Medullary carcinomas constitute fewer than 5% of most series of breast carcinomas (2,3,4,5). Patients as young as 21 years (6) and as old as 95 years (7) with medullary carcinomas have been reported. This broad range of ages notwithstanding, patients with medullary carcinoma tend to be relatively young. The mean age in several series ranges from 45 to 54 years (6,8,9,10). Medullary carcinoma occurs in the male breast only very rarely (11).

The anatomic and size distributions of medullary carcinomas do not differ from those of commonplace breast carcinomas. Medullary carcinoma is not especially common among patients with bilateral mammary carcinoma. On the other hand, bilateral carcinomas have been found in 3% to 12% of patients with medullary carcinoma (2,7,12). Synchronous or metachronous medullary carcinoma involving both breasts is very uncommon (7,12).

Because of the presence of reactive hyperplasia, ipsilateral axillary lymph nodes tend to be enlarged in patients with medullary carcinoma even in the absence of nodal metastases. This phenomenon may complicate clinical staging (13), and the greater ease of detecting enlarged hyperplastic lymph nodes accounts for the larger number of lymph nodes retrieved from axillary dissection specimens from patients with medullary carcinoma compared with specimens from patients with conventional carcinoma (14).

Radiologic images of medullary carcinomas typically demonstrate a dense, round or oval mass with indistinct or lobulated borders lacking calcifications (15). Because they have circumscribed margins, medullary carcinomas may be mistaken for fibroadenomas. Radiologic findings cannot differentiate medullary carcinoma from circumscribed nonmedullary carcinoma (16), but a mass with an irregular or jagged margin is unlikely to be a true medullary carcinoma.

The typical intact medullary carcinoma is a moderately firm discrete tumor that one can mistake for a fibroadenoma. A distinct margin usually outlines the tumor and demarcates it from the surrounding tissue; however, an intense lymphoplasmacytic infiltrate extending beyond the immediate perimeter of a small carcinoma may blur its border. The carcinoma has a lobulated or nodular internal structure. Hemorrhage and necrosis occur at times, even in medullary carcinomas smaller than 2 cm; however, the extent of necrosis is directly related to tumor size. As the extent of necrosis increases, so, too, does the likelihood that the tumor will develop cystic foci.

It is necessary to adhere to established morphologic criteria strictly if the diagnosis of medullary carcinoma is to be predictive of a relatively favorable prognosis (7,8). Although researchers have proposed modified criteria, those set forth by Ridolfi et al. (8) remain the most reliable for detecting survival differences between true medullary, atypical medullary, and nonmedullary carcinomas (17).

Medullary carcinoma is defined by the following constellation of histopathologic features: prominent lymphoplasmacytic infiltrate, noninvasive microscopic circumscription, growth in sheets (syncytial pattern), poorly differentiated nuclear grade, and a high mitotic rate. A tumor must display all of these attributes to merit classification as a medullary carcinoma. When most but not all of these findings are present, the tumor may be termed an infiltrating ductal carcinoma with medullary features. Tumors in the latter category have a syncytial growth pattern and certain other histologic features of medullary carcinoma but deviate from the definitive appearance by demonstrating one or more of the following structural variations: invasive growth at the periphery of the tumor, sparse or diminished lymphoplasmacytic reaction, well-differentiated nuclear cytology, low mitotic rate, or conspicuous glandular, trabecular or papillary growth with fibrosis. Invasive ductal carcinomas with medullary features can display immunohistochemical findings seen in typical medullary carcinomas, but they do so less frequently. One cannot make a conclusive diagnosis of medullary carcinoma from examination of a needle core biopsy specimen because of the limited nature of such samples. In these circumstances, it is appropriate to report that the findings raise the possibility of medullary carcinoma and that the final classification depends on evaluation of the completely excised tumor.

The lymphoplasmacytic reaction must involve the periphery and be present diffusely in the substance of the tumor. The internal lymphoplasmacytic infiltrate tends to be limited to fibrovascular stroma between syncytial zones of tumor cells. Uncommon medullary carcinomas seem to be largely devoid of stroma, and the lymphoplasmacytic infiltrate mingles intimately with carcinoma (Fig. 12.1). One can find it difficult to differentiate such a tumor from metastatic carcinoma in a lymph node, especially when the mass occupies the lateral aspect of the breast. The presence of uninvolved lymph node and nearby in situ carcinoma are useful distinguishing findings.

At the periphery of the tumor, the amount of the lymphoplasmacytic infiltrate may vary, but it should appear at least moderately intense at the interface of the carcinoma with the mammary parenchyma and in the adjacent tissue. In the usual case, the lymphoplasmacytic reaction encompasses adjacent ducts and lobules occupied by in situ carcinoma. The inflammatory cells also tend to surround more distant ducts and lobules, which do not contain identifiable tumor cells. These secondary peripheral alterations are so common in medullary carcinoma that the diagnosis may be questioned in their absence.

The lymphoplasmacytic infiltrate may be composed almost entirely of either lymphocytes or plasma cells, but one most often finds a mixture of these cells. Bässler et al. (18) reported that lymphocytes predominated at the periphery of medullary carcinomas, whereas plasma cells represented the preponderant inflammatory cell in the center of the carcinoma. This phenomenon does not occur in every case. Because intense lymphocytic infiltrates can occur in nonmedullary infiltrating ductal carcinomas, this finding does not have diagnostic significance. A predominance of plasma cells, on the other hand, favors the diagnosis of medullary carcinoma. A few neutrophils, eosinophils, and monocytes can be found in a medullary carcinoma, especially in the presence of necrosis or cystic degeneration, but they never dominate. Rarely, the lymphocytic infiltrate gives rise to germinal centers within the tumor or in the surrounding tissue. Hence, one cannot rely on the presence of germinal centers as evidence that one is dealing with metastatic carcinoma in a lymph node.

Noninvasive microscopic circumscription refers to the appearance of the border of the infiltrating carcinoma rather than the periphery of the surrounding lymphoplasmacytic reaction. In medullary carcinoma, the edge of the tumor should have a smooth, rounded contour that appears to push aside the breast parenchyma rather than to infiltrate it (Fig. 12.2). Consequently, non-neoplastic glandular or fatty breast tissue should not be found within the body of the invasive portion of the tumor. In assessing the margin of the carcinoma, one must not confuse the extension of inflammatory cells into the surrounding parenchyma with invasion of glandular tissue or
fat by the carcinoma cells. In the latter instance, the tumor cells tend to grow in trabecular, dendritic, or dispersed patterns, and the carcinoma typically lacks the cohesive syncytial structure of a medullary carcinoma.

A syncytial growth pattern refers to the formation of broad irregular sheets or islands of carcinoma cells in which the borders of individual cells are indistinct (Figs. 12.2 and 12.3). The appearance sometimes resembles that of a poorly differentiated squamous carcinoma. A tumor that is otherwise characteristic may be accepted as a medullary carcinoma if it has minor components of trabecular, glandular, alveolar, or papillary growth. Such regions may have a diminished lymphoplasmacytic infiltrate and fibrosis and thus appear distinct from the medullary growth pattern. It has been reported that overall survival and relapse-free survival are directly related to the extent of the syncytial component (19). Although there was not a significant difference in outcome between patients with 75% and 90% syncytial growth, survival was diminished when less than 75% of a tumor was syncytial, and the difference was most marked at and below the 50% level. These data justify the currently employed requirement for at least a 75% syncytial component for a diagnosis of medullary carcinoma.