[Critical review of 215 cases of non-palpable mammary lesions]

The natural history of human breast cancer shows that lesion size correlates directly with nodal metastases and distant spread. Nodal metastases are found in only 6% of cases in the preclinical stage of the tumor and therefore imaging must detect a breast cancer before it becomes palpable. We reviewed 215 nonpalpable breast lesions studied in the last 10 years to assess observers performance and ultimately improve the interpretation of suspicious mammograms, evaluating "cost" in terms of the ratio between benign and malignant lesions (B/M).

From 1988 to October 1998, two hundred and fifteen women with nonpalpable breast lesions suspected at mammography were examined. The lesions were removed after stereotaxic or US location and a radiograph of the surgical specimen was always performed. Mammographic patterns were interpreted retrospectively by two blinded radiologist experienced in breast imaging and specialized in locating nonpalpable breast lesions. Mammographic patterns were classified as poorly/highly suspicious calcifications, regular/irregular masses, spiculated masses, masses with calcifications and parenchymal distortions. Radiographic findings were compared with surgical results and the data used to calculate the B/M, positive predictive value (PPV) for malignancy and the trend of operator's performance.

Modern techniques permit to detect a very high number of in situ breast carcinomas. Nineteen of 22 lesions (86%) were detected by mammography as highly suspicious calcifications, 2/22 as spiculated masses and 1/22 as a mass with calcifications. No in situ carcinoma was detected as an irregular mass. All regular masses were proven to be benign at histology. B/M analysis showed a decreasing trend (from 1.94 in the first 3 years to .57 in 1994-96, to .83 in 1997-98) and an overall value of .90. The PPV for malignancy was 83.33% for spiculated masses, 65.5% for highly suspicious calcifications, 63.63% for irregular masses, 47.05% for masses with more or less dysmorphic calcifications, 32.65% for poorly suspicious calcifications, 8.33% for parenchymal distortions and 0% for regular masses.

All spiculated masses and highly suspicious calcifications and microcalcifications should be removed. Biopsy is recommended in parenchymal distortions, despite its low predictive value for malignancy, because these lesions are uncommon and the cost of biopsy is therefore acceptable. Needle aspiration or long-term monitoring can be reconsidered for irregular masses and poorly suspicious microcalcifications. Finally, relative to possible different interpretations of mammographic patterns by center and operator's experience, we suggest that the PPV for every single pattern be continually reassessed based on personal case records rather than on literature data. This holds true especially for microcalcifications.