While the Breast Cancer Risk Assessment Tool (BCRAT) predicts breast cancer incidence, the model's performance, re-purposed to predict breast cancer mortality, is uncertain. Therefore, we examined whether the BCRAT model predicts breast cancer mortality in postmenopausal women in the Women's Health Initiative (WHI). BCRAT 5-year breast cancer incidence risk estimates were calculated for 145,408 women (aged 50-79years) enrolled in the WHI at 40 US clinical centers to examine associations of BCRAT risk groups (< 1%, 1-< 3%, ≥ 3%) with breast cancer mortality using Cox proportional regression modeling in all participants and in those with incident breast cancer. Women with BCRAT ≥ 3% risk, compared to women with BCRAT < 1% risk, were older (age 70-79years: 38.3% versus 5.3%), less commonly Black (1.1% versus 40.2%), and had stronger breast cancer family history. With 20-years follow-up, considering all participants, with 8,849 breast cancers and 1,076 breast cancer deaths, breast cancer mortality in BCRAT group ≥ 3% was not higher versus BCRAT group < 1% (Hazard Ratio [HR] 1.06 95% Confidence Interval [CI] 0.80-1.40): percent without 20-year breast cancer mortality; 99.4% [group < 1%] and 98.8% [group ≥ 3%]. Considering women with incident breast cancer, breast cancer mortality was also not higher in BCRAT group ≥ 3% versus BCRAT group < 1% (HR 1.07 95% CI 0.79-1.45). The BCRAT model, at ≥ 3% 5-year incidence risk (US guideline threshold for chemoprevention), does not identify women with higher breast cancer mortality risk, with implications for breast cancer prevention strategies.
Read full abstract