Mammographic-histopathologic correlation of large-core needle biopsies of the breast.

Large-core (14g) needle biopsy (CNB) of the breast is a new diagnostic modality increasingly being used to evaluate patients with mammographic abnormalities. Two hundred twenty-four CNBs were performed on 198 patients. Surgical follow-up was available in 64 cases (28.6%). Overall concordance rate was 93.8% (60 of 64 cases). Of the four discordant cases, two were diagnosed as atypical ductal hyperplasia (ADH) on CNB; on excision, these cases showed cribriform ductal carcinoma in situ (DCIS); two remaining cases, diagnosed on CNB as ADH versus DCIS, showed invasive carcinoma (DCIS with invasive component and infiltrating cribriform carcinoma, respectively) on excisional biopsy. Malignancy, primary (52) or metastatic (5), was identified in 57 cases (25.4%); 47 of these patients underwent surgical excision, and the diagnosis was confirmed in all of these cases. Of 51 cases with radiographic evidence of microcalcifications, 48 (94%) had microcalcifications in the CNB: 30 (62.5%) were benign, 11 (22.9%) were malignant, and 7 (14.6%) were diagnosed as ADH. In the remaining three cases (1.3%), only benign breast tissue without microcalcifications was seen, and the lesion was considered to have been missed. Biopsy specimens were obtained from 173 lesions because of the presence of a mass: 125 (72.3%) were benign, 45 (26%) were malignant, and 3 (1.7%) were diagnosed as ADH. Follow-up was available in 118 patients with benign lesions: all were mammographically stable or decreased at 6 or 12 months; no follow-up was available for the remaining patients. CNB of the breast is a highly sensitive (96.9%) and specific (100%) technique for management of patients with mammographic abnormalities. The histologic findings should be correlated with the mammographic appearance, and an attempt should be made to achieve a specific diagnosis in all lesions, particularly masses. The diagnosis of ADH should always prompt excisional biopsy because of a high frequency of false-negative results caused by sampling errors or underestimation.