Feed

Purpose To classify the causes and pathologic characteristics of false-negative screening digital breast tomosynthesis examinations. Materials and Methods All false-negative screening examinations at a single center from March 1, 2019, through December 31, 2023, were collected. Seven fellowship-trained breast radiologists independently reviewed the false-negative screening and diagnostic mammograms preceding biopsy. For each mammogram, radiologists recorded the cancer as visible or not visible. Cancers were classified as mammographically occult (not visible on mammogram), missed (visible on screening mammogram), or true interval cancers (visible on diagnostic mammogram only). Between-classification comparisons were performed with the Pearson χ² test or analysis of variance. Results A total of 117 female patients (median age, 60 years [IQR, 49-70 years]) with false-negative screening mammograms were included. Most cancers manifested symptomatically (88 of 117, 75.2%) a median of 237 days (IQR, 187-301 days) after screening mammography. Among cancers, 47.9% (56 of 117), 26.5% (31 of 117), and 25.6% (30 of 117) were missed, mammographically occult, and true interval cancers, respectively. True interval cancers were more often estrogen receptor- (50.0% [15 of 30] vs 16.1% [five of 31] vs 21.8% [12 of 55]) and progesterone receptor-negative (66.7% [20 of 30] vs 22.6% [seven of 31] vs 35.7% [20 of 56]) than mammographically occult and missed cancers. Mammographically occult cancers were more often stage T0 (32.3% [10 of 31] vs 7.1% [four of 56] vs 10.0% [three of 30]) and detected at MRI (48.4% [15 of 31] vs 3.6% [two of 56] vs 10.0% [three of 30]) and developed in younger female patients (51 vs 62 vs 59 years of age, all P < .05) than missed and true interval cancers. Conclusion Half of false-negative screening mammograms were missed, one-quarter were true interval, and one-quarter were mammographically occult cancer, with differences in pathologic characteristics. Keywords: Mammography, Screening © RSNA, 2026 See also the commentary by Elahi and Slanetz in this issue.