Development and evaluation of a breast cancer risk stratification tool for women in Macau.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
BACKGROUND: Widely used breast cancer risk prediction tools are predominantly based on data from Western countries, and their predictive efficacy in Asian women is often questioned, partly due to ethnic differences in risk factors. This study aimed to develop and temporally validate the Macau Breast Cancer Risk Stratification Tool (MBC-RST), based on non-laboratory-detectable indicators. METHODS: This multi-stage study involved the development and temporal validation of the MBC-RST. Phase 1 involved a literature review and retrospective analysis of 14,111 local health questionnaires to create an initial 17-item tool (MBC-RST_Version 1) with expert consultation. In Phase 2, this version was pilot-tested on 10,691 women, while a more detailed cross-sectional survey (MBC-RS) was administered to 1,321 women. Machine learning (specifically LightGBM) and statistical analyses were used to identify additional predictive factors, leading to an optimized 21-item tool (MBC-RST_Version 2). Phase 3 involved the launch and performance evaluation of MBC-RST_Version 2 using data from 9,985 new participants. RESULTS: The initial retrospective analysis identified significant risk factors including age (OR = 1.09), delayed age at menarche (OR = 0.90), presence of a breast lump (OR = 22.2), and unilateral nipple discharge (OR = 2.14). The subsequent factor analysis and machine learning analysis on prospective data identified 'pregnancy history,' 'age at first childbirth,' 'number of miscarriages,' and 'exercise intensity' as highly important predictors. These factors were integrated into the final 21-item tool. The final MBC-RST_Version 2, evaluated on a total of 25,476 participants, demonstrated a sensitivity of 86.7% (95% CI: 72.4%-94.3%) and a specificity of 89.4% (95% CI: 89.0%-89.8%). The tool significantly enhances screening efficiency; it is estimated that only 136 imaging examinations are needed to detect one positive case in the high-risk group, compared to 326 in an age-based universal screening approach. CONCLUSION: This study developed and temporally validated the MBC-RST, a 21-item tool tailored for Macau women which is practical, accessible, and scientifically robust. The MBC-RST showed promising performance in a temporally separated cohort recruited within the same Macau healthcare system. For broader clinical or population-level implementation of this tool, validation in a geographically or institutionally independent population is required.