Rationale and objectives: To determine the limitations of mammography in symptomatic patients.
Materials and methods: Approximately 20,000 mammograms obtained during a 2-year period were correlated with the results of a retrospective review of the pathology database, which revealed 711 patients with breast cancer. A total of 496 of these patients underwent mammography at the authors' institution. Three observers reviewed the mammograms of patients with "normal" mammograms within 6 months of excisional biopsy. The parenchymal patterns were also assessed.
Results: Mammographic abnormalities were identified in 481 cases (97%), and 15 mammograms (3%) were interpreted as "unremarkable." Fourteen of the unremarkable cases were available for review, which indicated nine true-negative findings, one technical error, two observer errors, and two combined technical and observer errors. The technical error involved the lower inner quadrant of the breast, adjacent to the chest wall. All 14 mammograms were of dense breasts.
Conclusion: Mammographic abnormalities were identified in nearly all patients with proved breast cancer. Mammograms in which lesions were missed were of dense breasts, but one-third of the lesions were visible in retrospect. When examining symptomatic patients with dense breasts, radiologists should use optimal mammographic techniques and consider using nonmammographic breast imaging modalities.