Related Experiment Video
Updated: Jul 18, 2026
Optimization of Breast Biopsy and Mastectomy Sample Collection Procedures for Biobanking, Personalized Medicine, and Research Applications
Published on: September 2, 2025
Underestimation of malignancy of breast core-needle biopsy: concepts and precise overall and category-specific
Nehmat Houssami1, Stefano Ciatto, Ian Ellis
1Screening and Test Evaluation Program, School of Public Health, University of Sydney, Sydney, NSW 2006, Australia. nehmath@med.usyd.edu.au
Background:
A review of the literature indicated variable underestimation rates for breast core-needle biopsy (CNB) based on generally small series. In this article, the authors present precise estimates for overall underestimation and for categories of histologic underestimates (including categories that reflect contemporary classification) and examine the effect of lesion and sampling variables.
Methods:
Among 4035 consecutive CNBs, the authors examined women whose CNB outcome represented a potential underestimate of malignancy (benign but of uncertain biologic or malignant potential or B3 and ductal carcinoma in situ [DCIS]). From 889 eligible women, all 758 women who had excision histology available were included.
Results:
Overall underestimation of CNB was 27.7% (95% confidence interval [95% CI], 24.5-30.9%). The following category-specific rates were used: B3 underestimates, 36.2% (95% CI, 30.6-41.8%); B3 underestimates (excluding atypical proliferations), 17.9% (95% CI, 10.8-24.9%); atypical ductal hyperplasia underestimates, 29.0% (95% CI, 21.4-36.6%; upgraded to DCIS) and 44.2% (95% CI, 36.0-52.5%; upgraded to DCIS or invasive cancer); and DCIS underestimates, 22.8% (95% CI, 19.0-26.5%). There was a significant trend toward greater underestimation of malignancy with increasing lesion size on imaging studies for overall underestimates (P = .00008), B3 underestimates (P = .009), and DCIS underestimates (P = .0007). Underestimation rates did not differ between masses (27.9%) and microcalcifications (27.6%; chi-square statistic with 1 degree of freedom = 3.02; P = .98) and were significantly lower for vacuum-assisted CNB (11-gauge) than for automated CNB (14-gauge; P = .001). Underestimation rates, when sampling microcalcifications, decreased with increasing number of cores collected, but this was mainly for DCIS underestimates.
Conclusions:
CNB results that were not definitely negative or were not positive for invasive cancer were associated with high rates of underestimation of disease and with target lesion size. Identifying imaging or sampling factors that affect underestimation rates of CNB may assist in reducing their occurrence and better predicting excision histology outcomes.
