Model for Predicting Breast Cancer Risk in Women With Atypical Hyperplasia.

Area of Science:

• Oncology
• Biostatistics
• Pathology

Background:

• Women with atypical hyperplasia (AH) on breast biopsy face an elevated risk of breast cancer (BC).
• Current risk prediction models lack precision for individualizing BC risk in this specific high-risk population.
• Accurate risk stratification is crucial for guiding surveillance and management strategies.

Purpose of the Study:

• To develop and validate a novel breast cancer risk prediction model specifically for women diagnosed with atypical hyperplasia (AH).
• To designate this new model as the AH-BC risk prediction tool.
• To improve individualized risk assessment for women with AH.

Main Methods:

• Utilized retrospective cohorts from Rochester, MN, and Nashville, TN, comprising women aged 18-85 with pathologically confirmed benign AH.
• Employed L1-penalized Cox proportional hazards regression to identify key clinical and histologic risk factors.
• Applied a Fine and Gray regression model to estimate BC risk, accounting for death as a competing risk, and assessed model performance using discrimination and calibration metrics.

Main Results:

• The model-building set included 699 women with AH (142 BC events), and the external validation set comprised 461 women (114 BC events).
• The final AH-BC model incorporates age at biopsy, age at biopsy squared, and the number of AH foci.
• The AH-BC model demonstrated good discrimination (0.63) and calibration (0.87) at 10 years in the development set, with acceptable performance (0.59 discrimination, 0.91 calibration) in the external validation set.

Conclusions:

• A new, validated model (AH-BC) has been developed to refine breast cancer risk prediction for women with atypical hyperplasia.
• The AH-BC model exhibits robust discrimination and calibration, proving effective in an external validation cohort.
• This tool offers a more precise method for assessing individual breast cancer risk in women with AH, aiding clinical decision-making.