Bone marrow involvement by lobular carcinoma of the breast cannot be identified reliably by routine histological examination alone.

The aims of this study were twofold: (1) to evaluate the ability of pathologists to recognize infiltration of bone marrow core biopsy specimens by breast carcinoma, particularly lobular carcinoma, using routine hematoxylin-eosin (HE) sections; and (2) if indicated, to determine the reasons for difficulties in diagnosis. Thirty-six bone cores obtained before bone marrow harvest were involved by breast carcinoma and were confirmed by pancytokeratin immunostains. Thirty of the 36 were ductal carcinomas and six were lobular carcinomas. Fourteen negative bone core biopsy specimens (from patients with breast cancer or lymphoma) were included as controls. These 50 bone cores were reviewed by three surgical pathologists. Lobular carcinoma was correctly identified in only 39% of positive specimens as compared with 88% for ductal carcinoma. After instruction, sensitivity for the detection of lobular carcinoma improved to 61% but at the expense of an unacceptably high rate of false-positive diagnoses (18%). None of the three pathologists was able to achieve both high sensitivity and high specificity in recognizing lobular carcinoma in the bone marrow. Lobular carcinoma was difficult to detect because of tumor cell size similar to hematopoietic cells, infiltration as single cells, presence of bland cytological features, and paucity of tissue reaction to the tumor. Although the number of cases of bone marrow involved by lobular carcinoma is small, these findings suggest that pancytokeratin stains should be performed routinely in the evaluation of bone core biopsy specimens from patients with lobular carcinoma, and probably from patients with ductal carcinoma whose HE-stained bone core biopsy specimens are considered negative for tumor.