PNET-like features of synovial sarcoma of the lung: a pitfall in the cytologic diagnosis of soft-tissue tumors.

Fine-needle aspiration (FNA) cytology of soft-tissue tumors is evolving. As more experience is gained, we are becoming aware of potential pitfalls. We describe 2 cases of synovial sarcoma of the lung, primary and metastatic, in patients who had FNA biopsy performed on a lung mass. The cytologic smears showed extremely cellular groups of malignant small round cells, intersected by small blood vessels, with numerous loose single cells, in a background of macrophages and mature lymphocytes. The tumors displayed monomorphic cells forming rosettes and displaying occasional mitoses. A diagnosis of neuroendocrine tumor/primitive neuroepithelial tumor (PNET) was suspected. Furthermore, this suspicion was supported by immunohistochemical stains, which showed positivity for a neuroendocrine marker, Leu 7 (case 1), and for a neural marker, CD 99 (O 13 or HBA 71) (both cases); and negativity for cytokeratins (case 1). The resection specimen of case 1 had mostly tightly packed small round cells, with occasional rosettes, similar to the FNA biopsy, and focal areas composed of spindle cells, organized in a focal fibrosarcoma-like and hemangiopericytoma-like pattern. A balanced translocation between chromosomes X and 18, demonstrated by both karyotyping and fluorescent in situ hybridization (FISH), enabled us to make a diagnosis of synovial sarcoma, which was histologically classified as poorly differentiated. Case 2 was a metastatic biphasic synovial sarcoma of the arm, with a prominent epithelial component. Synovial sarcoma, when composed mainly of small round cells on cytologic smears, is a great mimicker of neuroendocrine/PNET tumors, with light microscopic and immunohistochemical overlap. Awareness of this potential pitfall may aid in preventing a misdiagnosis. Its recognition is of major concern, especially for the poorly differentiated variant, because it is associated with a worse prognosis. Copyright 2001 Wiley-Liss, Inc.