criteria, and so this research must now be carefully considered as part of the current review of its Family History guidelines. However, 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."
Study limitations
The authors acknowledge several limitations. Family history data on ovarian cancer beyond parents, grandparents, siblings, and children was unavailable, as were data on sarcomas, gliomas, and childhood adrenal cortical carcinomas included in NICE guidelines. Mammographic breast density and rare pathogenic variants were not available in the dataset. The study population was limited to White women, and additional studies will be needed for other races and ethnicities. The analysis also assumed full uptake of risk assessment, whereas the only published UK primary care study reported an uptake of just 16.1%.
The researchers estimate that if the full BOADICEA (搜索) model were implemented, up to an additional 3,000 breast cancer (搜索) cases could be detected earlier or prevented compared with current NICE guidelines. The next step, they note, is to test how multifactorial risk assessment can be implemented safely, equitably, and cost-effectively in routine care.