The cytologic features of this lesion correlated well with the histologic findings. Architecturally, the hypercellularity and haphazard cellular arrangements were clearly demonstrated, as were several of the cellular findings including the spindle cell morphology, nuclear pleomorphism, conspicuous nucleoli, and the presence of mitoses. A feature that appeared more prominent on cytology, however, were the epithelioid and plasmacytoid morphologies of a large number of the tumor cells, whereas on histology the neoplastic cells were predominantly spindle-shaped, albeit plump.
There has been one report describing the cytologic findings of an intraductal mammary MEC [1
]. The cytology, as in our case, showed cohesive cell groups composed of spindle cells with cigar-shaped nuclei showing atypia and mitoses. These spindle cells were also admixed with what they describe as "polygonal" cells, which appear quite similar to the epithelioid cells in our specimen. In contrast to our case, however, the authors report that the cells were organized in a fascicular pattern and that a subpopulation of cells with clear cytoplasm was present.
Furthermore, the cytologic features of a pure malignant MEC have recently been described by Sauer and are quite similar to our findings [2
]. In contrast to our case, however, the malignant cells presented mostly as single cells rather than in clusters. Moreover, we did not appreciate the presence of occasional nuclear inclusions as did Sauer.
A wealth of different spindle cell lesions of the breast exist, translating into an accordingly wealthy number of differential diagnoses for a cytologic specimen displaying a predominant population of spindle cells. But, because of the multiple worrisome cytologic findings in this case, the list of likely diagnostic possibilities heavily favored malignant entities over benign processes. Such cytologic features have been discussed by Darvishian and Lin in which 17 myoepithelial cell-rich lesions, though mostly of salivary gland origin, showed that pleomorphism, coarse nuclear chromatin, prominent nucleoli, mitoses, and necrosis were observed only in the malignancies; 89% (8/9) of these malignant lesions were eventually diagnosed by histology as myoepithelial carcinoma [3
Various mammary spindle cell malignancies typically show markedly pleomorphic histopathology and therefore may appear more or less indistinguishable from MEC on cytology. Metaplastic (sarcomatoid) carcinoma, spindle cell carcinoma, malignant fibrous histiocytoma, and other sarcomas all may demonstrate atypical spindle cells along with other common stigmata of cancer like necrosis and mitotic activity [4
]. In some instances, the presence of such features as a chondromyxoid background and atypical multinucleated giant cells in a case of metaplastic carcinoma may generously help in eliminating MEC from the differential.
Myoepithelial cells may adopt a number of different morphologies on cytologic specimens including spindle cell, clear cell, epithelioid, and plasmacytoid forms [3
]. In a study by Hornick and Fletcher, over half of their 101 cases of soft tissue myoepitheliomas actually showed a mixed pattern of these four morphologies [7
]. Our case did not reveal a subpopulation of clear cells but did display the other three morphologies whereas the MEC reported by Sauer described both the spindle cell and epithelioid ("polygonal") cell populations [2
]; the intraductal case reported by Tamai et al. lacked only the plasmacytoid cell type [1
]. The heterogeneity with which myoepithelial cells may present on cytology can therefore complicate the diagnostic process for myoepithelial carcinoma even further, obviating a need to expand the differential diagnosis beyond spindle cell lesions to include biphasic entities such as metaplastic carcinoma and malignant phyllodes tumor [8
In most cases of metaplastic carcinoma, the carcinomatous component is identified on cytology, allowing for a smoother diagnostic process. The problem arises when only the mesenchymal population is sampled. Strong keratin staining of the mesenchymal elements favors a metaplastic carcinoma over MEC, but differentiating between the two entities from a cytologic specimen may not always be possible.
It may be appropriate to entertain the diagnosis of lobular carcinoma or metastatic melanoma, simply because some of the tumor cells in our case displayed a plasmacytoid morphology. The presence of signet-ring cells in lobular carcinoma and intracytoplasmic pigment in melanoma can distinguish these entities from MEC [3
Ultimately, in order to subclassify a spindle cell lesion like MEC based solely upon cytomorphology, immunohistochemistry will most likely be needed. By showing that the lesion is reactive for the immunohistochemical markers SMA, S-100, cytokeratin, p63, and CD10, a myoepithelial origin for the tumor can be established [3
]. In our case, the immunohistochemical panel was performed on tissues derived from the mammotome biopsy, as the cytologic specimen was inadequate for staining procedures.