Ductal carcinoma in situ with spindle cells:: a potential diagnostic pitfall in the evaluation of breast lesions

Aims: To evaluate the morphological features of 11 cases of breast ductal carcinoma in situ (DCIS) with spindle cells and to propose an approach to distinguish it from benign mimics. The association with neuroendocrine differentiation was also investigated. Methods: Cases of breast DCIS with a spindle cell component diagnosed in the Department of Pathology, Singapore General Hospital, between June 1996 and January 2003, were included in the study. The histological characteristics were documented, and immunohistochemistry for neuroendocrine markers, hormone receptors, cerbB2, smooth muscle actin (SMA) and high-molecular-weight (HMW) cytokeratins, was carried out. Electron microscopy was carried out on reprocessed paraffin-embedded material in three cases. Results: Of 11 women diagnosed with DCIS with spindle cells, four presented with nipple discharge, six with a breast lump, while one was discovered to have a screen detected density. The tumour size ranged from 3 to 41 mm. The proportion of spindle cells varied from 10% to 80% of the in-situ tumour cell population. Nuclear grade was low in seven cases and intermediate in four. Necrosis was observed in two cases. Architectural pattern was papillary in six cases, and mixed in the rest. Microinvasion was present in two cases, with possible microinvasion in another two. Immunohistochemistry for neuroendocrine markers synaptophysin and chromogranin showed positive reactivity for at least one marker in all but three cases; one of these latter cases demonstrated ultrastructural neurosecretory granules. Oestrogen and progesterone receptors were expressed in 10 and nine cases, respectively, while cerbB2 was positive in only one case. HMW cytokeratin immunoprofile revealed a general lack of immunostaining within the abnormal cell population; likewise, no positivity for SMA of the cellular proliferation was detected. Conclusions: Almost all DCIS lesions with spindle cells disclose neuroendocrine differentiation. Although the distinction from benign florid usual hyperplasia may pose a diagnostic histological problem, the presence of diffuse neuroendocrine expression, in conjunction with the pattern of HMW keratin profile on immunohistochemistry, supports an in-situ neoplastic process. The absence of SMA immunostaining, in conjunction with negative reactivity for cytokeratins 5/6 and 14, makes the possibility of a myoepithelial proliferation unlikely.