Immunophenotypic overlap between adenoid cystic carcinoma and collagenous spherulosis of the breast: potential diagnostic pitfalls using myoepithelial markers.

Adenoid cystic carcinoma of the breast is a rare neoplasm whose cribriform architecture may mimic invasive cribriform carcinoma, cribriform ductal carcinoma in situ, and collagenous spherulosis. The diagnosis may be even more challenging in needle core biopsies. Immunohistochemical expression of p63 and c-kit distinguishes adenoid cystic carcinoma from invasive cribriform carcinoma and ductal carcinoma in situ. A formal comparison of the immunophenotype of adenoid cystic carcinoma to collagenous spherulosis has not been reported. Of concern is the overlap in myoepithelial markers between these two entities. Both may express S100, smooth muscle actin, and p63. This overlap may cause diagnostic confusion yet is under-emphasized in the literature. The expression profile of newer myoepithelial markers has not been studied in this setting. We evaluated smooth muscle actin, p63, calponin, smooth muscle myosin heavy chain, as well as c-kit, in nine cases of cribriform pattern adenoid cystic carcinoma of the breast in comparison to 12 cases of collagenous spherulosis. Both entities strongly expressed p63 and smooth muscle actin; in adenoid cystic carcinoma, the basaloid myoepithelial-like tumor cells expressed these markers, but the ductular epithelial cells did not. Neither calponin nor smooth muscle myosin heavy chain was expressed in adenoid cystic carcinoma but both were strongly expressed in collagenous spherulosis. Whereas the ductular epithelial cells of adenoid cystic carcinoma were positive for c-kit in all cases, collagenous spherulosis was negative for c-kit. Positive p63 expression by a cribriform breast lesion is not sufficiently specific to confirm a diagnosis of adenoid cystic carcinoma. A broader panel that includes calponin or smooth muscle myosin heavy chain and c-kit is required to exclude collagenous spherulosis in settings in which the distinctive morphologic features that separate these entities are not conspicuously present. Reliance on p63 or smooth muscle actin alone poses a potential diagnostic pitfall in evaluating cribriform breast lesions.