Assessing the Effect of Breast Density on Screening Performance in a Population‐Based Program

Abstract

Background Breast density is both an independent risk factor for breast cancer and a major determinant of mammographic diagnostic performance. However, its impact on overall screening performance within population‐based screening programs is less well understood. Methods This retrospective cohort study included women aged 50–74 years screened through BreastScreen NSW between 1 January 2016 and 31 December 2017, excluding first‐time screeners above 70. Breast density was assessed using Lunit INSIGHT MMG (v1.1.7) generating an ordinal density score (1–10) and corresponding BI‐RADS category (A–D). Screen‐detected and interval cancers were ascertained via the NSW BreastScreen Information System and the NSW Cancer Registry. Standard screening performance metrics were calculated stratified by age, breast density and screening round. Results The cohort comprised 564,681 screening episodes, with 4758 cancers diagnosed (3756 screen‐detected; 1002 interval). Cancer incidence rates and screening performance varied by breast density. The proportion of cancers that were interval cancers increased markedly with increasing density, from 10.9% of cancers in category A (least dense) to 46.6% in category D. Recall rates followed an N‐shaped pattern, peaking in density category C, while specificity mirrored this relationship. Conclusion The effectiveness of conventional two‐dimensional mammography is strongly dependent on breast density, with reduced sensitivity and increased interval cancer rates for clients with extremely dense breasts. By utilising a more granular measure of breast density in a large population screening cohort, this study identified nonlinear relationships between breast density and screening performance metrics, providing evidence to support future evaluation of more individualised screening approaches.